& wireless world · plotter for instart print-outs of displayed traces, and the ability to...

THE R AL R PR FE INAL E I EER

ELECTRON., :3& WIRELESS WORLDSEPTEMBER 1988

Worldstandard forh.d.tv?

IEEE 488interface forZ80 -basedcomputers

P.l.ds andlogic probes

Flywheel step-up switchingregulatordesign

Alec Reevesand pulse -codemodulation

Dimensionalapproach tounified theory

Denmark DKr. 63.00 Germany DM 12.10 Greece Dn. 680 Holland DEL 12.50 Italy L 6500 IR 62 86Main Phis. 700.00 Singapore SS 11.21' Switzerland SFr. 8.50 USA 55.95

£1.95

The Gould 1604

Give it ascreen testThe Gould 1604 digital storageoscilloscope has the memoryand performance to tackle anyrole in low frequency electronics,mechanical or physiological R&Dand test enviror ments.

With massive 10K wordmemories on each of its 4channels the 1604 can examinedetail with expalsion factors upto 200 times, and resolutiondown to 5Ons.

And it can do so much more.

The 1604 has a built-in colourplotter for instart print-outs ofdisplayed traces, and the abilityto archive up to 50 traces in thenon-volatile memory pods. Foreven more capability, interfacethe fully programmable 1604 toa computer; or plug in thewaveform processor.

The 1604 DSO has a wide rangeof automatic functions at anunbeatable price, and it's so easyto drive!

Why not let it audition for you?For further details of the 1604 orthe 2 -channel 1602, contact:01-500 1000.Gould Electronics LimitedTest & Measurement Sales DivisionRoebuck Road, Hainault.Ilford, Essex IG6 3UETel: 01 500-1000 Telex 263785

Immo.

Digita orage sci oscope

10101I '

,,4.

qt

MODEL

"(I 1K Good) 1604-DicitrALOta

osciu,og_oPE

10K WORDtAMoo

cN 010{ CRANKa

JillILL/

O

uoutu13S 423 Int,.

I sc..

FAA,atEN NEL

opozAn oN

')D Ms15 A

AN ANPAfiA

tqfri,15 Is cAPPlesiL0

o

0 Po --L9 PRo02004Hieteke

0 IIVrT-1,,coLourz

ikorreg

102

. Olt

CAL, VMS. Va. bin

T. vi

Alg "C""7

goompilmorip-

TIVEIDIV

474

LDc:j441

Available for rental fromLivingston Hire

I 11 It I UN 1{1:1.1.1MULE

Electronic.

SEPTEMBER 1988 VOLUME 94 NUMBER 1631

COVER

High -definition television will bring todomestic viewers pictures sharper eventhan those they can see in the cinema

(off -screen photographs by courtesy of SonyBroadcast).

EUREKA 95 -A WORLDSTANDARD

845European promoters of high -definition

television prepare for a crucialdemonstration

Tom lvall

IEEE488-TO-Z80 INTERFACE852

Dedicated instrumentation controller forthe SC84 and other computers

T.I. Vellacott

PC BUS PERFORMANCE856

Two new personal computers use entirelynew bus structures.

William Nowlin

LOGIC PROBE867

Fast pulse -detecting probe based on a p.l.d.B.J. Frost

PIONEERS873

Alec Reeves, inventor of p.c.m.W.A. Atherton

MULTIPROCESSORSYSTEMS

875This third article examines the issues

involved in linking processors togetherAlan Clements

A DIMENSIONAL APPROACHTO UNIFIED THEORY

882Dimensional relationships of common

physical conceptsS.K. Chatterjee

PHILIP SMITH'S CHART -2887

JW supplies a user's guideJoules Watt

STEP-UP SWITCHINGREGULATOR

891High -efficiency design for use with NiCd

cells.N.J. Chaffey and T.A. Perkins

L.F. NAVIGATION REVIVAL

897Tracking vehicles by radio

Richard Lambley

DEPENDENT SOURCETHEOREM

900Finding equivalent generators in a network

which has no independent sourcesHarry E. Stockman

PIEZOELECTRIC COAXIALCABLE

905Applications of a novel composite material

Peter Johnson

Loran brought up to date: a new nationwideradio position -finding system is describedon page 897.

KEYBOARD DESIGN

910Details that designers sometimes overlook

Jeff Wright

ANALOGUE CIRCUITS FORAUTOMOTIVE USES

915Spike -resistant chips for cars

Rocco Shah

PAN -EUROPEAN CELLULARDEMONSTRATION

919Digital radiotelephone success

BROADBANDINSTRUMENTATION

AMPLIFIER937

Wide bandwidth plus good c.m.r.r.Nanno Herder

SUPERCONDUCTING FETS 850

NOVEL POWER SUPPLY 885

ASSERT COMPETITION 918

SCI -TECH '89 918

OBITUARY 918

7`a 'VON 11111

COMMENT 843

EVENTS 854, 881

LETTERS 860

CIRCUIT IDEAS 864

BOOKS 869

APPLICATIONS SUMMARY 871

SATELLITE SYSTEMS 903

TELECOMMS TOPICS 908

NEW PRODUCTS 920, 927

RESEARCH NOTES 929

TELEVISION BROADCAST 932

RADIO COMMUNICATIONS 934

RADIO BROADCAST 936

ELECTRONICS & WIRELESS WORLD 841

TAYLOR RF/VIDEO MEASUREMENT INSTRUMENTS

MEASUREMENTSMADE EASY

UNAOHM FSM5987 T.V. FIELD STRENGTH METERINPUTSensitivity: from 20dBuV to 110dBuV (-40dBmV to 50dBmV) or 10uV to 0.3V,

in eight 10dB steps.Reading: dB reading proportional to peak value for video signals; proportional

to mean value for AM or FM sound signals. For botn signals scalecalibrated to rms value and expressed in dBuV Two mo-e scales areavailable: volt from 0 to 50, and ohm from 0 to 21300 ohm. Batterystatus is also provided.

Accuracy: +/- 3dB for bands I & III +/- 6dB for bands H & IVNImpedance: 75 ohm unbalanced; DC component blocked up to 100V

FREQUENCYRange: 46 to 860 MHz as follows:

Reading:Price:

Band I 46 to 106MHzIII 106 to 206MHzH 2C6 to 460MHz

IVN 460 to 860 MHz4 digit LCD readout. 100KHz resolution£378.00 exc. VAT and Carriage.

VISA

TAYLOR BROS (OLDHAM) LTD.BISLEY STREET WORKS, LEE STREET,OLDHAM, ENGLAND.TEL: 061-652 3221 TELEX: 669911 FAX: 061-626 1736

UNAOHM EP741FMSFIELD STRENGTH METER/SPECTRUM ANALYZERFrequencyRange:

FrequencyReading:

Function:TV Monitor

Panorama:

PanoramaExpansion:AnalogueMeasurement:

Continuously adjustable via a geared -down vernier as followsIF 38.9MHzBand I 46 to 106MHzFM Band 88 to 108MHzBand III 106 to 290MHzBand H 290 to 460MHzBand U 460 to 860MHzTV Bands -4 digit counter with 100KHz resolutionFM Band - 5 digit counter with 10KHz resolutionReading Accuracy: reference Xtal +/-1 digit.NORMAL: picture onlyZOOM : 2 to 1 horizontal magnification of picture

picture + line sync pulse (with chromaburst if TV signalis coded for colour]

panoramic display of the frequency spectrum within the selectedband and of tuning markerAdjustable expansion of a portion of the spectrum around the tunedfrequency.

20 to 40dB. Static measurement of received signal. Scale calib-rated in dBuV [at top of picture tube] to rms value of signal level.

DC/AC Voltmeter: 5 to 50VMeasurementRange:

MeasurementIndication:

Video Output:DC Output:TV Receiver:

AdditionalFeatures:Price:

20 to 130dBuV in ten 10dB attenuation steps for all bands.-60 to 130dBuV in nine 10dB steps for IF.ANALOGUE: brightness stripe against calibrated scalesuperimposed on picture tube. The stripe length is proportionalto the sync peak of the video signal.BNC connector 1Vpp max on 75 ohm.+12V/50mA max. Power supply source for boosters & converter

tunes in and displays CCIR system 1 TV signals. Other standardsupon request.

(1] Video input 75 Ohm. [2] 12V input for external car battery.(3] Output connector for stereo earphones.£1344.00 exc. VAT and Carriage.

UNAOHM EH 1000 TELETEXT AND VIDEO ANALYZERFunction:

RF Input:

Video FrequencyInput:Teletext Input:Teletext ClockInput:

Oscilloscope:

Price:

Eye Pattern: display of RF and video -frequency teletext signals bymeans of eye pattern diagrams both in linear representation andlissalous figures (0 and X). Line selection: display of video signalsand line by line selection. Measurement of modulation depth.Teletext: monitoring of teletext pages.

Freq. Range: 45 to 860MHz. Frequency synthesis, 99 channelrecall facility, 50KHz resolution, 30 channel digital memory. Level:40 to 120dBuV: attenuator continuously adjustable. Indication ofthe minimum level for a correct operation of the instrument.Impedance: 75 ohm. Connector type: BNC.Minimum Voltage: 1Vpp. Impedance: 75 ohm or 10K ohm in caseof a through -signal. Connector type: BNC.Voltage: 1Vpp/75 ohm.

Voltage: 1Vpp/75 ohm. Measurement, Aperture of eye pattern:linear or Lissalous figures, selectable. Indication: directly on thepicture tube. A calibrated scale shows percentage of eye patternaperture. Error: the instrument introduces an error of < 5% withvideo input and 20% with RF input. Jitter on regen'd clock: <25ns.Line selector: Selection of any TV line between the 2nd and the625th scanning cycle by means of a 3 digit thumbwheel switch.

VERTICAL CHANNEL: Sensitivity: 0.5 to 2Vpp/cm. FrequencyResponse: DC to 10MHz. Rise time: pre & overshoot < 2%. InputCoupling: AC. Input Impedance: 75 ohm/50pF.TIME BASE: Sweep Range: 20 to 10ms (1.1/2 frames); 32;64/192us [1/2; 1; 3 lines]. Linearity: +/-3%. Horizontal Width:10 divisions; x5 magnification£1870.20 exc. VAT and Carriage.

ENTER 40 ON REPLY CARD

COMMENT

Predicting theunpredictable

EDITORPhilip Darrington

EDITOR - INDUSTRY INSIGHTGeoffrey Shorter, B.Sc.

01-661 8639DEPUTY EDITOR

Martin Eccles01-6618638

COMMUNICATIONS EDITORRichard Lambley

01-661 3039ILLUSTRATIONRoger Goodman

01-661 8690DESIGN & PRODUCTION

Alan Kerr01-661 8676

ADVERTISEMENT MANAGERMartin Perry01-661 3130

ADVERTISEMENT EXECUTIVEJames Sherrington

01-661 8640CLASSIFIED SALES EXECUTIVE

Peter Hamilton01-661 3033

ADVERTISING PRODUCTIONBrian Bannister

01-661 8648Clare Hampton01-661 8649

MARKETING EXECUTIVERob Ferguson01-661 8679PUBLISHER

Susan Downey01-661 8452

REEDBUSINESSPUBLISHING

How technological advancements will affect the future is remarkably difficult topredict. There is no need to go back to the industrial revolution to find evidence of thisstatement. As many of you will remember, not too many years ago experts predictedthat the country would soon be run by a big brother computer.

Were it not for microprocessors we might now have a big brother. Admittedly thereare a few large central computers doing mainly specific tasks, and the day might stillcome when our destinies are determined by a central supercomputer, but the majorityof today's processing is done locally - in offices. laboratories, shops, factories and inthe home.

And, on the subject of predictions, who can say with any certainty whethercomputers will eventually create jobs or cause even more unemployment? Popularanswers to this question are frequently evaluated from political rather than computerknowledge, which is perhaps what flawed the central -computer theory.

One of the bolder predictions made recently comes from researchers at the GeorgiaInstitute of Technology who say that by the year 2000, operators will be instructingcomputers in plain language, not by entering commands on a keyboard. That in itselfis not a particularly profound prediction, but researcher David Roessner goes on tomake a point with much wider implications. As a result of advances in computertechnology, and in particular in the computer -to -person interface. he says that we willbe "thrown back on those skills that are uniquely human - skills such as knowing whatquestions to ask, how to interpret information, and how to judge when acomputer -generated answer does not make sense."

Our schools are certainly working hard at 'computer -education'. but mightemphasis be turning away from teaching the disciplines of computer technology andprogramming -which is essentially problem solving - towards training in how to typeand how to use programs such as word processors, data bases and spread sheets? Onehopes not. but since it has become so clear that the health of a nation relies on thesuccess of its industry and commerce such a situation is justifiable should you put thelong-term future second to current needs.

Getting and holding a job in the future, says Roessner, will require literacy, notcomputer literacy. If he is right, there will be problems for everyone should we decideto train computer use rather than continue to teach problem solving usingcomputers. Young people will leave school expecting to use their hard-earned trainingand be disappointed, employers will find a shortage of suitable staff, and the economywill suffer through the resulting short -fall in production.

Of course. nothing indicates that every future worker will need to be a computerprogrammer. It might be that our current information -technology managers andcomputer programmers will evolve side by side. and jobs for everyone else mightsuddenly appear, be created, or never have disappeared in the first place.

Two things are certain from this reminder that new technology might not producethe results that seem immediately apparent. One is that is that any child emergingfrom school with the ability to solve problems will make a better contribution tosociety than one emerging with a sound knowledge of how current computerprograms work, and the other is that any unbiased prediction of the effects of newtechnology is better than none.

Electrinars & Winless World is published monthlyUSPS687540. By post, current issue £2.25. hack issues ifavailable, £2.50. Order and payments to 301 Electronicsand Warless World. Quadrant House, The Quadrant. Sut-ton. Surrey SM2 5AS. Cheques should be payable to ReedBusiness Publishing Ltd. Editorial & Advertisingoffices:EWW Quadrant House. The Quadrant. Sutton, Sur-rey SM2 5AS. Telephones: Editorial 01-661 3614. Adver-tising 01-661 3130 01-661 8469 Telex: 892084 REED BP

EEP Facsimile: 01-661 :19414 (Croups II & 111, Beeline:01-661 8978 or 01-661 8986.:100 baud. 7 data bits. evenparity. ale stop -bit. Send ctrl -Q. then EWW to start; NNNNto sign off. Newstrade -- Quadrant Publishing Services No.01-661 :3240. Subscription rates: I year ,normal rate,I:23.40 UK and V.94.50 outside UK. Subscriptions: Quad-

rant Subscription Services. Oaklield House. PerrymountRoad. Ilaywards Heath. Sussex RH163 MI. Telephone 0444441212. Please notify a change of address. USA: $116.00airmail. Reed Business Publishing [USA,. SubscriptionsOffice. 205 E. 42nd Street. NY 10117. Overseas advertis-ing agents: France and Belgium: Pierre Mussard, 18-20Place de la Madeleine. Paris 75008. United States ofAmerica:Jay Feinman. Reed Business Publishing Ltd. 205East 42nd Street, New York. NY 10017. Telephone 12121867-2080 Telex 2:1827. USA mailing agents: Mercury Air -freight International Ltd. Inc.. 10,10 Englehard Ave.Avenel N.J. 07001. 2nd class postage paid at Rahway NJ.Postmaster - send address to the above.

©Reed Business Publishing I.td 1988. ISSN 0266-3244


"It's a pretty small battery -powered PROM programmer- so what?"

Tools which are convenient get used atheir existence. There is no wayexplain all the usefulness of S3 heInstead, if you're interested we'regoing to let you see it, use itand evaluate it in your ownworkshop. We went to a lotof trouble to design S3 justthe way it is - no otherPROMMER is all CMOS andall SMT. So we must be con-vinced that S3 would be a for-midable addition to yourarmoury. Now all we have to dois to convince you.

"Such a little thing can't bepowerful, like a big bench -programmer - er - can it?"Yes, it can. It is more powerful. S3 leavesother prommers streets behind. S3 hascontinuous memory, which means that youcan pick it up and carry -on where you left offlast week. S3 has a huge library EPROMS andEEPROMS. S3 can blow a hundred or morePROMS without recharging. S3 also works re-motely, via RS232. There s a DB25 socket on theback. All commands are available from your computer(through a modem, even). Also S3 helps you developand debug microsystems by memory -emulation.

"What's this memory -emulation, then?"It's a technique for Microprocessor Prototype Develop-ment, more powerful than ROM emulation, especiallyuseful for single -chip "piggyback" micros. You plug thelead with the 24/28 pin headerin place of the ROM/RAM.You clip the Flying -Write -Lead to the microprocessorand you're in business. Thecode is entered using eitherthe keyboard or the serialinterface. Computer -assembledfiles are downloaded in standard format - ASCII, BINARY,INTELHEX, MOTOROLA, TEKHEX.Your microprocessor can WRITE to S3 as well as READ. Youcan edit your variables and stack as well as your program, ifyou keep them all in S3.S3 can look like any PROM up to 64K bytes, 25 or 27 series.Access is l(X)ns - that's really fast. Memory -emulation ischeap, it's universal and the prototype works "like the realthing".

lot - that justifieswe couldre.

S3 loads its working programs out of aPROM in its socket, like a computer

loads from disk. Software expansion isunlimited. Upgrades will come in a

PROM. Programs can be exchangedbetween users. How's that for

upgradability?

"Can I change the way it works?"You surely can. We keep no secrets. System Variables can he"fiddled." New programming algorithms can be written fromthe keyboard. Voltages are set in software by DACs. If youwant to $et in deeper, a Developers' Manual is in preparationwhich will give source -code, BIOS calls, circuit -diagrams, etc.We expect a lively trade in third -party software e.g. dis-assemblers, break -point -setters and single -steppers for variousmicros. We will support a User Group.

0 "I'm bound to let the battery go flat."I. Quite so. But in practice it doesn't matter. S3 switches

off after a half-hour of non-use anyway, or when thetr battery gets low. You don't lose your data. Then a

low -charge overnight or boost -charge for threehours will restore full capacity. You can keep using!t when charging. So there really is no problem.

"I already have a programmer."Pity it doesn't have S3 features, eh? Buthere's a trick worth knowing. If you plug

S3's EMULead into the master socket of aganger then you get an S3 with gangcapacity. Isn't production separatefrom development anyway?

"It looks nice. Will I bedisappointed?"

Dataman tools are designed tobe used by Engineers. Notjust sold to Management.

Have you ever been misledby some mouthwateringad for a new product?Great artwork and ex-

citing promises whichfeed your fancy? Onimpulse you buy and

when the thing arrivesyou feel let down. The pic-

ture looked better. The claimshardly justified; not exactly

misrepresentation, justpoor implementation.But you've bought it.And you're stuck with it.

are

Softy3 is here!It stays in the cupboard,most of the time. So howabout this: buy S3 anduse it for up to a month.If you're not still thrilledthen you can have yourmoney back.

"Refund in the first month! How can you offerthat?"We trust S3 to fire your enthusiasm. We trust you not to use usas a free hire -service. We bet you won't send it back. Howwould you manage without it?

"These things cost a fortune and take months toarrive."We wouldn't get you all excited and then let you down. It Costs£495 plus VAT. That includes P & P, Charger, EMULead,Write Lead and a HELP program in ROM. S3 is in stock. Buyit today. Use it tomorrow. (That's a fair promise. But pleasereserve product by phone or telex to make it come true).

DAIAribinLombard House,Cornwall Rd.DORCHESTERDorset DT1 1 RX.Phone 0305 68066 Telex 418442If you purchase while this ad is current, you have 28 days toexamine the goods and return them for refund. Carriage will becharged at cost. The right to charge the cost of refurbishmentof damaged goods is reserved.

VISA


844 ELECTRONICS & WIRELESS WORLD

Eureka 95 -aworldstandard?

Research teams all over Europe are racing to make ready acrucial h.d.tv demonstration

demonstration to be heldatat Brighton in September

ill form a significant stepin the European attempt toestablish a compatible high -definition television stand-ard. It could also be crucialin an international decision.to be made eventually by theCCIR. on whether this standardwill be acceptable as a worldh.d.tv standard.

The demonstration is beingorganized by member countriesresponsible for the Eureka EU95h.d.tv project' (an outline of thiscollaborative European schemewas given in the March 1988issue. p.294). It will take place atthe site of the InternationalBroadcasting Convention, 23-27September. but will not be anofficial part of the conventionand will not be open to all IBCvisitors. It has a limited ob-jective: to demonstrate a working chain ofhardware based on an interim h.d.tv produc-ion standard of 1250 lines. 50 fields persecond and a 2:1 interlacing factor.

This hardware is designed as if for acompatible broadcasting system - an essen-ial feature of the European proposal. Thecompatibility will be with the MAC/packetd.b.s. transmissions to be started soon inEurope (e.g. in Britain from the BSB satel-lite now being prepared for launch). Sincethe MAC d.b.s. signals will be displayedthrough MAC decoders on existing standardPAL and SECAM television sets. the EU95h.d.tv transmissions would therefore becompatible with these 625/50/2:1 receivers.And, of course, the h.d.tv pictures transmit-ted via the d.b.s. channels would also bedisplayed on new high -definition televisionsets as and when these became available.

Demonstration of the interim productionstandard at Brighton is described as a "li-mited objective" because it is only one stepon the way to a complete and viable system.In the past the television production stan-dard was the same as the transmissionstandard and the receiver display standard.Then standards converters working be-tween, say 525 and 625 -line systems, showedthat a production standard need not, in fact.be the same as the transmission (= display)standard.

TOM IVALL

Bosch h.d.tv camera for Eureka 95 system (see July issue, 708710for a description).

Now, digital signal processing has adv-anced to such an extent, and its cost reducedthrough v.I.s.i. devices, that all three can bedifferent from each other where it serves apractical purpose. For example, large -areaflicker, resulting from a combination of fieldrate and screen brightness, is exclusively aproblem of the display standard. It occurswith screen areas subtending a large angle atthe eye because the outer part of the retina ismore sensitive to flicker than the fovea orcentral part. Component manufacturers aretherefore now beginning to supply chip setswith digital field memories for receivers toenable incoming 50 field's transmissionstandard pictures to be displayed at 100Hzfield rate in order to eliminate this flicker.

HISTORICAL PERSPECTIVE

As reported in March, the Eureka 95 col-laborators are proposing a single worlddigital production standard of 1152/50/1:1(sequential scan) with a 16:9 aspect ratio and1920 samples per active line for luminance(960 for colour). At Brighton they will onlydemonstrate an interim production stan-dard because it is as far as they can get in arace against time. An earlier demonstration,at the 1987 Berlin radio and tv show. waslittle more than a statement of intent and didnot include any encoding/decoding hard-

ware. The race results from theneed to meet an important dead-line - the next plenary assemblyof the CCIR in May 1990. As apermanent body of the ITU, theCCIR is the world organizationwhich makes international de-cisions on broadcast engineeringstandards.

Eureka 95 made a late start inthe h.d.tv stakes because thecollaborative organization wasonly formed in 1986, immed-iately after the last CCIR plenaryassembly held in Dubrovnik.Essentially it was a Europeanreaction to the possibility thatthe NHK non -compatible1125/60/2:1 h.d.tv standardcould become a de facto worldstandard for broadcasting. Thiswas the only working h.d.tvsystem offered as a candidate atDubrovnik and was officiallysupported by the USA. Japan,

Canada. Brazil and Chile. It includes abandwidth -compression technique calledMuse (multiple sub-Nyquist samplingencoding' outlined in the December 1984issue of E&WW, p.38.

Most European delegates at Dubrovnikstrongly opposed this 'revolutionary' non -compatible system because they felt it wouldbe disastrous for Europe. A major objectionwas the incompatible 60Hz field rate. Theystressed that they wanted an 'evolutionary'approach to h.d.tv starting from the otherend - the existing base of PAL and SECAMreceivers in Europe. These millions of 50Hzreceivers were not only a huge investmentin domestic equipment which should beutilized instead of ignored: in a compatiblesystem they would also provide animmediate audience for the first h.d.tvprogrammes - before new h.d. receiversbecame generally available - and thus givea commercial incentive to programmecompanies to produce material for the newmedium.

As a result of this disagreement the 1986CCIR meeting ended in a stalemate on thequestion of a world standard. The plenaryassembly decided to defer a decision onh.d.tv. It simply issued a report. No 801.giving details of the proposed 1125/60/2:1standard and adding a footnote expressingreservations. Even before the Dubrovnik


meeting there had been discussions betweenlike-minded European broadcasters, equip-ment makers and government officials onthe possibility of establishing an alternativeh.d.tv standard. The effect of the plenaryassembly was to turn these thoughts intoaction.

Immediately afterwards, four Europeancompanies with interests in televisionbroadcasting equipment and receivers gottogether to formulate a definite proposal foran alternative standard and the developmentof equipment to demonstrate it. These wereBosch (West Germany), Philips (Nether-lands), Thomson (France) and Thorn -EMI(UK). To obtain EEC backing and financialsupport, the proposal was put to a June 1986conference of EEC ministers concerned withthe Eureka scheme, which was founded in1985 for co-operation on advanced technolo-gy (E&WW April 1986, p.7). It was acceptedby the Eureka ministers and given theproject number EU95. A first meeting of thenew h.d.tv consortium was held in October1986 and included government officialsfrom the principal countries concerned,France, Germany, Netherlands and UK. La-ter, Belgium, Italy, Sweden and Switzerlandjoined as secondary, or 'B' participants.

The extremely urgent research and de-velopment effort needed to catch up with theformidable NHK/Sony engineering achieve-ment in h.d.tv -a product of some ten years'work - is being supplied by eleven teams indifferent countries. These are covering fun-damentals and psycho -physics, productionstandards and conversions, studio equip-ment, transmission, MAC encoding/decoding, display standards, receivers, homevideo recorders and players, programmematerial, bit -rate reduction, and overall co-ordination.

CRITERIA FOR HIGH DEFINITION

A primary question that has to be consideredby all these teams is: what is high -definitiontelevision supposed to be? Or, more precise-ly, what performance do we need from h.d.tvto justify the effort and cost of developing it?Fortunately, all countries in the world con-cerned with h.d.tv, regardless of support fordifferent systems, seem to agree on this.Everyone recognizes that a large improve-ment in picture quality must be achieved tomake it worthwhile.

CCIR report No801 formally states that,relatively to existing tv systems, h.d.tvshould have twice the vertical and horizontalresolution (thus four times the pictureinformation), with separate luminance andcolour -difference signals and improved col-our rendition. Also it should have pictures ofwider aspect ratio. Here, everyone seems toagree that a 16:9 aspect ratio is desirable. Ata given viewing distance from the screen, awider picture than in conventional televi-sion sets makes the viewer feel more in-volved with the programme material, poss-ibly because there has to be more movementof the eyes to take in all the picture content.There seems to be general accord here that aviewing distance of three times the pictureheight (3H) is optimum, as against thetypical 6H of conventional viewing. Finally,the CCIR report specifies multi -channelhigh-fidelity sound.

In general the Eureka 95 project is aimingat a resolution of about 440 000 pixels in the16:9 format (compared with about 110 000pixels in a typical 4:3 PAUSECAM receiver).This would be given by 900 horizontal pixelsand 490 vertical pixels. Ideally the receiverdisplay should have a screen area of about0.8m2.

Another general problem already studiedby the CCIR is that of finding sufficientbandwidth for world-wide h.d.tv services inthe satellite broadcasting frequency alloca-tions at present available. These allocationsare at 12GHz, 23GHz, 42GHz and 85GHz.The 12GHz allocation is available for all ITUregions and is planned for analogue trans-missions in 24 or 27MHz channels. It allowsbasebands of a maximum of 12MHz. So withh.d.tv production standards requiring initialvision bandwidths in the region of 30 to60MHz (depending on the interlace factor,2:1 or 1:1), considerable bandwidth com-pression is necessary for transmission inthese r.f. channels.

More attractive is the greater bandwidthavailable at 22.5 - 23GHz. This 500MHz,however, is available for satellite broadcast-ing only in Regions 2 and 3, not Europe inRegion 1. So there would have to be a newinternational agreement to allow it to beused world-wide. Here, estimates have beenmade that a possible r.f. channel bandwidthof about 40MHz would allow two h.d.tvchannels per country and that signal base -bands could be about 20MHz, again necessi-tating some degree of bandwidth com-pression.

In the 40.5-42.5GHz and 84-86GHzallocations, rain attenuation is a seriousproblem and much more developmentwould be needed to make these two bandssuitable for h.d.tv. All in all, says the CCIR,for h.d.tv on a world-wide basis there is arequirement for an all -region BSS allocationof about 500MHz, preferably not above23GHz. The second part of the latest WARCdevoted to the geostationary satellite orbit.which was started in 1985 (see December1985 issue), is now being held in Geneva, butthis is not constituted to make any substan-

1250/50/2:1

Vtr or telec mi

picturesources

(interlaced)

Digitizationat 54MHz

sampling

frequency

Analoguevideo

1250/100/2:1

8- bidigital video

Two 54 MHzsampling

frequencies

/Two 2-Dtiters shcpespectra formoving and

static detail

H d tv

display D -to -a Field up-

(interlaced) converter converter

16 9

1250/50/2:1

Hdtvdisplay

(interlaced)16 9

1250/100/2 1(1013MHz)

D- to -aconverter

1250/50/2 1

Two 13.5MHz

sampling 1250/50/2!1 625/50/21frequencies quincunx pattern

/Two

re -samplers

at

13 5MHz

Mode

selection(static ormoving)

Selection ofmoving or

staticbranches

Line

shut fler

Sound data

LMAC

encoder

Digital assistance data

Digital assistance data

54 MHz

/Interpolationor static and

moving modesremoves repea

spectra

54 MHz

LouJ

de-shuffler

7129/50/2:1 625/50/21(13 SMH2 sampling

frequency)

PAL /SECAM

iv set (4 3)MAC

decoder

Compositesignal

MAC

demultiplexer

rMAC

multiplexer

Ldemul* :Jexer

Sound data

To d.tIssatellite

uplinknotional)

Fromdbssatellite(unlink

notional)

From d bs satellite downlink(notional)

Fig.1. Simplified schematic of Eureka EU95 compatible h.d.tv encoding/decoding demonstration chain, based on interim productionstandard of 1250/50/2:1. Only luminance digital processing is shown.

846 ELECTRONICS St WIRELESS WORLD

tial changes in the d.b.s. allocations. Howev-er, the feasibility of making the 22.5

band into a world-wide allocation islikely to be considered at this Geneva confer-ence.

EUREKA 95 PRODUCTION STANDARD

The equipment to be demonstrated atBrighton assumes compatible transmissionin the 12GHz world-wide d.b.s. allocation.An outline of the chain of hardware based onthe 1250/50/2:1 interim production stan-dard is shown in Fig.1. For simplicity onlythe luminance signal processing is shown.Broadly, this chain is built around twoestablished international broadcast en-gineering standards. One is the CCIR 4:2:2digital coding standard for 625 -line studioequipment, with sampling frequencies of13.5MHz for luminance (y) and 6.75MHz forthe two colour components (u and v). All thedigital signal processing is related to thesesampling frequencies and has eight -bit re-solution (256 quantization levels).

The second broadcast engineering stan-dard is the MAC/packet family of transmis-sion systems3 adopted for d.b.s. by the EBUand recognized by the CCIR. In the Eureka95 scheme it is being employed as a develop-ment from HD -MAC, which itself followedfrom the earlier extended- or E -MAC, orenhanced C -MAC proposal (see E&WW De-cember 1984 p.37, and February 1986, p.7).Essentially the Eureka 95 version providesthe extra vision information (pixels) neededfor the 16:9 h.d.tv pictures, while allowingconventional PAUSECAM receivers withMAC decoders to use only the informationappropriate for their 4:3 pictures in a com-patible fashion. Extra digital information toassist the receiver, called digital assistancedata, is transmitted as well. From the EBUMAC family, Eureka 95 will be compatiblewith both the D -MAC standard as intendedfor Britain's d.b.s.4, and the D2 -MAC ver-sion, intended for cable television systems.

In studying Fig.1, some readers may bepuzzled by the block marked '2-D filters'(two-dimensional filters). This term derivesfrom the concept that television pictures aredistributions of visual information in spaceand time. More precisely, the individualpixels in a continuing sequence of picturescan be located in relation to a three -dimension set of co-ordinates: horizontal,vertical and times

In this context, a conventional filter suchas a low-pass type, whether analogue ordigital, is only a one-dimensional filter. Ifconnected in a video circuit it simply mod-ifies the vision bandwidth and therefore, onthe displayed picture, alters only the hori-zontal resolution.

A 3-D filter, which requires field delays,shapes the bandwidth horizontally, vertical-ly and temporally, so that useful bandwidthis represented by a volume.

Thus a 2-D filter operates in any two ofthese three dimensions. Sampling is applic-able to all three dimensions. Indeed italready exists in one dimension in conven-tional television - the vertical sampling ofthe picture by the scanning lines.

In Fig.1 a picture source working on theinterim production standard generates ana-

1250/50/21 formatodd field

Yin fielo

Samples in sequence of 625/50/2:1 formatfour fields 11,2,3,41

1 3 1 31, 3, 1, 3

4 2

3 3 1

2,4, 2,42 4

1 3 1 3

2 4 2

3 1 3

Fig.2. Line shuffling technique combines samples as shown to convert 1250'50,2:1signal into 625 50 2:1 compatible signal. Numbers indicate both sample positionsand successive fields in which these samples occw.

64p s

625 lines

1

Digital sound and data164 packets of 751 bits

0 -MAC and 02 -MACdigital assistance data

Vertical blanking interval21 lines

Line 625 sync and)service ident. data

Vision signalmultiplexed analogue components

. for 16.9 and 4.3 pictures

Fig.3. D MAC format of Eureka 95 compatible h.d.tv signa I (not to scale).

logue video signals in component form. Thisworking standard was chosen because itssimple relationship to the 625 -line televisionstandard allows efficient and economicalconversion. At a later stage a sequentialsource will be introduced to give greatervertical resolution and better digital proces-sing performance. At the demonstration,Thomson will show a 1250/50/1:1 camera asan example.

Analogue luminance signals from thesource are digitized to eight -bit resolution ata sampling rate of 54MHz. This is four timesthe CCIR digital coding sampling rate of13.5MHz for luminance and has been chosenbecause of the 4:1 bandwidth compressionthat follows, thus retaining the CCIR codingstandard.

BANDWIDTH COMPRESSION

The next four blocks of Fig.1 are concernedwith bandwidth compression and conver-sion to a 625 -line signal format. Bandwidthcompression is based on the general princi-ple of first removing redundant visual in-formation from the picture and then spread-ing selected portions of the resulting videoinformation over an extended period of time.This follows the Hartley -Shannon theory onthe maximum communications capacity of achannel in bit/s: a trade between bandwidth,time and signal/noise ratio is possible.

Here the redundant information that isdiscarded consists of diagonal image fre-quencies and components of scene move-ment in which the eye is not sensitive to finedetail. Two 2-D filters perform this spectrumshaping task. Their mode of operation will bedescribed in more detail later with the aid of

Fig.4, which is an elaboration of this part ofFig.1.

The reduced -bandwidth signals from thefilters are then re -sampled at a lower fre-quency, namely the 13.5MHz CCIR digitalcoding frequency for luminance. What arebeing re -sampled here are the eight -bitrepresentations of the signals. Re -samplingat a lower rate, called sub -sampling, has theeffect of spreading the samples over a longerperiod of time, in this case over four fields. Itis effective for bandwidth compression withstationary pictures and static parts of pic-tures containing movement but introduces aloss of resolution on moving parts. So thetwo sub -sampled signals are next passed intoa section of the bandwidth -compressionsystem that exploits the eye's insensitivity tohigh spatial frequencies on moving parts ofthe picture. The purpose of this section is toselect dynamically from the two filteredinputs an optimum signal for transmission,depending on the nature of the scene con-tent. It produces a digital assistance signalwhich carries instructions for controlling adecoding system in the receiver (see later).

At this point the luminance and colourcomponents are still on the 1250/50/2:1standard, though in sampled and digitizedform. The samples of the sub -sampled signalare distributed over four fields as shown inthe middle part of Fig.2. Here the positionsof the samples are marked by numbers. Eachnumber (1,2,3,4) indicates the field, in atime sequence of four, in which that particu-lar sample has been placed by the sub -sampling process. It will be seen that thesamples are arranged in a quincunx pattern(like the arrangement of five dots on one face


wpw

Re -samplerat 13 5 MHz

2-D filter(vertical -

temporal)

8 bit digitalvideo 54MHzsampling

frequency

Grp.,

SchTr lcto-

Video at 13.5MHz sampliig frequency(optimun for static areas)

Local reconstruction

2-D filter(spatial)

Re -samplerat 135 MHz

Errormeasure

Original digital signal

Errormeasure

Locol reconstruction

Selection ofsmallest

coding error

Video at 13.5 MHz sampling frequency(optimum for moving areas)

Combinerand

encoder

8 -bit digitalidea 13.5MHz

samplingfrequency

Digitalassistance data

Fig.4. Motion -adaptive bandwidth -compression algorithm used in 1250/50/2:1 encoding chain. This functional diagram is a moredetailed version of the third, fourth and fifth blocks of the encoding chain in Fig.1. Interpolators correspond to those in Fig.1.

of a gaming die), a result of the diagonalfiltering. The process is similar to that usedin the NHK Muse system.

Next, the 1250 -line signal is convertedinto a 625 -line format. This is done by aprocess called line shuffling: the principle isillustrated in Fig.2. Solid lines indicate onefield while the broken lines indicate the nextfield interlaced with it. As can be seen, in afour -field sequence the line shuffler re-arranges samples from two successive linesin each field of the 1250/50/2:1 picture intoone line of a 625/50/2:1 picture. Thus sam-ples from odd -numbered fields (1,3) aretransferred into one 625 field while samplesfrom even -numbered fields (2,4) go into thenext, interlaced, 625 field.

After this line shuffling, the resulting625/50/2:1 luminance signal shown in Fig.1passes into a MAC encoder as the v compo-nent, along with the u and v colour compo-nents (not shown), of a complete videosignal. The v. u and v components are timecompressed and arranged in time -divisionmultiplex, all by digital signal processing.Extra picture information is, of course,generated with the wider, 16:9 h.d.tv aspectratio relative to the standard 4:3 aspect ratio.In the high -definition MAC encoding equip-ment contributed by the IBA this isaccommodated by time compressing theluminance by a greater amount -a factor of2:1 (and colour components 4:1) instead ofthe 3:2 factor (colour components 2:1) usedin a conventional D -MAC encoder.

As a result of these higher compressionratios the image frequencies of the analoguecomponents multiplexed in the transmis-sion signal are increased correspondingly.Consequently the high -definition MAC sig-nal has a wider baseband width, of about10.5MHz, than that of the conventionalD -MAC signal, which is about 8.4MHz.However, this can still be accommodated inthe normal d.b.s. 27MHz r.f. channel. Pre -and de -emphasis techniques are being consi-

dered to improve the signal/noise perform-ance.

After the MAC encoding of the v. u and vvideo components they and the digital soundand data packets are time -division multi-plexed to produce the complete high -definition MAC waveform for modulatingthe transmitter. At this point the digitalassistance data signal mentioned above isalso included in the multiplex. This repre-sents a data rate of about 1.1Mbit/s and isinserted in the 21 lines of vertical blankingperiod for D -MAC and D2 -MAC compatibletransmissions (Fig.3).

The receiving/decoding chain in Fig.1 isalmost self-explanatory as it reverses theprocesses described above for the encoding/transmitting chain. Two alternative 1250 -line h.d.tv displays are available, one with50Hz field rate and the other with 100Hzfield rate. Reception of the compatible sig-nals on existing PAUSECAM television setsis indicated at the bottom right. Here, theMAC decoder, in addition to expanding thetime -compressed analogue vision compo-nents, also receives a data signal telling itwhat portion of the 16:9 picture to select toform the standard 4:3 PAUSECAM picture.

MOTION -ADAPTIVE ENCODING

Returning to the bandwidth -compressionpart of the system, the main functions of thisencoding algorithm are shown in moredetail in Fig.4. It is a motion -adaptivescheme producing digital assistance signalsfor transmission to the receiving system.6The incoming eight -bit digital luminancesignal with a sampling rate of 54MHz isapplied to two 2-D filters for bandwidthreduction as explained above. One functionsin the vertical and temporal dimensions andgives optimum performance for stationaryareas while the other functions in the hori-zontal and vertical dimensions, or spatialdomain, and gives optimum performance formoving areas. After re -sampling at 13.5MHz

there are two reduced -bandwidth signals.One gives good resolution on stationarypictures but poor resolution on movement,i.e. blurring. The other gives only moderateresolution on both stationary and movingpictures.

One or other of these two signals is

automatically selected for transmission -and this is the motion -adaptive part of thesystem. To decide which is the better at agiven moment, the total picture is dividedinto a large number of small blocks. Withineach block the system first reconstructs twodisplayable versions from the two re -sampled signals. This is done by the twointerpolators shown in the diagram. It thencompares these two local reconstructions ofthe block with the corresponding part of theincoming signal. This is done by the subtrac-tors shown, which give the accumulateddifference, or error, between the originaland the reconstruction. The two errors aremeasured; and the version of the blockwhich shows the smaller error, i.e. the betterrendition of the original picture, is selectedfor transmission.

In stationary areas this system uses fourfields of samples (Fig.2) to reconstruct animage of high spatial resolution but poortemporal resolution. In moving images ituses one field of samples to reconstruct eachimage, resulting in reduced spatial resolu-tion but good temporal resolution.

If the receiver is to display optimumpictures by reconstruction from theselected, more accurate, blocks, it must be'told' for each block which of the two 2-Dfiltering processes has been used in themotion -adaptive encoder. This is necessarybecause the receiver decoder must apply thereverse process to that of the selected 2-Dfilter in each case. Here the reverse processis interpolation, or post -filtering as it issometimes called (they are functionallyequivalent).

To tell the receiver which of the two

848. ELECTRONICS & WIRELESS WORLD

interpolators to apply to the incoming video,the encoder in Fig.4 sends a digital controlsignal. Called 'digital assistance' data, thiscontrol information is produced by thecoding -error selection block in Fig.4. Asshown in Fig.1, it is multiplexed as data inthe D -MAC signal and then de -multiplexedin the receiving/decoding chain.

This digital assistance technique, origi-nally investigated by the BBC ResearchDepartment and here adopted in the Philipsencoding/decoding algorithm, has two im-portant benefits. One is that all processingdecisions are taken at the encoding end.using the high quality, full bandwidth sig-nals available at the source, rather than atthe receiver where the signals are compress-ed in bandwidth and may be corrupted byinterference and a relatively poor signal/noise ratio. In this way the performance ofthe system is enhanced. The second benefitis that the highly complex and expensivesignal processing hardware is concentratedat the encoding/transmitting end, so thatthe receiver is kept relatively simple and itscost minimized.

MEASURING SCENE MOTION

A serious limitation of this motion -adaptivebandwidth compression system as it standsis an objectionable loss of resolution onmoving areas of the picture which the eye isable to follow - say a slow sweep of a person'shand. The Brighton demonstration stopsshort at this point and does not provides aremedy in the encoding/decoding alorithm.But the Eureka 95 development teams arenow addressing this problem and hope tohave overcome it by the time of the nextdemonstration in late 1989.

The method being adopted is to provide atthe transmitting end a refined method ofresponding to motion. This will measuremotion and supply information to the re-ceiver that will result in improved resolutionon moving areas.

Essentially the refined system will mea-sure speed and direction of movement in thepicture and will transmit this 'motion vec-tor' information to the receiver. Themeasurement will be made in each of thesmall blocks mentioned above and the resultwill be sent as data to the receiver via thedigital part of the Eureka 95 signal. In thereceiver the motion vectors will be used tocontrol digital processing functions whichwill enable moving areas to be reconstructedwith the same resolution as stationary areas.

Several techniques for television motionmeasurement have been investigated overthe years. One that looks promising at themoment is called phase correlation. Herethe correlation measured is between succes-sive images. A two-dimensional Fouriertransform is performed on each image, thecorresponding frequency components aremultiplied together and a reverse Fouriertransform is done on the resulting array.This gives a set of numbers called a 'correla-tion surface', which has a peak at thecoordinates corresponding to the shift be-tween the successive images. The correla-tion surface shows by its shape the directionand velocity of the movement responsible forthe shift.

X velocity(pixels/field period)

Fig.5 Correlation surface computed fromsuccessive mages. as one way of derivingmotion vectors (BBC Research Depart-ment). This example results from a camerapan, where the amount of displacement ofthe main correlation peak from the originmeasures the velocity, here mainly in ahorizontal direction.

111

Useful screen

Fig.6. Screen dimensions of a new SonyTrinitron cc lour tube providing a display of16:9 aspect ratio. Outside diagonal of thetube is 1048mm.

Figure 5 is an example, a correlationsurface calculated from successive imagesobtained by panning a camera across ascene. Here, the large peak showing highcorrelation occurs at a certain horizontal (X)velocity, while there is virtually no vertical(Y) velocity. If there had been no movementat all (e.g. before panning) the large correla-tion peak would have been stationed in themiddle of the surface at the origin, indicat-ing zero horizontal and vertical velocities. InFig.5 the amount by which the peak isdisplaced from the origin is a measure of thevelocity of the pan.

Because only phase information is used,the process is not confused by brightnesschanges and also has good noise immunity.It is therefore a robust method of measure-ment. But as described so far it is onlycapable of measuring overall motion. Realtelevision picture sequences, of course, con-tain a variety of different movements occur-ring at the same time.

The BBC Research Department has in-vestigated by computer simulation an exten-sion of the basic phase correlation principlethat would be able to measure the motion ofmany separate objects in a scene7. In this thepicture is divided into blocks of, say, 64pixels square. Phase correlation is per-formed between corresponding blocks insuccessive pictures. This gives a number ofcorrelation surfaces describing the move-ment present in different areas of the pic-ture.

Each correlation surface is then searched

to locate several dominant peaks - not justone - resulting from the motion of objectswithin each block. So several motion vectorsare measured by each correlation process. Afurther stage in the method is to assign theprincipal motion vectors to particular areasof the picture.

Experiments have shown that the phasecorrelation technique does work well. TheBBC says that for simple panning and onemoving object, typical vector measurementaccuracies of 0.05 pixel can be obtained, and0.015 pixel by use of a windowing process.They also say that in most cases it is

impossible to detect three independentlymoving objects. Hardware to implement thetechnique is being constructed.

WIDE-SCREEN DISPLAY

Whatever h.d.tv standard is eventuallychosen, the spread of any new broadcastingservices based upon it will depend on theavailability of receivers with large, widescreens. Most demonstrations so far haveused projection displays but wide-screencathode-ray tubes are beginning to appear aswell. To judge from a recently introducedSony Trinitron tube for h.d.tv, it would seemthat these can provide the required aspectratio of 16:9 but not yet the desirable largescreen area of about 0.8m2.

Figure 6 shows the 16:9 screen dimen-sions of this new tube. From these figuresone can calculate that the useful screen areais about 0.35m2. Deflection angle is 90° andthe overall length, following from this, is798mm, while the neck diameter is 36.5mm.An anode voltage of 32kV is required. TheTrinitron, of course, has a screen formed bytrios of vertical phosphor stripes, with acorresponding aperture grille behind it.Across the 16:9 screen format there are 1861phosphor -stripe R.C.11 trios, while the aper-ture grille has a pitch of 0.45mm. Themakers claim a horizontal resolution ofmore than 1000. lines and a screen lumi-nance of 95cd/m2. A display monitor con-structed with this tube operates on 1125/60/2:1 and has a video bandwidth of 30MHz.

Wide-screen tubes of this kind mustobviously be bulky and heavy (here 05kg).As such they provide a strong incentive tofurther development of flat -screen colourdisplays for future h.d.tv domestic receivers.Work is continuing in this field, for exampleat Philips Research Laboratories.

OUTLOOK FOR EUREKA 95

The success of the Eureka 95 system insupporting a possible world h.d.tv standardwill depend mainly on whether it achieves asatisfactory compromise between the twinrequirements of compatibility and highquality pictures. A compromise cannot beavoided. In practice there is no such thing asperfect compatibility with no impairmentswhatsoever to the existing broadcast pic-tures on standard receivers. And in picturequality, with the practical limitation im-posed by bandwidth on channel capacity,perfect definition, equalling the resolutionof the human eye, is unobtainable. Tworesearchers examining the psycho -physicalbases of h.d.tv have concluded8 that if a


television system were to be "transparent" ata viewing distance of three times the pictureheight it would have to operate on a standardof at least 2270/80/1:1. This is obviously notpossible with existing bandwidth con-straints.

Furthermore, in all practical, bandwidth -limited systems the two requirements ofgood compatibility and high quality picturesconflict with each other. You can improveone, but only at the expense of the other.

There are several other requirementswhich a h.d.tv system has to meet. It mustallow satisfactory recording on video tape:simple transfer from tape to film and viceversa: the convenient exchange of high -definition programmes: easy conversion be-tween the h.d.tv production standard andexisting production standards: and straight-forward transcoding from the h.d.tv produc-tion standard to the h.d.tv transmissionstandard (see The h.d.tv studio, E&WW July1988, p.710).

At the present stage of Eureka 95 develop-ment there is a serious problem in obtaininggood resolution on movement, as explainedabove, and this will have to be overcomebefore the system can be properly assessed asa whole. In the meantime, a belated recogni-tion in the USA that compatibility is. afterall, important has given rise to several moreh.d.tv system proposals. Some of these havealready been outlined in E&WW (see March1988 issue. pp.226-229 and 294). Includingthe NHK/Sony system already mentioned.no fewer than eight h.d.tv proposals arecurrently being reviewed in the USA alone.Clearly the Eureka project is up against a lotof competition. But the demonstration atBrighton should at least give a better idea ofwhat chances it has of becoming interna-tionally accepted as a basis for a worldstandard.

Postscript. Just before going to press. in-formation received from the Eureka HDTVDirectorate at Eindhoven suggested that thedemonstration equipment at Brightonmight differ in a few details from that shownin Fig.1 . However, the interim productionstandard and the principles. parameters andtechniques to be demonstrated will remainas described.

In Britain, the DTI has now increased itsfinancial support to UK participants by £1.7million to make a total of 14.8 million. This£1.7M will go partly to Quantel Ltd, todevelop a range of h.d.tv editing and imageprocessing equipment, and partly to PhilipsResearch Laboratories for research into pic-ture analysis and coding techniques. The 1988 International BroadcastingConvention will take place at Brighton.23-27 September. Further information ab-out the technical sessions, the exhibitionand other aspects of the event can beobtained from the IBC Secretariat. Institu-tion of Electrical Engineers, Savoy Place,London WC2R OBL. telephone 01-240 1871.

The author thanks the MI Technical Cen-tre. the BBC Engineering Information De-partment and the IBA Engineering Secretar-iat for their valuable help in the preparationof this article.

Further reading

A useful. though not comprehensive survey ofEuropean work on h.d.tv is a collection of fivepapers under the heading EBU Studies on High -definition Television, reprinted from the EBUReview (Technical), No219, October 1986. MainlyAmerican h.d.tv work, with some European andJapanese. forms the entire contents of the IEEETrans. on Consumer Electronics. vol. 34. Nol.February 1988. Earlier papers on an internationalfront are to be found distributed in the Proceed-ings of the International Broadcasting Conven-tions for 1982. 1984 and 1986 IIEE conferencepublications).

1. R.W. Young. Eureka and production standardsfor li.d.tv. Image Technology. December 1987.527-528.

2. Y. Ninomiya et al.. A single -channel h.d.tvbroadcast system - the !%luse. NI1K LaboratoryNote No:10 '1. 1984.

3. Specifications of the systems of the MAC/packet

family. Ellt. document Tech 3258. 1986. amendedby document Silt i:18.4. M.D. Windram. G.J. Tonge. R.C. Hills. The1) -MAC packet transmission system for satellitebroadcasting in the t.nited Kingdom. ERE' Review(Technical). No227. February 1988.5. G.J. Tonge. Extended definition televisionthrough digital processing. Proc. Int. Broadcast-ing Convention. 18-21 September 1982.6. R. Storey. HDTV motion -adaptive bandwidthreduction using DATV. Proc. 11th Int. Broadcast-ing Convention. 19-2:1September 1986.7. GA Thomas. Television motion measurementfor MTV and other applications. BBC ResearchDepartment Report RI) 1987/11. Sept. 1987.8.1. Childs and R. Melwig. HDTV standard -settingon psycho -physical bases. Ent' Review (Technical). No219. October 1986.9. R. Hopkins. Advanced television systems. IEEETrans. on Consumer Electronics. vol. :It Nol.February 1988. Some of the systems describedhere were listed in the Television Broadcastcolumn. EW1111*July l N :,age 724.

Superconductive j-fetsIt is broadly assumed that many semicon-ductor devices are suitable for use atcryogenic temperatures. Indeed, many

semiconductors may work at cryogenictemperatures, but few actually offer thesensitivity available at room temperature.The junction field-effect transistor, howev-er, shows little shift in performance overchanges in temperature, and offers a uniquesolution for operation at cryogenic tempera-tures, writes Ed Oxner of Siliconix.

Most cryogenic instrumentation relies onindium -phosphide or germanium -galliuminfrared sensors to monitor activity. Thesesensors, like the common electret sensor,present an extremely high impedance. Aslong as the sensors/detectors operate at highimpedances, whether at cryogenic or atroom temperatures. j-fets should be the firstchoice.

The reason why j-fets outperform bipolartransistors when input impedances are highis simple. The term noise really defines apower phenomenon, represented as E2/R or12R. When j-fets are operated with a gatevoltage, the noise power is reduced byresistance (E2/R). However, for the bipolartransistor, which uses a base -drive current,the noise power is a multiple of this highresistance (12R). Hence, the j-fet easily out-performs the bipolar transistor when high -impedance sensors are involved.

Problems increase below the temperatureof liquid nitrogen (77K). At 40K, a phe-nomenon known as (silicon) 'carrier freeze -out' halts all action of the semiconductor,whether it is a bipolar transistor or a fet. Asthe critical superconductivity temperaturerises, transistor action returns. The extent towhich it returns depends on several factors,but the most crucial appears to be related tothe doping concentration.

Siliconix has responded to the currentinterest in the superconductivity, and isactively pursuing a low -noise development.Although already offering a commerciallyavailable j-fet, the Siliconix Si4338(gm= 750µS, Cgd = 1pF, typically) that offersoutstanding noise performance at 80K of25nV/VHz at 10Hz and of 104nV/VHz at

1Hz, the company says it is developing ahigher gain j-fet with even lower cryogenicnoise.

In addition to superconductivity, thereare other applicaitons for cryogenic fets, forNASA, for orbital astronomy, and for themilitary. All these rely on high -impedancesensors, so they must use fets to gain theadvantage of increased sensitivity and dyna-mic range. For example, radio astronomers,who seek and send signals far into the solarsystem, have historically pushed the tech-nology, seeking fets with a 1/f noise voltageof 20nV/VHz at 10Hz at 50K. Today, theyhope to halve that figure. As the noisevoltage is halved, the signal/noise perform-ance of the system is doubled. Therefore.focusing on astronomy, the limits arepushed out by offering double the sensitivityof former non-fet systems.

In electrical utility transmission lines, forexample, up to 15% of the generated electricpower is lost during transmission because ofresistive wire losses. In fact, the US -basedPacific Gas & Electric Company is reportedto lose more than $200M each year throughtransmission resistance. It is speculated,however, that when superconductivity be-comes fully practicable the web of electricallines will be replaced with buried tubularcryogenic dewars containing low -voltage,high -current superconductive transmissionlines engulfed in liquid nitrogen. Conduc-tion in zero -resistance superconductivetransmission lines, achieved with high cur-rents rather than high voltages, would alsosave energy by eliminating high -voltagecorona discharge. The initial costs of such asystem would be staggering. but these costsmust be weighed against the improved effi-ciency that would result in tremendousenergy savings.

Other commercial applications could alsosee revolutionary changes. Motors for heavymachinery that today run with large heatexchangers to limit entropy might be sub-mersed in liquid nitrogen to eliminate theheat exchangers. Power -station chimneysand acid rain might both disappear. Theelectric car might become a worthy reality.


i NOW IN UK!SINGLE SOURCEFOR ALL YOUR

ELECTRONIC CHEMICALS IIRequired reading for all service pro-

fessionals! Chemtronics latest cata-loguelogue is packed with over 200 topquality chemicals and cleaning products.ructs. Includes CO2 powered aerosol

of high -purity solvents, fluxIremovers, circuit refrigerants,precision dusters and conformalcoatings. Also features wipers,

Iapplicators, premoistenedpads/swabs, antistatic com-pounds, lubricants, adhe-

I.fives, desoldering braids andsolder. Complete with technicalspecifications and applications guide.

ICALL TODAY FOR FREE CATALOGUE AND SAMPLES.

Tel: (0322) 846886I Fax: 10322) 846549 Telex: 917667 CHEMUK G.

Chemtronics UK16 Swanscombe Business Centre,

London Road, Swanscombe, Kent. DA10 OLH

IIMIIM SIM IMMN MEMENTER II ON RE PT) ( N131)

111

02§164

QUALITYQUARTZ

CRYSTALSQUICKLY

Our frequency ranges are:r.A.P .U. Crystals

+A .P U Oscillators

1.

I I10kHz 50kHz 103kHz 500kHz 1MHz 100MHz 250MHz 360MHz

P of essional Crystals

We also supp.'y quartz crystal filters,oscillators of all types andcommunication antennae.

VV

Winter ElectronicsILMINSTER,SOMERSET TA19 9QA,

ENGLAND

TEL: (0460) 57166TELEX: 46571 FRQNCY GFAX: (0460) 57865

ENTER 26 ON REPLY ('Alt I)

COUNTERTIMERS

PFM200A £75.50 -f VAT

20Hz - 200MHz in 2 ranges, 4 gate times. 10mV sensitivity.8 -digit LED display. Battery/mains operation.

TF200 £175 + VAT

10Hz - 200MHz in 2 ranges. 5 gate times. Frequency. period.period average. totalize. 10mV sensitivity. 8 -digit 0.5" LCD.200 hour battery life.

TF600 £139.50 + VAT

5Hz - 600MHz in 3 ranges, 3 gate times. OmV sensitivity. 8 -digit0.5" LED display. Batteryimains: complete with adaptor/charger

TF1000 £450 + VAT

DC - 100MHz on both channels, 6 gate times. Frequency,period, period average, time interval, time interval average,frequency ratio and totalise. 20mV senVivity. HF filter,attenuator, trigger controls. tugger hold -off. 8 -digit 0.6" LEDdisplay. Mains operation.

TF1100 £550 4- VAT

As for TF1000 but with 70MHz to 1GHz orescaler (frequencyonly). 10mV sensitivity

TP600 £55 + VAT

-.- 10 prescaler. 40MHz - 600MHz 10mV sensitivity.

TP1000 £75 -t- VAT

10 prescaler, 100MHz - 1GHz. 10mV sensitivity.

thandarELECTRONICS LIMITED

Thandar Electronics Limited.London Road. St. Ives. Huntingdon.Cambridgeshire PE17 4HJ. EnglandTelephone (0480) 64646 Telex 32250 Test

THE LOGICAL CHOICE

ELECTRONICS & WIRELESS WORLD

It ON REPLY ('.X1{11

851

IEEE488-to-Z80interface

A dedicated IEEE488 controller and bus buffers greatlysimplify the interfacing of instruments to the SC84 and other

Z80 -based computers.

John Adams presented his SC84 compu-ter design in E&WW just over four yearsago. In the computer world, this is a

long time and most applications for whichthe SC84 would have been considered idealthen are now almost universally im-plemented using PC -compatibles. But SC84still has advantages over the more modernapproach: in particular, it is adaptable, andfacilities can easily be added on plug-in cardsdesigned for a specific purpose, such as thisIEEE488 interface.

Since the design was first published, theSC84 Z80 c.p.u. has been upgraded to themore powerful 64180. which is capable ofrunning Z80 software. This article describeshardware, and outlines software require-ments. for an IEEE488 interface based onthe original Z80 SC84. It was produced forthe satellite -instrument testing facility atOxford which uses SC84s to perform re-latively low-level control and data collectionto reduce the processor load on a PDP11/23,which has overall control of our experi-ments. Some modifications may be requiredto adapt the interface for use with the 64180.

Only one modification to the existing setof three cards is required so the interfaceboard is very easy to install. I wire -wrappedthe prototypes, but construction is nowmuch simpler since a p.c.b. is available(details later). For our purposes theIEEE488 socket was bolted to the end of theEurocard away from the backplane connec-tor, obviating the need for a flying leadconnecting to a header on the card.

IEEE BACKGROUND

Originally, the IEEE488 interfacing stan-dard came from Hewlett-Packard, which iswhy it is often referred to as the HPIB(Hewlett-Packard interface bus): it is alsoknown as CPIB (general-purpose interfacebus). Data passes through the interface inbit -parallel, byte -serial form to and from upto thirty slave devices, each with a uniqueaddress. One cable connects all slave devicesto the controller. As a parallel printer inter-face, IEEE488 is usually very fast. It is

becoming increasingly common to findIEEE488 interfaces on equipment includingoscilloscopes and stepper -motor controllers.

The bus consists of eight data lines(1)101,) and control/handshake lines (slot.ATM. EDI. OM. WED. NDAC. irc and KEN). Controlline DM ('data valid') is used by the sendingdevice to indicate that the data lines contain

T.J. VELLACOTT

a valid byte. Signals MUD ('not ready for data')and No.w ('not data accepted') are activated bythe devices for which the byte is intendeduntil the data has been accepted and thedevices are ready to receive the next byte.

Pull -down open collectors drive the inter-face lines so the logic is active -low. Also, aline remains low until all the collectorsconnected to it turn off. For this reason. thespeed of transfer is determined by theslowest listening device.

Bytes sent on the data lines are eithercommands or data. Commands are used bythe controller to order the source anddestinationls) of transfer. or (quasi -asynchronously) to investigate the status ofthe devices on the bus.

At any moment, the controller may desig-nate a particular device (including itself) tobe a 'talker', and others as 'listeners', bysending the address of the device in the lowfive bits along with the appropriate com-mand in the high three. Under this condi-tion, all those addressed to listen receivedata from the talker, and send handshakingafter each byte. Similarly the controller maypass control of the whole bus to anotherdevice, but the master controller can regainit on demand.

With five bits for address, why are thereonly thirty-one possible addresses, ratherthan thirty-two? The reason is that address31 (decimal) is used for universal negation -for example. telling address 31 to listen hasthe effect of making everything ignore thedata (nothing can ignore commands whileits interface is active).

If more than 31 devices are required, thereare two choices. One is to use anothercomplete interface (not so silly as it sounds)and the other, which involves a lot morework. and is probably only possible if build-ing a complete system from scratch, is to usesecondary addressing. which effectively di-vides one device into different sections.

Remaining control lines are used as fol-lows. Attention line ATN is activated by thecontroller to indicate that the data lines(No) _s) contain a command. rather thandata. Service -request signal silo is activatedby a slave device to indicate to the controllerthat it has a problem. In a normal response,the controller asks each device in turn (serialpolling) or in groups of eight (parallelpolling) which one initiated the servicerequest. End -or -identify signal Eui is usuallyused to indicate the last byte of a block

transfer: it is optionally sent with the lastbyte and can be used to simplify and speed upa command system.

Interface -clear command IFe is held activeby the controller for a specific period. usual-ly only when starting up the bus. Interfacesof slave devices on the bus are activated bythe HEN command. Many fuller descriptionsof the IEEE488 interface standard can befound in previous articles in E&Wtt' andelsewhere.

DESIGN

At the centre of the design is a TMS9914A.which connects to the 75160 and 75161buffers designed specially for IEEE488 inter-facing. Almost all of the IEEE488 interfacingis performed by the 9914 alone, and it hasthe particular advantage of being usable as acontroller: other i.cs do not have this capa-bility.

Connection of the 9914 to the Z80 bus isdone by the remaining three i.cs through theSC84 backplane, buffering the bus betweenthe 9914 and Z80 and decoding the link -

selectable port -address range of the 9914.For interrupts, the 9914 has an open -collector line which in the SC84 connects toline 1$ of the serial timer and interrupt i.c.(STI) using a spare line on the SC84 back -plane on the STI card. On other computersthis should go to a spare interrupt line. Theopen -collector output is converted to avoltage level with a resistor to the positivesupply rail.

Line Iti is unused in the basic SC84, and isone of the eight lines on the STI that may beprogrammed either as a parallel i/o port or asa general-purpose interrupt input. Here I, isused as an interrupt line, and the STIregisters are reprogrammed to allow inter-rupts from this source.

Address decoding is straightforward.There are eight registers in the 9914, therequired one of which is selected by threelines. Texas Instruments' numberingsystem for buses is somewhat confusing.since the least -significant bit for theregister -select pins is numbered zero, whilethat for the data bus pins is numbered seven.

Since lines connect directly to thelowest three bits of the address bus,respectively, the eight registers are mappedto eight consecutive ports in memory.Address lines .5_3 are sent to a 74LS138one -of -eight decoder, the highest two. .,7.going to active low mum.: inputs. and the


.5V

ov

2c

, 140A3 m= i 1 A0 Vcc

A2

6 lc674 LS138'

8GND

10clozal

11W1 cMi2

RD 11;+

Nim

A2I

12a

13a

14 1

2

- Group selectlink

9

10

5

or interrupt line(see text)

3a I

DO

07

Z80 bus/SC84edge connector

23c

16cRESELT.

4k7

Ma=

Vcc

1C3

74 LS 245

GNO

13 14

3IC 7

7474

1

ACC Ra CC

ACC GR TR

CE 0101

WE

DBIN

RS0

- RS2

INT

D7 D108

IC 4

TMS9914A

remaining three. \5_3 being used to selectone of the eight output lines used to enablethe 9914.

Which output line to use is chosen fairlyarbitrarily. Number 1 is used here, so theregisters are mapped to ports 10-171h. Bychanging the order of the inputs to theLS138 to any eight unused consecutiveports, the range is selectable.

An often overlooked characteristic of theZ80 is that the line not only signifies aninput/output request, as its name implies.but also an interrupt acknowledgment.When i-04 is active (low), therefore, it doesnot necessarily imply that i/o is taking place.Whether i/o or an interrupt acknowledge isrequired is determined by the state of suwhich, when active (low), indicates that thec.p.u. is in the first machine cycle of aninstruction (normally the op -code fetchperiod).

Input/output requests cannot occur dur-ing Mi. so MI being active concurrently withWW) indicates interrupt acknowledgement.while MI being inactive (high) designates ani/o request. This is why mi is connected to theactive -high enable input of the 74LS138. Ofcourse, it is not a problem for computerswhich do not use interrupts.

Output of the 74LS138. indicating thatthe address bus holds an address in the rangeof the 9914 registers and that mi is notactive, is gated with by one gate of the74LS02 quadruple two -input Nor package.This is inverted by a subsequent gate to give

- D0

0

RES

20GND

CONT

SRO.

