worldradiohistory.com€¦ · awnt le &the wireless engineer vol.. iii. october, 1926. no. 37....

AWNTLE

&The WIRELESS ENGINEER

Vol.. III. OCTOBER, 1926. No. 37.

Editorial.The B.E.S.A. Glossary of Terms used in

Electrical Engineering.ON page 612 of this issue will be found a

brief review of this publication andparticulars as to price, etc. In a paper

read before the Engineering Section of theBritish Association at Oxford, on Wednesday,iith August, we made a vigorous protestagainst the unmerited slur which the BritishEngineering Standards Association has castupon the fellow countrymen of Faraday,Clerk -Maxwell and Kelvin by giving itsauthoritative imprimatur to the publicationof such a collection of definitions as are to befound on pages II to 18 of this book, and weexpressed the hope that the Association wouldtake immediate steps to recall it and issuea note asking those who had purchasedcopies to delete these pages pending theirrevision. The outcome of this has been theappointment by the British Association ofa committee to co-operate with the B.E.S.A.in the revision of the definitions. We cannothere go into the definitions seriatim* butwe may indicate briefly the sort of confusionwhich exists among the definitions of eventhe most fundamental magnetic and electricterms. Electrostatic Flux is defined as" The number of unit electrostatic tubes offorce traversing a given surface. The total

* The paper was published in full in the Elec-trician of 2oth August, and in Engineering of2oth August.

flux over a surface enclosing a charge isequal to the charge." Here we have a hope-less confusion between two different things,viz., the electric force and the flux or dis-placement produced by the force. In acondenser, for example, if the P.D. is fixed,the electric force is fixed, but the flux orcharge depends on the dielectric constant orpermittivity of the medium. How then canthe electrostatic flux be defined as the numberof tubes of force ? Exactly the same con-fusion occurs in the magnetic units, forMagnetic Flux is defined as the number ofunit magnetic tubes of force traversing agiven surface. The impossibility of thisdefinition is obvious from the fact that whenmagnetic flux passes from iron to air themagnetic force undergoes an enormouschange ; in fact, another definition tells usthat permeability is the ratio of the magneticflux produced by a given magnetic force inthe material or medium to the magneticflux which would be produced by the samemagnetic force in a perfect vacuum. Howcan the ratio change if flux is defined as thenumber of tubes of force ? The reason forthe muddle is to be found in the failureto distinguish between cause and effect,between the force and the resulting displace-ment, between the stress and the resultingstrain.

There is another trouble, however, whichis caused by a real difficulty, although tojudge from the definitions, those who drew

October, 1926

them up were not aware of the difficulty.The term "potential difference" is based on thepotential theory, and is only strictly applic-able to fields in which the work done inmoving unit charge or pole from one pointto the other is independent of the pathfollowed. In such a field the work done inmoving round any completely closed curvemust be zero. Now when the magneticflux through any closed curve is changing, thework done in moving unit charge around thecurve is not zero, but is equal to the E.M.F.induced in the path. Strictly speaking,therefore, the potential theory is not applic-able to alternating current circuits, but inthe majority of practical cases it can beapplied with little ambiguity. When onespeaks of the P.D. between two transmissionlines one assumes that the voltmeter wiresby which it is measured ran approximatelyfrom one line to the other by the shortestpath, and not via the power station and thestator slots. In radio -telegraphy and allhigh frequency work, one has to be verycareful about the voltmeter leads whenmeasuring the so-called P.D. between twopoints, although if the P.D. were equal tothe work done in moving unit charge fromone point to the other, irrespective of thepath followed, such precautions would notbe necessary. As an example of the confusioncaused by the misapplication of the potential

588 EXPERIMENTAL WIRELESS &

theory, we may choose an example whichwill appeal to our least mathematical reader.Electromotive Force is defined as " anelectric condition tending to cause a move-ment of electricity in a circuit. It is measuredby the sum of the potential differences frompoint to point round the circuit." One of thefirst things we do in teaching students theprinciples underlying electrical engineeringis to try to get them to distinguish clearlybetween electromotive force and potentialdifference, but in this definition the two aremixed up together. If one makes a closedring of copper wire and suddenly plungesthe pole of a bar magnet into the centre of it,a current is produced around the ring becausean E.M.F. is induced in it, but from con-siderations of symmetry every part of thering is at the same potential ; there is anE.M.F. but there are no potential differences.One cannot therefore define the E.M.F. asthe sum of a lot of non-existent potentialdifferences.

These few examples will serve as a warningto those readers who may turn to the Glossaryfor enlightenment on fundamental magneticand electric terms. They may even servea more useful purpose by causing manyreaders to look to their own foundations andsee if they have clear conceptions of themeanings of these fundamental terms whichoccur on every page of electrical literature.

THE thermionic voltmeter has becomea very powerful weapon in variousH.F. measurements, and little infor-

mation concerning the operating conditions of;he various forms of the voltmeter appearsto have been published since the introductionof the direct deflection voltmeter by E. B.Moullin.* The present paper gives theresults of an experimental investigation ofthe performance of five different types ofthermionic voltmeter from the points ofview of :-

(a) Sensibility and range.(b) Possible frequency error.(c) Wave form error.(d) Input load.(e) Stability of calibration.

Type 1.-Deflection Type, Rectifying onCurvature of the Anode Characteristic.The essential connections of this well-

known type of voltmeter are given in Fig. Z.The rectified current through the micro -ammeter G will vary with the A.C. voltage

3--

Fig. x. Connections of deflection voltmeter,, Type I,rectifying on curvature of anode characteristic.

applied across the input terminals, and thescale of the instrument may be calibratedto give a direct reading of the input voltagefor given values of the filament current,steady bias voltage, and anode voltage dueto the battery B.

* I.E.E. Journ., Vol. 6i.

THE WIRELESS ENGINEER 589 October, 1926

The Thermionic Voltmeter.By W. B. Medlam, B.Sc., A.M.I.E.E., and U. A. Oschwald, B.A.

[R261The voltmeter will function without the

anode battery B if the galvanometer lead istaken to L.T.± as shown by the dottedconnection, but a certain amount of extraanode voltage is a decided advantage as isshown later. The condenser K (of the orderof IpF) prevents losses due to H.F. currentsflowing through the galvanometer and anodebattery. It also steadies the reading of theinstrument when the input voltage carriesa varying modulation. To put the volt-meter into operation its input terminals areshorted, and the filament resistance isadjusted until the reading of G is a definitevalue, taken to be the zero of the voltagescale.

(a) Sensibility and Range.The smallest voltage which can be read

with this type of voltmeter depends uponthe values of anode voltage and negativegrid bias. The value of the latter is more orless definitely determined by the maximumvoltage it is desired to measure on the volt-meter. In order to avoid grid current thebias should be about one volt greater thanthe peak value of this maximum voltage.Using this bias, the lowest voltage readableon a microammeter will be about two voltsless than the maximum voltage. That is,the scale of the voltmeter covers a range ofabout two volts, commencing at any desiredvoltage that may be safely applied to thegrid, say roo volts or so. From this it willbe seen that the sensibility decreases con-tinuously as the negative bias is increased.This point is illustrated in Fig. 2 which showsa series of calibrations for values of the biasbetween -2 volts and -6 volts. Forexample, suppose the voltmeter is to measureup to 5 volts (peak value), the bias would befixed at -6 volts. With this bias the changeof deflection becomes too small to read withless than 1.41 volts peak, giving an effectiverange from I to 3.53 volts (R.M.S.). To readinputs of the order of r volt (R.M.S.) moreaccurately the bias may be altered to -3

B2

October, 1926

volts, changing the effective range to 0.2 to1.4 R.M.S. volts, and so on. To measurevery small voltages most accurately nonegative bias should be used, but in this

90

cC 80a.

g 802Z 60

5ccce

0

0w 20

010<

70

40

30

I8

I-II

/ -2 /- 3 1 -4 1-5 ,I-6iI

I I I, I

II I

I II

I1

I1

II

8

. ,

II

8II I I

II

8II

1

II I I

. i I/ II II I*

I IIr I

III

, 8I i

0 2 3 4 5

R.M.S. VOLTS

6

Fig. 2. Calibration curves of deflection voltmeter,Type I.

case the range should be restricted to themeasurement of a few tenths of a volt only,owing to the increasing load produced by gridcurrent as the A.C. voltage is raised. Thelowest voltage measurable with this typeof voltmeter and a microammeter is of theorder of o.i volt. It may be mentioned thata different calibration is necessary for eachrange. The calibrations may be convenientlyarranged in linear form, on a postcard, asshown in Fig. 3. The A.C. volts correspond-ing to any anode current may be read

0

ANODE OURRENT BIAS -6.05 10 15 20

LI/IIIII kith 11 I

0-5 41.5 2.5I

3.5 3.71 2 3

R. M .S VOLTS

Fig. 3. A convenient method of recording voltagecalibrations of a microammeter scale.

directly from the lower of the two scalescorresponding to the particular bias in use.

The general effect of anode voltage on thesensibility is illustrated in Fig. 4, in whichthe relation between anode current and anode


voltage is shown, for zero applied voltage bycurve A, and for a certain constant A.C.voltage by curve B. The change in anodecurrent (i.e., the vertical difference betweencurves A and B) is shown to an enlargedscale by curve C. In this particular case,u9

ui 800

05cc0

za

cr5 200cc

0 100

0 0

IIIIIIIIM1111111111111111111111111111, , 1111111111111111111111111.111111113/111

1111011111111111111101111111111111211111111111111111110111111111111P2C21111

111111111111111111P11111/21121111111111101211111111MIKIIIIIIIII

11111111511EMOMMIIIIS101111111161111ITAMIC111111111111110

IIIIMIIIIEM111111111111111111B1

Fig. 4.

10 20

ANODE VOLTS

Effect of anode voltage on theof the Type I voltmeter.

cc

3120 w

O 8

80

60

4030

sensibility

in which no grid bias was used, the best H.T.was 15 volts. The best value, however,increases with the bias. Thus, although thevoltmeter may be used with the anodeinstrument connected direct to L.T. -I-, the

8

7

6

5

4

3

2

010 20ANODE VOLTS

SO

Fig. 5. Effect of anode voltage on grid current.

sensibility is considerably increased with acertain amount of extra anode voltage,particularly when the grid is given a steadynegative bias. This extra voltage is alsobeneficial in another respect : it reduces the

THE WIRELESS ENGINEER

possibility of grid current to some extent, asshown by the curve in Fig. 5, which refersto the same valve with zero grid bias.

(b) Frequency Error.There does not appear to be any appre-

ciable frequency error in this type of volt-meter up to a frequency of 2 X 108 cycles.A calibration at this frequency agreed withinthe limits of experimental error with one at4o cycles,* and further, the A.C. calibrationunder any given conditions is calculableexactly from a static characteristic of thevalve. Fig. 6 shows how close is the agree-ment between a calculated and observed

In

ccw 6020

40

20cccc

O 10

0O 0

50

30

10 20 30A.C. VOLTS.(PE AK)

Fig. 6. Close agreement of a calculated and observedcalibration curve. (See Table Ma.)

40

calibration. The points on the curve locatedby X's were obtained from an A.C. test,those shown by circles were calculated fromthe static characteristic. The calculationsare given in detail in the Appendix.

(c) Wave Form Error.Errors may arise if the instrument is used

to measure voltages having a wave formdifferent from that on which it is calibrated.The. errors become more serious as theamplitude of the voltage is increased, i.e., asthe instrument reads nearer the top of itsscale. The lower part of the scale followsa square law approximately and no waveform error occurs where this law holds, butover the middle and upper parts of the scale

* Care was taken to eliminate harmonics in thewave forms.

591 October, 1926

the calibration approximates more to a linearlaw and large errors may be introduced ona distorted wave. To give some idea as tothe magnitude of these errors the results ofa comparative test are given in Table I. and

Fig. 7. Wave form (A)-a sine wave.