ATN

E01

DAV

N RFD

NDAC

I FC

REN

TE

40

.5V

1:5 7516011- GNO

_ 010i

31 \_20

2

21

22

D 09

CC

10

DC GND

sRaATN

EOI

DAV

NRFL

NDAC

I FC

REN

Vcc

ICA 75162

11

OSee text

Z80 to IEEE488 interfacing is simplified bythe 9914 dedicated controller. Apart froma little control -signal gating. the remainingi.cs simply buffer the various buses.

the correct (active low) sense for the chip -enable input (pin 3) of the 9914, and thissignal is also gated with the line from theZ80 to give an active -high IMIN signal (pin 5of the 9914) to indicate that the Z80 wantsdata from the chip.

The active -high IMIN signal is also sent tothe 74LS245 data buffer to indicate that itshould send data from the 9914 to the Z80.Care must be taken not to allow data from aperipheral onto the data bus spuriously, asthe consequences are generally more dis-astrous (and less difficult to trace) than aspurious write to a peripheral.

Design of the data -bus buffer is simple. Itschip -enable input is held permanently active(low) so the 9914 data inputs usually followthose of the Z80. Data is ignored by the 9914until imiN is high and a are low: the decodingensures that there are no spurious writes tothe chip. When the 9914 address is selectedand the W) line from the Z80 is active (low).the direction of data transfer in the buffer isreversed and output from the 9914 is con-nected to the bus.

Unused gates of the 74LS02 are tied to+3V through a resistor. As mentioned ear-lier the 75160 and 75161 buffer output of the9914. with the control lines using the 75161.

EEE 488connector

1

D101

0111

13

16 0108

10

11

5

9

17

SRO.

ATN

E01

DAV

NRED

NDAC

1FC

REN

12Shield

24Logic GND

18-23GND

and the data lines using the 75160. Input 1E.pin 1 of the both the buffers ttalk enable),determines the sense of the flow through thebuffers. A fuller implementation of theIEEE488 standard would require a 75162instead of the 75161 to allow an overallinterface controller to take precedence overtemporarily -assigned controllers, but therearc relatively few applications where this isnecessary.

SOFTWARE

Since the IEEE488 interface is so versatile,and its applications so diverse. it is difficultto give much indication of what software todrive it entails. Consequently I have notincluded examples. Reference to the TexasInstruments data book on the 9914 is essen-tial.

Under SciDOS and CP/M operating sys-tems it is relatively easy to set up the bus forread/write transfers using the often other-wise redundant PUN: and RDR: logical de-vices. Where these devices have been allo-cated, for RS232 input and output ports.accessing the IEEE488 interface from sepa-rate programs is more tedious. It is usuallybetter to have one block of general-purposecode used in different ways than to rewritesource code for each application.

Further complications arise when adv-anced features such as service requests andsecondary addressing are used, but withrelatively little effort the necessary informa-tion can be extracted from the Texas lnstru-


ments book. For operation of the interfacefrom SC84, the STI data sheet will also benecessary to find out how to interrupt themain processor.

Software sets the IEEE488 address bus bywriting to one of the registers in the 9914,which is why I have not included the cus-tomary dip switches. For our purpose thiswas more appropriate since the variousSC84s could have their software exchangedto change their functions. As the address wasspecified in software, it moved with thefunction of the computer, not with thehardware.

TESTING

The easiest way to test the interface is toconnect a printer with an IEEE488 interfaceto it. Epson provides plug-in cards for someof its models, which allow the printer tobehave both as an addressed device (printingonly when told to listen by the controller)and as a bus monitor. when it prints (inhexadecimal form) the contents of data linesna)1_8 as each byte is transferred (slowingthe bus down to do so). Each command sentis also printed in bold type, which is veryuseful.

Connecting other devices to the bus isthen quite straightforward, but some inter-faces do not seem to behave as expected, so itis as well to have tried your software with adevice that is known to work and is under-stood.

Further readingAdams. 1.11.. SC84 Microcomputer. Electronics

September andOctober 1984.TMS9914A General Purpose Interface Bus (GPIB)Controller. Texas Instruments. Manton Lane,Bedford MK41 7PA.

Timothy Vellacott is 23 and in the secondyear of a Doctor of Philosophy course in theDepartment of Atmospheric, Oceanic andPlanetary Physics at Oxford. He is involvedwith building part of the PMIRR satelliteinstrument to be flown on NASA's MarsObserver mission in 1992.

Timothy's 1EEE488 interface was de-signed in 1985 when he was a vacationstudent in the Department. The satelliteinstrument mentioned in the article is!SAMS, a stratospheric and mesosphericsounding pressure -modulator radiometerbeing built by Oxford. Rutherford AppletonLaboratory and British Aerospace. and to beflown on the Upper Atmospheric ResearchSatellite in the early 1990's. ISAMS willmeasure temperature and concentrations ofgases involved particularly in the chemistryof ozone on a global scale.

Eurocard p.c.bs for this design are availablefrom Combe Martin Electronics. KingStreet. Combe Martin. North Devon for £9fully inclusive (UK or overseas). The boardsare single sided, silk screened and compati-ble with SC84.s Z80 backplane.

Conferences andexhibitions

5-9 September, oxford: Science 68. 150th annualmeeting of the British Association for theAdvancement of Science. Special events includeexhibitions in the University and at OxfordPolytechnic, concurrent conferences organizedby other learned societies, a scientific film festival.a day of debates on problems now confrontingscience, a family science day - and a re-enactmentby BBC Television of the celebrated confrontationbetween Bishop Wilberforce and T.H. Huxley overevolution and the origin of Man. which took placein Oxford at the BA meeting of 1860. Among thehighlights of this year's conference are a day'ssymposium on molecular electronics, a session oneducating the engineers of tomorrow, and alecture by Dr David Jones ("Daedalus" of the NewScientist), who promises to demonstrate to hisaudience some intriguing physical phenomena. Aprogramme is available from the BA in London.01-734 6010 ext 381.

6-8 September, Wembley, London: Coil WindingInternational exhibition and conference. Detailsfrom Bush Steadman and Partners. 0799-26699.

6-11 September, New Delhi: Electronics India88. Details from the general manager. Trade FairAuthority of India. Pragati Maidan, New Delhi.India: telex (India) 031-61022 COMX IN.

9-11 September, Eastbourne: Tecdoc'88, tech-nical documentation and communication. organized by the Society of Electronic and RadioTechnicians. One of this year's themes is translat-ing documents into other languages - 1992 is notfar off. Details from Consert, 57-61 NewingtonCauseway, London SEI 6BL, tel. 01-403 2351.

12-15 September, University of York: IERE con-ference on electromagnetic compatibilityle.m.c.). The conference will be preceded by atutorial day and accompanied by an exhibition.Details from the conference secretariat on 01-2401871.

13-15 September, Novotel. London W6: PowerUK 88 and Enclosures 88 (note change of dates).Conference topics include battery technology.r.f.i./e.m.i. shielding, smart power, connectiondesign for the enclosure industry. Contact: SCSExhibitions Ltd at Portsmouth on 0705-665133.

13-16 September, Alexandra Palace. London:EPoS/EFTPoS 88 - retail automation show andconference. including technical sessions. Detailsfrom RMDP Ltd in Brighton. 0273-203581/3.

14-17 September. Washington D.C.: NationalAssociation of Broadcasters' Radio '88. Detailsfrom NAB at 1771 N Street. N.W.. WashingtonD.C. 20036. tel. (202) 429-5420.

19 September, London: one -day conference ondigital cordless telephones ICT-2. telepoints etc.).Fee: £245 +v.a.t. IBC Technical Services. 01-2364080.

20-24 September, Besancon. France: interna-tional exhibition of microtechniques, Micronora88, with the theme cad/cam stamping andmachining. Details from French Trade Exhibi-tions. London. 01-225 5557.

21-25 September, Olympia. London: PersonalComputer World Show. Details from MontbuildLtd. 01-486 1951.

23-27 September, Brighton: InternationalBroadcasting Convention I IBC88). with 100 tech-nical papers covering d.b.s. and cable systems,recording. studio and o.b. equipment. measure-ment techniques and new services. and an exhibi-tion supported by 200 companies from Britain andabroad. Details from IEE Conference Services on01-240 1871 ext.222.

27-30 September, Birmingham NEC: Test +Transducer '88 exhibition and conference. Elec-tronics in Engineering Design and Design En-gineering shows. Details from Trident Interna-tional Exhibitions Ltd in Tavistock. 0822-614671.

28 September, London: one -day conference.Liberalisation of European telecommunications -preparing for 1992. IBC Technical Services. 01-236 4080.

3-7 October, Cascais. Portugal: internationalconference on the theme New Directions for Unix.with tutorials and showcase. Organized by theEuropean UNIX systems User Group, telephone( Royston. Hertfordshire) 0763 73039.

5-11 October, Cologne: photokina - world's fairof imaging systems. KolnMesse. Messeplatz I.Postfach 210760. D-5000 Kiln 21 (Deutz): telefax(0221) 821 2574.

4-6 October, San Diego. California: internationaldisplay research conference, organized by IEEEElectron Devices Society and others. Details fromthe Advisory Group on Electron Devices. 201Varick Street. suite 1140. New York. NY 10014.

17-19 October, IEE. Savoy Place. London: fourthinternational conference on satellite systems formobile communications and navigation. Detailsfrom IEE Conference Services. 01-240 1871 ext.222.

18-20 October, Brighton: Southern ElectronicsShow. Details from Trident International Exhibi-tions Ltd in Tavistock. 0822-614671.

24-28 October, Monte Carlo: international sym-posium on automation, concentrating on cellcontrol and quality management systems for themanufacturing industries: accompanying tradeexhibition. Details from ISATA in Croydon. 01-681 3069 or 01-686 1329.

26-27 October, Wembley Exhibition Hall. Lon-don: Instrumentation Wembley. Details from Tri-dent International Exhibitions Ltd in Tavistock.0822-614671.

1-3 November, Barbican. London: ITEX - In-formation Technology Exchange. New eventwhich will display the latest commercial applica-tions of i.t. research and development, linkingAlvey. Esprit and other research programmes.Supported by DTI. Cahners Exhibitions. 01-8915051.

3-6 November, Los Angeles. The DTI is againproviding support for British companies wishingto exhibit at AES: details from the Sound andCommunications Industries Federation inSlough. tel. 06286-67633.

14-18 November, Paris: International Exhibitionfor Electronic Equipment and Products, PRONIC.Details from French Trade Exhibitions, 01-2255566.


The Hotels

ELECTRONICA-MUNICH`1°

8-12 NOVEMBER 1988

WIN A WEEKEND IN MUNICHElectronics & Wireless World in association with Electronics Weekly is offering everybodywho books one of the business trips listed below to Electronica, a chance to win a TRIP FOR

TWO to MUNICH for the weekend following the show.The leading electronics newspaper and technical monthly, in associaticn with Commercial

Trade Travel, will fly your partner over on Friday morning via Gatwick, giving you a fullweekend in Munich. We will then fly you and your partner home on Sunday.

The draw will take place on Friday the 21st October and the winner announced in theNovember issue of Electronics & Wireless World and the 26 OctoberElectronics Weekly.

Good Luck

The ToursMUNICH SHERATON HOTELThis de -luxe hotel is situated in the fashionable Bogenhausen district ofMunich, close to the English Gardens and 10 minutes from the airportand downtown area All rooms have private bath/shower, WC, hairdryers, direct dial telephone. TV. radio and video service and =thatThe hotel has fashionable bars. restaurants. and night club as well as alarge swimming pool. sauna and fitness centre Opposite the hotel thereis a pedestrian shopping centre with several restaurants. bars and acinema In the Autumn of 1988 is proposed that the U-bahn(underground train) will be opened linking the hotel downtowr areaand exhibition grounds

REGENT HOTELThis brand new first class hotel is situated in the city centre, close to themain railway station and exhibition grounds It is possible to walk to thefair grounds in 20 minutes or on the U-Bahn (underground train) it is lust2 stops The hotel has a fine reputation. all rooms have privatebath -shower. WV, radio and TV. =mbar and direct dial telephone Thehotel has a fashionable bar and elegant restaurant. It also has a sauna.solarium, whirlpool and coiffure.

CONDOR HOTELThis 5 -star hotel is situated in the City Centre, close to the main railwaystation and within 15 minutes walk, or lust 2 stops on the U-Bahn(underground train) from the exhibition centre. All rooms have privatebath/shower, WC, direct dial telephone. radio and minibar. The hotelhas a breakfast room. bar and spacious lobby area.

MODERN HOTELThis 3 -star hotel is situated in the City Centre. close to the main railwaystation and within 15 minutes walk, or lust 2 stops on the U-Bahn(underground train) from the exhibition centre. All rooms have privatebath/shower. WC, radio and TV, minibar and direct dial telephore. Thehotel has a breakfast room and bar

INCLUDED IN THE ABOVE PRICEScheduled flight services of either AIR EUROPE or LUFTHANSA, in flight catering as appropriate to the time of day travelling. all airport taxes,accommodation on a bed and breakfast basis inclusive of service charge and taxes: private coach transfers airport to hotel and return, a cocktail partyat the Exhibition Grounds, season ticket entrance to ELECTRONICA including catalogue :the entrance to the Exhibition is £29 per person), assistance

upon arrival and departure in Munich of a Commercial Trade Travel representative

FLIGHT DEPARTURES FROMGATWICK

(Scheduled services of Air Europe)

FLIGHT DEPARTURES FROMHEATHROW

(scheduled services of Lufthansa)

TOUR A -3 days (2 nights) TOUR D -3 days (2 nights)Tuesday, 8 November Tuesday, 8 November

dep Gatwick 07 15 dep Heathrow 09 45

arr Munich 10 00 an Munich 12 20

Thursday. 10 November Thursday, 10 Novemberdep Munich 11 25 iep Munich 12 20

an Gatwick 18 00 arr Heathrow 15 20

TOUR B -3 days (2 nights': TOUR E -3 days (2 nights)Thursday, 10 November Thursday, 10 November

dep Gatwick 07 15 dep Heathrow 09.45

arr Munich 10 00 arr Munich 12.20

Saturday. 12 November Saturday. 12 Novemberdep Munich I I 15 dep Munich 12 20

arr Gatwick 12 00 arr Heathrow 15 20

TOUR C -3 days (4 nights) TOUR F -5 days (4 nights)Tuesday. 6 November Tuesday, 8 November

dep Gatwick 07.15 dep Heathrow 09.45

arr Munich 10 00 arr Munich 12 20

Saturday. 12 November Saturday. 12 Novemberdep Munich 11 15 dep Munich 12 20

arr Gatwick 12.00 art Heathrow 15.20

The CostsINCLUSIVE PRICES PER PERSON(TWE - sharing twin bedded room with facilities)(SWE- single room with facilities)

HOTEL MODERN CONDOR REGENT SHERATON

Accommodaticn TWE SWE TWE SWE TWE SWE TWE SWE

TOUR A and b 295 340 305 360 340 450 360 450

TOUR C 575 450 385 490 450 650 475 650

TOUR D and E 420 470 455 495 470 575 485 565

TOUR F 490 575 499 599 575 750 599 750

COMMERCIAL TRADE TRAVEL are the only company offering tours via GATWICK

To make your reservation please ring Ray Gardner on 01-491 1312 or fill in the coupon andsend to:

Ray Gardner, Commercial Trade Travel Ltd, Aspen House, 25 Dover Street, London W1X 3PA.

Name Name

Address Address

Tel. No: Tel. No:



INSIGHT EXTRA

PC BUS PERFORMANCENUBUS VS MICRO CHANNELT

\ vo new computers have beenintroduced recently whichuse entirely new bus struc-tures for add -in cards. Apple

Computer's Macintosh II uses a personalcomputer version of the NuBus. IBM's newPersonal System/2 family uses an improvedpersonal computer bus called the MicroChannel Architecture. This article presents acomparison of the two buses and the twocomputers, and includes a technical ex-amination of certain measured system char-acteristics using GPIB/IEEE488 boardswhich have been developed for both buses.

The Apple Macintosh II and the IBMPersonal System/2 computers are attractinga great deal of attention from add -in boardmanufacturers. The article covers some ofthe technical aspects of designing cards forthese machines.

The Macintosh II is a 32 -bit 68020 -basedcomputer which uses a personal computerversion of NuBus a general-purpose 32 -bitcomputer backplane bus conceived at MITand developed further by Texas Instruments.At present, the IEEE P1196 committee isformalizing the NuBus specification into aform suitable for acceptance and publica-tion.

The IBM PS/2 family currently consists ofsix computers, three of which use the newMicro Channel Architecture (MCA). The

Moco

za.

NMR

This analysis of two

new personal computer bus

structures includes

comparative measurements

based on plug-in

GPIB boards.

Micro Channel is a general-purpose compu-ter bus which supports 8. 16, or 32 bitmicroprocessors, peripherals, and memory.PS/2 Models 50 and 60 are equipped with16bit 80286s, and Models 70 & 80 use the32bit 80386. Models 50 and 60 support the16bit version of the MCA, and Models 70 &80 support both 16 and 32 -bit add -in cards.

Performance examplesThe performance specification listed in thepanel are all theoretical in nature. To deter-mine how each bus actually performs, Iexamined two add -in card examples anddiscuss the measurements obtained.

Table 1 gives a comparison of the twocomputers used: a Macintosh II computer

MasterInterlace

1

Sa

1

RTSI BUS

max

V

DATA BUS

DMAC

GPM DMA Requota

LL

SlaveInterface

ADDRESS BUS

DROControl4

clodsPnsocale,

Trigger BusInterface

to

Counter/Timer System

NMR

Turbo488

GPIBInterlace

Controller

GPIB Direct Access PortGPIB

Transcewers

GPIBTransceivers

V

TABLE 1. EXAMPLE COMPUTER SYSTEMS

NuBus Micro ChannelComputer Apple IBM PS 2

Macintosh II Model 80Microprocessor 68020 80386Clock speed 16 MHz 16 MHzCard slots 6 8Add -in card NB -DMA -8-G MC GPIBOperating syst Macintosh PC DOSWait state 100 ns 62.5 ns

and a PS/2 Model 80. The Macintosh II is runby a 16MHz 68020 microprocessor and theModel 80 by a 16MHz 80386. The d.m.a.controller in the PS/2 runs at 8MHz, whilethe Mac is equipped with a National Instru-ments NB -DMA -8-G board containing an82380 d.m.a. controller running at 10MHz.

The Macintosh II has six card slots, one ofwhich is used by a display adapter andanother which is occupied by the NB -DMA -8 -G card. The Model 80 has eight slots, withone taken up by a disc controller and asecond slot used by National Instrument'sMC-GPIB card.

The Macintosh II runs the Macintoshoperating system and the Model 80 runs PCDOS3.3. Slave cards on the Macintosh II canintroduce wait states of 100ns each, and onthe Model 80 each wait state adds an addi-tional 62.5ns to the transfer.

The NB -DMA -8-G, in addition to its 82380d.m.a.c. and bus master circuitry, has anIEEE488 interface which responds as aNuBus slave. The measurements describeduse the NB -DMA -8-G both as NuBus d.m.a.master and slave.

MC-GPIB is an lEEE488 interface for theMicro Channel. It can respond as a bus slaveand also contains arbitration circuitry toallow it to be used in conjunction with thed.m.a. controller on the mother board.

Slave performanceIn this example, I treat the NB -DMA -8-G andthe MC-GPIB cards as bus slaves and mea-sure some of the timing parameters whichare of interest and which affect i/o perform-ance.

A block diagram of the NB -DMA -8-C isshown alongside. For performance testing, Ihad the Macintosh II's 68020 execute asimple i/o write test. The test programs,shown on this page, transfer data to andfrom the 16x 16 bit, fifo memory in theTurbo 488 integrated circuit on the card.

The same test was performed on the PS/2Model 80 using the code shown. A block


INSIGHT EXTRA

diagram of the MC-GPIB card, right, identi-fies the Turbo 488 and circuitry that performslave functions which are quite similar tothose on the NB -DMA -8-G.

Table 2 gives a summary of the perform -

TABLE 2. SLAVE TIMING PERFORMANCE

Macintosh II PS/2 Model 80and NB -DMA -8-G and MCGPIB

Type 16 bit i o 16 -bit i oRead cycle (min) 403 ns 375 nsWrite cycle (min) 403 ns 375 ns

(byte s)Measured' o read 703K 533K

rate (byte's)Measured r'o write 703K 533K

rate bytc /s)Measured interrupt 35 ps 7.3 !, s

latency

ance obtained using the test programs andthe two different systems. The fifo read andwrite cycle times are theoretical best -casevalues. Note that the NB -DMA -8-G adds200ns to the fastest possible transfer cycle onthe NuBus. This extra delay is due to fifosynchronization arbitration and access time.Similarly, the MC-CPIB adds approximately125ns to the minimum Micro Channel cycle.These delays establish maximum theoreticaltransfer rates of 2.5 x 10616 -bit transfers persecond (5Mbyte/s) for the NB -DMA -8-G and2.67 x 106 16 -bit transfers per second(5.3Mbyte/s) for the MC-GPIB.

The actual programmed i/o rates obtainedusing the i/o read and write test programsare 0.7Mbyte/s for the NB -DMA -8-G and0.533Mbyte/s for the MC-GPIB. As you cansee, bus speed is of less importance in thisexample than microprocessor performance(execution time, program memory speed,on -chip cache, instruction set, etc).

One final slave timing characteristic wasmeasured for both systems: interrupt re-sponse time. For this test, the time betweenwhen the card generated the interrupt re-quest on the bus to when the interruptservice routine was entered when measured.

The 35µs interrupt latency for the Macin-tosh II is indicative of the interrupt servicequeue used by the Macintosh OperatingSystem rather than hardware performancefeatures of the 68020. The 7µs interruptlatency for the PS/2 Model 80 is a closeindicator of actual hardware performance,since no intervening interrupt service soft-ware overhead is incurred in PC DOS.

Bus master performanceTo measure bus master performance and, inparticular, d.m.a. performance, we use the82380 d.m.a.c. on the NB -DMA -8-C to trans-fer data to and from the Macintosh IImotherboard and the fifo on the card. Forthe Micro Channel, we use the arbitrationcircuitry on the MC-CPIB in conjunctionwith the PS/2 Model 80 d.m.a.c. to transferdata to and from the motherboard memoryand the fifo on the card. The results of thetest are shown in Table 3.

Burst mode fetch and deposit d.m.a.transfers were used on both machines.Transfers on the NuBus were 16 bits wide fori/o accesses (fifo reads and writes) and 32 bitswide for memory. (The 82380 automatically

The two boards and computers compared in these tests are National Instrument's MC-GPIBinterface for the IBM PS/2 (above and below), and the NB -DMA -8G combined d.m.a.controller and 488 interface (page 856).

U)

CO

-Jw

OCC

AddressLatches

DataTranscervers

BusInterlace

Logic

POS

Turbo4t8-0

AddressDecode

DMA andInterrupt

Log c

CPS.nterface

Controller

GPI8"franscervers--


INSIGHT EXTRA

TABLE 3. BUS MASTER PERFORMANCE

Macintosh II PS.2 Model 80

DMA transfer and NB.DMA-8-G and MC-GPIB

type burst dual burst dualData width 16/32 16i16Master latency 300ns 500nsI/O read 500 ns 750nsI/0 write 500 ns 750 nsMemory read 1.1 p s 375 nsMemory write Ups 375 nsTransfer rate 1.9M 1.8M

(byte, s)

performs source and destination hit widthadjustments. The master latency times showhow long it took from the time the cardrequested use of the bus until the time itreceived bus control.

The i/o read and write times indicate thetime taken to access the fifo by the d.m.a.controller. The memory read and writetimes show how long it took for the d.m.a.controller to access system motherboardram. Note that it should he possible tosubstantially increase the performance ofthe Macintosh II in this example by using ahigh-speed ram. PC -style NuBus add -incard.

The data transfer rates are given last. TheMacintosh II is able to overcome its relative-ly slow memory access time by transferring32 bits as opposed to 16 bits on PS/2.

ResultsBoth machines provide powerful platformsfor add -in i/o cards, and both have distinctadvantages and disadvantages, as well asunique characteristics. Some of the featuresthat are common to the Macintosh II and thePS/2 Model 80 are high speed. 32 -bit microp-rocessor. memory, and bus: automatic cardconfiguration at power on (no jumpers):powerful interrupt mechanism: and supportfor multiple add -in cards. Some of thedissimilar features are listed in Table 4.

TABLE 4. MACINTOSH II AND PS/2 MODEL 80FEATURE COMPARISON

Macintosh IINuBusindustry standard

connectors'one standard busmedium board areasynchronous busparallel bus arbitrationno integral d.m.a.refresh circuitry on cardsslow memory access

PS 2 Model 80Micro Channelcustom edge connectormultiple versions of bussmall board areaasynchronous busserial bus arbitrationintegral d.m.a.refresh circuitry on

fast memory access

I feel that the 96 -pin DIN connector usedby the PC -style NuBus cards offers advan-tages over the edge fingers defined by theMicro Channel Architecture. primarily interms of mechanical real iabi I i ty.

The multiple versions of the Micro Chan-nel make it more difficult to develop add -incards that can be used on all machines and atthe same time offer maximum performance.

For example, if you were developing a highperformance. intelligent peripheral proces-sor card for the Micro Channel, you may beforced into developing two versions: onewith a 16 -bit interface and one with 32 -bitinterface.

The small board area available on MicroChannel cards makes development moredifficult. In many cases. the developer will beforced into designs which require gate arraysor other asics and/or surface -mounted com-ponents. Both of these solutions lengthendevelopment time and, except in very highvolume cases, increase manufacturing costs.

At National Instruments, it is easier todesign for the NuBus because of its synchro-nous nature. The separate address and datalines on the Micro Channel do not turn outto be of advantage because timing require-ments force addresses to be latched much inthe same way that they are latched on NuBusimplementations.

Although I don't have quantitative, mea-sured data to support this next conclusion, itis obvious that a portion of the MicroChanne! bus bandwidth is utilized by thearbitration process, and that the percentageused increases as more devices contend forthe bus. Unlike the Micro Channel. NuBusarbitration occurs in parallel with normalhus traffic.

For i/o intensive applications involvinglarge blocks of data, I feel that it is essentialfor the computer system to provide d.m.a.transfer capability. The Macintosh II re-quires that this capability be provided by theadd -in cards, themselves. The PS/2 uses aclever scheme which allows cards to use ad.m.a.c. on the motherboard. Unfortunate-ly, the one on the PS/2 Model 80 is limited to16 -bit transfers. Perhaps future members ofthe PS/2 family will offer full 32 -bit support.

The Micro Channel provides refresh sup-port for dynamic ram. add -in cards. Thiseliminates the expense of refresh circuitryon each ram card. It does, however, utilizebus bandwidth (less than 5% on the PS/2Model 80). This characteristic is probablyparalleled on the NuBus, however. by in-creased access times during memory refreshand bus access collisions.

Accessing the Macintosh ll motherboardram from an add -in master NuBus card isagonizingly slow by today's standards. ThePS/2 family offers acceptable access to itsmotherboard memory. Perhaps future ver-sions of the Macintosh ll will correct thisdeficiency.

Nowlin is engineering vice-president withNational Instruments Corporation. Austin. Texas.represented in the UK hy Amplicon Electronics.

ComparativeoverviewThe table lists some of the primary featuresof the PC -style NuBus and the Micro Chan-nel bus. The NuBus is considered a synchro-nous bus. All bus transactions are referencedto a single. 10MHz clock signal. Transac-tions on the Micro Channel bus are asyn-chronous and are not referenced to a specificclock signal.

The 10MHz + 0.01% clock on the NuBushas an unequal duty cycle of 25:75. TheMicro Channel bus provides a 14.31818MHz+ 0.01% clock signal but does not use it forbus timing.

Micro Channel Architecture provides forseparate memory and i/o address spaceswhile NuBus provides for a single addressspace. I/O addresses on the Micro Channelare 16bits wide. providing 64K bytes of i/oaddress space.

Memory addresses on the NuBus are32 -bits wide and allow up to 4Gbytes of ramor rom to be individually addressed. The16 -bit version of the Micro Channel uses a24 -bit memory address, which translates to16Mbytes: the 32 -bit version provides a full32 -bit. 4Gbyte memory address space.

Data can be transferred on the NuBus asbytes. 16 -bit halfwords, or 32 -bit words. The16 -bit version of the Micro Channel cantransfer 8 or 16 -bit data. while the 32 -bitversion can handle 8, 16. 24 and 32 -bit data.Bit, byte. halfword, and word significance isthe same on both the NuBus and the MicroChannel.

On the NuBus, cards are given 1/16th ofthe total 4Gbyte address space 1256Mbyte).NuBus allows up to 16 cards in a system andeach card is given I/16th of the total 256Mcard address space (16Mbyte per card). Thecard automatically assumes its properaddress by sampling four signals on theNuBus connector whici, provide a cardidentification code.

Micro Channel cards use the Programm-able Option Select IPOS) feature to deter-mine i/o addresses. Add -in card addressinformation is stored in a non-volatile r.a.m.available to the microprocessor. At power -on -time, the microprocessor reads theaddress informationi from the n.v. r.a.m.and configures each card in the system.Using the POS mechanism, add -in cards mayoccupy any free portion of the availablememory or i/o address space.

Data transfers. A minimum single datatransfer on the NuBus consists of two clockcycles (a start cycle and an acknowledgecycle, of 200ns, which translates to a20Mbyte/s transfer rate see Tables. TheMicro Channel minimum cycle is 200ns.which, for 32 -bit data transfers, translates to20Mbyte/s. The 10MHz 80286 microp-rocessors used in the PS/2 Models 50 and 60support a minimum cycle of 300ns. whichtranslates to 6.7Mbyte/s. The 16MHz 80386Models 70 & 80 have a minimum cycle of250ns, or 16Mbyte/s. The higher -performance 20MHz 80386 version will beable to support the full 20Mbyte/s transferrate of the Micro Channel.


INSIGHT EXTRA

Bus masters. NuBus arbitration priority isfixed and depends on the card slot ID. withslot number 15 having the highest priority.Figure 1 shows the relationship of theNuBus slot IDs for the Macintosh II.

The Micro Channel defines 18 levels offixed priority, see Table. The arbitration levelused by an add -in card is normally controlledby a Programmable Option Select registeron the card which allows a limited form ofprogrammable priority.

DMA transfers. The Micro Channel compu-ters are equipped with an eight -channeld.m.a. controller. Add -in cards can use thed.m.a. controller to perform data transfersusing the standard Micro Channel busarbitration mechanism. The d.m.a. control-ler on PS/2 Models 50 and 60 runs with a10MHz clock. and the 16MHz Model 60d.m.a.c. runs at 8MHz.

The Macintosh II does not provide a d.m.a.controller on the backplane. For d.m.a.comparison, we used the NB -DMA -8-G card,Fig.2. This provides the Macintosh II with ageneral-purpose d.m.a. controller for thePC -style NuBus in addition to a 488 businterface. It uses an Intel 82380 controllerrunning at 10MHz.

An analysis of the d.m.a. capabilities of theMacintosh II equipped with an NB -DMA -8-Gcard as compared to a 16MHz IBM PS/2Model 80 with built-in d.m.a. controller isshown in the Table. The analysis is theoretic-al and assumes best -case conditions.

Each configuration can support eightindependent channels. The 82380 on theNB -DMA -8-G can perform 8. 16, or 32 -bitdata transfers: the PS/2 Model 80 can per-form 8 or 16 -bit transfers. The Model 80d.m.a.c. transfers data using a fetch anddeposit technique. called flow -through ordual -addressing. The 82380 normally usesthe same fetch and deposit method, but canalso support single address or flyby d.m.a.transfers.

For this theoretical analysis, I assumedthat the data transfers can be performed inthe minimum time possible: for both MicroChannel and NuBus.

Burst capabilities allow the d.m.a.c. tokeep the bus as long as necessary. Assumingthat no higher -priority master is using theNuBus, we can determine the maximumtransfer rate obtainable using burst mode.For the NB -DMA -8-G, this time is 600ns per32 -bit fetch and deposit cycle 16.67Mbyte/s)and 400ns per 32 -bit flyby cycle (10Mbyte/s).

On the PS/2 Model 80, the memoryrefresh arbitration level will override anyburst transfer in progress. Since a memoryrefresh takes place an average of once every15.8µs,this will limit the number of 16 -bitfetch and deposit transfers that can beperformed during a burst to 25. Using thisfact. the maximum burst data transfer ratethat can be achieved is 3.1Mbyte/s.

The 82380 d.m.a.c. on the NB -DMA -8-C isequipped with a 24 -hit transfer counter.allowing it to transfer a maximum of16Mbyte/s with one d.m.a. transfer opera-tion.

The custom d.m.a.c. on the PS/2 Model80. and on Models 50 and 60, contains a

CARD SLOT ID

9 t0 11 12 13

CARD SLOTS

MAIN LOGIC BOARD

ID -c

DISKDRIVES

16 -bit transfer counter, providing the abilityto transfer 64K 16 -bit words (128K bytes) or64K bytes with a single d.m.a. operation.

Interrupts. The PC -style Nubus provides aninterrupt request signal for slave cards callednon -master requests. or NMRQ. As im-plemented on the Macintosh II, each cardgenerates a separate interrupt request signalto the 68020 microprocessor on the mother-board. Each slot's interrupt has a prog-rammable priority. Interrupts on the Macin-tosh II are level sensitive as opposed to edgesensitive.

The Micro Channel implements a share-able. level -sensitive, multiple -line interruptmechanism. A total of 11 shareable interruptlines are available on the bus. Priorities areshown in the Table.

MAIN FEATURES OF NUBUS AND MICRO CHANNEL

TypeFrequencyI 0 address widthMemory addresswidthData widthCard addressing

Data address linesMultimasterBus parityNumber of cards

NuBus

synchronous

32 bit

8/16/32 bitGeographicslotMultiplexedyesOptional16

MicroChannelasynchronous

16 bit24:32 bit

24/32 bitPOS

Separateyesno8

Theoretical maximum bus transferratesSingle cycleBlockBlock size

20Mbyte/s 20Mbyte/s37.6Mbyte/s -16 (64byte) -

Bus mastersNumber (max.)ArbitrationPriorityFairness

16ParallelFixedYes

16SerialProgrammableProgrammable

MicroChannel arbitration priorityArb level Assignment

-2 (highest)-10.78-EF

Memory refreshn.m.i.d.m.a.ReservedSystem microprocessor

MicroChannel interrupt prioritiesNMI

IRQ 0IRQ 1IRQ 2

Parity. watchdog tinter.Arbitration time-outChannel checkTimerKeyboardCascade interrupt control -IRQ 8 Real time clockIRQ 9 Redirect cascadeIRQ 10 ReservedIRQ 11 ReservedIRQ 12 MouseIRQ 13 Math coprocessor exceptionIRQ 14 Fixed discIRQ 15 Reserved

RQ 3 Serial alternateRQ 4 Serial primaryRQ 5 ReservedRC 6 DisketteRC 7 Parallel portRQ 8-15 are cascaded through IRQ 2

Mechanical comparisonNuBus Micro Channel

Bond area 52 in' 35 inNurnber of i.cs 74 50(i 0.7 in2,i.c.)Intncard spacing 1 in 0.8 inComponent height 0.55 in 0 6 inConnectors 96 pin DIN Card edge

Theoretical maximum d.m.a. transferrates (byte/s)

Bus connectors and signalsNtiBus Micro Channel

16 16 VE 32 32 MMS16MHz

Macintosh IIwith NBDMA8G

PS/2Model 80

Total active pinsDigital bus signals

9651

11276

13276

174107

182107

Number of :Flannels 8 8 8 Total power 23 12 12 18 18DMA transfer type dual single dual 5V 11 7 7 10 10Cycle steal transfer 12V 2 3 3 6 6rate 5.0M 6.7M 2.0M 12V 2 2 2 2 2

Burst mode transfer 52V 8 0 0 0 0

rate 6.7M 10M 3.1M G-ound 22 17 22 26 28

Maximum transfersize 16MB 16MB 128KB

Other andrrserved

0 7 22 23 29

ELECTRONICS & WIRELESS WORLD s19

Better, nicerconnector

The review of 'Electronics andcomputer acronyms' by PhilBrown IEIVW July) mentionsthat BNC (as in connector)stands for Bayonet Nut Cou-pling.

A more interesting version isgiven by Chet Heyberger andMarshall E. Pryor in 'The XYZs ofusing a Scope' (Tektronix.Beaverton. 1981) in which BNCmeans "Bayonet Neill-Concelman": named for PaulNeill. who developed the N seriesconnector at Bell Labs, and CarlConcelman, who developed the Cseries connector.Roy Fursey.Netherburv.Dorset.

I am afraid I must take issue withthe statement that BNC standsfor Bayonet Nut Coupling (bookreviews. &Wit' July 1988). It isin fact short for Bayonet NavyConnector, being a developmentof the "N" or Navy type coaxialconnector. SMA. SMB and SMCare variants of the Sub Miniatureconnector. All this informationcan he found in Hewlett Pack-ard's professional r.f. testgearcatalogue - which contains agreat deal more very useful in-formation (and doesn't costc£24!).

Andrew Emmerson.Northampton.

School leaversMr Linsley -Hood's letter in yourApril issue touched a responsivechord.

In the early 'eighties I wasemployed by a major South Afri-can recruitment company, in-volved in placing technicians andengineers.

During this time we saw anumber of young persons whowere both "inspired and inquisi-tive". but lacked paper qualifica-tions. Even in this country of(electronics) opportunity. theyhad difficulty finding positionscommensurate with their abili-ties.

In the end most of them optedfor the only other option. and

FEEDBACK

started their own companies.Thank you for expressing

these very necessary sentiments.Clyn S. Craig.Microdyne c.c..Edenvale.Transvaal,South Vrica.

F.m.broadcasting

Regarding the comment in RadioBroadcast in your July issue -"listeners have failed to respondto the 'listen on f.m. exhorta-tion' "- the basic reasons for thisare the design of domestic re-ceivers (the proposed BBC re-ceiver mentioned in the Juneissue would be a giant leap for-ward) and the programmes thatare available on f.m.

The BBC must take the blamefor the latter, with their policy offrequency -hopping. Imaginehow long we would have beenstuck with 405 -line televisionhad the same policy been appliedthere. You switch on BBC( on625 at 5p.m. on Saturday and theannouncer says 'BBC] program-mes continue on 405 only: on625 we have the third in theseries "Spanish for the Hard ofHearing".

Reinforcing the 'listen ona.m.' attitude is the fact that,after twenty years, the BBC'smost popular channel has nof.m. service except in the Londonarea and the Radio Times listsprogrammes as on a.m.

What was surprising from thesurvey mentioned was that 40%do listen to f.m. at some time,despite all the efforts to discour-age them from so doing.S.J. Hampton.Shoreham -by -Sea.West Sussex.

Pseudo audioI read with interest Mr Self'sarticle (July) on the contempor-ary attitude to audio engineeringand its products. and I am grate-ful that EMI' has provided aforum in which this matter canhe debated.

Since Mr Self invites com-ments. I would like to offer a fewthoughts on this topic. Firstly. I

am wholeheartedly in favour ofhis basic premise. that engineer-ing progress must be founded onobjective and reproduciblemeasurements, and on specifictechnical goals. Opinions.however august their author.cannot be anything other than apossible corroboration that thetechnical targets sought by theengineers are correct, or an in-dication that the desired speci-fication is, in some way. less thanadequate.

However, the fact that the'subjective sound' pundits evergot their foot in the door of thestandards laboratory in the firstplace is the engineers' own fault,in that when transistor ampli-fiers were first introduced, as an'improvement' on the existingvalve -operated designs. they suf-fered from large amounts of re-sidual crossover distortion - lessthan 0.1%. perhaps, at full out-put power, hut several percent atnormal listening levels. The en-gineers. and the manufacturers'sales blurbs. said they were ex-cellent. hut the man in thestreet, and his wife, thought theywere nasty to listen to. as indeedthey were.

We, as engineers, havingtardily admitted our error in thismatter, then went on to stabilizeour amplifier designs, which hadmassive amounts of negativefeedbck to improve the steady-state linearity, with dominantlag compensation, usually acrossthe second gain stage within theoverall feedback loop, becausethis particular technique led tothe hest high frequency t.h.d.figure. This led, in turn, to aud-ibly objectionable slew -ratelimiting effects, the existence ofwhich was again overlooked inthe specification, but made theamplifier sound poor, especiallyat higher output levels.

Once again, the man in thestreet noted that there were dif-ferences in sound quality be-tween similarly specified de-signs. and that what we said wassuperb might not provide thesound quality he expected. This,and a few dozen other similaroversights gave the potentialbuyer of audio equipment theincentive to seek opinions basedon listening trials, conducted byrespected figures among the

'golden -eared' fraternity. andprovided a ready market for thepublishers of the magazineswhich catered for this area ofpublic disquiet.

While I arrived on the transis-tor audio amplifier design scenerather too late to fall into thepitfall of class -B crossover distor-tion and I was aware enough tooffer designs which were notslew -rate limited. I am sure thereare other things which I did notdo as well as I should. simplybecause I do not know sufficient-ly fully the performance stan-dards I should seek to meet, orthe tests which I need to apply. todisclose the faults which a care-ful listener with good program-me material and ideal listeningconditions might perceive.

Taking one specific point aris-ing from Mr Self's article. I thinkhe. like the audio engineeringworld in general, places toomuch reliance on steady-statetests. such as his low signal -levelt.h.d. measurements on differenttypes of capacitors. which appearto have convinced him that thereare no effective performance dif-ferences between different capa-citor types. This may. indeed. heperfectly true under steady-stateconditions.

It is my opinion that the pur-suit of steady-state linearity hasnow been carried to such lengthsthat the hulk of the residualdifferences between specifiedelectrical performance and appa-rent sound quality lie in the -normally inadequately specified- behaviour of the audio systemunder discontinuous or tran-sient signals, which make upsuch a large proportion of musicor speech. It is in this respectthat many components. particu-larly capacitors. exhibit a rangeof known and measurable de-fects.

With regard to the transientperformance of audio equipmentin general. it is certainly possibleto distinguish between systemshaving adequately low steady-state waveform distortion, butwith large differences in theirresponse to step changes orother waveform discontinuities.

However, since these differ-ences in transient response areseldom measured and the per-formance. in this respect. is sel-


dom defined. it is possible that inthe eyes of the lay user thediffering units will appear to beidentically specified, which leadsto a consequent. but dis-appointed. expectation of iden-tical sound quality.

I would point, in this context.to the substantial improvementswhich have been made in soundquality during the developmentof both cassette recorders andcompact disc players, as betweenearly and later designs of thesame systems, of which much ofthe improved performance is dueto significant improvements inthe accuracy of reproduction oftransient -type signals ratherthan to any lessening of harmo-nic distortion.

Overall. I simply do not believethat we yet know enough aboutthe importance. in audio terms.of some of the electrical short-comings of components, or ofthe waveform distortions arisingin the signal chain due to compo-nent imperfections or injudi-cious circuit design - which leadto effects which can be bothdemonstrated and measured -for us, as engineers, to be able tosay that there will be no audibledifference between component'A'. having one set of knowndefects, and component 'B'which has another. different, set.or between systems of whichonly some of the performancecharacteristics are identical.

This is not intended as a justi-fication for high -tensile -steelcontrol knobs or mains wiringmade from single crystal copper.but as a plea that we should try toassess the significance of theelectrical phenomena we doknow about, and can measure.before either saying that defectsdue to this cause don't exist orthat the effects that they dointroduce must be inaudible - inview of the fact that most avail-able programme material is

already degraded.J.L.Linsley Hood.West Monkton,Somerset.

Could it be that my stiff neck isdue to the vigorous nodding ofmy head whilst reading DougSelf's article? Alas. I fear that inyour august journal, his wellreasoned arguments are largely

FEEDBACK

directed at the converted. It is

safe to assume that, had he offer-ed it to the editorial office of anyof the well-known hi-fi comics, itwould have met with the enthu-siasm accorded to a dead rodent.

I cannot but agree with everyword, and I say this with feeling,having contributed regularly tothe emerging hi-fi magazineworld in the distant past. review-ing as well. But what l find muchmore disturbing is the steadyerosion of ethical standardswhich the review "industry" (Iuse the word advisedly): the de-terioration of technical compe-tence and what appears to becovert corruption as well. Strongstuff.' Yes, it is - but there ismore than enough evidence of itstruth.

On the question of standards,for a start: amongst those re-viewers who actively lobbymanufacturers and importagents for work, very few haveany academic qualifications ortraining whatsoever: I usuallyrefer to them as the "ex -shopassistant" brigade. From them.we cannot really expect anythingother than their own concept ofwhat is acceptable. Sadly, thattiny minority who do have anengineering background havenow been fully corrupted by whatI refer to as the "Emperor'sclothes" syndrome. They eitherjoin the Subjectivist chorus - orlose work. Simple as that. It isn'thelped, either, by universityacademics who lend a gloss ofrespectability by contributinglargely unreadable treatises fromtheir ivory towers.

We must remember, and thismust not be interpreted as ajustification for all this non-sense. that hi-fi is now aboutconsumer durables. like washingpowder. Change the perfume andthis is the New. Improved Hi-FiAmplifier: and a sycophantic re-view press obligingly dances tothe industry's tune.

Ethics? Who pays for thosetrips to shows abroad? Not themagazine, that's for sure. Whyare some advertisements refusedby at least one well-known maga-zine. with no explanation given?Why does at least one of ourleading high -quality audiomanufacturers, Quad. openlydiscourage reviews of their pro-

ducts (Letters. July. HiFi News)?Could it be that, bearing in mindthe questionable expertise ofthose to whom their product isentrusted. they regard a "good"review just as damaging as a"bad" one? Looking at a couplerecently, a very expensive ampli-fier was measured as having anoverload of 23dB on a CD input of100mV sensitivity. Another, inwhich a CD player measuredpoorly. yet sounded the "best". Inboth cases, the reviewers hadsome degree of engineeringqualification - yet not one wordof explanation. These two exam-ples are far from isolated. Read-ers of this magazine will need noexplanation from me, as to theabsolute failure in this instanceto justify public trust.

Like Doug Self, I fear thisnonsense is going to be with us awhile yet. Experienced engineerswill run a mile, rather than takeon a review commission thesedays. This is not to say its endcould not be hastened. Reputablemanufacturers could go publicand disown all reviews. until themagazine business puts its housein order. Those reviewers who dostill enjoy some respect and areclinging to the wreckage -mere handful, sad to say - shouldcease their ambivalence. Aban-don all that euphemistic custardpoured over a dubious product:emulate the little girl in the fairytale and cry "But this is rub-bish!". I'd like to think it waspossible, but you know they havea living to earn. Meanwhile. I

shall go on doing my best at theuniversity where I teach and nipthis rubbish in the bud as soon asI see it.Reg Williamson,Kidsgrove.Staffordshire.

Three cheers for Doug Self andhis spirited attack on the lunaticfringe of the hi-fi fraternity.More power to his elbow.

I would like him to know thatthere are plenty of us out herewho are strictly amateurs so faras audio engineering and hi-fiare concerned, but who arenonetheless enthusiasts and whohave some sort of rigorous scien-tific training, who find it im-possible to swallow some of thenonsense currently propounded.

I. for one, believe the follow-ing (amongst other things - thelist is purely representative, butone could go on and on):

(i) any subjective audible dif-ference between, say. amplifiers.which can be shown to exist bymeans of a statistically well -controlled double-blind trial.can be explained in terms ofmeasurement and current scien-tific knowledge:

(ii) it is possible to make one'sown perfectly competent if mod-estly powered amplifier, whoseperformance need not sensibly tobe improved any further. for lessthan .150:

(iii) loudspeaker cables have asmall resistance, even smallerreactance, and that is the end ofit - discussions of skin -depth, orworse. transmission lines. areludicrous at audio frequencies:

( iv) differences betweenelectro-mechanical transducers.and recording techniques (mic-rophone placement etc.) far out-weigh anything else in subjectiveterms.

All this seems really very ob-vious, and yet such is the currentseizure of the high ground by thesubjective extremists. that I. anamateur at that. feel obliged tostate it. Can I put in a plea forpeople to stand up and say so ifthey honestly can't tell the differ-ence between amplifier A or B.and not be intimidated intothinking that they have got clothears. rather than golden ones.Let us have the courage to shout"The king isn't wearing anyclothes!".John Howarth,Hull,Humberside.

It was with some interest andrelief that I read the article byD.R.G.Self. As a professional en-gineer with an interest in high -quality sound reproduction, I

have always thought of subjectiv-ism as with little technicalfoundation and even less com-mon sense.

I would not, however, havebeen moved to put pen to paperwere it not for an article pub-lished in the June 1988 edition ofHi-Fi Answers. The item entitled'Free of charge' in my view trans-cends by orders of magnitudeany other writings on subjective


audio. In it the author (MrJ.Hughes) puts forward thetheories of a Mr P.Belt claimingthat the sound of your hi-fi isaffected by such perverse thingsas the number of CDs on yourshelf, your wristwatch and evenwhether your sofa has three orfour legs. etc.

Much more worrying is thetheory that all unused mainspowered appliances in the homemust be "neutralised" by short-ing the live and neutral pins ofthe mains plug with a piece ofwire and crocodile clips.Although the author does statethat appliances must be unplug-ged, the implications for safety inthe home are clearly worrying ifreaders believe the rest of thearticle. After telephoning theeditorial department of Hi-FiAnswers to check if the articlewas genuine (which it was) I havewritten to both the magazine andthe publishing company to com-

If. as Mr Belt claims, allitems in the listening room thatcarry a static/electromagneticcharge should be neutralized,perhaps the listener should re-tire to another part of the housewhenever the hi -ti is operating,as he/she could be carrying alarge static charge and is de-finitely a source of electro-magnetic interference!

In conclusion. subjectivism inthe area of hi-fi has now movedway beyond the realm of goodsense and one wonders where itcan possibly go from here.S.J. Shorey,Dartford.Kent.

Diagrams,please

I am writing on behalf of theCalderdale Business and Innova-tion Centre (CBIC) with a specialrequest to your readers.

CBIC is a member of a Euro-pean network of centres set up tohelp small companies in allaspects of their business develop-ment. In particular we deliber-ately encourage such companiesto use modern technology totheir advantage. As part of ourwork we are planning to stage

rFtEDBACK

several events and seminars forcompanies in the Yorkshire andHumberside area. We also takepart in events which bringtogether other members fromEurope. At present I am prepar-ing a programme which will(hopefully) demonstrate thechanges in electronic technologyfrom the turn of the centurythrough to the present day. Theconcept is to illustrate, by meansof a series of circuit diagrams.not only how the technology haschanged but how the "technolo-gical information" has changed.i.e. the way the presentation ofcircuit diagrams as well as theinformation content therein haschanged. The period 1908 to1988 in 10 year segments hasbeen chosen to illustrate thistheme.

I am therefore requesting anyof your readers who has, or canget. access to circuit diagramsfrom these years if they would beso kind as to send me a copy. lamsure the engineers working forthe larger corporations would hewilling to dig amongst their arc-hive material to find a goodexample of their past work.These diagrams need not beoverly complex, indeed simplic-ity is probably a more effectiveway of getting our messageacross.

Any contributions to thisprogramme would be gratefullyaccepted and of course yourname and company would beacknowledged in the final pre-sentation.

Thanking you in anticipation.Bill Herdman.Dean Clough.Halifax HX3 5AXWest Yorkshire

Wagner,Debussy and

electro-magnetism

I have read the letter by T.R.Boag in the July issue of EWW.concerning the constant lc in F= k,.q (tar. the expression link-ing the magnitudes of the force Fbetween charges q1 and q2 sepa-rated by a distance r. and I am at

a loss to understand his difficul-ty. The constant is usually writ-ten as 1/flare and has the value9x 10" Nm2/C2 where F is innewtons. (41 and q., in coulombsand r in metres. This number iseasy to remember and is wrongby about 0.1%.

In my experience, eft occurs byitself far more frequently than1/47reft: hence I would not like tobe writing 1/4irke whenever Imeant the constant representedby eft if I were constrained to beusing standard notation. If MrBoag wants another symbol asstandard, I suggest he advocatessomething looking vaguely likeeft_ as is the case with Planck'sconstants h and h = h/2.

Finally. the equation

F= 1 '1012.

flare r2

expresses a relationship betweenpure numbers. A mathematicianwould not accept, for example,an answer F = 10 newtons - hewould insist that the force is Fnewtons, the distance apart is rmetres. etc., ensuring the quan-tities are pure numbers and giv-ing the answer F = 10. Giventhis, I do not understand whatMr Boag is trying to do at the endof his letter by dividing the mag-nitudes of a charge by a coulomb,of a distance by a metre, and soon. I suggest he consults themaths department of his uni-versity for help.Peter F. Vaughan.Coutant Electronics Ltd.Ilfracombe.Devon.

T R Boag's vision in the Julyletters of a madman wanderingabout at a Royal Society Soiree.and introducing himself to alland sundry as 'the Coulomb Con-stant'. is perhaps even moreappropriate than he realises. Inthe original absolute version ofthe electrostatic units there is nodimensional constant in theCoulomb equation. since theequation itself defines the unit ofelectrical charge in terms ofpurely mechanical quantities.Instead there is a purely numer-ical factor, which is specified indefining a particular absolute

system. The symmetry of theforce law demands that, if chargeis to be given an electricaldimension, then not one but twoequal dimensioned constantsmust be introduced, one associ-ated with each charge in theequation, i.e. that Coulomb's lawmust be written

F = qi q2

r2 7Tc k

where an/k is equal to 1/4 -n -e0. Ifthe consequences of this split aresystematically followed through,and the fact that some over-familiar equations were original-ly derived in the Absolute Elec-trostatic System and containsuppressed dimensional factorsis not forgotten, it can be shownthat E and D have the samedimensions, and that at anypoint in a vacuum both take thesame value.

The usual introductions of SIunits fail to take account of thesesuppressed dimensional factors.and introduce some legerdemaininto the handling of Gauss'stheorem. errors which lead to anumber of unexpected dimen-sional distinctions. Thus E and Dare assigned different dimen-sions. and take different values atthe same point in free space. Thisresult, which physicists of thegeneration of Jeans and Edding-ton would for good physicalreasons have rejected out ofhand, leads to the mystic notionof the 'impedance of free space'.

The logical structure of the SIelectrical units is sound enough.and their utility for practicalpurposes is unquestionable.However the expedients adoptedin deriving them created a single'Coulomb Constant' sufferingfrom incurable schizophrenia,which imposes a heavy burdenon anyone trying to understandrather than merely to use elec-tromagnetism. Nevertheless I

believe that this burden can herelieved by changes falling wellshort of complete destruction ofthe SI Valhalla for practicalunits.C.F. Coleman.Grove.Oxfordshire.


No charge to evaluate repairs.No extra premium

for emergency repair work.No limit on the type of

motorised products we'll repair.No need to test repaired items.

No high levels of inventory.No lengthy downtimes.

No problems, in fact.That's what Omega

service ability means.Yes, we have been repairing and refurbishing all your motors for

over a generation.Yes, our service expertise covers all your electronic and

electromechanical requirements.Yes, we are a BSI 5750 registered company.Yes, we do have a highly -skilled technical team to solve all your

difficulties.And yes, we are in the forefront of technology .. . sc anticipating

your future needs poses no problems.

Omega Dynamics, Kingsbury Trading Estate, Kingsbury Road,London NW9 8UP Telephone: 01-200 8844 FAX: 01-200 4816TELEX: 8954958G

The dynamic force in motion. ENTER 52 ON REPLY CARD

15.5 =1A4 Hi " /HZ + /H4 ;increment centre 101 UeVelOpIllelll Ul I CICIIIMILS

A3 * /A1 + /A4 * A3 bit3 Bangalore/AO + /A4 * A3 * India

HULL HLLLH L L L ; 17 ;LoopHLLLH HLLHL L L L ; 18 ;hereHLLHL HLLHH L L L ; 19 ;until reset


FCIRCUIT IDEAS

CIRCUIT IDEASFM quadrature i.c.works as a p.1.1.detectorNormally, the CA3089 and its many deriva-tives are used as quadrature f.m. detectors,and work well where distortion is the prim-ary design criterion. In some applications,however, noise and threshold performanceare more important. This idea was designedfor demodulation of weak wideband audio

11

-612V

100

subcarriers on satellite tv broadcasts, butwith modified values, it may be useful forn.b.f.m.

Inside the chip are a limiting amplifierand phase detector, so by adding an externalv.c.o. and loop filter a phase -locked -loopdemodulator can be realized.

In the circuit shown, i.f. input is con-nected as usual so the a.g.c. and signal -leveloutputs function normally. One input of thephase detector connects internally to the i.f.;the other input comes from the externalv.c.o. Control voltage for the v.c.o. is pro -

A =150kHz Ci =68P C2 omittedA 50kHz Ci = 22P C2 =10P

Fromi.f. filter 330

3

If amplifierand

limiter

13 15

a.g.c. meter

Phasedetector

CA3089IHA12411etcI

41 51 14j

Bias

47k

10

6 1

10u

II1

1

1

1

-1

1n

15k

4n7

1k

Loop filter

vided by the 'audio output' through the loopfilter. With values shown, the simple Col-pitts voltage -controlled oscillator operatesaround the i.f. of 10.7MHz.

Values of C1,2 were determined empirical-ly for various frequency deviations and theloop filter was optimized for high-fidelitymusic. Threshold performance is consider-ably better than the quadrature detector, butof course distortion is much greater.M.L. ChristiesonCrowboroughEast Sussex

Audio outputto de -emphasisstage

47k

10k Ci 22p

Oscil ator

NEXT MONTHMagnetic heading sensor. Using a pen-dulous flux valve detector to obtain adirect -current heading signal for air-borne telemetry. The device is compact,can be remotely sited and is free frome.m.i. emissions.

ast Fourier transforms of sampledsignals. A procedure to enable theaccurate determination of unknown fre-quencies. The author suggests that thef.f.t. algorithm does not require a re-peating waveform and that the waveformin the window is "considered" in isola-tion.

Remotely controlled Wien oscillators. Anovel Wien -bridge oscillator design us-ing analogue switches and photoconduc-tive cells to provide a coverage of 10Hzto 1.3MHz.

ELECTRONICS& WIRELESS WORLD

Telemetryheading sensor

Fast Fouriertnnsforn ofsampledwaveform,

RemotelycontrolledWien oscillators

Optocouplingin automatictest equipment

Pioneers -J.C.Mameell

Interfacing and signal processing withC. Programming in C has many advan-tages; this article describes how toexploit the language in interfacing amicrocomputer to the outside world.

Butterworth low-pass filter withequalization. Kamil Kraus shows that alow-pass Butterworth with equalizationleads easily towards the implementationof a higher -order crossover filter.

Bi-directional opto-coupling. If youneed to avoid trouble with earthing andnoise, opto-coupling could be theanswer. But if there is a computer and atext fixture at each end, it might not beeasy; R.A. Beck shows the way.


Fast logic probePutting the logic of a logic probe into a p.I.d. results in a fast

and versatile unit no larger than much simpler designs.

Recently, in a popular compendium ofelectronic circuits I found no fewerthan 24 logic -probe circuits. With so

many logic -probe designs around, I feel thatI must first justify my version. Its mainfeature is that it captures pulses too short todisplay on an oscilloscope. Sometimes simp-ly connecting the capacitance of an oscillo-scope probe removes erroneous pulses andcauses a faulty circuit to work perfectly -until the oscilloscope is removed. Beingbuilt around a programmable -logic device.the probe is small and simple, yet it includespulse -analysis features.

Two types of pulse that I frequently en-counter are problematical enough to justifythe probe's pulse -detection facility. They areglitches. i.e. pulses that cause erroneouscircuit operation, and pulses such as proces-sor chip -select signals. Neither type of pulseis easily made repetitive, and although pro-cessor strobes can be made repetitive bysuitable programming, some means of de-tecting these as single events certainly justi-fies adding to the logic -probe circuitry.

A major feature of the probe is its ability toclassify pulse widths into several bands; theprobe differentiates between fast glitches,typical processor strobes and strobes that aretoo long to be valid. Of these pulses theglitch is the most important since the probemust illustrate the presence of pulses thataffect present-day advanced logic such as74F and 74AC series.

Most logic probes have a pulse -indicationfacility based on either a monostable multi -vibrator or pulse -stretching principle. Ifonly edge indication is required, a 74F or74AC bistable i.c. with a delayed reset pro-vides about the fastest detector. This sim-plicity diminishes though when detectionfrom any input edge is required.

Any circuitry that responds to both edgesmust preserve the bandwidth needed forglitch -detection. When the basic logic -probelevel -detecting circuitry is added to thepulse -detection circuits, it becomes difficultto accommodate the number of chips re-quired inside a reasonably -sized case. It isnot surprising that simple transistor -basedprobes are so popular.

Each probe function requires only simplelogic, but brought together, these functionsrepresent an untidy mixture of gates andinverters. This mixture is ideal for im-plementing in programmable logic. Myprobe design has a single programmable -logic array providing the logic -probe func-tions and pulse -analysis facilities with fewadditional components.

13.J. FROST

Equivalent circuitry inside logs block PAL

Logic input

Logic valid

Logic valid

IC2A 2

02 r-7478

7404

7404:C c

Logic delayed

7404

10

5

7408

2

Pulse set inL.

7486

IC2B

3

7404

111B

6

1C3A

IC20

7404

Low

Logic pulse

Reset pulse lat- 2

114A

3 13

7600

IC4B6

7400

IC2F

7404

12 Pulse indication

This circuit repeatedthree times

Fig. 1. Logic of the logic probe is program med into a pal to save space.

Probe circuitry falls into two sections,namely the logic -state indicator and thepulse analyser. Circuitry shown in Fig. 1 andthe peripheral components of Fig. 2 will giveyou an insight into the operation of eachsection. Logic -state indication is simply amatter of determining whether the probeinput is high, low or. with the aid of awindow comparator, open circuit. Leds in-dicating the logic state are driven directly.

Being based on a set -reset bistable device,the pulse analyser is almost as straightfor-ward. Probe input passes through anexclusive -Or differentiator which produces alow -going transition on any probe inputtransition. This output feeds directly to thefirst set -reset bistable device which is im-mediately set on the shortest input pulse andso provides glitch detection.

Four more set -reset bistable devices arefed with the differentiator output but viaincreasing CR time -constants, such thatonly pulses that exceed each time -constantwill set the corresponding bistable deviceand illuminate its led. The glitch bistable

device will always be set on any pulse and isused together with a reset time constant toclear all the bistable devices after a predeter-mined display period.

PROGRAMMING THE PLD

Other articles have described the principlesbehind programmable gate arrays so I willnot give the full programming process. List1, produced using a simple text editor, ispassed through a proprietary logic -compilerprogram and a programming file is producedforsending to a device programmer.

Advantages of this method of logic imple-mentation are apparent from List 1, whereyou can see that the device inputs andoutputs are declared first with a pin nameassigned to a pin number in quite arbitraryorder and with a polarity that is indicated byan exclamation mark. Thus pin 1 is thelogic -probe input, true high, and pin 2 thefeed from the window -comparator circuitry,but true low. The two leds indicating thelogic state connect to output pins 12 and 19,


Tri-state check and input protection

100pR13

2k2

R20 100k

8

LM393

LM393

R14

IDA

1C1B

.5VInput supply to probe

P1

C7

0A47 Low drop -out regulator

LM2931Z-5.0

1 2

C8 C9 C10

1p 10n

Logic input 1

Logic block

UI

Valid logic 2

IlTe 3

Logic delayed 4

RI

1M5

C41pOV

Logic state constantAllows probe to detect pulsesof either polarity

R12

1M5C5

T

1p

OV

Pulse reset constantDetermines length of timethat pulse indicators are lit

Pulse

Pulse Ilse 6

Pulse T2 set 7

ulseT3 set 8

DulseT4set 9

11

R9

100RC2.rIn

VO

141-C1.r 1n

+0V100k

R10

c3.r1n

P11- OV

C6 100n

. 0V

Pulse time -constants

19

OV

16V8

5V

10n

OV

12

Pulse display

5V

Pulse Pulse Pulse Pulse

14 13 12 Ti Glitch

/16

Ds

13 RaTie 3

330 R 330 R 330 R 330 R5

D

14 PT f215 NE-re16 Pulse GI

17

18 krrilTeTT:

which are both true low, and the equationsthat relate their indication to the inputstates are:

/* drive HL display leds directly from logicinput'/high_led = logic_input & logic_valid:low_led = !logic_input & logic_valid:

An ampersand indicates the logic And func-tion and an exclamation mark indicatesinversion. Thus the 'high_led' pin is true(pin 19 goes low) when the 'logic_input' pinis true (i.e. pin 1 is high) And thelogic_valid' pin is true (pin 2 low). Com-pare this with the logic -gate equivalentcircuit of Fig. 1 and the similarities shouldbe clear.

This principle can be used to synthesizeany logic function and the bistable devicesare constructed in a similar way, but needslightly more explanation. When seen on acircuit diagram a set -reset bistable device isoften drawn showing a cross -coupled gatepair which conveys both the essential sym-metry and the circuit connections. Somedrawings do not, or cannot illustrate a gatepair in this way and place one gate followed

by another with correct connectivity butwith a more hidden actual function.

The logic equation for implementing sucha bistable device tends to look like this:

/* latch any pulse at all */pulse_C = (pulse_C & !reset_pulse) #pulse_C_set:

Here. the output pin of the bistable device('pulse_C') appears on both sides of theequation, indicating its latching capability.The hash sign represents the logical Orsymbol, and the equation output must be-come true if the input pin 'pulse_C_set'becomes true, irrespective of the state of theoutput: this is the SET action. Presence of atrue 'reset_pulse' causes the expression'!reset_pulse' to become false and anyself -latching cannot be maintained: this isthe RESET action. This equation is directlyequivalent to a cross -coupled Nand -gate pairand displays exactly the same instability ofoutput when both input lines are simul-taneously low.

As with any program listing these equa-tions and pin definitions are easily changedwithout changes to hardware, and it is this

ogic display

.5V

R7 330 Pe 330

06/1

07Low High

Glitch is any pulse

T1 is pulse >100nsT2 is pulse >1ps

13 is pulse>100psT4 is pulse >10 ms

.Eg Lattice GALI6V8-25LP

Fig. 2. These simple interface circuitsconnect to the pal to form the completelogic probe.

flexibility that makes pal devices so attrac-tive. Another advantage of this logic defini-tion is that the logic compiler used totranslate these equations into device dataalso allows a file of test highs and lows to behypothetically placed on the device inputpins and a print-out of the effect on theoutput pins to be obtained prior to program-ming the device. Such simulation provides agood chance of first-time success.

Implementing discrete logic with pal de-vices is an interesting and rewarding sub-ject. A useful publication containing dataand applications information is the AMDProgrammable logic handbook, availablefree to designers.

A fast mos version of the popular 16L8 palproviding up to 18 inputs or 8 outputs isused here. Its mos inputs provide littleloading to the probe tip and allow conve-nient CR network component values: itshigh speed provides pulse detection down tosingle nanosecond figures.


List 1. These equations, representing the logic within the logic probe. are used to producethe fuse plot for the programmable logic device.

I" Inputs/**Pin 1 = logic_input;Pin 2 -- !logic_valid:Pin 3 -- !reset_pulse,Pin 4 - logic_delayed:Pin 5 - !pulse_G_setPin 6 - !pulse_t1 setPin 7 = !pulse_t2_setPin 8 - !pulse t3 setPin 9 - !pulse t4 set

I" Outputs

Pin 19 !high_led:Pin 12 !low led;Pin 17 !logic pulse:Pin 16 - pulse_G:Pin 15 = !pulse_t1;Pin 14 = !pulse_t2;Pin 13 = !pulse_t3:Pin 18 = !pulse t4:

I' high is high. low is low -/* low if a valid logic level ..'/'true to reset the pulse capture *I' delayed i/p provides auto pulse info I' set pulse glitch latch / set pulse tl latch / set pulse t2 latch *.I' set pulse tl latch *.rI' set pulse t2 latch :

**/// high display led /' low display led/' pulse circuit drive'/1' low if pulse at all '// low if pulse _AI 'I/ low if pulse >t2'1/ low if pulse >t3% low if pulse >t4'/

/ Logic Equations/; drive HL display leds directly from logic input /high_led - logic_input & logic_valid:low_led = !logic_input & logic_valid./' generate pulse drive from exclusive -Or of input and delayed input /logic pulse - (logic_input & !logic_delayed)

# (!logic_input & logic_delayed):/ latch any pulse at all*"pulse_G = (pulse_G & !reset_pulse) # pulse_G_setI' latch any pulse ti ipulse tl - (pulse_t1 & !reset_pulse) # pulse_tl_set/* late). any pulse .A2*/pulse_t2 = (pulse_t2 & !reset_pulse) # pulse_t2_set/ latch any pulse :>t3 /pulse t3 - jpulse_t3 & !reset_pulse) # pulse_t3_set/ latch any pulse >t4pulse_t4 (pulse_t4 & !reset_pulse) # pulse_t4_set

HARDWARE DETAILS

All of the circuitry can he built into a probecase like the one available from RS and othersuppliers, and is easily accommodated onthe piece of perforated board that comeswith the case. The leds can he glued intosmall holes drilled in the case in groups oftwo (for the state indicators) and five (for thepulse display). All components are easilyobtainable except for the pals; I will he happyto supply these for readers without program-ming facilities (details later).

A small amount of power supply circuitryis shown in the diagram. Any 5V logic supplycould he used to power the probe directly butto ensure that it does not suffer damage, thesupply circuitry shown should he included.It consists of a low voltage -drop regulatorwhich only loses about 0.2V at full currentand a germanium diode to provide reverse -voltage protection.

Probe input thresholds remain at 1/3 and2/3 of 5V whatever the supply voltage, hutsupplies above 8V should not be used forextended periods. Do not omit the luFdecoupling capacitors on both sides of the

regulator. Due to their current -mode out-put. low -dropout regulators are prone toinstability if the required decoupling capaci-tances are not present.

For the best glitch -catching performance.ensure that the shortest possible groundwire connects the probe circuit to the circuitunder test. This connection performs thesame function as the ground clip on anoscilloscope probe and has the same charac-teristics: i.e. it can be ignored for general.uncritical checks, but if working with nar-row pulses with wide bandwidth it is essen-tial. Remember that all common logic fami-lies except 4000 series c-mos have an edgerate -of -change that is equivalent to band-widths in excess of 100MHz.

A programmed pal for this design can heobtained by sending £9.99 to Dorset Designand Developments at 8 Rohinswood Drive.Ferndown. Dorset BH22 9RZ. This priceincludes postage and packing.

Brian Frost is ttith Dorset Design andDevelopments

OOKSCircuits, signals and devices by MichaelJulian (of Bristol Polytechnic). LongmanScientific and Technical, £14.95. Well -filledtextbook on electronic engineering for firstand second year degree and diploma stu-dents, designed to meet the need of readersapproaching these topics for the first timealthough not geared to any particular sylla-bus. The author's treatment is primarilydescriptive. but he supplies extensive mathe-matical support as well. Text is comfortablyset out with many diagrams. An attractivefeature is the statement of learning objec-tives which appears at the start of eachchapter and the list of key points to remem-ber at the end. Soft covers, 515 pages.

Programming Windows: the Microsoftguide to programming for the MS-DOSpresentation manager. Windows 2.0 andWindows/386, by Charles Petzold. MicrosoftPress (Penguin Books). £22.95. Weightytechnical manual for the C programmer,providing detailed coverage of all aspects ofWindows in comprehensible form. Sectionheadings include basic concepts (creatingand displaying windows), getting input(from keyboard, mouse and timer. usingresources (memory management, icons.cursors, menus, dialogue boxes), thegraphics device interface (bitmaps. meta -files. printers and fonts), data exchange andlinks. Soft covers. 852 pages.

The programmer's PC sourcebook by ThomHogan. Microsoft Press (Penguin Books),£22.95. Excellent all -in -one reference workfor the PC user, covering chips and hard-ware, interfaces, DOS commands and util-ities, disc usage, function calls and supporttables, interrupts, mice, extended memory(Lotus -Intel -Microsoft) support. and Micro-soft Windows. The information is exten-sively indexed and cross-referenced; detailsof sources are given throughout. Softcovers, 525 pages approximately A4 size.

The early British radio industry by RowlandF. Pocock. lanchester University Press.£27.50. Early means pre -1900. the era whenthe main customers for radio equipmentwere Europe's navies. The author, tracingthe development of the infant art and thegovernment response to it (the British Gov-ernment's, mainly), provides a political andeconomic history rather than a technicalone: no diagrams or photographs illustratehis text. Did you know that Britain's indust-rial decline was recognized as long ago as1867, and was even then being attributed toa failure to encourage scientific and technic-al education? Those who ignore the lessonsof history are doomed to relive them. Hardcovers. 184 pages.


TECHNICA.

SYSTEMS LTD

The DIGIAC1050 has beenspecificallydesigned toprovide a singleself-containedtrainer that coversa.c./d.c. fundamentals,semiconductor circuits and digital electronics.For full details on the DIGIAC 1050 send forour full colour brochure or call us on Norwich(0603) 748001.

DIGIAC 1050 The Complete

Introductory

Trainer

The DIGIAC 1050 features: -

Self cor tained module with steel case and integralpower upplies.

Two curriculum texts available -FT02 - An AcCelerated Introduction to

E ectricity/ElectronicsFT04 - A Comprehensive Study of the Fundamentals

o' Electricity/Electronics 26 independant circuit blocks with different

configurations covering resistors, capacitors,inductors, op amps, diodes, relays, transformers anddigital logic circuits.

On -board switched faults - concealed by a lockablecover.

Screen -printed circuit layout. Comprehensive Instructor's manuals available for

both curriculum texts.

LJ Technical Systems Ltd.Francis Way, Bowthorpe Industrial Estate.Norwich. NR5 9JA. England.


HIGH QUALITYLOW PRICES

4 1-11GHz SATELLITE TVRECEIVING EQUIPMENT

RECEIVERS, LNB's, LNC's,FEED HORNS, ANTENNAS,ANTENNA POSITIONERS,

POLOROTORS, LINE AMPLIFIERS, ETC.

For further details contact:

HARRISON ELECTRONICSCentury Way, March,Cambs PE15 8QWTel: (0354) 51289

Toroidal& E.I.TransformersAs manufacturers we are able to offer a range ofquality toroidal and laminated transformers athighly competitive prices.

Toroidal Mail Order Price Listprices inclusive of VAT & Postage15va 9.12, 30va 9.48. 50va 10.16, 80va 11.02, 120va12.23, 160va 14.44. 225va 16.37, 300va 18.05, 500va26.46, 625va 30.66, 750va 34.14, 1000va 49.40.Also available 1k2, 1 k5, 2k, 2k5, 3k. Prices on request.

Available from stock in the following voltages: 6-0-6,9-0-9, 12-0-12, 15-0-15, 18-0-18, 22-0-22, 25-0-25,30-0-30, 35-0-35, 40-0-40, 45-0-45, 50-0-50, 110, 220.240. Primary 240 volt.

Quantity prices and delivery on request

Air Link TransformersUnit 6, The Maltings, Station Road,

Sawbridgeworth, Herts. Tel: 0279 724425ENTER 22 ON REPLY CARD ENTER 25 ON REPLY ('MU)


APPLICATIONS SUMMARY

SLEEP

I pk=VRs Vref /RS

Vref .(R3/(R243))X5VChopper frequency=036/(R6xCi)

*I unused inputs hgh) IC 2a

SN74HC''.

STEP

DIRECTION

1PR

10

1CK

Motor 28Orn4/phase with seriesresistance and inductance of 3711and 80mH respectively

5V 10k

39k

SN74HC112'

ICsa

SN74HC 386

1C5b

1PR

10

10

2PR

202CK

2Q

ICTh

L TLC 3702

Oscillator

104

10n

4

Vref 620 10k

20( 20

2K IC2b

IC 6b

.lcc1 \la 23,4EN

1Y

3Y4A

3A

SN754411L293

1A

2A

1,2EN

OR33 R

I 31-4;41 1N4935

a

VRS

Ds -8

All 1N4935

Constant -currentstepping -motor driveA brief stepping -motor overview, a circuitfor normal and half-step motor control, andthis constant -current driver circuit are con-tained in the application brief for TI'sSN754411/L293 stepping motor controller/drivers.

Motor -winding LR time -constant causesmotor current to respond slowly to steppedinputs. Consequently, during high-speedoperation, motor speed may never reach itsrated value. This chopping circuit allows themotor he be operated from a high -voltagesupply, so a faster current increase ispossible.

Audible motor noise is avoided by correctchoice of frequency -determining compo-nents R and C 120kliz). Peak motor currentis set by Vim.. and R,.

Osc

Vref

VRS

01w

OV

EMI suppressionResidual inductance in a capacitor, whichdepends on electrode construction and leadlengths. combines with the capacitance toform a resonant network. In e.m.i. suppres-sion applications, this resonant point cansignificantly reduce the capacitor's effective-ness.

These illustrations, from Murata's "EMIcountermeasure" guide, give residual in-ductance values and insertion losses oftypical two -terminal capacitors.

For a publication of its size. the guide doesnot contain much text, hut it iswell providedwith large illustrations. It presents a sum-mary of ti ,e fundamentals of e.m.i. suppres-

Capacitor type

SeriesCapacitance inductance(P.0 (nH)

Leaded monolithic ceramic 0.01 5Leaded monolithic ceramic 1 6Leaded disc ceramic 0.0022 4.5Polyethyleneterephthalate film 0.03 9Mica 0.01 52Polystyrene film 0.001 12Polystyrene film 0.1 100Solid tantalum 16 5Aluminium electrolyticfor r.f. use 470 13Aluminium electrolytic 470 130

sion techniques and explains the advantagesof three -terminal capacitors. Data sheets formany of the company's suppression devicesare included.

0

20-

60-,

80-

100,05 1 2 5 10 20 50 100 200 500 1000

FREQUENCY (MHz)

Tantalum C

Al electrolyticC 147p)

Monolithic ceramicC (1p)

Monolithic ceramicchip C1101

Monolithic ceramicchip C (10n)

Monolithic ceramic

- C (10n)- -


APPLICATIONS SUMMARY

ri nterdrivernode

Net wor kaccessnode

Ram

2Kx8

Net workaccessnode

Printerdrivernode

Addressswitches

Rom8Kx 8

Net workaccessnode

1421=116

Cl

Small -areanetworkingHardware for a simple small -area networkcan consist of little more than a number ofuniversal asynchronous receiver/trans-mitters capable of operating in 'multi -drop'mode. Such a network is the subject of noteAN990 from Motorola, entitled "A small -areanetwork using the MC68681 duart."

In multi -drop mode the serial frame bit.normally used for parity checking, signals anaddress when high, or data when low. Thisfeature allows a number of terminals withreceiver/transmitters to be linked serially tothe same controller and individually writtento. With a simple polling mechanism, theterminal controller can also determinewhether individual terminals have informa-tion to send back to it.

Optical shaft encoderwith five -secondresolutionNormal rotation encoders rely on sensingradial coded slots or holes on a disc. Incontrast, a new encoder technique de-veloped by Muirhead Vactric involvescharge -coupled sensors arranged in a ring sothat sensing over the whole disc surface ispossible.

Resolution is further increased by takingadvantage of the c.c.d's analogue sensingnature. The results, described in a brief

Minimum chargetransfer

Maximum chargetransfer

brochure called "The Microgon Encoder", isa 38mm rotary position encoder with aresolution of 4.94 arc seconds (the first unitsshould be available early in 1989).

On the encoder disc are two concentrictracks -a pseudo -random binary track and abar track - which form the moire fringes byinteracting with the c.c.d. elements. Thelight source is pulsed to form instantaneouspattern images on the c.c.d., resulting in a

Addres

MotorolaSt Martin's Way, Cambridge RdBedford MK42 OLF0234 47188

Muirhead Vactric38 Croydon Rd. BeckenhamKent BR3 4BE01 650 4888

Texas InstrumentsTechnical Enquiry ServiceManton Lane, Bedford MK41 7PA0234 223000

Murata Electronics5 Armstrong MallSouthwoodFarnborough. Hampshire GU14 ONR0252 523232

Ire

set of charges within the cells. These repre-sent the position of the pattern, and itscorresponding moire fringes. The c.c.d.correlates the pattern position with a prede-termined reference position on the disc.

872ELECTRONICS & WIRELESS WORLD

Tune in or fade out: that was the advicegiven by Alec Reeves in 1971 to en-gineers and managers to help them

face the next fifty years of electronics. Reeveshimself was a man highly tuned to hissubject. He was the inventor of pulse -codemodulation and conceived the idea of thewartime Oboe precision navigation system.His keen sense of the way ahead for telecom-munications was evident again in 1960 whenhe led a team on a then futuristic project toachieve communication using optical waves.The advent of lasers and fibre -optic cableseventually made that possible. C.K. Kao, thefather of fibre -optic cables, was employed atthe same laboratories and has recalled theinspiration of working among men such asReeves (see Pioneers 16, Electronics & Wire-less World April 1988).

For that next half century Reeves foresawpersonal global telephone numbers whichwould go with you anywhere in the world,and 700 -line opto-electronic picturephonesin colour and with stereo sound. For massstorage he suggested 1012 bits per cubiccentimetre as an achievable long-termtarget. That would enable an hour of colourtelevision programmes to be stored in acubic centimetre of material. The ultimatewould be .one bit per atom, and there arearound 1022 atoms per cubic centimetre in asolid. Generally what was needed to face thenext half century, Reeves suggested, was arevolution in thought.

Pulse -code modulation is one of thefoundation stones of modern telephony. Ithas especially good immunity to noise andhas been used to transmit pictures and datafrom space. The first American photographsfrom Mars came via a p.c.m. system.Reeves's patent was filed in Paris on 3October, 1938, though he had conceived theidea the previous year. As you read this,p.c.m. is 50 years old.

It may seem strange that a British inven-tor filed his invention in France but theexplanation is simple enough. At the time hewas working in Paris at the laboratories ofthe International Telegraph and TelephoneCorporation (ITT).

Alec Harley Reeves was born at Redhill inSurrey, on 10 March, 1902. He graduatedfrom Imperial College, London, in 1921 andtwo years later joined the InternationalWestern Electric Company. This was part ofITT and the forerunner of Standard Tele-phones and Cables (STC) and its laborator-ies, STL. There he worked on the firstcommercial transatlantic radio telephonecircuit and was one of the first to use theEccles -Jordan bistable flip-flop to build

Pioneers21. Alec H. Reeves, 1902-1971:

inventor of pulse -code modulation.

W.A. ATHERTON

Alec Reeves at the age of 63, holding acircuit board from a demonstration modelof a p.c.m. system (by courtesy of STCTechnology Ltd).

digital timing and counting circuits. In 1927ITT opened new laboratories in Paris towhich Reeves was transferred. There hestayed until the German takeover in 1940.

P.C.M.

In Paris the ITT group was looking at ways ofusing the s.h.f. and u.h.f. bands for radiotelephony links. The main problem wasnoise. Ever since A.G. Bell's invention of"speech -shaped current" the telephone hadused analogue circuitry. Frequency -divisionmultiplexing had made multiple channeltransmissions possible. But to Reeves.telegraph -style pulse techniques, used withtime -division multiplexing, seemed anattractive alternative to analogue methods.Pulse -amplitude modulation was tried andfound to work, but the varying amplitudes ofthe pulses suffered from noise as badly as didanalogue techniques.

In 1937 Alec Reeves came to the conclu-sion that what was needed was a return tothe fixed amplitude pulses of telegraphy.They had great tolerance to noise since theywere either there or not there. Digitaltelephony became his target.

Today's preoccupation with digital techni-ques makes it hard for us to imagine themental struggles that such ideas posed fiftyyears ago. Reeves first tried using the widthof the pulses to represent the amplitude ofthe signal (Fig.1). This is now known aspulse -width modulation (p.w.m.) He soonrealised that the width could be representedby short pulses for the leading and trailing

edge. or better still, by just one shortfixed -duration pulse whose position relativeto a fixed time represented the amplitude ofthe signal at that instant (Fig.1). This tech-nique, known as pulse -time modulation orpulse -position modulation, was used inmany early s.h.f. radio links.

Pulse -time modulation was an importantinvention but it did not solve the originalproblem since noise could affect the positionof the pulses and so produce distortions.These effects would be cumulative if passedthrough many repeaters. Reeves still wantedto approach the telegraph situation whereonly the presence or absence of pulsesmattered: for that he needed not only tosample the signal but also to convert theamplitudes of the samples into some sort oftelegraph. To do this he needed analogue -to -digital and digital -to -analogue converters.

He now fell back on his earlier work atdesigning counting and timer circuits usingthe Eccles -Jordan bistable. These he com-bined with pulse -time modulation (p.t.m.).By converting the analogue signal to p.t.m.and measuring the time intervals with hiscounter he designed an analogue -to -digitalconverter. Similarly he showed how to con-vert back from digital to analogue. In his1938 patent Reeves suggested a samplingfrequency of 8kHz (twice the baseband fre-quency of 4kHz ) and 32 amplitude steps (25).

P.c.m. did all that Reeves had asked of itexcept for one thing: it was difficult toachieve with the technology of the day.Widespread implementation had to await theera of cheap transistors and integrated cir-cuits. By that time the STC patents hadexpired and STC made no money fromthem2. After the war p.c.m. was developed ina number of centres, especially at BellLaboratories where its feasibility was provedin 1947-1948 by a team led by the famousHarold S. Black, inventor of the negative -feedback amplifier.

OBOE

In 1940 Alec Reeves fled from France leavinghis belongings behind - although, I am told,a suitcase was returned thanks to help givenby "a man he met in a pub". He joined theRoyal Aircraft. Establishment at Farn-borough but was transferred to the Telecom-munications Research Establishment thenlocated at Swanage. There he was in chargeof a number of groups of which one, Group17A, had the task of developing a newlong-range blind bombing system conceivedby Reeves and code -named Oboe. By now hehad laid aside previous qualms about work -


ing on offensive systems. The operationalsystem was worked out by F.E. Jones indiscussion with Reeves in April -May 1941but the principles came from Reeves. Air-borne trials began in September. a remark-able tribute to the small team which hadproduced all the equipment'.

An enormous amount of work and re-sources were needed to develop and useOboe even though its principle was remark-ably simple. Oboe relieved the bomber crewof the responsibility of deciding when torelease the bombs and gave that job toground -based radar operators hundreds ofmiles away from the action. The entiresystem relied on very accurate distancemeasurement of the aircraft by two widely -separated radar stations and on knowing theexact range of the target from the twostations. This latter requirement necessi-tated the accurate alignment of the Ordn-ance Survey with the continent of Europe -whilst the war was raging. A raid onFlorennes in Belgium (described by R.V.Jones in his book Most Secret War) helpedcheck this alignment.

Oboe made use of the fact that Britishradar could make range measurements bet-ter than it could make directional ones. Asthe bomber neared its target it flew on acircular course at a fixed distance from radarstation A (Fig.2): this would ensure that itpassed over the target, or rather just short ofit to allow for the tangential motion of thefalling bombs, which were usually droppedfrom 30 000 feet. Signals from station A toldthe pilot whether he was on track or to theleft or right of it. A second radar station. B.also tracked the bomber: when this was atthe correct range from the station the signalwas given to release the bombs.

Oboe was tested over bombing ranges inWales and near Stranraer, the latter some250 miles from the radar stations. Practicebombs dropped by Mosquito aircraft from30 000 feet gave an average radial error of150 yards.3

The first Oboe -controlled sortie againstenemy targets took place on 20 December.1942. By the end of the war the RAF hadflown about 10 000 Oboe -controlled raidswhich had marked the targets for nearly120 000 bombers. The US Air Force had alsoused Oboe on more than 1600 missions. Air

Signal

Pulse amplitude Imodulation

Pulse widthmodulation

Pulse timemodulation

Pulse codemodulation

11 111 1 111 11 111

L 1. 1111 11 1 1 1 1

1000 00. ,, Bic; 1c00

Fig.1. The path to pulse -code modulation.Over a couple of years in the late 1930sReeves progressed towards digital tele-phony by a series of steps including mod-ulating the amplitude, the width, and thenthe position (timing) of pulses. Finally, inp.c.m., only the presence or absence ofpulses mattered.

"If I could find theman responsible forthis fantastic Oboe Iwould sack him..."

Fig.2. Oboe. By means of a transponder inthe aircraft, radar stations A and B couldboth monitor accurately the distance fromthem to the bomber. Flying along a curvecentred on A. the bomber released itsbombs when instructed to by station B. Forthe Florennes raid in December 1942station A was at Trimingham and B atWalmer.

Marshal "Bomber" Harris later wrote, "Bom-ber Command's main offensive began at aprecise moment - the moment of the firstmajor attack on an objective in Germany bymeans of Oboe". It was a far cry from acomment in one headquarters file written atthe time when the principles of Reeves's ideawas being considered. Perhaps worried bythe transponder needed in the aircraft.whose transmissions could be detected bythe enemy, a Government scientist wrote ofOboe. "If I had the power I would discoverthe man responsible... and sack him so thathe would no longer waste not only his timeand effort but ours also by his vainimaginings"1. Instead of being sacked.Reeves was honoured with the OBE in 1945for conceiving the most precise bombingsystem of the war.

SAINTLY

After the war Reeves rejoined STL where hestayed until his retirement in 1970. Afterthat he formed his own consultancy andperformed work on optical communicationsystems for the British Post Office until hisdeath on 13 October. 1971 at Harlow inEssex. He had no next of kin. As well as histwo main claims to fame, p.c.m. and Oboe,he held about 100 patents and had worked onthe first short-wave radio telephone linkbetween Spain and South America.

Reeves has been described as "saintly". Itis a poetic description and one which doesnot spring easily to the lips of engineerswhen describing colleagues. Professor K.W.Cattermole has written that anyone whoknew him would remember his kindness andgenerosity, his charitable work, personalintegrity and sincerity. He devoted time.money and emotional commitment to help-ing rehabilitate young delinquents back intothe community and bore the breaking oftrust with the faith that "the right thing todo was the right thing to do". He was a keenmountaineer and encouraged a love for sportin young people, and did not mind recallingbeing beaten at discus -throwing by a 12 -year -old girl.

He received many honours besides hisOBE: a CBE, honours from the AmericanIEEE and the Franklin Institute, and atrophy and cheque from ITT. The mostunusual one however was a British postagestamp and first-day cover issued in 1969which commemorated one of the most im-portant inventions in modern telecom-munications - pulse -code modulation.

ReferencesI. K.W. Cattermole, Proc. IEE. vol.126. 889-892.September 1979.2. P. Young. Power of speech: a history of STC.Allen & Unwin, London.3. F.E. Jones. unpublished paper lby courtesy ofSIC).4. R.V. Jones, Most secret war, !famish Hamilton.London. 1978.

Next in this series of pioneers of electricalcommunication: James Clerk Maxwell -ABritish scientist second only to Newton.

Tony Atherton works at the IBA HarmanEngineering Training College. Seaton.Devon.


Multiprocessorsystems

In this third article, Alan Clements looks at howmicroprocessors communicate with each other.

up to now I have been looking at thetopology of multiprocessor systems\ vith little or no consideration of the

actual connections between the processors.Here I describe some of the ways in whichprocessors are linked together and examinethe various issues involved in linking them.

COUPLING

Possibly more than any other factor, therequired degree of coupling between proces-sors in a multiprocessor system determineshow the processors are to be linked. Atightly -coupled multiprocessor system passesdata between processors either by means ofshared memory or by allowing one processorto access the other processor's data, addressand control buses directly.

When shared memory, sometimes calleddual -port ram. is employed to couple proces-sors, a block of read/write memory is arrangedto be common to both processors. One pro-cessor writes data to the block and the otherreads that data. Data can be transferred as fastas each processor can execute a memoryaccess. Later I will look at the design ofdual -port ram and the problem arising whenboth processors try to access the same block ofmemory simultaneously.

The degree of coupling between processorsis expressed in terms of two parameters: thetransmission bandwidth, and the latency ofthe interprocessor link. Transmission band-width is defined as the rate at which data ismoved between processors and is expressed inbits per second. For example, if a 68000 writesa byte of data to an 8bit parallel port every 40.the bandwidth of the link is 8bitf4lits or2Mbit/s. However, if a Ifibit port is used tomove words at the same rate, the bandwidthrises to 4Mbit/s.

Latency of an interprocessor link is definedas the time required to initiate a data transfer.

Fig.1. Processors linked by a shared bus.

That is. latency is the time that elapsesbetween a processor requesting a data transferand the time at which the transfer actuallytakes place. A high degree of coupling isassociated with large transmission band-widths and low latencies. As you might expect.tightly -coupled microprocessor systems needmore complex hardware than loosely -coupledsystems.

The way in which processors are linked canbe listed in order of their degree of coupling asfollows:0 Serial data link

Fig2. Processors linked by a block ofshared memory common to both proces-sors.

0 Parallel data link0 Parallel data link with fifo buffers between

processors0 DMA channel0 Shared bus0 Shared common memoryHere. I will describe only interprocessor linksinvolving shared memory or buses.

SHARED BUS

One of the most popular ways of tightly -coupling processors is to permit two or moreprocessors to share the same bus in anarrangement of time -division multiplexing.That is, the bus is allocated to a processor onlyfor a fraction of the available time.

It is possible for more than two processors

CPU A

A D

A -addressD- dataC- control Jai

Ram

A


RAW.]

ENABLE.]

ADDRESS,'

Addressbuffer

1

D1R

Databuffer

1

Accessrequest

RIW

R/171 CS

Dualportram

It

Arbiter

Arbitrationclock

Mer

1

DIR

Databuffer

AccessMyles+

F2/1,;2

ENABLE 2

ADDRESS2

AS2

il5S2

DS2

'T4C1(2

Fig.3. Structure of a shared memory common to two processors.

to share the same bus, allowing the generalbus topology to he realized: this is why theshared bus is so popular. It should he im-mediately apparent that if a number of proces-sors do try to access the common bus all thetime, the throughput of the individual proces-sors will be very low.

Figure 1 shows how a shared bus system isarranged. Each of the three processors. A. Band C. has its own local buses through whichit accesses its own local memory and. possibly.i/o ports. For much of the time. the processorsexecute programs stored in their local mem-

ory and operate on their own data. When oneprocessor wishes to communicate withanother. it takes control of the system busthrough its bus -control unit. Now it can

access the resources of another processor bycontrolling the system bus.

When I introduce the VIEhus in a laterarticle, you will see how the 68000 bus controlfacilities hilt lei and ismck) simplify the designof multiprocessor systems.

Fig.4. Timing of an indivisible test and set(TAS) instruction.

CK

A1-23

S7 So St Si 53 Sz, S5 S6 57 $8 59 SlO S11 S12 513 S1: S15 Sx, St2 Sts Si*si_

AS asserted for duration cc TAS cycle

Data from memor y

Indivisible read /modify/write cycle

Read memoryInternal

10 ,1operation

1

Data from 68000

Write to memory

SHARED MEMORY

In the final method of linking two processors ablock of memory common to both processorsis used. As I have already stated, commonmemory is often called dual -ported rani.Fig.2. It is, in fact. possible to have n -portedram that is shared between n processors.Because multiprocessor systems based onshared memory have very h 1W latencies due totheir tight coupling. I will discuss the designof multiprocessor systems with shared mem-ory in more detail.

DESIGNING MULTIPROCESSORSWIT! I SI IARED MEMORY

In the example of Fig.2. the two processorshave entirely separate buses together withtheir own local memory and i/o resources asappropriate. However, each processor is alsoable to read from or write to a block ofdual -ported memory common to both theiraddress spaces. Figure 3 provides furtherdetails of the arrangement of this memory.For the sake of simplicity, byte -selectiondetails are not included.

Each processor in Fig.3 has its own dataand address bus buffered on to the dual -portram buses by means of two pairs of three -statebuffers. Processor I accesses the dual -portedmemory when ENABLE! is asserted and proces-sor 2 accesses the dual -port ram when ENABLE.,is asserted. As the outputs of the buffers areconnected together, only one set of buffers at atime may he enabled. Therefore. both proces-sors cannot access the memory simul-taneously.

Inputs cs and RAV to the dual ram are alsobuffered by three -state bus drivers. However,

s76ELECTRONICS & WIRELESS WORLD

in this case it is necessary to pull up both Lsand kW when neither processor is accessingthe bus, in order to avoid spurious memoryaccesses.

ARBITRATION

So far so good. The real problem in designinga shared memory system lies in the selectionof a suitable mechanism by which a processoris granted access to the shared memory. Eachprocessor in Fig.:3 may attempt to access thedual -port ram by asserting its AN. 1.1.N and Loslines and by placing the appropriate addressonAl

Access requests from both processors arefed to an arbiter circuit to decide whichprocessor will he granted access to the sharedmemory block. Some arbiters have a clockinput (not necessarily the system clock) thatdetermines the points at which the requestinputs are sampled. That is. the requests aresampled periodically by a sampling clock. Thisclock has important implications for thereliability of the system as you shall soon see.

If one processor requests the memoi-y andthe other does not, the requesting processor isgranted access. Equally. once a processor hasbeen access, the other is locked out

RECITgE1

Arbitrationclock

REQUEST2

IC3

K

0

CK

CK 1C2a

CK

BB

PRE

D

CK !CR)

a

PRE

aCK 1C2b

a

ENABLE 2

Fig.6. CircLit diagram of a two -input asyn-chronous arbiter taken from a Motorolaapplication note. Latches ICI, are drivenby a non -inverted clock signal and latchesIC2,1, by an inverted clock to ensure thatREQUEST? and REQUEST? are not sampled at the

same time.

Fig.5. Arrangement of dual -ported ram in a multiprocessor system.

and must wait until the memory becomesfree. When both processors request a memoryaccess almost simultaneously, the arbitermust decide which processor may proceedand which must wait.

The arbiter is an interesting device, notleast because it is impossible to design a

Bus to CPU1

IIIM11114

LS245

Databuffer

1

DIR

R/W1

Bus ' PU2

41ff

DIR

Databuffer

2

LS245

LS244U

Address071 and

control

0-11 "1buffer

5

Ram

A

32k x 8Upper byte

Ram

32k x 8Lower byte

CS 1,T.'

R/W

LS245

Databuffer

3

E DR

Databuffer

4

LS24

UDS

LDS

E E

Addressbuffer

6

Lc24.

Addressdecoder

1

CS(active whenshared memoryaccessed I

ENABLE 1 I I ENABLE 24 1.

Request)

DTACK1

Arbiter Reques

LS244

Addressand

controlbuffer

7

DTACK2

Addressbuffer

8

LS244

Addressdecoder

2

ELECTRONICS & WIRELESS WHOA) 77

8Ons

125MHzCK

12.5MHz -CK

1_,T5111 -L-11_11

Idle Contention Request? gets bus

REQUEST1

RequestiIdle gets bus Idle

REQUEST2

A

B

AA

BB

E NABLE

ENABLE?

Fig.7. Timing of a dual -ported ram arbiter.

perfect arbiter. It is an asynchronous devicebecause a processor may request a memoryaccess at any instant if its clock is independentof the arbiter's clock. Therefore, there is afinite probability that a processor may requestaccess at the exact moment ( in practice afinite sampling window) that the arbiter ismaking a decision. In this case, the state of thesystem is undefined.

No matter how well an arbiter is designed.the problem of a simultaneous access requestand sampling clock edge can never be elimin-ated. Simultaneous changes at a bistabledevice's data and clock inputs are called meta-stability. It is possible however to reduce theproblem to acceptable limits. For example, anarbiter may be designed that fails once every.say, fifty years. A mean time between failure of50 years represents, of course, a high degree ofreliability by everyday standards. Note that afailure due to a metastability problem is a softfailure.

Another problem associated with sharedmemory, common to almost all forms ofmultiprocessor systems. is in ensuring reli-able communication between the processorsvia the shared memory. For example, supposethat processor A writes a block of data to thedual -port memory for processor B to read.Processor A will want exclusive use of theshared memory for the entire duration of thisoperation. Since processor A cannot lock outprocessor B by a hardware technique, it is

necessary for A to pass a message to B.

indicating to it that it wishes to have soleaccess to the memory.

Typically a flag bit, called a semaphore.forms one of the memory locations and. whenset, indicates to a processor that the memory

is in use by another processor and is thereforeunavailable.

In order for processor A to take control ofthe shared memory. it first reads thesemaphore. Remember that both processorscan physically access memory. subject only tothe state of the arbiter. If the semaphore is set.processor A must wait. since B is already usingthe memory. If the semaphore is clear. proces-sor A may take control of the shared memoryby setting the semaphore and thereby lockingout B.

Simple? Well. not really. Assume that pro-cessor A reads the semaphore and finds itclear. Processor A will then perform a writeoperation to set the semaphore and to gainpossession of the shared memory. Now sup-pose that at an infinitesimally small periodafter processor A requests the memory, pro-

Fig.8. Generating DTACK in a dual -portedram.

cessor B does the same. Processor B will readthe semaphore after A has read it and it toowill set the semaphore: each will think that ithas sole possession of the shared memory.

The above is not a theoretical, abstract orirrelevant problem. It has serious consequ-ences in multiprocessor systems. For exam-ple, in a geographically distributed system atravel agent may aks for a seat on a certainflight. The computer system interrogates thedata base and finds, say, a single seat free.which the client then hooks. But suppose thatin another town a second client also requests aseat on the same flight. He or she may be toldthat the flight is free because the first clienthas not yet booked the seat. The outcome isthat the seat becomes double -hooked.

Fortunately, most data bases have softwarethat avoids this dilemma. Once a client hasasked if a seat is available, that section of thedata base cannot he read by another user untilthe original transaction is fully completed.

The 6800(1 microprocessor has a test -and -set instruction. TAS. designed specially formultiprocessor applications. A TAS <effectiveaddress> instruction reads the contents ofthe specified effective address and, if clear. setsthe least -significant hit of the byte tested inone indivisible operation. That is. it does notrelease the memory between the action ofreading the semaphore and that of setting it.The action carried out by a T.Ns instruction iscalled an indivisible read. modify and writeoperation.

Figure 4 provides the timing diagram of aTAS instruction. which lasts a minimum of 20clock states and includes a conventional writeaccess followed by a conventional read access.Like any other memory access, both read andwrite operations can he extended by delayingdata -transfer acknowledge signal ifrAcx as re-quired. However, unlike other read-followed-hy-write cycles. the address strobe is notnegated between the read and the writeaccesses. Therefore, in any multiprocessorsystem using shared memory, the arbitermust he arranged so that it will never recind aprocessor's access to the shared memorywhile its address strobe is asserted.

MEMORY ARBITRATION

An arrangement of a practical dual -port ramwith 64Kbyte of shared memory is given inFig.5. Two 32K -by -Shit i.cs provide the 32Kwords of shared memory and eight bus driversform the electrical interface between the ramand processor I or processor 2. Processor data

REOUEST1

(E1)UDS1

LOS1

ENABLEfrom arbiter

12.5MHzarbiter clock

Iii*t) when CT.,1 andU ,t/CM) asser ed

4

SHIF T

High when UDS1/ LD'i asserted duringaccess by Processor 1

Vcc

tiltlSelect l;nk for aDpropriete delay

Ut

DIACK1


lines are buffered by 74LS245 bidirectionaltransceivers, whose data direction inputs arecontrolled by the BAN signal from the processorbus to which they are connected. The databuffers are enabled by ENABLE' when processor1 has access to the memory and by EMMY,when processor 2 has access.

Address lines .s! 13 from processor 1 are

buffered by two 741.5244 octal three -state busdrivers enabled by Emmy'. Lines Al 13 fromprocessor 2 are similarly buffered and enabledby EMI

Three 74LS244 three -state buffers are re-quired to buffer the 15 address lines and the

Ills and LDs control lines from the proces-sors. Each of these three lines is pulled up to alogical one state (on the ram side of thebuffers) by resistors, to avoid accessing theram when neither ENABLE' nor ENNBLE2 is

asserted.Two Or gates (And gates in negative logic

terms) permit the processors to perform byteor word accesses to the ram. These gate upperdata strobe VI PS with cs from the arbiter toselect the upper byte. and Los with cs to selectthe lower byte.

Address decoders 1 and 2 are driven fromprocessor 1 and processor 2 address busesrespectively. Each decoder is enabled by itsown processor's address strobe. For example.if processor 1 requests a memory access byaddressing the 32K block selected by addressdecoder 1. the decoder output (op.:sr' is

asserted. Assuming processor 2 does notrequire the bus. the arbiter asserts ENABLE' andbuffers 1. 2. 5 and 6 are enabled. This permitsprocessor 1 to access the memory exactly as inany other access to static ram. Note that esfrom the arbiter is asserted to enable the ram.if either ENABLE' or ENNBLE., is asserted. Shouldprocessor 2 request the ram. the action isidentical to that above, except that buffers 3.4.7 and 8 are enabled by ENNRLE.,.

Figure 6 is an example of an arbiter fordual -ported ram. taken from Motorola ap-plication note AN881. Each processor em-ploys its address bus together with Ns andstrobes ros is to produce a memory accessrequest whenever it wishes to write to or readfrom the shared memory. Memory requestsignals BEQuEsri and worEs-L. are sampled bytwo positive -edge triggered latches. The latchto he clocked first locks out the other latch. sothat the latter cannot respond to a request atits input. The secret of the arbiter is its use ofan antiphase clock to provide different sam-pling points for the request inputs.

Timing for the case in which both proces-sors request the bus simultaneously is shownin Fig.7. As you can see, processor 2 wins therequest and processor 1 must wait untilprocessor 2 has relinquished the bus. That is.processor 1 does not have to try again, itsimply waits for the memory to become free.Processor 1 determines that the bus is oncemore free by the eventual assertion of itsdata -transfer acknowledge signal.

Initially, the arbiter is in an idle state withboth request inputs inactive -high. Therefore.both 1) inputs to latches are high and ina steady state condition. Outputs A. in. ENABLE'

and EN.Vil Esi are all high.

Suppose that Itic.(tasii and IfEquEsT., are

asserted almost simultaneously when theclock is in a high state. This results in the

outputs of both Or gates (A and B) going lowsimultaneously. The cross -coupled feedbackinputs to the Or gates (ENABLE' and EN 11iI.E., ) are

currently both low.On the next rising edge of the clock, the Q

output of latch ICL, (i.e. As) and the Q outputof latch (i.e. BB) both go low. However, as

latch sees a rising edge clock first. its Qoutput goes low one half a clock cycles (40ns)before latch IC is output also goes low.

Earlier. I said that when a latch is clocked atthe moment its input is changing, it goes intoa metastable state in which its Q output is

indeterminate. The metastable state can lastfor up to about 75ns before the output of thelatch settles into one state or the other. Forthis reason a second pair of latches is used tosample data at the input latches after a periodof 80ns.

Thus. 80ns (one clock cycle) after REouEsr,has been latched and output Bis forced low, theoutput of latch IC'h. ENABLE., goes low. Itscomplement. EMISI.E.i. is fed hack to Or gateIC.;, forcing input .s high. After a clock cycle isalso goes high. Because ENAISI.E2 is connected tolatch IC h's active -low preset input, latch IC

Fig.9. Timing of DTACK during execution of a TAS instruction.

CPU1CK -11-1:11-1-5111-L-ULFCPUI-HAi_:3

AS1 j

LIDS1

[5Si J

R/1741

C St

REQUEST1)

F11 1

(reset shift reg I

ENABLEI

SHIFT

AArbitration delay

OTACKI

DTACK idelay

\A 1

DTACK delay

Fig.10. both arbitration logic and memory refreshing arecontroller.

provided by the 74LS764

Portrequestandgrantsignals

Memory

address

DTACK

SE Li

REQ1

SE L2

RECI2

GNT

RCK (Refresh ck)

Arbitrationlogic

CK

AccessD ramcontrol

RAS/CAS/MP>,timing

A2

ROW/COL

Address latchand

row/colmpia

MA0-6

Address tod -ram

ELECTRONICS & WIRELESS WORLD879

is held in a high state.At this point. ENUL.F:I is negated and

asserted, permitting processor 2 to access thebus. The access is completed by the genera-tion of a ofwK I (described later) and by thenegation by processor 2 of its AN and rl,Jl.usoutputs.

When processor 1 relinquishes the mem-ory. HEouEs-r., becomes inactive -high. causingfirst then nu and finally i svtl.Ey to he negatedas the change ripples through the arbiter.Once ENABLE, is high. ENARLE2 goes low. causing

the output of Or gate IC:; (i.e. s) to go low. Thisis clocked through ICI,,_,, to force Es 11tI lowand therefore permit processor 1 to access thememory. Of course. once I'.\.\III.EI is asserted.and assertion of mocEsT., is ignored.

The remaining aspect of the dual -portedram to he considered is the generation ofirrAciii and la by the arbiter. Many circuitswhich produce a 1,1 uK response to the asser-tion of the processor's address strobe follow-ing a valid address would nut he suitable for adual -ported ram. A glance at Fig.4. the timingdiagram of a TAs instruction. reveals the diffi-culty unique to TAs. This is the only instructionthat requires two handshake cycles (one toacknowledge the read cycle and one to ack-nowledge the write cycle) while As is con-tinually asserted.

RECO

SELL

GNT

RAS

Assert SE Li togrant REri.1

J

WG(Write gate(

CAS

DTACK

5

Fig.11. Timing of the LS764 arbitration sequence.

Fig.12. Dual 68000 system using a 74LS764 controller.

Vic

WALK

Databuffers74F 245

d Refresl-clock

3 -statebuffer

74F 1244or

74F 244

C P DTAIK R PRE01 RE 02

LS764

SELL 57.

A1-18 CASEN RAS WG MA 7.8

T

Decode

A

F139

t

3 -statebuffer

74 F1244Or

74F2,4

IUS WE A0..8 CZ

256k x8

RAS WE A0.8

256k x 8

Ri a A0.8 CZ

758k x 8

RAS tit. A 0.. 8 CAS

256k x8

Databuffers

74F 245

SS() III( Iiji)\ It. N.A. uillt1 I)

Figure 8 shows how DTACK1 for processor I isgenerated from ups' (or UN), REQUEST', andENABLE' from the arbiter. Basically, orAcki isgenerated in the normal fashion by holding ashift register (741S164) in a clear statewhenever Koms-ri (i.e. csil is inactive -high.and then using the shift register to shift aclock pulse along when it is enabled. In thiscase, the shift register is also reset when ill's'and Los' are negated. Consequently, the cir-cuit generates the two consecutive MACKI.Srequired during a TAS operation. Figure 9provides a timing diagram for the IYMKgenerator. Note that irmk., is produced in anidentical way. except that the shift clock isderived from the complement of the arbiterclock.

Now the hardware design of a dual -portedram is complete. In use it may be accessed by aprocessor as follows:

Shared -memory

Sem

EQU $F00000 Location ofmemoryblock

DS.W I First word=semaphore

LEA Semi, (AO)

Wait TAS (A01

BM1 Wait

Point tosemaphoreTest and setsemaphoreRepeat untilsemaphoreset

}Criticalsection

CLR.B (AO) Clearsemaphore

The address of the semaphore is loaded into\IP and the TAshimi loop repeated until thesemaphore is found to have its seventh hitclear (i.e. mil w ur not taken). When hit sevenof the semaphore is clear it is set by the rs toenable the processor to take control of theshared memory.

The next part of the program is called thecritical section because it is the region thatmust he executed to completion without anyother processor accessing its data. Once thecritical section has been completed. theshared memory is de -allocated by cry r I vi)which resets the semaphore.

DYNAMIC RAM CONTROL

Both arbitration logic for dual -port memoryand the controller and interface to a block ofdynamic ram are included in the 74LS764dynamic -ram controller. Fig.I0.

Request signals ur:oi and wo are theaccess -request inputs from ports 1 and 2.and sm.! and sEI. are the correspondingaccess -grant outputs. All other inputs andoutputs are largely self explanatory. Typicaltiming of an access by processor 1 is shownin Fig.11 and a circuit diagram of a dual68000 system based on the 74LS764 isshown in Fig.12.

A microprocessor requests access to thedynamic ram by asserting the appropriatemi) input. If a dynamic -ram refresh cycle is

not in progress and the other request inputis not active, the so. output corresponding tothe active iico will go low to indicate that theaccess will he granted. Output (.NI then goeshigh and its low -to -high transition indicatesthat a memory cycle is about to begin.

If an access or refresh cycle is in processand the other processor has not requestedaccess by asserting its xi:o signal, the NELoutput corresponding to the active Ho willgo low to indicate that access will be granted.However. ,Nii. will not go high until thecurrent cycle has been completed. Aftercompletion of the current cycle. (NT willautomatically be asserted high.

If access to the dynamic ram is requestedby both processors. the initial arbitrationphase determines which processor is to gainaccess to the memory by asserting theappropriate SEI. output. This arbitrationtakes place irrespective of whether or not arefresh cycle is in progress at the time theaccess is requested. Request contention isarbitrated, as discussed earlier. by samplingthe loio inputs on opposing phases of thearbitrer's clock.

When GNT becomes active -high, addressinputs II 1, are latched by the 74LS764 andthe NI 9 signals propagated to the NIN

row -address outputs. One half clock cyclelater. column -address inputs on AI -1,; arepropagated to m and the active -highcolumn -address strobe. cAs. asserted. Onceos is asserted, the controller will wait threeclock cycles before negating los. making itfour clock cycles in duration.

As RAS is negated, the MACK output becom-es active -high to indicate that data on thedynamic -ram data lines is valid (or that theproper access time has been met). All con-troller outputs are held in their final stateuntil the processor accessing the dynamicram negates its REvsignal. When the requestis negated, all controller outputs assumetheir inactive states and any impendingrefresh cycles are serviced.

A refresh cycle is triggered by internallygenerated refresh requests. The refreshclock is divided by 64 to produce periodicrefresh cycles and then arbitrated with the

outputs in the second stage of arbitra-tion.

Refresh cycles always take priority overprocessor accesses. At the start of a refreshcycle, the 9bit refresh address is placed on>1%,,,; for a half clock cycle and then its isasserted for four clock cycles, followed bythree inactive -high cycles. The inactive -highperiod of 1( %, allows time for the dynamic ramprecharge period.

I introduce VMEbus and look at some ofthe special-purpose i.cs that simplify inter-facing 68000s to VMEbus in the final sectionof this series.

Alan Clements BSc.. PhD.. is a Reader in theSchool of Information Engineering at Tees-side Polytechnic.

Conferencesand Exhibitions

8-12 November, Munich: Electronica 88, 13th

international trade fair for components andassemblies in electronics. Associated conferencesinclude sessions on quality assurance, componenttesting techniques and sensor elements for mass -market products. Organizer: Miinchener Messe-und Ausstellungsgesellschaft mbl I. Messe-gelande, Postfach 12 1009. 0-8000 Munchen 12.te1.189) 51 07-0.

22-24 November, IEE, Savoy Place. London:fourth international conference on electrical safe-ty in hazardous areas. An accompanying exhibi-tion is planned. Details from the conferencesecretariat. IEE Conference Services, 01-240 1871ext. 222.

25-27 November. Windermere: Institute ofAcoustics 1988 autumn conference on noise in

and abound buildings. Institute of Acoustics. 25Chambers Street. Edinburgh Eli1 1HU.

20-22 December, University of Nottingham: In-stitut.. of Physics annual solid-state physics con-ference. Details from the loP in London. 01-2356111.

23-25 January, 1989, Bahrain: Middle East Elec-tronic Communications Show and Conference.incorporating the Middle East Commputer Show.Contact: Overseas Exhibition Services Ltd inLondon. 01-4861951.

23-26 January. 1989, Bahrain: Middle East Elec-tronic Communications Show and Conference.incorporating the Middle East Computer Show.Contact: Overseas Exhibition Services Ltd in

London, 01-4861951.

25-26 January. 1989. Fleming Park. Eastleigh:Instrumentation Southampton. Details from Tri-dent International Exhibitions Ltd in Tavistock.0822-614671. Further dates in 1989 for exhibi-tions in this series are 22-23 February at theHarrogate Exhibition Centre. 21-22 March atFilton. Bristol. and 14-15 June at The Forum.Livingston.

7-9 February, 1989. Wembley. London: surfacemount and related technologies exhibition andconference. Smartex. Details from MGB Exhibi-tion Ltd. 01-302 8585. Prospective authors ofconference papers should submit a 200 -wordsynopsis to the chairman of the technical commit-tee. Smart Group. 3 Lattimore Road. Wheathamp-stead. I lertfordshire ALA 8OF.

20-23 February. 1989, Genoa. Italy: EuropeanAdvanced Manufacturing Systems Exhibition andConference. Cahners Exhibitions. 01-891 5051.

21-24 February, 1989, Birmingham NEC: WhichComputer? Show. Cahners Exhibitions. 01-8915051.

6-9 March, 1989, Swiss Federal Institute ofTechnology. Zurich: 8th international Zurichsymposium and technical exhibition on electro-magnetic compatibility. Details from EMC Sym-posium Zurich. ETII Zentrum - KT. 8092 Zurich:teL (Switzerland) 01-256 2790/2788.


A dimensional approach toa unified theory

It is possible to reduce the conventionaldimensional expressions, in terms oflength, time and mass (inertia) or force, to

terms containing length and time only. Someadvantages accrue and tabulating the result-ing expressions in the way shown in Table 1clarifies relationships previously obscured bythe complexity of a three-dimensional format.

The chart does not show all the manyscientific concepts now in use, since it isintended primarily to be introductory andindicative: but it will. I hope, interest thosefamiliar with the dimensional techniques nowincreasingly being used.

It is obvious that all the non -dimensionalparameters and numbers, such as Reynolds.Froude, Densitometric Froude, Richardson.etc., are merely ratios of the values tabulatedand that they are finite in number, forpractical purposes. Although the tabulationcan extend to infinity, the concepts suscept-ible to measurement are confined to the firstsix powers on each axis.

What is perhaps not so obvious is that thetable also shows the laws or rules of Natureand their relationships as interlinked byNewtonian calculus. One can make severaldeductions from the tabular array, but it isenough to point out here that it might havebeen preferable for scientists to use a naturalunit of measurement based on Napierian e, ITand a representative velocity such as c or,more accurately, the constant for the ratio oflength to time for the particular case; and ingeneral in the study and measurement ofnatural phenomena.

The Michelson-Morley experiment, andsubsequent endeavours to establish the velo-city of light and other forms of electromagne-tic radiation, are open to another interpreta-tion - that of establishing a ratio of length totime in this universe. In this paper themeaning of these experiments is to be taken asestablishing the connection of the dimensionlength to that of time in an absolute sense,instead of its being measured in units familiarto the scientist. Those units are arbitrary,despite the attempts of Wren and the AbbePittard in 1670 AD to relate the unit of lengthto the unit of time by the beating of apendulum in a gravitational field. That theseexperiments, such as the Michelson-Morley,have also established that the value soobtained is a physical constant of the universeis accepted.

Dimensional relationships of commonphysical concepts.

S.K. CHATTERJEE, M.A.

The expressions for Newton's Law P = mfand Einstein's Law E = mc2 can be expressedas P = m1114 and E = mL2/T2 and, substitut-ing m = L3/T2, we obtain for these two laws

and

n L

T-3P = X

L=

L'T- T

E.G = A

T- T- I

The expression for m = L3/T2 is derived byClerk Maxwell' as follows. If (as in the metricsystem2) the unit of mass is defined by itsattractive power, the acceleration of a mass mat a distance r is by Newton's Law m/r2.Suppose this attraction to act for a very smallinterval of time t on a body originally at rest,and to cause it to describe an interval of spaces, then by the formula of Galileo

2 rI m

t.2

srtra TOMMY,.

1111Cyle011011 TOLIA111

:ATTIC canna,

lataLarran ear=L. ranarrwrn

mammaVOLUM

IRIOCITTI I111TONSIMLITT

TLOTTIRICAL CallTOCITTITT

TNITETTON I

WOGS TYCTITOOTTRACT

110,011, TOTIPLTION

ULF MOM=

I

11111111110

T.

SPESTFIC VOLLTIT

TITS 141.11(0araterrnrr

COMITITTS1 LTTV WON TOMO6161111102

Sin MI

L. I L T

VOLANI. MT TANA

IMKT= FULD

TES'M 1

YOOLILII ATTSLTIOATIONMUT 011.1111/01TT ilegATISVOLUM 11.11811TMIKA =MTV MB/

0 0 L. 0 L. 0 L. T.

L. 0 L. 0 0 0 L. 0

0 0 L. 0 L L. T.


whence9f s .m - , I.e. M = -r.

T -

that is, volume (L') and density (1/12) con-jointly, which is in line with Newton's defini-tion for mass ("Quantitas materiae est mesoraejusdam orta ex ilius densitae et magnitudineconjunctem") in his Proposition I in ref.' (seealso p.5 in ref)). In practical unitsHeisenberg4 confirms this when he takes themeasured values of Planck's constant and thevelocity of light, assumes a length of roughly10-13cm. and forms an expression which in itsdimension corresponds to a mass, its valuebeing of the order of magnitude of the massesof the elementary particles. In dimensionalterms this becomes

velocity of lightPlanck's Constantlength

whence

cm/serg/scm

IltxLI=f

L/TL 'ITL

In these circumstances where mass has thedimensions of L'/T' the Gravitational Con-stant G is non -dimensional. since':

but

P=G -Gmm x

=c11::11L2=6 L!1 V / 1 T'

hence G = I (non -dimensional). Alternative -L ly. GM = cm3/s2. but M = m = L VV. so again

= I .

T' 1 T- / 1 T-

Although Heisenberg believed at least threefundamental constants were necessary todetermine the masses and the other prop-erties of elementary particles (e.g. atomicnuclei of the lighter elements) it is onlynecessary. as will be shown later, to have two-

T.

MCNMOMAL

ACMILIDATIOO

rinnrrir viSODorreOtennrrireCISCOLOTION

T noon. vir.OrTriVORT:CtTlMDCMCNCT

10110DTN.

MTPOITTIRENNAL COPYL111.DOLTTIONN 6 COORTJUIT

TONfonason =WIWI

MNVIIIONTATONIC

name rieninarTre. Of Intern

fLNCTRIC MTNITSITYFelarnnOlr.inaTtwoucTue=

Mmmc ruarOLO Mann

a..comer CONDUCTITIT,MONITTC INDUCTIONNACMITIC ODIN IITIMMTINI

MY STUMM!leallirrre me

Olanerie ROWnnenetarrinfnefe

CannlailefirrtTereionne°Men,' or naCTUCTIT

tcrorno oterucerrereiticrolc rifOocelomrowan money

Ottnillien COIrrirenerneem .01.11.rreIW MIOILITT

elerairre

IT 0

Tr Lt T. 0

L

ireneelan, Morenonoonlocenoternina Ibl

TfLOMITMONITIC MOOTTa

NOLDDOLDNININTIOLT T.7

MR.IMMNIMMr TOLL* C NOMINT

MCOONISATINO OOMPTICIMItNMOTI

yl° =TM OATINT ° I

--f-111:111TCP.MIDOMM Or INSTOTULLAND MOOONNTIoN 10COMM TOLTO CoMMOIMILTMC MTIOMTV OMM

MN MR MOW I

N -TM 011001/M

NIMOBTICIMINT

Tjt

Om VINT Of MOIL

CLIOONT ONONINTV 101

a.

nomneram anMonne mean tan Itileilite IND.IMMO orneerenn MOM ONO4 MMnoneno renal anan maMOMne ma nmoncOanOrte Onforree

MINNS 1TINOITTNIOtOrIT NoONLENTMASTIC NONNI=TOACTITIMINOCM/UNIT LIMN

MOTO.,Mina TOMB.ONCOMMON COMM,

.1jmO, MMOMMeo mo.

MOIR.

MOM

nworscomunNome or anon=WWI OMMTMACTION

MEM

T L

TI Le

JJ

Mechanical

Diffusivity L2/T

AccelerationL/T2

L3/T2

Dynamic viscosity L2/T3

Magnetic fieldL/T2

Electromagnetic

L2/ T

Electric flux

///'' \

Magnetic p.d.L2/T2

.."///

Electric fieldL2/T3

Thermal

Thermal diffusivity

Temperature

gradient

Magnetic fluxL3/T2

Entropy

Thermal conductivity

Electric field

Electrostatic

Magnetic flux

Electrostatic Electric flux

P.

Magnetic field

ELECTRONICS & WIRELESS WORLD883

namely length and time, and their ratio.So far I have dealt with the dimensions of

mechanical quantities. and it must be stressedthat only those properties susceptible tomeasurement are being investigated. Othermeanings of the word dimension are notapplicable to this enquiry.

The dimensions of entropy, the heat weightof Zeuner (defined as i)(,)/01'1. are derived fromthe equation (dQ/Blx = work done orenergy, where itC) is the quantity of heat.whereas those of temperature t) are derivedfrom PV = mRt) where P is pressure, Vvolume, and R is the Universal Gas Constant (apure number). Thus using P = IL '/T1) x UV,V = L3, m = 1,3/T2. R = 1. gives = L2/T2, andentropy = LIT2. These are then substitutedin the dimensions for thermodynamic con-cepts.

Since electrostatic permittivity p. andmagnetic permeability E are essential lyof thesame kind or quality. and p. E = 1/C-,, it isreasonable to substitute p. = 1/c and E = 1/c inthe electrical entities.

There remains one more step before I canestablish that Table 1 shows the interconnec-tions of the entities by means of the Calculusof Newton. Consider the volume of a sphere4/377 r3, and its surface area 47r r2. One can see.by extending these two expressions backwardsand forwards that we have an infinite sequ-ence

4 1

3titr 8;Tr Orr'

3Wr

or

8r 8rr 87r2 Kw,' 87r' 8zr"1 1 1.2 1.2.3 1.2.3.4 n!

of terms found in the expansion of theexponential expression 8 r er or 8 r exp(r), i.e.8 r exp (LI. Thus, on the horizontal axis inTable 1, there is shown

L L3 L1

as a shorthand for the terms 8 w L"/n!n= 12.... + xOn the vertical axis of Table 1 there is similarlyshown

1 1 1 1 1

T T2 TI TxThe inverses of these L and T terms are also

shown and yield Table 1 divided into quad-rants indexed by L". 1/L". I. 1/T". Into thistable can be inserted all the modified dimen-sional expressions in terms of L, T, and theirratio c, and the permutations and combina-tions of these expressions which for mostpractical entities seem to be confined to thefirst six or eight powers.

An additional analogy which bears on theinterpretation of the tabulations is the Col-umn Analogy of Professor Hardy Cross' whereit is the mathematical identity between themoments produced by continuity in a beam.bent or arch, and the fibre stress in aneccentrically loaded short column: this hasthe dimension L/T1.

Similarly, the equations of Poisson andLaplace have applications in several branchesof physics due to their dimensional identity ineach of the disciplines. They are equations ofenergy and have dimensionality L5/T1.

Thus the laws (e.g. Newton's) or rules (e.g.Column Analogy) are shown in generalized

dimensional form in Table 1 - which is to heunderstood as one entity embracing all otherswhich are manifested as Hilbert geometries inspace and time.

From the time of Newton, for whom theBook of Nature was written not in mathema-tical symbols but in the form of a cryptogram,and of Boole, who believed that dynamical andother actions seemed not only to be measur-able in themselves but also to be connected toeach other, even to the extent of mutualconvertibility, it is perhaps only now that it ispossible to display - through tabulation interms of L, T and c - the fundamentallyabstract nature of the entities of science in theform of Hilbert spaces of many dimensions. Asa consequence, one obtains a consistenttheory applicable to the very large and the verysmall. uniting the many concepts susceptibleto measurement by a dimensional approach toa unified theory.

Omitted so far are Planck's large numbersfor L, T and M, which relate all three entitiesthrough Planck's Constant. the velocity oflight and the gravitational constant. innumerical terms measured in the c.g.s. sys-tem. as a further confirmation.

There are. I believe, a number of otherexpressions which are relevant to the under-standing of a unified theory based on dimen-sions. and the ones which spring to mindmost readily arise in Heaviside's ImpulseTheory and his fascination with the specialcase of Euler's Integral:

-f -6!e

=e'[ X6 X5 X X

6!+

5!+-+-+

4! 3!+ X + I

2!

Heaviside's final (unpublished) researches"led him to formulate a unified field theory inwhich electromagnetism is co -related withmass properties and in which there is a

reciprocal relationship between radiation andmatter. This is a somewhat similar idea toProfessor Jennison's model of an electron asa phase -locked cavity: mass equals a volume(L3) and two frequencies 1/T,etic and 1/

Of equal interest is Euler's GammaIntegral and its relation to the LogarithmicSpiral, the Golden Section and the FibonacciSeries.

There is one more relationship which ispertinent to this discussion, and which is asfollows:

G = 1,NEWTON

G = K = kc2/87,EINSTEIN'"

and hence

k= 1 /c2 =CLERK MAXWELL

and is the result of G being 1. i.e. non -dimensional. This indicates that instead ofwriting E = mc2 one should now write

E= m.

As 11 depends on the physical characteristicsof the media, it is more accurate for experi-ments carried out under terrestrial conditionsto calculate the value of the masses of theelementary particles using these values ratherthan the constant c, that is, m = E I.LE ratherthan m = E/c2.

Table 1 of the unified theory extends to plusand minus infinity. It still has many gaps suchas the entities of chemical, biological andsocial sciences. However, provided that thesecan be expressed in terms of dimensionalequations including, if unbalanced, one ormore constants, the values of which aredetermined by dividing one side into theother, it will be found that these constantseither turn out to be non -dimensional num-bers or to have dimensions in terms of L, T orc, which are then substituted into the originalequations to give a dimensionallyhomogeneous expression. In many problemsthe time element (1/1") can be disregarded,but it becomes significant in studies of thevery large (cosmology) and the very small(nuclear physics), and in problems involvingfrequency.

I now express my great appreciation to DrJohn Amson and Dr Christine Crow for itshelp, to St Andrews University for the use ofits libraries, and to the members of ANPAfor their courteous and critical appreciation ofthe views now made public.

Symbols useddimen-sional variables standardM Mass m mass 0 temperatureL Length r,s length µ magnetic

permeabilityT Time t time E electrical

permittivityf acceleration G.K Universal Gravit-

ation Constantc.v velocity k Einstein

constantQ heat R Universal Gas

ConstantP pressureV volume

References

1. J. Clerk Maxwell. A Treatise on Electricity andMagnetism. Vol. I. lst.Ed. Clarendon Press. Oxford,1873. lArt.5.61. Vol. 1. 3rd.Ed.Reprinted. Dover.1954. [Art.5.61.2. Anon. Going Metric. HMSO 1967.3. Isaac Newton. Philosophiae Naturalis PrincipiaMathematica . Third edition. London 1726.IDefn.l. ProP.1I4. W. Heisenberg. Physics and Philosophy. Geo.Al-len & Unwin, 1958.5. G.W.C. Kay & T.FI. Lahy. Tables of Physical andChemical Constants. Longman, 13th Ed. 1966[pages 15.1616. J.A. Ewing. The Steam and other Ileat Engines.C.U.P. 1922. [page 12217. Hardy Cross. 'The Column Analogy'. BuL215.Eng. Exp. Stn. Univ.of Illinois, 1930.8. H.J. Josephs. 'Some unpublished notes of OliverHeaviside'. Heaviside Centenary Volume, I.E.E.1950.9. R.C. Jennison. What is an Electron?', WirelessWorld. June 1979. [page 471.10. A. Einstein. The Meaning of Relativity. Methuen& Co. 1950. [page 861


APPENDIX A - PROPOSITIONS1: TI IAT the ratio measured as the velocity of light. c, can be interpreted as a ratio of dimensions I.Length and T = Time in the study of natural phenomena, and that 'c' is TI IE basic physical constant.2: THAT the dimensions of physical entities can he expressed in terms of a representative length Land/or a representative time T.3: TI IAT these concepts so expressed can he tabulated as shown in Table I.4: TI IAT the tabulation indicates that the entities are connected by a calculus (of Newton) and can hesplit to form entities of a lower order: for example:

1.1 L'` L L I- = x or -x -T T- T- I T

5: THAT each concept so defined exists in a space-time of a multi -dimensional geometry consisting ofscalars, vectors, quarternions (Hilbert spaces, in general). For example: (a) mass exists in a spaceconsisting of three dimensions of length and two dimensions of time (or frequency = I/T):(b) the quantum of action LIT.' exists in a space with five dimensions of length and three of time.being the time integral of energy f E dt.6: TI IAT each of the geometries is conservative. e.g. exhibits conservation of mass. or of power. or ofenergy: hut that interchange occurs in physical systems.7: TI IAT Table 1 provides a metric for an aether in which all the entities operate.8: TI IAT throughout the tabulation there are time and length integrals and differentials connectingthe concepts, and that these operate in general from minus infinity to plus infinity.9: TI IAT is c is a constant. then LIT is constant. (Thus, if L increases. then T must increase to preservethe constant ratio. When T increases. la decreases in numerical terms. Therefore the greater thedistance over which L is measured. (i.e. frequency) diminishes. Thus there is a shift to the red inthe observed spectrum which cannot he equated to a velocity of recession as in the Doppler Effect).

APPENDIX B - SIMILAR OPERATIONS

Operation = Integration Differentiation

dxxT

Mathematical L io

Electrical * V = --1- V,.dt-CR

V - CR

Mechanical Babhage 13abhage

Unified Theory xLor xT x I/T or x

N.B.I.- = C and x - L or T

* Equivalent electrical circuits

Integration

Vo

TVs

Differentiation

C

Vs R Vc,

Examples

fP dl. = 11T' x I. = - energy or work done.'

L' I L'P - _ -x - = . -impulse or01, T' L T' momentum.

1E dI - L ' L 'x T - Planck's constant.T''

L' 1 L'E x -

T- T. -= power.

N.B. The quantities on the right -hand -sides areconstant or conservative in any closed system.

The equation of conservation of mass (sometimescalled the convection -diffusion equation)may be written as

m 1H2- - [tIX (IX

la) ( ) (13) (y)

where

A ...area: C = L = V...velocity:

m = -L ...mass: I)_diffusion

Then:

lo -L L 1 L

x = m I (

I T T T-

(13) -

- L'< -

I Tx x

I T

1-y1 - WC )lx

x xL 1 T T

, ac aUT 17 n.b. - = - -ax a L T

Ill

If the substance is decaying at the rate K.proportional to its concentration. then KAC isadded to the right-hand of the equation. i.e.

K L .

KAC - -1 T

x-x-- IT2 .e.K

If there are two (lows in opposite directions then

= LT, ±()

where is the main flow. and A is thecross-section under review. The correspondingequation for conservation of volume. i.e. L'. is

then;1

;it ax

i.e. constant diffusion/concentrationL -

(T)/

Novel power supplyneeds sponsorshipBy switching capacitors using mosfets 't ispossible to build a power supply that is'isolated' from the mains. yet does not need atransformer. Transferring charges viaswitches is not a new method of providingisolation - take the flying capacitor forexample - but applying the approach topower -supply design probably is new.