Fig. 10 for (a) a sine wave, (b) a wave witha pronounced third harmonic, and (c) a wavewith a second harmonic. The wave formsare given in Figs. 7, 8 and 9 respectively.The equations to these wave forms, neglectingunimportant higher harmonics, are :-

(b) E= sin cut- .31 sin Scot + .05 sin 5wi

(c) E= sin cot +.26 sin (2a)t2 )

In (b) the ratio of peak value to R.M.S. is1.83, and in (c) the ratio peak/R.M.S. is 1.05

Fig. 8. Wave form (s)-has third and fifthharmonics.

for the flat half and 1.77 for the reverse halfof the wave, the ratio of the peaks beingli.7.The errors in the last three columns ofTable I. were deduced from the sine wavecalibration.

October, 1926

The results in Table I. show that the largethird harmonic (b) increases the readingabout 2 per cent., the flat half of the secondharmonic (c) produces a more variable andrather larger error, and the peaky half of (c)produces unexpectedly large errors of theorder of 22 per cent. for readings greaterthan i volt.

TABLE I.H.T. =18 volts. Grid bias -3.o volts.

R.M.S.volts.

Anode current (pA) withwave forms. Per

b

cent. error.

a b Peakyhalf

makinggrid +

Flathalf

makinggrid +

Peakyhalf

Flathalf

0 14.5 24.5 24.5 24.50.5 17.5 17.7 27.8 x6.8 +0.5 + 0.6 -26.o1.0 24.0 25.3 28.0 24.5 +7.0 +22.0 + 2.0X.5 35.0 374 47.3 37.2 +6.0 +25.0 + 5.32.0 52.4 54.0 71.0 55.7 +2.0 +21.0 + 3.22.5 75.0 I 77.0 101.5 78.5 +2.0 +21.0 + 3.23.0 100.0 1101.4 237.0 105.0 +0.8 +23.0 + 3.3

These wave are great enoughto indicate the necessity for care in the use


Fig. 9. Wave form (c) -has a second harmonic.

of a valve oscillator, the supply from whichcontains both even and odd harmonics.Such a supply may be effectively filtered bypicking up through a tuned circuit looselycoupled to the oscillator. For instance, theE.M.F. wave (b), Fig. 8, was applied acrossan inductance loosely coupled to a circuittuned to the fundamental. The wave formof the voltage across the condenser of thistuned circuit did not differ perceptibly froma sine wave.

(d) Input Load.

The power factor of this type of voltmetercan be reduced to a fairly low figure whenthe voltmeter is operated under properconditions as regards the value of the steadybias in relation to the peak value of the A.C.voltage, and with a moderate value of H.T.voltage. The power factor, however, risesvery rapidly once the grid is allowed to attaina positive potential during any part of thecycle of the input voltage, A typical set ofpower factor measurements on an "R"type valve are given in Table II. In thistable the second column shows the input

TABLE II.

Conditions. Capacity.(212F)

Powerfactor.

Relativepower lost.

A 2.4 .011 IB 5.7 .019 7C 8.o .025 13

capacity, and the fourth column the powerlost expressed as a ratio to that lost underconditions (A). The conditions are asfollows :-

(A) Voltmeter panel with valve holder,terminals, and all internal wiring. Ex-cluding the valve and external connectionsto H.T. and L.T. batteries.

(B) Same as (A), but with the valve inits holder.

(c) Voltmeter under working conditions,with an input of 2 volts R.M.S. at A=480 metres. Bias -3.3 volts. H.T. 25volts. Mica condensers of .02µF con-nected across the bias and from anode to

From the above results it will be seen thathalf the total loss with the voltmeter workingnear the top of its scale is due to losses inthe dielectrics used in the construction ofthe valve. These losses vary considerablyin valves produced by different firms and ifone has a number of suitable valves to choosefrom, it is worth while making a preliminarytest of their relative losses. The testnecessary to classify the valves in their orderof losses is a very simple and rapid one. Onhigh frequencies a condenser connected asshown in Fig. r may materially reduce thepower factor.


(e) Stability of Calibration.Error due to some internal change in the

valve itself appears to be very small. Forexample, one valve used continuously for12 months showed no appreciable change in

180

140

120

100

80

80

40

20

ab

ci(FtArr

010 30

R.M.S. VOLTS

Fig. io. Calibration curves of the Type I voltmeteron wave forms (A), (B), and (c).

calibration over this period. In fact, anyinternal change is very largely compensatedfor when the voltmeter is operated byadjusting the filament current to produce aconstant value of anode current with no A.C.volts on the grid. This method also partially

20

55

45

35

40

35

3014 18 18 20 22

ANODE VOLTS

Fig. 11. General effect of a change in anode voltageon the calibration of the Type I voltmeter.

40

compensates for changes in anode voltageand grid bias, but serious errors may ariseif these voltages differ much from theiroriginal values.

593 October, 1926

The effect of change in anode voltage isshown in Fig. II by the full line curve, whichwas obtained under the following conditionswith a 235 -type valve : Constant grid bias- 2.8 volts, constant anode current withno A.C. volts Is:44A (the filament currentnecessary to give this reading with variousanode voltages is shown by the dotted curve).Constant input of 2 volts R.M.S. Curvesobtained under other conditions were ofsimilar shape, showing increasing slope asthe H.T. was increased and the filamentcurrent reduced.

Unless the filament current is low the anodecurrent rises linearly as the anode voltageis reduced. The slope of the curve in Fig. IIis such that i volt (say 5 per cent.) change inH.T. alters the anode current by 1.65µA.

55

W

W Wcc a.

S0 t) 45 zp00 E

z35

2.5 3 5 4 5GRID BIAS IN VOLTS

Fig. 12. General effect of a change in bias voltageon the calibration of the Type I voltmeter.

From a calibration curve it was found thatthis change in anode current correspondedto a change in A.C. volts of .04 volt, i.e., from2.00 to 2.04 volts, or 2 per cent. Thus nearthe top of the scale the voltmeter reads

per cent. high when the anode volts drop2.5 per cent. With low filament currentsthe error becomes greater.

The effect of a change in steady grid biasis shown in Fig. 12 for the same valve withthe same anode current with no A.C. voltsand with a constant input of 2 volts as before.The anode current with the A.C. voltageapplied was noted for a series of values ofthe bias (after adjustment of the filamentcurrent to give the same " zero " of ioiLA).The results for five different values of theanode voltage from 16 to 24 volts are asshown in Fig. 12. It will be noted thatthe general slope of these curves is verymuch less than is the case in Fig. II

October, 1926 594

(horizontally they cover about the samepercentage change in voltage) indicating,in general, a much smaller error due toa change in the bias than is producedby the same percentage change in anodevoltage. Fig. 12 shows that as the bias isreduced the curve may rise, keep level, orfall, according to the operating conditions;thus these may be chosen so that little erroris produced by, say, a 20 per cent. fall inbias voltage.

Type 1L-Deflection Type with LeakyGrid Rectification.

The connections of this type of voltmeterare shown in Fig. 13. In this type thegalvanometer reading decreases as the inputvoltage is increased. Its obvious advantagesover the former type are that it does notrequire a closed input circuit, i.e., it can beused to measure the voltage across one oftwo condensers in series, and it is not affected

Fig. 13. Connections of the deflection voltmeter,Type II, with leaky grid rectification.

by any D.C. voltage across the input providedthe grid condenser has an insulation resis-tance very much greater than the resistanceof the leak. The importance of high insula-tion resistance in the grid condenser cannotbe emphasised too strongly. To take anextreme, but quite practical case, supposeone wishes to measure the A.C. voltage acrossan anode resistance of 50,000 ohms in aresistance -coupled amplifier, and suppose theA.C. voltage' to be measured is of the orderof I volt. The anode current (D.C.) may beof the order of 2 milliamps, giving a D.C.drop on the resistance of roo volts. A simplecalculation will show that a D.C. voltageexceeding r per cent. of the A.C. voltagewill be applied to the grid unless theinsulation resistance of the grid condenserexceeds the huge figure of 20,000 megohms,

EXPERIMENTAL WIRELESS &

with a grid -leak of 2 megohms. Thenecessary insulation resistance of the con-denser increases with the leak resistance.

400

300

200

100

010 2-0 30 40

R.M.S. VOLTS

Fig. 14. Specimen calibration curve of the Type IIvoltmeter.

(a) Sensibility and Range.A specimen calibration for a voltmeter of

this type is shown in Fig. 14. From thiscurve it is evident that the sensibility isgreatest between the limits of 0.5 and 2 volts,and falls rapidly for voltages exceeding 2.

laZ 20CC CACC LiM IX 180 wa.aaa00Zcc<0 8141 Z-

12

4

008R.M.S. VOLTS

Fig. 15. A low voltage calibration curveof the Type II voltmeter.

Thus the range is restricted to the measure-ment of not more than about 3 volts, and wehave not been able to devise any simple wayof extending the range, such as can be donein the first type by adjustment of the grid bias.

018 0-24


A low voltage calibration is shown in Fig.15 by the full line curve. For comparison acalibration over the same range for a Type I.voltmeter is shown by the dotted line. Thegrid -leak type appears about three timesmore sensitive than the other, and readsdown to about .04 volt with a microammeter.Owing to the large steady anode current withno applied voltage it is usually necessary tobalance out this current in order to avoidhaving to shunt the microammeter and thuslose entirely the extra sensibility of thismethod. The usual balancing arrangementsare shown in Figs. 16 and 17. A low resis-tance potential divider is connected across thefilament accumulator (Fig. 17) or separateaccumulators (Fig. 16). Current through thegalvanometer from the potential divider is

Fig. 16. A method of balancing out the steadycomponent of anode current.

in the opposite direction to the steady anodecurrent, and may be made equal and oppositeto it by adjusting the potential at the tappingend of the resistance R. The resistance Rshould be made large compared with that ofthe instrument and potential divider tominimise loss of sensibility due to its shuntingeffect. The multiplying power of the shunthas usually to be determined from a separatecalibration.

Unlike the first type of voltmeter, theleaky grid type will not operate when theanode voltage is below a certain value. Withlow anode voltages rectification on the lowerbend of the characteristic occurs and theanode current rises-instead of falling-withincreased applied volts. It is, in fact,possible to adjust the H.T. so that no changein anode current occurs when a voltage isapplied to the grid. These points are wellshown in Fig. i8, in which the full line curvesshow how, in an actual test, the anodecurrent varied with the H.T. with (a) no

595 October, 1926

voltage applied to the grid and (b) an alter-nating voltage of R.M.S. value 1.61 voltswas applied. The change in anode currentdue to the A.C. voltage is shown to anenlarged scale by the dotted curve.

Fig. 17. Another method of balancing out the steadycomponent of anode current.

The best potential to be applied to thefilament end of the grid -leak has to be deter-mined by trial, but it may be mentioned herethat any gain in sensitivity due to a positivepotential on the leak may be more than offset

200

100

0

a

\,10 20 40 80 80

ANODE VOLTS

40 cccc

+

30 °Wret0 2

2 0st cc10

0 Wz2

0

20

10

20100

Fig. IS. Effect of anode voltage on the sensibility ofthe deflection types of voltmeter.

by the rise in power factor of the voltmeterdue to this positive connection, and by thepossibility of a change in calibration due toalteration of this potential as the cells rundown.

October, 1926

(b) Frequency Error.This type of voltmeter is not entirely free

from frequency errors, although these maybe made small by suitable choice of valuesfor the grid capacity and leak resistance.For present purposes the voltmeter is equiva-lent to the arrangement shown in Fig. 19,in which R is the leak resistance, K, the gridcondenser, and K2 the sum of the grid -filament and grid -anode capacities, includingthe capacities of the leak and wiring. K.will be of the order of xotg.LF, having areactance of 20,000 ohms at a wavelengthof 377 metres, rising to goo megohms onabout so cycles. As the leak resistance Rwill be between the limits of 1 to io megohmsit is evident that at high frequencies theeffect of R is negligible, whereas at verylow frequencies that of K2 is negligible.