Aubrey Sandman. who has applied for apatent for the charge -transfer supply, saysthat switching -capacitors will simplify p.s.u.design and result in lighter products. And, ifhis assumptions are correct. the theorycould also result in efficiencies approachingthose obtained from switch -mode supplies.

t' ut without the same r.f.i. problems., Mains 'isolation' provided by the circuit is

, of what most people would envisage asisolation, hence the inverted commas. Byisolation, Aubrey means that the positiveand negative rails of the supply are alwaysisolated from the mains by a power mosfet inits off state. As a result, either the positive ornegative sides of the supply could potentiallybe connected to earth without causing cur-rent in the neutral line.

/ W

He argues that by including protectioncircuits, such as crowbars. the switching -capacitor supply could be as safe as aconventional isolating supply. We pointedout that BS has difficulties with the opto.isolator, which puts a piece of plastic be-tween the mains and the user, so it wouldcertainly find a semiconductor isolator hardto accept. Remaining optimistic. Aubrey said"I'll have to have a chat with BS".

Although Aubrey has a working circuit, heWill has one or two problems to sort out. For

xample, to get the best results the powermosfet pairs need to operate as close aspossible to their safe -operating limitsthroughout the entire capacitor -chargingcycle, providing a few milliamps while theirdrain -source voltages are high and a fewamps when their capacitor is almost fully

;aged. He says that he has this problem!most solved.

Another problem is that d.c. output fromthe circuit is at 300V. On paper. Aubrey'ssolution to this problem looks interesting.but he has yet to work out the practicaldetails. To develop his ideas further, heneeds sponsorship. Is anyone interested? Ifso. telephone 01 608 1282.


PHILIPS PM3217 50MHz dual -trace. duallimebase oscilloscopes complete with carryingcases. handbook and probe C400 each VATAlso in stock PM3212 25MHz £250 PM321425MHz £295 HP17076 75MHz £400 Tek 221560MHz £475 Tek 2235 100Mhz £750

TF1152A/1 RF power meter 0-25W 250MHzTF106613/6 AM/FM signal generator 10-470MHzTF1245/1246 0 -Meter and oscillatorTF201 1 FM signal generator 130-180MHzTF2012 FM signal generator 400-520MHzTF2016/2173 AM/FM signal gen 10KHz- 120MHzTF2162 MF attenuator 0-111db in 0 ldb stepsTF2603 RF MillivoltmeterTF2001 AF signal source/monitored attenuatorTF2300 modulation meter AM/FM to 1GHzTF2300A as above with deviation to 1 5KHz IsdTF2300B modulation meter as aboveTF2356 level oscillator 20MHzTF2501 power meter 0 3W Isd DC-1GHzTF2600 millivoltmeter AF 1mV-300V tsdTF2600B video voltmeter ImV-300V IsdTF2604 electronic multi -meterTF2807A PCM multiplex testerTF1313A Universal Bridge2828A/2829 digital simulatoranalyser2833 digital in -line monitorTF2908 blanking 8 sync mixer60558 signal source 850-2150MHz6460 RF power meter6460/6420 power metermicrowave headTF893A audio power meter 1mW-10W IsdTF99513/5 AM/FM signal generator 0 2-220MHzTF2213A XY crt displayTF2015 AM/FM sig gen 10-520MHz500W 50ohm RF load with power meter2092C noise receiver. many filters available20912092A noise gen'receiver 8 filtersTF2370 110MHz spectrum analyserTF2100 audio oscillator6600A sweep generator 8-12 4GHzTF 1066B AM,FM sig. gen 10-470MHz

£75£350E500C325C325C600£100£250£200£250£350£450£400£150£75

£175£150£400C350

£1500£275£250£750£350£495£75

£250£100£595£100£500£7503.500£125£400£295

AMPLFIERS RF band II 100W outputAVO 8151 LCR universal bridgeBRUEL 8 KJAER 2033 spectrum analyserBIRD 4370 RF wattmeterBIRD Termaline RF loads 500W 50 ohmBIRD Termaline RF loads 150W 50 ohmBIRD 4340 standard thruline wattmeterDAR TINGDON oil -air frequency standardsDOLCH LAM3250 logic analyserJJ Lloyd CR500 G P pen recorderLYONS PG73N Bipolar pulse generatorDRANETZ 305 digital phase -meter 2Hz-700KHz £400KEITHLEY DMMs 178 8 179 £125. 0145FEEDBACK SFG606 sweep function generator £200MOTOROLA HDS200 hardware development stationPHILIPS PM2554 audio millivoltmeterPHILIPS PM5590 TV IF ModulatorPHILIPS PM5597 VHF TV ModulatorsPHILIPS PM1590 1mHz 2mHz synth function gen.PHILIPS PM8235 multipoint pec recorderSINGER 40MHz spectrum analyserSTOLTZ A G prom programmer M2 MaestroRACAL 9911 120MHz counters in carrying casesRACAL 9905,04C 200MHz counter timerRACAL 9102 DC-1GHz 30W power meterRACAL 9912 120MHz counterRACAL 9024 600MHz counterRACAL 9083 2 -tone signal source

£500£250

£3000£325£100

£75£200

C' zlist£500£175£295

£125£175

£1500£250£950£495C650C250£125C225

£150£150£150£300

ALL OUR EQUIPMENT IS SOLD IN EXCELLENT. FULLYFUNCTIONAL CONDITION AND GUARANTEED FOR 90 DAYSMAIL ORDERS AND EXPORT ENQUIRIES WELCOMED. PLEASETELEPHONE FOR CARRIAGE QUOTE. ALL INSTRUMENTS AREAVAILABLE EX -STOCK AS AT CCM DATE GOOD DUALITY TESTEQUIPMENT ALWAYS WANTED FOR STOCK PRICES QUOTEDARE SUBJECT TO ADDITIONAL VAT

21GHz Spectrum Analysersystem type 1821 mainframe with 8559A plug-inunit covering 0.01-21GHz

£5950

1122A power unit for let probes11602B transistor fixture355E 8 F attenuators. pair80078 pulse generator8733A pin modulator8600A 8 86632A Sig gen mod section

400F milli -voltmeter529A Logic comparator10529k 10526T Logic troubleshooter4342A 0 -Meter331A distortion meter334A distortion meter5300B/5305B 1300MHz 8 digit counter3518A wave analyser 15Hz-50kHz3400A millivoltmeter3132A1PIP-band attenuator 0-50db415E swr meter4204A decade audio oscillator43188C/478A microwave power meters from6516A power supplies 0-3KV 6mA7046810712 -pen XY plotter high-speed8018A101I serial data generator8672A synthesized sig. gen 2-18GHz5005A signature multi -meter85188551B spectrum analyser 10M-40GHz

50111

EX -STOCK

RACAL 9084 104Mi , , ,,,n GPIB £1500RACAL 9303 digital RF millivoltmeter £950TREND 1 8 data transmission test set £200SYSTRON-DONNER 410 sweep function gen £395TEKTRONIX 2901 time -mark generator £250TEKTRONIX 7511 sampling plug-in C750TEKTRONIX 7012/M2 A/D converter

£C£14329

50WAVETEK 157 synthesizer 100uHz-MHz £3255

WAVETEK 1503 sweeper 450-950MHzWAWAVETEK 2001 sweeper 0-1400MHz

TEXSCAN 9650 0 4-350MHz spectrum analyser £950WAYNE KERR RF Bridge 8601 £100WAYNE KERR 0642 Auto Balance bridge £295VALRADIO inverters 24V DC 230V AC from £75SHIBASOKU envelope delay measuring set 0250MICROWAVE selection of P 0 X Band waveguide slotted linescouplers attenuators etcRHODE 8 SCHWARZ IkW 50 ohm load N -types £250

£195C395£150£495£250

£1000£250£275£295£850£750£950£295

£1500£175£250£295£295£250£250

£1000£1000

£p o a£495£750

£1000£500


(E) EPROM PROGRAMMER

AT LAST! Over 50 Generic Device Types. . . .

.' 1 506 50ms15.276416-2764A

29-874930-8750

43-874444-8051-

1 2516,10ms 17-27128 31-8748H 45-8052'' 6 6166, 18-27128A 32.8749H 46,8044'

19-27256 33-8750H 47-87C5120-2725621V 34-8741 48-63701V21-27512 35-8742 49-637010

250.4 S33,/ls 22-27513 36-8041' 50-63705V4 2758 23-87C64 37-8042' 51-637052

2716 24-87C256 38-8048' 52-63701Y7 2732 25-8755 39-8049' 53-2816A

'2 2732A 10ms 26-8755A 40-8050' 54-2817A1 2732.450ms 27-8355' 41-8751 55-2864A7 2764-50ms 28-8748 42-8752/21V 56-EMULATOH

. . . . at a price to suit any budget!THE MODEL 18 PROM PROGRAMMER* Types include 27C parts: EEPROMs now programmed'* Supports our new EPROM Emulator Automatic Data Rate setting 300-192000 Baud. Two independent Communications Protocols built in. Use with

any host computer with RS232 port and Terminal Emulator.our PROMDRIVER Advanced Features User Interface Package available for all

MS-DOS. PC -DOS and CPM-80 computers.* Fast interactive algorithms automatically selected as appropriate.* Upgradable for future types* Designed, manufactured and supported in the UK.* Comprehensive User Manual.* n b Devices other than 24,28 pin require low cost socket adapter

NEW PRODUCTS!!!!8048,41 Cross assembler for MS-DOS Introductory Offer Price £99.50 VATEPROM EMULATOR 2716 to 27512 £149 80 VAT

li

Write or telephone for further details:ri Fri ELECTRONICS, UNIT 2, PARK ROAD CENTRE,JL MALMESBURY, WILTS SNI6 OBX. Tel: 0666 825146

Used equipment - with 30 days guarantee. Supplied with Manuals This is a very smallsample of stock SAE or telephone for LISTS Please check availability before ordering

Carriage all units £16. VAT to be added to total on Goods and Carriage.

OSCILLOSCOPES1200:02 5..114o.(I 0,.II I'

16t.,15.15=1st,,Is.110

, ,,110

/85/100

L9012251150

11 250150013501600

SPECIAL OFFER AT ONLY £300 EACH

' 15 000[1000

13 500

13000

(1000(120011 000

TRIO OSCILLOSCOPES£ 1 200

08001600

MULTIMETERS150

; :. .

190

165

15 (pip 131, .- . .,

:10 (pip /31

7000 RANGE TEKTRONIX

P0 A

131108 1 KJOER FQUIPMF NT AYA11481F

/900

' -', (15. (P&P 0711650 earn

1 NLRATORS(2000

.. 5011 550145019001600

(1 200

18001600

16001400

L30013001150

160 .., : :

180 '

/260

...... - 11001125,,

L150

,"- - . ,. ,., ,. - , -. 'i , 135 each 1,61.p 17)

ynwionly L351747707

only 145 W4.{/ 17)

NEW EQUIPMENT

/575

/314

'' 1219L2S5

A ,,

19911261175

. .., .:, 11100199

44

,

STEWART OF READINGFA 110 WYKEHAM ROAD,

Tel: 0734 68041Fax: 0734 351696

READING, BERKS RG6 1PL 114Callers welcome 9am to 5.30pm. MON-FRI. (UNTIL 8pm. THURS

ENTER 18 ON REPLY CARD ENTER 37 ON REPLY CARD


Philip Smith's chartHaving described the Smith chart in detail in the first part of

this article in August. JW concludes by giving some examples ofits usage

Last month. I traced the development ofthe Smith chart and showed how it isapplied to the presentation of

transmission -line impedance problems in avisual manner. although it has many uses instability, matching and related problems inother types of circuit. Now, we can have alook at a few examples of the chart inimpedance plotting and stub matching.

EXAMPLES

You should find plotting impedancesstraightforward. For example, the impe-dance z1, of a certain radio aerial might varywith frequency according to Table 1. Animpedance plot crossing the real axis of theSmith chart shows where your resonancesare. In this example, label the resonantfrequency "1" on the aerial impedance plot,Fig.11. We have normalized values again -this time with respect to frequency. If younow connect an inductance with normalizedreactance j1.4 at f = 1, in series with theaerial, the new impedance plot. zi.' 0on the chart, shows resonance qkt40,lowered to about 0.875f. If 05

you draw a maximumtolerable s.w.r.

60

JOULES WATT

circle on the chart, you get the bandwidthover which this aerial will operate for thecondition before you connected the seriesinductor, as well as that after.

The next example shows how a shift inposition along a transmission line changesthe impedance. Suppose you have a zi. of 1.8- j0.5 and you want to know the impedanceat a point 1.1 back along the line. Enter thechart at this point in Fig.12. By the way. theangle from the entered point to either the 0or x position in the direction towards thegenerator defines the electrical length of thetermination. This length gets you to a pointof pure resistance, either the high value(choose x) or the low one (choose 0). I havechosen the open line, so the length is 0.28.The "S = 2" circle upon which z1, sits, showsthe v.s.w.r. on the line. Move round thechart a further 1.1 towards the generator.The impedance locus moves along the S

S LOAD

44Vietfrtiv; 4.:Alltl! 1&.: Al#0 4 0

circle to position z,. Read off the impedanceas 0.72 - j0.53 which, on a 5011 system,would be (36 - j26.5) ohms.

Now have a go at stub matching. A singleshort-circuited transmission -line stub pos-sesses two variables - its length and whereyou have connected it across the line. Bothparts of the admittance. the real and theimaginary, come under control because ofthis. In particular, you can adjust the realpart to 1 and eliminate the imaginary partaltogether. Across the line is the operativeword. It indicates that we should work interms of admittance, and that is how thechart should be interpreted now. I have leftthis example as the general case; see Fig.13.Enter the chart at y!.. Immediately, you havethe electrical length of y,, and the v.s.w.r. onthe line, if you want them. As you move backalong the line towards the generator. even-tually you will strike the circumference ofthe g = 1 circle. At this point yi = 1 + jb. Ifyou connect the stub at this point, arrangingfor it to contribute a susceptance of - jb.then the line sees a perfect match at this

junction and the job is done. Youcan easily see this form,

= gi. + jb!.y! = 1 + j 1Y. =y.. = yi + >7 =

(matched)

O.

Si0-4"

8 -o

0

0

The Smith chart as it is familiarly seen. Youcan buy books of them from the engineer-ing graphical suppliers


Table 1. A typical list of impedance valueswith frequency for a microwave aerial.

I t

0 8 080 116 11 1: 0.80 JO 480 9 090 11.26 0.90 10.261 0 120 100 1140 1 20 10 401 1 150 11 54 1.50 4. 12 541 2 220 117 1168 220 13 38

1 1 GHz L - 11.2 nH

You will not find sliding a stuh up and downcoaxial transmission lines very easy. A stan-dard method to overcome this problem usestwo variable short-circuit stubs at fixedpositions on the line, as Fig.14 shows.

A number of people beginning the use ofthe Smith chart find the treatment of thisdouble stub difficult to visualize (geometryagain?). hut following it through the illus-trates the usefulness of the chart very well.Moving towards the generator through thedistance di transforms the admittance of theload. yi., to some value yi across 1 - 1: seeFig.14 inset. Stub No 1 adjusts yi so that, at zifurther distance 1.2 hack along the line, anadmittance value y., appears across 2 -2 witha normalized real part equal to 1. Thereforeat 2 -2.y.. = 1 + jb., so that stub No 2 has theduty of adding susceptance - jb, there. sotuning out the + jh2 to leave a matched line.

We do not yet know the value of b2, butwhatever it is, y., at 2 -2 sits somewhere onthe "g = 1" circle of the chart. To arriveforward at position 1 - 1. the whole of the "1circle has to move round the chart towardthe load through the distance d., inwavelengths, carrying all its circumferentialpoints - including the unknown point re-levant to the problem. At this stage you cansee how much susceptance stub 1 has to addacross I - 1 to move the admittance yi alonga constant conductance circle to get on tothe shifted "g = 1" circle. This move arrivesat and identifies the relevant point. (Theremight be two points. How would you inter-pret the other one? On the other hand,suppose the move along the constant con-ductance circle failed to intersect the shifted"1" circle at all. How would you handle that?)

Notionally, you now journey back alongthe transmission line to position 2 - 2carrying the known point back to the normal"g = 1" circle position. Stub 2 tunes out theremaining susceptance, as already men-tioned, and this final move takes the pointinto the centre of the chart. In other words,you have matched the line.

The Smith chart, then, after a littlepractice, answers questions about impe-dance very quickly and with engineeringaccuracy. I hope you now agree that it formsone of those 'analogue' applications thatcome through the 'digital' revolution un-diminished - in fact enhanced, as its regularappearance on computer v.d.us testifies.

AERIAL

.J025

5

+J05 .J20

.J30

fik4atltelaJ5 0

!V04is _J50*op 130-10 25

1111-j1°5

J2JO 5111.

-1

Fig.11. The Smith chart allows the locus plots of impedances and admittances overwide bandwidths: the effect of changes can be instantly observed (especially withautomatic plotting equipment see last month's bibliography).

ITowardsgenerator

-J0 25

z=

Zin

0125+1

0

111:r.01240:71i0t<.)

.J5 0

j2

041dhlr4Zin

4111111#4015 °

j05

z=-1653/1110_5 '311A

Fig.12. Finding the whole range of impedances anywhere back along a transmis-sion line (in other words, all the points on the circumference of the standing wavecircle). as shown here, points up the value placed on the original use of the chart.


.j0 25

Towardsgenerator

-j0 25

%A 0.15X

Towardsgenerator

0

-j0.25

Wandt- movements-j05

along the line'Band if susceptoncesshunted at 1-1 and 2-2by the stubs

shifted

Fig.14. The double -stub tunerthrough. The chart helps greatly. Follow the argument in the text and you shouldsee what is happening.

Fig.13. The effect of the single shunt stubplaced across a transmission line forms aclassic first example of the use of theSmith chart

d wavelengths

-12

.j 3 0

.j50

1s.

0

-j3-0

Y- 00

needs a surprising amount of thinking to see it

Seminars, trainingcourses20-23 September. University of SouthamptonMicroelectronics Centre. Practical course on cir-cuits and systems in silicon. Contact the confer-ence office at Sira Ltd. 01-647 2636.

18 October, Chislehurst. Kent. Tutorial on adap-tive and self -tuning controllers: principles andlimitations of conventional p.I.d. loop controllers.Sira Ltd. 01-647 2636.

27-28 September. University of Essex. Colches-ter: STEbus applications seminar and trade ex-hibition. organized by the STE Manufacturers andUsers Group. Topics include networking, multi-proce:.sing. rom-based target systems, data-comms. system development using intelligentslaves. Details from STEM on 07357-4976 (fax51551 or at P.O. Box 149. Reading. Berkshire RG631113.

Among the summer events organized by theInstitute of Electrical Engineers are vacationschools on switching and signalling in telecom-munications networks 14-9 September. Birming-ham): measurement technology. d.c. to v.h.f.(4-10 September. Birmingham): telecommunica-tion network design and performance evaluation(11-16 September. Glasgow): electromedicaltechnology 118-24 September, Canterbury). Aschool on plasma processing of semiconductormaterials is being arranged for December. Detailsfrom the IEE in London. 01-240 1871.

A series of one -day workshops on p.c.b. designtools is being staged at Cambridge by Iii-Tek CAE.Rate; this year are 22 September. 21) October. 24November and 15 December. Fee is £200. Iii-TekCAE. Dillon Walk. Cambridge CBS 8QD. tel.IC23-213333.

Commercial courses covering a wide range ofsubjects including artificial intelligence. super-computers. digital processing. communicationsand technology management are organized byIntegrated Computer Systems. Fee for each two-day course is £250. Details from ICS PublishingCompany on 0372-379211.

European Silicon Structures offers courses oncustom silicon design at is own premises or thecustomer's. Details from ES?. at Bracknell on041-525252.

Force Computers. which specializes in VMEbusproducts, has produced a brochure describing itstraining courses and seminars. Contact ForceComputers (UK) Ltd. 1 !lolly Court. 3 Tring Road.Wendover. Buckinghamshire 111)22 (iPE. tel. 0296625156.

EVENTS14-16 March. 1989. Birmingham NEC: Internep-oil'. Semiconductor International. Special topicsare hybrid manufacture. p.c.b. sub -contracting..ind custom silicon. Both exhibitions will heaccompanied by conferences. Details fromCaliners Exhibitions in Twickenham on 01-8915051.

16.17 March, 1989. Olympia. London: Cable &Satellite 89 exhibition and conference. The ex-hibition continues fm a further two days duringwhich it is open to the general public. Details fromMonthuild Ltd. 01-486 1951.

17-19 March, 1989, Brighton: Defence Oceanol-ogy International 89. embracing technology fromsub -sea to satellite. Spearhead Exhibitions, 01-5 la 5831.

5 -ti April. 1989. London: fifth InternationalLogistics Conference, organized by the Society ofI.gistics Engineers. Details from Consert. 57-61Newington Causeway. London SE1 6131.. tel. 01-

!32351.


60 MHz Universal Oscilloscope2 Channels, max 1 mV div.. Delay Line Component Tester.

Timebase- 2.5s-5ns incl. x 10. Sw a DeloTriggering DC to 80MHz,e

The HM 604 is a new innovative general purpose oscilloscopesatisfying a wide range of exacting requirements inlaboratory, production. and service. The dual -channelmeasurement amplifier can be readily checked on the built-infast-risetime 1MHz Calibrator from probe tip to CRT screen.

Another important feature is the internal delay line forobservations of the leading edge of a signal. As in dual -timebase oscilloscopes. the HM 604 features a calibrated sweepdelay mode, allowing smallest waveform sections to beexpanded up to 1000 times.

The HM 604's most outstanding feature is the automaticAfter -Delay Trigger mode to nsure extremely stable displaysand jitter -free measurements of asynchronous signal sectionsand bursts or pulse trains, independent of amplitudefluctuations. An active TV -Sync -Separator further enhancestrigger quality and in the alternate trigger mode, two signalsof different frequencies can be compared.

At only £575 plus VAT the HM 604 from HAMEG again sets anew price performance standard representing decades ofsuccessful design experience!

another winning scope from Hameg.HAMEGrot' 0105e who COINIpare

All Homeg Oscilloscopes are suppled with 2 probes rooms loud andyours wall amity on in sir unlentt

70-78 Collingdon Street, Luton, Bedfordshire LU1 1RXTelephone (0582) 413174 Telex 825484 Fax (0582)456416


TIME AND FREQUENCYThe Company is Registered to Def-Stan 05-2; (AQAP- 1

RADIO CLOCKS* Synchronisation of remote sites.* Time Stamping GMT/BST.

CHRONOMETERS * Quartz master/slave systems.

STANDARD FREQUENCY Accurate off -air standard (MSF Rugby).* Calibration and reference for timers, counters,

RECEIVERS frequency meters.

TIME CODE Generators, Readers with high speed tape searchand control. Timecodes IRIG A, B, vela, EBU,

INSTRUMENTATION NASA, XR3.

Analogue, digital and self -setting analogue types

DISPLAYS (desk, wall or console mounting).* Public time displays for airports, bus, railway

8 stations and factories.7 5

A computer network monitoring and management

MIWATCHDOG system for synchronisation and fault reporting of

up to 64 independent computers.

1.1 CONSULTING AND * Feasibility studies and consultancy.* Small quantity manufacturing and test services.

MANUFACTURE * System design.

All the above can be supplied with a wide range of options and interfaces including Airborne, Military andCommercial versions. Customised systems available.

European Electronic Systems Limited,Maldon, Essex CM9 GSW, UK.

Telephone: 024 541 5911 Telex: 995917 EULEC G Fax: 024 541 5785



Step-up switchingregulator

An f.m. switching regulator producing 44V from seven NiCdcells at an efficiency of over 90% for currents of 1.5 to 60mA

any of the portable stimulation con-trollers produced by the MRC Neuro-logical Prosthesis Unit require a

supply of between 25 and 44 V. To save on thenumber of cells in the battery. flywheel -type.step-up switching regulators have been usedfor most of our stimulation controllersdesigned in the last two years.

We found that commercial switching -regulator control i.cs were disappointing -particularly in stability, efficiency, quiescentcurrent and the large number of outlyingcomponents needed. For scarcely any morecomponents a much more predictable andefficient voltage converter may be based onstandard op -amps or comparators. The f.m.switching regulator described here is de-signedto 10 V battery (seven NiCd cells, giving 8.5 Vfor most of the discharge curve) whilstmaintaining an efficiency of over 90% forloads from 1.5 to 60 mA. Useful efficiency180+'/6%1 is maintained down to only 400 µAload -a dynamic range of 150:1. which is dueto the low quiescent current of 368 ILA for8.50 V input. Since very little has beenpublished on switchers for use below 5 Wand as 80% seems generally accepted as"good- efficiency - usually quoted at full loadwith a load -current dynamic range of about5 : 1 - we thought our design might be usefulto others. Output voltage can readily healtered by changing the ratio of a potentialdivider. Best of all, the design is simple andcan be made compactly, most of the compo-nents being available in surface -mount ver-sions.

The circuit shown in Figure 1 consists offour main parts: the parametric feedbackcircuit which sets the output voltage: oscil-lator: the flywheel voltage converter: and alow -battery -voltage cut-off.

TIIE FLYWHEEL CONVERTER

Taking the flywheel converter first, there arethree phases of operation. Initially the in-ductor is charged from the input supply: inthe second phase it is discharged to theoutput: and in the final phase it is essentiallydisconnected.

Charging the inductor. In the first phase,when the mosfet Tr, is turned on, energy ELis stored in the choke. Presuming that thecurrent rises linearly with time (true for aconstant inductance with no series resist-

N.J. CHAFFEY and T.A.PERKINS

R f

1M

0112 gpA

Tri 13 8V2

44V

Rp

3M3

25V

Si

ZTX 5040204

fM9

1N4448

1N4448

1 "I'150p

Rti\,.0.324pA

142k5Do OR1

11DQ06

IRFZ 12

Vo lo

C2 =-=220,u

7V cut off

Rs R71M8 100k

16 60)JA15 3pA C3

62 051M

25V

V, I,

4700

Fig.1. Switching -regulator circuit diagram.

ancel. E1. is given by

El. = L 11,2/2 = V, I, T (1)

Where L is the inductance of the choke. IL isthe peak current in it. 1, is its mean currentaveraged over the whole cycle of period T,and V, is the input supply voltage.

11, can be obtained from:

11, = V, T,/l. = 2 I, (2)

T, being the input pulse width. L is related tothe number of turns N in the inductor

L = A1, N2 (3)

where AL is taken from manufacturer's data.From equation (3). equation (2) may berewritten

11, = V, T,/(Ai. N2) 14)

and hence

Ei. = V,2T,-/2 AL N2.

By definition

V, I, = V11100

where ri is percentage overall efficiency. Vthe output voltage. I the output current.and 1, the mean input current (includingthat of the converter control circuit). At fullload in this design, the control circuitcurrent I, is quite insignificant (about 0.1%of I,,, - see Tables 1 and 2). I, may then betaken as the same as I, and the overallefficiency the same as the converter efficien-cy -rt..

By using equations (1). (5) and (6) andcrudely approximating rl = 100%. a roughcalculation can now he carried out. SupposeV,= 8.5 V. Al. = 250 nil, N = 35. T, = 351.Ls(keeping it simple). V = 44 V and I = 60mA: then E1 = 144 µJ and T, turns out at55p.s. Note designs with T, < I, are notfeasible! For Fig. 2. the situation is T, =


Fig. 2. Operation at 8.50 V,. 43.5 Vo, 727 ohmload : 7l = 93%. (a) oscillator output at5V/div; (b) choke charge current at 1 A/div;(c) Tr2 drain/choke at 20 V/div: (d) Vo at 0.1V/div.

33.5µs, T, = 59.5µs. V = 43.5 V and =

93%. Table 1 summarizes key data for theflywheel converter, along with the requiredformulae. To simplify assembly of the potcore, the adjuster was omitted: a 6% induct-ance reduction can be expected for this.

Saturation. The usual relationship betweenthe voltage across a coil and the rate ofchange of magnetic flux linking the turns ofthat coil may be written for our purpose(bearing in mind the flux is initially zero)

B = V1T,/(AN) (7)

where B is the flux density and A the effectivecross-sectional area of the core. If the pulsewidth Ti exceeds the time T, at which the fluxdensity reaches saturation both magnetichysteresis losses and current increase rapid-ly. Using the 'rough' example quoted above,a peak flux density estimate of 340 m Tesla isobtained, which seems safe compared withthe manufacturer's saturation flux densityvalue of 390 m Tesla (see again Table 1). Thislast figure, however, is for a uniform toroidalcore. For a pot core of non -uniform cross-section, it may be more pertinent to substi-tute the minimum cross-section A,,, (ifknown) for A in equation (7). Where theferrite data is not available, a more conserva-tive limit of 300 m Tesla is probably safe IcfK.L. Smith. 19854).

With V, = 10 V, Fig.3 illustrates the startof saturation where the current changesslope dramatically upwards after 29p.s. some10% before the end of the input pulse.Fortunately the reduction in efficiency from93% to 92% is acceptably small. However, if

NW.

Fig.3. Operation at 10.0 V,. 44.1 Vo, 727 ohmload: = 92% (a), (b), (c), (d) as for Fig.2.

Table 1: choke :14mm x 8mm N48 pot core) and switch (IRFZ12)

PARAMETER SOURCE/FORMULA THEORY SOURCE/FORMULA MEASURED

nominal A, /u,nominal ARN turns 0.375 mm Cu wire

Data Book'Data Book'

250nH/159115 u!)35

L zero bias inductance L AM' 306.25µH no adjuster -6% 284µHR, coil resistance R, AaN2 0.14 1! 0.2 1!I,,, mean coil current 329 mAE, energy per cycle V,272/(2ALNI 144 µ.I 166 u.II, peak coil current

d.c.-biased inductance1/,T,/(ALN2) 0.97A

20 ram

-21:1, T,"V,T, I,

1.17 A244µH

I magnetic path length Data'A cross sectional area Data' 25 mm2A, minimum cross section Data' 19 mm2

B mean flux density u, IL *V,T, (AN) 340 mTesla -V,T, (AN) 325 mTeslaB, maximum flux density Data' 390 mTesla (A,k1) 436 mTesla1,, saturation current

Data Sheet' 50 V

-V ,T L,, 1 19 A

max. Ve, for IRFZ12R2,,,,, for IRFZ12C,,,, output capacitance

Data' (ll 101c,Data' in 8.5Ve.

0.20hrn Typ240 pF

(il 8.51/(47.-2f1) Co

0.2309 pF

VR ripple voltage pk-pk E,N.C IL(e.s.r.) 15 mV ? -50 esr 75 mVR, series res, Tr2 on R5, Rdson Rita 0.5 ohm 0.11.! wiring! 0.61!R,.. series res. Tr2 off R,, R,05, 0.3 ohm

Notes 1.2 and 3: see references.Rough estimate based on: V,- 8.5V. V. 44V.10 60mA. T, - 3511s. nominal At and II 100%. then T, - 55 is

-Conditions: V, 8.5V. V. 43.5V.1 60mA. T, - 33.5us. then - 93% and T, - 59.5us-Based on: V, 10V. Le, 244µH and see Fig. 3 for 29µs:'1 92%

Table 2: oscillator, regulation and low -voltage cut-off

PARAMETER DATA USED & (EQUATION) VALUE

7 time constantV lower trip pointV,, upper trip point

R,C. 142.5 kohm & 150 pF (Fig.1)Fig.1 & (15) 0.661V, V, = 8.5VFig,.1 & (16) 0.920V, V,= 8.5V -

21.38 is5.62 V7.82 V

delay correction for V. ,

V.,' corrected V,,,time from V., to ViiTr, oscillator output low

t 2.5µs. T 21.38irs. V,,, V, V, & (17)V, 0.071V, V,0 V0. V & (18) T 21.38us & (0.3407 2.5µs)

0.071V,0.929V,0.34079.8µs

time from V' to V,,,T, oscillator output highT oscillation periodexpected mark: space ratio

---1111"PHS. 721.38µs. V,. Vi, & (17) Yr.-V, V. Vr, - V, Vol & (18) 7.21.38115 & (1.6427 2.511S) T. + T,T,/T,

0.588V,1.642-r37.6 is47.4 u.s3.84:1

nwastil vd Tmeasured T,measured T,,

Tr, drain disconnected. V, 8.5V Tr, drain disconnected. V, 8.5V

Tr2 drain disconnected. V, -- 8.5V

9.5,s37.7 s

472 ,s

I,,, measured supply currentI, static currentIre free osc. dynam. currentlei dynamic current,10=60mAI, control circuit current

Tr2 drain disconnected. V, 8.5VFig.1:14 15 16 lic Tr, drain disconnected: V, 8.5V. I,,, I.

V. - 43.5V & V, = 8.5V : 40(47.2,59.5)10 - 60mA. V, = 43.5V & V,- 8.5V: I,

373 ir A91 Ok282 irA224µA315 RA

V, set voltage - highV,, set voltage - lowV. measured output voltageV. measured output voltageV,., low battery cut-off

Fig.l. V, 0.6, Si open. h,t 150 & (19)Fig l Vb. = 0.6.51 closed, N.= 150 & (19)V= 44.1V, I = 0 - 60mA 8. V, -7.5 -10.0VV,.,=25.0V.1.=0- 34mA & V, = 7.5 -10.0VFig.1 & (20)

44.1V25.0V42.7 44.7V24.2 24.6V7.0 V

As free -running oscillator. n.b. timing independent of V,

much greater than 10% of the pulse is insaturation, the efficiency falls alarmingly! Asshown in Table 1. equation (7), with T, alsosubstituted for Ti, then yields a maximumflux density for Fig.3 of 436 m Tesla.agreeing with the manufacturer's value towithin 12%.

It should be noted that as the coil currentapproaches the saturation current theinductance is not constant. For IL just shortof 1,0,. Table 1 shows the inductance 1.d, is

some 14% less than that without d.c. bias. Aby-product of this inductance reduction isthat the energy EL is correspondingly in-creased, equation(1) still applying if Ld, issubstituted for L and 2 In, T,/T, for

From equations (2). (4) and 15). it may beseen that to vary N while maintaining thesame power (hence holding I,,, and l con-stant too), the timings T, and T, should bothbe kept proportional to N2 and then equation(7) may be rewritten

B x N/A when T,.sc T, sc N2. (8)

and V, constant.

Thus. if saturation is a problem, the fluxdensity may be reduced for the same powerby lowering N and reducing the timingperiods as N2.

This approach is limited by the speed ofthe control circuitry. Conversely if no reduc-tion in inductance is detected, N may heincreased to take advantage of the reductionin quiescent current possible with slowercontrol circuitry. Alternatively, a smaller potcore may be used in this case.

Discharge phase. When Tr, is turned off, thechoke current continues to flow (i.e. itflywheels), discharging the stored energythrough the only pathway available - via thediode D3 to the output capacitor C2 at theoutput voltage V,,. In this phase of the cycle.


the step-up voltage V, is across the inductorin such a direction as to reduce its current tozero in time Td

Td = Vi TiNsu

where

Vs = V + Vd - Vi

(91

(10)

Vd being the forward bias voltage of the diodeD. For Fig.4 a 0.1 ohm resistor was tempor-arily placed in series with Di and a pair ofdifferential probes connected across it toshow the discharge current. Superimposedon this is the charge current waveformderived from the voltage on the 0.1 ohmresistor in series with Tr.,'s source, giving agood illustration of how closely equivalentare the peaks of the charge and dischargecurrents in the choke and how close to linearare the slopes for these currents. Taking Vdas 0.4 V for a Schottky diode, equations (9)and (10) indicate a Td value for Fig.2 of8.2µs, some 7.7p.s being observed.

Ripple. It may he shown that the peak -to -peak capacitative ripple voltage V, in capaci-tance C may be approximated as

V, EL/V C (111

by presuming that the entire energy E1, isdelivered to C in negligible time, and byequating energies before and after the chokedischarge. At the low operating frequencyhere (< 20 kHz) the action of the capacitor'seffective series inductance le.s.1.1 may beignored. The total ripple VR may then beestimated by adding in the contribution dueto a presumed peak discharge current 11,

through the capacitor's effective series re-sistance (e.s.r.)

VR = ELNC + (e.s.r.). (121

Unfortunately it may not be easy to predictthe e.s.r. value prior to using the compo-nent. As shown in Table 1. an e.s.r. estimateof 50 mil accounts for most of the 75 mV p -pripple seen in Fig.2. The value of C mayseem unnecessarily large, but is required inour case to cope with the heavily pulsatileload of a stimulation controller as well as toreduce e.s.r.

Final phase. The inductor has across itparallel capacitance Cp made up of theoutput capacitance of the mosfet. thereverse capitance of D,,, the self capacitanceof the coil. etc. It is. therefore, a paralleltuned circuit, one end of which is per-manently connected to the input supply V.At the end of the discharge phase, the chokehas no current but still has voltage V, acrossit, kicking it into a damped oscillationcentred on V. Here, as > V,. the firstdownstroke of this ringing is caught atground by the integral diode in Tr, (restor-ing a tiny amount of charge to the inputsupply). Thereafter, ringing proceeds with apeak -to -peak swing of about 2V,. graduallydecaying away. The frequency f of thisringing (537.5 kHz for Fig.2) provides thevalue of Cp by

Cp = 1/47r2f2L) (13)

Here. Table 1 shows that C constitutesthe overwhelming bulk of Cp. The energyneeded to charge Cp with voltage V, is

= Cp Vsu2/2. (14)

For Fig.2, Ecp is 0.2 which is negligiblecompared with the El, value of 166 p..1 butprovides a warning of possible significantloss at higher operating frequencies.

THE OSCILLATOR

The relaxation oscillator in Fig.1 is builtaround one of the two comparators in a TLC3702. This c-mos i.c. typically has quiescentcurrent consumption of 110 per compara-tor and 2.5 is switching delay for 5 mVoverdrive at its input. The push-pull outputswhich swing fully rail -to -rail make it idealfor driving the mosfet Tr2. Ignoring for themoment the switching delay, the capacitorC, should he charged and discharged via R,between a lower trip point V1, and an uppertrip point Vu, given by

V1, = R, R,V,/(R, R2 + + R,12:1)115)

and

Vu, = (R, R2 + Rd WW1 + R, R.,+ R2 R, (16)

The voltage on C, may be obtained from

Vr, = Vr e-1/1. T = R, C, (17)

where Vr, is the voltage across R, at time tand V, is the initial voltage across Ri.Conversely, the time taken to reach voltageVrt is given by:

t = T In (V,.Aird (181

Using equation (17). Table 2 shows howthe trip points are affected by the switchingdelay and hence, using equation (18). howthe high and low output times may beobtained for the free running oscillator.Agreement between the theoretical andmeasured periods is within 0.3

The mark:space ratio for the free -running oscillator was chosen to be about 4 :1, virtually as high as is practicable with thiscircuit. The intention was to maximize theEl, available per cycle (hence making regula-tion easier) and also to minimize therequired for a given power level so as toreduce resistive ( i.e. 12R) losses.

REGULATION

The parametric feedback circuit of Fig.1limits V by turning on Tr', hence forcingthe oscillator and Tr, off. whenever Vexceeds V, which is given by:

V, = (R, + Rb)(V, + + R,, I3/11,,119)

where V, is the zener voltage and V, is thebase bias voltage of Tr,. h11 its current gainand its collector current when turned on.

be

a

I

Fig.4. Choke charge and discharge cur-rents (a) oscillator output/at 5 V'div; (b)choke charge current at 0.5 A div: (c)choke discharge current at 0.5A div.

The 13 term may be ignored in a roughestimate. When the load current 1 dis-charges C below V. Tr, is turned off,allowing a high pulse from the oscillator todeliver energy El. to C after the end of thatpulse. In equilibrium, pulses are permittedjust often enough (f.m. control) to maintainV oscillating around V,. Fig.5 illustratesthe off-load condition where only the inter-nal load of the parametric feedback circuititself applies. For all but about 200 p.,s out ofthe 52 ms final phase of the cycle, V is suchthat Tr, supplies only sufficient currentthrough R, to hold C, just above VH, keepingTr., off as long as necessary. Wasted currentis minimal. a powerful demonstration of theintelligence of negative feedback.

In practice. Tr, does not turn off instantlyas V,, falls below and the residual currentslightly shortens the choke charge pulse T.The speed with which Tr, turns off is directlyrelated to the speed with which V fallsbetween choke discharges. i.e. I,,. Thus itmay be reasonable to expect pulse width todecrease linearly with decreasing l (seeFig.6 for measurements). Although thispulse width modulation causes the incon-venience of making pulse -width predictionmore involved, it does have the considerableadvantage of extending the dynamic range ofthe regulator. Rapid optimization of timingmay be carried out by setting V, to itsminimum value, monitoring V at highsetting off load, then applying the maximum1 and varying R, till the drop in V isacceptable. A check should be made atmaximum V, that the off load V is stillsufficiently restrained and that the core isnot then excessively saturated with max-imum I (see again Fig.3 ). Table 2 shows the

Fig.5. Operation of parametric feedback at8.50 V,, off load: T, = 52 ms. (a) oscillatoroutput at 5 V/div; (b) Ct/lit at 5 V/div. N.B.trip point at 5.6 V; (c) Tr2 drain/choke at 20V/div.


- 96

- 95T

30fqn.y tr tes f tn c t 21s - 94

VI

- 93

2 92

50 60

L.;,r2 (m;.,

Fig.6. Pulse width T, and voltage converter efficiency II, for 8.5 V, to 44 Vu conversion -versus load current.

t.

E so.

Measured

TheoreticalImaximum difference

2%)

r.101.4JA 1mA 10frr.

LOAD CURRENT :

Fig.7. Total efficiency Ti for 8..5 V, to 44 Voconversion -versus load current.

measured range of V for both the 44 V andthe 25 V settings. While regulation at thehigher voltage is merely adequate for ourpurposes, that at the lower voltage is re-latively much better because of the lowerpower required. Fig.6 shows a line for pulsewidth T, versus l for V1 = 8.5 V and the 44 Vsetting, this line being linearly interpolatedfrom measurements at the four points V, =7.5 V and 10.0 V with I = 0 and 60 mA.Agreement between the line and the pointsmeasured at V, = 8.5 V is within 0.2 !Ls.

LOW VOLTAGE CUT-OUT

The fourth section of the regulator is almostself explanatory. When V, is below Ku,. thecomparator output is high and, being wire-Ored via a diode to the same input to theoscillator as for the parametric feedback,forces the oscillator off. V,, is given by

V, = 2.5 (R5 + Ro)/R6. (20)

As well as protecting a flat battery fromlarge currents, this circuit also ensures that.at power on. Tro is held off until V, is highenough to ensure proper operation of theregulator. Consequently. the circuit may beswitched on at the 44 V setting with max-imum load with equanimity (power on sett-ling time is then 140 ms).

LOSS ANALYSIS

Presuming that the choke current I in-creases linearly on charge, the energy ER,,dissipated in the total resistance R,,,n inseries with the inductance is given by

T,

Elton = R" dt =1L' Rsn T,/3. (21)

Hence from equations (1) and (21) the

percentage PRn of the input energy EL lostin RSrin is

Piton = 200 R,n .173L.

the oscillator being just held off by Tr,. Inthis circumstance, the internal load current1, drawn from V by the parametric feedbackcircuit is

= + 13 = + Rh) + V/R,. (28)

The consequent input current In neces-sary to maintain I,,, is:

= V1,/VoiC (29)

which for the above conditions comes to 265

The remainder of the control circuitry inthese circumstances consumes a static cur-rent L of 91 µA (see Table 2). The totalquiescent current In can then be estimatedfrom

(22) l = I, + 10 130)

During the choke discharge, Tr2 does notcontribute to the series resistance. so a lowervalue Rsidr applies. Thus from equations (9).(10) and (22) and presuming PR0 is small,the percentage PRo. of EL lost in FLA- is, to afirst approximation

--- 200 R,01; T; Vi/I3L (V + Vd VA.(23)

The percentage Pd lost in the flywheel diodeD may be written as

Pd = 100 Vd/V (241

Assuming all these losses are small theymay be combined by the Binomial Theoremto give a simple estimate of converter effi-ciency Tle

= 100 -P- Ronp- R..it Pd (25)

If the coil is tightly and evenly layer -wound and the enamelled copper wire dia-meter and N are chosen so as to exactly fillthe bobbin, then the coil resistance R can beobtained from

R,= AK N2 (26)

where AR is a constant given by the manufac-turer. In practice, although equation (26) isa useful guide, not all the above conditionsapply and R is consequently rather higher(see Table 1). R,n may be written as

R,,, = R, + Rd,n + Rte + R55. (27)

where R. is the wiring and connectionresistance which is not negligible unlessgreat care is taken with the circuit layout. Ina production circuit, R, would be omitted.As PR,f is much smaller than PR,,, a roughestimate for R,ff is sufficient (see againTable 1). Using equations (22) to (25) and theT, values indicated by the solid line in Fig.6.lc may now be obtained for 44 V output with8.5 V input as shown by the dotted line inthat figure. For instance, the off loadestimate is 94.5% and that at full load 93.8%.

In Fig.1, the current values indicated byarrows apply for V, = 8.5 V. V = 44 V and for

which here amounts to 356p.A. This theore-tical value is just 121JA short of the measuredquiescent current. The total input current I,for any external load current I can now beconfidently predicted by

I, = V (I + 1;)Nr,1c + I, (31)

and hence the overall efficiency TI by equa-tion (6).

The resulting efficiency curve for I from20p..A to 60 mA is shown in Fig.7, togetherwith seven measured points which agreewith the theoretical curve to within 1.2%.

It would appear that other losses arenegligibly small.

DESIGN PROCEDURES

The design procedure may be summarized asfirstly finding an EL Auch that Tc> Tcf(equations (1), (5) and (6) and Table 2).Secondly equations (7) and (8) are used sothat saturation is avoided. Note the full -loadT, in this design is about 10% less than thatof the free oscillator's Th. Once the timing isknown, efficiency may be predicted by equa-tions (22) to (31).

Having obtained the appropriate valuesfor the parametric feedback componentsfrom the intended V and equation (19). theregulator's performance may then be testedin practice. Finally. it should be noted thatthis circuit is tolerant of considerable com-ponent variation. Hence a wide range ofinput and output voltage and current ispossible.

References1. Siemens Ferrites Data Book 1982/83, pp.153.154.2. Ibid. p.38.3. International Rectifier Data Sheet No PD9440.4. Smith, K.L.. 'Switched -Mode Power Supply'.Electronics & Wireless World. October 1985.pp.61-64.

AcknowledgementOur thanks to P.E.K. Donaldson for his en-couragement.

N.J.Chaffey, Ph.D. and T.A.Perkins. M.Sc.are both at the Medical Research Council'sNeurological Prosthess Unit.


SUPERKIT ELECTRONICS, INC.BUY DIRECT FROM THE USA AT WHOLESALE PRICES AND TAKE

ADVANTAGE OF THE $. JUST A SAMPLING OF OUR PRICES!

SN74LS323AN TI 51 MK68000P6A MOSTEK 2.70TL081CP TI 13 ON74LS366A PAN .08 S6802P AMI 1.35MC26$10L MOT 35 DN74LS386 PAN .08 HD6845P HIT .97LM3I9NLM320K5MC34002P

SIGNSCMOT

.123204

DN74LS90SN74S02J74S08PC

PANTIFSC

050408

AD7503JND8048Cn411061

ADNECSGS

1.08.46

2.16UA3406PCMP03467

FSCMOT

17.26

SN74S195N74S32PC

TIFSC

11

09Z8OHCPUP8224

SGSINT

9765

LS37 HIT 03 SN74S51N TI 11 D8259AC NEC 54LM392N NSC 13 ULN2032A SPG .21 P8287 AMD .70MN4035B PAN .04 74S240N SIG 11 P87481 -I INT 2.97MN4042B PAN .04 SN74S241N TI 15 N8 X305N SIG 4.32HCF4052BEY PAN .03 UA7812UC FSC 09MN4078BMN4082BMN40856HCF4086BEYMN4519BMN453913MC I 4584BCP

PANPANPANSGSPANPANMOT

040404030506.10

UA7912UCN8T245NN8T26AFMC8T26APN8T97NN9401N96LSO2DC

FSCSIGSIGMOTNSCSIGFSC

0911

11

11

11

8130

TRANSISTORS 6 SCR'sTIC206C)TIC216DTIP2955GI752

TITITIGI

15152311

MN4585 PAN .08 DIODES6331.1J NMI .46 MICROPROCESSORS 1N4001 ITT 01MP06700 MOT .26 WD1015PL -00-02 WD 6.75 1N4005 ITT 01SN74145N TI 15 FDC1797 SMC 1 62 1N4006 ITT 01SN7433N TI 09 WD1943M00 WD 1.35 1N5225BRL MOT 05SN7445NSN74ALSO4AFN

TITI

1503

D2147D2COM2601

NECSMC

.46

.541N58181N6263

MOTITT

0604

74HOON SIG 19 A12625-5 BURR BROWN .38 1N753A 0274H01 N NSC/SIG 19 SCB2673BC51.440 SIG .8174H108N74H10PC74H2ON74H40N74H51N74H55N74H74N

TIFSC/MOTNSCSIGTITISIG/TI

1919191919

.2019

D2758TMS2764-25JLTMS27C256-25JLTMS2732A30JLAM27S181DCTBP28L22NA14437

INTTI/SGSTITIAMDTIAMICON

.271.542.161.72

.65

.971.62

OPTOSH21A2DL 500HDSP5523HDSP5533TP$603

GELITRONIXHPHPTOSHIBA

4116654908

74LS240N FSC/PAN 10 MM5290N4 NSC .1574LS24IPC FSC 10 SSI580CP SSI 2.16 COMPUTER74L$26 PAN 04 R6500-IEAB3 ROCK 14.04 5' Floppy disk drive - Chinon 35SN74LS322AN TI .51 R6503AP ROCK 1.30 Keyboard 101 AT enhanced 29

ALL PRICES IN STERLING

WE HAVE 10M US DOLLARS SUPERKIT ELECTRONICS, INC.IN STOCK! NEXT TIME, 7905 NW 60th Street, Miami, Florida, 33166 USA.TRY US. Tel: 305 477 4069 Telex: 153265 KIT UT Fax: 305477 4116

1.:NTER 12 ON REPLY CARD

R.S.T. LANGREXSUPPLIES LTD

One of the largest stockists anddistributors of electronic valves, tubesand semiconductors in this country.

Over 5 million items in stock covering morethan 6,000 different types, including CRT's,

camera tubes, diodes, ignitrons, imageintensifiers, IC's, klystrons, magnetrons,

microwave devices, opto electronics,photomultipliers, receiving tubes, rectifiers,tetrodes, thryatons, transistors, transmitting

tubes, triodes, vidicons.All from major UK & USA manufacturers.

Obsolete items a speciality. Quotations byreturn. Telephone/telex or fax despatch

within 24 hours on stock items. Accounts toapproved customers. Mail order service

available.

LANGREX SUPPLIES LTD1 Mayo Road, Croydon, Surrey CR0 2QP.

Tel: 01-684 1166Telex: 946708

Fax: 01-684 3056

'HENRY'SELECTRONICS FOR TRADE, INDUSTRY,EXPORT, EDUCATION AND RETAIL

*INSTRUMENTSSCOPES II COUNTERS

DMMS PSU'S

I GENERATORSETC

*PUBLIC ADDRESS

II SPEAKERS II AMPLIFIERS

111 MIXERSMICSETC

COMPONENTSHUGE STOCKS

ALSO

TOOLS II CABLS ET(

*ACCESSORIES 1V -VIDEO AMPLIFIEES

AUDIO III TV /VIDEO

II SECURITYCB RA.)10

ACCESSORIES

COMMUNICATIONS INTERCOMS 7,B RADIO

*SECURITY PANELS PIRS SIRENS DOORPHONES STROBES

44 p -

FREE!ILLUSTRATED

CATALOGUES WITHRETAIL DISCOUNT

VOUCHERS Instruments/Security

Computer General Catalogue

Is SW 11,.

Please state Trade/Education(,[ Retail/mall order Send 9" x 4"

00 each or £1.50 for both

HEnRYZ4Te0l4. 0E rma roe3R2o3aRoad..London WI2 1 ED

'ALSO AT Audio Electronics 301 Edgware Road W2 01-724 3564

SALES OFFICE 01-258183' Telex 298102 Fax 01-724 0322ENTER 17 ON 10:1'1.1 I ARO


Fl It I. 1'1 `i \ 11

il,k11.)1111. III

NW.

0 Co) U. ;)

LIVINGSTON HIREFIRST AGAIN -

FIRST FOR TEKTRONIXPORTABLE DSOs

you can be the first to rent thenew TEK 2440 portable digitalstorage oscilloscope because

we are the first UK company to addit to our rental inventory.TEK2440: 500 Msamples/s digitising speedM 2ns glitch capture with 100%

confidence at any sweep speedIII 300MHz analogue bandwidthTEK2230: 100MHz analogue bandwidth 20Msamples/sTEK2210:MI 50MHz analogue bandwidth 20Msamples/sYou won't find a greater range ofDSOs from any other rentalcompany!SOUTH

uTH977 8866NORTH

998 7900Clg 66 415757

M Livingston Hire

EUROPE'S No. I IN ELECTRONIC EQUIPMEN" RENTAL

N.1 Eli 130N REPLY c.Itl)

VersatowerA range of telescopic towers in staticand mobile models from 7.5 to 36metres with tilt -over facility enablingall maintenance to be at ground level.

Designed in accordance with CP3 Chapter V. part 21972 for a minimum wind speed of 140 kph inconditions of max mum exposure and specified byprofessionals wor d -wide where hostileenvironments demand the ultimate in design, qualityand reliabilitySuitable for mounting equipment in the fields of.CommunicationsSecurity surveillar ce - CCTVMeteorologyEnvironmental monitoringGeographical sun.eyDefence range -findingMarine and aero r avigationFloodlightingAirport approach ightingFurther details available on request

STRUMECH ENGINEERING LIMITEDPortland House. Coppice Side. BrownhillsWalsall. West Mid ands WS8 7EX. EnglandTelephone Brownhills (0543) 452321Telex 335243 SE_ GFax 0543 361050

gig wysepte tr, Was

WA a.

0

U

ENTER 36 ON REP).'' CARD

RAYCOM COMMUNICATIONS SYSTEMS LTD 1'

EAr00Khz

to 950MHz IN ONE BOX

Manufacturers. importers and suppliers of world famous communications products1!INTERNATIONAL HOUSE. 963 WOLVERHAMPTON ROAD, OLDBURY 869 4RJTEL 021-544 6767 FAX 021-544 7124

WAmateur Radio. Business Radio Telephones. Sales. Service Accessories and antenna systems

OlICOml YAESU V c FOR THE LATEST,IN SCANNING

RECEIVERS

THE YAESU FRG9600 SCANNING RECEIVERSAre you looking to' a commercial grade monitor receiver with options to suit your

requirements both 117 cost and performance?

An Band Capability. LW MW SW VHFlow Air -band. 137 -Satellite HF High Maitary UHF Low. UHF -High TrY

UHF -CB Cellular (All Brox1Cast An Marne. Amateur. Business Rada Military TV Cellular up to 950MHD This unit is

rgw suitable to use in any country as all bands channel steps are now available dependeng onoptions

All Mode Capability. AM FM (wide and narrow modes) LSB USB Ode° optron a.aomble ,NTSC PAL 5 56MH1 IFI

Selectable Frequency Slaps. Most modes have selectable tuning steps *Inch allow last and easy frequency channel

selection (conventional Nom( 100 MinnorieLScanning. Any frequency in the range can be entered stored giro a fnernoq and selected scanned onbanks dowdy funclion aliows monitorrig Of omportant channel frequency whole searching Other bands channels Modes are.

also stored in memory

Fully Computer Compatible. Yaesu s CAT system is a standard Mature and a RS.?32 interface is available plus otherinterfaces lOr popular HOME computers Wrth this 'carte memories can be expanded and the unit can be remote

controlled via a modern

Clock Function. CIOCk unction allows lime logging angaul0 switch on off of the receiver

High litacinvar Sonsiiiviy. in Ore range 60.950MHz a typical recerver is I Suy for 120B SINAD IFMI or, the rangeI 6MHz-60 0006011U -2uV to 1208 SINAD (FM, below I 600MHz sensanify is nominally EouY S meter funchOn allows

nornemi held strength rneaSuremenl

13a volt DC Operation. This unit Opt,fres af f 3 BV DC of , A mat whorl, alonws mobile portable base opera.,

Matching Antenna Options. Mal, r ' -.4,, are available , ,thdong

Please ask for lull colour brochure and detailed spec, at or, Mk.' vet see evadable with coverage60-950MHe. all units now have eeeeleclors hired toe VHF UHF

s,onS. LF HF

T, o YAESU Mkt S RECEIVER ...maple ,ore

Tel: 021-544 6767 1246ranswerphonel Export enquiries welcome

%mesa Telex: 336483answerback IDENTI-G -\15321CINI.2?.:==,



L.f. navigation revivalBritain's first new long -wave radio network for decades is

being set up by a Swindon company, Datatrak. It forms onehalf of a system for tracking vehicles on the move.

For a fleet operator, knowing eachvehicle's exact location provides aworthwhile boost to efficiency. At the

very east it can help improve scheduling.But with valuable cargoes such as cash orbullion, or tobacco or spirits, the informa-tion could reassure you that a load was safelyon it way; or it might warn you thathijackers were at work.

Datatrak's system is centred on a smallelectronic module carried by each vehicle.Using positional information obtained bymonitoring Datatrak's I.f. signals, this unitdetermines its exact location and reports itautomatically by radio to a monitoringpoint. Any v.h.f. or u.h.f. radio link is

suitable (incluchng the new Band Ill trunkedsystems), though the company offers its owndedicated data network: this can handlesome 2000 vehicles on a single radio chan-nel. For dealing with the flow of positionreports Datatrak offers a bureau service.operated by Securicor (which is one of itsparent companies - the other is the GeorgeWimpey group, whose interests in offshoresurveying were the source of much of theradionavigation expertise behind the sys-tem). At the bureau, information is collatedand sent out to individual customers viamodems and telephone lines.

For land vehicles, an 1.1. radiopositioningsystem offers important advantages over thesatellite systems increasingly being used inthe air and at sea. To obtain a fix from theCPS satellites, it is necessary to have a pathto three of them - and that is difficult toarrange in city streets overshadowed by tall

RICHARD LAMBLEY °lice bLildings. Besides, hardware for an LEsystem is much cheaper.

Since last November the system has beenr innini. in the London area, though Data-t-ak is now expanding it to cover the entirecountry. Eventually the long -wave networkwill comprise 18 sites in a grid layout:station: covering the south-east are atHeathrow Airport to the west of London.Pomner Marshes, Selscy Bill. Kineton (nearStratford-upon-Avon) and St Neots. Powerlevel in each case is about 1.5kW delivered tothe antenna, though it is difficult at suchfrequer cies to achieve a radiation efficiencybetter tan about 20 per cent.

These stations, operating in the 135-145k1h-, region, are organized in groups ofeight: each transmits in turn, sending a pairof frequencies within its time -slot. A com-plete sequence of transmissions. the primary-iavigation cycle, occupies 1.68 seconds.Time -dots for a further eight stations arenterleived between the first set.

On the vehicle, the receiver module looksfor the six strongest signals and then com-putes is position by measuring the trans-mission path delays. With three signals, itcan geL a rough fix: with six, the intersectinglines box it in more closely and the fixbecomes highly reliable. A status signal fromthe urit indicates the degree of confidence.An output, in the form of map grid coordin-ates. is available from the unit on a V.24

One of Datatrak's If antennas: the com-plete If chain will consist of 18 stations.

'I VrT11411111re V. WIDVI VCC WADI ri 897

-VOMOMMIIMA

MIIMININO MM.11111111111111111

11111111111011 EMI a..1111MR911111111111111111111116.111111111111111111111011111NOMOMMOWOMNIIIMINOMMORNMEIMMOMMIUMIIMMOROMMOMOOMMOMMOMMY

MOWNIMMOMMOMMOMM

mumilMI

mor11111UOMME Mr

1111111111I11111011111rNNWNM,IWILmaMIIII.

11111111116. AMIE MINN111/111'I 41111MIIIIIIIIIIIIIIIIIII

mom iottamiimrQ,_ib... AIIIIIMMIIIMIIIIIIMImm....dammuu,rrommirik/l IMMIIIMIllik......masal.._

IIIIIIIIIIIIII quill`MIEU :"". ' ' -Nil 'p d111111111110111111111111111111111111 Ilr' - , , , ,-, ' - 'W lE1111111111111VIIIIIMIIIMEMBh. allik : 1 AMMER= `

1111WIRMINIMM1111b...,...millIMMIIM/ - it AINNIIMINIIIIIIIIIEINORIIIIIIIIIIIIIMMIIIMIIIIIIIIMIIIIP: .illho.1011111MMIIIIIIIIIIIMIII"T11IIIIM1111111111111=111111111111,:s 11111111111111111111111111111111M111111V

'ffiiiMMEEMEMMIENF: ' Iii11.1111'. "iliMMI1

AaMR

41111111

.ft0111111 MUMEMMUOMAIL,

lila ARM MMUMMUMUMUMUUMMA.

AWE. fEM UMMUUMMUMUUMMUI1111111.11111111111M111111111111M11111

ABM I 14111111111=111111111111111111111111111111rNEEL,.MEW

ARMORS,ARM MUnummilmMMUUMUUMMUMMURIBUt

'N1111111111111111111111111111111111111111111111I1E1111111111111111'

11111M11111111M1111111RIST11111111P"'14111111=

The vehicle -mounted module is fully auto-matic. A 68008 processor computes thevehicle's position by reference to the Ifsignals.

serial port for users who wish to take it nofurther.

A special feature of the Datatrak system isthat the pairs of I.f. signals are frequency -modulated slightly: the complex signalwhich results can yield a much greaterpositional accuracy than was possible withearlier systems. A fix can be to within 50 oreven 10 metres. In additional, the receivercan initialize itself from cold anywhere -every fix is an absolute one.

For relaying data back from the vehicle,Datatrak has two frequencies in the 170MHzregion and a network of 25 base stations.mainly at IBA sites. Expansion outside thesouth-east will entail a move to u.h.f. chan-nels around 400MHz. Each vehicle has itsown pre-programmed time -slot. synchro-nized by the 1.1 navigation signal, duringwhich it sends a 30ms burst of data contain-ing its position, status. direction and speed.Direction and speed (accuracy is to withinfive miles per hour) are obtained by compar-ing the two most recent fixes. Some 150-160base stations will be needed to cover thewhole country, each linked by private linesto a data centre and interconnected to allowroaming. In England there will be four datacentres, in London. Loughborough, Swin-don and Leeds.

How often your vehicle transmits a fixdepends to some extent on what you arewilling to pay: for a bullion van, a fix everyfew seconds might be needed, since anydeviation from its planned route could meantrouble. With other types of vehicle, a fixevery minute would be enough. Datatrak'sfull service includes the use of an emergencyalarm: when the driver hits the button, thesystem reports it immediately and the vehi-cle continues to transmit its position everyfour seconds until the emergency is over.

One early customer for the system isCellnet. which will use the system for com-piling field -strength maps for its cellularradiotelephone network. Measuring vehicleswill be able to log signal measurementsdirect to tape or disc together with theirexact location, cutting out a lot of tediousmanual data entry. Another user could beLondon Regional Transport, which has beeninvestigating the system's potential for help-ing to reschedule buses caught in trafficjams.

The vehicle -mounted unit for Datatrakcosts just over £1000: and the running cost.including the bureau service, is about £85per month. Further details from DatatrakLtd. Hargreaves Road. Croundwell Indust-rial Estate. Swindon. Wiltshire SN2 5AZ: tel.0793-722549.

Left: top -of -the -range display is a high -resolution colour screen on which eachvehicle's position is re -plotted every fewmoments. Advanced software will make itpossible for a warning to be given auto-matically if a van deviates from its plannedroute. Less ambitious users can opt for asimple telex -style printer instead.


The Archer Z80 65CThe SDS ARCHER - The Z80 based single boardcomputer chosen by professionals and OEM users.* Top quality board with 4 parallel and 2 serial ports,

counter -timers, power -fail interrupt, watchdog timer,EPROM & battery backed RAM.

* OPTIONS: on board power supply, smart case,ROMable BASIC, Debug Monitor, wide range of I/O &memory extension cards.


Thc 5owinan 6800065CThe SDS BOWMAN - The 68000 based single boardcomputer for advanced high speed applications.* Extended double Eurocard with 2 parallel & 2 serial

ports, battery backed CMOS RAM, EPROM, 2 counter -timers, watchdog timer, powerfail interrupt, & anoptional zero wait state half megabyte D -RAM.

* Extended width versions with on board power supplyand case.


acrd DafncSustem6 Ltc.Sherwood House, The Avenue, Farnham Common, Slough SL2 3JX. Tel. 02814-5067

"WE CAN TONEYOU UP FAST."

RETROFIT TONE

SIGNALLING SYSTEMS

For

RADIOTELEPHONES

O CTCSS I/S

O 5/6 Tone

O Selective Calling

O ANI Microphones

O Control & Display

FOR ANY GENERATION & MAKE OF EQUIPMENTDESIGNED, BUILT & FITTED

PMRPrivate MobileRentals LimitedIndustrial Estate, Gwaelod-Y-Garth,Cardiff, CF4 8JN, United Kingdom.Telephone: 0222 810999Telex: 497244 NOVCDF GFax: 0222 813369

DISTRIBOOR

ADVERTISEMENT

PCB Manufacturers -Which to choose?With scores of PCB manufacturers fallingaver themselves to make your

conventional boards, it can be very difficult tochoose the right one.

You could however, choose a PCBmanufacturer with more to offer. Such asassembly, panel printing, final build and design.

Slee Electro Products can provide you withthese services and will still offer you competitiveprices on your PCB's.

"We're hot on quality and delivery too. Andbeing a member of the Printed CircuitAssociation, means we have to try harder."

Slee Electro ProductsUnit 4, Grange Lane Industrial Estate,

Can -wood Road, Barnsley, South YorkshireS71 5AS, England.

Tel: 0226 200717. Fax: 0226 731817ENTER 25 ON REPLY CARD



ti99

Dependent source theoremFinding equivalent generators in a network which has no

independent sources.

many of today's networks contain de-pendent sources only, in fact themajority of the semiconductor -

equivalent networks belong to this group.While the Thevenin-Norton theorem readilyyields equivalent generators when the net-work contains at least one independentsource, it fails when the network is free ofsuch sources. Still, we wish to be able toconstruct equivalent generators in all cases;methods to obtain such generators are dis-cussed in the following. Although the net-work is put in the handy configuration of agenerator, in reality it remains passive, withits "source" controlled by a current or avoltage outside the port. The dependentsources we are concerned with presently areof the basic kind kels) and kils).

A simple example of a network whichcontains a dependent source only is shownby the one to the right of the port in Fig. I (a),and this network might well represent atransistor. One method to construct anequivalent generator is that of the SourceTransformation Theorem, in which we col-lapse the given network in steps from rightto left, thus ultimately obtaining a series -form generator and a parallel -form gener-ator like the ones shown in Fig.1 (b) and (c).Such equivalent generators, however, arenever independent of the driving network onthe other side of the port, and we mustconsult the equations for the entire systemwhen we wish to switch variables, for exam-ple to make the source a function of e insteadof i. Since the driving voltage and currentare the same before and after the trans-formation, we may express the equivalentsource as a function of either.

Another method is contributed by theDependent Source Equivalent -GeneratorTheorem, which makes the segregated"dead" network alive by prescribing that thesource retains the initial value it has in thegiven network. This means that we deter-mine the value of i in the given totalnetwork, and then treat ki as an independentsource when determining the open -portvoltage and the closed -port current. Ex-pressing the equivalent sources ex and ix ascurrent sources, we find, with R23= li.,+ R3and 12112,-F R 1 R3+ R,R3=A,

e' = k 12, i/R.,3 IIIix = k R, i/A (2)Rx = A/R, 3 (3)

We note that the relationship ex/ix = Rx,typical for the Thevenin-Norton theorem,also holds for this theorem.

We may use either equivalent generator to

HARRY E. STOCKMAN

R1N

calculate quantities such as RIN and v/e.obtaining

RINI= (A+ kR2R23 (4)v/e=R31R,-k1/(A+ kR21 (51

A most direct way of obtaining RIN from theconstructed generator in Fig.1(b) is to applythe Compensation Theorem, change ex intoa resistance and then just sum the tworesistances.

There exists a short-cut to the determina-tion of input and output impedance, contri-buted by the Port Immittance Theorem,assuming, however, that the transfer func-tion is known, and in this case it is given by(5). The theorem states that the denomina-tor of (51, set to zero, contains both the inputand the output impedances. It states that theinput impedance is the negative of RI, thus-R1 = RUN and that the negative of R: is theoutput impedance, thus -R3 = Rout.Whenever the proper transfer function isknown, this theorem saves time. To find RINwe must add R1 to RUN'. We find for Rorr:

+k)124R (6)-R3= RorT= (RI

If the transfer function is not known. wemay still use the indicated short-cut byintroducing -RUN in the given network as areplacement for e and R1 and, in a subse-quent operation, -Rorr as a replacement forR3. An analysis of the network yields theformulas for RUN and R017. This is in accord-ance with the Theorem, which is a collateralto the Tellegen theorem, and the proof restswith the Tellegen theorem. As we replace eand R1 by -RUN. we cut off the signal supply tothe network under investigation, but at thesame time we introduce a negative resist-ance, and indeed one that equals the positiveresistance it "sees". Accordingly, we createinstability and thus the currents and vol-tages we need to analyze the network.

The Dependent Source Equivalent Gener-ator Theorem has the following formulation:

A linear network with dependentsources only. provided with an accessi-ble port for a driving generator. may at

RIN

(bl

this port be replaced by a generatorimpedance in series with the open -portvoltage or in parallel with the closed -port current, respective voltage andcurrent appearing with the initialvalues they have in the given network.and the generator impedance being theport looking -in impedance when allsources are removed.

The driving network may be representedby a series -form or parallel -form generator,and may consist merely of the source to sucha generator, as is the case in Fig.1(a).Additional network theorems are listed inthe reference.

ReferenceStockman I I.E. "Transient Analysis Aided byNetwork Theorems". Sercolab. 1984. Box 767. E.Dennis. MA 02641, USA.

Dr Stockman was born in Sweden. Hereceived his M.S. degree from the RoyalInstitute of Technology in 1938. continuingas Assistant Professor of Radio Engineeringat the Institute. During the Finnish -Russianwar. he served as a radio adviser and wasawarded the Liberty Cross by Field MarshalMannerheim. In 1941, Dr Stockman went tothe United States on a Ford Fellowship andgained a S.D. from Harvard. He eventuallybecame Head of the USAF CommunicationsLaboratory and later formed his own com-pany, Sercolab. in the electronics educationfield.


FIELD ELECTRIC LTD. 01-953 6009. Fax: 01-207 63753 SHENLEY ROAD, BOREHAMWOOD, HERTS WD6 1AA.

SUPER SPECIAL OFFERNes' 12. Green Phi, Monitor 751! comp vide' s input. high ,,scu 119.95 p 4NEC 9" Green Phos Monitor 7511 oornpvideo input high r es 240V AC input cased with built in handle £19,95180 test.- :

12" 7`.!! , 1. ,sete Video Input Monitor. 22MHz bandwidth 230V AC input, new 8 boxed in case green phos datasupplied £59.95 c p 5 00

Motorola HI. Mondor Chassis r green lobos. 22MHz bandwidth. 120 DC input 12A new 8 boxed. circu't dia 8 datosupplied compatible to BBC IBM r omputers dia supplied for connection to BBC 7511 comp video circuit diasupplied. discount for 10 r 120.60 r p 2 00

Qume Model QVT 102 Monitor c wan model QVT 1020 keyboard green phos monitor coverty k board 65232 inputspivot 8 tilt VDU /50 up 500

5' '4" ; 7 height disk drives in case 4 . I t £46

MPI Micro Peripherals ice 5ba hiu l- .1. ; , . : le 120.60 Inc c p Some l/2 height 80tracks sin stock

Winchester 8' hard disk roes Q2020 20 meg supplied with data new ex equipment £100.00 p 600

tabor Corp F loopy disk drives 83rbrf disk 34 way IDC edge our cooper for new 8 boxed £18.50 p 200

I inlay Microfilm F MI. portable mu ro fiche reader. 240V AC or I20 DC input c with 6V 1 6A AC adpt 12! DC 6V ACodor fiche inc le is. gates or Nat Pan sealed lead acid cells c 3. we cannot offer guarantee cells 6V 9 watt Quartz

o. ova, I,. I r arryIng case size 810 7la c 5 hard vinyl 124.00 new and boxed

; .. E single pen recorder 5V DC sensitivity electric writing. complete with 3 rolls paper new rinboxed112500 1. 503

it 0 2770 AC 7400 A(' 101,11.1n/ I5A cased new £160.00 p F 00

miler Yellow splash proof . work etc £39.00 c p 5 50. I £36.50 . ' 75 240V Primary 110 se, - £44 00 ; 6 75

Pmverultra Sirring we

switch mode units 2400 AC input 5V 2G° 118 50£25.00 12V 25A ultra small £38.00. Multi ra c

y vast stocks of PSU

£25.00 frOP /22.00 larnell 6V 5A4 see your requirement Please

: hes listed Tekh000crs 4- DU eh.

TEST AND MLAIsiiniNi NTHP new L230.H P Multifunction mete,HP 62605M DC PSU 50 i £135H P 6518 Test Oscillator 101r.: h, , ,l biz 1195HP Voltage divider probe 10004D. new with manual. I Om! I I Opt £85H P 9895A 8" Disk Drive. cased with PSU etc. new and boxed £225 complete with manuals etcH P 829015 twin 5124" disk drives cased with PSIJ new 1225H P 5000A Logic State Analyzer /230H P C34 431C Ruggedized Power Meter 10MHz 0 40GHz 180H P 4288 Clip on Milometer 1 MA to 100150H P 33308 Automatic Synthesizer 0- I3MHz /1150H P 161IA Logic State Analyzer c w probes etc 1850H P 6824A DC PSU 'Amplifier 50 50V DC DiH P 8120 1082 Thermistor Mount Cable New £80H P 478A Thermistor Mount 20011 Neg £125H P P486A Thermistor Mount 10011 Neg 1125H P 1803A D C Din Offset amplifier Pin £75Tektronix 286 sampling head multiplex unit /1211Tektronix 178 linear IC test fixture amTektronix F ET Probe 6045 £90Tektronix S3A Sampling Head £120Tektronix 7403N Rackmount 0 scope Mainframe no guarantee tube 06 1200Tektronix DI I Storage 0 scope Mainbarne. no guarantee tube 06 £200Tektronix 7S11 Sampling Amp Pim £450Tektronix 7892 Dual time base Pm £450Tektronix IA4 4 Trace Plug in £75Tektronix I A5 Comparator Plug in /45Exact Model 337 Digital phase gen 4µHz 100Mkrz 1300Keithley Inst 610C Solid State Electrometer rnea. ore V 1.R Q and IS a current source £850Bell 8 Howell Datatest Calibration Und for I M sylems £175Marconi Sanders Microwave Osc 'w 26 (;Hz Plug in t???Mar, on: Inca 0 Ill universal bridge IF 13130 £280

0, ipm P a II it 7400 AC input all metered Kingshrll 501 0 50V 0 10135 0-40V 0-3A - 2 1115 0 200 0-10A £115185 I 02A'2/125 Weir Maxireg 762 0-600 0 2A 1140 Lambda 0 400 0 3A 198 0 WV 0 IA

f 125 - . 0 IA 2 /45 H P 6824A 50V IA 175 Oltronix B401 0 400 0-1A 150 13817S3 10V 0 7A£98 - 614,112 0 40V 0.12A 1345 60.40 60V 0-4A /260 Lambda LMG 12 12V DC 5'. 65A DC Lin /345

p details Please ring

Chr.r6 111 Numeric ASCII Coded Keyboard including 8 colour coded graphic keys. I )8 keys f:an, X Y math.. ..nr ode keys. 9 graphic control keys. 5V rail teak 8 black all case new and boxed 124.95 1

£22 00 .

All listed cards 1/2 shown price this month. Card No 1 I 280A CPU 1 c D8255 AC5 in holders. I cs M118264 15 I c SN74198N 53 carious chips new exequipment 116.50 Card No 2 I c WD19331

16 various new ex equip £1225 Card No 3 2/ D8255 AC 5.2 c HLCD0437P in holders 10 various 14.95N., 4 LCD 6 digit display 12 momentary plain keyboard rocker switches 4 bar LEDs Green yellow red Flat top type£6.95 Card No 5 Peripheral Comms Controller .8 MC68661 8 MC14891 8 c MC1488P 13 various chips all inholders new exequipment 118.95 Card No 6 ;100 Backplane 20 100 pin connectors new £29.95 Card No 7Hard disk floppy disk controller SIO0 type 'Ili 0766AC D8237 AC5 8253 8085/1 2764 64K [prom SN74L240N224N 373 etc. Inc block dia new ex equiprtent 126.50 Card No S lobo red Remote Controller IM -3-84700Encoder IC I Infra Red Emitter 16 nymbrone ke,troor I 13.50 Carol'.,, 1 NIChnla MI12 Motorola ceramic CPUI 16MHz Xtal I various chips in, block I , s 1111 new Y. ,j111: 11,`. £29.95 Card No 1010 MasterBoard 17.50 Card No 1110-M BOLI, £7.50 c a No Is MC601511. - r./ o Ceram, £12 various chips

Frew art! F'or k ard eVA Personal Computer with built in interfaces for 2 disc drives and centrcnirs compatible printer6411 built in user memory. 14 user definable keys display capacity 16 or 24 lines c 80 characters c w system demoI tc I I 'I : F.' lull user Manila: etc r uruplete new in sertecl L, v es £350.00 mm molt fi.!

Full height MFE tape streamer 30 .C.S. Interta, stremner tape £65.00 0 1 50

1 2V 1 8Ah NiCad new. 42 25n,n £2.00 .: JAtr N. Cad new 4: /2.00 rfc c

Vac win, Ascend' S 7400 0i130 : -

P .62,0 PPM r aoth snot vac h

We would like the opportunityOfficial Orders/Overseas Enquiries Welcome/Order by phone or post Open 6 daysinclude 15% VAT unless stated. Phone your order for quick delivery. Access. Amex.

and aerial equipment. much more

to tender for surplus equipment. Postal rates apply U.K. mainland only. All test equipment carries warranty All pricesDiners. Visa accepted. We can supply telephone and some audio equipm,1 PIP r ,3l

than is shown in our ad. Please ring.

ENTER 150N REPLY CARD

INTEL 82786 GRAPHICS ENGINE Hardware Windows Drawing at 2.5 Million Pixels/sec Fill at 30 Mbit/sec BitBlt at 24 Mbit/sec

TT786- HARNESS THE POWER -£395 IBM PC Add -In 100% CGA, EGA, and VGA Compatible 512 Kbyte to 4Mbyte Memory

\f"42\' TT786 SOFTWARE LIBRARY ..f75 C and BASIC Comprehensive Documentation

1N

111111111111001IMMO MEMO

MINIMMO MOMMIn_111111M1 IMMOMmomMEMOINIMMEM III MEMOS

111:MMillOMMUMM

TEKTITE LTDPO BOX 5FELIXSTOWE, IP11 7LWSUFFOLK, ENGLAND0394 -672117TELEX: 987458

M & B RADIO (LEEDS)THE NORTH'S LEADING USED TEST -EQUIPMENT DEALER

OSCILLOSCOPESTektronix 7603 main frames £750Tektronix 7604 main frames C650Tektronix 7704Plugs in stock 7A18 7A17.7B52. 7653. 7A17Telegurpment D83 50MHz D T E325Telequipment 075 50MHz Dual beam £235Telequipment D66A 25MHz £175Tektronix 465 100MHz £650Tektronix RM529 Wavetorm Mondor £150Quantity of Cossor CDU 150 Compact Solid StateOscilloscopes. Dual Beam 35MHz with delayedtirnebase and probes each one tested and cinchedfor calibration £175Tektronix 603 Storage Monaco, E550HP 182 Scope Main Frame £350Systems Video Pal Vector Scopes £550

£195HP 191A Waveform MonitorsFamed C13 1214 E195Hameg HM605 60MHz £395

SIGNAL GENERATORSMarconi TF2015 I TF2171 Sync10 to 520MHz £600Marconi TF2016 TF2173 Sync AVVFM

oliz to 120MHz £400Marconi TF1066rB 10,470MHz AM FM £175Marconi TF144H 101-1, to 72MH: £75Marconi TF2012 AM FM 400 to 520MHz £175Marconi TF995 A5 1 5 to 220MHz E175Marconi TF2333 Transmission T ast SetLyons PG73N Bo Polar GeneratorHP202C LF Oscillator £85HP 8620A Sweep Generator t 2 to 12GHz 0£££.000300E£3,985:55

HP33t IA Function Generator £175HP333A Distortion AnalyserMarconi TF2330A Wave AnalyserHP3200 VHF Osc 10 to 500MHz £195Level) TG150 Audio Osc £25Gen Rad 1362 UHF Osc 220 to 920 MHz E150Texscan WB71 t 500MHz Sweeper £325Fluke 6160A Freq Synthesizer 30MHz £195Advance B4B 30MHz Sig Gen £45AVO CT378B 2 to 250MHz AM CWRadiometer SMG1 Stereo Generator £175AE Codasyn 201 Display £145Ferrograph R T S2 £200Marconi TF2331 Distortion Factor Meter £150Pndips PM5234 100611,1 10MHz £375

Marconi TF 1245 1246 0 MeterMarconi TF2700 LCR BridgeMarconi TF 1313A 1% BridgeMarconi TF2701 In Situ BridgeRadiometer MM2 LCFI BridgeHP 3465A DVMHP 3403C True RMS VoltmeterHP 3400A RMS VoltmeterHP 400EL vortmeterHP 8170A Logic Pattern GeneratorHP 1601A Logic Analyser Plug-inHP 1600A 1607A Logic AnalyserHP 1600A Logic AnalyserBird Termaline 6254 100MW meterBird 8329-300 30DB 2KWBird 8327 30DB 1 000 wattsBird 8201 500 wanHP 3761A Error DetectorHP3670A Data GeneratorHP2722A Noise GeneratorHP 6516 Test Osc

E195£150C150

C75£1750225£495£275£200

CI 000£250£475£300

£60£300£150El rt"

£1 75

SPECIAL OFFERSRacal RAt 7 Communication Receivers 5001(CS to30MHz. ex -mod quantity available £140Bradley CT471C Electronic Multorneter AC:DCVolts MA II with RF probes to 1100 MHz Batteryoperated £25Marconi Freq Meter Resonant 50010t000MHz E15RiS Polyscops II 1200MHz E100AVO Meters Type 8 £50

SPECTRUM ANALYSERS

HP 8558B 1 to 1500MHz 182 FrameHP 85588 I to 1500MHz 18114HP 141T 8554L 8552A 1250MHz

£4.250£5 000C2 500

Marconi TF2300 Mad MeterMarconi T F 2300BQuad 50D Audio Power AmpsJVC Umatic Edit SuitShlurnberger Attenuator 4980

C150£350

£45C1.000

£250

All instruments are tested and sold in full workingorder

ALL PR'CES PLUS VAT AND CARRIAGE

86 Bishopgate Street, Leeds LS1 4BB.Tel: 0532 435649


ELECTRONICS& WIRELESS WORLD

Nii -,0\ 1'11 I 51111

NEW FROM NUMBER ONE SYSTEMS NEWPCB CAD, FOR THE PC/XT/AT, THAT YOU CAN AFFORD

- HAVE YOU BEEN PUTTING OFF BUYING PCB CAD SOFTWARE?- ARE YOU STILL USING TAPES AND A LIGHT BOX?- HAVE YOU ACCESS TO AN IBM PC/XT/AT OR CLONE INCL. AMSTRAD

1640 8 15129- WOULD YOU LIKE TO BE ABLE TO PRODUCE PCB LAYOUTS UP TO 17'CD

SQUARE?- WITH UP TO 8 TRACK LAYERS AND 2 SILK SCREEN LAYERS?

PLUS DRILL TEMPLATE AND SOLDER RESIST?- WITH UP TO EIGHT DIFFERENT TRACK WIDTHS ANYWHERE IN THE

RANGE 002 to 531"?WITH UP TO 16 DIFFERENT PAD SIZES FROM THE SAME RANGE?

- WITH PAD SHAPES INCLUDING ROUND. OVAL. SQUARE, WITH ORWITHOUT HOLE. AND EDGE CONNECTOR FINGERS?

- WITH UP TO 1500 Cs PER BOARD. FROM UP TO 100 DIFFERENTOUTLINES?

- WITH AUTO REPEAT ON TRACKS OR OTHER FEATURES - IDEAL FORMEMORY PLANES?

- THAT CAN BE USED FOR SURFACE MOUNT COMPONENTS?- WITH THE ABILITY TO LOCATE COMPONENTS AND PADS ON GRID OR

TO02" RESOLUTION?- WITH AN OPTIONAL AUTO -VIA FACILITY FOR MULTILAYER BOARDS?- WITH THE ABILITY TO CREATE AND SAVE YOUR OWN SYMBOLS?- THAT IS AS GOOD AT CIRCUIT DIAGRAMS AS IT IS AT PCB's?- THAT CAN BE USED WITH EITHER CURSOR KEYS OR MOUSE?- WHICH. WITH "EASY -PLOT". CAN ALSO OUTPUT TO A PEN PLOTTER? (A

PHOTOPLOT DRIVER WILL BE AVAILABLE SHORTLY)- WHERE YOU CAN LEARN HOW TO USE IT IN AROUND HALF AN HOUR?- THAT ONLY COSTS £275.00 . VAT?

PLEASE CONTACT US FOR FURTHER INFORMATION

..`

CD CD

CD CD

CD

CD

CD

CD CD

CD CII

CD CD""ss

fai CA

o

b

o

CD

CD

CD

CD CP

CD

0-CD

0 CD

0 CD

CD

0 "cik

c/.

^"0 Cit

" C 0 0 0 :fC0 0 0 0 C 0 0 0 "'

i:

^/"'"':0 : 0 CCI:

0 0

0 0

Output on dot matrix printer reduced from 2:1

EASY -PC, EASY -PC, EASY -PCNUMBER ONE SYSTEMS LIMITEDREF: WW, HARDING WAY, SOMERSHAM ROADST IVES, HUNTINGDON, CAMBS PE17 4WR Telephone: 0480 61778

ENTER 10 ON Ill CARD

RUGBY TIME?MSF CLOCK is EXACT - never gains

or loses. SELF SETTING at switch -on. 8 digits show Date, Hours,Minutes and Seconds. ALSOparallel BCD output for COMPUTERand audio to record and show timeon playback, receives Rugby 60KHzatomic time signals, only £97.90,get the RIGHT TIME.

60KHz RECEIVER, as in MSFCLOCK, serial data output, withBasic listings to decode Years,Months, Date. Weekday. Hours,Minutes and Seconds, narrow bandsuperhet receiver, built-in antenna,1 ,000Km range, only £35.30.

Each fun -to -build kit (ready-made toorder) includes all parts, case. pre -

wound coils. fibre -glass pcbs,instructions, by -return postage etc andlist of other kits.

CAMBRIDGE KITS45(WJ) Old School Lane, Milton, Cambridge. Tel 860150

GPIB for PC/XT/ATHIGH QUALITY half-lengthcard - fits all PC -compatibles

.62). Full IEEE 488.2 extensions forall popular languages

Easy to install hardware -powerful, easy -to -use software

These are some of the advantages provided by the newhigh-performance IEEE 488.2 controller card and softwarefor PC -compatibles from M A INSTRUMENTS.

migiim Versatile FORMATTED I/O: mix hex and text!

24 MONTH guarantee

ignitT Lifetime telephone HOTLINE support

Enhance your existing -FMS9914 oruPD7210-based GPIB card with ourIEEE 488.2 software upgrade.Phone us for details.

TBASIC AVAILABLE NOW!

Designed, manufactured and supported in the UK by:M A INSTRUMENTS LTD.Freepost, Yelverton,

IEEE -488.2 Devon PL20 6BR1-r Tel: (0822) 853585Tlx: 45441 IMAGE (MAI)

902



FSATELLITE SYSTEMSGeostationary

loops?A novel idea to solve the problemof congestion in the geostation-ary orbit is the geostationaryloop. It doesn't use the conven-tional Clarke orbit at all hut hasthree satellites moving in threeelliptical orbits as shown in Fig.1(a). The orbital planes are angu-larly spaced by 120° from eachother. The orbital elements arechosen so that the three satel-lites follow each other in thesame track in a co-ordinate sys-tem rotating with the Earth. asshown in Fig. 1(bl.

As a result of this arrangementthere are two small geostation-ary loops. 180° apart, positionedabove particular points on theEarth's surface. Observers on theEarth would 'see' a satellite ineach loop, moving round it ineight hours. There would be asatellite in each loop at all times:as one satellite leaves a loop, thenext satellite in the sequence ofthree enters it. They pass in andout of the intersection pointsseen in Fig. 1(b).

As mentioned briefly in ourMay issue report on an IEE dis-cussion on the geostationaryorbit (p.489), this proposal com-es from Germany and has beenput to the European Space Agen-cy. It is the joint work of threeWest German organizations: theaerospace company MBB/ERNO,the communications engineer-ing firm ANT Nachrichtentech-nik, and the country's centralaerospace research and develop-ment body DFVLR. They havecalled the system Loopus (geo-stationary loops in orbit occu-pied by unstationary satellites).

For each of the elliptical orbitsin Fig. 1(a). the orbital elements(defined in the November 1987issue. p.1158) are as follows.Apogee is 39 117km. perigee is1 238km and inclination is 63.4°.The semi -major axis of theellipse is a distance of26 585.7km while the eccentric-ity is 0.7132 and the argument ofperigee is 90°. In this path eachsatellite has an orbital period of11 hours. 57 minutes. 47seconds (there's German preci-sion for you!). One version of thissystem would produce geosta-tionary loops above Europe at10°E and above the Pacific Oceanat 170'W. Both loops would be at

latitudes between 45'N and63.4°N.

As a means of providing sat -corn services the Loopus systemhas been designed to meet twomain requirements. One is togive global coverage for themaritime mobile satellite service- that is. coverage of areas onEarth where ground stations cansee the appropriate satellite at aminimum elevation of 5°. Theother requirement is to providespot beam coverage for the fixedsatellite service (FSS). Here thecomsat beams would give fivespots on the Earth's surface, fourwith a diameter of about 3000kmeach and one with a diameter ofabout 1000km. All the spot -beamfootprints provided by one suchLoopus would be in the northernhemisphere, but it would bepossible to have another, similar.independent system serving thesouthern hemisphere as well.

One problem that had to beconsidered in devising this un-usual technique is that thecoverage areas change as eachsatellite moves round its geosta-tionary loop. This is because ofthe varying distance and angle ofthe spacecraft relative to a givenpoint on the Earth.

The coverage areas are definedas those areas in which the powerflux density on Earth, during theentire passage of the spacecraftfrom the intersection to the topof the loop and hack again, cor-responds to at least the p.f.d.generated at the top of the loop atthe 3dB contour. A study hasshown that by taking the p.f.d. asa basis for determining theperiphery of the coverage area.the changes in path attenuationare more or less compensated bychanges in the satellite antennagain which affect the edges of thecoverage area. So the size ofthese areas remains practicallyconstant and only minor de-formations occur as a result ofchanges in the satellite's anglerelative to the Earth.

Another problem arising fromthe varying position of each com-sat as it travels round the geosta-tionary loop is that it affects theparameters of the radio link be-tween Earth and space. As aresult, various compensationtechniques have to he introduced- especially, for example, for thechange in received carrier fre-quency caused by the Dopplereffect. Further complications are

(0)

Fig.1. Loopus system using three satellites to form two geostation-ary loops: (a) three elliptical orbits as seen by a fixed observer; (b)the satellites' track as seen from within a frame of co-ordinatesrotating with the Earth. A second, independent 'upside down'system could be used as well, to give good coverage in thesouthern hemisphere.

made necessary by the fact thatin each geostationary loop thecommunication task has to behanded on, every eight hours.from one satellite to the next.

All these difficulties have beenlooked into by the originators ofthe Loopus scheme and solu-tions have been proposed. It re-mains to be seen whether theseinevitable complexities are justi-fied by the benefits offered by thesystem. Apart from the relievingpressure on the limited resourceof the geostationary orbit, theLoopus system would also pro-vide two other benefits. It wouldgive better coverage of polar re-gions than the g.s.o. does andwould allow single -hop connec-tions between Earth stations onopposite sides of the world pro-vided these were in the samehemisphere.

New amateurradio satellite

One of the three spacecraft laun-ched simultaneously on the firstflight of the new Ariane-4 rocketwas a further communicationssatellite for radio amateurs.Amsat IIIC. It becomes the latestmember of the well known Oscarseries (orbiting satellites car-rying amateur radio) and will beknown as Oscar 13.

The new Oscar was built byAmsat-Germany under the inter-national programme run by theAmsat organization. It is a 150kg

spin -stabilized space -craftshaped like a solid three -pointedstar (pictured overleaf), with so-lar arrays fixed on the six edges ofthe points of the star. Like allothers in the series, Oscar 13 isbasically an educational aid,allowing amateur radio com-munication over much greaterdistances than would normallybe possible in the v.h.f. and u.h.f.bands used.

There are four transponders inthe spacecraft. One, with435.5MHz uplink and 145.9MHzdownlink, has a bandwidth of150kHz. The second has a1269.45MHz uplink and a436.85MHz downlink, with abandwidth of 250kHz. The thirdtransponder operates on a144.45MHz uplink and a435.95MHz downlink, while thefourth uses a 435.6MHz uplinkand a 2400.7MHz downlink. Adigital communications experi-ment is included in the last-mentioned transponder. Theantennas are a mixture of v.h.f.rod, u.h.f. dipole and helicaltypes.

Because most of the world'sradio amateurs live in the north-ern hemisphere between the lati-tudes 30° and 60°. satellites inthe third Amsat generation havebeen put in orbits designed to bethe most useful for these re-gions. Amsat IIIC/Oscar 13. asthe third member of this thirdgeneration, carries on this policyand has an elliptical orbit whichgives much better coverage ofthese northern regions than


[SATELLITE SYSTEMS

Fig.2. The Amsat IIICor Oscar 13 amateurradio communica-tions satellite laun-ched in June this year.

would be possible from, say ageostationary one. This ellipticalorbit has an apogee of 36 000km.a perigee of 1500km. an inclina-tion of 57° and an orbital periodof about 11 hours. Immediatelyafter launch from Kourou,French Guiana, the satellite wasplaced in a transfer orbit with aninclination of 10° and was subse-quently moved into the finalelliptical orbit by the reaction -propulsion motor mentionedabove.

The Oscar series started inDecember 1961 with a spacecraftcarrying just a simple 145MHzbeacon transmitter. Later satel-lites introduced transpondersand became progressively moresophisticated in spite of cost con-straints. Britain's contributionso far to the internationalamateur programme is the Uni-versity of Surrey's UoSat-1. cal-led Oscar 9 and launched inOctober 1981. and UoSat-2. cal-led Oscar 11 and launched inMarch 1984. The new Oscar 13 isan improved version of Oscar 10.which was built by Amsat-Germany and launched in JunePis:1

Space stationagreed

Elaborate radio communicationsbetween space and ground willbe needed when the Westernworld's permanently -mannedspace station eventually goesinto operation, possibly in the1990s. This will be a very largeartificial satellite, orbiting theEarth at a low altitude of 450kmand an inclination of 28°.

As a proposal it has been dis-cussed and argued over for sever-

al years. but now seems morelikely to go ahead as a definiteproject. After two years of nego-tiations, a basis for internationalco-operation on design, develop-ment, manufacture and opera-tion has been agreed between thefour main partners: USA,Europe. Canada and Japan.NASA will be responsible for theAmerican work and ESA. repre-senting eight member states. forthe European contribution.

The USA will provide the over-all space station framework:operating sub -systems includinglife support and 75 kilowatts ofpower: laboratory and livingquarters: and an unmanned free -flying platform that will beplaced in polar orbit for Earthobservation.

Canada will provide a mobileservicing system to be used inconjunction with the assembly.maintenance and servicing ofparts of the space station. Japanwill contribute an 'experimentmodule', which is a permanentlyattached pressurized laboratory.ESA, through the associated Col-umbus programme, will providea pressurized laboratory modulepermanently attached to themanned base: an unmannedfree -flying polar platform towork in conjunction with the USpolar platform: and a 'man -tended free flyer' to be serviced atthe manned base.

Some time ago there was asharp rift between the Amer-icans, who seemed to want to usethe space station partly for milit-ary purposes, and the Euro-peans, who were firmly opposedto this. In recent statementsNASA describes the project as a"civil" space station while ESAgoes further and adds the words"for peaceful purposes".

New satcomsoperator

The first private company to ownand operate an international sat -corns service outside of the offi-cial Intelsat. Eutelsat and Inter -sputnik organizations is the USfirm Pan American Satellite, asubsidiary of Alpha Lyracom ofGreenwich. Connecticut. It ownsa single spacecraft, the PAS -1comsat, which was launched bythe first Ariane-4 rocket alongwith Amsat IIIC (see report else-where) and Meteosat P2 (Januaryissue p.58). From its geostation-ary slot of 45°W, PAS -1 will beable to relay international trafficbetween Europe and USA andbetween Latin America and theUSA. as well as covering theCaribbean area.

Although providing an inter-national service, Pan AmericanSatellite will not be competingdirectly with the main worldorganizations in public telecom-munications. The company willbe concerned principally withprivate business satellite com-munications and will not linkinto public telephone networks.In the UK, uplinks to PAS -1 willbe provided by British Telecom.However, the comsat is also like-ly to be assisting with nationalservices for Latin Americancountries which might other-wise be unable to afford them.

In engineering design PAS -1 isfairly conventional, being a mod-ified version of a standard RCAAstro-Electronics Series 3000satellite. It has 24 transpondersand operates in the C and Kubands, normally working tosmall and inexpensive Earth ter-minal antennas. The three satellites launchedon the first Ariane-4 flight on 15Jung had a combined mass of3513kg. As already reported inthe July issue, the Ariane-4 de-sign. which is much morepowerful than the precedingAriane-3, allows various com-binations of engines and boos-ters to be used to suit differentloads. In this case the 44LPversion was employed, with twoliquid and two solid propellantboosters. The third stage of thisrocket was first injected into atransfer orbit with an apogee of36 359km, a perigee of 221kmand an inclination of 10°. Subse-quently the three satellites prop-

elled themselves into their re-spective final orbits as statedelsewhere.

A further success for theAriane -3 space vehicle came on22 July. with the launch of theIndian satellite lnsat IC (seeJanuary issue. page 57) and thelast of the five European Com-munications Satellites. Whencommissioned. ECS-5 will be re-named Eutelsat 1-F5 and willprovide a range of telecom-munications services to Europeincluding telephony, businessservices. and radio and televisiondistribution. ECS-5 will replaceEutelsat 1 -Fl, which will then beavailable to carry new services inaddition to acting as an in -orbitspare. ECS-5 is one of the threeEutelsat craft built by BritishAerospace.

Intelsat VIItransponders

The new Intelsat kit series ofcomsats announced last year(December 1987 issue. p.1232)will each carry 36 transponders.For C band there will be sixteen72MHz and ten 36MHz trans-ponders. For Ku band com-munications the spacecraft willprovide six 72MHz and four112MHz transponders.

These high -power satelliteswill allow the use of smallerEarth station antennas and willreplace the present Intelsat V -Acomsats at the end of their life inthe Pacific Ocean region. Theywill have the ability to switch Cband capacity between the east-ern and western hemispheresand will give simultaneous Kuband coverage of three regions inthe Pacific area which can bealtered while the spacecraft arein orbit. Later on, The first twospacecraft are expected to belaunched in mid -1992 or early1993. The penultimate Intelsat V -Aspacecraft, F-13, was successful-ly launched on an Ariane-2 rock-et in May this year and is sta-tioned at 53°W., The last in thisseries, F-15, is due to go up inJanuary 1989 (the F-14 havingbeen launched out of sequence).

Satellite Systems is written byTom lvall.


Piezoelectric coaxial cableNovel packaging should make piezo-rubber more usable

Most applications which require apiezoelectric transducer, for exam-ple underwater hydrophones or

pressure sensors, use a ceramic materialcalled PZT. (lead zirconium titanate). Whensubjected to a mechanical strain, this pro-duces a voltage of the order of 311V/Pa - ausefully high value that requires nothingmore sophisticated than a charge amplifierto buffer the extremely high source impe-dance of this capacitive transducer to ananalogue signal -conditioning circuit. Fig. 1 .

PZT has many advantages: it is chemicallyinert, it has a reasonably high sensitivity andit can he used over a very wide range ofpressures without serious non -linearity.There are no problems in making thisceramic in almost any shape or thickness atrelatively low cost.

The biggest drawback of the material is itsmechanical fragility: like all ceramics. PZT isbrittle and must not be subjected to impactor excessive flexure. In underwater hyd-rophone design. where maximum sensitivityis obtained by using a very long transducerassembly, mechanical flexibility is essential.so a large number of PZT elements are linkedelectrically within a long PVC tube. Thistube may be filled with an insulating oil toobtain the required overall specific gravityand minimize the support needed for thehydrophone in the water.

Such a hydrophone incorporates a greatmany lengths of wire and soldered jointswhich will be subjected to considerablewhiplash if used at sea -a labour-intensiveand potentially unreliable technique.

The technical ceramics division of NTK.one of the world's largest spark plug manu-facturers. supplies PZT transducers tooceanographic hydrophone constructors. Ina paper presented to the 7th Symposium onUltrasonic Electronics in Kyoto (piezoelec-tric transducers are widely used at ultra-sound frequencies), four NTK researcherspresented a paper describing a new compo-site material consisting of fine particles ofpiezoelectric ceramic material embedded ina synthetic rubber. Chlorobren.

By attaching flexible electrodes madefrom carbon -loaded Chlorobren, the resear-chers were able to show that this compositehas excellent piezoelectric properties, but iscompletely flexible. Further, the rubberused has good acoustic coupling propertiesfor signals in sea water, because the speed ofsound in both media is similar.

This table gives some of the relevantproperties of piezoelectric rubber comparedwith PZT and PVDF (polyvinylidine dif-luoride. a synthetic rubber with inherentpiezoelectric properties).Note that both the new composite and PVDF

Pl:'l'l;R Jot INsoN

Rf

Piezoelectrictransducer

-Vcc

Signal earth

Fig.1. Lead-zirconium-titanate is easy touse as a pressure transducer. It has arelatively high voltage output and needsonly a charge amplifier like this one toproduce a signal suitable fo- feeding toanalogue signal -conditioning circuits. Re-sitor Rf would typically be around 20M!!given an LM308.

Output

Nemorubber I'VD1, l'ZT-

Density 1.8 1.78 7.5 tonneSpeed of Si 0.9-1.8 lonisec

Dielectricconstant :to 8-11.5 1175 at IklloSensitivity ig;,1 150 $:t-282 2.$ niV.in/N

have similar sensitivities: however, since thevoltage generated by a piezoelectric trans-ducer is given by:

v=gtPwhere P is the pressure applied, g is sensitiv-ity and t is thickness of the transducer in thedirection of the pressure, electrical output ofa piezo-rubber device can be made greaterthan that of a PVDF transducer by increasingthe thickness: PVDF is not easy to manufac-ture except in thin sheets.

Underwater hydrophones are typicallyused in sonar systems to monitor the returnpulse of acoustic energy reflected from theseabed or other submerged object. This is asharp -edged pulse when transmitted. butwill normally show "clutter" after reflectionwhich gives the experienced sonar operatorsome indication of the type of surface fromwhich it was reflected.

With a PZT transducer, which has a verylow damping factor. this characteristicshape is masked by the ringing of thetransducer element itself. Fig 2(h). Therubber matrix material of the new compositeimparts a much larger damping factor to thehydrophone. greatly reducing ringing. Fig.2(a). PVDF. a ferroelectric material. alsoexhibits ringing oscillations when used todetect a single ultrasonic pulse.

PIEZOELECTRIC CABLE

Flexible materials are not easy to makecontact with, especially unsolderable sub-

400 -

300-

E

2C0-

100-

0 -I

0 1 2

TIME (us)

( al

1Ic-

800-

E00-E400

700-

o-1 1

0 1 2

TIME (us)

(b)

Fig2. Piezo-rubber (a) and PZT (b) tran-sient responses. Amplitudes are not direct-ly comparable due to different thicknessesof transducer.

PVC sheath Braided Piezoelectric(con be square screen rubber compositein cross section) ( PR -302

I j

Centre conductor

Fig.3. Piezoelectric coaxial cable is similarin appearance to ordinary 50i! coaxialcaole.

()versed

+10.zoO,e0P6p,itmo oo'160'e,.

.attPE.....?',-;°,

;a0,?°°40.7qoo4'.2t.9

Protective Piezoelectricex trus on coax

Sealant

Fig.4. Piezoelectric coaxial cable used as avehicle sensor requires a protective extru-sion, but it has a longer life than electricalcontact strips. Compared with electro-magnetic sensors, a piezoelectric cablesensor detects smaller vehicles and iseasier to install.

(10Af:oa.6:0Ny.--A:°`e 0 0. O%


stances like piezo-rubber. In existing ap-plications of this still new material, theactive element is sandwiched between layersof conductive rubber of similar elasticity:contact is then made to the conductiverubber by means of beryllium -copper leafsprings.

Clearly, this is an inconvenient method ofconnection, so NTK investigated more ver-satile techniques which did not require themechanical precision of earlier methodssuch as the one described. The result of thiseffort was a coaxial cable, similar in appear-ance to normal 5011 coaxial cable, but withthe insulating layer between the screen andcentre conductor made of a piezoelectricrubber composite designated PR -302. Fig.3.This cable can be made in continuouslengths of up to 20m and is compatible withsoldering or with standard coaxial plugs,making it much easier to use.

Electrical properties of this cable are asdetailed in the table above.

APPLICATIONS OF PIEZOELECTRICCOAXIAL CABLE

This form of the material has only been inproduction for a short time, so there are, forinstance, no hydrophone applications so far.However, the cable has been used in Japan asa non -contact self -screening pickup for anelectric guitar, and research is being carriedout into its incorporation into a blood-pressure/pulse-rate monitor.

As a vehicle sensor for traffic lights, thecable was found to perform well because ofits resilience and high sensitivity. Suchsensors receive severe battering so the cablewas protected by embedding it in toughsynthetic rubber extrusion, shown in Fig.4.Compared with vehicle sensors relying onelectrical contact strips, this method has alonger life and is insensitive to the ingress ofrainwater.

Electromagnetic sensing systems - themost popular alternative - are reliable buthave problems detecting cyclists and mopedriders, and require a large area of the roadsurface to be removed for installation. Thepiezoelectric cable sensor needs only a nar-row transverse trench to be excavated bypneumatic tool.

Piezoelectric coaxial cable is still, to someextent, an interesting solution looking for aproblem. It is here, in an easily -used form.and available off -the -shelf to the develop-ment engineer. It remains to be seen whatthe ingenuity of R&D engineers will make ofit.

Further reading

Hisao Banno. Kohji Ogura. Hideo Sobue & KanjiOhya: Piezoelectric and Acoustic Properties ofPiezoelectric Flexible Composites. Japan JournalofApplied Physics. Vol.26 11986).

RY Ting. Ferroelectrics Vol.67 p.1-13. 1986.

H. Ohigashi. K. Koga. M. Suzuki et al. Ferroelec-tricsVol.60 p.263.1984.

Peter Johnson is with Quantelec inHenley-on-Thames.

WIRF1 %ORLI) & WIREI IRS NORLD & WIRELESS WCIRLD WIRELESS VIORL

three year subsc lectronics & Wireless World costs only £53.delivered free to the address of your choice; that's a massive saving of

almost 25% - ti - equivalent of eight issues free!

NM mi EN im Ns NESUBSCRIPTION ORDER FORM

El Please send me Electronics & Wireless World for three years at the special price of £53.00''Includes posrage and Pa,,'"17

I enclose a cheque/PO to the value of

Please debit my credit accountAccess Barclaycard/VisaAmerican Express Diners Club Expiry date

NM MI

made payable to Reed Business Publishing Limited

Name Signed

Address

Please return to Electronics & Wireless World. Room L302 Quadrant House. The Quadra,Sutton. Surrey SM2 SAS

III IMO MI IIIIII MI MN NM 1E1 IM alb


INTERNATIONALBROADCASTING CONVENTIONBRIGHTON UNITED KINGDOM23 - 27 SEPTEMBER 1988

jaiINTERNATIONAL

BROADCASTING CONVENTION

FURTHER INFORMATION can be obtained byreturning the reply coupon below.

The 1988 IBC Technical programme will cover allaspects of broadcast engineering with particularemphasis on emerging technology including satelliteand cable distribution, enhanced and high definitiontelevision systems, as well as multi -channel soundsystems and associated information systems.

The IBC EXHIBITION complementing the technicalsessions will have the latest professionalbroadcasting equipment on display and demonstrationby leading world manufacturers.

The SOCIAL PROGRAMME during the Convention willinclude a Reception and a special Ladies Programmeof talks and demonstrations and visits to places ofinterest.

The IBC Secretariat, Tte Institution of ElectricalEngineers, Savoy Place, London, United KingdomWC2R OBL. Telex: 261176 Telephone: 01-240 1871

Please send further details of IBC 88 to:

Name & Position

Company/Organisation

Address

Post Code


-TELECOMMS TOPICSTowards ISDN- 2

Adrian Morant continues his introductionto integrated services digital networks.

ISDN basic access consists of two 64kbitis Bchannels for user information together with a16khit/s I) channel. While the former arecircuit -switched. the latter is a packet -orientated channel whose main purpose is toexchange signalling between terminal equip-ments and the local exchange.

Under normal circumstances. when used forvoice, once the call has been set up the Dchannel is free until the call is to he terminated.For this reason. while the basic access consistsof 2B+ D. at the next level in the hierarchy(primary rate access or PRA) there is only one Dchannel with either 30 or 23 B channels. Thisdepends. respectively. on whether the system isoperating to the CCITT or the American "Bell"standard: the former uses a 2Mbit/s transmis-sion channel while the other uses 1.544Mbit/s.

From this it can be seen readily that a Dchannel is shared between a number of Bchannels. This is a "common channel," and thesignalling it carries is therefore common chan-nel signalling. It forms an essential part of thenetwork and provides control and supervisoryinformation regarding the routing and setting -up of a call as well as status information.Consequently, when attempting to initiate anISDN call. the calling terminal will use thesignalling channel to make its request and willuse the CCITT Common Channel SignallingSystem No.7 - often just referred to as C7. Notonly must this request cite the number of theparty being called, it must include a variety ofother information. For example. if the proposedcall is just voice from one digital telephone toanother. only one 64khit/s B channel would beneeded. If. on the other hand, the required callwas between two combined voice and dataterminals. two B channels would be requiredspanning the entire end -to -end path. Differentdevices require different network resources: thecalling party must therefore initiate negotia-tions to ensure that it will obtain the necessarychannel capacity.

OSI Protocols

The D channel signalling protocol consists ofLayers 1 to 3 in the ISO Open Systems Intercon-nection (OSI) Reference Model. In Layer 1. theelectrical and physical characterstics of the Sinterface are defined. In addition. a multipleaccess mechanism is specified which allows theD channel to he used by a number of terminals.

Layer 2 of the I) channel protocol (LAP I)) canhe considered to he an adaption of the LAP Bprocedure defined in CCITT recommendationX.25 to the specific needs of ISDN basic access.Owing to the bus structure of the S interface.LAP I) has to support multiple simultaneousLayer 2 connections. Additionally. LAP D en-ables the higher layers to broadcast a message toall terminals connected to the S bus.

Layer 3 defines procedures for controlling the

supplementary service facilities as well as sig-nalling procedures for the basic call handling,e.g. call establishment and release for B channelconnections.

1Vere the D channel used only for signalling itwould frequently be idle. Consequently it canalso he used for the transmission of packets ofdata. The transmission of data by packet switch-ing makes for far more efficient use of re-sources. Instead of a circuit being allocated andheld exclusively during a message transfer it canbe time-shared between many.

Access to a packet network is via a packetassembler/disassembler (PAD): this is reachedat present via the public switched telephonenetwork. but in due course it will he via theISDN. Initially it will use a B channel - beinganalogous to a PSTN voice channel - buteventually the]) channel will be employed. Thiswill mean a more efficient use of resources.After all, the D channel is transparent to thepackets being sent and is not concerned withtheir content. It will transfer them as requiredirrespective of whether they are signallinginformation or actual user data.

In packet switching the message is transmit-ted as a series of discrete blocks or packets. Inaddition to the data itself, each packet mustcontain address information to indicate itssource and destination. It must also include asequence number to enable the received mes-sage to be re -assembled correctly. This isanalogous to a series of postcards which eachcontain part of an entire message. By markingthem with a serial (or sequence) number it ispossible to ensure that they are in order and sodetermine whether any have been lost in thepost. Should this occur, the recipient wouldrequest the sender to repeat the lost informa-tion.

Whereas in circuit switching a path throughthe network is "transparent" to the user once acall has been set up. in packet switching thenetwork must supervise the whole operation.

To send a message from a character -orientated terminal. such as a teletype or adumb v.d.u.. it is necessary to dial in to a PADand then log -on with one's Network UserIdentification (NUD and password. The PADmaintains control and integrity of individualmessages and integrates them with otherswhich are also within the system. Havingobtained access to the network. the user speci-fies a destination address for the message andthen inputs the message. The PAD takes themessage and breaks it into a series of datapackets. To each of these packets it adds framinginformation and then the whole packet is

launched into the network. Both header anddata portion of the packet have a checkwordattached, to enable the integrity of each packetland thus of the entire message) to he verifiedon receipt. Should an error he detected thereceiving end will request re -transmission.

Skyphone takesto the air

Europe's first public satellitetelephone call from an aircrafthas been carried out by Racal andBritish Telecom International.The test link -up took place he-tween Racal's company Jet -stream aircraft and BTI's satelliteEarth station at Goonhilly inCornwall.

The in-flight tests are pavingthe way for a trial commercialservice for passengers travellingon British Airways Boeing 747transatlantic aircraft later thisyear. This will he followed by aregular manually -connectedtelephone service and subse-quently by a ioil commercialdirect -dial Skyphone serviceavailable to all airlines by mid -I989.

[sing the Inmarsat satellitenetwork. Skyphone will offerpassengers a direct -dial tele-phone service to 185 countriesworldwide. as well as facilities forthe transmission of aircraft oper-ational telephone and datatraffic.

The equipment is designed tooperate to Inmarsat specifica-tions in the 1.5-1.6Gliz I. hand.The low noise amplifier is housedin a small package to permitmounting on the inside of theaircraft skin close to the receiv-ing antenna. It has sufficientgain to allow inexpensive andking cable runs to the radiofrequency/intermediate frequen-cy unit mounted in the aircraftequipment racks. without im-pairing receiver noise figure.I..n.a. noise figure. includinglosses from pre-l.n.a. filters. is

lower than 2.5d13 even a1 aircraftskin I emperatures C.

Research intopersonal

communicationThe Government has announcedthe eighth 'Link' programme.which aims to stimulate the pre -competitive research towards anew, higher capacity, third generation personal communica-tions system. Its contribution tothe three-year Personal Com-munications Programme is

£6.35 million £3.35l from theDTI and £3M from SERC.


-411LECOMMS TOPICSCompanies which have indi-

cated that they would like to heinvolved in this programme in-clude Aerial Facilities. Auto-phon, British Aerospace. BT Re-search Laboratories. GEC. Mar-coni, Philips. Plessey Research,Racal. STC and Thorn EMI. In-terested academic institutionsinclude the following universi-ties: Aston. Bangor. Bradford.Bristol. City. I leriot-Watt. Leeds.Liverpool. Southampton, Sur-rey, Swansea and York.

Lord Young. Trade and Indus-try Secretary, speaking in theHouse of Lords, said: "The needfor the programme is promptedby the convergence of telecom-munications and radio technolo-gies. which is generating a newindustry of personal com-munications. The number ofsubscribers is increasing. withgreater demand for accessibilityboth indoors and outdoors -whether stationary or mobile -and internationally.

"There is therefore a need toextend the range of servicesavailable and to integrate privatemobile. cellular and satellitemobile systems. Much basicwork needs to he done to achievethis. It will he a key technologicaldevelopment for the 1990s."

Flexible AccessSystems

SIC has been appointed one ofthe suppliers for the nationwideextension of British Telecom'sFlexible Access Systems (FAS).for completion by the middle of1991. SIC expects to receiveorders in excess of .ti I1)0 million.

FAS is based on the use ofoptical fibres rather than coppercable connections to businesscustomers. This will allow largenumbers of private circuits to heprovided. and the evolution towide bandwidth services such asvideo conferencing in the future..\t the customer's end, a multi-plexer connects the various cir-cuits on to the fibre. These mul-tiplexers are controlled by net-work management softwarewhich enables the mixture ofconnections to he varied to ser-vice customers' changing needsand to provide connection to anyother location required.

The first Flexible Access Sys-tems have been successfully de-ployed in the City of London

Fibre Network, for which SICwas the prime contractor, andSIC is also supplying equipmentfor part of British Telecom's Lon-don Docklands project.

STC claims to have been thefirst company to appreciate thestrategic possibilities of distri-buting intelligence in the net-work.

Flexible Access Systems willextend to local communicationsthe benefits of the major mod-ernization programme that Brit-ish Telecom has been undertak-ing in the trunk network. SICand its partner Northern Tele-com have established a 200 -strong team to work on thedevelopments and products re-quired. This team is deployed inthe UK and North America.

SIC and NT are also jointlyworking on the extension ofoptical fibres to small businessesand residential customers to pro-vide high quality telephony ser-vices, and the future possibilityof television -based services.Another joint programme onsynchronous multiplexers andswitches will allow circuits of anybandwidth to he carried efficient-ly ar, pond the whi,le network.

Payphones upthe Canyon

Visitors to the Grand Canyon cannow dial honie from remoteareas of the park following anagreement between US West Inc.in Arizona and the British com-pany GPI Telecoms Products.

Under the agreement. GPT hasinstalled the first phase of intelli-

gent payphones throughout theGrand Canyon National Park.These payphones connect to avariety of transmission lines andcarriers.

GPT payphones generate coinaccounting reports and they dia-gnose malfunctions and attemptto clear them. 1Vith the heap of adot-matrix display and voicechip, the telephones also com-municate in two languages.which will help some of thethousands of international tour-ists who visit the park each year.

"We are currently investigat-ing ways to extend even moreservices to communicationpoints such as the Grand Can-yon", said Peter Brown, manag-ing director of GPT TelecomsProducts. "We can foresee thetime when we will be providingdirect international dialling."

Third WorldTelecoms

By the first anniversary of thestart of its field activities, theCentre for TelecommunkationsDevelopment ICTD). establishedin 1985 by the ITU Administra-tive Council in Geneva. had re-ceived some 40 requests forassistance from developingcountries.

As the Centre employ:. only asmall core staff at its headquar-ters. the major part of its work inthe field has so far been carriedout by experts and consultantswhose services were obtained bya combination of cash contribu-tions and support in kind.

CTD's work is in preparing

preliminary documentation forrural telecommunications pro-jects in several developing coun-tries and of telecommunicationdevelopment master plans inothers. including some regionalprojects. This groundwork willenable the countries to submittelecommunication projects forfinancing. Some of the countriesconcerned are Burma, Gambia.Tanzania. Somalia. Egypt. Nep-al. Malta. the Yemen Arab Re-public, and People's DemocraticRepublic of Yemen.

The Centre relies entirely onvoluntary contributions fromboth the private and public sec-tors in the developing and in theindustrialized parts of the worldwhich stand to benefit from theCentre's activities. To date,pledges amounting to about Sw.Fr. 5 million (very little in view ofthe huge task) are 60% in cashand 40% in kind for the currentyear 1988. broadly distributed asfollows: 45% from industry. 45%from administrations and therest from development financeagencies. Several developmentfinance institutions. as well asdeveloping countries them-selves. have so far respondedfavourably.

Twelve missions have beencarried out so far and six moreIto Bolivia. Chile. China. Iran.Mozambique. Nigeria) are sche-duled during 1988. It is expectedthat these missions will lead toadditional requests for projectsand specific assistance.

Teleconuns Topics is compiledhy.ldrian Morant.


Keyboard designFiner details of keyboard and keypad design are often

overlooked by design engineers.

There are hundreds of keyboard applica-tions, not only in computers, but alsoin typewriters, telephones, remote

controllers, washing machines, microwaveovens etc. In many of these applications asingle -chip microcomputer, or controller,reads the keys and acts on the data entered soa good understanding of keyboard -to -micro-computer software and hardware interfacetechniques is essential for anyone involvedwith keyboard design.

There are two main types of microcom-puter -based keyboards. In the first type,found for example in telephone and washingmachine controllers, keyboard reading isonly a part of the microcomputer's function.Alternatively more complex systems, such ascomputers and typewriters, often have adedicated microcomputer controlling onlythe keyboard. These applications mightseem quite distinct but the software andhardware techniques for any size of keyboardare similar.

KEYBOARD TYPES

There are several different keyboard con-figurations. each with its advantages anddisadvantages. The simplest type ofcontroller -based keyboard has one keywhich is assigned to each line of an inputport. Fig.1. Keys can be read very quicklyusing this method, and the software neededis trivial, but it can only be used where thereare many spare input lines. Direct keyreading is therefore restricted to smallkeypad applications.

Much more common is the matrixkeyboard where the keys are arranged inrows and columns. This method makesmuch more efficient use of the microcompu-ter's input/output lines, but it requires morecomplex software for scanning the matrix.Scanning is normally accomplished by pro-ducing an output pulse on one column at atime and reading in all the rows to determinewhether a key in that column has beenpressed. For a given number of i/o lines. thenumber of keys in the matrix is maximizedby arranged it as a square: for example an8 x 8 matrix reads 64 keys whereas a 4 x 12matrix reads only 48.

Roll-over of a keyboard is defined as thenumber of keys which can be pressed at anyone time and still he individually recognized:if all keys can be pressed at once andrecognized the keyboard is said to have n -keyroll-over. This is the most desirable solutionbut is more expensive than simpler matricesfor reasons explained later.

Figure 2 shows a basic 4x4 resistivematrix. In the case of a resistive matrix eachkey has a contact resistance of the order of500 when pressed and a potential divider

JEFF WRIGHT

effect is thus produced. This matrix is said tohave two -key roll over because if any threekeys in the corners of the an imaginarysquare or rectangle are pressed then the keyin the fourth corner will appear closed aswell la ghost key) and it is impossible to tellwhich three keys are actually pressed. There-fore the maximum number of keys that canbe pressed without confusion is two.

Although this matrix is defined as havingtwo -key roll-over it would be more accurateto say that it had a minimum of two -keyroll-over as often many more keys can bepressed before a ghost key is produced. Forexample, all keys on one row can be pressedat the same time and still be successfullydetected. It can therefore be advantageouswhen designing a keyboard of this type tomake the matrix as rectangular as possiblewith the maximum number of output scanlines and also to arrange the layout so thatthe keys which are most commonly pressedtogether are on the same row.

A drawback of making the matrix rec-tangular is that more i/o lines are requiredfor the same number of keys, but for manyapplications the gains obtained make theextra lines worthwhile.

Another enhancement which can be madeto the basic matrix is to give important keys.such as 'shift' on a computer keyboard, theirown input line. This then gives the two -keyroll-over minimum and additional specialkeys.

Software for this matrix is actually themost complex since to ensure correct opera-tion, ghost keys have to be detected. Also, i/oswitching is needed to scan the matrix toavoid contention caused by a scan pulse ofone logic level being connected via pressedkeys to another pin outputting the oppositelogic level. This latter point also means thatpull-up resistors are required on both sidesof the matrix as shown to avoid floatinginputs (these pull-up resistors also act ascurrent limiters).

When a ghost key is detected the keyboardis locked to prevent the false key from beingprocessed, but the designer then has anoption as to when to allow processing tobegin again. One option is to start proces-sing keys once the ghost key has beenreleased but this can lead to key strokesbeing missed -a problem for the fast typistwho may frequently have several keys press-ed together momentarily. The safest optionis to delay processing until all keys arereleased - a warning of the delay could begiven to the user if desired.

This basic matrix is used extensively inlow-cost computer keyboards, terminals andmany keypad type data -entry systems wheremultiple -key presses are uncommon. Try

MCU input port

Fig.1. Using one controller input for eachkey results in fast key -press reading andvery simple software, but this does notmake efficient use of controller lines.

Col

on

// // //Syr4 4/ // //4 //,, /

1-',/1/ // // //i // // 3 /Iy/Tie 2

mns/o port

f I I 1

z-3V V V

Rows on input port

Fig.2. Arranging keys in a matrix makemore efficient use of controller ports -only 8 lines are needed to read 16 keys.This produces two -key roll-over sincepressing more than two keys at once canresult in an erroneous reading e.g. press1,2,3 and 4 appears as a ghost key.

I

4 /

Columnson output port

/ ///

///

//y)e"

V V V

Rows on input port

Fig.3. Adding diodes to the matrix stopskeys interfering with each other, produc-ing n -key roll-over.


your PC keyboard for quality: some don'teven have ghost detection so that by pressinga certain three keys you get four characterson the screen!

N -KEY ROLL OVER

In higher quality computer keyboards and inapplications where a large number ofmultiple -key entries are required, an n -keyroll-over matrix is used. There are twopopular methods of achieving an n -keymatrix hut both are more expensive tomanufacture than the hasic matrix type.

The first and most common method is toadd diodes to the hasic matrix. Fig.3. Thesediodes prevent any one key from interferingwith another. thus producing n -key rollover.Voltage drop across the diodes is an impor-tant hardware constraint when designing amatrix of this type since it can cause viola-tion of the controller port input thresholds.Excessive voltage drop can normally heavoided by choosing the correct value ofpull-up (current -limiting) resistor hut inlow voltage applications it can be more of aproblem.

The second main type of n -key roll-overmatrix has capacitive keys Fig.4. This con-figuration gives n -key roll-over without thecost of the diodes but with the additionalcost of comparators. Each key forms theplates of a capacitor and when the key ispressed the plates move closer, increasingthe capacitance and allowing a short low -voltage pulse to reach the comparator input.Impedance of the other scan lines and thenumber of keys pressed also affects the levelof this pulse. Pressing a ghost key producesthree capacitors in series, which reduces theeffective capacitance. thus degrading thepulse which would appear as the fourth keysignal.

Mien a key is pressed the comparatorrestores the incoming differentiated signalto a true logic level, so producing an outputpulse which is detected by the controller.Since this pulse is normally shorter than thescan pulse, care has to be taken in thesoftware not to miss any keys by reading toolate. Also, the comparator reference must beset up carefully so that the low -amplitudesignal can he detected reliably, and thebiasing circuit's impedance should be largeso that the time constant for the key press islong enough to produce an acceptable pulselength for detection.

With both these n -key matrices the soft-ware is simpler than in the previous two -keyroll-over case, since the scan lines canremain as outputs all the time and ghost -keydetection is not required. They also avoid therequirement for pull-up resistors on theoutput scan lines -a minor cost saving.

KEY BOUNCE

In all keyboard types there is another vitallyimportant criteria - key bounce. Keyboardscan often he installed in very noisy environ-ments and the scanning of a large keyboardwith long p.c.h. tracks can produce its ownnoise and cross -talk problems. For thesereasons p.c.h. layout and software dehounc-ing must be handled carefully.

One common method of dehouncing is to

I

Columnsoutput ;;

\se.i -

// // ///

- / // // // -LI// // // // T1-1

;// // // se

.(>,

Camper

.... .. _-L.

Input tocomparator

-I I -

tor outputli-ey pressed

Key released

-r1

Rows on input port

Fig.4. Capacitive keys and comparators prcduce n -key roll-over without the use of diodes.

Up

Key state

Down

High

Port pin

Low

Debounce time11

Keyboard scanned here-key press detected

lot Valid key press

Second scan-key press confirmed

1.6

Debounco time

First scan here- such o noise pulse could be

interpreted as a key press.

A

Second scan- false key press

rejected

lb) Noise rejection

Fig.5. Contact bounce problems are commonly solved by reading a key's state twice in

quick succession and comparing the readings.

EATURES IMPORTANT TO KEYBOARD DESIGNERS

Bid.rectional i .o lines are suitable for both two -key and n -key roll-over type keyboardsand

available in numbers sufficient for any application from 4 - 4 matrix to a 120 key or largerdesigns. Selectable pull-up down resistors on some devices are an added bonus.

Timers can be used to determine a fixed polling rate or, in the case of a computer keyboard.used to provide timing for the auto -repeat featire. Timers can also be used for the datatransmission.

High current ports are a prime requirements for many control applications. On computerkeyboards they can be used to directly drive leds which are often required to indicate thestate of certain keys such as the lock for capital letters. Two bits can also be used for serial

line driving.Low power consumpticn modes and wake up features are important for portable

applications such as remote control. telephones and portable computers.Serial Interfaces can be used for the serial Link between keyboard and computer or to

incorporate a magnetic card reader or mouse.-

Large ram holds matrix images for debouncing and forms buffers to hold characters to besent when required. On board eerom could be used to provide user programmable function keys on a computerkeyboard independent o' the actual computer. Only an additional program key on thekeyboard would be required.

ELECTRONICS & WIRELESS WORLDtill

read a column, store it. and then read thecolumn again. after the anticipated bouncetime. Fig.5. When the column is read for thesecond time it is compared with the firstresult and if there are any differences thenthose keys are considered to be in theunpresssed state. A genuine key press will hedetected on subsequent scans.

The problem with this method whenapplied to large keyboards is that theaccumulated debounce times from eachcolumn check can produce an unacceptablyslow scan rate. A simple way around thisproblem is to scan the entire matrix in onego. storing it in ram. After the debouncetime the matrix is scanned again with eachcolumn being compared to its stored imageas before. This can use a lot of ram depend-ing on the size of the matrix and so acompromise between scan rate and ramusage can be obtained by debouncing thematrix in sections.

Up to now it has been assumed that thekeyboard will process any key that is found tobe pressed, hut on some keyboards a key isonly processed when it has changed state sothat no repetition takes place. Alternativelysome computer keyboard protocols send akey break code when the key is released.

Both of these situations complicate thedebounce and key -handling software sincean image of the entire matrix has to bemaintained in ram so that the present stateof the keys can be compared with thatobtained on the previous scan. Figure 6shows the actions taken by a computerkeyboard depending on the type of operationand the states of the keys.

With complex keyboards a good method ofdebouncing is to keep a new and an oldimage of the keyboard in memory (called.say. NEWBUF and oiniii,r). Then in the scanroutine the entire keyboard is read intoNEWItIT and after a desired debouncing timethe matrix is scanned for a second time. Onthe second scan, as each column it is readand compared with the first scan result. Itany of the keys have changed state they areconsidered to have bounced and the oldvalue of that key is extracted from ouniiT andinserted into NIAtlit'F. In this way only keystates which are valid for the two scans areaccepted. Later in the program NFA1111T andounitiF are compared. the valid key changesare processed, and °miff is updated ready forthe next scan (details later).

CHOOSING A CONTROLLER

In most control -type applications the choiceof microcontroller will depend largely on theother functions that the device has to per-form, with the keyboard being a secondaryconsideration. Despite this there are a num-ber of features available on some deviceswhich can be a great help to the designer.

Shown in Fig.7 is the MC68HC04.13.which has two features which make it parti-cularly suited to control applications incor-porating small keyboards. The first of theseis the option of internal pull -down deviceson all or some of the port pins, and thesecond is a wired -Or interrupt option. Thissoftware controlled feature allows the deviceto be brought out of its low -power WAIT andsroi, modes or to be interrupted from run

Keyboardtype

Old keystate

New keystate

Actiontaken

Simpleauto repeat

X

X

On

Off

Send key code

Do nothing

Nonauto repeat

Off

Off

On

Off

On

X

Do nothing

Send key code

Do nothing

Complexauto repeat

Off

Off

On

On

Off

On

On

Off

Do nothing

Send key code

Auto repeat

Send break code

Fig.6. Software in controller -based keyboards can produce three different types ofoperation.

keyboardmatrix

Fig.7. A dedicated controller as opposed to a microprocessor simplifies keyboard designsince it has i o lines that can be connected directly to the keys. Some controllers alsohave internal pull-up or pull -down resistors and internal logic that produces an interruptwhen a change of i o line state is detected.

mode when a key is pressed. thus eliminat-ing the need for continuous polling.

Quite often there are not enough i/o lineson the controller chosen to handle all thecontrol functions plus a keyboard. but thereare ways round this problem by sharing orexpanding input/output pins. In Fig.8(a). forexample the MC6805 forming a tv controllerdoes not have enough i/o lines to drive boththe display and keyboard. and cost preventsthe use of an external display driver. Thesolution to this problem is to multiplex thetwo functions on the one i/o port. Three portlines A1, disable the display while thekeyboard is scanned and the high -valueresistors are to prevent shorts betweensegments caused by pressed keys.

Figure 8(h) shows a situation where alow-cost controller is being used to scan afull computer keyboard. Again, this partdoes not have sufficient i/o lines to scan thekeyboard and so an analogue multiplexer isused to compensate. Note the dual use of themultiplexer address lines as inputs for spe-cial keys that need n -key operation.

Mass produced computer keyboards needto be made as cheaply as possible. It could beargued that a computer keyboard with itsown intelligence is not justifiable, but de-signing a keyboard with its own processorbrings several advantages.

In systems with a separate keyboard aserial link between the processors improvesreliability and saves on cable cost and bulk. Aseparate processor also allows data to beinput independent of the main processor -abuffer in the internal ram -of the keyboardcontroller can store key inputs to be trans-mitted to the main processor when it is free.On power -up an independent keyboard pro-cessor can perform checks on the keyboardto detect stuck keys. etc.. while the mainprocessor checks out the rest of the system.