Fig. 19. Equivalent input circuit of theType II voltmeter.

Thus we have to make the reactance of K,small compared with that of Ka at highfrequencies and also small compared withR at low frequencies. The first conditioncan be met by making K, say, i,000 timesKa, or about .00rpF, but this value is notlarge enough to satisfy the second condition.Suppose we wish to limit the error on 5ocycles to 0.25 per cent. using a 4-megohmleak, then the minimum value of K, may becalculated from the formula

If, -107_ IA',con.% j2e

in which e is the percentage error assumedto be small. In this case e = .25. As w

2ir X 50 = 314, and R = 4 X 106, thevalue of K1 = If the leak resistanceis reduced to 2 megohms., K, becomes0.022pF, and for a 1 megohm leak K, rises too. 045 ELF.

The final choice of grid -leak resistancevalue will be determined from the fourfollowing considerations : (1) The resistancemust be high to minimise frequency errorwith a reasonable capacity in the grid con-denser ; (2) The D.C. resistance must be


high so as not to add very appreciablyto the power factor of the voltmeter.* Fromthis view -point it should not be less than 2megohms ; (3) It should not be so high thatthe grid takes appreciable time to steadydown to its steady potential. With tomegohms the grid becomes sluggish, and thevalue should be below this figure ; (4) Itshould be as low as possible to prevent errordue to D.C. potentials leaking through thecondenser insulation, as has already beenmentioned above. In view of the last twoconsiderations the leak should lie between2 and 4 megohms.

(c) Wave Form Error.In order to make a direct comparison

with Type I. of the effect of wave form on

TABLE III.

Anode current (nA) withwave forms.

R.M.S.volts.

a b Peaky Flathalf half

making makinggrid + i grid +

Per cent. error.

b

Peaky Flathalf half

o 230 230 230.0 230.00.5 x88 x8o /76.0 200.02.0 150 130 138.0 170.01.5 128 100 107.0 154.02.0 x12 82 93.0 143.52.5 102 72 87.5 136.83.0 96 66 84.0 x32.8

+14+45+75

>xoo*>xoo"

xoo*

+ 24 -22+ 30 -30+47 -37+70 -43>xoo* -49>100* -54

3.54.04.5

929089

* Outside the range of the sinewave calibration.

the calibration of the Type II. voltmeter,the leaky -grid type was tested on exactlythe same wave forms and over the samerange as described above in connection withType I. The results are collected in TableIII., the letters (a), (b) and (c) having thesame meaning as before. These calibrationsare shown graphically in Fig. 20.

The errors are enormously greater in theType II. voltmeter than in Type I. Infact there is only one single observation inTable III. with an error less than 20 per

* This must be distinguished from the A.C.resistance, associated with the self -capacity of theleak, which is in parallel with the D.C. resistance.The former resistance causes most of the losses onhigh frequencies unless the D.C. resistance is quitelow.

THE WIRELESS ENGINEER 597

cent. Some of the errors are too great to beevaluated on the sine wave calibration. Forexample, with 3 volts R.M.S. the anodecurrent with the third harmonic wave b

250

200

150

100

50

a

(FLAT1

EAKY)

0 1.0 20 30R.M.S. VOLTS

Fig. 2o. Calibration curves of the Type II voltmeteron wave forms (A), (B), and (c).

is 66. The sine wave calibration cannotusefully be continued down to this anodecurrent as it flattens out in the region of4 volts and 9o/LA.

In the above test the frequency was 82cycles, the leak resistance 3 megohms, andthe grid condenser o.3µF.

(d) Input Load.The power factor of this type of voltmeter

is very considerably higher than that of the

40

TABLE IV.

Conditions. Capacity.(.41/F)

Powerfactor.

Relativepower lost.

Grid -leak D 1.0 0.19 12.0Grid -leak E 2.0 0.52 68. oGrid -leak F 0.4* 0.42 11.0Grid -leak G 1.2 0.12 9.oVoltmeterwith leak D. 9.o o.o66 39.0

Voltmeterwith leak E to 0.13 85.o

Voltmeterwith leak F 8.4 0.066 36.0

* Estimated value.

October, 1926

first type. This is due partly to the gridcurrent necessary for the action of thevoltmeter, and partly to losses in the leakresistance. Some typical results are givenin Table IV. These results were obtainedwith the same valve mounted on the samepanel and with the same input conditionsas for the results shown in Table II. sothat a direct comparison of the two setsof results may be made. The power lost(Table IV.) is relative to A (Table II.). Theresults are shown for the same voltmeter,using three different leaks (connected to apotential of + 2 volts), and the results forthe leaks alone are given separately.* Theresistance element in the case of leak D wasa plastic compound, the D.C. resistancebeing 4 megohms. Readjustment of thisresistance to about 10 megohms did notaffect the A.C. losses appreciably. LeakE was a piece of some ebonite substitutesold for use as panels in wireless sets.The D.C. resistance was 4 megohms. LeakF was made from sodium silicate which

4.0

ILIu

2 3.0

0cc

X

3 2.0

Ui

ccIY

0 1.0Otr

00

GRID OURRENT

\ANODE CURRENT

1 2 3 4 5 8 7

A.C. VOLTS (R.M.8)

Fig. 21. Effect of input voltage on grid currentin the Type II voltmeter.

170

180 coI

150

140 0cc0

130 2

120z

110

100 0

I -zI!

90 t

80

70

I

was contained in an ebonite tube carryinga terminal at each end. The D.C. resistance

* If the leak behaved on A.C. as a pure ohmicresistance of 4 megohms, the power lost in it wouldhave been represented by the low figure of 2.7.

October, 1926

was 3 megohms. Leak G was a pencil line onebonite. The D.C. resistance was 3 megohms.The power factor of the voltmeter appearedto be practically independent of the inputvoltage. Any change due to change ingrid current was probably swamped by thelarge constant losses in the circuit. Themanner in which the grid current varies with

a) 100cca.Q.

0cc02z

et 80

8aZ 40et

14

a

b

a

15 18 17 18 19

ANODE VOLTS

20

06

0.5

04

21

Fig. 22. General effect of a change in anode voltageon the calibration of the Type II voltmeter.

the input voltage is shown by the straightline in Fig. 21. The curve shows the anodecurrent calibration obtained at the sametime.

(e) Stability of Calibration.The calibration of this type of voltmeter

may alter due to a change in (1) the valveitself, (2) the anode voltage, (3) the resistanceof the grid -leak, (4) the potential appliedto the end of the leak when this is not con-nected to L.T.

Regarding (I), we have no definite ex-perimental results, but it appears probablethat internal changes are practically com-pensated for by the initial adjustment ofthe filament current to give a definite falsezero of anode current. The general effectof change in anode voltage is shown in Fig.22. These results were obtained by adjust-ing the filament current, for each value of


the H.T., until the same anode current(I2oµA) was obtained with no A.C. volts.The anode current with A.C. volts on thegrid was then noted. It is the relationbetween this anode current and the H.T.which is plotted in Fig. 22, for the followingA.C. inputs (a) 0.83 volts, (b) 1.64 volts and(c) 2.44 volts. The dotted curve shows thecorresponding variation of filament current.The valve was of the 235 -type. A changein H.T. of 1 per cent., causes an error ofabout i per cent. in the reading at 1 voltinput, under the conditions of this particulartest.

The effect on the calibration due to achange in leak resistance varies considerablywith the type of valve and the conditionsunder which it is operated. A typical setof results is shown in Fig. 23, from whichit will be seen that with leaks less than 4megohms the calibration changed much morerapidly than it did with leaks greater than4 megohms. The calibration in the regionof 2 volts R.M.S. changed 1 per cent. whenthe leak changed 1.3 per cent. when set at1.5 megohms, and 6 per cent. when set atnot less than 4 megohms. In the test to

400

Oz

W cc

13 3 100

0

<Z

X

ce 200

0o2 4 8 8 10 12

LEAK RESISTANCEIN MEOCHIAS

Fig. 23. General effect of a change in leak resistanceon the calibration of the Type II voltmeter.

which these figures refer the anode currentwith no input was maintained at a fixedvalue, as the leak was varied, by adjustmentof the filament current. The A.C. inputwas kept constant at 2.19 volts R.M.S.

If the leak is connected to L.T. H- theremay be a further change in calibration asthe accumulators run down.

(To be continued).


A High -Tension Rectifier for a Low -PowerTransmitter.

By T. S. Skeet.

THE apparatus about to be describedworks from 200V, 5o cycle A.C., anddelivers smooth D.C. at approximately

62oV and any current up to 6o milliamps.Other voltage and current loads may be

obtained as required ; the output at highvoltage being limited by the flash -over pointof the commutator and by the voltage whichcan safely be applied to the smoothingcondensers.

The apparatus consists of (a) A VaVhouse lighting transformer suitably re-wound ; (b) A small iooV A.C. motor

STARTINGRESISTANCE

[R355.51

The circuit arrangement of the outfit is asshown in the diagram, and the photographsgive a general idea of the equipment.

The motor, which is a two -pole machinewound for iooV (A.C.) was first fitted withring oiler bearings, the particular form ofhollow pedestal casting being very suitablefor this alteration.

A light slip ring made from the shadecarrier ring of a lamp -holder was next care-fully forced over the armature, empire clothbeing used for insulation. The ring wasconnected to one commutator segment and

Circuit arrangement of the H.T. rectifier described in this article.

converted to a synchronous motor and fittedwith separate rectifier commutator ; (c)Starting switches ; and (d) A smoothing unit.

The transformer and smoothing unit wereused for a considerable time with chemicalrectifiers, but a device whose voltage outputwas less dependent on load current was.desired, and the synchronous rotary rectifierwas evolved.

the diametrically opposite segment wasconnected to the shaft. To ensure that thecurrents shall not flow through the bearings,a steel ball was placed in the shaft centrehole, and a piece of brass rod with springtension is pressed on to the ball, the frameconnection being made to this spring contact.The other end of the shaft has a thrustbearing, so that the armature is not displaced

October, 1926 600 EXPERIMENTAL WIRELESS &

by this end tension. The various experi-ments with the machine occupied severalmonths of spare time, and it may be advisableto describe briefly the ideas which weretried, and the mistakes which were made.

The field was first rewound for separateexcitation from 2V, the field input beingabout ijr watts. This necessitated D.C.being used for running up to synchronousspeed ; the D.C. was obtained from the2ooV A.C. by means of a full -wave 8 -cellaluminium -lead rectifier, with a solution ofborax as the electrolyte. The curious beat

This alteration necessitated a re -arrange-ment of the switching system, and theplunger switch which had originally only sixcontacts now has nine.

The new system also necessitated D.C.for starting, and as the chemical rectifierwas not available at this period, the currentwas obtained from the receiving H.T.battery (accumulators, home-made, of ap-proximately o.i A.H. capacity), the machinewas found to synchronise quite as readily aswhen separately excited.

The necessity for starting up from the very

The complete apparatus.

note given out by the machine indicatesapproximately when synchronous speed isreached, and at this instant the plungerswitch is operated which applies the iooVA.C. to the slip ring and frame connection.After a little practice it is quite easy to makethe machine synchronise. This arrangementwas quite satisfactory in practice, but theadditional accumulator and bulky chemicalrectifier were objectionable. The next move,therefore, was to eliminate the accumulator.Once the machine was started, the brusheson the commutator were idle, and as a D.C.voltage of approximately 13o could beobtained from them, the field was rewoundto absorb about 2 watts at this voltage.

small H.T. battery was the next objection,and efforts were made to start up from A.C.A separate winding of comparatively fewturns of 38 gauge wire (D.s.c.) was put onthe field, and attempts made to start up asan A.C. motor. Great difficulty was experi-enced in getting the motor to run up tospeed, and the first starting winding wasburnt out in the attempt. The reason forthe difficulty was not obvious and occupiedconsiderable time in finding.