In the past it was common to find micro-computers with external memory in compu-ter keyboards but today the single -chipmicrocomputer is almost universal in newdesigns due to its inherent cost advantagesin control applications. A list of featuresfound on various controllers which are of


Fig.9. Most of the design work for a PC -compatible keyboard using a controller isdone by the software writer.

use to the keyboard designer is shown in thepanel: some of them apply only to computertype keyboards. others to any type.

DESIGN EXAMPLE

This example forms the basis of a PC -compatible keyboard and has been chosenbecause it demonstrates most of the princi-ples described previously. Figure 9 showsthe keyboard implemented using a low-costMC6805U3 with mask -selectable pull-up re-sistors on all ports. It has 100 keys arrangedas an 8x 12 two -key roll-over matrix plusfour special keys, it also incorporates statusleds and a serial link to send characters tothe main computer. Software shown in List1 is 6805 assembly language for scanning an8 x 8 section of the matrix: this allows theprinciples to be shown clearly while keepingthe software to a minimum. Note thatwherever possible the number of row inputsto the controller should he kept to a multipleof four as this simplifies the scan softwareconsiderably.

The ram required for these routines con-sists of two buffers used to hold images of thematrix. Each has the same number of bytesas there are columns in the matrix plus onestatus register and two temporary storagebytes.

At the start of the program would be aninitialization routine that sets up the ports

4!

12V

BJ

8.220

.------

%A,tiw8x2k2

NAN

/2' Main matrix

I I I I I I I I I I

pedal key

INT

Porto 0 Po-tA 0 7 Port

MC 6805U3 with internal pull-ups

Porte 0 Vss Vcc R EXTAL XTAL NUM TIM

t5V

I I I

Clock Data Spore

StatusSerial l nk indicators'

5V

Ceramic resonator

SHFTp

Doto

Clock-1\AA

OV

CA 11 1,1

NUM

Vd d

TIM

Reset

Pal

INT

TIT

1PC, PC -

PAC ST

r-FT 1E7-

B.

C.

V-

B-

S N

PE3

PB2

PBO

PC0

PC3

MC 613HCO4P3

with internalpull -downs PA

EXTAL

XTAL

MDS

V55

'

16

9

SHF IL10

PA0

PB7

P84

A LTp11

ALT}

MC1 4 0 S 1

3

15

12

1

5

2

4

Bx12 2 -keyrollover matrix

Fg.8. In the tv controller of (a) below,reading key presses is only part of theniicrocontroller's job whereas in the com-puter keyboard of (b), the controller's onlyjob is to read keys and turn the decodeddata into serial form.

Vd

Vss

Din

a

MC 144111 1:12

0 -tc-o a3

CL 04

BO e7

A0AlA2

Vcc MC 6805 T 2Reset

NUM

A4Ay

Ac.

0

VssINT EXTAL XTAL A Co C1 C2 FIN

Ilipre amp

for remocontra! LI

Mute' I

selection

Standard TStandby

V'tv4x 47k

Contrast

Brightness

Colour

Volume

SCL SDA Vd d

M C M 2814

VCprop store

CSO [Si M Vss

1111 Tuner

atom Dotoout VDR CL

NEG

OUT 5ND

UAA 2001Loop filter and

bond driver

B1

B2

B3

1 L Vc 2 VccP OS t 84

ri 30V 12V

A/V

Uhf

III


and buffers, etc. This section is not shownhere as the software is trivial and controllerdependent. The main loop of the programconsists of a scanning subroutine, sc %mut!), aghost detect subroutine, cmcfiosT, and a

routine, SENDKEY, that decodes the keychanges and sends the data to the computer.updating the 01.1)111,F in the process. This lastroutine is only shown in skeleton form as itis unimportant for this example.

Routine scANKB1) scans the matrix twiceusing i/o switching and also detects andcorrects bounce. It begins by setting a flag toidentify which scan pass is being executed.Port A data register is cleared as is TEMPI

which is used as an image for Port Adata -direction register bum (this is necessarysince certain commands cannot be executedon the port data -direction registers).

BRO of TEMPI is set and the byte is writtento ma forcing port A bit° to output a low:column 0 of the keyboard matrix is then readin vai Port D in the subroutine RE NI Oil.. ValueTEMPI is then shifted left and if it has notoverflowed then it is written to Ifllk\ thusshifting the scan pulse to the next bit of PortA. this continues until TEMPI overflows. Afterthe scan the loop flag is tested to see if it isthe first or second scan that has beencompleted. If it is the second then the scanroutine is exited: otherwise the loop flag istoggled, the required debounce delay ex-ecuted and the keyboard scanned for asecond time.

During the first scan the subroutine READCol, simply reads the column in via Port D.stores it in NEWIIIIF and increments the bufferpointer. In the second scan however RE.wcoiis a little more complex as it checks for keybounce. As each column is read an exclusive-

Or operation is performed with it and theresult from the first scan to see if anychanges indicating possible bounce haveoccured. If there are changes then the resultof the exclusive -Or operation is used as amask to replace the "bouncing" keys withtheir previous state (held in oumur).

If a valid key state is wrongly interpretedas bounce it will he picked up next timeround the main loop. Note that if an n -keyroll-over matrix were used instead of the oneshown then SOWS!) could be simplified, asshown at the end of the software listing.since no i/o switching would be required.

Also of interest is the ghost checkingroutine cIIKGiIosT (not required for n -keyroll-over matrices). The detection techniqueused relies on the fact that when a ghost keyis produced there will always be two columnswhich show two or more keys pressed in thesame row positions. Thus if two columns areequal and contain more than one key pressthen a ghost -key exists.

Each column in the matrix image NB% Iis only compared with those residing above itin memory to avoid time -wasting duplicatecomparisons. Temporary register TEMPIstores the contents of the column that isbeing compared to the others and TEMP holdsthe pointer value for the same column. Asdiscussed earlier, once the ghost flag hasbeen set it is only cleared when all keys arereleased (not just the one causing the ghost).

Jeff Wright is with Motorola in East Kilhride

List 1. Example software in 6805 assembly language for scanning 8 8 rows of thePC -compatible keyboard.

ORG RAMBSENEWBUF RMB COLS

ENDBUF EQU

OLDBUF RMB COLSFLAG RMB 1

TEMP RMB I

TEMPI RMB 1

ORG ROMBSERESET EQU '

JSR INIT

New and Old matrix imagebuffers1 byte in each for eachcolumn

Status register2 temporary storageregisters

Set up ports initialisebuffer and status reg etc

MAINLOOP EQU 'JSR SCANKBD Go scan keyboard twice

store in ramJSR CHKGHOST Check image in ram for

Ghost keysBRSET GHOST. FLAG if detected then wait for

MAINLOOP keys to be releasedJSR SENDKEY Go and process any valid

key changesBRA MAINLOOP

NXTCMP LOX TEMPINX

CPX ENDBUFBLO STOREX

ALLOFFGHOSTEX

SENDKEYTSTCNG

LDAINCABNE

BCLRRTS

TEMPI

GHOSTEX

GHOST.FLAG

Check if we havecompared every columnto every other if not,update pointer and doloop again

If all keys are releasedthen TEMPI shouldcontain FFSo clear ghost flag.

EQUJSR TSTCHNG Look for differencesBNE SENDKY between NEWBUF

OLDBUF. if none foundreturn with Z 1

RTS No key changes toprocess so return

SENDKY JSR TSTKEY Process each key changein this column and updateOLDBUF with its new state.

BRA TSTCNG Look for next column witha change in it

Modified scan routine for an N -Key rollover matrix

SCANKBD EQU

BSET LOOPFLG. FLAGLDA SFF

STA PORTA Set portA all output high.SCAN CLX

BCLR 0.PORTA Pull first scan line lowSCANA BSR READCOL As before with 2 keyroll

over versionLSL PORTA Move scan pulse to next

columnBCS SCANS If not all done yet then

continueBRCLR LOOPFLG. FLAG Rest same as for 2 key

OKRTN rollover versionBCLR LOOPFLG.FLAGBSR DEBNCEBRA SCAN

OKRTN RTS

SCANKBD EQU

BSET LOOPLG.FLAG Initialise loop flag.SCAN CLRX

STX PORTA Clear data register andSTX TEMPI Data Direction image

registerBSET 0 TEMPI

EQULDA PORTD Read selected columnBRCLR LOOPFLG FLAG If second scan then go

VERIFY check for bounce otherNEWBUF. X wise store column

Increment the bufferpointer

STAINCX

RTS

CHKGHOST EQU

SCANA LDASTA

BSR

LSL

BCCCLRA

TEMPI Switch selected column toDDRA output so producing a low

scan pulse.READCOL Now read in that columnTEMPI Move scan pulse to next

columnSCANA If not all done yet then

continue else return alllines to input.

STA DORA

BRCLR LOOPFLG. FLAG 112 scans complete thenOKRTN return

BCLR LOOPFLG. FLAG Clear loop flag.BSR DEBNCE Wad for desired debounceBRA SCAN time then scan keyboard a

second timeOKRTN RTS

READCOL

COLINC

VERIFY FOR NEWBUF.XBEQ COLINC

STA TEMPCOMAAND NEWBUF.XSTA NEWBUF.XLDA OLDBUF.X

Compare with first scanvalue if no differences(bounce) then return.

otherwise manipulatescan results with old valueso that only key changesthat were valid for bothscans are included theothers are returned totheir old value

AND TEMPORA NEWBUF.X Equation for NEWBUF isSTA NEWBUF.XBRA COLINC (SCANI EOR SCAN2)

OLDBUF ISCANIEORSCAN2) SCAN ILook for Ghost keys in

LDX NEWBUF matrix image stored inNEWBUF

LDA . SFF Initialise pointer andtemporary register

STA TEMPI

STORE% STX TEMPLDA X

CMP SFFBEQ NXTCMPSTA TEMPI

INCRX INXCPX , ENDBUFBEQ NXTCMP

LDACMPBNE

SEC

X

TEMPIINCRX

ROTA RORA

BCS ROTACMP SFF

BEQ INCRXBSET GHOST.FLAGRTS

Store latest pointer valueGet associated byte fromNEWBUFIf no keys pressed in thatcolumn then try the nextone else store it

If end of table reachedthen go compare nextcolumn with all thoseabove itGet a column from bufferand compare it with thestored one if not equaltry next columnBytes are equal -possibleghost set carry for test

Have got the 1st 0 in carryso dace now not FF

then there must have beenmore than 1 key in thesame position in 2columns and so a Ghostsituation set flag andreturn

914ELECTRONICS& WIRELESS WORLD

Analogue circuits forautomotive uses

These i.cs, designed to survive the harsh operating conditionsfound in motor vehicles, can provide substantial cost savings.

With the increasing use of micro-processors for engine control, trans-mission control, anti -lock braking

and multiplex load control the need hasarisen for interface and power switchingcircuits for interfacing from logic level oranalogue signals to high current, high vol-tage loads.

Development of these circuits has beenmainly driven by the search for improve-ment of total system performance. protec-tion against the harsh automotive environ-ment. minimum component count andhence lower module costs.

Several environmental factors must beconsidered in the selection of automotivedevices.

TEMPERATURE

Passenger compartment electronics have amore limited range requirement of -40°C to+85°C, whilst under -bonnet electronicsmust withstand -40°C to +125°C or evenhigher. Remember that the junctiontemperature will be above ambient. hencethe trend to the lower power dissipationtechnologies such as high-speed c-mos.where the junction -to -ambient temperaturedifferential is generally lower.

SUPPLY VOLTAGES

For passenger cars the normal battery vol-tage is + 12V. although +24V systems maybe introduced in the future. Unfortunatelythis 12V is not constant: essential systemshave to operate at 6V. the low point reachedduring cranking. Once the engine is started.the charging voltage at the battery mayreach +16V. With the battery disconnected,the alternator will probably produce a con-tinuous +18V: and during a two -batteryjump start +24V is applied. Not only mustintegrated circuits protect themselves, theyare now expected to protect their loads fromdamage also.

TRANSIENT VOLTAGES

Transients can appear on any line, withvarious voltages and durations: the probabil-ity of each varies (Table 1). This hazard ismuch more difficult to assess but neverthe-less has to be taken into account if anythingother than a worst possible case design(which is not necessarily the most econo-mical) is required.

Designing for transient survival probably

ROCCO SHAH

TABLE 1. - TYPICAL TRANSIENTS ENCOUNTERED IN THE AUTOMOTIVE ENVIRONMENT

Length oftransient

Cause Voltage Energycapability

Occurrence

Steady state Failed voltage regulator 18V Infrequent

5 minutes Booster starts with 24V battery 24V Infrequent

4.5-100ms Disconnection of battery duringhigh charging rates

125v 10J Infrequent

0.32s Inductive load switching 300V to 80V 1J Often

0.20s Alternator field decay 100V to 40V LI Each turn-off

90ms Ignition pulse - disconnected battery 75V 0.5J 500Hz. several timesin vehicle life

1.0ms Mutual coupling in wiring harness 20C V 1J Often

151, s Normal ignition pulse 3V 0 001J 500Hz continuous

Accessory noise 1.5,/ 50Hz to 10kHz

Transceiver feedback 20mV R.f.

causes as many problems as any otheraspect, and reliable data can be difficult tocome by. The load dump transient oftenquoted is actually a worst case. It occurswhen a battery so flat as to be effectively ashort circuit is disconnected whilst beingcharged by an alternator driven by an engineat maximum engine speed. It can be morecost-effective for a car manufacturer tochoose lower load dump criteria based on hisexperience. He could then save money on allunits, but would trade this off against theincreased claims that resulted.

The car manufacturer may well decidethat central transient suppression at thealternator is more cost-effective than ahigher degree of protection at each and everyunit - particularly as these proliferate.

IGNITION INTERFACE CIRCUITS

Basic building blocks for the interface cir-cuits are power supply regulator. resetgenerator with over -voltage protection. coildriver outputs and sensor input circuit.

Figure I shows a simplified diagram of anignition injection system based on theMC68HC05-B6 or MC68HC11, interface cir-cuit UAA1550. the microprocessor inputprotection circuit TCF6000 (Fig.3) and thehigh voltage Darlington BU323.

The power supply regulator provides anoutput of 5.1V +3% and current capabilityof 400mA, adequate for the microprocessorplus other peripheral circuits. Its designenables it to cope with the voltage range and

noise level of the vehicle's battery supply.The reset generator provides a reset signal

to the microprocessor if the regulator out-put falls to 7% below normal level. In theevent of broken spark plugs or leads. theover -voltage protection circuit provides pro-tection to the Darlington. The coil driveroutput stage has outputs for two coils withsteering logic and constant current logic.

Finally. the sensor input circuit has beendesigned for either variable reluctance orHall -effect sensors, both of which are popu-lar for speed/position sensing.

In the system shown, the interface regula-tor provides a regulated output with abattery voltage down to 6V. The diodeprovides protection against reverse battery.and. in association with a capacitor. en-sures an uninterrupted supply during nega-tive transients.

PRESSURE SENSOR AMPLIFER

Load sensing is normally effected by measur-ing absolute pressure. Most measuring de-vices suitable for automotive use produce ananalogue voltage output.

Figure 2 shows a manifold pressuremeasuring system with a semiconductorpiezoelectric device MPX100 and a dedicatedlow power dual operational amplifier. Theoperational amplifier has been designed foran output voltage swing from OV to V. a lowoffset voltage (2mV). a large current drivecapability and an operating temperaturerange of -40 to +125°C.


Lom

bdo

sens

or

Dia

gnos

tic

Eng

ine

tem

p

Air

tem

p

Kno

ck s

enso

r

0 0"v

"\/\,

Air

flow

01\

/\/\

I

tV B

AT

---e

V

Thr

ottle

posi

tion

Mag

netic se

nsor

(Hal

l effe

ct)

TC

F 6

000

Mic

ropr

oces

sor

MC

68H

C05

-B6

MC

68H

C1I

forn

ily

Vcc

5 1

V

Res

et

Driv

er s

elec

t

Dw

ell

Sta

ll

DE

SA

T4

ZC

5V r

egul

ator

Vol

tage

mon

itor

Pow

er -

onre

set

Out

put A

Con

trol

logi

cU

AA

1550

Out

put B

Cur

rent

lim

it

Sen

sor

inpu

t

VC

L A

*VB

AT

IA

VC

L B

Out

put d

river

A

o

B

IE3

LIL

BU

323

Coi

l A

O R,

Spo

rpl

ug

To

coil

B

-V B

AT

Driv

ers

Inje

ctio

n va

lves

Spe

edom

eter

Con

trol

light

Idle

spe

edp.

Fig

.lign

ition

and

inje

ctio

n m

odul

e w

ith d

ual -

coil

outp

ut, b

ased

an

eigh

t -bi

t mic

ropr

oces

sor

(sim

plifi

ed).

MICROCOMPUTER INPUTPROTECTION NETWORK

Normally mos and bipolar circuits are sensi-tive to negative and positive voltage spikes.and so care has to be taken to ensure thatvoltage transients do not reach the micro-computer's i/o pins.

Conventional protection units consist ofdiode resistor networks which do not ingeneral meet the specification over thewhole temperature range.

Figure 3 shows one of the six inputprotection circuits developed as an inte-grated circuit. TCF6000. and capable ofsinking 10mA at +5.75V and -0.3V. Thisperipheral clamping array has been opti-mized for microprocessor -based automotivesystems requiring minimal componentcount and low hoard space. Figure 4 shows atypical ignition system protected in this way.

DATA CONVERSION CIRCUIT

In several automotive systems, independenta -to -d converters are used for convertinganalogue signals from the sensors (such aslambda 'oxygen'. knock, load and tempera-ture) to provide digital inputs to the micro-processor.

A low cost c-mos eight -bit a -to -d conver- Eac cell

ter MC145040. with serial interface parts toprovide communication with micropro-cessors. operates from a single power supplyand gives a maximum non -linearity of ±1/sI.s.b.

The device provides a successive approx-imation conversion time of 10 micro-seconds. has 11 analogue input channelswith internal sample and hold, and has Analogue inputs

separate reference voltage and analogueground pins for noise immunity. Rin

TCF 7000

v

Output

Fig.2. Manifold pressure -measuring system based on a special-purpose dualop -amp.

Vrefgenerotor

Vdd

HIGH SIDE DRIVER SWITCH

In engine and multiplex control modules.high -side driver switches are increasinglyused to drive loads from the positive side ofthe power supply.

The main advantages of high -side switch-es are corrosion resistance and protectionagainst accidental short-circuits duringmodule inspection or repair.

Figure 5 shows a high -side driver switch,MC:3399T, fabricated in a power BiMOSprocess which combines the hest features ofbipolar and mos technologies. The devicehas been designed to drive low current loadsdirectly and for intelligent control of low -power relays or solenoid valves. Its output iscontrolled by t.t.l-compatible enable pin. Inthe on state, the device exhibits very lowsaturation voltage (400mV typical) for loadcurrents in excess of 750mA.

The device also protects the load frompositive- or negative -going high voltagetransients I± 100V1. With its current limit-ing and thermal shutdown, it can drive allkinds of loads including inductive ones overa wide temperature range 1-40 to +125°C).

DASHBOARD DISPLAY

Independent display drivers are still used inlow cost microprocessor -based dashboard Fig.4. Using the TCF6000 to protect a microcontroller's i/o pins.

Digital inputs) -1VV\.,-V \A,

T T

MicrocomputerIFig.3. Both mos and bipolar inputs in motor vehicle systems need to be guardedagainst transients: the TCF6000 is made up of six protection networks.

VW'

TCF 6000

Engine tempera ure

Pressuresensor

Battery volts

Ccr

-411-Vcc

111111.

3° Cin

V cINT1 BO

D6

MC6805 S2

DO

2

V s

Coil

drive

Coilfeed-back

Serial diagnostics

Ignition module


Vs 2 5 to 26V

± 125V

Thermollimit

Voltagelimit

Input t t l threshold

Ground

CurrentI

Saturationde ect

(lb l mit)

Bi moscontrol circuit

Idrain.25mAIcut. IA

Vio= 0 5v

@lout=1A

ri© 0-mos

Fig.5. Driver switch based on hybrid bipolar-mos technology.

BPout BPin

-- 16

Oscillator

Osc in

Data

Clock

Out33 Out 2 Out)

33 goted buffers

33 latches

Bit 35 Bit34 Bit33 Bit2 Bits Bit 0

36 -bit shift regis er

loutto 1 A"

Load

Control logic

Reset

Fig.6. Low -power driver for liquid -crystal displays on thedashboard. Data input is in serial form.

systems because they allow flexibility andfree the microprocessor for other tasks.

Liquid crystal displays, as used for clocks.trip meters. information, climate controlsystems etc.. need low -power display drivers.capable of handling several I.c.d. segmentsper package.

Figure 6 shows a low -cost c-mos displaydriver for automotive Dashboard applica-tions. It consists of a 36 -stage shift registerwith 33 latches and output buffers, whichmay drive up to 33 non -multiplexed I.c.d.segments per package. Devices may be paral-leled to increase the number of segments perpackage with minimal interfacing betweenthe display and data source.

Further readingMicroprocessor -controlled engine management.by Pat Jordan. Electronics & Wireless World.September 1987. p.876.Advanced engine management systems. by PatJordan. Electronics & Wireless World. December1987. p.1225.

Rocco Shah is in product marketing atMotorola Semiconductors S.A.. Toulouse.France.

Radio -nay pioneerHarvey Schwarz, joint developer withW.O'Brien of the Decca Navigator sys-tem, has died.

He came from the American mid -West and worked at GEC on the de-velopment of radio reception, arrivingin England in 1932 where he went toDecca. As chief engineer, he was re-sponsible for the development of theradio navigation system, spendingsome of his time in California workingwith O'Brien and much of the restpersuading the Admiralty that his sys-tem was worthy of their attention.Fortunately, he was sufficiently persua-sive and it was adopted in time to beused in the invasion of France in 1944.

After the war he turned his attentiointo Decca's new long-playing records.He leaves a widow, who was also in-volved in the development of the DeccaNavigator.

ASSERTcompetition

All the entries for our 1988 writingcompetition are now in and are beingread by the judges. This is no light taskand will take some time to get through.but we will he as quick about it aspossible. Results will he announced in acouple of months.

The first impression is that the quali-ty of entries is good. so that a numberwill eventually be seen in print after thecompetition is over.

SCITECH 89The British Science and Technology Trusthas accepted an offer from SCS ExhibitionsLtd to stage SCITECH 89 at Alexandra Palaceand Park from 17 to 21 May. 1989. Theorganizers intend that the exhibition shallhe housed in the new location each May.

Invitations to exhibit are to be restrictedto science -based, British companies or thoseforeign companies which carry out a majorpart of their research and development inBritain.

The organizers will allocate free space to alimited number of schools, universities andhospitals to allow them to show their scien-tific and technical development.

Enquiries should be made to SCS Exhibi-tions Limited, 212 London Road. North End,Portsmouth. Hampshire P02 91E.

Professor Paul Cook of BSTT talking toscience students


Pan -European cellulardemonstration

Research teams at Plessey and Racalgive themselves until 1991 to shrinkthis equipment into a telephone

handset. Inside the 19 -inch rack in thephotograph is the digital part of a mobiletelephone for CSM, Europe's second -generation cellular telephone system. GSM(standing for Groupe Speciale Mobile, astandards -making committee of the Euro-pean PTTs) is a 900MHz system which willuse digital speech encoding to achieve aboutthree times the spectrum efficiency possiblewith the present-day analogue networks.

The core of the first GSM network has nowbeen ordered by Racal-Vodafone. Equipmentto be supplied by Orbitel, the manufacturerowned jointly by Racal and Plessey, will serve60 000 subscribers in London and the south-east of England. Work on installing it is

expected to begin in 1990, to provide aninitial service in June 1991. Vodafone hasordered a second system, from EricssonRadio Systems, to serve the Midlands andNorth. Eventually GSM will become thelargest single -standard mobile telephonesystem in the world, with 17 networks allover Europe. Some 9M subscribers are ex-pected to be on it by 1999.

The prototype seen here, built by Plesseystaff at Roke Manor and demonstrated inLondon in July, contains most of the fea-tures of a GSM mobile unit. Missing were thefrequency -hopping capability, which willenable the system to live with interference,and the long-term predictor section of thedigital voice coder.

At the distant end of the radio link was aroving light van, fitted with broadly similarequipment though cast paradoxically in therole of a base station. A call put through byOrbitel's technical director, Dr David Bal-ston, to Bernard Mallinder of GSM's Perma-nent Nucleus in Paris, was clear and freefrom interference - though troubled at firstby delayed echoes returned through a publicaddress system at the Paris end. This call wascoupled at the van to the Vodafone cellularnetwork and was thence routed over ordin-ary international lines.

In an operational system, each CSM radiochannel would be time-shared by up to eightconversations, giving a total data rate of270kbit/s. The speech rate leaving the voicecoder (top shelf of the demonstration unit isabout 13kbit/s per channel, but to thischannel codec adds error -protection andcontrol data. Because of bandwidth com-pression. the speech quality sounded dis-tinctly gritty: but according to Orbitel thelong-term predictor will improve it to thepoint where it hardly differs from the aver-age analogue cellular quality. The coder isdesigned to accept a bit error rate of 1 in 100on speech, but the system could manage

9.6kbitis of data with a b.e.r. of 10'.Propagation delay through the system canbe up to 90ms.

A crucial feature of CSM radiotelephoneswill be the equalizer (bottom shelf of thecabinet). By means of advanced digital signalprocessing techniques, this extracts thewanted signal from the mass of multi -pathreflections which are inevitable at u.h.f. in acity. With the aid of the equalizer, the digitaltelephone gives acceptable reception with acarrier -to -noise ratio of 12dB: analoguecellular needs 30-35dB. Improvements inequalizer design could enhance perform-ance still further, and would of course enablethe networks to pack more paying customerson to the same frequencies.

On the centre shelf of the unit is thechannel codec. which handles the complexsignalling procedure between base stationand mobile. The 900MHz r.f. unit, runningabout 5W. is in a separate cabinet on the top.Frequencies used for the demonstrationwere in the band reserved for CSM - 955MHzmobile -to -base and 910MHz coming back.

Work is now in hand to design the all-important v.I.s.i. chips for the voice codec,

This equipment has proved that Europe'snew cellular telephone system is workable:now the task is to miniaturize it. Thecomplex digital speech coding will result ina s3ectral efficiency three times betterthan that of Britain's present cellular net-works. Some 17 countries are alreadyinvolved in the project. By 1999, the sys-tenrr is expected to cover a population of300M and to have 9M subscribers.

channel codec and equalizer, each likely tobe the equivalent of about 600 000 transis-tors. One potential hitch is that GSM is stillstruggling to agree on the details of acommon radio interface standard; but if noagreement can be reached. Orbitel and itsassociated companies are prepared to goahead with their own joint standard.

Base station hardware for both VodafoneGSM systems will include Matra designsalready in use in the French Radiocom 2000mobile telephone network. Switching will bebased on the Ericsson AXE -10 exchange.

Further details from Orbitel Mobile Com-munications Ltd, Keytech Centre, AshwoodWay, Basingstoke RG23 8BG.


NEW PRODUCTSFault finding in i.csFast repair at component level inproduction or field service is possibleusing the diagnostic informationobtained from the DIT-24XP digitali.c. test system.

Thurlby Electronics' device candetect manufacturing defects such asdry joints and faulty i.cs. identifyunknown i.cs: and test i.cs in -circuitusing a self-contained library.Operation is simple: a test clip is

connected to each i.c.. the i.c.number keyed in. and after pressingthe test key the results are displayedon the monitor. The normal librarycovers t.t.l. c-mos memory andinterface devices. Standardinterfaces are provided for RS232.Centronics type printer and anexternal keyboard. ThurlbyElectronics Ltd. New Road. St. Ives.Huntingdon. Cambs PE17 4BG. Tel:0480 63570.

Cursor measurement of waveformA cursor measurement system on theHitachi V1100A, 100MHz readoutoscilloscope allows easymeasurement of voltage. time.frequency, phase difference andratio.

The results are numericallydisplayed on the screen to eliminatethe errors associated with visualinterpretation of waveform

parameters. There are four channelswith varying sensitivities on the unit,which is available from ThurlbyElectronics. A signal delay linepermits viewing of the leading edgeof the displayed waveform on an8 x 10cm display with autofocus.Thurlby Electronics Ltd. New Road.St. Ives. Huntingdon, Cambs. PE174BG. Tel: 0480 63570.

Programmabledisplay moduleA programmable vacuum fluorescentdisplay has been introduced byPulseview.

The PVFM 35-05-20 can be pre-programmed with up to 32 fixedmessages or 24 fixed and 4 movingmessages: either by themanufacturer or by the user with astandard eeprom programmer. Thedevice provides a single row of twentycharacters each comprising a 5 x 7dot matrix, and each 5mm in height.The module measures128 x 34.5mm. Pulseview Ltd. Unit 1& 2, Suffolk Way. Drayton Road,Abingdon, Oxon 0X14 SJY.

Inductive proximitysensorsContact -free detection of metalobjects in industrial automationapplications is offered by Kuhnke'snew inductive proximity sensors.With 8 switching distances between 1and 15mm, the series has bouncefree switching with a typicalrepeatability of 2%. The switches arecapable of operating from either 10-30V d.c. or 80-250V a.c. H. KuhnkeLtd. Lane End Industrial Estate.Lane End, High Wycombe, BucksI IP14 3JC. Tel: 0494 882829.

Colour graphicsSingle -hoard layout, smoothhardware scroll, flash facility. 2MHzoperation and a Centronics printerport are among the advantagesclaimed for the CU -Graph II colourgraphics controller.

Compatible with its predecessor, itis designed by Control Universal foruse with EuroBEEB, the BBC Basic -based single -board computer. Threeversions of the controller areavailable: colour and monochromeeach with a Centronics printer port;and a low-cost monochrome withoutthe printer port. Control UniversalLtd. 137 Ditton Walk, CambridgeCB5 8QF. Tel: 0223 244448.

Multi -pinconnection with nosolderingHighland Electronics' Swisstacrange of front -panel componentsincluding switches, pushbuttons andlamps can be easily installed andreplaced using a new multi -pin plugassembly.

All the wires are pre -assembledinto a three -element contact blockwhich includes up to 14 flatconnectors. The connector assemblyis plugged onto the rear contacts ofthe switch unit with no need forsoldering. Each flat connectormeasures 2.8 x 0.5mm includingcatch and is tin -coated. HighlandElectronics Ltd. Albert Drive.Burgess Hill. West Sussex RH159TN. Tel: 04446 45021.

Complete p.c.b.design systemThe entire design process of p.c.bsfrom schematic generation throughto board artwork can be automatedwith Project p.c.b.

This software system is designed tobe easy to learn and to use. A mouseis used to select menu options, drawlines, define symbols and enter partslocations. Screen windowing helps tospeed progress through the packagewithout long delays or indirect inputsequencies. Clwyd Technology hasoffered to refund the purchase priceto any customer not satisfied withthe package within 28 days. ClwydTechnology Ltd. 99 Leman Street.London El 8YE. Tel: 01-480 5447.

Small surge arresterstops big surgesAn inert gas -filled cylindrical surgearrester measuring only 5x 5mm iscapable of diverting a surge currentof up to 2.5kA and an alternatingcurrent of up to 2.5A. Typicalapplications for the unit designed bySiemens are the highly concentratedmain distribution frames fortelephone lines. Siemens Ltd.Siemens House. Windmill Road.Sunbury -on -Thames, MiddlesexTW16 7HS.


ANEW PRODUCTSDigital storageoscilloscopeSampling rates of 100M samples/seccan be achieved simultaneously onthe two channels of Hitachi Denshi'sdigital storage oscilloscope.

The unit has a memory capacity of4000 words/channel and two 4000word save memories relative tochannel 1 which has energy back-upto hold data for 72 hours. It canhandle four samples/period thatenable storage of single shotphenomena at 13.5MHz or 25MHzon one or two channels.

Three different modes - envelope,averaging and roll - allow detailedanalysis of waveform. The VC6265has as standard: plotter output; CPIB(binary or BSC II) for computer datainput/output; and analogue outputfor a chart recorder. Hitachi Denshi(UK) Ltd 13-14 Garrick IndustrialCentre, Garrick Road. Hendon.London NW9 9AI'. Tel: 01-202 1311.

Low -consumptionpower supply i.c.A c-mos chip that can supplytelephony equipment with 40V to thefour -wire network termination inaccordance with CCITT (S interface)recommendations consumes lessthan 10mW.

This chip has been designed bySiemens to replace discrete powersupply circuits and is suitable forvoice terminals which are line -fedand require an internal regulatedvoltage of 5V. It is also suitable forterminals that operate in emergencypower conditions, for example, afteran a.c. power failure. Siemens Ltd.Siemens House, Windmill Road,Sunbury -on -Thames, MiddlesexTW16 7HS. Tel: 0932 752323.

Fast digitizing oscilloscopePhilips has announced its fastestoscilloscope, the PM3340, with full2GHz bandwidth and triggering,post -capture waveform processing,sensitivity down to lmV/div and adynamic range of 125 to 1.

The PM3340 uses the principle ofsequential sampling which allowsfaster signal build-up than is possiblewith random sampling at u.h.f.frequencies. The oscilloscope has

four memories in which thewaveforms can be subjected toprocessing and automaticmeasurements. Operation is simplethrough on -screen menus and softkeys plus auto set. Up to eight signalscan be stored and comparativemeasurements performed easily withon -screen annotation. Philips Testand Measurement, Colonial Way.Watford, Herts WD2 4TT. Tel: 0923240511.

Increased data capture on logic analyserData -capture capabilities have beenincreased with the introduction of alogic -analyser system which makesuse of a high-speed dram module.

The K450M system provides up to10Mbit of data capture on each of its64 logic -analyser channels at speedsof up to 50MHz and is front panelconfigurable up to 40Mbit perchannel of data capture across 16channels at 200MHz. GouldElectronics claims that theinstrument has up to 4000 timesmore full -speed data capture

memory than any other logicanalyser on the market.

It is supplied complete in a 48inrolling cabinet. which contains thelogic analyser with dual disc drivesfor storage, the memory module, andthe PC/AT compatible chassis with80Mbyte hard disc. The PC colourscreen and keyboard may he placedon top of a convenient work surface.Gould Electronics Ltd. InstrumentSystems, Roebuck Road, Hainault,Ilford, Essex IG6 3UE. Tel: 01-5001000.

Colour displaymodulesColour display modules offering achoice of user -selectable line ratesfrom 15 to 35kHz. combined withhigh performance video inputs areavailable from Nevin Developments..

The units measure183 x 142 x 120mm and the completescan electronics, including the linetransformers are mounted on asingle p.c.b. for ease of installation.User controls include: height; width;horizontal phase and frequency;vertical shift and frequency;linearity; brightness; and contrast.Nevin Developments Ltd. CharltonRoad, Andover. Hants SPIO 3.1L. Tel:0264 332122.

Microwavemodulation analysisMore accurate analysis of complexmicrowave modulation is possiblewith Hewlett-Packard's HP 8981Avector modulation analyser.

It adds a broadband, coherent I/Qdemodulator, covering the 50 to200MHz intermediate frequenciesand carrier range, to the HP 8980Avector analyser introduced in 1987.which provides time, vector andconstellation analysis modes. Theadditional front-end vectordemodulator processes a 35MHz I/Qbandwidth (for up to 70MHz channelmeasurement). External filter portsare available for use with 200MHzchannel measurements.

After digitally processing thesignal the analyser displaysconstellation parameters,quadrature error, l/Q imbalance andd.c. offsets in visual as well asdigitally annotated form. Hewlett-Packard Ltd. Eskdale Road.Winnersh Triangle, Wokingham.Berks RGII 5DZ. Tel: 0734 696622.


NEW PRODUCTSReal-time controlfor BBCmicrocomputerA real-time control interface for theBBC microcomputer has beendesigned with schools, laboratoriesand home users in mind.

The Termite interface from PaulFray provides eight terminals,independently programmable to anycombination of i/o. and it connectsdirectly to the user port of any modelBBC computer. It is powered eitherdirectly from the BBC or an externalpower supply and can be used to drivemotors, relays, detect sensors orsimply light up bulbs. Paul Fray Ltd,Willowcroft. Histon Road.Cambridge CB4 31D. Tel: 022366529.

Signal analyserallows custom -measurementMany test and automation featuresthat traditionally have required theuse of an external computer areincluded in the Hewlett-Packardtwo -channel dynamic signalanalyser.

The HP 35660A fast Fouriertransform -based analyser hasapplications in electronics.mechanical testing, acoustics andother low -frequency areas in bothproduction/process and r. and d.environments. It provides spectrumanalysis from zero to 102.4kHz andnetwork analysis from zero to51.2kHz.

Using HP Instrument Basic whichruns inside the analyser without theneed for an external computer, testsequences are simplified. Advancedusers can create custom algorithmsand result formats to provideanswers not directly available fromthe standard analyser. HewlettPackard Ltd. Eskdale Road.Winnersh Triangle. Wokingham.Berks RG11 5DZ. Tel: 0734 696622.

Verbal communicationbetween man and machineVerbal command and responsebetween user and personal, micro,mini or mainframe computers isachieved by the Verbex 5000 voice i/osystem.

The system from DecadeComputers uses a two-way serialconnection at up to 19 000baudthrough a standard RS232 -C port on

the host computer. The speechrecognition technology allows acontinuous flow of verbal input atnatural human speeds, regardless oflanguage or dialect. Voice cartridgescan be easily changed by frontloading. Decade Computers Ltd.Ringway House, Kelvin Road,Newbury, Berks RG13 2BD.

Prototype p.c.b. is also production boardTransition from a prototype p.c.b. toa final production board has beenmade easy by the introduction ofAugat's Unilayer II board modules.These have a high -density, plated -through hole pattern on a grid ofstandard pitch. Supply and groundbuses are already etched on eitherside of the board and, forprototyping, it is fitted with wire -wrap pins.

For production, the same patternof module is used by the wirewrapping is replaced by solderedwires produced by the Unilayer IIwiring system to give high -density

layers. Insulated wires allow signalsto cross and provide the shortestpaths. Each layer is soldered to theplated holes and then bonded to theboard with adhesive. A finished boardmay have six or more layers. Boardsused with the system are available inall the popular formats for use inP.Cs, VMEbus and other systems.

Augat manufactures the modulesfrom the prototypes and says that itsservice is much faster thanconventional multi -layer boards.Augat Ltd, Sunrise Parkway, LinfordWood East, Milton Keynes MK146LF. Tel: 0908 676655.

Power transistormodulesSeparate driver stages in a.c. and d.c.motor controllers, power supplies.converters and other industrialequipment are not needed with thenew range of power transistormodules from MitsubishiSemiconductors.

These high -gain versions of theQM series consist of six packconfigurations for three-phasebridges as well as dual driver stages.Voltage ratings are 600 or 1000V. andcurrent ratings range from 10 to200A. Mitsubishi Semiconductors.Mitsubishi Electric UK Ltd,Electronics Division. TravellersLane, I latfield. I lertsAL10 8XB. Tel:07072 76100.

Diode arrays reduceassembly timeWhere large numbers of diodes needincorporating on densely packedp.c.hs. assembly time can be reducedusing Iskra's sil or dil diode arrays.

The dual in -line arrays are housedin standard 16 -pin plastic packagessealed in flame-retardant epoxyresin. Each pack contains eightindependent 1N4148 epitaxial signaldiodes or the BZY 88 range ofdiffused junction zeners.

Single in -line networks consist ofeither common -anode or common -cathode 1N4000 series diode chipswhich are bonded on to a ceramicsubstrate. The lead -outs are on astandard 0.1in pitch forcompatibility. Iskra Ltd, Redlands.Coulsdon, Surrey CR3 2HT. Tel. 01-668 7141.


rNEW PRODUCTSMulticore cableanalyserThe latest industry requirements fortesting cables up to 1000V d.c. aremet in the upgraded multicore cableanalyser available from Cable CheckSystems.

The analyser can be used for acomplete serviceability test onmulticore cables on or off site.Results are displayed digitally, alongwith core numbers for easyidentification. A hard copy of faultlocations and test acceptance levelscan be obtained using the integralprinter. Cable Check Systems. 18Quay Lane, Gosport. HampshireP012 4L1. Tel: 0705 528396.

Digital wattmeterA high resolution digital wattmetermanufactured by Infratek is suited tothe measurement and analysis ofdistorted waveforms.

The single phase 103A availablefrom Lyons Instruments features awide frequency range from d.c. to100kHz. voltage ranges to 1000V andcurrent to 50A without externalaccessories. Power ranges are from9mW to 50kW and the unit has asix -digit display with a basic accuracyof 0.3%. Lyons Instruments Ltd.Hoddesdon. I lerts. Tel: 0992 467161.

Power supply unitsThe first of a series of compact,o.e.m. power supply units capable ofdelivering up to 1500W d.c. outputhas been introduced by WeirElectronics.

The series, based on an advancedswitched -mode regulator, is toinclude single output and multipleoutput configurations to meetindustry requirements. All modelsare housed in a metal enclosuremeasuring 279 x203 x 127mm withan integral cooling fan.

Now available is the SMM 1500 0000 which has five output rails. Thecombined load rating is 500W but ifhigher power levels are required theunits can he connected in parallel toprovide outputs as high as 3kW. WeirElectronics Ltd. Durban Road.Bognor Regis. Sussex P022 9RW.Tel: 0243 865991.

use -Girl

minify iwanel.lo

MAIN MINU[leer screenVierporlArchiveDisbraztAnalyzeP200m t FenConFillerWereImage editProgramMternel Prole

lo

Real-time video digitizationThe Il3M PC IN turned into an trudgeprocessing workstation using ahardware and software packagecalled PCVisionplus Frame Grabber.

The package from AmpliconElectronics is composed of fourmajor sections: video sourceinterface; frame memory: displaylogic and host computer interface. Itoccupies only a single expansion slot

in the PC and features an eight bita -to -d for 256 grey levels/pixel;advanced phase -lock for stablesynchronization with v.c.rs:programmable gain and offset and a1024 x 512 x 8 bit frame memorystores multiple images. AmpliconElectronics Ltd, Richmond Road,Brighton, East Sussex BN2 3RL. Tel:0273 608331.

Probe identifies metalsIdentification of the carbon contentin ferromagnetic steels, thechromium content in stainlesssteels, the alloying elements incarbide tool alloys and the amount ofretained austenite in austemperedductile iron can be carried out simplyusing the Metal Analyser availableform Kontron Electronics.

The analyser's probe is placedagainst the material under

investigation and the metallurgicalcharacteristics are identified bysensing the electrical effectsproduced in the probe. Changes ofinductance and the inductance -to -resistance ratio of the sensing coilare all measured. KontronElectronics Ltd. Blackmoor Lane.Croxley Centre. Watford. Hots WD18XQ. Tel: 0923 45991.

Four -channel oscilloscopeThe simultaneous observation ofmultiple signals is possible using theGrundig MO 100 four -channel100MHz oscilloscope available fromInstrumex.

The trigger level position on thescreen is shown by a cursor and canbe selected to trigger from a specificpoint. An automatic time -range

switch selects the appropriate timebase for the signal to be measured. Aseven segment led digitally displaysthe time range. The conventionalstepping switch for setting the timesweep is replaced by a potentiometer.Instrumex Ltd, Dorcan House,Meadfield Road, Langley, BerkshireSL3 8AL. Tel: 0753 44878.

Electronic key forsecurityAn electronic key has been designedby Dallas Semiconductor as part of arange of miniature security systemsfor software and other data.

Constructed to be carried aroundin the pocket, the device stores 64bits of user -definable identificationcode and a 64 bit security match codewhich protects 128 bits of read/writenon-volatile memory. Theidentification and security matchcodes are programmed into the keyafter which it follows a specialprocedure with a serial format toretrieve or update data. DialogueDistribution Ltd. Wicat House. 403

Stacking dot matrixdisplayA two colour 101mm dot matrixdisplay element designed for easyp.c.h. mounting is stackable bothvertically and horizontally to formlarge alphanumeric displays.

The LTP-4357AA 5 x 7 dot single -plane device is intended forapplications such as computer -controlled message hoards. It

incorporates green galliumphosphide and orange galliumarsenide phosphate elements on atransparent substrate. The elementshave individual pins for each colour.producing 28 pins on the rear face ofthe package. The light outputensures that either colour is visiblein normal artificial lightingconditions. Selectronic Ltd. The OldStables. 46 Market Square. Whitney.Oxon OX8 6AL. Tel: 0993 73888.

Flat electrolyticcapacitorsHalf-size versions of the Petchaconrange of thin electrolytic capacitorshave been introduced by ECCElectronics. The 7mm thick devicesare suited to applications whereheight is restricted. They areavailable with capacitances rangingfrom 180 to 15 000µF ECCElectronics (UK) Ltd, 9 BlenheimRoad, High Wycombe. Bucks HP123RT. Tel: 0494 450716.


rN-Ew PRODUCTSFixtureless testsystemAn in -circuit test system called theTakaya APT -2200N uses threemoving probes and functions as animpedance signature analyser toperform tests on a loaded hoard.

Contax supplies the fine -tippedprobes for contact spacings down to11.1mm. so the latest surface mountand hybrid devices can be tested on anon -fixed test bed. Also the probeshave a height clearance of 15mmallowing either side of a loaded hoardto he probed.

Board sizes up to 400 x 350mmcan he handled and each test takesabout 0.5s. making the systemsuitable for low volume production.Using bar coding techniques, testprograms are called off disc or fromanother computer station as eachhoard arrives at the test station.Contax. 362 Spring Road. Sholing.Southampton SO2 7PB. Tel: 0703434141.

Two -wire pressuretransmitterExcitation voltage and currentoutput are transmitted over the samepair of wires in the model BL flushdiaphragm pressure transmitter.

Control Transducers offers theunit in various pressure ranges from0.35 to 1400 bar. Overload is twotimes rated pressure without damageand five times without bursting.Excluding the connector. the unitmeasures 24 x 98mm and is suitablefor applications where ease ofcleaning and sterilization areimportant, such as film processingand the food and drink industry.Control Transducers. North Lodge.25 Kimbolton Road. Bedford MK402NY. Tel: 0234 217704.

Improved reception for shortwave listenersListeners will appreciate theimproved quality of the reception ofthe BBC World Service using theLiniplex F2 receiver.

This receiver system, availablefrom Phase Track. is based onreceivers supplied for off airbroadcast relays to BFBS in theFalklands. Belize. Gibraltar. Cyprus.Hong Kong and Nepal. Now asynthesizer covering 2-22MHz isoffered in addition to the standard

eight preset frequencies. which arechosen according to the user'slocation.

A linear phase -locked fullyautomatic synchronousdemodulator reduces the effects offading and interference. Totalfrequency range available is from0.15 to 26.1Mliz. Phase Track Ltd. 16Britten Road, The Robert CortIndustrial Estate. (Elgar Road).Reading RG2 OAU. Tel: 0734 752666.

Television and video testingTrouble -shooting and testing oftelevision, monitors and videoequipment are easily carried outusing the Philips PM5514 andPM5514V colour pattern generators.

The MP5514 is intended formonochrome and colour televisiontesting and the PM5514V with CVBSand RGB outputs meets therequirement of the video equipmentand computer monitor markets.

Available from Instrumex. both offera choice of over 70 colour patternsand combinations which are selectedusing push -buttons on the displaypanel. Sound modulation is alsoprovided, with a choice of internaland external modulation modes anda switchable carrier signal.Instrumex (UK). Dorcan I louse.Meadfield Road, Langley. Berks SL3Ml.. Tel: 0753 44878.

Modular logic analyser has 112 channelsA multi -level conditional trigger -trace sequencer controls the 112channels of timing and state analysison the Thandar TA3000 modularlogic analyser.

Complex acquisitions can be set uprapidly using the soft -key guidedmenus and displays. Instrumexsupply the analyser which has a

general purpose computer with aCP/M Plus operating system.1.4Mbyte of storage on dual floppydisc drives, qwerty keyboard and fullIEEE -488. RS232 and Centronicscommunications capability asstandard. Instrument Rentals (UK)Ltd. Dorcan House. Meadfield Road.Langley. Berks SI,3

Input switches forp.c.b. mountingA range of input switches designedfor mounting on to p.c.bs is availablefrom Highland Electronics.

The polyester switches can beilluminated with a choice of colouredleds, 3mm in diameter. Key caps areavailable in ten colours and haveeither a pulse or latching function:they are pushed on to the switch as amodule. There is a single break.self-cleaning, snap -action switchingelement and the contacts are goldplated to give a mechanical life of5 x 10' operations. HighlandElectronics Ltd. Albert Drive.Burgess Hill. West Sussex RH159TN. Tel: 04446 45021.

One GAL replaces 21palsTwenty-one different 20 -pin paldevices can he replaced by one ofThomson Microelectronics 16V8series of Generic Array Logic devicessupplied by ITT Multicomponents.

The series uses a floating gateconcept which means the user canelectrically erase and reprogram thedevices several times. Other featuresof the 16V8 include an active supplycurrent of 90mA (compared with180mA for pals): an i/o propagationdelay time of 15-20ns and a 5Voperating voltage.

The GALs have logic copying andhulk erase facilities and a 64 hitregister which can he programmedto contain user defined data. Theyuse ABEL and CUM. compilers bothof which run on the IBM PC andcompatibles. The series is available indual in -line plastic packages. ITTMULTIcomponents Ltd.:146Edinburgh Avenue. Slough. BerksSI.1 4TU. Tel: 0753 824131.


ftk' COMMUNIQ

art

COMMUNIQUE UK LTD, COMMUNICATIONS HOUSE,PARLEY AVENUE, LONDON NW2 1SB.TEL: 01-450 9755 TELEX: 298765

HX307 2 ChannelHX307V 4 watt 160-174MHzHX307U 3 watt 450-470MHz

UKV62/2 25/40 w t 66-88MHz425/40 1:)0-175M

25/35 wa 420-470M) izUKV62/4 Mini 10 watt 1-170MHz

H X417 25 ChannelHX4107V 5 watt 160-174MHzHX4)7U 5 watt 450-470MHz

SEE US ATCOMEX '88

STAND No. Al

COMMUNIQUE: MEET THE NEXT GENERATION. . . .

I\ 11 1 1 1'1 ( Jill

VALVES 51 k ..ia,QUALITY Please phone tor firm cluotatoon V A 1 ircluded

Atut.s 1.40A2293 7.00A2900 12.75AR8 1.15ARP3 1.15ARP35 1.15ATP4 0.90B12H 6.90CY3I 2.40DAF 70 1.75DAF 96 0.90DET22 32.80DF92 0.65OF% 1.1501176 1.15DL92 1.450986'87 0.65D9802 0.70E92CC 2.80E 180CC 11.50E1148 0.75EA76 110EB34 1.15EB91 0.60EBC33 2.20EBC90 0.90EBC9I 0.90EBF80 015EBF89 0.80EC52 0.65EC9I 4.00EC92 1.85ECC8I 1.25ECC82 015ECC83 1.10ECC84 160ECC85 0.75ECC88 1.10ECC189 1.20ECC804 0.65ECM 1.25ECF82 1.15ECF002 1.80ECH42 1.65ECH81 1.25ECH84 0.90ECL80 0.65ECL82 0.75ECL85 0.75ECL86 1.10EF9 3.50E F22 3.90E 1-378 2.15I It 30 1.10

LI bu 0.65E F83 3.90EFFIS 0.90EF86 1.45E F89 1.60E.F.31 1.60

EF92 2.15EF.95 1.90EF96 0.60EF183 0.75EF 184 0.75EF812 0.75

..,`,..--`,.,,'-'9 ...fr-'"' 0.85

E L32 0.85EL34 3.25P980EL34' 5.95EL82 0.70E L84 1.35EL86 1.45E L90 1,75E1,9 1 6.50E L95 1.80E L504 2.70E L519 7.70E L821 8.05E L822 995ELL8OSE 4.50em80 1.35

EM87 100190

E Y81 1-2[988 87 0-'.E 988 0.65E180 0.80EZ81 0.80GM4 8.90GN4 6.30G9501 1.50GZ32 1.90G133 4.200134 2.45GZ34' 4.40GZ37 3956177" 14.006T88" 25.03ML4 3.20ML6 120M012001 29.50N78 9.90t 332 1.803., . 8. 0 95

PCL84 0.85PCL86 0.80PCL805 85 0.95PD500 510 4.30PF L200 1.10PI -L200' 2.80PL36 1.60PL81 1.30PL82 0.70PL83 0.90PL84 0.90PLSO4 1.25PL508 2.00PL509 5.65PL519 5.85PLE10251 3.45

0.70P981 800 0.85P982 0.75P988 0.60P9500A 2.1000V03 10 5.9500V03 10' 7.5000V03 20A 27.500006 40A 28.5000001,409' 54.100V03 12 5.75SP61 2.501121 45101122 45.00UABC80 0.75UBF 80 0.70UBF 89 0.70UCC84 0.85UCC135 0.70UGH42 2.50RCH81 0.75UCL82 1.60UF41 1.85UF 80

160,,LUF 85 , '

11-,UL84 1.50.,,..

UM802-31°.

UM80' 8UM84 0.95UY82 0.70U985 0.85VR105 30 2.45911150 30 2.45061M 1.70X66 4.951749 0.75.37'.0 19.90

3.453.75

Z803U 21.15Z900T 4.301A3 3.801L4 0.651115 1.001S4 1.001S5 0.751 T4 0.751U4 0.802X2A 3/03A4 1.303A T 2 3.4031328 17.5031328' 19.50306 0.603E29 21.85354 1.454E132 40.25511409 3.35SU4G 1.855V4G 1.90SY3G I 1.90513 4.855740 1.2551401 2.206 30L2 0.806AB7 0.706AC7 1.156AG5 0.606A65 1.906AK6 2.856ALS 0.606ALSW 1.506AM5 6.506AM6 1.606AN8A 3.806A05 1.756A05W 2.306AS6 1.156AS7G 4.956AU6 0.906AX4GT 1.306AX5GT 1.306BA6 1.406B86' 1.856BE6 1.40

66E6' 1.856BG6G 1.60,.

61336 1.'"6B07A 0.8568137 4.806BW6 6.106BW7 1.506C4 1.20

6CH6 7.506CL6 2.756CW4 7.006CX8 4.608oys 1.156D6 2.5061-60 1.9561-6GB 1.106F7 2.80

6F8G 0.8561-12 1.6061-14 1-,26F15 3-`96F17 3.1061-23 0.756F24 1.156F33 10.506FH8 18.806GA8 0.656GH8A 0.906H6 1.606J4 1.956J4WA 3.106.15 2.306.15G I 1.50636 2.206J6W 2.806JE6C 8.456JS6C 8.456JU6 6.3561( 7 1.456KD6 8.106L6 7.106L6GC 8.106L6G I C 2.9061,18 0.706LD20 0.706L06 8.45607G 1.306SA7. 1.806SG7 1.806S -i7 1.806SK7 1.856SL 7G T 3.506SN7G I ISO6507 0.9565117 4.606V6G 1.506V6GT 1.40604 1.50605G I 0.75696G 2.806Z4 1.75714 1.90906 2.15

11E2 19.5012A6 1.0012416 19512417 1.2512AU7 0.9512A07 1.1012BA6 1.801213E6 110129H7 14511 19.9512J5GT 0.5512K7GT 1.151210301 1.251207GT 1.15125C7 0.80125117 1.2512S.17 1.4012SK7 1.4512507G / 2.201294 0.701303 2.801306 0.9019A05 1.8519G3 11.5019G6 10.3519H5 38.002001 0.8028E1 1 30

2V I 0.6025L6G I 1.602324G 0.7585A2 1.408iA2' 2.55572B 6110807 3.45897. 4.308.1 IA 13.50812A 32.00813 28.50913' 44.00829B 16.0082913' 24.00866E 14.955314 13.95o31A' 19.80964 1.10555 1.10956 1.20

L,.,57835.75

'''''"Ln' 1.956080 7.306136 2.80614613 11.909001 1A09002 6.509003 8.50

VALVES AND TRANSISTORS NEW HEADSETS ..,,,,J. ill microphonesCAROL HP.339M at E10.00

1,101,1,1 .19 103.1 1,111,1, I , " Headphone imp 4-16 ohms Microphone imp 31600 olys,FIELD TELEPHONE. CABLE TYPE DIO E-80 at 215.00FIELD TELEPHONES TYPE .£. Tropical In metal cases Headphone Imp 4.16 ohms Microphone imp at tun10 -line MAGNETO SWITCHBOARD Can work with ohms

Surplus VIDICON P8211 and othets at f 18 00

POSTAGE AND PACKING CHARGES I 't 1.'355p £3 I:5 65o CS C10 85p £10 1,15£1 OS 3.1 5 1 201'1 60 OVel£2011.' 1,l I , .. . , ..5 at Cot.3

COLOMOR (ELECTRONICS LTD.) 170 Goldhawk Rd. London W12Tel: 01-743 0899 or 01-749 3934. Open Monday to Friday 9 a.m. - 5.30 p.m.

Forole:. 0061 t2061co, 651

P

SEND FOR DETAILS ABOUTA BRILLIANT NEWPRODUCT WHICH

GREATLY INCREASES THEUTILITY & FUNCTIONS OF

THESE RECEIVERS

AIRCASTLE PRODUCTSP.O. BOX 78

BOURNEMOUTHBH1 4SP

(0202) 666233PLEASE STATE WHICH RX YOU HAVE




925

COUNTERS & OSCILLATORS

COUNTERS MET100/600/1000,15008 digit 0.5" LED. 5Hz up to 100/600/1000/1500MHz.Resolves 0.1Hz. Sensitivity 5mV up to 10MHz. Low passfilter. Mains/rechargeable battery powered.

LEVELL RC OSCILLATORS TG152D/D613Hz-300kHz. 5 ranges, acc 2% + 0.1Hz up to 100kHz,3% at 300kHz. Sine or square <20014V to 2.5Vrms. Distn.<0.2% 50Hz-50kHz. TG152DM has an output meter.

LEVELL RC OSCILLATORS TG2000/DMP1Hz-1MHz. 12 ranges. acc 1.5% +0.01Hz to 100kHz. 2%at 1MHz. Sine or square outputs <20014V-7Vrms.Distortion <0.05% 50Hz-15kHz. Sync output >1V.TG2000MP has output meter and fine frequency control.

LEVELL FUNCTION GENERATORS TG302/30.02Hz-2MHz in 7 ranges. Sine, square, Mangle, pulse andramp 2OrnV to 20Vpp from 500. DC offset 0/±10V. TTLoutput. TG303 also has a CMOS output and 6 digit 10MHzcounter with INT/EXT switch.

TEST METERS

WELL AC MICROVOLTMETER TM3B16 ranges 15yVfs,500Vfs. accuracy 1%+ 1%fs 1uV- 20dB/ + 6dB scale ±3dB 1Hz 3MHz 150rnVfs output.

LEVELLfor INSTRUMENTS

* REDUCED PRICES *

DIGITAL MULTIMETERSHC5040T £32HC4510 £52

FUNCTION GENERATORSTG302 £112TG303 £185

FREE UK DELIVERY VAT EXTRA

LEVELL BROADBAND VOLTMETERS TM6B16 LF ranges as TM3B - 8 HF ranges 1mVfs/3Vfs.accuracy 4% - 1 °'ofs at 30MHz 3dB 300kHz-400MHz

HC DIGITAL MULTIMETERS HC5040/5040T3'zr digit 12 7mm LCD Up to lkVdc, 75OVac,104 20MS2 Resolution 100µV.100nA. 10m015040T: 100mR) Buzzer dcV 0 25% Battery life2000hrs 5040T: has a TR testHC DIGITAL MULTIMETER HC45104 digit 11mm LCD Up to 1kVdc, 75OVac, 10A, 20M0Resoln. 100V. 100nA, 10m0 Buzzer dcV 005%

LEVELL DIGITAL CAPACITANCE METER 77050 1pF 2000oF acc 0 5% 3'i digit 12 7mm LCD.

LEVELL INSULATION TESTER TM14Log scale covers 6 decades 10M010112 at 250V, 500V,750V, 110/, IM17S2 at 25V -100V: 100k 100GS2 at2 5V 10V. 10k 10G1-2 at 1V Current 100pA 100srA

OSCILLOSCOPES

CROTECH SINGLE TRACE 20MHz 3031/362mV10V/div. 4Ons-0.2s,div. Cal 0.2V Component test.3031: CRT 1 5kV 5x7cm 3036: CRT 1 BkV 8x10cmHAMEG DUAL TRACE 20MHz 102mVI HM203-62mV 20V cm Ch2-± Chl XY. Cal 0 2V 2V 1kHz sq20ns-0 2srcm. Auto. normal or TV tng. Component testHAMEG DIGITAL STORAGE 20MHz HM205-22mV 20V cm Chl Ch2 Single shot and X -Y modes2Ons-0.2s cm. Auto, normal or TV trig Component test.Cal 0 2V 2V 1kHz.5Mfiz sampling Two 1K memonesDot Joining feature Y out CRT 2kV 8x10cm.

HITACHI DUAL 20MHz V2121222/223lmV-12Vicm. 20MHz at SmV. Ch1±Ch2. X Y Chloutput. 100ns-0.5s/cm. Auto, normal or TV trigger.Cal 0 5V 1kHz square. Z input. CRT 2kV 8x10cm.V222: Plus DC offset and alternate magnify function.V223: As V222 plus sweep delay 14-100ms.

LEVELL DECADE BOXES

C8410/610 4 6 decs. 10pF steps. acc 1% :t2pF.R401/410 4 decs 10 or 1011 steps, acc 1%, 2.5WR601/610 : 6 decs. 112 0(100 steps, acc 1%, 2.5WR601S : 6 decades. 111 steps, acc 0.3%. 2 5W8701 : 7 decades. 10 steps, acc 1%, 2.5W

LEVELL ELECTRONICS LTD. Moxon Street, Bonet, Herts., ENS SSD, EnglandTelephone: 01-440 8686 & 01.449 5028 Telex: 266016 Fax: 01-449 7217


SMALL SELECTION ONLY LISTEDRING US FOR YOUR REQUIREMENTS WHICH MAY BE IN STOCK

Latest bulk Government release - Cossor OscilloscopeCDU150(CT531/3) £150 only. Solid state general purposebandwidth DC to 35MHZ at 5MV/Cm - Dual Channel - Highbrightness display (8-10cm) full delayed time base with gatedmode - risetime 1 ONS illuminated graticule - Beam finder -Calibrator 1KHZ squarewave power 100 - 120V 200V - 250 voltsAC - size W 26CM - 14CM deep - WT 12.5 KG - carrying handle,colour blue, protection cover front containing polarized viewer andcamera adaptor plate - probe (1) - mains lead. Tested in faircondition with operating instructions -£150.00.Racal RA17L Communications Receivers. 500KC/S to 30MC/Sin 30 bands 1MC/S wide from £175. All receivers are air tested andcalibrated in our workshop supplied with dust cover operationinstructions circuit in fair used condition - Racal Ancillary Unitsfor all receivers mostly always in stock - Don 10 Telephone Cable1.2 mile canvas containers or wooden drum new from £20 - ArmyWhip Aerials screw type F sections and bases large qty availablenow P.O.R. - Test Equipment we hold a large stock of modernand old equipment. RF and AF Signal Generators - SpectrumAnalysers - Counters - Power Supplies - Oscilloscopes -Chart Recorders all speeds single to multipen - XY Plotters A4A3 - Racal Modern Encryption Equipment - Racal ModernMorse Readers and Senders - Clark Air Operated Heavy DutyMasts P.O.R. All items are bought direct from H M Governmentbeing surplus equipment price is ex -works. S.A.E. for enquiriesPhone for appointment for demonstration of any items, alsoavailability or price change V.A.T. and carriage extra.

EXPORT TRADE AND QUANTITY DISCOUNTS

JOHNS RADIO, WHITEHALL WORKS,

84 WHITEHALL ROAD EAST, BIRKENSHAW,

BRADFORD, BD11 2ER TEL NO. (0274) 684007.

WANTED: REDUNDANT TEST EQUIPMENT VALVES PLUGS SOCKETS,

SYNCHROS ETC. RECEIVING AND TRANSMITTING EQUIPMENT

IN VIEW OF THE EXTREMELYRAPID CHANGE TAKING PLACE

IN THE ELECTRONICSINDUSTRY, LARGE QUANTITIES

OF COMPONENTS BECOMEREDUNDANT. WE ARE CASH

PURCHASERS OF SUCHMATERIALS AND WOULD

APPRECIATE A TELEPHONECALL OR A LIST IF AVAILABLE.

WE PAY TOP PRICES ANDCOLLECT.

R. Henson Ltd21 Lodge Lane, N. Finchley,

London N12 8JG.5 mins. from Tally Ho Corner

Telephone:01-445 2713/0749

I \ II (I\ ) I \1;l1 ENTER ON RE:1'1,11 CARD


NEW PRODUCTSDesk -top flatbedplotter

high level of accuracy in photo -plotting is brought to PC CAD usersby the compact Flashscan flatbedplotter.

Measuring only 16 by 20in, it ismade in Switzerland by PatekPhilippe and marketed in the UK byKontron Electronics. It uses afirmware -controlled variable -aperture system which projectsimages onto a film. These imagescreate pads from 0.12 to 2.54mmwith 135 steps in between. The padsare used to create lines of any widthas well as user -defined graphicsymbols. Kontron Electronics Ltd.Blackmoor Lane, Croxley Centre,Watford, Herts WD I 8XQ. Tel: 092345991.

Compact mosfetmodulesInternational Rectifier's new CPX200series 250 and 500V H -bridgemodules replace four TO -220 deviceswith a single in -line nine -pinmoulded package.

Designed for high -density p.c.b.applications including medium/highpower supplies, u.p.s. systems, highvoltage motor drive systems, andservo amplifiers the fully isolatedpackage measures only 0.2in thick by0.75in high. It can be mounteddirectly on to the chassis or heatsink.International Rectifier, Hurst Green,Oxted, Surrey RH8 9BB. Tel: 0883713215.

Memory gets batteryback-upThe memory in Instrumatic's ICDemulators can be protected by abattery back-up to overcomeproblems associated with memoryvolatility.

The incoming memory Vcc ismonitored and if an out -of -tolerancecondition occurs a lithium powersource is automatically switched onand write protection isunconditionally enabled to preventcorruption of data. The battery back-up memory option supports standardZAX memory of 128K or extendedmemory board of 512K. lnstrumaticUK Ltd, First Avenue, Globe Park,Marlow, Bucks SL7 1YA. Tel: 0628476741.

.....IL

aim q.).)owt)ove eocivile#40 ittiiihr#E#V4:4:4:066)0,42,44.eilfectid) eeee&tteOte

Compact instrumentation recorderUp to 24 channels are available in theV -Store portable video cassetteinstrumentation recorder. RacalRecorders claims this instrument isthe most compact recorder of its type.

Recording signals from all types ofsensors with electrical output fromd.c. to 100kHz, the unit is ideal foruse in medical, military, automotiveand general industrial research

including vibration. noise andperformance monitoring. Therecorder is simple to operate, useseasily available VHS tapes and has ahigh -contrast electroluminescentdisplay screen with associated softkeys. Racal Recorders Ltd, HardleyIndustrial Estate, Hythe,Southampton, Hants SO4 6ZH. Tel:0703 843265.

68030 computerExisting systems can be updated withthe Plessey Microsystems PME 68-32, a single -board computer which isoptimized for disc -based, single andmultimaster system architecturesthat run on Unix 5.3.

It is suited to real-timeapplications with multi -processorsupport. The board is based on the68030 processor and gives about twoto three times the performance of

previous 68020 -based products.Clock speed ranges from 16 to33MHz. The system has a 4Mbytedram which is dual ported, givingfast access from the processor andVMEbus. The memory is able tosupport the cache burst fill of the68030 processor, giving up to a 15%improvement over cards not havingit. Hi-Tek Electronics, Ditton Walk,Cambridge CB5 8QD. Tel: 0223213333.

Sustained performance at 20mipsA continuous operating rate of20mips with a 25MHz clock rate isoffered by the LR3000. LSI Logic'slatest 32 bit hc-mos riscmicroprocessor chip set.

The chip set consists of theLR3000 c.p.u., LR3010 floating pointaccelerator, and LR3020 writebuffer. The c.p.u. includes a high-speed arithmetic unit and co-

processor interface. It contains a 64entry translation look -aside buffer.support 512Kbytes of cachememory which only requiresmedium performance s. rams. Binarycompatibility with existing R2000software is maintained. LSI LogicLtd, Grenville Place, The Ring,Bracknell, Berks RG12 IBP. Tel:0344 426544.

Fast i.c. designEngineers can specify i.c. designs inan easy -to -use high-level language,simplifying the design process andreducing the chance for error withVLSI Technology's enhanced state -

machine compiler.This version offers the user fast

compilation, automatic stateassignment, readable outputinformation, and an improvedsystem of control over fan -out thatprevents signals from beingoverloaded. The company claims thatin less than a day a designer can entera complex design by simply writing afew dozen lines of instructions at anengineering workstation. Usingtraditional design methods, the samedesign would have taken weeks ofdevelopment. VLSI Technology Ltd,486/488 Midsummer Boulevard,Saxon Gate West, Central MiltonKeynes MK9 2EQ. Tel: 0908 667595.

Plug-in terminalblocksModular terminal blocks that can bebuilt up to any number ofterminations have been introducedby IMO Electronics.

The 20.900 Series has two- andthree-way modules in horizontal andvertical plug-in units for attachmentto on -board soldered pin connectors.To prevent incorrect connectionwithout losing a pole a tiny key isinserted into the block. This avoidsthe need for blank terminals or endcheeks to achieve polarization. IMOElectronics Ltd, 1000 North CircularRoad, Staples Corner, London NW27.IP. Tel: 01 452 6444.

High -voltagemosfetsThere are three new high -voltagen -channel mosfets specificallytailored with on -resistance ratingsthat support the requirements of testequipment and telecommunications.

Siliconix claims the VN50300,which is an ideal voltage -spikeprotector, is the first 500V deviceavailable in the SOT -23 surface -mount package. The other twodevices are 400 and 350V modelsproviding high -voltage, low on -resistance operation in systemsusing low-level power supplies orc-mos logic. Siliconix Ltd, 3 LondonRoad, Newbury, Berks RG13 1.1L. Tel:0635 30905.r High -Voltage Small -Signal MOSFETs

550

400

1. 250

WW2Obit 10-101

4antA

10 50

r0u05,(Olns)500 1000


Low-cost Programmable PSUs from

Thurlby

GOO OCG,

0 III,

3000 -.gx

II

PL -GP series IEEE -488 controlled PSUs30V/2A and 15V/4A single and twin unitsSeries/Parallel connectable for higher outputProgrammable to 10mV and 10mA resolutionReadback of current demand via the busConstant voltage or constant current operationBench mounting or 19" rack mountingSingles £445 + vat, Twin units £650 + vat

Thurlby Electronics Ltd., New Rd.,

("Thurlby' SptE.11v7e4sB, GHuinteinipdason,8Cco

63570ambs.

A

ck

Low-cost Precision DMMs from

Thurlby

1905a 514 digit computing DMM £349+ vat0.015% accuracy; 1µV, 1mi2, lnA sensitivityAveraged AC standard, true RMS optionalNulling and relative measurements; linear scalingPercentage deviation; digital averagingdBV, dBm, general logarithmic displayLimits comparison; min. and max. storage100 reading timed data loggingRS -232 and IEEE -488 interface options

Thurlbyivs, EHlecttrionngidcosn cLtd.a,mNbews Rd.,

/I Thurlby StePE17 4BG. Tel: (0480) 63570

ADVANCEDACTIVE AERIAL

ALL THE

We are not decryingI. No longer2. No longerbehaviour statistically.

.

must youmust

SCOPE

today swait

you wait

YOU MAY EVER NEEDON PAPER!

excellent test instruments but we think you ought to knowuntil a circuit is built before you can test it and

until it is in production before you can sample its

. . .

- .-1-----77-___H_.:,

..... ,.

ICI s,..

,

-....

The aerial consists of an outdoor head unit with a controland power unit and offers exceptional intermodulationperformances: SOIP +90dBm, TOIP + 55dBm. For thefirst time this permits full use of an active system aroundthe If and mf broadcast bands where products found areonly those radiated from transmitter sites. General purpose professional reception 4kHz -

30MHz. - 10dB gain, field strength in volts/metre to 50 Ohms. Preselector and attenuators allow full dynamic range

to be realised on practical receivers and spectrumanalysers.

Noise - 150dBm in 1Hz. Clipping 16 volts/metre. Also50 volts/metre version.

SURREY ELECTRONICS LTD.,The Forge, Lucks Green, Cranleigh.

Surrey GU6 7BG.Tel: 0483 275997

Waveform through switching circuit Effect of

Those Engineers have always offered the highest performancesimulation. The latest version of ECA-2 is stacked even higherprogrammable signal generator. Monte Carlo and Worst Caseanalysis of transients) and is up to four times faster thannodes and 2500 components can be modelled. Its non-linearswitches. diodes (including zeners). JFETs. MOSFETs.motors and etc to be modelled complete with any saturationand signal conditioners can also be modelled usingimportantly. despite its capabilities. ECA-2 is easy to usenow

-available

-'"-ble

to schematic and PCB design software.

,

- - - i,

I

.....

Internal generator programmed to produce sine Waveformwave with exponential amplitude decay

II you would like to see now easy it is to debu your designscost of many a scope. phone us for a free IBM PC compatibleCOMING SOON: GES1 graphic entry system, LCA1 logicPlease send for details.

Moo®

component

before.

BJTs.characteristics.polynomial

and

IIII,ttill

through

circuit

tolerances

inwith useful

toleranceCircuits

representationsSCRs.

user-friendly.

modulatingoffset

on paperdemonstration

analysis.

analogue

analysis.

discharge

functions.

for

-,in a filter

circuitfeatures (fully

Fourierof over 500

allowlamps.

TransducersMost

Interfaces are

III .

1111

.

circuit with

less than thedisk.

..,

- o.:.laavItIA@LTo 112c1000c ROPCSaffloce Ansa) Mae °C;VAII V,Inrdpastioi ° &,c7.0,quo NCA3 019)3Tel 01135 2771 E 1.4.. One to One B,. 23332001 Ti. P950511 tansbY ON F ONf G, Quol,n9bo. 23332001

ENTER -13 ON ItERLY CARD

928

39 :PIA' CARD


RESEARCH NOTESSpace-agevision aid

Researchers at NASA and at theJohns Hopkins Wilmer Eye Insti-tute are to adapt a vision systemdesigned originally for spacerobots to restore sight to thealmost -blind. It's essentially aportable closed-circuit televisionsystem that will produce brightpictures directly in front of theeyes via what looks like wrap-round sunglasses. Instead oflenses, the sections covering theeyes will actually he small flat tvscreens that the person will lookat rather than look through.

What's even more ingeniousthan the display system is thecamera system to pick up imagesof the outside world. Obviouslyno normal person would want towalk around with a standard tvcamera mounted on a hat. and soa method had to he devised ofdisguising this inevitably bulkyitem. The answer researchershave come up with is to use tinylenses and bundles of opticalfibres embedded in eachthe spectacle frame where thefront and ear pieces join. Light isthen conveyed along the fibres toa c.c.d. camera in a pack worn onthe hip or shoulder.

NASA's existing black -and -white version designed for re-mote handling applications inspace will form the basis of thenew vision aid. But before it canhe of practical use for handicap-ped people on Earth the systemwill have to be converted to fullcolour and also made capable of adegree of video processing.

The team emphasizes that theportable vision aid is not in-tended to restore totally lost vi-sion nor to correct the manyoptical defects that are more -or -less adequately dealt with byspectacles. The main function isto provide an increase in sensi-tivity for those who are almostblind. It's therefore a sort ofpicture amplifier.

Dr Robert Masoff. a researcherat the Johns Hopkins Wilmer EyeInstitute says the new deviceshould he able to help an esti-mated 2.5 million people in theUSA alone. It won't however.only help those who have lowvisual sensitivity. Because of thevideo processing capability. itwill be possible to enhance con-trast or to accentuate edge

effects or even introduce falsecolour to help those with specificforms of colour blindness.

The whole project, lasting fiveyears and costing II 1 millionshould. say its developers, lead toa practical device that is afford-able by elderly people on fixedincomes. In more concreteterms they talk in terms of thecost of two colour television sets.

Memories thatlast for ever?

New memory chips being de-veloped by various US companiescould largely supersede eproms.That's the view of some expertswatching the emergence of ex-perimental memories based onthe ferroelectric effect. This is aproperty whereby certain crys-talline materials such as leadzirconate titanate become pola-rized when subject to anexternally -applied field. Thispolarization is non-volatile, butcan be reversed by the applica-tion of a field of opposite polarity.In practice, ferroelectric mate-rials behave rather like bistablecapacitors with zero leakage. Theonly problems in developing apractical application for suchseemingly ideal materials hasbeen the slow 'write' speed andthe rather indefinite polarizationthreshold.

Research into new and as -yetsecret materials is now comingup with some highly promisingprototype rams offering highread/write speeds and non -volatility. Raintron Corporation,one of the companies concerned,predicts that data stored in someof its chips will remain uncor-rupted for 100 years or more.

As well as developing newmaterials, one of the importantgoals of research into ferroelec-tric rams has involved findingways of integrating such mate-rials into the memory circuits ofexisting technology. Here differ-ent researchers adopt differentapproaches, some favouringdirect substitution of conven-tional capacitor elements, othersuse the ferroelectric material as a'power -down' memory backup.

Either way, it looks as if it'sjust a matter of time before chipsappear on the market that aremuch faster than today's epromsand which can tolerate manymore write cycles before they

wear out. It also appears that aferroelectric memory isradiation -hard, which will give itinstant applications in space aswell as the nuclear industry andmilitary projects.

Electronicspeech therapy

Significant advances in the use ofa computer -based technique toaid speech therapy were de-scribed at the first National Sym-posium on the Clinical Applica-tions of Electropalatography inReading.

The symposium reviewed thework of 28 speech therapists andexperimental phoneticians whouse a technique called electropa-latography (EPC), whereby anartifical palate with 62 electrodesis placed in the mouth and thecontact of the tongue with thematrix of electrodes is recordedon a personal computer and dis-played on the screen. The EPGsystem gives access to detailedand previously unobtainable in-formation about tongue positionon the roof of the mouth.

In speech therapy. a patient'sabnormal pattern of tongue con-tact can be displayed alongside amodel pattern which the therap-ist has produced using a separateEPG palate. This visual feedbacktherapy technique has beenfound to be particularly effectivefor children with developmentalspeech disorders, for those withcleft palate, and for stroke pa-tients who have been left withspeech problems.

An improved version of theEPC has resulted from progressmade recently by a team com-prising researchers from the IBMUK scientific centre in Winches-ter and from the department oflinguistic science at ReadingUniversity.

Dr Bill Hardcastle, Reader inLinguistic Science at ReadingUniversity, told the symposiumthat advances in the EPG systemmean that it is now a viable andproven technique for both ther-apy and speech research. EPCrecords have shown that indi-vidual speech sounds such as 't','k' and 's' have specific tonguecontact patterns and that thesepatterns can change in predict-able ways depending on a num-ber of factors like how fast theperson is speaking, and whichsounds precede or follow, and soon. By comparing these patternswith those produced by a speakerwith a speech disorder, the clini-cian is able to pin -point thenature of the errors more pre-cisely than by using traditionalprocedures. The extra precisionin diagnosis means that the clini-cian can now not only plan moreefficient therapy but can alsooften gain valuable pointers as tothe underlying cause of the dis-order.

Computerbiology?

Further evidence that today's se-quential von Neumann compu-ter architecture may be on theway out comes from the relative-ly recent re-emergence of in-terest in neural networks -attempts to model electronicallythe behaviour of neurons, ornerve cells in the brain. Part ofthe aim is to try and gain insightinto how the human 'computer'.based on an incredibly slow logicelement. can outperformmachines at so many complextasks. In fact. the more complexthe task. the greater the brain'sadvantage. Try getting amachine to recognize someone ithasn't seen for 20 years!

One of the fascinating aspects

Wrong (left) and right: visual feedback from tongue and palate. MIL

OOOOO


RESEARCH NOTESof neural networks is their un-predictability. Based as theyusually are on thousands of in-terlinked processors, they aren'talways susceptible to beinganalysed by conventional Bool-ean algebra. In fact neural net-works - like their biologicalcounterparts - are self -programmed to the extent thatthey can sometimes learn newand unsuspected tricks. All onehas to do, it seems, is teach themachine by its own mistakes.

A recent example of thisapproach was an attempt by re-searchers at MIT and the Uni-versity of California to build aneural network capable of im-itating the brain's ability to lo-cate objects in space. The net-work consisted of three layers ofprocessors, one to accept inputdata. an uncommitted middlelayer and an output layer toassemble the answers. Input dataconsisted of electronic analoguesof eye position and visual fieldand. initially, the network was'told' what it was looking at. i.e.what 'answer' to expect. Afterabout 1000 trial runs, the net-work had 'learned' to recognizeand locate some remarkablycomplex shapes.

Another intriguing aspect ofthese neural networks is theirtheoretical basis.

As I mentioned earlier. theydon't readily yield to convention-al mathematical analysis. Theirproperties do, however, showstriking similarities to those ofthe materials that comprisebiological brains. And just asbulk properties of matter such astemperature or pressure can bededuced from molecular be-haviour. the behaviour of neuralnetworks can sometimes be de-duced from the way the indi-vidual constituent processors in-teract.

If all this seems a flight ofjournalistic neuronal fancy, con-sider one of the more advanceddevelopments in the USA. a'biological' computer atCarnegie-Mellon University. Notonly does it outperform sequen-tial processors at a variety oftasks, its behaviour can to someextent be predicted by the princi-ples of thermodynamics! Tomor-row's programmers, it seems.will have to forget all they everlearned of high-level languagesand take a few lessons in biophy-sics.

Enter theelectronic

dentistAn unexpected and unwelcomevisit to the dentist recentlyprompted me to take a closerlook at what seemed at first sightto be an excellent d.i.y. methodof spotting rotten teeth.

Studies at three U.K. dentalcolleges (Br. Dent. J. vol 164 No8) have shown that simple resist-ance measurements can spotearly caries in 85% of cases.

Research over 30 years agoshowed that, when dried with ablast of air. a sound tooth ex-hibits a tip -to -gum resistance ofgreater than 600kf 1. Under simi-lar conditions. a tooth with pitsor fissures is likely to read lessthan 2501(11. But because lots ofwell -established methods ofspotting dental caries already ex-ist, the plug-in approach hasnever really caught on.

There is. moreover. a realproblem with it: electronic tes-ters have often given low read-ings on teeth that appear perfect-ly sound. In fact three out ofevery four teeth fall into thiscategory. A dentist can't juststart drilling and filling everytooth that appears normal butmeasures less than 250ki1.

What this latest study did wasto apply the technique prospec-tively to patients due to haveapparently sound teeth extractedfor cosmetic reasons. Fifty teethwere measured in vivo and alsoinspected in the normal way.After extraction the same teethwere sliced up and subject todetailed microscopic examina-tion.

Result? Six were bad accord-ing to the dentists, 28 accordingto the electronics and 37 inactual fact. The electronic tech-nique therefore appears muchmore likely to spot early decaythan standard dental techniques.It also generates relatively fewfalse positives. But as the authorspoint out. this doesn't meandentists are doing a bad job. Notmany of the electronically identi-fied cases of decay would yet beready for filling. Nevertheless thenow -validated technique offersgreat potential for catching earlydecay and treating it in otherways. But before you start tryingto eat a set of Avo probes, I

should point out that for accu-rate results the probe needs to beprovided with a coaxial blast ofdry air - a bit like a miniatureMIG welder. Without that essen-tial item the results are likely tosend you rushing off premature-ly to the nearest false teeth sup-pliers!

Smashing spaceparticles

Physicists at the Los AlamosNational Laboratory have used aVan de Graaff generator toaccelerate particles of iron10-hm in diameter up to veloci-ties of 105m/s, a hundred timesfaster than a high -velocity bullet.The idea is to simulate whathappens when a spacecraft is hitby showers of micrometeorites.

When we think of collisiondamage, most of us think morenaturally of cars, aeroplanes orperhaps missiles. But out inspace these macro -scale collis-sions aren't always the best mod-els, nor are they necessarily agood basis from which to ex-trapolate. Free from atmospher-ic drag, space dust can reachtruly enormous velocities.accelerated by the gravitationaland electrical fields of the solarsystem.

Any 0 -level physics studentknows that, in terms of damagecaused, the size and velocity of aprojectile are in no way equiva-lent. If you double the size of acolliding particle then it has dou-ble the inertia and causes rough-ly double the damage. On theother hand if you double thevelocity of a given object it hasthe potential to inflict four timesthe damage. So even a particlethe size of fine smoke can belethal if it's travelling fastenough.

But in devising their latestexperiments to acceleratesmoke -sized particles of iron, theLos Alamos team were well awarethat you can't just use 0 -levelequations to extrapolate to theconditions found in space. A one -micron -sized particle travellingat lIrm/s will, for example, vapo-rize a target up to 1000 times itsown weight. So it's scarcely justan extension of what happensevery day on the M25!

By observing exactly whatdoes happen under these ex-treme conditions, the physicists

hope to develop mathematicalmodels to predict in detail theprocesses of melting and vapor-ization that take place. Prior tothe construction of their 140 tonVan de Graaff machine. whichhas an energy level of 8MeV, thebasic data was simply not avail-able.

Although there is obviously amilitary interest in how to 'har-den' spacecraft against theeffects of natural micro -meteorites, this latest work isdefinitely not the beginning of anew type of beam weapon. Parti-cles travelling at 10'm/s can sur-vive only in the vacuum of space.And as far as the prospect ofputting a particle accelerator inorbit, the Los Alamos team ques-tion the desirability of launchinga 140 ton gun that fires nothinghilt d putt

Bog standardIn this enlightened age of sexualequality there are some thingsthat obstinately refuse to beequalized. A statistical surveyundertaken by the EngineeringDepartment at Cornell Universi-ty has revealed an importantdiscrepancy between the relativetime men and women spend intheir respective 'rest rooms'. Re-searcher Anh Tran. who lurkedfor hours with a stopwatch out-side motorway loos in Washing-ton State, says that men, onaverage, take 45 seconds (stan-dard deviation 3.73) to use thefacilities while women take 79s(s.d. 4.47). A university spokes-man described this as 'signifi-cant'.

Tran has developed a simplecomputer program to make iteasier for architects and desig-ners to calculate the mostappropriate ratio of male andfemale excretory facilities in anygiven area. The program runs onan Apple Macintosh and needs tobe supplied with only the bareminimum of data.

One immediate conclusion isthe need, in most situations, toalter the generally -accepted50:50 ratio to something nearer60:40 in favour of the ladies. Rollon the paperless office.

Research Notes is written byJohn Wilson of the BBC ExternalServices science unit at BushHouse. London.


NEW 8051 DEVELOPMENT CARD

0ICOM

The new Cavendish Automation development card carries a full symbolic Assembler and text editoras well as the MCS-BASIC 52 package. It will allow the user to write applications programmes ineither BASIC or Assembler.The text editor supports ORG, LOC, HIGH and LOW directives as well as the current location ($) andthe - and - operators. Full source text editing is included, and the source file as well as assembledcode may be blown into PROM/E2PROM on -card. A powerful feature of the system is that a functionlibrary of over 60 routines within the interpreter may be accessed using assembly language CALLinstructions, enabling simple negotiation of floating point, logical operations, relational testing andmany other routines.

iArr:

ITINC-111

cc-CCgC OOOOO CR

soccosiss :

"KIMORPROPRP17PP 'u.=.

Mamma

<St:. VV.Ciotti Mdxr

" " I1 \ ' '

FEATURES:

Only requires 4-5V supply anddumb terminal

Save assembled code or sourcetext in PROM on -card

Card I/O includes 9 x 8 -bit portsand 2 serial lines.

II Very fast interpreterspecifically written to accesscapabilities of '51 Family

II 32K user RAM, 16K user PROM(RAM jumpered to access codeor data space)

Card supported by over 50other types of CA I/O and CPUtarget cards

So, for professional implementations at super -low cost, call us on (0480) 219457.

Cavendish Automation, 45, High St., St. Neots, Huntingdon, Cambs PE19 1 BN. Tel: 0480219457. Telex: 32681 CAVCOM G


COMMERCIAL QUALITYSCANNING RECEIVER

soma IC -hood

ae

The IC -R7000, advanced technology, continous coverage communications receiverhas 99 programmable memories covering aircraft, marine, FM broadcast,Amateur radio, television and weather satellite bonds. For simplifiea operationand quick tuning the IC -R7000 features direct keyboard entry Precise frequenciescan be selected by pushing the digit keys in sequence of the frequency or by tur ningthe morn tuning knob FM wide/FM narrow/AM upper and lower SSB modes with6 tuning speeds 0 1, 1 0, 5, 10, 12 5 and 25kHz A sophisticated scanning systemprovides instant access to the most used frequencies. By depressing the Auto -Mswitch the IC -R7000 automatically memorises frequencies that are in use whilst it

in the scan mode, this allows you to recall frequencies that were in use Readout.s clearly shown on a dual -colour fluorescent display. Options include theRC.12 nfio-redremotecontroller voice synthesize, and HP - I headphones

tees. (UK) Ltd. Tel 0227 363859 Telex 965179 ICOM GN B Authorised Welsh distribution byMRS Communications Ltd. Cardiff Tel 0222 24167

Please send information on 'corn products & my nearest !cam dealer.1

IName/address/postcode:

Status: Tel:Lt to:kom (UK) Lid. Dept. WW, FREEPOST, Herne Bay, Kent CT6 8BR j

ENTER 80N REPLY CAR I)

HART - The Firm for QUALITYLINSLEY-HOOD 300 SERIES AMPLIFIER KITS.

Ultra high quality. Mosfet output. Fully irtegrated Hi -Fr amplifier kits by thisfamous cesigner Two models. 35 and 45 watts per channel Capable ofsuperb sound quality with greater delic icy and transparency of tone thanmost commercial amplifiers. Building is very easy with our comprehensivebuilding instructions as most components lit on the PCBs and setting -uponly needs a multimeter15300.35 Total Parts Cost 0138 28. Discount Price for Complete Kit 198 79 K300-45 Parts Cost C142 74 KitDiscount Price C102 36RLH485 Reprints of Original Anicles I rem Hi-Fi News CI 051FREE with Kit)

LINSLEY-HOOD SUPER HIGH QUALITY AMFM TUNER SYSTEMA combination of his ultra high quality FM tuner and stereo decoder described in 'ETI and the SynchrodyneAM receiver described in WW' Cased to match our 300 Series amplifiers this kit features a ready builtpre -aligned FM front end. phase locked loop IF demodulator with a response down to DC and an advancedsample and hold stereo decoder This tuner sounds better than the best of the high-priced exotica but thanksto HART engineering. remains easy to build1(400 -FM FM Only Kit Complete C134 61K400AM'FM Full AM/FM Kit C205.92

GREAT NEWS FOR KIT BUILDERSAs a result of popular demand we are low offering BLACK Irontplates with matching collet knobs on our 300Series amplifiersPrice as option with kit 15 20Price as upgrade for existing kits C 15 90Tuner versions will be ready soon

SOLENOID CONTROLLED FRONT LOAD CASSETTE DECKHigh quality 10 08°,W8F) successor to our very popular SF 925F Offers allstandard facilities plus cue and revue modes all under remote logic ofsoftware control Power requirements are simple with 12v solenoids and12v motor with built in speed control Logic and driver chips are availableDeck with stereo head, other heads to order £42 20Full manufacturers data £2 90

HIGH QUALITY REPLACEMENT CASSETTE HEADSDo your tapes lack treble, A worn head could be the problem Tape headsare constantly improving and frying ome of our latest replacement headscould restore performance to better than new, Standard mountings fitmost decks and our TC1 Test Cassette will make it easy to set theazimuth spot on As we are the actual importers you get prime parts atlowest prices All our heads are suitable for Dolby machines

6,0HX100 Standard Stereo Permalloy Head C2 49HC20 High Quality Permalloy Stereo Head £7.66HS16 Sendust Alloy Super Head Ouie simply the best Longer life than permalloy higher output than ferrite.fantastic frequency response metal tape capability C14 86H0551 4 -Track head for auto -reverse or quadrophonic use. Full specification record and play head C14 60Full data on these and other heads in nur range are contained in our free list

HART TRIPLE -PURPOSE TEST CASSETTE TC1One inexpensive test cassette enables you to set up VU (Dolby) level, head azimuth andtape speed without test Kw pment Viral when fitting new heads Complete with instructions [4.66

Send for your FREE copy of our lists with full details of our complete range of Kits. Components, PCBs.Casserte Heads and Decks - Overseas please send 5 IRCs for Airmail Post

ELECTRONIC KITS LTD1, Penylan Mill, Oswestry, Shropshire SY10 9AF24 hr SALES LINE (0691) 652894 Please add VAT



-TELEVISION BROADCASTA licence to lose

money?Among all the media hype sur-rounding d.h.s. from Astra andthe BSB satellite. it is easy toforget the extent to which, only afew years ago. we were all beingsimilarly persuaded of the won-ders and imminent coming ofthe wired city with its widehandinteractive electronic grid. Therewere endless articles and pro-grammes about Warner's QUBEsystem. Far fewer marked itsdemise after a few years whenWarner found they could notprovide enough popular prog-ramming. while subscribersfound the control pad too com-plex and did not take kindly tothe pay -per -view concept ratherthan the monthly subscriptionbasis of most American cablenetworks. Furthermore thescrambling-descrambling sys-tem degraded the picture qualityto a noticeable degree.

Similarly, the INDAX systemrun by Cox Cable in San Diego.offering banking. shopping andpay -per -view facilities, was un-able to convince many viewersthat its programmes justified theexpensive converter boxes.

Nevertheless. less ambitiousforms of interactive cable areagain expanding in the USA.most of the "return" paths usingthe normal telephone network orworking on a store -and -forwardbasis rather than instantaneousservice. Some 16 million homeswere connected to such systemsat the beginning of the year. hutthe full potential of the electro-nic grid seems as far away asever. Even the efforts of Amer-ican cable companies to providedata links for urban businessusers have so far failed to pro-duce the predicted revenues.

The success of any televisionchannel or video system dependsin the long term on providingprogrammes capable of attract-ing and satisfying large numbersof viewers as well as being re-latively cheap and easy to use.Provided that it is easy to oper-ate, technology appears to playonly a minor role. The Govern-ment's suggestion (since with-drawn) of changing BBC2 andChannel Four Television intosatellite subscription services,while filling terrestrial channels

flew deregulated, competingadvertisement and sponsorfunded channels flew in the faceof American experience - andwas a sure way to lower thequality of British television.More choice does not necessarilymean worse "free" programmes- but I would not bet on that!

The efforts over the past de-cade of Australian andAustralian -born media mogulsto move into UK commercialbroadcasting have been paral-leled by their efforts to provideservices in the Pacific islands.The Niugini Television Networkin Papua. New Guinea, set up bythe Parry Corporation at a cost of$A 9M closed down after fourmonths as a non -viable opera-tion. The Packer Group, takenover by Alan Bond. won the rightto set up a service in Fiji in theface of stiff competition. This wasdue to start late last year but wascancelled following the militarycoup on the island. Now TV NewZealand is anxious to establish aPacific regional service centredon Fiji. but finds that the BondGroup retains exclusive rightsfor twelve years.

US masttragedy

Failures of very high televisionmasts are fortunately rare inEurope and when they haveoccurred have usually not re-sulted in casualties. The Amer-ican experience seems lesshappy.

Early in June a tragic accidentbrought down the new 610 metretriangular lattice mast of KTVO.Kirksville. north-east Missouri.resulting in the death of threeerectors who were fitting newbracing rods at a height of 146m.

The o£1.1M mast. one of anumber of American televisionmasts reaching the maximumheight permitted by the FCC.was comparatively lightlyloaded. It carried two circularly -polarized antennas for the100kW transmitters, some mic-rowave dishes and an f.m. radioantenna. It served viewers inparts of Missouri. Illinois andIowa. Services were restored onthe earlier, lower mast whichhad not yet been dismantled.

Soon after the new mast was

commissioned in September1987. cracks were found in twobracing rods. After furthercracks were detected, the deci-sion was taken to replace all 113(1rods. It was during this work thata sudden slippage of the chainwinch used to provide temporarybracing caused the mast to twistand begin to collapse. Two of theriggers were crushed beneaththe top 464m section of mast as itfell to the ground. becomingembedded to a depth of up to 5m.The third man jumped to hisdeath from the work -basket.

A detailed report in New CivilEngineer (June 9) by SimonMontagu notes that "the fre-quent collapse of US masts dur-ing maintenance or because oficing has in the past failed toraise the self-governing associa-tion EIA to action". He forecaststhat lessons from the collapse ofthe KTVO mast seem unlikely tobe passed on to the US engineer-ing fraternity.

Four -bit digitalaudio

With linear p.c.m. coding. 14

hits with 32kHz sampling of a15kI1z audio channel results in ahit -rate of 448khit/s. Withd.p.c.m. coding this reduces to256kbit/s. Delta modulation withadaptive pre -emphasis brings therate down to 203khit/s. In prac-tice the CD stereo data rate isabout 1.4Mb/s. Nicam compand-ing to 10 hits provides a stereohit -rate of 728kbit/s. As noted inthe February "Radio Broadcast"column. the Thomson Group hasdeveloped in Germany a form ofadaptive transform coding pro-viding a stereo data rate of256kbit/s.

S.M.F. Smyth and J.V. McCan-ny of Queen's University, Belfastare making strong claims (Elec-tronics Letters, 14 April, 493-5)for four -bit a.d.p.c.m. coding togive a bit rate of 128kbit/s for a15kHz channel directly applic-able to the European ISDN tele-communications hierarchy.They claim the results they haveachieved are of major import-ance not only for ISDN andbroadcasting, but also for otherdigital audio technology includ-ing CD records and digital audiotape.

Their coder consists of a two -hand quadrature mirror filterhank incorporating backwardadaptive prediction and quan-tization in each sub -hand. Theaudio signals are sampled at32kIlz and each 16 -hit sample isfiltered into two frequencybands, coded to four bits fortransmission. Work is also inprogress to implement thismusic coder in real time. It hasalready been shown that a fullduplex implementation is possi-ble on a single 1(10 nanosecondfixed-point d.s.p.

The authors claim that theirresults clearly demonstrate thevery high coding efficiency ofsub -hand a.d.p.c.m. when com-pared with digitally compandedand a.d.m. schemes applied tohigh-fidelity music. They write:"We believe these findings arevery significant and represent amajor advance in digital audiotechnology, having immediateimplications for ISDN. broad-casting and d.h.s. digital audiodistribution ... In additiona.d.p.c.m. is known to offer biterror immunity down to 1 in 10'without any form of protectionor concealment. four or fiveorders of magnitude better thanmost p.c.m.-based systems ... weseriously question the con-tinuing viability of p.c.m. as areliable and economic means ofstoring or transmitting high -quality digital audio".

The Electronics Letters reportincludes details of subjectivelistening tests which appear toshow that music recovered fromthe four -bit coding is essentiallyindistinguishable from the ori-ginal material. It would be in-teresting to see independenttests carried out, preferablybased on the EBU test record (see"Radio Broadcast". page 9361 onboth the Thomson 256kbit/sstereo system and this latest con-tender. While the lower bit rateswould be attractive for audiocontribution and distributionnetworks, one wonders what thereaction would he to any sugges-tion at this late stage of changingexisting standards for d.h.s. orterrestrial broadcasting! Alreadythe Nicam 728 system is underthreat from Rupert Murdoch'sPAL plans for his Astra channels.

Television Broadcast is writtenby Pat Hawker.


KESTRELELECTRONICCOMPONENTS LTD.

All items guaranteed to manufacturers spec. Many other items available.'Exclusive of V.A.T. and post and package'

s 50+ 1+ 50t74LS125 0 15 0.12 8 Meg 0 45 0.35

74LS244 0.30 0.21 16 Meg 0.45 0.35

74LS245 0.30 0.21 IL0-74 1.50 1.35

74LS373 0.30 0.21 1488 0 25 0.16

6809 2.50 2.20 1489 0.25 0.16

6821 1.25 0.95 LM2901 0.80 0.60

6845 2.50 2.20 LM2917-8 1 50 0.75

6850 1.60 1.40 2732-A 2.50 2.30

6502 2.80 2.40 2764A-25 2.20 2.00

6522P 2.80 1.85 27C64-20 3.00 2.60

8031 2.80 1.70 27128-25 3 70 3.40

8085 1 60 0.95 27C128A-25 3.40 3.10

8255-5 1.90 1.30 27128A 2.98 2.72

LM319N 0.50 0 40 27256-25 3.30 3.12

32.768Khz Crystal 0.25 0.16 27C256-25 3.50 3.22

1.8432 MHz 0.95 0.70 43256C 12L 10.00 9.00

2.4576 MHz 0.75 0.65 ICL8211CPA 0.80 0.60

3.5795 MHz 0.50 0.35 TL084CN 0 50 0.45

4 Meg 0.45 0.35 Z8OACPU 1 00 0.80

4 194304 0.45 0.35 Z8OACTC 1 00 0 80

6 Meg 0.45 0 35 Z80AP10 1.00 0 80

All memory prices are fluctuating daily. please phone toconfirm prices.

178 Brighton Road.Purley. Surrey CR2 4HATel: 01-668 7522. Fax: 01-668 4190

1 Eli I 1 U\ 11E1'1.1 .%1t1)

R.F. COMPONENTS AND

ELECTRONIC VALVES/TUBESPRODUCT RANGE

CRT's R.F. RESISTORSCAMERA TUBES R.F. SCREENED

CONNECTORSR.F. RESISTORS

(MRF. SD, 2N, 2C etc)R.F. INDUCTION/

DIELECTRIC SPARESRECEIVING TUBESRECTIFIERSSPARK GAPSTETRODESTHYRATRONSTRANSISTORSTRANSMITTING TUBESTRAVELLING WAVE TUBESTRIODESTUBE SOCKETSVACUMM CAPACITORS

CHOKESCARBON RESISTORSDIODESHEATSINKSIGNITRONSIMAGE INTENSIFIERSINTEGRATED CCTS

(Linear. Audio)KLYSTRONSMAGNETRONSMICROWAVE TUBESPHOTOMULTIPLIER TUBESPLANAR TRIODESR.F. CAPACITORSR.F. CONNECTORS

Mail order service available. Please specify which productsyou are interested in when replying.

Export enquiries welcome.

RAEDEKELECTRONICS CO.

BANNERLEY ROAD. GARRETTS GREEN.BIRM NGHAM B33 OSL

Tel 021-784 8655. Telex: 333500 CROSAL G.Fax: 021-789 7128

I \ I I I; 11, \ 1;1 \I;1)

The Circuit Designer'sDream . . .

L.

O

I dreamt that thecomputer was laying thetracks automatically-

right there on my IBM PCscreen...

/ dreamt that thecomputer was drawing

perfect schematicdiagrams for me . .

I . becomes a reality!The PROTEL family is the answer to all these

dreams - at prices you can afford!When you look at its price, specification andease of use, it's not surprising that PROTELis fast becoming a world standard for LOWCOST PROFESSIONAL CIRCUIT DESIGN

CAD on any IBM PC or compatible.

TRY FOR YOURSELF!For £25 per program you can try our "full

facilities" evaluation packs.

arn becomes

realitywithPrOtelin action

Engineering Solutions LimitedKing's House, 18 Kings Street,

Maidenhead, Berks. SL6 1EF

Tel 106281773444 Telex 849462 Fax (0628)74928

ENGINEERINGSOLUTIONSLIMITED

ENTER .12 UN IMPLY It It

ELECTRONICS & WIRELESS 11ORLD 933

RADIO COMMUNICATIONSCoping with

e.m.c.?An open day at the DTI's KenleyRadio Technology Laboratoriesnear Whyteleafe, Surrey, re-vealed measurement facilitiescapable of providing accuratetype -approval checks and meet-ing all current and proposede.m.c. radiation and immunitytest needs. The facilities, nowbeing offered on a commercialbasis to industry for the pre-testing of equipment, are seen asan essential requirement whenthe European Community EMCDirective is finally approved.(The draft is still being revisedand it now looks unlikely that theCouncil's common position canbe reached until at least Septem-ber.) UK legislation will also benecessary if certain British Stan-dards are to become enforceable.While Kenley is ready, the DTIhas been one of the principalobjectors to some clauses of thedraft Directive. arguing inter aliathat it cannot be enforced on allapparatus in the absence ofnational standards.

Yet as one of the Kenley de-monstrations showed clearly,home computers currently onthe UK market are failing by awide margin to meet the stillunenforceable BS6527:1984 cov-ering radiation from informationtechnology equipment. A popu-lar computer was shown produc-ing 40dB(µV) radiation at128MHz. 10dB above the BS6527limit, measured under the speci-fied conditions.

RTL's facilities, covering 9kHzto 22GHz, lean heavily on Hew-lett Packard and Rohde &Schwarz spectrum analysers andmeasuring receivers. ChaseElectronics assisted with a de-monstration of V -network andclamp techniques for measure-ment of cable -conducted emis-sions, showing clearly the widespectrum of emissions from aswitch -mode power supply. Ples-sey Assessment Services collabo-rated in a demonstration of elec-trostatic discharge testing inaccordance with BS5967 Pt 5.Fundamental and harmonicradiation from microwave ovens(potentially a source of interfer-ence to 12GHz d.b.s. reception)is measured in a reverberationchamber with rotating vane,

although RTL is not responsiblefor investigating the possiblehealth hazards that could arisefrom defective door seals in oldovens.

Immunity testing of televisionreceivers and other appliances iscarried out with a large t.e.m.cell; although with no v.h.f. tele-vision in the UK, it seems oddthat this is suitable for a sourceonly up to 150MHz. A largeenvironmental test chamber isused for bulky equipment suchas marine m.f./h.f. transmitters.Case radiation from transceiverscan be measured down to a nano -watt to 4GHz by substitutionmethods using a large cornerreflector to minimize the effectsof wall reflections etc. Receiveroscillator radiation from mobileand c.b. transceivers can bechecked over the range 100kHzto 2Gliz to determine that thisdoes not exceed 2nW below1GHz (20nW above I GHz).

Kenley is also the base for thenew Radio Investigation Servicemobile laboratory, covering9kHz to 40GHz and fitted withtwo pneumatic masts. one ofwhich can extend up to 28m.There is a quiet -running on-board 3.5kW generator. Used toback-up the RIS field teams, par-ticularly for the investigation ofindustrial and radar interfer-ence, it surely represents goodvalue at the standard RIS rate of141 per engineer per hour.

Essentially, Kenley remains aprecision measurement labora-tory rather than an e.m.c. designadvice centre. But what may beneeded most by small electricaland electronic appliance makers,when finally the EMC Directiveand new UK legislation comeinto force. will be expert advice atan early design stage. The openday left one wondering whetherKenley is yet set up to meet thisrequirement, which calls fore.m.c. design specialists ratherthan precision instruments.

DTI has recently introducednew type -approval testing proce-dures for land -mobile radioequipment. Type approval test-ing of MPT1300 series specifica-tions can also be carried out byERA Technology at Leatherhead;Plessey Technology Services atFareham: and RFI Ltd at Basing-stoke. DTI is introducingcharges for all type approval test-ing at Kenley at rates subject tocontract.

Superconduct-ing dipoles

Theoretically, an electricallyvery short dipole antenna ele-ment could be as efficient as aresonant half -wave element. Inpractice, this cannot be achievedsince there are always high r.f.ohmic losses in the element andmatching network relative to theradiation resistance, which fallswell below 10 for an electricallyvery short element. Clearly, ashort element and associatedmatching lines without measur-able r.f. ohmic resistance wouldgreatly improve the efficiency.

A team at the University ofBirmingham supported by ICItechnologists has reportedachieving a gain at 550MHz of10.56dB (as against a compara-ble copper wire dipole) by usingone of the new "rare earth"superconducting ceramics:(Electronics Letters. 14 April,1988, 460-1).

At room temperature theradiation resistance of the testantenna was 0.2611 and the lossresistance 3.512. When the cop-per element was reduced to thetemperature of liquid nitrogen.without of course superconduct-ing, a gain of 6dB was achieved.

Rising abovethe waves

Although e.l.f. waves can pene-trate to a considerable depthunder the surface of the sea, theyprovide only a one-way radio linkwith submarines and then onlyat an extremely low data rate. Fortwo-way communications, sub-marines still depend on raisingan antenna above the waves. Yet.today, as A.J. Meaden points outin PESL New Technology(Spring 1988): "The strategic im-portance now assumed bynuclear -powered submarines.particularly of the ballisticmissile -armed type, has placedthem in a role where an indi-vidual vessel can have interna-tional impact undreamt of at theend of the second world war. Itfollows that the ability to com-municate with these submarinesat virtually any moment duringthe day or night is completelyessential to their operation."

As the events surrounding the

sinking of the Belgrano in 1982showed, radio communicationwith submarines is still oftenonly intermittently successful.A.J. Meaden notes that com-munication depends on either abuoyant antenna tower towedbehind a deeply -submerged ves-sel. with problems of manoeuv-rability and recovery, or with amast -mounted antenna raisedfrom the bridge -fin when thevessel is at periscope depth.

Plessey has developed a newintegrated communicationsmast. MoD designation MU,which is free -flooded and ofglass -reinforced -plastics con-struction. Unlike earlier designs.it does not penetrate the hull andhas a faired profile that overcom-es lateral vibration due to shed-ding of Karmon vortices. It cancarry m.f., h.f.. v.h.f.. u.h.f. andsatellite navigation and satellitecommunication antennas. Two -band radar -absorbent materialtuned to anticipated "enemy"frequencies has resulted fromcollaboration with Plessey Mic-rowave Materials, Towcester.The u.h.f./D.-band antenna com-prises a conical log spiral designat the top of the mast with abroadband dipole for line -of -sight. low -angle radiation alongthe axis of the cone. An h.f. whipantenna extends above the mast\ \lien rk.viirctl

RSGB's 75thanniversary

By the time these notes appear.the 75th anniversary of thefounding by four pre -first worldwar experimental amateurs ofthe Wireless Society of London(it became the Radio Society ofGreat Britain in November 1922)will have been duly celebrated bya series of events starting with athree-day convention at NEC inBirmingham, opened by thesociety's patron, Prince Philip,Duke of Edinburgh. This year.even the society's annual generalmeeting, until now always heldin London. may depart the capit-al - perhaps a sign that theamateur radio movement, withits need to regain the interest ofthe younger generations, is find-ing it increasingly difficult todecide where its future lies.

Radio Communications is writ-ten by Pat Hawker.

934 ELECTRONICS& WIRELESS WORLD

RACKSCA8INEIS

& F ANS

SURPLUSBOUGHT&

SOLD

ELEC1RENt9 &

c'SDO 0LE-cf.OSS

MONITORSVDU'S

& PLOTTERS

1000'S OF

IT EMS

IN STOCK

COMPUTERpsrEmsBS

RELAYSmOIORStSTEPPERS

SDPID0WERiN PL/Es

VER &TORS

oveES PRINTERS %0E0akt. SI DISKECOfOP*AE

HIKE YeIV

DRRSES &

&DAfaouiPtEnr-TIS

THE 'ALADDINS' CAVE OF ELECTRONIC & COMPUTER EQUIPMENT

COLOUR MONITORS16' Decca. 80 series budget range colour monitors features indude PIL tube attractive teak style case uaranteed 80 columnresolution only seen on monitors costing times our once readyto connect to a host of computer or video outputs Manufacturersfully tested surplus sold in little or hardly used condition with 90 dayfull RIB guarantee 1000's Sold to date.DECCA 80 RGB TTL SYNC input for BBC type interface etcDECCA 80 COMP 75 It composite video input with integral audioamp & speaker ideal for use with video recorder or TEL ST orany other audio visual use Only £99.00 (E)

HIGH DEFINITION COLOURBRAND NEW CENTRONIC 14' monitors in anractNe style mouldedcase featuring hi res Mrtsubushi 0 42 dot pitch tube with 669 x 507pixels 28Mhz bandwidth Full 90 dly_guaranteeOrder as1004-N2 for TTL sync RUB for BBC etc £150.00 (E

1003-N1 for IBM PC etc fully CGA equiv £189.00 E1005-N2 RGB interlace for OL 85 columns £169.00 E11

20 " & 22" AV SpecialsSuperbly made, UK manufacture. PIL tobe. all solid state colourmonitors complete with composite video and sound inputs. attrac-tive teak style case. Ideal for a host of applications including Schools.Shops Disco's. Clubs etc Supplied in EXCELLENT little used con-dition with 90 day guarantee

20' Monitor £165.00 (F) 22 Monitor £185.00 (F)

MONOCHROMEMOTOROLA M1000-100 5' CRT black & white compact chassismonitor measuring only cm 11 6h .12w. 22d. ideal for CCTV or com-puter applications Accepts standard Composite video or individualH & V syncs Operates from 12v DC at apprx 0 8a Some units mayhave minor screen marks but still in very usable condition Fulltested with 30 day guarantee & full data Only £29.00 (C)Fully cased as above with attractive moulded. desk standing swiveland it case Dim cm 12h.14 5w.26d £39.00 (C)JVC type 751-7 5" ultra compact black & white chassis monitor for12v 0 7a DC operation Dim cm 11h.14w.18d Simple DIY circuit dataincluded to convert data and separate sync input to composite videoInput Ideal portable equipment etc. Supplied with full data

Brand New £65.00 (B)KGM 324 9' Green Screen. Little used fully cased. mains poweredhigh res monitors with standard composite video input Futy testedand in excellent condition £49.00 (E)20' Black & White monitors by AZTEK, COTRON & NATIONALAll solid state. fully cased monitors. ideal for at types of AV of CCTVapplications. Units have standard composite video inputs with in-tegral audio amp and speaker Sold in good. used condition- fullytested with 90 day guarantee Only £85.00 (1)

FLOPPY DRIVE SCOOPDrives from Only £39.95

A MASSIVE purchase of standard 525 disk drives enables us tooffe you prime product at all time super low prices All units unlessstated are removed from often BRAND NEW equipment. fullytested and shipped to you with a full 120 day guarantee. AJI unitsoffered operate from 5 and 12 volts DC. are of standard size andaccept the common standard 34 way interface connectorTANDON TM100-2A IBM compatible 40 track FH double sided

Only £39.95 (B)TANDON TM101-4 FH 80 track double sided Only £49.95 (B)JAPANESE Half Height double sided drives by Canon, Tec.Toshiba etc. Specify 40 or 80 track Only £75.00 (B)TEAC FD55-F 40-80 track double sided Hail Height

Brand New £115.00 (B)DISK DRIVE ACCESSORIES

34 Way interface cable and connector skigle E5.50, Dual £8.50 (A)5 25 DC power cable £1.75. Fully cased PSU for 2 x 5 25 Drives MASTER SYSTEMS type 212 mlcroproces-£19.50 (A) Chassis PSU for 2 x a drives £39.95 (B) sor controlled V22 lull duplex 1200 baud. This

811 DISK DRIVES fuNy BT approved modem employs all the latestfeatures for error free data comms at the stag-

SUGART 800.801 single sided refurbished £175.00 E) genng speed of 120 characters per secord,SUGART 851 double sided refurbished £2e0.00 E) saving you 75% of your BT phone bills andMITSUBISHI M2894-63 Double sided switchade Hard or Sort sec data connect time II Add these facts to ourtor BRAND NEW £275.00 (E) give away price and you have a superb buy "SPECIAL OFFER Dual 8 drives with 2mb capacity in smart case Ultra slim unit measures only 45 mm high withwith integral PSU ONLY £499.00 (F) many integral features such as Auto answer.

Full LED status indication. RS232 interfaceRemote error diagnostics. SYNC or ASYNCuse. SPEECH or DATA switching., integralmains PSU. 2 wire connection to HT line etcSupplied fuly tested. EXCELLENT slightly usedcondition with data and full 120 day guarantee

PRINTERSi.dda. un51,, Inxc or pi mien 1.11111411,4Th(

1,..ipenr: .111he r.crn di I MI South I ...don Shop

HAZELTINE ESPRINT Small desktop 100 cps print speed with bothRS232 and CENTRONICS interfaces Full pin addressable graphicsand 6 user selecable type fonts Up to 9 5 single sheet and tractorpaper

rTIFIONIR 150 series A reel workhorse for cBand

worCiltriYuou"s99uslv!ilitractor feed paper, either in the office. home or factory desk standing,150 cps 4 type fonts and choice of interfaces Suppled BRAND NEWOrder as

150 -SW up to 14 5 paper handling £225.00£185.00

E150-SN up to 9 5 per handling

150-GR up to 14 5 paper plus full graphics £245.00 EWhen ordering please specify RS232 or CENTRCNICS interface

Ultra Fast 240 cps NEWBURY DATANDR 8840 igh Speed Printers

Only £449A special purchase from a now defunct Goverment Dept enables usto offer you this amazing British Made, quality punter at clearanceprices, SAVING YOU OVER E1500 II the NDR8840 features highspeed 240 cps print speed with integral, fully adjustable paper trac-tor giving exceptional fast paper handling for multi Dart forms etc Theunit -features 10 selectable type fonts giving up to 226 printablecharacters on a single Nne. Many other features include internalelectronic vertical and horizontal tabs Self test. 9 needle head. Up to15 5 paper 15 milion character ribbon cartridge lie and standardRS232 serial Interface Sold in SUPERB tested condition with 90 duguarantee

model 512 40 column 3 5 wide roll?en.Z.Eh14.7.sPiermatrix (3 lines per second) printer mechanism for incorporation inpoint of sale terminals. ticket printers, data loggers etcUnit features bidirectional printhead and integral roll paper feed mech with tear barRequires DC volts and simple parallel external drive logic Completewith data RFE and tested Only £49.95 (C)EPSON model 542 Same spec as above model. itit designed to beused as a slip or flatbed printer Ideal as label. card or ticket printerSupplied full y cased in attractive, small. desk top metal housing Com-plete with data RFE and tested Only £55.0) (0)PHILJPS P2000 Heavy duty_25 COI bi directional daisy wheel printerFully DIABLO, OUME, WORDSTAR compatible Many features In-clude full width platen - up to 15 paper. host of avalLable daisy wheels.single sheet paper handling superb quality print Supplied completewith user manual & 93 day guarantee plus FREE dust cover & dais_ywheel BRAND NEW Only £225.00 (E)

Most of the items in this Advert. plus a whole range of otherelectronic components and goodies can be seen or purchasedat our

** South London Shop **Located at 215 Whitehorse Lane, London SE25. The shop ison the main 68 bus route and only a few miles from the mainA23 and South Circular roads Open Monday to Saturday from9 to 5.30 parking is unlimited and browsers are most wel-come. Shop callers also save the cost of carnage

MODEMSModerns to suit all applications and budgetsPlease contact our technical sales staff if yourequire more information or assistance

SPECIAL PURCHASEV22 1200 baud MODEMS

ONLY £149 Ii

COMPUTER SYSTEMSTATUNG PC2000. Big brother of the famous EINSTEIN theTPC2000 professional 3 piece system comprises Quality high resGREEN 12 monitor Sculptured 92 key keyboard and plinth unit con-tairerg the Z8OA CPU and all control electronics PLUS 2 integralTEAC 5.25 80 track double sided disk drives. Many other featuresinclude Dual 8. IBM format disk drive support. Serial and paralleloutputs full expansion port 64k ram and ready to run software Su

age Bied

compRAleteNDNEW

with M. WORDSTAR, BASIC and accounts pack -

Full 90 day guaranteeOriginal price OVER £1400 Only £299(E)EQUINOX (IMS) 5100 system capable of running either TURBO orstandard CPM Unit features heavy duty box containing a powerfulPSU. 12 slot S100 backplane. & dual 8 double sided disk drivesTwo individual Z80 cpu boards with 192k of RAM allow the use ofmulti user software with upto 4 RS232 serial interfaces Many otherfeatures indude battery backed real time dock. al IC's socketed etcUnits in good condition and tested prior despatch. no documentationat present hence price of only £245.00(F)S100 PCB's IMS A465 64K dynamic RAM £55.00 (BUMS A930 FDCcontroller £85.00 (B). IMS A862 CPU & Uo £66.00 (B)

SAE for lull list of other S100 boards and accessories

LIMITEDQUANTITY Only £149 (D)

CONCORD V22 1200 baud as new £330.00(E)CONCORD V22 1 200 2400 BIS £399.00 (E)RIXON Ex BT Modem 27 V22 1200 £225.00 (E)DATEL 4800 RACAL MPS 4803 EX BTmodem for 4800 baud sync use £295.00 (E)DATEL 2412 2780/3780 4 wee modem unitEX BT fuNy tested £199.00 (E)MODEM 20-1 75.1200 BAUD for use withPRESTEL etc EX BT fully tested £49.00 (E)TRANSDATA 307A 300 baud acousticwith RS232 1'0 Brand New £49. (ElRS232 DATA CABLES 16 ft long 25w D dug to25 way D socket Brand New Only f9.95 AI

As above but 2 metres long f 4 99 A)BT plug & cable for new type socket f2.95 A)

POWER SUPPLIESAll power supplies operate from 220-240 v AC Many other types from3v

dower10Kv in stock Contact sales office for more details

PLESSEY PL12'2 Fully enclosed 12v DC 2 amp PSU Regulated andprotected Dim cm 13 5 x 11 x 11 New £16.95 (B)AC -DC Linear PSU outputs of 5v 5 5a, 5v 0 6a 24v 5a Fully regu-lated anc short proof Dim cm 28 x 12 5 x 7 New £49.50(C)POWER ONE PHC 24v DC 2 amps Linear PSU fully regulated

New £19.95 (B)BOSHERT 13088 switch mode supply ideal disk drives or completesystem 5v 6a 12 2 5a-12 0 5a. -5v 0 5a Dim cm 5 6 x 21 x 108

New £29.95 (B)BOSHERT 13090 same as above spec but outputs of 5v 6a. 24v15a 12v 05a 12v05a New E39.95 (B)GREENOALE 19ABOE 60 Watt switch mode outputs 5v 6 a 12vla 12v ia 15v taD 11 x 20 x 55 RFE Tested E24.95 (B)CONVE9 AC130-3001 High grade VDE spec compact 130 wattswitch mode PSU Outputs give 5v 15a, -5v la 8.12v 6a Dim 6 5x 27 x 12 5 Current list price £190 Our price New £59.95.00 (C)FARNELL G6 40A Compact 5v 40 amp switch mode fully enclosed

New 140.00(C)CFARNELL G24 5S Compact 24v 5 amp switch mode fully enclosed

New £95.00 (C)

Special OfferEXPERIMENTORS PSU

ONLY

£16.95 (c)Made to the highest spec for FIT this unit gives several fullyprotected DC outputs most suited to the Electronics Hobbyist. 5v 2a. 8,12v la.- 24v la and 5v fully floating at 50ma

Ideal for school labs etc Quantity discount avalableFully tested with data RFE Removed From Equipment

The AMAZING TELEBOXConvert

COLOURnitorintp a

QUALITY TELEVISIONBrand new high quality fully cased. 7 channel UHF PAL TV tuner sys-tem Unit simply connects to your TV aerial socket and video morelorturning same into a fabulous colour TV Dont worry if your monitordoesn t have sound. the TELEBOX even has an integral audio amp fordriving a speaker plus an auxillaryoutput for Headphones or Hi Fi sys-tem etc Many other features LED Status indicator. Smart modeledcase. Mains powered Boit to BS safety specs Many other uses for TVsound or video etc Supplied BRAND NEW with fill 1 year guaranteeCarriage code (B)

TV SOUND& VIDE°,TUNER

ONLY

£29.95TELEBOX ST for monitors with composite video input £29.95TELEBOX STL as ST- but fitted with integral speaker £34.95TELEBOX RGB for use with analogue RGB monitors £59.95

(RI ffr i11,1,00 lo.11151.I 11151moniti. I) \ I \ rcrikr,1 \I I 1011,1 .\ 1 1

RECHARGEABLEBATTERIES

Maintenance free sealed longite LEAD ACIDA300 12v 3 Ah f 13.95A300 6v 3 Ah

RFE £5.99£9.95 AAA300 6-0-6 v 1 8 AhNICKEL CADMIUM

Quality 12 v 4 Ah cellpecA Originally madefor the TECHNICOLOUR video companythis unit contains 10 high quality GE raced, Dtype cells, configured in a smart robustmoulded case with DC output connector Dimcm 19 5 x 4 5 x 12 5 Ideal portable equipmentetc BRAND NEW £24.95 (B)12v 17 Ah Ultra rugged all weather. virtuallyindestructable refillable IIICAD stack byALCAD. Unit features 10 x individual typeXL 1.5 cells in wooden crate Supplied to theMOD and made to deliver exceptionally highoutput currents & withstand long periods ofstorage in discharged state Dim cm 61 x 14 x22 Cost over E250 Supplied unused & testedcomplete with instructions £95.00 (E1EX EQUIPMENT NICAD cells by GERemoved from equOmerft and believed ingood but used condition. 'F' size 7th 6 for£8 (B) Also D' size 4Ah 4 for £5 (B)

Dil3sFIAVieNsEfft451a..End of line purchase enables this brand newunit to be offered at an all tfrne super low priceThe NEC D2246 6' 80 Mb disk drive featuresfull CPU control and industry standard SMDinterface, Ultra high speed data transfer andaccess times leave the good old ST506 inter-face standing Supplied BRAND NEW withfull manual Only £399.00 (ElDual drive. plug in 135 Mb sub system for IBMAT unit In case with PSU etc £1499.00 (F)Interface cards for upto 4 dives on IBM ATetc available Braid new at £395.00

COOLING FANSIscep hot 1,1..111111 .tna RI I I \ 141 I star,

r of 1112.5511 *s. I \AC FANSn Specfv 240 or 110 v3 Fandim 80x 80x39 £8.503.5 ETRI slimline 92 x 92 x 25 £9.95 B4' Fan Dim 120x 120 x38 £9.95 BAs above - TESTED RFE Only £4.9510' round x Rotron 10v £10.95 BDC FANSPapst Miniature DC tans 62.62x25 mmOrder 812 6-12v or 814 24v £15.95 A4" 12v DC 12w120x 120x38 £12.50 B4" 24v DC flw 120x 120 x25 £14.50 B°UHLER 12v DC 62 mm £12.95 A

'SPECIAL INTEREST'Please call for availability or further into

RACAL-REDAC real time. colour draftingPCB layout system £3950DEC VAX11750 inc 2 Mb Ram DZ and fulldoc etc Brand New £8500HP7580A 8 pen digital Al drum plotter withIEEE interface As New £4750CHEETAH Telex machine £9951.5 kw 115v 60 Hz i)ower source £950500 watt INVERTER 24v DC to 240v AC sinewave 50 Hz output £275SOLDER SYSTEMS tin lead roller tinningmachine for PCB manufacture 310CALLAN DATA SYSTEMS multi user INTELbased UNIX system complete with softwareand 40 Mb winchester disk drive £2750WAYNE KERR RA200 Audio. real time fie-guency response analyzer £3000TEKTRONIX 1411.'R PAL TV test signalstandard 900TEKTRONIX R140 NTSC TV test

E6signal

standard £875HP 3271A Correlate system £350PLESSEY portable Microwave speech / datalink 12v DC 70 mle rangeThe pee £275.0017' Rack cabinets 100's in stock trom £15.00

DISPLASTS-ELECTRONICS-

All prices for UK Mainland UK Customers must ADO 15% VAT to total order value. Minimum order, cash CS. Credit Card C10. Official account onionsfromGovernment Depls. Universities Schools 8 Local Authorities welcome - minimum account order value C25. Carriage charges (A) Cl 50 (B) C3 50. (C)C6 50. (D) C8 50, (E)C10 00 (F) C15, (0)Call All goods are supplied subject to our standard conditions if sale All guarantees given 011 a return to base basisWe reserve the right to change prices 8. specifications without prior notice Bulk trade & export enguffir s most welcome

MAIL ORDER & OFFICESOpen Mon -Fri 930-5 30

32 Biggin Way,Upper Norwood.

London SE19 3XF

LONDON SHOP10004 of Bargains for callers

Open Mon -Sat 9-5 30215 Whitehorse Lane.th Norwood. London SE2

DISTEL g The ORIGINALFREE of charge dial up data base

1000 s of Items info oN LINE NOW!!300 baud 01 679 1888 1200 75 01 679

6183 1200 FOX 01 679 8769

ALL ENQUIRIES

01 679 4414FAX 01 679 1927TELEX 894502


RADIO BROADCASTA.m. stereo:what went

wrong?Robert Streeter. who developedthe Magnavox medium -wavestereo system, has commentedin IEEE Spectrum (June 1988)on the reports that a.m. stereo inthe USA is in the doldrums, withlittle interest being shown by thevast majority of listeners ("Radiobroadcast". July 1988). The Mag-navox AM -OM system was initial-ly selected by the FCC in 1980 asthe American standard: hut. fol-lowing strong opposition fromthe other contenders, the rulingwas withdrawn and in 1982 it wasannounced that the choicewould he left to the market-place.

Streeter writes: "If all in-terested parties had settled downto producing an a.m. stereo sys-tem in 1980. then a.m. broad-casting would have had a muchbetter chance to he equal to f.m.stereo in the ears of the listeningpublic before its fall to nearoblivion."

He contends that none of thea.m. stereo systems authorizedby the FCC in 1982 is flawless butthat the same could he said ofNTSC colour television or f.m.pilot -tone stereo - "Yet theseother systems have served theconsumer well and have made agreat deal of money for manufac-turers ... There are ways to hidethe flaws in each of the a.m.stereo systems sufficiently toprevent the consumer frombeing annoyed."

The National AM Stereopho-nic Radio Committee was set upin 1975 charged with assessingpossible systems. Ever since thenthe technology has been sur-rounded by controversy. withsome proponents strongly criti-cising rival systems, often withlegal action threatened or initi-ated. Of the four systems autho-rized by the FCC in 1982 -Magnavox AM -OM. Motorola C-QUAM. Harris V-CPM and Kahn/Hazeltine i.s.b. - only C-QUAMand i.s.b. remain in contentionin an atmosphere of fading in-terest.

As the time approaches for theextension of the North Americanbroadcast band from 1605 to1700kHz, H. Dickson Norman,chairman of Global Radio Cor-poration (NDXE) of Opelika. Ala-

hama. who for some years hasbeen endeavouring to set up aninternational h.f. service witha.m.-stereo. has now proposed tothe FCC that clear -channel fre-quencies in the extended m.f.hand he allotted to his companyfor the establishment of a "super-power. a.m.-stereo. nationaldirect broadcast service" cover-ing continental USA. Alaska andHawaii. Traditionally, the FCChas always limited transmitterownership so that the maindomestic radio networks arebuilt around chains of "affiliate"stations. Power has been limited.even for the so-called "clearchannels", to a maximum of50kW. The ambitious Globalplan calls for chains of synchro-nized transmitters at powers upto one megawatt. Programmeswould he to a news and informa-tion format, funded by advertis-ing. It remains to be seenwhether FCC will respondfavourably to Global's proposalsfora first "national" channel.

EBU CD testrecord

For many years both SMPTE andEBU have produced sets of col-our slides carefully selected toprovide very critical tests of tele-vision systems. Such slides play-ed an important role in the com-parison of colour encoding tech-niques and later in devising thedigital standards.

The EBU's ad hoc Group V3/SQAM has now produced a Com-pact Disc record containing 60minutes of material for the sub-jective assessment of both ana-logue and digital audio systems.The 70 segments intended foruse in accordance with the proc-edures detailed in CCIR Recom-mendation 562 comprise: align-ment signals, artificial signals:36 segments of single instru-ments: vocal segments with andwithout orchestral accompani-ment: solo instruments withorchestra: orchestral and popmusic segments.

The EBU committee believesthat subjective listening tests areoften essential when assessinghigh -quality audio systems andthat the choice of test sequencesprofoundly affects the result. Itcontends that only if the sametest sequences are used can diffe-rent listening tests he compared

meaningfully. The EBU recordcontains sequences selected toreveal a variety of defects. Therecording was made directly indigital form using a SonyPCM1630 coder and V05080 U-miak.. recorder.

CD life span?It has been the claim of therecording industry that CD re-cords not only provide near -

immaculate sound quality hut.having no physical contact withthe laser. will last indefinitelywithout degradation. However.there have been rumours that afew optical discs have appearedto degrade rapidly, due to whathas been called "laser rot" - amisleading term, since the prob-lem seems to arise from exposureof the aluminium layer to the airas a result of defective applica-tion of the protective lacquer.

A more serious allegation wassprung on the public in TheGuardian (June 29) suggestingthat a significant proportion ofexisting discs can be expected tofail in eight to ten years time as aresult of contamination of thelacquer by the ink used to printthe "labels", leading to pin holesand consequent oxidation of thealuminium surface. Sony andPhilips. who receive royalties onall CD records, insist that noproblems are likely to arise withrecords made correctly to theagreed specification. While thereis clearly the noise of commer-cial axes being ground within theindustry, with Nimbus claimingthat accelerated life -tests haveindicated that there could be aproblem, those who have builtup a library of CD records mustbe counting the years with someanxiety.

Digital audio tape (DAT) maynot he an answer since RafailSandu. chief engineer of theOIRT Technical Centre in Pra-gue. recently wrote of digitalaudio: "Certain aspects requirefurther study... It is still notknown what causes partial signaldropouts in magnetic tapes usedafter a storage time of severalmonths".

The French consultant en-gineer Marc Chauvierre. who hasworked in television since 1930.has discussed the futuristic pos-sibility of a sound and imagerecording system entirely free of

electromagnetic elements (IEEETrans. on Consumer Electro-nics. May 1988. 298-301). Hedismisses the solution of electro-nic memory because of the diffi-culties of congestion of thememory. its connections to thereading device and the prohibi-tive costs of manufacture. Hisproposed solution adopts theprinciple of analogue modula-tion, after conversion to digitalform, photographically recordedas a matrix on a square or rec-tangular film or transparency.

He recognizes that the prac-tical implementation of such asystem for more than a fewseconds of audio signals dependson reducing the size of the pic-ture element (pixel) from thefraction of a millimetre possibletoday to only a few microns. Hecomments: "This would seemutopian today. hut tomorrow...?"

Solar radioset -back

Lack of help from specializedplastics firms, which are oftenassociated with battery manufac-turers, is blamed by UNESCO forthe problems experienced in de-veloping a low-cost radio receiv-er powered from the sun andsuitable for manufacture andsale in Third World countries.The UK withdrew from UNESCOin 1985.