The machine has laminated fields and avery small air gap, and is therefore a fairlyefficient transformer. The running windinghas sufficient turns to give an induced E.M.F.of 3,000V when 200V are applied to the

THE WIRELESS ENGINEER 6oz

armature and starting windings. The endsof this running winding were connected toa switch contact on the slate base and aterminal mounted on a fibre strip on themachine frame, and whilst this insulationwas quite O.K. for 13oV D.C., it was uselessfor 3,000V. The effect of the leakage currentwas practically to short circuit the runningwinding, and therefore to reduce the fieldflux to a very small amount. Increase ofstarting current merely increased the leak-age and resulted, as before stated, in a burnt -out starting winding. After the insulationwas improved the motor started up andsynchronised without further trouble. Itwas thought that all trouble was now over,and that it would merely be necessary tofit a commutator and slip rings to the shaftto enable loads at any voltage to be takenfrom the rectifier.

An account of the further failures mayprevent other experimenters holding thesame absurd ideas. A four-part commutatorwas first constructed, the live segments beingabout two-thirds the length of the deadsegments. Gauze brushes were used for theslip rings, and carbon brushes in carefullymade boxes for the commutator. Thisarrangement was quite satisfactory foraccumulator charging, particularly whengraphite brushes were used, but was hopelessfor high-tension rectification, anything above250V causing violent flash -over, with con-sequent shorting of the transformer. Gauzebrushes were next tried, and whilst matters im-proved, results were anything but satisfactory.

A 4 -part mica insulated commutator,with three live segments connected to eachslip ring was next tried, and as this gave fiveseparate and somewhat thick mica sectionsbetween the live portions, considerablygreater success was obtained. Foliatedcopper brushes were next substituted andthese seem far superior to anything yettried. It is presumed that a commutator,with live segments let into an insulating

or cylinder, would be ideal, but such anarrangement would not be particularlysound mechanically, and would be subjectto rapid wear. The present arrangement hasbeen tested with a smoothed output of 7ooV3o mA and 53oV 70 mA. Larger output at7ooV was not tried on account of lack ofsuitable artificial load, and an output of800V gave occasional flash over.

October, 1926

As the input from the mains to the trans-former never exceeds loo watts, a o.5kNVtransformer is altogether out of proportion,the transformer was, however, obtained froma lighting contractor for 7s. 6d., and had thecore size been reduced, a greater length ofwire would have been needed, and the costof the wire was of course much greater thanthe cost of the transformer. The trans-former when obtained was " auto " wound,and as the wire occupied the whole of thewinding space, it was necessary to remove

The rewound transformer.

the original zooV winding and rewind withsmaller gauge wire to make room for thehigh voltage secondary. The transformer inquestion is of the " core " type, the core beingcomposed of rectangular plates. In takingthe transformer apart, only the end platesshould be removed, the portions inside thecoils being carefully preserved in theirstaggered formation by tying with tape,this will save considerable time duringreassembling.

It has been observed that 3 turns per voltis a standard winding for 5o cycle trans-formers having iron of 3 sq. in. cross section,so that should the core be reduced the turns


must be increased proportionately, and alittle consideration will show that the actuallength of wire required will be greater thanfor the larger core. The transformer underdiscussion was rewound with 300 turns of22 gauge D.C.C. on each limb, and thewinding taped and varnished. A rectangularformer was now constructed approximately

in. long, and the same size as the outsideof the primary winding, the removable metalcheeks were slotted to take four threads fortying up the coils preparatory to taping andto facilitate the removal of the coils from thetransformer. Twelve coils were wound, eachconsisting of 34o turns of 35 gauge D.s.c.wire, and were thoroughly taped with adhe-sive tape. The coils were not varnished, astheir flexibility assists in the placing ofthem in position over the primary. Sixcoils were placed on each limb and con-nected in series, a tapping being taken offat every connection and brought out to one ofthe 13 terminals mounted on the transformer.

The transformer was required to give amaximum voltage of 1,200, the extra 40turns for coil being included to compensatefor voltage drop on load. The centre con-nection of the zooV winding is brought outto provide the necessary IooV for runningthe synchronous motor.

Whilst the set is running on load, with thefield at full strength, the voltmeter and milli -ammeter needles have a beat swing in stepwith the beat note given out by the machine.This beat effect is believed to be due to somefluctuation at the generating station, asthe same note can be heard in a telephonereceiver by induction from the mains. Thisbeat trouble can be cured by mechanicallyloading the motor, i.e., by applying frictionto the shaft, but as this is very undesirable,the adjustable 7,000 ohm resistance shownin series with the running winding is varieduntil the beat ceases and the output issteady. The actual beat is caused by thearmature changing its position periodicallywith respect to the phase of the supply ;this fact was proved by inspecting therunning machine by the light of a Neonlamp (" Osglim " type) lighted from thesame supply main. The armature, when soinspected, appeared to be merely rockingbackwards and forwards at the beat fre-quency, but the writer is unable to explainthe reason for such behaviour.

The change -over switch shown on thesmoothing unit is for the purpose of reversingthe connections to the output circuit, thischange over being often necessary owing tothe fact that the direction of the D.C.depends on chance, i.e., upon the mannerin which the motor synchronises.

The smoothing unit employs Mansbridgezi.LF paper condensers, eight of which werepurchased for 12s. (Ex. Govt.). Thesecondensers, as purchased, are useless forany purpose, except as variable (very)grid -leaks, and special treatment is requiredbefore they can be used as smoothers. Theeight condensers referred to were placedupright in a large iron vessel and moltenwax poured in, until the condensers werecovered. The wax was then maintained ata temperature of 330°F. until all bubblingceased (this took five hours). The waxwas allowed to cool, fresh wax being addedto make up for that drawn into the conden-sers. When quite cold the condensers caneasily be broken out of the wax, particularlyif they are tied together before boiling. Thewriter has treated many faulty condensersin this manner, and finds that they standup to severe treatment far better than newcondensers of similar make, owing, no doubt,to the fact that they are truly hermeticallysealed, provided that they are totallyimmersed in the wax during the coolingprocess. The eight condensers are groupedas shown in the diagram, giving an effectivecapacity of 2pF across each end of thesmoother, with the voltage of the supplydivided between the two capacities in series.The smoothing choke is of original design,and has a closed core of 4 in by 4 in. crosssection.

Two windings are used, each consisting of8 ozs. of 34 gauge enamelled wire. The closedcore is of rectangular form, and one windingis placed on each of two opposite limbs.One coil is connected in each leg of thesupply and smoothing is very effective ;in fact this type of smoothing choke withtwo windings is the only type which thewriter has found satisfactory, either fortransmission or reception H.T. supply. Thefact that the core is closed and of liberalcross section, enables smoothing to be effectedwith comparatively few turns of wire, andconsequently limited voltage drop.

The unit described above has not yet been


used for transmission purposes, but it waslearned that reception of Daventry on aneighbour's crystal set was seriously im-paired whilst the rectifier was running onload in the workshop.

As it is desired to run the machine for testpurposes, and for work with a non -radiatingaerial during broadcast hours, this inter-ference had to be eliminated. The startingand running of the machine did not cause

October, 1926

It was noticed (on a test receiving circuit)that the interference varied with the sparkingof the rectifier, and after considerablerunning of the commutator, previously des-cribed, it was found that the flash -overvoltage was reduced, and with the commu-tator at rest the flash -over occurred at800 (R.M.S.) V, due to the copper dustembedded in the mica. It was probablethat the minute sparks due to this leakage

The synchronous motor (running end).

the trouble, which only occurred whilst therectifier was on load, it was assumed thereforethat shock excitation of some portion of theapparatus was responsible.

A case lined with tinned iron was made toaccommodate the whole of the apparatus,and the joints soldered, except, of course, thedoor joints. The apparatus was enclosedand the lining earthed, the machine was puton load and the interference was considerablyincreased, but with the earth connectionremoved there was a slight improvementover the original arrangement.

were accentuating the shock' excitation, andin spite of previous remarks, it was decidedto construct on ebonite commutator, withbrass segments let into the surface.

This new commutator does not flash -over,and is far more satisfactory than the othertypes, but it is early to judge of its wearingqualities. The result of this latter alterationwas a further slight diminution of theinterference.

From the results obtained by earthconnecting the screen, it was deduced thatthe excitation was being induced into the


lighting system, the outdoor cables for whichrun overhead between the houses, and thenext step was to provide H.F. chokes in thesupply leads.

A coil of approximately 13o turns of 24gauge wire on a 3j< in. diameter former wasincluded in each leg of the supply, and theinterference was eliminated so far as thecrystal set was concerned, but was stillaudible on the writer's valve set, the H.T.for which is obtained from the mains.

A Iit&F condenser was next placed acrossthe mains at each end of the chokes, and thetrouble was completely cured. The energysupplied to the transformer with the secon-dary on open circuit is only 9 watts, so thatthe condensers are causing very little loss.

The excitation should be equal in eitherleg of the supply leads, and therefore thecondensers should not make any difference,but the slight improvement noticed isprobably due to, the fact that one main isearthed at the sub -station transformer, withconsequent inequality of the currents induced.

The casting carrying the brush -rockerdisc, which may be seen in the photographs,may be recognised as the frame of an oldBlake transmitter, of the carbon button andplatinum contact type which saw servicemany years ago in the early days of telephony.

If this article has pointed out some of theerrors to be avoided in the construction of arectifier system, its object will have beenachieved.

A Method of Calibrating Microphones andLoud- Speakers.By H. J. Round, M.I.E.E.

CONSIDERABLE difficulties occur inthe calibration of microphones. Aconstant sound pressure over wide ranges

of frequency is difficult to obtain, andreference is usually made to some theoreticalbasis for the purpose.

I have not seen the following methoddescribed before. It is comparatively simple,and avoids a good many of the usual diffi-culties, and is applicable to almost any typeof microphone which employs a diaphragm.I have recently used it extensively to studythe action of. various microphones. Ingeneral the idea is to apply the necessaryforce to the diaphragm by an electrostaticforce instead of an air force.

The microphone diaphragm, if not metallic,is first coated with aluminium leaf, whichleaf is extended all over the microphone,and well earthed.

If the microphone has a metallic dia-phragm fixed in a metallic case, there willbe no necessity to do this if the case can beearthed. (The W.E. electrostatic microphoneis an example of this.)

Near the diaphragm surface is placed a

[R290

solid flat -faced metallic block, the flat faceof which can be brought to about 2 mm. fromthe diaphragm surface. Various shapes ofblock may be used to check whether any airpocket resonance is present.

If the block be now statically charged to ahigh voltage (from 600 to 1,50o volts I findconvenient), a force is applied to the dia-phragm, and if now superimposed on this

Fig. 1.

voltage an alternating E.M.F. of any fre-quency is applied-the E.M.F. of which canbe measured-the alternating componentof the force produces a current in the micro-phone circuit which can be measured andplotted against the applied force.

Fig. i is an overall representation of theapparatus used.


A is a musical note producer, which willbe described later.

B is a resistance amplifier using a platevoltage of 30o volts to get large undistortedamplitudes.

C is the solid metallic block.M is the microphone, modified if necessary

as described above.D is the microphone amplifier.G is a polarising battery.E and F are A.C. voltmeters of the slide

back type placed where indicated, and thecurve is plotted between their readings.

October, 1926

tone, magnified up to the value desired, is ofalmost constant amplitude, and, in fact forrough experiments, can be assumed constant.

Calibration of note frequency is done onthe condenser of the second oscillator andmust be done frequently by means of one ortwo tuning forks. Fortunately the fre-quency scale is nearly linear.

The resulting beat tone is magnified upby means of a carefully constructed resistanceamplifier. Harmonics are, of course, intro-duced, and for very accurate work must beremoved by cut-off filters after the last stage

Fig. 2. Circuit arrangement of the note producer and amplifiers.

This curve can be taken as the curvebetween air pressure at the diaphragm sur-face and output open circuit volts. Absolutevalues can, of course, be obtained.

It may be of interest here to give a moredetailed description of the note producerand amplifiers (Fig. 2).

A weak 3,000 meter fixed oscillator 0,fitted in a shielding box, is fed through oneloosely -coupled circuit-to remove harmonics-to a valve V, and thence induces into arectifier R.

Another more powerful oscillator P, thefrequency of which is Variable and withouta coupled circuit attached also induces intothe circuit attached to the rectifier valve R.

The resulting beat tone in the plate circuitof the rectifier R is fairly pure and free fromharmonics. A filter circuit F, to remove allH.F., is then inserted, and the purified beat

of amplification, but for a lot of work theycan be neglected.

The above method for producing notes ofconstant amplitude was suggested to me byMr. Norman Lea, and the arrangement ofapparatus, including the addition of a coupledcircuit for one of the H.F. oscillators, wasmade by my assistant Mr. N. M. Rust.

The idea of this coupled circuit is veryinteresting. If two oscillators are beatingtogether and being rectified, and both produceharmonics the resulting beat tone will havesimilar harmonics, but if only one set ofharmonics is removed the result is afreedom from harmonics in the beat tone.

A modification of the test with which I amnow experimenting, is to balance the inputand output in a bridge system which willgive the phase angle as well as the amplitudecurve.

I.-Previous History.THE technics of wireless telegraphy

throughout the whole world haveexperienced a great surprise during

the past two years as the result of theunanticipated effectiveness of short wavesfor Transoceanic traffic. In this connection,waves of about ioo metres and less are tobe understood as short waves, i.e. electricaloscillations of frequencies of about threemillion cycles per second and over, whichhave produced quite extraordinary resultsover great distances.

Although wireless telegraphy can nowlook back upon a lifetime of 25 years, itmust nevertheless, as comparedindustries and branches of technology, beaccounted a very young industry, in whichdevelopment has proceeded with extra-ordinary rapidity and constant change,but one which has in no way reached astate of equilibrium. There does not yetexist any generally recognised best systemof transmission, and in reception also thereare still countless variations, even although,in the fundamental principles of wirelesstelegraphy, in the electro-dynamic actionin the apparatus itself, as well as in theprocess of radiation and reception withaerials, there are hardly any differences ofopinion worthy of mention. The explana-tion of the propagation of the waves overour earth has, however, always proved anextremely difficult matter. This propa-gation has been extensively dealt with inexhaustive theoretical and practical investi-gations from the very first years of wirelesstelegraphy (J. Zenneck, A. Sommerfeld,H. Poincare, J. W. Nicholson, L. W.

1Translation of a paper read at the meeting ofthe Deutsche Gesellschaft fiir Technische Physik,Danzig, in September, 1925, published in theTelefunken Zeitung, January, 1926.

October, 1926 6o6 EXPERIMENTAL WIRELESS &

Recent Developments in Short -Wave WirelessTelegraphy.'

By H. Rukop(Telefunken Company).

Austin, V. Rybczinski, L. F. Fuller, W. H.Eccles2), and there is no doubt that the trueconditions have been depicted with approxi-mate correctness by the formulae of therespective authors. The formulae ascer-tained by methods of calculation relatingto the propagation along the surface of theearth take account of numerous factors :the properties of the waves themselves, theconductivity and dielectric properties ofthe earth's surface, the divergence of theradiation away or " Dispersion " from theearth in consequence of the spherical formof the latter, but they do not contain at theoutset any terms which take account of theabsorption, reflection or refraction on theearth itself or in the atmosphere. Now, ifit is desired to express the intensity of theelectrical field prevailing at a certain dis-tance from a known transmitter and pro-ceeding from its radiation, one uses forthe purpose preferably the Austin - Cohenformula :

[R401

'fh 8 o ,ooi5d microvolts

E= 12077-8 S e /-Ad sin d v A per metre,

which, as compared with the calculatedformula of the above -mentioned authors,contains a small correction ascertained bymeasurements. In this formula, J, signi-fies the intensity of current in the trans-mitting aerial, h, its effective height ofradiation, A the transmitted wavelength,d the distance of the place of reception fromthe transmitter measured along the greatestcircle of the sphere, all in km., S the anglesubtended at the centre of the earth by thetwo stations, and E the electrical field

2For publications on this subject see : J.Zenneck and H. Rukop, Lehrbuch der drahtlosenTelegraphie, 5th Ed., p. 294 ; A. Hund, Hoch-frequenz-Messtechnik, p. 204 and following ;Zenneck, E.u.M., 43, 1925, p. 593 ; G. J. Dias,E.N.T., 2, 1925, p. 351.


strength at the place of reception, in micro-volts per metre. If, now, it is desired toestimate the dependency of the receivingfield strength upon the wavelength for twogiven points, special conditions regardingthe transmitter must also be added, forexample, that with a given aerial a certainnumber of kilowatts are available, or thatthe aerial is being utilised to the limit ofpermissible voltage. If one combines suchconditions with the Austin -Cohen formula,a function is obtained which has approxi-mately the form of curve i in Fig. 1.

100.000 10,000 1,000

X METRESFig. 1. Received field strength on various wave-

lengths at a distance of about 6,000 kilometres.

100 10

Now, this function is generally found tobe confirmed in practice with approximatecorrectness. One finds in fact an optimumwith certain medium wavelengths, which asa rule are about the 500th part of thedistance between the two stations, that isto say, at a distance of about 6,000 km.(Nauen-New York) a wave of 12,000 metreshas been found to be very satisfactory.If much longer waves are used, it is foundthat the strength of reception drops veryconsiderably in consequence of reducedradiation. It is likewise found that whenmuch shorter waves are used, there is alsoa greater diminution in the intensity, not-withstanding the increased radiation, thisbeing due, as is known, to failure to bendround the globe (dispersion) on the onehand, and to absorption on the other hand.In fact wavelengths below about i,000metres are found to be very unsuitable forTransoceanic communications, for whichreason, during the last decade, one has beenenabled safely to conclude that curveshown in Fig. i reproduces the true con-ditions with absolute correctness, and notechnician who cared about his scientific ortechnical reputation would have ventured,in connection with ranges of io,000 km. orsimilar ones, to speak of wavelengths farbelow ioo metres. And this is preciselywhat has now become the accomplished fact !

October, 1926

The discovery of the extraordinary effec-tiveness of these short waves was madeaccidentally, namely, by the fact that inthe United States of North America therewere allotted to amateurs, who also wishedto devote their attention to transmitters,regulation wavelengths under 125 metres,regarding which it was assumed that theywould not cause further interference toanybody. It was very soon found, however,that such short-wave transmissions werereceived at quite considerable distances,in Europe for example, whereupon theattention of all wireless technicians wasimmediately focused upon this remark-able phenomenon.

The question will now at once be askedhow it was that so important a fact couldremain concealed for so long. The ex-planation of this can easily be deduced fromthe history of wireless telegraphy.

There can be no doubt at all that thenewly found good 'results with short wavesare to be attributed solely to the era of thevalve transmitter and valve receiver, since,in principle, transmission with so shortwaves is in no way a novelty. It shouldbe recalled that in the early days of wirelesstelegraphy, Righi oscillators were used,and, as a general rule, these were of suchdimensions that they were only able to givewavelengths of a few metres at the most.At least, it was the intention at that periodto work with such oscillators and suchwavelengths, even if, in fact, much longerwavelengths resulted when the aerial wascoupled direct to the Righi oscillators.

There is no doubt that at that time thetechnical means available, and especiallythe receivers, were too inadequate for suchresults to be even thought of, as can nowbe witnessed with short waves.

A few years later, during the era of themusical note transmitter, such waves couldperhaps have been conceived, from thestandpoint of the transmitter. But eventhen no results would have been obtainableas regards reception, since there was not afully elaborated receiving and amplificationtechnique as would have been necessaryfor the purpose. The Poulsen arc andmachine transmitters were quite unsuitablefor the production of such short waves.

It was only after the elaboration of trans-mitting valves, when the production of

October, 1926 6o8 EXPERIMENTAL WIRELESS &

undamped oscillations for wavelengths downto about i metre was made comparativelyeasy. that trials with short waves wereagain made. In fact, it was not until about1916 that the technical auxiliaries, such astransmitting valves of suitable output,heterodynes, and all the necessary receivingand amplification circuits were developedfar enough to lead to the discovery andutilisation of the singular qualities of shortwaves. But meanwhile theory and experi-ence had apparently shown unanimouslythat propagation over the earth is subjectto the laws represented by Fig. 1, curve 1.As however this representation is absolutelycorrect, even now, down to about A 15ometres, and as not even a trace of deviationor exception has ever been found, it waslogically held to be correct for still shorterwaves, and it would have been a waste oftime to go still lower with the wavelength.Only for special purposes, such as directionaltransmission with reflectors at shorter dis-tances, did the short waves seem likely tobe successful and attempts were in factoften made to use them.

.Special mention should here be made ofthe series of experiments on a large scalethat were carried out by C. S. Franklin andthe Marconi Company, about the year 1919and succeeding years.3 Nevertheless, owingpresumably to the lack of other receivingobservation stations for these wavelengths,these experiments did not reveal the enor-mous importance of short waves, althoughall the requisite conditions were present inthe transmitting apparatus of C. S. Franklin.Even the Marconi Company's receivingstations, at the time when the fabulousranges of short waves had been discoveredelsewhere, had not yielded correspondingresults, so that the fame for this discoverywas lost to it in consequence of the results thathad been secured accidentally by amateurs.

Therefore, when the first signs of theextraordinary ranges of short waves hadbeen justified by observations in severalquarters, interested circles directed all theirenergies to the elaboration of this importantnew branch of high -frequency technics.

The " Telefunken-Gesellschaft fur draht-lose Telegraphie," which together with the

" Transradio - Gesellschaft fur drahtlosenUberseeverkehr " is interested in the trans-mission of messages by wireless telegraphyover the greatest ranges, proceeded imme-diately with the study of these problemsand, by building and putting into operationseveral valve transmitters for short wavesbetween 100 and Ico metres, for communi-cations between Nauen and far -distantreceiving stations, such as Buenos Aires(12,000 km.), Bandoeng, Java (ii,000 km.),Osaka, Japan (9,000 km.), has achievedimportant results and progress, and acquiredextremely valuable experience. Further,in several other quarters connected withthe high -frequency industry, as well as inthe State laboratories of various countries,important results have been obtained andpublished. Furthermore, amateurs, andespecially those of the United States of NorthAmerica, have obtained important data onthe propagation of short waves, and theeffect of such variable factors, as the timeof the day, time of the year, wavelength,etc. One can therefore form a provisionalpicture of the strange phenomena of short-wave wireless telegraphy.

Before I proceed to the chief theme ofthe present paper, namely, the results whichthe Telefunken Company in co-operationwith the Transradio Company has obtainedin the attempt to bridge long distances bywireless, it is necessary, for a better com-prehension of the subject, to explain thosephysical phenomena which form the basisof the action of short waves, as contrastedwith long waves, as far as they appear tobe known at present.

11.-Method of Propagation..

The propagation and transmission ofshort waves, as compared with long waves,shows some striking differences, which havealready been partly described by A. Meissner4and A. Esau.3 In tfhe case of shortwaves, the ollowing properties are especiallynoticeable : Fading, i.e., sudden strongfluctuations of intensity, as, for example,are represented in Fig. tz ; the greatdifferences between the day and nightsignal strength ; the existence of suitable

4 A. Meissner, Zeitschrift j. Tech. Phys., 5, 1924, 485.drahtlose Tel., 21, 1923, 58. 5A. Esau, Zeitschrift j. Tech. Phys., 5, 1924, 538.