UNESCO has spent about$25 000 in developing the pro-totype of a simple receiver in-tended to he retailed at less than$20. AEG of West Germany. witha factory in China, has developeda suitable polycrystalline siliconsolar array. Porsche agreed todesign a special plastics cabinetfor $5000 but UNESCO has beenunable to find a specialist manu-facturer willing to produce aproduction mould for thecabinet within its budget. Theorganization claims that Sie-mens was "not interested".Alcatel. the French multination-al company which now owns ITT.Telefunken and Schaub Lorenz.did not respond. Thomsonwanted $1 million.

In recent years UNESCO haspioneered solar -powered 10 wattv.h.f./f.m. transmitters forKenya. Burundi. Sri Lanka. Togoand Ghana.

Radio Broadcast is written hr PatHawker.


Broadbandinstrumentation

amplifierThis design avoids the compromise between c.m.r.r. and

bandwidth that characterizes conventional instrumentationamplifiers made from op -amps.

Instrumentation amplifiers made fromop -amps generally consist of an input -buffer pair followed by a differential

amplifier stage. Fig.l.It is rather difficult to realize a differential

amplifier with a high rejection factor forcommon -mode signals combined with awide frequency range, mainly due to theop -amp's non -symmetrical structure.Therefore, it is desirable to make use of aninput -buffer pair which already has a highcommon -mode rejection ratio. However, acommon -mode component always remainsequal to that at the input I (V; +V.0/2. Fig.l I.So a high c.m.r.r. calls for a high differentialgain of the buffer (a11. but with thedrawback of a reduced bandwidth.

This common -mode component is greatlyreduced in the new circuit of Fig.2. Op -ampIC:; provides a large amount of feedback forcommon -mode signals in the buffer stage. Inthis way a high rejection factor can heobtained with low gain. maintaining a highbandwidth. Output of each of the two buffersconsists primarily of the difference of theinputs; as a result the following differentialamplifier is less critical.

Amplifier IC3. whose non -inverting inputcan be considered as a virtual earth, forcesthe sum of the buffer -outputs to zero. It canbe shown that with unequal R, resistors thecommon -mode component at the output ofthe buffers remains low. even in the extremecase that one R3 is removed. However. thenon -symmetrical structure of the circuitthen leads to a reduced c.m.r.r. at higherfrequencies.

Despite having more components. thecircuit needs no extra adjustments for a goodbalance. and the two additional resistors. R:;.do not need very careful matching. Additionof IC; does not degrade noise performancenor does it increase offset errors. Care mustbe taken to guarantee the stability of thecircuit.

Circuits have been built using NE5534op -amps which have high output drive capa-bility. low noise and high open -loop band-width. The latter feature is desirable for agood high -frequency common -mode rejec-tion figure. In the practical circuit. IC.;

NANNO HERDER

o(Vi -V21Vi

Vo -V21

Fig.1. Conventional instrumentation amplifiers made from op -amps usually consist of aninput buffer pair followed by a differential stage.

0 .1-20 (V1. V2

p

V2

needed a stability capacitor (with a= 1). Withcareful circuit -hoard lay -out and power -supply decoupling, a c.m.r.r. of 100dB at100Hz and 80dB at 50kHz and a bandwidthof 1MHz is obtainable.

I recommend the use of low -valued resis-tors ( 1 kit) to reduce balance errors due tostray capacitances and to maintain a lownoise level. The new buffer circuit can alsobe used as a precision balanced driver. e.g.for audio purposes. With one groundedinput, two 180° out of phase signals can beobtained from the outputs of the buffers.

820

820

820

820

V0rK2IVI-V2)

Cc= 22p

All op amps 5534

FIg.2. Adding an op -amp to the convention -a, instrumentation amplifier configurationprovides a high common -mode rejectionratio over a wide frequency range. Outputof IC3 is proportional to the common -modecomponent at the input and can be usedfor an overrange indication.

ELECTRONICS & WIRELESS WORLD 9:17

01-208 1177 TECHNOMATIC LTD 01-208 1177BBC Computer & Econet Referral Centre

AM815 88C MASTER £344 (alAMC06 Turbo 165C - 021 F cpansron Module

ADC08 512 Pron.,ADE 14 Rom Can, .1,,ADJ22 Rel Manual Pnt"

(195(b)C13(b)C14(c)

ADJ24 Advanced Ret ManualADF10 Econel ModuleADJ23 Rel Manual Pail itBBC Master Dust Cove,

BBC MASTER COMPACTA free packet of ten 3 5 OS discs with var.', GnorparSYSTEM I t 280 Single 134011 Dore am bundled software (385 I1

YS TEM 2 System I wan a 12 H. Res N(irki Monaca (469 (a)SYStEM 3 System i wen a II Med Res RGB Monitor (5990)Second Once Ka £99 (c) .ienson Cable tor em 5 25 drive 012 50 (0)

view 3 0 User Guide C10 (d)BBC Dust Coca VI 50 (0)aof S ROM Ow B roln t 770 01)S A /1 Fnusr (26 (0)ACORN 180 2nd Processors C329110MUL f IF ORM 180 2nd Processor (219 (b)

ORCH 280 2nd Processor ZE P 100TZDP 240 /EP WO with I echnornalc PLMOOP dual drive wan burn r

AMB12 BBC MASTER Econet C315 (a)(b)

(111.50(()(41 (c)(14(e)75 (d)

Viewsheel User Guide C101411770 DE S Upgrade lor Model 13 (43 50 (01

205 ROM t151111)ACORN 6502 2.1t1 Processor (173101

ACORN IEEE interface C269 (a)0229 lai

n monitor stand 0431 IC

META Version III - The only package available in the micro market that willassemble 27 different processors at the price offered Supplied on two 16Kroms and two discs and fully compatible with all BBC models Please phonefor comprehensive leaflet £145 (b).

We stock the full range of ACORN hardware and firmware and a very wide range ni otherperipherals for he BBC For let alert Sp, at.ns and pricing please send tor Our leaflet

DISC DRIVES5.2S Single Drives 40,50 switchable.TS400 400K 640KPS400 400K/640K with integral mains power supply5.25" Dual Drives 40/80 switchable:TD800 800K/1280KP0800 800K1280K wito integral mains power supplyPD800P 800K 1280K with integral mains power supply and monitor stand3.5" 80T DS Drives:TS351 Single 400K 640KPS351 Single 400K640K with integral mains power supplyTD352 Dual 800K 1280KP0352 Dual1300K1280K with integral mains power supplyPD853 Combo Dual 5 2513 5 drive with p s u

0114 (b)(129 (b)

(199 (a)0229 (a)C249 (a)

(99(b)f1 19 (b)(170(b)(1 87 (b)0229 (a)

3M FLOPPY DISCSIndustry Standard floppy discs with a lifetime guarantee Discs in packs of 10

51/4 Discs 31/2 Discs40T SS DD E10.00(d) 40T DS DD (12.00(d) 80 T SS DD C20.00 (a)807 SS DD (14 50 (d) aOT DS DD E15.50 (d) 80 T OS DD (25.00)d(

PRINTERS & PLOTTERSEPSONEPSON LX136 Elp1a1Optional Tractor Feed 1080 86 EN ICISheet Feeder LO80 86 US ICIF X800 (319(s)

1000 C449 (a)[X800 C409 (a)L0800 (80 col) (439 (a)L01000 (589(a)

TAXANKP815 (160 cps) C249 (a)KP915 (180 cps) (369(a)

JUKI6100 (Daisy Wheel (259 (a)

NATIONAL PANASONICK X P1080 (80 con (149 (a)

STAR NL10 (Parallel interface)STAR 9110 (Serie' Interface)STAR Power Type

BROTHER HR20

COLOUR PRINTERS

Dotprint Plus NLO Rom forEpson versions for F X RX. MXand GLP only)

PLOTTERSHitachi 672Graphics Workstation

(A3 Plotter)Piotmate A4SM

Ma la/Oral.)an (a)

(329 (a)

(28 (d)

(459 (a)

(599 (a)E450 (a)

Single Disc Cable C5 (d)0 Disc Library Case El 80 (d)

50 5' Drsc l oCir 4444- Hit. C9 00 Ic)

FLOPPICLENE DRIVEHEAD CLEANING KITFLOPPICLENE Disc Head Cleaning Kit with 28 disposable cleaning discsensures continued optimum performance of the drives 5' .t £12.50 (d)

3. (14.00(d)DRIVE ACCESSORIES

Dual Disc Cable (8.50 (d)30 5' Disc Storage Box £6 (c)r,K, 5 Disc Lockable Box (13 (c)

PRINTER ACCESSORIESWe hold a wide range of printer attachments (sheet feeders. tractor feeds etc)in stock Serial parallel. IEEE and other Interfaces also available Ribbonsavailable for all above plotters Pens with a variety of tips and colours alsoavailable Please phone for details and prices.Plain Fanfold Paper with extra fine perforation (Clean Edge):2000 sheets 9 5' x 11* (13(b) 2000 sheets 14 5- x I 1' (11).50(b)Labels per 1000s Single Row 3' x 1 7/16' (5.25(d) Triple Row 2-7/16' x 1 7/16' (5.00(d)

RGB 14"1431 Std Res1451 Med Res1441 Hi Res

MONITORS

(179(a)(225(a)(365 (a)

MICROVITEC 10" RGEUPALAudro1431AP Std Res C199 (a)1451AP Std Res C259 (a)All above monitors available in plastic ormetal case

TAXAN SUPERVISION II12 Hi Res with amber green optionsIBM compatible C279 (a)Taxan Supervision III (319 (a)

MITSUBISHIXC1404 14 Med Res RGB. IBM 8 BBCcompatible 0219 (a)

MONOCHROMETAXAN 12 HI-RESK0 1201G green screenKX1203A amber screen

PHILIPS 12 HIRESBM7502 green screenBM7522 amber screen8501 RGB Std Res

C90 (a)(95 (a)

(75(a)(79 (a)

(139(a)

ACCESSORIESMicrovitec Swivel Base (20 (c)Taxan Mono Swivel Base WithClOCo C22 (c)Philips Swivel Base (14(c)BBC RGB Cable C5 (d)Microvitec (3.50 (d)Taxan (5 (d) Monochrome 03.50 (d)Touchtec 501 (239(b)

MODEMSAll modems carry a full BT approval

MIRACLE TECHNOLOGY WS Range

WS4000 V21/23 (Hayes CompatibleIntelligent Auto Dial/Auto Answer( (149 (b)

WS3000 V21/23 Professional As WS4000and with BELL standards and battery back upfor memory £245 (b)

WS3000 V22 Professional As WS300 V21 /23but with 1200 baud full duplex £450 (a)

W53000 V22 bis Professional As V22 and2400 baud full duplex (595 (a)

WS3022 V22 Professional As WS3000 butwith only 1200/1200 £350 (a)

WS3024 V22 Professional As WS3000 butwith only 2400/2400 (450 (b)

WS2000 V21/V23 Manual Modem (95 (b)

DATA Cable for WS senes/PC or XT (10 (d)

DATATALK Comms PackageIf purchased with any of the above

modems T70 (c)

PACE Nightingale Modem V21/V23Manual (75 (b)

(Offer limited to current stoCkS)

RT256 3 PORT SWITCNOVERSERIAL INTERFACE

input, output a I input 3 notarynudl channel seleclion input

impul baud rates independentlyselectable 7 bilAltal odd even noneparity Hardware 01 softwarehandshake 2560 butler trainspowered C375 (1s)

PB BUFFERinternal butler to most Epsonprinters Easy to instaiisuppliedvain 128K (99(c)

UVERASERSUV 1 T Erase, with bwllm timer and mans indicatorBuilt-in safety interlock 10 avoid accidental exposureto the harmful UV raysIt can handle up to 5eproms at a sure with an averageerasing time of about 20 mins C59 + [2 pip.UV1 as above but without the timer (47 + (2 p&p.For Industrial Users we otter uv140 6 UV 141 era-sers with handling capacity of 14 eproms UViitt hasa built in timer Both otter lull bunt in safety eaturesUV140 C69. UV141 £85. p&p C2.50.

EXT SERIAL/PARALLELCONVERTERS

Mains powered Cur,,elersSerial to ParallelParallel to SeriaiBidirectional Converter

(48 (c)(48 (c)

. (105)b(

Serial Test CableSerial Cable switchabie at both endsallowing pin options to be re-routed oflinked at either end -- making it possibleto produce almost any cableconfiguration on siteAvailable as kA,M or M r (24.75 (d)

Serial Mini Patch BoxAiiows a easy method toretonlig re pin lunCtiOnSwithout rewiring the Cableassay Jumpers can be usedand reused (22 (d)

Serial Mini TestMonitors RS232C and CCITTV24 TransmisSiOnSindicating Status with dualColour LEDs on 7 mostSignificant Ines Connects inline (22.50 (d)

CONNECTOR SYSTEMS

I.D. CONNECTORS(Speedbloca Type'

NO Or Header Recep Edgevt,;4 Plug 'acre Conn

10p 859 120pI0 14Sp 125p 115,26 178p 1509 240p34 200p 1110p 320140 220p 1110p 340p50 235p 200p 310p

SOFTY IIThis low cosi inteliigent eprom programmer can program 2'16 25162532 2732 and wah an adaptor 2564 and 2764 Displays 512 bytepage on TV has a serial and parabet PO routines Can be used as an emulator cassette intenaceSofty Il C195.00(01kOaolo to 27642564 (25.00

SPECIAL OFFER2764-25 £3.00 (d);27128-25 £5.00 (d);6264112-15 £4.00 (d);

D CONNECTORSNo of Ways

9 15 25 37MALE.Ang PinsSolderiDCFEMALESI PinAng PinsSolderIDCSt HoodScrewLOCK

12060

175

10016090

19590

130

18085

275

14021013032595

150

230 350125 170325

210275t95375100t75

380440290

120

EDGE

CONNECTORS2 6 way ,,,,,,,, suer.2 10 way2 12 war tr, 20,2 IS war

23 way i258ii2. 25 war,2.28 way iSPeCit2 36 .2,

43 wax2.22 wax2. 43 way

77 war,2 50 waciS 100r on.,

3005SOp

350p140p

i ?Sp 220p225p 220p200p25092609.90p3959400p soup6004,

AMPHENOLCONNECTORS

II, way plug Centronicssolder 500p i IDC i 475p36 way ski Centronicst solder 550p tiDCl 500p24 way plug IEEE , solder I475p (IDC 475p24 way ski IEEE I SOlder500p IDC 500pPCB Mtg Skt Ang Pin:))1 Aar, lOOp 36 Wa,, 750p

RIBBON CABLE,9re, we're

IC way 400 3,14ar 180016 -wan 100 40 oar 110920 Way Op 50 -war, 200926 )ear 1209 64 -ear 2809

TEXTOOL ZIFIOCXE1S 24 pin (7 se28 pin CI 10 40 0,412 10

EURO CONNECTORSDIN 416122 x 32 way St Pin2 K 32 way Ang Pm3x 32 way St Pin3 x 32 way Ang PinIX Skt A + BIOC Skt A + C

Piug Ski230p 275p275p 320p

375p 40260p 300p0p

400p400p

Fox 2 x 32 way please spool),sawing (A + B. A 4 C)

GENDER CHANGERS25 way 0 type

Male to MaleMaw to Femme

(10C10(10

DIL HEADERSS,zocier IOC

4 pin 409 100p16 pin 5Op 110p18 pin 60p20 pin 75p24 pin 1009 Isop28 Pin 160p 200p40 pin 200p 225p

RS 232 JUMPERSits wAn Di

24 6ingie end Mae24 Sangre end Female24 Female Female24 Male Male24 Male Female

[500ES 25

(1000950re 50

MISC CORNS21 pin Scar! Connector 200p8 pin Video Connector 200p

DIL SWITCHES4 way 9012 6 -way 105P8 -way 120p 10 way 150p

ATTENTIONAM poems Vi thle double pepsodvenIsernent ere misled to

Wove teldexd nodes.ALL PRICES EXCLUDE VAT

Please odd =Mope 50punless Indicated as Mow(a) CS (b) C2.50 (c) 01.50 (d)

01.00

TECHNOLINE VIEWDATA SYSTEM. TEL: 01-450 9764Using 'Prestel type protocols. For informationand orders- 24 hour service. 7 days a week


/ I SI III S

7007401

7401

74031404

1405140674077401

14011

n101.1,/4127413

74,674,

142.7423

/616742,1421

7430

143161337437

743.70110074010114746414

74617445706661474

70447057446

709741134

1491

74924'4134

2.32

A

3 A

5453

Si451

'0

,3

7$

74176741717417174110706174771MI -741107419314191741921093141074195141.9

14197741907410142217423174290742074207427474170

I%OM0 30

a10074074040040074OM074

30

0 10

05070

0730

a 300 Go

OMOn

0740 11

0

0 10

0740 300 10

04000040ON0 70

co

10

0 701%

00

00

M/5

03.OM0 SS

ON46

0400 SO

0100450 OS

OD

10

10

70

90

10

50

75

95

10

70

55

%S

SS

PS

70

SO

10

0

51271 ON74213 IN7474/ 3 II74790 IN14293 ON/421. I N743S, 290613654 I74300 IN743674 ON74379 1 111

74310 1

/4313 124147743 1 60

741 5 SERIES74.50074501745027415037µ5-14S05145-74509741 510

745111451374/51474515741570/052174152274152474152474527745707453074/53/745331453/145747415407415427454374/571.5074051

7415547435%14573*74074474515741 5110

741511741034741515741507459074159274593?ANN7415%745107

7451-74511114 5113143511471511274151737,45,757451271.513274513374(5123705131745131745145/451477415161/41.515174,51527415153746154745155745156'451571051501451404741516141451624741 516347415164

14516414510647451074.516914.51707451716/4/517414 51757451117451637415l%6151,174/5192741519074511156145/1174511177415221

14 57074 Sill745242745243741 520745745741524774510"'Si"lots»,.525376.52S674.524574

76,53511474(5151744520074.5266

O 24

424124024024IMIll124O 24

OM014

noil

0 24

0240500202.0240140 24

24

02424

050ISOON,N02.0 24

0240)00730460740420730730350 0on

073OM

ON00040

046045

65

0 70OwON084O a056546ONINOn1 11

110N2%

1 4061

0 ON00ON073

75

PG

0 7S1 10190

1 n0%l..1%

n073200190

75

073 VI073In

10

O%110

1-010

OnON073 1101 10190190

175InO N670O 1In075060

71527374. 52 n1452107402137415210741571;1415793741 52%

74/5?9,74151907415799745321741532727415323'415324ak S346'65%2'415351

5316'415/63

5X07453657415366741536174.50874537,74/5374745375741531'

74153 II7453n745311'74/5307453%745393745304745301743045,41.50657415467

7415610140550.5541141500014561074156,27415624/4562474151326

74056217456407µ5610I7415641745642740662 1745643745613 174066474156457415615 '

74154130

705469745670741561074115,03

41564474156.7'4154%7415703

osOn1900900

1. 02ON0

61 00

00

221732901001 /0290

ZO

1732 10

10

00

0 so0 So

0 52

0500 700 70073

30

0951 X450on0.9

20

19040

NZe

InI90

N16$7%

SOBana

216225126240190'502501%25019010000

090090

1 70

2901

SO

SO

16 00

105 SE R 1E S

Specialoffer

toPE

readersonly -

10% offonall

TTLSand

CMOS

74ALS SE RIES

,443504am 501

'441.510744(52010153274435/4'445130,44i.513974.41510`Mt 5245r441557144557466, SSW

Oa046OM050040OM0460 70150104%

290490

GO

74500

14542745-74505746047451074511

74520745:2745307453174537741.31

74540/455'/456.751.7454574506/451127451117451/474512474513274513374S13161513974514070151701537015774579074533745161745171)4517S745110701.970166745195745191745200745701"Son/45240/4524,/4514474S251

/45257"Si,.145290745261

245231452011452.174520"SRA74537345374

ON014050

O0050n

0 63

0 50050000

ON0500%0 45

o nS so

00

ON70

I 10

DO

00

ON190110IN190IN290100590

S01%120

sooa

300ISO4N1245.104040402.01.142...10IN

70

2252N

490229

4N4%211

4000 St RIO S400040014002400640D7

400040094010

401,601240134014

40154016AP 701140,940704021

6022

4023

440240/56/26

2/63211

.029030403,4032403)403.40364073

474633,40n604006147424043Din40454146

40474.40494050

405,4052405340544055405640%40634000676000007007100774073.075

6214

0 240250 70

anON0 45

0007.09SOM0100 PO

0)80 ffON010ONON0 70036OM

24

010040ON073n'n190

ZS

SO

20071

SO

4076407/40714040

404260.5MOO

400940934034409540%03740916.1/9

450345-450$45064507

4S08

as,452,4572

4526457/451945794531

4532

4534

45364544539454'4543

451145534555

45564557

4004566456045694572

4034504

4054724mu1012.16

611914001411514500314599

22100221012110240014

400054009740396601-40101401024010140104

4017540106

40107401M401094011040114401074016340173

401/440,754019240193WIN00700245mono40373

40374100%10[9710C -

LINEAR ICs40001147473100040103

1 SOSO

y 1 13501 09103.912

*2015*4302043020004643059*301000.0(43005ANNAMIE40904 0

4130E413014140EC631401C63146C431(10CARVECA.3,622Cella/C4.3240(C43190,Di002DAC 1406

0.001013INC016053.309001366(17106/0.741,(Loon(M7555ICM7556

1.C7120

1:1130IC1131IC71311.7341

1045111353113551135661

173570711.(MIXER1743610IM)07111300074

074370LM111

.M3101113191.54324

24413421.335.2EM330174139174340

(MAN1037701300 80.310.104313131104101306.1

1.1131/1139110.19250.3131.5.3941,.14D9

11N7310

00

00

503490

5%00

50

io

010InIN

0 7119400ISOInonINO aIN22516620O N27910390

00

390

290

100l%1005150%410

10ON

3001103.

50

IM0%0%010

no1%410

50

074041073273460190lo

166T IS

1%

00SO

DO

Sa

In10

i00n

90

1 10

0%4 00

073

174710 0440171/ 90

01123 0 %17412505 10001733 0 66

17474, 0 u1.0774? 0 70

17474$ 0101.1410/1 4001141014 1 50

011101 3 %.1130 27.0011171 1 DO

1611077 00

1M7.16 690LAMM 450Loom, ]00.3302 010013900 ON1143929 441747111 1 00

013914 ISO.3915 344

1747116 340(MONO I90M515131 2

11515161. $0

M63712 210101310 1 0413 onMC/40 045MC1495 100MC 1496 0 70

MC3000 000MC340, 0 70

MC3404 01674010(51 41031050140 1NM1902 SN741.922 4%741162214 1 mE5711 7107405.11 in.544 I n740555 0726150 ONNUM 4%NE565 124741514 154NE547 7251E570 4 %%SP IN74E592 On11553211 190NT 55330 IN745530 1211553440 INOP 07E0 150011024 5 %AC4136 0 SS

PC415' 1 %PC4195 50

1,1C450 0 SS

550240 9005096361 BOO

51690 0%504760135 5%55760330 510557411500 21555764119 4%55060% 4%$576660 12950070612 710505,5 70,7170 120071,0 14007204 19007205 ON107222 1,. 'Pi 50

T6011114103,I6110'11420'661200'6020(By.(CVOs1(6170(CanD41010"D.610/2'007024104,1705/13420021042003'0423-'04206'0420)0'005631042653

'043111,D01:030/01001,1011CI,?LayS06.'LOP.1017S074',OF71-2'LODT00.4

SO%,100C04759042240IC500,41.7157001401047002*11152003411152004416N/064LP 52002u 02033011424.4UPC 575

uPC592.100(11560UPC1115601210072206002207012711012216001749

214404

104,4/16419017402X1542*/'147501;542640

420841119f449E r

14.191,41.411

:,1,

nOwOw

01,

073200n

59067n23

Ya

10

00

142

244320290

14

70354

011IN190

a

0 NON0190731 10

ON050073190290129129,n690

nn

073on290110Nn

715200390590

490490

7,5

573173

20N

0%InN30n

36$10040]n

00290

00

7 SO

00

290690

21 009 so

65024650111

6.100

002ban6.0%6001GOON(

6100018-3SBOCA

1039DX"

1010.403154110.5403161017IONION074'111/0

75,151601

n15003711595NOV3067%7/104rowno.

COMPUTER COMPONENTSCPUs

1102C2 IN2004 1.906502 50

65002 210612 Oa

SO

1401 SO

65010 00

10 00

2 014 00

SO

6 00 70

00

SO

I%156

X 001170

ION1790IS CO

101017 %1490II II

11N190ONSNIN

SUPPORTDEVICES

VOLTAGE REGULATORS

of

5971056V 710669 7012. 1,7

nv /111S

1.4 71'1240 7024

50 MOS6V 711.05

IS 14-190 .1715011.75

le FIXED V017006 MASTIC 11.220

vi

0

00Oµ

0 00 sa050

7905

706nos1112MIS79111

7974

14 11100 VOL 18GI MANIC TON

03007403eOMON

5v 794061 20 19417

156 PAIS

050050o seo»

4soo4.

enON

324732,15

6470

6522652246532%MA

612/

11821s00006000615060006.525.5401351

8,M9,550156

62051212

82,68224

9/16

92.3

13( S,,,,Sof

..s7C

61/X51201

52041207WOOD

1750

,7 0000

4 SO

I 001 SO

150490

00

INSO

INGO

SO

156SO

690

So

3 10

190

n190

ea0,3

/5sa

60

n 00

so120

MN5600

00

n 00 004 0a

.0%

6 so7300

OTHER REGULATORS

9100 MIEGULAT006M 30.M3214

MCKa.

SV

3.4 Sv

SA SV

11416411.1 44006.470014,S11056.,1743117 TO 270

1413176 103174317'1113501111173.4

009104011 6N0AUET0446(ONO50352471460

nal7054071'95

450790

so

laIn490

IN

254JNSOO

2nSeso

OTHERS6061n6014121

0421706610Oen.51020576316Ti 6,

3116

1 00

10

173

20

DISP1 AVS

010357 IN770050015770

IN04050/ X0771

1N

MAN" 0170700

11441360 1 7,1

50474460 2%

1695040T705500,T74500213159014

/ORD/604-10MX IC2906, TC/9004517/A41/4111T

14%SN5 00

14 %

290n

2 SO

27311 SO

7 00

111001144 01

11104.3744 750

70345,05 1110

/90400ISOPPO S NIMEE IC 5%/9010401 SN

ME MORII S

211,3 1 004112 N410 15 2S04170.15 7004145455 600447315 00051013 50061114.2615 29062294 12 0000 2 50

PROMS

,/ 490

00

225110125

SO

190173

IBM AMSTRAD

UPGRADES

x37557_107 a60287 500207807.7 no

731

11101110on*22.11

1173'

112010.1070 61201114211 IN1 Toro onlEn

A &X IN001011 06961,

I6$101(064. 0/40,644 In

1466610 00NNW, 5 79110311 6071112 1 N1II.7100 16$6.41.910 194.0/0140 210

DISPI AYDRIVERS

450490

SO

1, PROW.

2716-052732.4527.'227054271n ZS

27C2515275127751327(121327246

4,4490120590440710In

SO -550

513

PS., 11

0441231 71111

C

C (1%100)116 R

CA75037CP16545091641193651113660936709369mo.6105

VIC064550MCl/14750%364Tm59276

12 00

I%1200MNNN

09

590694650190

10 00

INTERFACE ICs

1090,,So 10 004255,0 010NAPA 5152,

174

0942515257150

4042615)1 120414261532 1241479,00( n%00113, 6 00

011304 4 SO

D5369 4 SO

DUSK 1 40

051631 7 561

056632 150251833 225061136 1 SO

00136 12507001 GOO

74(106 60

...con ON74(3446 29004(3459 490MC53470 4 7574(3400 19410340 In74C3117 22060402. 5900(4644 9906011113 160MC14411 75474(14472 15075107 09000011 en75109 11$75110 090

/5117 71075113 12075114 14075115 1 60

Z5121 1 40

75122 1.75110, 120

Sr

nolo'62221102713

0 IS0 1.'U

CUUr511Rs

75,56051015103

75,61

?5,6215171

75161

75l%7511/2

75365

7545075451

1S452

7$1153

7560'woo75491

'54920176172.ST%ST96

1197

011.0,1515111516111$97411597SRNS1111.51n

990761441370

96.36

onn

129711

S%$ SO

SO

.N010010ON

ON0 SO

0500 70

a ro1 50

a do0 46

I JO

20

211

In12012910I001.0140140690ININ110SOO

1110

0%

DISC

CONTROLLER

ICs

,666 10NMAI 5100272 1090(07771 le Oe10/791 SO %401753 211 %

307797 22 %4601710 UDCAWN, 1500*D2743 II OaW027713 2/ 0

}110

CHARACTERGENERATORS

KEYBOARDDECODERS

1 Do

6 0

BAUD RATEGENERATORS

UARTS

1 00

, 1. 1 DC

[Oat' N1741407 4 50

74(9257.9767497121,41000

1013,314

1.43915.3916311)746111

1113101114

0157003

ut1.20043301106.01042102111162103

0071107091'SAW

190990IN

70

30IN150In32IN

IN20INININ4760 /0

TELETEXTDECODER

bo.4,6/,5050X504504'505050

GOO

7016110

9N

MODULATORS

SOUND ILVISION

CRYSTALS3211406$ 1%1 6025016 225

0,010 121451.0.064111

WI

15760066So

So

56

.550. DO

So

00

00

SO

54

5056, 75

p?

Ps

Am, /Is w Sa

SO

%SO

. 75

' mA/'5640, SO

XS..? 03

3,046. 10

13416 8-746$ SO

'84327473 SO

96141.0 77

3000.1 SO

300.06 75

000.416 IS

1,0

130,00 I 00

Please 0010All prices ars sublect tochange vothout notice

Only current prime gradecomponents stocked

We also stock a Widerange of TransostorsDiodes. Triacs Piss,

Br.dge RectifiersTn,r.sto,s and Zenois

Please phone for details

OPTO ELECTRONICS

604341.1662,CO3,21

0403,317441474 Di 704MAN71 0.707440.4640144616610

114140100 V56106.11011012

5.030T6334T32Tan11311

18 100'non

190In0%

INt6$1%202 123

20

12010

SC

DPID 1,11 ATORS

11071000729P.C57400M/ X20

1N1NIN19012

1111111,051215,13151+666'1,

179973

11

1179

10

LOW PROFILE SOCK' TS BY TI WIRE WRAP SOCKETS BY TI

TECHNOMATIC LTDMAIL ORDERS TO: 17 BURNLEY ROAD, LONDON NW IO IED

SHOPS AT: 17 BURNLEY ROAD, LONDON NWIO(Tel: (11-2(18 1177). Telex : 922800 Tel: 01-723 0233

305 EDGWARE ROAD, LONDON W2

/Go., 110 1114, 210

409,/

TURNED PIN

LOW PROFILE SATS

lko,, nolop. no110,,

'IX. NO 24/20A 057 290" 607

22w 0.61 40,/ ,1106

111, 00.71p0

1110.Ikon no

1001. 4600 Ian 154240e73 560 006 ON

PLEASE ADD 50p p&p & 15 % VAT(Export: no N .NT. p&p at Cost)

Orders from (iovernment 1)epts & Colleges etc. welcome.

Detailed Price List on request.Stock items arc normally by return of post.



APPOINTMENTSAdvertisementsaccepted up to

12 noon 25th Augustfor October issue.

I

DISPLAYED APPOINTMENTS VACANT: £27 per single col. centimetre (min. 3cm).LINE advertisements (run on): £5.50 per line, minimum £48 (prepayable).BOX NUMBERS: £15 extra. (Replies should be addressed to the Box Number in theadvertisement, c/o Quadrant House, The Quadrant, Sutton, Surrey SM2 5AS).PHONE: PETER HAMILTON, 01-6613033 (DIRECT LINE)

Cheques and Postal Orders payable to REED BUSINESS PUBLISHING and crossed.

s///1 slay on the World CirC°1/4

Increasingly multinational, high technology companies areconsidering their Australian subsidiaries an integral part oftheir global strategy So it is with our client a world leaderin telecommunications The career positions briefly detailedbelow are all based in Sydney and offer an intellectuallystimulating work environment. and of course a life stylesecond to none, Competitive salaries, relocation assistanceand a real opportunity to broaden your career and personalhorizons make this an opportunity you should not miss.

RF DESIGN ENGINEERAn experienced RF or Analogue Design Engineer to join thedevelopment team for new Radio Communications NetworkProduct, is urrit,r1t6i snunht

SYSTEMS ENGINEER (PABX)A senior n ' attract an engineer , Henced w,11

' vitalhardware cosign and real-time assiimbler programmingExcellent management career prospects

TECHNICIANSAssisting the above design engineers with prototyping, laband field testing, and pre/post production activities ischallenging and career building Experience in one or moreof the following fields is essential RF /Digital Hardware/Software Maintenance/PABXFor further details telephone Phillip Campbell B.E. (Comms)on 010 61 2 95 95 111 between 8 30 a m and 5 30 p.m localtime. or write to Execute Consulting Services P/L, Level19 Northpoint. 100 Miller St. Nth Sydney NSW 2060. Australia(Fax 010-61-2-9594312)

All applications are treated in the strictest confidence.

X F. Cl 'FIVEC(1\ St I T I NG

New South Wales Licenced Private Employment AgentNo. 779.

DEPUTY CHIEF ENGINEERTelemotive UK Ltd is a growing company within the Davy Corporation andleads the market for industrial remote control. We are extending ourproduct range and markets with an exciting new range of electronic controlequipment for the 1990s and we require an engineer, preferably chartered,to lead this important development and to liaise with marketing,commercial and production activities.

Experience in radio frequency and infra -red transmission.microprocessor application and software development will be ideal, as willbe proven capabilities in project management. Familiarity with industrialmotor control and drives is very desirable. The ability to evaluate and solveour customers' application problems is essential, as is an appreciation ofother engineering disciplines such as mechanical and drawing office work.

This new positions will carry an attractive remuneration packageincluding car, pension scheme and relocation expenses where appropriate

If you can meet the challenge we offer, please write enclosing CV to:Tim Rowley. Managing Director (Ref: ELW1).Telemotive UK Ltd, Riverdene Industrial Estate.Molesey Road. Hersham. Walton -on -Thames.Surrey KT12 4RY.

657

Davy

PAGING YOUR FUTURE!Our Client, one of the leaders in National DigitalPaging Systems, is seeking additional Engineers aspart of its expansion plan.

Field Service EngineersLondon, S. East & ManchesterWith 2-3 years RF equipment servicing experienceyou will service and support transmitter stationsthroughout designated areas of the UK. Somecomputer/peripheral knowledge would be helpful.Good customer skills are essential as is the ability towork on your own initiative. Excellent careerdevelopment prospects will also be available.

Peripheral Test TechniciansBased at the London, you willundertake component level repairs on a range ofperipheral devices associated with digital pagingnetworks. Some RF experience would be useful, butnot essential.

For further information contact Ray McKenney on(0908) 670066 or send your CV to SouthernRecruitment (Milton Keynes) Ltd, 516 Elder Gate,Elder House, Central Milton Keynes.a /southern recruitment

Hardware/Software/Systems

£9,000 - £25,000

Recruitment

As a leading recruitment consultancy wehave a wide selection of opportunities forhigh calibre Design. Development. Systemsand supporting staff throughout the U.K.If you have experience In any of thefollowing then you should be talking to usfor your next career move.ARTIFICIAL INTELLIGENCE IMAGEPROCESSING ANALOGUE DESIGN

MICRO HARDWARE & SOFTWARE GUIDEDWEAPONS C PASCAL ADA RF &MICROWAVE ELECTRO-OPTICS SIMULATION C 3 I REAL TIME PROGRAMMING SYSTEMSENGINEERING ACOUSTICS SONAR RADARSATELLITES AVIONICS CONTROL ANTENNA VLSI DESIGNOpportunities exist with National, Internationaland consultancy companies offering excellentsalaries and career advancement.To be considered for these and otherrequirements contact John Spencer or StephenMorley or forward a detailed CV in completeconfidence quoting Ref. WW/ 101.

STS Recruitment, Telephone: (0962) 69478 (24hrs),85 High Street, Winchester, Hampshire 5023 9AP.


Wanted urgentlyPractical people for theThird World.Many people want to help the Third World.But relatively few can offer the kind of helpwanted most: the handing on of skills andprofessions which lead to self-reliance.You could make this priceless contributionby working with VSO.Current requests include:ElectronicsInstructors CJ

StudioElectronicEngineer ElHospitalElectronicsEngineersLecturers inPower andCommunicationEl

Ultra -soundTechnicianRefrigeration/Radio/TV EngineersElectrical Engineersfor instruction/installationMaintenanceand repairTechnician

For more details, please complete and returnthis advertisement to: Enquiries Unit, VSO,9 Belgrave Square, London SW1X 8PW.Conditions of work: Pay based on local rates Postsare for a minimum of 2 years You should bewithout dependants Many employers will grantleave of absence.I'm interested. I have the following training/experience:

Name

Address --EWW9 88

Helping theThird World help itself.30p S.A.E.appreciated Charity N0,313757

HighburyCollege

COMMUNICATION & MEDIA STUDIES

RADIO STUDIOTECHNICIAN

Technician required to service the Radio Journalism course at theCollege. Duties will include maintenance and repair of studoequipment, demonstration and instruction in the use of equipment.Applicants should have relevant experience and be willing to

undertake training.Salary £6,376-£6,957 pa (pay award pending)

Hampshire County Council pursues a policy of equality ofopportunity and welcomes applications from people with disabilities.

Full details and application form from Personnel Office (ext 225).Closing date: 8th September 1988.Cosham, Portsmouth P06 2SA

Telephone: (0705) 383131 656

E

L

CAPITALtihu I,fir I Il arr. 71 \\ Lint., 1,1111(11111 \ 17 (1SF

30511

'I'llE I. It's No. I 11,1,,CTR(1NICS.%(;EN(lYELECTRO \ ENGINEERS it gnu are looking for

jol) ill DESIGN. FIELD SER\ WE. TECHNIC \l, SALESlir SOFIA\ RE ENGINEERING.

Telephone \M\ for one of our FREE Jobs lists orsend a full (\ to the address below

acancies throughout the I h lo I li.11OO p.a.

Capital \ppointinents Lid.. EREEPOST London \ 17 ORR.

Please send 1111' list for

\ante

\ddress

!Nisi Code

Higin('iTs

01-808 3050 - 24 HOt RS

\\NI

CLIVEDENTechnic&Recruitment

TEST ENGINEER EssexRepair and fault find radar modules andsubsystems to C1'.000INDUSTRIAL ENGINEER EssexExperience of electronics production aidprocesses. Knowledge of computerisedstock control systems. Vetting ofsubcontractors. C13,000

INSTALLATION ENGINEER BerksWiring and assembly of computer cablingIRS232/%422) for a wide variety ofcommercial companies. Travel througnoutThames Valley. cC9,0004 car

TEST ENGINEER BarksFault find and repair computer add-ons.Analogue,digital background. C10,000 -

R.F. ENGINEERS throughoutSouthern England

Several opportunities exist withincommerc al and military companies farengineers with any experience of thefollowing: Cellular radio, speech synthesis.VHF/UHF comms telephones. to 0/7.000FIELD SERVICE ENGINEER SurreyRepair and service of IBM compatible pcs.Some software skills useful. 018,000

Hundreds of other Electronic vacancies

Roger Howard, C.Eng, M.I.E.E., M.I.E R.E.CLIVEDEN TECHNICAL RECRUITMENT

92 The Broadway. Bracknell, Berks RG/2 1ARTel: 0344 489489 (24 hour)

ARTICLES FOR SALE

BANKRUPT STOCK OF HITACHILM236 640x200 dot matrix LCD dis-plays. Modules include 20 IC drive andinterface logic board. Current distribu-tor price £312 each. Our price forsealed cartons of 15 display£435+ VAT. Samples available for£35+ VAT. Carriage £11.50. Any order.Send S.A.E. for data. Tel: 0600 3715.Specialist semiconductors. FoundersHouse. Redbrook, Monmouth, Gwent.

640

UNIVERSITY OF LANCASTER

ELECTRONICSTECHNICIANRequired in the Department ofPsychology. Duties will include thedevelopment, construction.modification and maintenance of awide variety of standard andspecialised electronic equipment,included micro -computers. Amechanical aptitude and the abilityand willingness to tackle a variety oftasks are requirements of the post.Experience in electronics (analogueand digital) and appropriatequalifications are required.Salary on the scale:C8.088-0,549 pa.Informal enquiries to Mr C. S. Barker(0524) 65201, extn 4052. Applicationforms, returnable by 9 September. areavailable from theEstablishment Office,University House. Bailrigg LA1 4YW.653

ARTICLES FOR SALE

FIGOLLEDGE1----

ELECTRONIC_Sj

QUARTZ CRYSTALS OSCILLA-TORS AND FILTERS of all types.Large stocks of standard items. Spe-cials supplied to order. Personal andexport orders welcomed - SAE for listsplease. OEM support thru: designadvice, prototype quantities, produc-tion schedules. Golledge Electronics,Merriott, Somerset TA 16 5NS. Tel:0460 73718. 124721

BRIDGES waveformnitransistoranalysers. Calibrators, Standards.Millivoltmeters. Dynamometers. KWmeters, Oscilloscopes. Recorders.Signal generators - sweep, lowdistortion, true RMS, audio, RM,deviation. Tel: 040 376236. (2616)


ARTICLES FOR SALE

HAVING DIFFICULTY OBTAINING AN OBSELETEVALVE/TRANSISTOR/LC.?

We try harder to locale rare types! Delivery either ex stork 11-21 day. ar haw our averveav source. 15.8 doetU

WE ALSO STOCK ALL POPULAR TYPES - USUALLY THE LOWEST PRICES ANYWHERE(comport our prices listed below!) we'll nearly always beat any written quote.

Large discounts for wholesale quantities phone/fax/telex for on immediate quote.Free advice re egulualent etc. (we've specialised In valves and translator. since 1982).

New boed guaranteed valves. Good quality brands (IlluLBri.STC etc.). quote n11111.

RM. add IS% VAT en and 4r IS pap guard*, Telephone...AIL et send cheque 6, prompl despatch by let Claes Pon All

tk.p, r6 for Mar., KIEntronyour or dersuotina VISA. RARCIATCARDInot ACCESS -

Wasted damagepanels iai sat exogamy) ageism M banal..

IS.,.,. le.,,, 6.6. lot MO valve ratalog. of a.6. ,.,63641 220 ,84 95 VVV0640A 24 006066 120 4I34 240 110142 2 00 ,,,,,,,,,,.,,E88CC 300E88CC Mv 450

,2731 Y. 3401,166 "4,0

10-41 11011141 700

E8113) 2400 RCA 9.00 Un41 300ECC81 80 h188 USA 10 75 4CO2508ECC82 80ECC83 93

M8136 Ma 6.00M8137 Hui 700

Ernst 56.0058,254M 1400 BILLINGTON VALVES

ECM Mu, 100EE86 210

M6162 Mug 500140

58'255M 1900604688mm 6.90 GOOD QUALITY LOW PRICE RARITIES A SPECIALITY

EF81 140 140 68A6 100 59 Highland* Rand. Horsham. Sussex II/113 5LSEE92 185 80 6133 SiC 900 (Callus writ on.* but In appointment only)EE95 80 12E1 1700 Piton* 0403 210729 I.. 0403 40214 Totes 87271(134 280 445314: 270 2300

57b1 575

Chlikh 111 logant b 30 penlAnsisenihone tele* and la* !eh on o.ernighe .,dMagnetrons. KI,11.er aye tubes .loo altrektni

TO MANUFACTURERS, WHOLESALERSBULK BUYERS. ETC.

LARGE QUANTITIES OF RADIO. TV ANDELECTRONIC COMPONENTS FOR DISPOSAL

SEMICONDUCTORS. all types, INTEGRATED CIRCUITS, TRANSISTORS,DIODES, RECTIFIERS, THYRISTORS, etc. RESISTORS, C/F, M/F, W/W, etc.

CAPACITORS, SILVER MICA, POLYSTYRENE, C280, C296, DISCCERAMICS, PLATE CERAMICS, etc.

ELECTROLYTIC CONDENSERS, SPEAKERS, CONNECTING WIRE, CABLES,SCREENED WIRE. SCREWS. NUTS, CHOKES, TRANSFORMERS, etc.

ALL AT KNOCKOUT PRICES - Come and pay us a visit ALADDIN'S CAVE

TELEPHONE: 445 0749/445 2713R. HENSON LTD.

21 Lodge Lane, North Finchley, London, N.1215 minutes horn Tally Ho Corner) 11613,

WANTED

TURN YOUR SURPLUSICs transistors etc into cash

Immediate settlement. We alsowelcome the opportunity to quotefor complete factory clearance.

Contact:COLES, HARDING & Co

103 South Brink, Wisbech, Cambs.Tel' 0945 584188 Fax: 0945 588844

F STABLISIIED OVEfi ISVE ARS 1,92

STEWART OF READING110 WYKEHAM ROAD,

READING RG6 1 PL.TEL: 0734 68041

TOP PRICES PAID FOR ALLTYPES OF SURPLUS TESTEQUIPMENT, COMPUTER

EQUIPMENT, COMPONENTSetc. ANY QUANTITY

Service Manual for M.E.L.Automatic Antenna Tunertype BA 1020/01 (part ofClansman Radio Set)

wanted. High reward.Please contact:Helmut Singer,

Feldchen 16-24, D-5100Aachen, W. Germany.

Tel: 010-49-241 155315.Telex: 832504 SITRO D.

PCB ASSEMBLY

(Conventional) & TEST,iality work, attention to detail

and follow through are our forte. Nota single dissatisfied customer in tenyears. Over 18.000 boards built andtested. Low overheads and efficiency

ensure very competitive prices.

Evem EOM SEMIS 1.11Gwel An Wheal, St Ives, Cornwall.

Tel (07361794670652

WANTEDTest equipment, receivers,valves, transmitters, com-ponents, cable and electro-nic scrap and quantity.

Prompt service and cash.

M & B RADIO86 Bishopsgate Street

Leeds LS1 4BB0532 435649 995),

'LAT1NUM, GOLD, SILVERSCRAP. Melted assayed and paid forwithin 24 hours relay contacts, thermocouples, crucibles. Also printed circuitboards, plugs, connectors, palladium,rhodium, tantalum and ruthenium. Wehave the technology to do the difficultrefining jobs that others can't handle.Totally free sampling service. Sendsamples or parcels (Regd post) orcontact Eric Henderson, 0773 570141.Steinbeck Refineries (UK) Ltd,Peasehill Industrial Estate, Ripley,Derbyshire DE5 3JG. No quantity toolarge or small. 495

Marconi phase -locked frequency synthesiser.TF21708/2002. accuracy I ppm. £175 APTstablised PSU. variable 0-50 V. 2A. £39. 0-50V 12A, £30 Philips DMM. £35 Advan,Audio/Ultrasonic Generator 15c/s-50KcJs. siresquare £19 Radiation Counter. MP!.readout, light pulse indicator. audio pulp .sensitivity better than 0.1p Sv/hr.safety as well as danger. don t wait '

expected increase on sunspot activity. nuci- r.accidents etc. monitor home and workplac. hotspOts - imported materials. use for rr.i.surveys. etc etc.. £135 Avo 7 meter mo,.ment (new) El 5 Five decade DC calibrator£98 Dual stabilised PSU 0-30V. 1A. £69 Avo7 Testmeter £39 Philips portable counter/frequency meter £69 Stereo cassettemechanism, dual drive £10 60W integratedcircuit amplifiers £6 50 High intensity pre -focus lamps. microscopy. optical AF transmis-sion. signalling etc. £2 Vanaciranstormercontrolled AC variable PSU. 0-240V. 2A. 0-7V.5A, £35 Electrostatic EHT meter 0-6KV £8MIG welder. portable. single power supply.£95. dual power supply £110 Oscilloscopes.generators bridges etc. etc

040-376236. 2016

SERVICES

PCB CAD DESIGNPCB designed on computer system fur

as little as 10 60 per pin.Input information can be schematic

nett list or existing and produced PCB.

For a speedy efficient service contactus now on:

Tel: 0934 22044 orFax: 0934 416238

CALIBRATIONNeed not costa ! For a free quote on

your DV M ,,....,..,scope or frequencycounter Send type of instrument, make and

6, ,,0! dErOils tocalibration Room

3' 7 DIGIT Bluehorrow LimitedD.M.M. Brookside Cottage

FROM £5.00Main RoodBrighstone

(Calibrator to 55/)4.1) IOW P030 401 649

NORTH CORNWALL DESIGN LTD- Software and Electronic design toyour requirements. In-house CAD andprototyping facilities available. Initialconsultation free. Ring: 0566 2958 andask for Pete Hurford.

CLASSIFIED ADVERTISEMENTSUse this Form for your

PLEASE INSERT THE ADVERTISEMENT INDICATED

To "Electronics & Wireless World" Classified Advertisement Dept., Quadrant

Rate £6 PER LINE. Average six words per NAMEline. Minimum £48 (prepayable).

Name and address to be included in charge if ADDRESSused in advertisement.

Box No. Allow two words plus £15. Cheques, etc., payable to "Reed Business

Publishing" and cross "& Co." 15% VAT tobe added.

Sales andON FORM

House, The Quadrant,

WantsBELOW

Sutton, Surrey SM2 5AS

REMITTANCE VALUE ENCLOSED

PLEASE WRITE IN BLOCK LETTERS. CLASSIFICATION NUMBER OF INSERTIONS

'12 ELECTRONICS & WIRELESS WORLD

What Do Next Year'sGraduates

Think Of YourCompany's Image?

The fact is good graduates are spoilt forchoice. To stay one step ahead in the race for goodgraduates it is ESSENTIAL that YOUR companymaintains the highest profile possible.

Technology Graduate is the ONLY maga-zine for ALL graduates pursuing hi -tech careers.32,000 full -colour copies of Tecnnology Graduatehit EVERY single campus in the UK at the three mostcrucial points throughout the academic year - that'severy term.

To help YOU maintain YOUR all-importantcorporate image, Technology Graduate has de-vised a superb financial discount package.

Whether you need one or one hundred, ifyou're serious about graduate recruitment call usnow:

PAUL SISSONS 01-661 3406/3112CLARE BUTLER 01-661 3402

Winter issue delivery date is 24th October 1988(Copy Deadline - 17th October

teehnolog/GRADUAT-ftCAREER OPPORTUNITIES FOR ALL GRADUATES


MATMOS LTD, 1 Church Street, Cuckfield, West Sussex RH17 5JZ.Tel: (0444) 414484/454377.

COMPUTER APPRECIATION, 30/31 Northgate, Canterbury, Kent CT1 1BLTel: (0227) 470512.

TRIUMPH ADLER/ROYAL OFFICE MASTER 2000 DAISY WHEEL PRINTERS.20 cps. FULL IBM AND DIABLO 630 COMPATIBILITY, CENTRONICSINTERFACE. Features include underscore. bold. subscripts. superscripts.underline etc. 132 column: micro proportional spacing. Complete with typewheeland ribbon, manufactured to highest standards in West Germany by Europe'slargest typewriter manufacturer and offered elsewhere at over £350 00

£119.50 (carr. £6.50) - £99.50 each for quantities of 5+VICTOR SPEEDPAK 286. 80286 based speed up card for IBM PC and mostcompatibles. Features cache memory and runs 6 to 7 times faster

£99.50 (carr. £3.00)PANASONIC Model JU-363 3'/2" floppy disc drives Double Sided DoubleDensity 80 track 1 megabyte capacity unformatted. Latest low component '/3height design. SHUGART compatible interface using 34 way IDC connector. Willinterface to lust about anything. BRAND NEW (We can offer at least 20%discount for quantities of 10 plus). Current model. We can supply boxes of 10discs for £15.95 plus £1.50 carriage £59.50 (carr. £3.00)PLESSEY Model T24 V22/V22 bis 2400 Baud MODEM. Including free softwaredisc for IBM or MATMOS PC Compact. automatic modem featuring the latesttechnology and the highest possible data rate over the ordinary phone system.Offers; both V22 and V22 bis compatibility. 1200/2400 Baud operation with auto bitrate recognition, operation on both ordinary phone (PSTN) and private circuit (PC).auto call and auto answer, duplex operation allowing simultaneous transmissionand reception of data at 2400 Baud in both directions over a single phone line.compact size (9" x 9" x 2'12"). BT approved and suitable for new PRESTELV22bis service. Software is included for IBM PC, MATMOS PC. and (includinghigh speed Prestel) for BBC MICRO. BRAND NEW. NEW LOW PRICE

£119.50 (carr. £5.00) - £99.50 each for quantities of 5DUPLEX Model 100 green screen 12 high resolution monitor with compositevideo input With tilt and swivel stand BRAND NEW £39.50 (carr. £5.00)ITT SCRIBE III WORKSTATION. Monitor sized unit with high quality highresolution 12" green screen monitor (separated video and sync), 5V and 12Vcased switchmode power supply, processor electronics incorporating TEXAS9995 and Z8OH processors with 128kbytes and associated support chips. allBRAND NEW but with only monitor and power supply guaranteed working.Original cost at least £2.500 £29.95 (carr. £5.00)ITT SCRIBE III KEYBOARD. Low profile keyboard for above with numerickeypad. serial interface BRAND NEW but untested .......... £8.95 (carr. £5.00)TRANSDATA Model 307 ACOUSTIC MODEM. Low cost self-contained modemunit allowing micro or terminal connection to BT lines via telephone handset V24interface. up to 300Baud. originate/answer modes. etc BRAND NEW withmanual £14.95 (carr. £3.00)

FUJITSU Model M2230AS 51,4 WINCHESTER disc drive. 6.66mbyte capacityunformatted. 16 32 sectors. 320 cylinders With ST506 interface. BRAND NEW

£47.50 (carr. £3.00)DRIVETEC Model 32C high capacity 5,/a" disc drives. 3.3mbyte capacity drive -same manufacturer and same series as KODAK 6.6mbyte drive 160 track Nofurther info at present. BRAND NEW £25.00 (carr. £3.00)ASTEC SWITCH MODE PSU. 5V 10 8A: r 12V it, 3A: 12V or 0.3A - to a total65W Compact cased unit. Ex -equipment. tested £14.50 (carr. £3.00)HEWLETT PACKARD Model 59307A dual VHF switch. DC to 500MHz 50 Ohmswitch for HP-IB £185.00HEWLETT PACKARD Model 5045A digital IC tester with CONTREL ModelH310 automatic handler. With IEEE interface and print out of test results eitherpass/fail or lull diagnostic including pin voltages at point of failure. With fullcomplement of pin driver cards and complete with substantial library of magneticcard test programs for 74 series TTL and other ICs. CONTREL handler allows fullyautomatic testing of ICs which are sorted into 2 bins. Price includes a secondHP5045A (believed fully operational) for maintenance back-up f550.00HEWLETT PACKARD Model 28C pocket calculator with graphics £1 00.00CASIO Model FX8000 calculator with graphics and interface £50.00ICOM micro2E 2 metre transceiver with carrying case £150.00TIME ELECTRONICS Model 9810 programmable power supply £160.00ITT PERFECTOR TELEX MACHINE. With 32k memory. screen with slow scrollingetc £350.00HEWLETT PACKARD MODEL 5501A LASER TRANSDUCER. Withpiezoelectric tuning for precise control of wavelength for measuringapplications £350.00VICKERS INSTRUMENTS MODEL M17 METALLURGICAL MICROSCOPE withbinoculars micrographic head and all eyepieces With 4 Microplan objectives andNomarski interference contrast £1,250.00KRATOS MS30 DOUBLE BEAM MASS SPECTROMETER. Approximately 8years old with negative ton capability and fast atom bombardment (FAB). With gasand direct introduction sample probes and with gas chromatograph inlet system.Output spectra are available directly via a HEWLETT PACKARD storage displayand a UV recorder. An on-line DATA GENERAL DS60 computer system, whichincludes a graphics printer and two TEKTRONIX 4014 terminals. analysesoutput ... p.o.a.LUMONICS SYSTEM 2000 RUBY LASER with 0 -switch and frequency doubler0.3 Joule per pulse. 6 ppm. Suitable for holography A low power He/Ne laser formirror alignment is included £3,500.00SPECTRA PHYSICS Model 164-08 Argon ion laser 4 watts output (all lines)Complete with 3 phase transformer £3,500.00Please note: *VAT & carriage (also - VAT) must be added to all prices.* VISA and ACCESS orders accepted

ENTER 19 ON RENA' CARD

INDEX TO ADVERTISERSAppointments Vacant Advertisements appear on pages 836-839

PAGE PAGE PAGE PAGEAircastle Products 925 Field Electric 901 Level) Electronics 926 Sherwood Data Systems 899Airlink Transformers 870 Livingston Hire 896 Slee Electro Products 899

Gould Electronics I FC Stewart of Reading 886M & B Radio (Leeds) 901 Strumech Engineering 896

Cambridge Kits 902 Hameg 890 M A Instruments 902 Superkit Electronics Inc 895Cavendish Automation 931 Harrison Electronics 870 M Q P Electronics 886 Surrey Electronics 928Chemtronics UK 851 Hart Electronic Kits 931Colomor Electronics 925 Henrys Audio Electronics 895 Number One Systems 902CommuniqueComputer Appreciation

925944 Icom I UK) Limited 931 Omega Dynamics 863

Taylor Bros. (Oldham)Technomatic

842/IBC938/939

Dataman DesignsDisplay Electronics

OBC935

John's Radio

Kestrel ElectronicComponents

926

873

Private Mobile Rentals

R Henson

899

926

TektiteThandar ElectronicsThose EngineersThurlby Electronics

901851928928

Ralfe Electronics 886Engineering solutions 873 LI Electronics 870 Raycom CommunicationsEuropean Electronic Systems 890 La ngrex Supplies 895 Systems 896 Webster Electronics 851

OVERSEAS ADVERTISEMENT AGENTSFrance and Belgium: Pierre Mussard. 18-20 Place de la Madelaine. Paris 75008United States of America: Jay Feinman. Reed Business Ltd.. 205 East 42nd Street. New York. NY 10017 Telephone 12121867 2080 - Telex 23827

Printed in Great Britain by E.T. Heron i Print/ Ltd, Crittall Factory, Braintree Road. Witham. Essex CM8 :NO. and typeset by Graphac Typesetting, 181 191 Garth Road, Morden, Surrey SM4 4LL.for the proprietors, Reed Business Publishing Ltd, uadrant House. The Quadrant. Sutton. Surrey SM2 5AS. ©Reed Business Publishing Ltd 1988. Electronics and Wireless World can be obtainedfrom the following: AUSTRALIA and NEW ZEALAND; Gordon & Gotch Ltd, INDIA. A. H. Wheeler & Co. CANADA: The Wm Dawson Subscription Service Ltd., Gordon & Gotch Ltd. SOUTHAFRICA: Central News Agency Ltd; William Dawson & ,on !S.A.) Ltd. UNITED STATES: Worldwide Media Services Inc.. 115 East 2:1rd Street, NEW YORK. N.Y. 10010. USA. Electronic &Wireless World $5.95 (745130.


TAYLOR R.F. EQUIPMENT MANUFACTURERS

PERFORMANCE& QUALITY

WALLMOUNT DOUBLE SIDEBANDTELEVISION MODULATORPRICES FROM ONLY £104.53 (excluding VAT & carriage)

PricesCCIR/5-1 1 ModulatorCCIR/5-2 2 ModulatorsCCIR/5-3 3 ModulatorsCCIR/5-4 4 ModulatorsCCIR/5-5 5 Modulators

£104.53£159.99£226.28£292.56£358.85 ENTER 2 ON REPLY CARD

19" RACK MOUNT CRYSTAL CONTROLLEDVESTIGIAL SIDEBAND TELEVISION MODULATOR

PRICES FROM £203.93 (excluding VAT & carriage)Prices CCIR13 £203.93

CCIRI3-1 £260.64

19" RACK MOUNT VHFIUHFTELEVISION DEMODULATOR

PRICE AT ONLY £189.00 (excluding VAT & carriage)

CCIR/3 SPECIFICATION

Power recimement - 240V 8 Watt (available other voltages)Video Inpit - IV Pk -Pk 75 OhmAudio Input - 8V 600 OhmFM Sound Sub -Caner - 6MHz (available 5 5MHz)Modulation - NegativeIF Vision - 38.9MHzIF Sound - 32.9MHz (available 33 4MHz)Sound Pre-Emphass - 50usRipple on IF Saw Filter - .6dBOutput (any chann.l 47-860MHz) - +6dBmV (2mV) 75 OhmVision to Sound Power Ratio - 10 to 1

Intermodulation - Equal or less than 60dBSpurious Harmonic Output - -40dB (80dB if fitted with TCFL1 filter or

combined via TCFL4 Combiner/Leveller

CCIR/3-1 - Specification as above but output level60dBmV 1000mV Intermodulation 54dB

Other Options Avalable - I F Loop/Stereo Sound/Higher Power Output

Alternative Applications - CCTV Surveillance up to 100 TV channelsdown one coax, telemetry camera controlsignals. transmitted in the same coax in thereverse direction

802 DEMODULATOR SPECIFICATION

Frequency Range - 45-290MHz. 470-860MHzA.F.C. Control - +/- 1.8 MHzVideo Output - IV 75 OhmAudio Output - 75V 600 Ohm unbalancedAudio Monitor Output - 4 Ohms

Tunable by internal presetAvailable for PAL System I or BC

Options - Channel selection via remote switchingCrystal Controlled TunerStereo Sound.

CCIR/5 MODULATOR SPECIFICATION

Power Requirement - 240VVideo Input - IV Pk -Pk 75 OhmsAudio Input - IV rms 30K Ohms Adjustable 4 to I .

Vision to Sound °ower Ratio - 10 to 1

Output - 6dBmV (2mV) 470-860MHzModulation - NegativeAudio Sub -Carrier - 6MHz or 5 5MHzFrequency Stability - 25 Deg temperature change 150KHzIntermodulation - less than 60dBSound Pre -Emphasis - 50usDouble Sideband Modulator(unwanted sideband can be suppressed using TCFL4Combiner/Level'er)

CHANNEL COMBINER/FILTER/LEVELLERto combine outputs of modulators

TCFL2 2 Channel Filter/Combiner/Leveller. Insertion loss 3.5dBTCFL4 4 Channel Filter/Combiner/Leveller. Insertion loss 3.5dBTSKO Enables up to 4.TCFL4 or TCFL2 to be combined

TAYLOR BROS (OLDHAM) LTDBISLEY STREET WORKS, LEE STREET,OLDHAM, ENGLAND.

TEL: 061-652 3221 TELEX: 669911 FAX: 061-626 1736

"It's a pretty small battery -powered PROM programmer- so what?"

Tools which are convenient get used a lot - that justifiestheir existence. There is no way we couldexplain all the usefulness of S3 here.Instead, if you're interested we'regoing to let you see it, use itand evaluate it in your ownworkshop. We went to a lotof trouble to design S3 justthe way it is - no otherPROMMER is all CMOS andall SMT. So we must he con-vinced that S3 would be a for-midable addition to yourarmoury. Now all we have to dois to convince you.

"Such a little thing can't bepowerful, like a big bench -

programmer - er - can it?"Yes, it can. It is more powerful. S3 leavesother prommers streets behind. S3 hascontinuous memory. which means that youcan pick it up and carry -on where you left offlast week. S3 has a huge library EPROMS andEEPROMS. S3 can blow a hundred or morePROMS without recharging. S3 also works re-motely, via RS232. There's a DB25 socket on thehack. All commands are available from your computer(through a modem, even). Also S3 helps you developand debug microsystems by memory -emulation.

"What's this memory -emulation, then?"It's a technique for Microprocessor Prototype Develop-ment, more powerful than ROM emulation, especiallyuseful for single -chip "piggyhack" micros. You plug thelead with the 24/28 pin headerin place of the ROM/RAM.You clip the Flying -Write -Lead to the microprocessorand you're in business. Thecode is entered using eitherthe keyboard or the serialinterface. Computer -assembledfiles are downloaded in standard format - ASCII, BINARY,INTELHEX, MOTOROLA, TEKIIEX.Your microprocessor can WRITE to S3 as well as READ. Youcan edit your variables and stack as well as your program, ifyou keep them all in S3.S3 can look like any PROM up to 64K bytes, 25 or 27 series.Access is 1(X)ns - that's really fast. Memory -emulation ischeap, it's universal and the prototype works "like the realthing".

S3 loads its working programs out of aPROM in its socket, like a computer

loads from disk. Software expansion isunlimited. Upgrades will come in a

PROM. Programs can be exchangedbetween users. How's that for

upgradability?

"Can I change the way it works?"You surely can. We keep no secrets. System Variables can be"fiddled." New programming algorithms can be written fromthe keyboard. Voltages are set in software by DACs. If youwant to get in deeper, a Developers' Manual is in preparationwhich will give source -code, BIOS calls, circuit -diagrams, etc.We expect a lively trade in third -party software e.g. dis-assemblers, break -point -setters and single -steppers for variousmicros. We will support a User Group.

I, "I'm bound to let the battery go flat.".0 Quite so. But in practice it doesn't matter. S3 switcheslit off after a half-hour of non-use anyway, or when the7o battery gets low. You don't lose your data. Then a

slow -charge overnight or boost -charge for three2 hours will restore full capacity. You can keep using

it when charging. So there really is no problem.

"I already have a programmer."Pity it doesn't have S3 features, eh? Buthere's a trick worth knowing. If you plug

S3's EMULead into the master socket of aganger then you get an S3 with gang

capacity. Isn't production separatefrom development anyway?

"It looks nice. Will I bedisappointed?"

Dataman tools are designed tobe used by Engineers. Notjust sold to Management.Have you ever been misledby some mouthwatering

ad for a new product.Great artwork and ex-citing promises whichfeed your fancy? Onimpulse you buy and

when the thing arrivesyou feel let down. The pic-

ture looked better. The claimsare hardly justified; not exactly

misrepresentation, justpoor implementation.But you've bought it.And you're stuck with it.

Softy3 is here!"Refund in the first month!that?"We trust S3 to fire your enthusiasm.as a free hire -service. We bet youwould you manage without it?

It stays in the cupboard,most of the time. So howabout this: buy S3 anduse it for up to a month.If you're nut still thrilledthen you can have yourmoney hack.

How can you offer

We trust you not to use uswon't send it hack. How

"These things cost a fortune and take months toarrive."We wouldn't get you all excited and then let you down. It Costs£495 plus VAT. That includes P & P, Charger, EMULead,Write Lead and a HELP program in ROM. S3 is in stock. Buyit today. Use it tomorrow. (That's a fair promise. But pleasereserve product by phone or telex to make it come true).

Lombard House,Cornwall Rd.DORCHESTERDorset DT1 1RX.Phone 0305 68066 Telex 418442If you purchase while this ad is current, you have 28 days toexamine the goods and return them for refund. Carriage will becharged at cost. The right to charge the cost of refurbishmentof damaged goods is reserved.

TOOLSFr )1?

. VIC/0 ,E.VG/NEERIA

VISA


& wireless world · plotter for instart print-outs of displayed traces, and the ability to...

Documents