3See, for example, under G. Marconi, Jahrb.

THE WIRELESS ENGINEER 6o9 October, 1926

wavelengths in the immediate neighbour-hood of which very unfavourable wave-lengths may lie ; the variation of thefavourable wavelengths with the time of theday and year, and finally the so-called deadzones. Before I go into details, it mustfirst of all be shown how the earlier laws(curve 1, Fig. 1) have been modified by thenew discoveries.

According to the results obtained up tothe present, there is no doubt that, inaddition to the range of wavelengths ofCurve i in Fig. i, there exists a furtherfavourable range, somewhat as shown incurve 2, Fig. i, of wavelengths of aboutzoo metres and below, but that betweenthese two ranges there is a range which isuseless for Transoceanic distances. Thatcurve 2 is not given by the above -mentioned formulae is due to the fact thatthe latter are only calculated for a directtransmission along the earth's surface, andthat Austin's experimental results wereonly obtained on longer waves, whilst theeffect of the short waves-as can now bemaintained with the greatest certainty-is an indirect one. By this it should beunderstood that the short waves travel,on the one hand, along the surface of theearth, but that the energy which is measuredat distances of several thousand kilometresdoes not proceed frbm a propagation alongthe earth's surface, but has returned fromthe higher layers of the atmosphere. Thereasons of this assumption are justified byvarious experimental results of the pastyear.

One of the most important reasons whichcan now be adduced for this is the so-called" dead zone " which occurs in certain caseswhich are more fully described below. Thestrength of reception, in fact, has often aremarkable distribution around a trans-mitter, in that, first of all (see Fig. 2(a)) thereis around the transmitting aerial (S) a zoneof considerable intensity (A). Beyond thiszone is found a second one (B) in which thereceiving strength is exceedingly small,whilst at a certain distance a zone of in-creased receiving strength (C) is again per-ceptible, and this may be of enormousextent. The simplest explanation of this,then, is that zone A is that of direct trans-mission of surface waves, which the Austin -Cohen formula represents, and that these

surface waves at the beginning of zone Bhave so diminished in their intensity byabsorption and reflection that they can nolonger be detected. A considerable amountof radiation energy has, however, directedits course obliquely upwards and, in conse-quence of physical conditions to be explainedlater, returns again to the earth, where itbecomes perceptible in zone C.

As regards the existence of the deadzones, exhaustive statistical investigationshave been reported by J. L. Reinartz 6,and also by A. H. Taylor and E. 0. Hulbert.?Exact data regarding this are given in thelatter paper, from which it appears that deadzones' do not occur generally in wavelengthsabove 9 metres, that in the neighbourhoodof 40 metres they may first of all assumebreadths of about 100-200 km. and thattheir breadth down to a wavelength of 15metres increases extraordinarily and mayreach far beyond i,000 km.

A

(a)

A

(b)

Fig. 2. Distribution of the receiving intensity arounda short-wave transmitter.

The direct receiving zones A, Fig. 2(a).which are obtained with various wave-lengths, may, depending on the conditions,vary in width from a few kilometres (occa-sionally estimated at less than 10 km.), toioo or 200 km., and the direct zone is infact probably reproduced with sufficientcorrectness by the original Austin -Cohenformula.

But even where there is no dead zone,there is direct as well as indirect trans-mission. This means that with the longerwaves of range 2 (approximately 6o to ioometres) the direct range is so large that itoverlaps the zone of the indirect range.

6J. L. Reinartz, Q.S.T., 9, 1925, No. 4, p. 9.7 A. H. Taylor and E. 0. Hulbert, Q.S.T., 9, 1925,

No. to, p. 12 ; A. H. Taylor, Proc. I.R.E., 13,1925, 677.

October, 1926

There then results, instead of the distributionshown in Fig. 2(a), a distribution as shownin Fig. 2(b), and the signal strength in areceiver, at increasing distances from thetransmitter would, in the case of the distri-bution shown in Fig. 2(a), be represented bycurve a in Fig. 3, whereas in the case ofthe distribution shown in Fig. 2(b) it wouldbe given by curve b. The former (curve a),would, therefore, be expected with wave-lengths of less than 40 metres, the latter(curve b) with waves over 6o metres. Itwould appear that the fading phenomenaare especially great at those places wherethe direct range overlaps the indirect. Onthe other hand, they are in no way limitedto this zone, since even at distances of 10,000km. and over, where indisputably only theindirect rays are concerned, they are alsovery pronounced. A further reason forthe assumption that short waves do not go

>-

U)zw

z0

a

DISTANCE

DISTANCE

Fig. 3. Variation of signal strength with distanceof the receiver from the transmitter.

by a direct route over great distances is thefact that when bearings are taken along theroute, either they are extraordinarily lackingin sharpness or it is completely impossibleto effect them, the received waves being ofa muddled or diffuse character.

A very interesting discovery regardingthe return of waves from the upperatmosphere has been made by E. V.Appletong who found, by means of areceiving set some 150 km. distant fromthe transmitter, that at the respectivepoint of reception there were both directand indirect rays, since interference between

8 E. V. Appleton, Tidj. Ned. Rad. Gen., 2, 1925,


the two could be demonstrated. He like_wise found, on varying the wavelength,that there were very close maxima andminima of reception, from which he wasable to deduce that there was a differenceof path between the two rays amountingto about a hundred wavelengths. Fromthe distance of the station and the differentlength of path (about 8o km.), he was ableto conclude that the indirect ray returnedfrom a height of 8o-go km., a height atwhich, from our knowledge of meteoro-logical conditions, a considerable conduc-tivity is extremely probable.

If we assume that the radiation returnsfrom the upper layers of the atmosphere,the conditions for the realisation of thisphenomenon must be very complex. Severalexplanations are possible. A simple as-sumption would be that the electrical wavesare totally reflected at the naturally notvery sharp limit between the atmosphereand the space which is either empty or inter-spersed with frozen gas dust. It can,therefore, be easily deduced, from the con-figuration of the atmosphere and the earth,that the rays in order to traverse a routesuch as that from Nauen to Buenos Airesi.e., about ioo degrees, would necessarilyundergo numerous total reflections. It ismanifest that this explanation is not ade-quate, however, since if it were only aquestion of the dielectric properties of theatmosphere, there would be no reason whywaves of from ioo metres up to the longestare not also totally reflected to the sameextent. This would necessitate a disper-sion, which is in nowise inherent in theair.

Instead of a total reflection, on passingfrom the air -space (atmosphere) to the emptyspace, one can secondly assume a reflectionat the much discussed Heaviside layer, i.e.,a layer which, as compared with the airlayers on the earth's surface, is said topossess a considerably increased conduc-tivity in consequence of ionisation. Thesame here applies as 'has been said in the,previous case, regarding the necessary mani-fold reflection.

The possibility both of reflection and ofselective reflection for short waves of onlya definite range, is here present. Thephysical bases of this assertion will also bediscussed below when treating of the fourth


possible explanation. In any case, theexplanation just given is to be countedamong the more probable ones, even ifpreference is now given to the fourthexplanation set forth below.

A third possible explanation is refraction,i.e., such a bending of the rays in the upperlayers of the atmosphere that they followa path parallel to the curvature of the earth,so that when the waves have once reachedcertain heights, they now direct their coursewithout reflections to distant points of theearth, and the energy, less what is loston the way, returns to the earth. Theonly physical prerequisite for such a curvedroute is that the speed of propagation inthe upper air layers must be somewhatgreater than in the lower ones, namely, ofcourse, to such an extent that the curvatureof radiation is approximately the same asthe curvature of the earth. Here also onemight at first assume purely dielectricdifferences. For this, the dielectric con-stant of the lower air layers, in comparisonwith that of the upper ones, would have tobe greater than that actually caused bythe difference density, and one wouldbe compelled to have recourse to the aqueousvapour in the lower layers or somethingsimilar, as has already been suggested bythe above -mentioned authors. But, hereagain we should be in a dilemma in thatwe should have to assume, with a purelydielectric explanation, a dispersion of theair layers, in order to explain the extra-ordinary difference in the propagation ofthe short waves as compared with thelong. This explanation seems thereforeto be as little tenable as that of totalreflection.

A fourth explanation is now chiefly heldto be the correct one, namely, that a curva-ture of the wave rays takes place, thereforenot reflection but refraction, and that thiscurvature is effected by the effectivedielectric constant being diminished in theupper layers by conductivity, and the speedof propagation consequently increased. Theexistence of such a conductivity cannot, ofcourse, be held to be contrary to physicallaws, and it seems entirely justified toassume a " Heaviside Layer " of such con-ductivity, so long as nothing better is knownto explain the peculiar facts. The assump-tion of 0. Heaviside is to the effect that in

611 October, 1926

the upper atmosphere there are found layerswith considerable ionisation and thereforeof considerable conductivity, and that thedivergence of the electrical waves from theearth and their radiation into universalspace is thereby prevented, and the receivingintensity is increased far beyond that givenby the ordinary diffraction formula. Now,whether this conservation of radiation iseffected by reflection or refraction, wouldprobably matter but little as regards thedesignation " Heaviside Layer." It can inany case be maintained that the greatscientist, with intuitive glance, at oncerealised the favourable influence of atmos-pheric conductivity on the propagation ofelectrical waves, even if his explanationshould not be found to be correct to a fullzoo per cent.

The assumption of a radiation curvaturedue to diminution of the dielectric constantof the upper layers, in consequence of con-ductivity, is not only capable of explainingthe extraordinarily favourable ranges andthe conservation of the radiation energy,but also the difference in the propagation ofshort waves, as compared with the longones. Even before the discovery of theshort-wave phenomenon, it had alreadybeen pointed out from many quarters that,if the effective dielectric constant of the airis lessened by ionisation or conductivity,it must become noticeable in wireless tele-graphy (H. W. Eccles, J. Salpeter, B. vander Pol, etc.).9 The assumption which hasto be made regarding the density of ionisa-tion in the upper layers, in order to obtaina suitable variation of the dielectric con-stant by conductivity, is exceedingly pro-bable and agrees with what is believedregarding the ionisation produced in theupper atmosphere by ultra -violet light,Cathode rays, a rays, penetrating radiation,etc. It may, however, be further statedthat such a diminution of the dielectricconstant without considerable losses in therespective layers, can only take place ifthe ionised particles are able to followsufficiently freely the electric field of theoscillations. An examination of these

9H. W. Eccles, Proc. Roy. Soc. A., 67, 1912,p. 79 ; Electrician, 79, 1912, 1015 ; Jahrb.Drahtl. Tel., 8, 1914, 253, 282. J. Salpeter, Phys.Zietschr., 14, 1913, 201 ; Jahrb. Drahtl. Tel., 8.1914 ; B. van der Pol. Dissertation. 1920, Utrecht.


conditions has recently been made by J.Larmor,1° who has shown that in this directionthe results are quite satisfactory. As willeasily be seen, the dielectric constant in theionised air is in itself dependent upon thefrequency, since the phase of the electronsin the electric field is dependent upon thefrequency. According to this, the dielectricconstant for the shortest waves ought toshow the smallest decrease, and for longerwaves a greater decrease, i.e., the radiationof the shortest waves will be less curvedthan those of the longer waves. How farthis is in accordance with the facts is notyet exactly known, but in any case A. H.Taylor and E. 0. Hulbert7 have maintainedthat with wavelengths shorter than io to15 metres a return of the radiation from theupper atmosphere hardly takes place. In

10 J. Larmor, Phil. Mag., 48, 1924, 1025. Seealso Jahrb. Drahtl. Tel., 25, 1925, 141.

the case of waves over 15 metres one wouldobtain, according to knowledge gained hither-to, a sufficient curvature by the conductivityin order to be able to explain the great ranges.Furthermore, in the direction of longerwaves, there is found a limit which, as isshown by Fig. I, lies in the neighbourhoodof a wavelength of about 125 metres, afterwhich favourable propagation decreases.This is to be explained by the fact that,from these frequencies downwards, the ionsand electrons respectively, in the timebetween two collisions with gas molecules,no longer have time to follow these relativelyslow oscillations in such a way that a curva-ture of the ray can take place without loss,i.e., for wavelengths beyond a certain valuethe energy not only returns towards theearth too early in consequence of greatercurvature, but also is consumed by thecollisions of the ions, so that large rangescannot be obtained.

(To be continued).

Book Review.THE B.E.S.A. GLOSSARY OF TERMS USED IN ELEC-

TRICAL ENGINEERING.-Published by CrosbyLockwood & Son for the British EngineeringStandards Association, pp. 263. Price 5s.

This Glossary was prepared by a Sectional Com-mittee, a sub -Committee, and a number of panels ;it was finally adopted by the Sectional ElectricalCommittee, which is the British National Com-mittee of the International Electrotechnical Com-mission in January of the present year. Theobjects in view are stated to have been (a) tostandardise and co-ordinate the electrotechnicalterms used in the British Empire, and (b) to p o -vide a basis for the British portion of an Inter-national Vocabulary, in course of preparation bythe International Electrotechnical Commission.It is divided into eleven sections, two appendices,and an alphabetical index. Each section is sub-divided into a number of sub -sections, for example :Section X., Radio Communication, is sub -dividedinto seven sub -sections under the following headings :

Ether and Ether Waves ; Aerials and Aerial Con-struction ; Transmission ; Reception ; Valve Con-struction and Properties ; Circuits and theirProperties ; Amplifiers and Relays. Appendix I.gives a table of symbols for quantities and units ;Appendix II. classifies the contents in the orderof the international decimal classification. Anattempt has been made to standardise as far aspossible the termination " or " for pieces of appa-ratus or machines for accomplishing a certainpurpose, leaving the termination " er " for theperson who carries out an operation, but the schemecould not be carried right through. The Com-mittee adopted " divertor " and even " startor "but drew the line at such things as " milkor,"" feedor " and " voltmetor."

The most unsatisfactory sections in our opinionare those dealing with fundamental electrostaticand magnetic terms. These appear to be veryconfused and we cannot understand how theyreached the stage of publication. G. W. 0. H.


Mathematics for Wireless Amateurs.By F. M. Colebrook, B.Sc., A.C.G.I., D.I.C. [510

(Continued from page 563 of September issue.)

5. Logarithms.(A) Preliminary Note on Functions, and their

Graphical Representation.NY algebraic letter symbol can beconsidered to stand for any positive ornegative number or fraction. Starting

with some such symbol, x for instance,other numbers can be built up the magni-tude of which will depend in some specifiedway on the magnitude of x. The number3x-4 is a simple instance of such a built-up number, and, being a number, it canbe represented for convenience by someother single symbol, such as y. This numbery is then defined by the statement (orequation)

= + 4In such a connection the number x is termedan " independent variable," the idea beingthat it is at liberty to wander at its ownsweet will over the whole range of magnitude.

...-- x --... p

i

Y

0 X

Fig 7.

The number y on the other hand has no willof its own and has to go where x tells it.This is expressed rather grandiloquently bythe phrase " y is a function of x." Inmathematical shorthand this is written

Y =i(x)Notice that the letter symbol f does not

in this instance represent a number, butexpresses a functional relationship betweeny and whatever symbol is written insidethe brackets. Different functional relationships can of course be represented by otherletters, e.g., F(x) or 0(x). Once a particularfunctional relationship has been specifiedthe same letter should be used throughoutany piece of work for that function.

The dependence of y upon x can beemphasised by assigning various values to xand finding the corresponding value of y.Such values can be tabulated thus :-

x y

-2 -2-I I

0 4

I 7

2 I0

An extensive set of tables could be drawnup in this way, but such tables would notrevc_.1 the distinctive character of the depen-dence of y on x. This, however, can be madequite clear by means of a method of graphicalrepresentation invented by the philosopherDescartes while he was lying in bed onemorning (which just shows the unwisdomof too early rising).

Draw two lines OX and OY at rightangles (Fig. 7). Any related pair of numbersx and y can now be represented by a pointsuch as P, which is situated x units of lengthperpendicularly to the right (if x is positive)or left (if x is negative) of OY, and y unitsof length perpendicularly up (if y is positive)or down (if y is negative) from OX. (Thereis no need for the units of length to be thesame in the two directions.) The numbersx and y are called the " co-ordinates " ofthe point P, which can be referred to asthe point (x,y).


In Fig. 8 are shown all the points repre-sented by the pairs of numbers listed in theabove table. A new aspect of the matteris at once revealed. All the points are seento lie on a straight line. Moreover, it will

Y 1

MI I l/

9 PAMEAimII 6.

5 IIII4)(0,

NMI1121

WWINMI o

1xraj

0

Fig. 8.

be found that any other related numbersbelonging to this set will give rise to pointswhich also lie on this straight line. Thestraight line can therefore be regarded asa complete representation of the function

y = 3x + 4since all the points of the function will befound somewhere on this line. Conversely,any point on the line satisfies the functionalrelationship, and the value of y for anygiven value of x could be read off the dia-gram or a suitable extension of it. Forx=1.7, for instance, the corresponding valueof y is seen to be 9.1.

The idea of functional form can now bedescribed. There is no special magic aboutthe numbers 3 and 4 in the above function,so it is inherently probable and is true infact that any other pair of numbers a and bwould give rise to a similar picture for thefunction, i.e., a straight line. The function

y = axis for this reason called a straight linefunction of x.

Obviously much more complicated num-bers than this could be built up out of xand other constant numbers. For instance

y=3x2+4x+5or more generally

y = ax2 +bx +c

2

The points of any such function for givenvalues of a and b could be similarly plottedon a diagram as in the above simpler case.It would be found that the points do notlie on a straight line for this rather moreelaborate function. Nevertheless by plot-ting a sufficiently large number of pointsclose together a dotted line will be producedthrough which a smooth curve can bedrawn, and it will be found that any otherpoints of the function within the range ofthe diagram will lie on or very close tothis curve, and the greater the number ofpoints calculated and plotted, the morenearly will any other points be found to lieon the curve.

(n) The Exponential Function.The above is no more than a very brief

introduction to the idea of functions andtheir graphical representation. From thispoint there will be considered a particularfunction closely related to the subjectmatter of Section 4, namely

y = axwhere a is some constant number.

6

5

y=2x

2

-4 -3 -2 -1 0x

Fig. 9.

2

To fix ideas, let a have some simple value,such as 2. That is

y = 2zThe first thing to notice is that x can have


any positive or negative fractional orintegral value, for the significance of such ageneral index has been determined. Forinstance, if x = .5

y 2'5 = 24 = 1/2=1.414and if x = - .5,

.y = r/x-5 = 1/1.414=-- .707

Proceeding in this way, it will be possibleto draw up a table of related values of xand y covering any desired range. Forinstance,

x y

-3.0 .125 1.0 2.00

-2.5 .177 2.5 2.83

-2.0 .250 2.0 4.00

-1.5 .355 2.5 5.66

-1.0 .500 3.0 8.00

0 1.00

These and similar values can be plotted ona diagram and a curve drawn. The curveis shown in Fig. g.

From such a curve any other corre-sponding pair of numbers within the givenrange can be determined fairly accurately(to about r per cent. or so). For instance,for the point on the curve for which y---4.8,x is seen to be 2.26, so that

4.8 = 22.26

(This illustrates a very useful applicationof the graphical representation of functions.Given that

4.8=2"the determination of x by any method ofdirect calculation would be a troublesomebusiness, and would puzzle most people.)

Theoretically the above curve could beextended indefinitely beyond the limitsshown in either direction, and it would befound to be a smooth continuous curve.This means that for any finite value of yit will be possible to find a value for x suchthat

yThere is no special magic about the number

615 October, 1926

2 which has been used for a in these calcu-lations, and it may therefore be said thatin general, for any finite positive value of a,and any finite value of y, it will be possibleto find a value for x such that

y ar[Notice that a must be positive. The readeris recommended to try to tabulate a set ofrelated values for y and x similar to theabove for

y = (-2)x ].

(c) Logarithms,

When x, y, and a are so related thaty = ax

x is called the logarithm of y to the base a,and is written

x = logoThe two statements

y=axand

x = logay

therefore mean the same thing. The numbercorresponding to a given logarithm is calledthe antilogarithm of the given logarithm.Thus in the above y is called the anti-logarithm of x. (These, by the way, arethe full ceremonial titles. Log and antilogare what one actually says.)

Having now shown that the logarithmof any number to the base a exists, andhaving indicated the possibility of deter-mining it by simple arithmetic and drawing,it will be assumed that there is available aset of tables or curves recording the logarithmof all numbers to the given base a, theintervals between whole numbers being sub-divided to any desired degree of fineness.For reasons to be given later, io is the basechosen in practice, and tables of logarithmsand antilogarithms to the base ro are easilyobtainable. What, now, is the use of suchtables ?

Suppose it is required to find the productof two numbers y , and y.. The logarithms ofthese numbers can be found in the tables.Call them x, and x 2. Then

xlyiand

so that512 = ax.

Y1 X Y2 = X axe = dx,+x.)

October, 1926

This means that (xi+x,) is the logarithmof (y, x y2), or that (y2 x y2) is the anti-logarithm of (x,--Px2). To find the product(y' xy2), therefore, it is only necessary tofind the logarithms of yl and y2, add thesetogether and find the antilogarithm of thistotal. For instance

/og10 3.412 = .53300/ogia 796 = 2.90091

Sum = 3.43391antilog = 2715.8

Therefore3.412 X 796 = 2715.8

Thus multiplication is reduced to a simpleoperation of addition. In a similar manner,division can be simplified to subtraction,for, by an obvious modification of the aboveproof it can be shown that

log a(y o/y2) = logay o-logaY 2

Thus to divide 796 by 3.412,log10 796 = 2.90091log10 3.412 = .53300Difference = 2.36791anti/ogio = 233.29

Therefore796/3.412 = 233.29

Again, since/oga(y2 xy2) = logay logaY 2

it follows that/oga(y2 X Y2 X Y3 Yn) =

logay logay 2 logay3 logaynand if

y2 =y2 =y3 =yo etc. =ythis becomes

loga(y x y x y n factors) = loga(yn)logay logay + logay n terms

= n x logay.Thus a number can be raised to anyintegral power by simple multiplication.For instance

/ogio 3.412 = .53300/og10 3.4125= 5 X .53300 = 2.66500antilog lo 2.66500 = 462.38

Therefore


3.4123 = 462.38

Furthermore, the general formulalogayn =n x logay

proved above for an integral index, is truealso for a fractional index. Let n=p/q,

where p and q are integers. Then by thedefinition of a fractional index,

p

( Q) = YPTherefore

Butlog. (r)q = logay' = p x logay

loga (Yq)q =q x loga(yq)Therefore

q x log (y4) = p x logayOr

toga (yq) -=- (p/q) x logay

In particular the nth root of any numbercan be calculated by logarithms, for

log. '&61= logayn= (r/n) x logay or (logay) n

Thus, to find the fifth root of 796,log10796 = 2.90091/og10;/796 = 2.90091 -5

= .58018antilogo. = 3.8037

Therefore6796 = 3.8037

Here an arithmetical process which is toocomplicated to be admitted into ordinarytext -books is reduced to a simple matterof division.

The full advantage of the logarithmicmethod of calculation is seen in the deter-mination of a more or less complicatednumber such as

2.433 x 191 X V347Y. - t/1.5193

The determination of this by direct calcu-lation would be a weariness to the flesh,for which weariness the flesh would probablyretaliate by slipping into error. Logarith-mically, however, the calculation is simplifiedto

logay = 3 X log.2.43 +loga347 - Voga1.519

The whole of the foregoing propositionswith regard to the application of logarithmsare independent of the particular base towhich the logarithms are referred. Thechoice of the base to be used in practice is


merely a matter of convenience. As stated the use of this other system of logarithmsabove the base m is actually used for the will not necessarily involve a new set offollowing reason. Any whole number or tables, for a logarithm to any one base candecimal can be expressed as some number be readily converted to the logarithm tobetween 1 and 10 multiplied by some power some other base in this way. Supposeof io. For instance logay = m

.143 = 1.43/10 = 1.43 x 10-1 logo, =3179.8 = 3.1798 x moo = 3.1798 x 103 i.e., y = a'" = b"

and so on. Thus any number can beexpressed in the form

yxrowhere n is a positive or negative integerand y is some number or fraction between

and ick. Nowlog joy x 101= logibY logiolo"

logibY + nThus all that is needed in a table of loga-rithms to the base io are the logarithms ofall numbers between 1 and io, subdivideddecimally to any desired degree. (Fivefigures are generally quite sufficient forexperimental work.) Such logarithms willall lie between o and 1, e.g.,

log,o2 = .30103

The whole number to be placed in front ofthe decimal point will be n, determined asshown above. Thus logi0200 would be2.30103, and /og10.002, i.e., /ogio2 x 10-3would be 3.30103. In practice the negativesign is put as a bar over the whole number,as in this example, to show that it refersonly to the whole number and not to thedecimal part which follows. It is moreconvenient to keep the decimal part positive.Thus 2.30103 is the same as (-2-1-.30103),i.e. (-1.69897), but the first form is alwaysused in computation.

Logarithms calculated to the base Ico

are called Common Logarithms, and thebase is not expressed. Thus if no otherbase is specified, it is always assumed tobe 10.

There is another set of logarithms whichis in occasional use. These are calculatedto the base 2.71828, for which apparentlyarbitrary number the symbol e or the GreekE is always used. There is method in thisapparent madness. So much so, in fact,that such logarithms are called " naturallogarithms," though at first sight nothingcould appear less natural. The naturallogarithm of y is written logo/. Actually

Suppose further that the logarithm of b

to the base a is k,i.e., b = akThen

y = bit = (ak)n = aanand since also

y = a'"m= kit

so thatlogay = logab x logby

In particularlogey = logelo X log 1057

and since /ogeio is a constant number(2.3026) the conversion reduces to

logey = 2.3026 x log10y

i.e., Natural Log. = Common Log. x 2.3026.If the foregoing description of logarithms

and their application is thoroughly under-stood, the reader should have no difficultyin using this method of computation. Itdoes not pretend to be a complete set ofworking instructions for the manipulationof log tables, but if the theory is reallyappreciated, the reader should be able tofollow the detailed instructions which aregenerally included with any such tables.(To guard against any misconception onthis point, it should be made clear that thetabulated logarithms found in the usualpublished tables have not actually beendetermined by the method described inthis article, which would not be nearlyaccurate enough. The actual method ofcalculation will probably be encounteredlater in this series.)

Finally, one practical point. The loga-rithmic method can only be applied to anyexpression which consists exclusively ofproducts, quotients, and powers. There-fore, before setting out on any series ofcalculations it is well to arrange them ascompletely as may be in a form suitablefor logarithmic computation. An instance

October, 1926

which may frequently occur in connectionwith alternating current problems is thedifference of two squares, i.e., a2-b2,where a and b have certain specified numericalvalues. The form a2-b2 is not suitable forthe use of logarithms, but, as already shown(see Examples, p. 450, July issue),

a2 b2 = (a -b) (a + b)and the form on the right hand side, con-sisting of the product of two factors, isbetter adapted for calculation. For example,

8742- 27.82 = (874 -27.8) (874 + 27.8)= 846.2 X 901.8

No general rules can be laid down, but theexercise of a little ingenuity in this matterwill often save a great deal of time and labour.


Examples. --Logarithms.

1. Given that log2 = .30103, show thatlog5 is .69897.

2. Log3=.4771. Show that/op/3 = 1.5228log 81 = 1.9084log31 .5228/0g33° = 14.313

3. What are the whole numbers in thelogs of 21, .021, 99918, .00073 ?

4. Show that 2100 is a whole number of301 figures.

5. Show that log (logioi°x)=I±x.6. Show that logal = o for all finite values

of a.

(To be continued.)

Note on a System of Atmospheric Elimination.By S. Butterworth.

(Admiralty Research Laboratory, Teddington.)

IN a recent patent specification* M. L.Levy has described a circuit which isintended to eliminate atmospheric or

similar parasitic shock excitations. Someeight years ago the present writer hadoccasion to study a very similar circuit whichhad been suggested to him for the same pur-pose, and came to the conclusion that a cir-cuit of this type was incapable of eliminatingdisturbances due to shock excitation of anykind.

As the circuit has been once more putforward, and as the arguments in its favourseem very plausible, it may be of interestto your readers to point out why it does noteliminate atmospheric disturbances, and whyit is inefficient in regard to increasingselectivity between two continuous wavesignals.

The circuit in question is essentially asshown in the figure. Two resonant circuits,I and II, receiving equal E.M.F.s from the

* British Patent No. 225,570, described inE.W. & W.E., June 1926. p. 398.

[11431

required station, are slightly detuned, sothat I resonates at a frequency slightlyabove, and II at a frequency slightly below,the frequency of the incoming signal. Be-cause of the rapid reversal of phase as we

pass through resonance, it is clear that forany frequency in the band lying between theresonances of I and II the potentials of thegrids of the valves a and b will tend to be inantiphase, while for any frequency outside

THE WIRELESS ENGINEER 6i9 October, 1926

this band they will tend to be in phase.Hence, if we couple the anode circuits ofthese valves to a third resonating circuit III,which is tuned to the required signal, we canso arrange that while the induced E.M.F.sin circuit III. due to the wanted signal, arehelping, those due to unwanted signals arein opposition. From this argument itappears, therefore, as if we can eliminateeverything but the required signal.

We will first deal with the fallacy in the- reasoning when the unwanted disturbance is

due to shock excitation. A little considera-tion will make it clear that when a resonantsystem is subject to shock, the resultingdisturbance does not merely repeat the shock,but there is also set up a damped oscillationhaving the same period as that of the resonantsystem. In the case of sound, of course, thisprinciple is made use of in the constructionof all stringed instruments. Thus in thecase of the present circuit system we see thatan atmospheric disturbance will cause circuitsI and II to " ring " each at its own fre-quency. Hence, the net E.M.F. induced incircuit III by an atmospheric will consist oftwo damped E.M.F.s of slightly differentfrequencies, and these, by virtue of theirrapid alternation in relative phase fromassisting to opposing, will produce a dampedbeating effect in circuit III. This effect willhave longer and longer duration as theefficiency of the circuits I and II is in-creased. The circuit is thus useless in so faras elimination of atmospheric disturbancesis concerned.

We now consider the selectivity of thecircuits in regard to continuous waves. Letthe circuits I and II have equal capacities C,equal resistances R, and inductances L -1and L +1 respectively. Let similar E.M.F.s,each of value E, be impressed on the circuits,the frequency being w/27r. Then, assumingequal transformers T linking the valves withthe circuit III, and taking the common overallvoltage amplification (valve +transformer)

as m, the net E.M.F. impressed on circuit IIIis

E

m EjwC

jcoC[R ico(L-1)-a}

R ±j {co (L +1)-t.,}1

2jcol

[R±j{co(L+1)-(7-coi }1[R±j{co(L-1)- }]

the usual vector notation having beenemployed.

Hence the R.M.S. value of the E.M.F. is2 (mE)1

C .ER2 tco(L +1)-7-112]

[R2 + {co (L 1) ac}

If we are concerned only with stations ofslightly different frequency, the quantityunder the square root sign is the only onewhich changes appreciably with co. But ifwe had employed the same circuits I and IIon the input and output sides of a singlevalve, using a weak coupling on the outputside, the frequency variation of the systemwould be governed by the same factor.It is concluded that the same circuit materialcould be used to give the same C.W. selec-tivity in a far more efficient manner withoutdeparting from common practice. In additionthe small factor 1 occurring in the numeratorof the expression for the E.M.F., acting oncircuit III, reduces the intensity of the re-quired C.W. signal, while in the case of theatmospheric disturbances it merely governsthe rapidity of the beats. This considerationimmediately shows that the system increases,instead of diminishing, the relative import-ance of the atmospheric disturbances.

(E c) -

October, 1926 62o EXPERIMENTAL WIRELESS &

The National Radio Exhibition. [R064

Review of Components and Accessories of Special Interest to our Readers.

AFTER a careful examination of the new appara-tus to be seen at the first National RadioExhibition held this year at Olympia, one is

ready to admit that the wireless industry hasalready passed through its infancy. There hasprobably been more real progress made in thedesign of broadcast receivers and their associatedcomponents and accessories during the past twelvemonths than during any other year since broad-casting commenced. It must not be thought fromthese remarks that apparatus embodying funda-mentally new ideas has been evolved, but progresshas essentially been made in regard to refinementsin design which are the outcome of a better under-standing of the underlying principles.

Osram gas discharge tube giving full wave rectificationfor use in an A.C. mains battery eliminator.

The Osram U5 full wave rectifier.

The New Valves.Many new valves have been introduced specially

designed to suit the requirements of variousamplifying circuit arrangements, as well as othersfor use in connection with rectifying equipment.In the 2 -volt group of Mullard PM valves thereare now three types, the PMI H.F., PMI L.F.,and the PM2. Each of these valves is operatedwith a filament potential of 1.8 volts, the PMItype passing a current of o.T ampere and thePM2 0.15 ampere. The high impedance PM'H.F. has an amplification factor of 13.5 with animpedance of 20,000 ohms, showing the mutualconductance to be o.48mA per volt. The lowimpedance L.F. valve has an amplificationfactor of 8.9 with an impedance of 18,00o, the

mutual conductance being o.5mA per volt. ThePMr H.F. is suitable for use in high -frequencyand resistance -coupled amplifiers, and the PMIL.F. is intended for transformer -coupled low -frequency stages, whilst the PM2, which has an

B.S.A. power valves types P425 and P612.

amplification an impedance of 8,75o,is suitable for the last L.F. stage where a largergrid bias is required.

Another new Mullard valve is the DP425 whichis a dull filament power amplifier, and especiallydesigned for use with loud -speakers. The filamentrequires 0.25 ampere at 3.8 volts, and it has thevery low impedance of 3,500 ohms with an amplifi-cation of 3.15, giving a mutual conductance ofo.9mA per volt, with an anode voltage of Too.The available grid swing is approximately 3o volts,

Burndept rectifying valve type U695 for chargingH.T. accumulators from A.G. supply.

so that a grid bias of at least 15 volts can be used.With this grid swing of plus or minus 15 volts anda mean anode current of 12 milliamperes at roovolts, the largest loud -speakers can be operatedwith a minimum amount of distortion.

Two new valves have been added to the Cosmosshort path series, the SP55R and the SP55B.The former is a power valve having an impedanceof 3,50o ohms with an amplification factor of 6,


givinguan exceptionally high mutual conductanceof approximately x.7mA per volt. The filamentvoltage is 5.5 and the current is 0.25 ampere. Thenew SP55B is a moderately high impedance valvehaving the exceptionally high amplification factor

The Burndept Ethotron gas discharge rectifier foruse in an A.C. battery eliminator.

of 35 and a mutual conductance of o.65mA pervolt, the impedance being 55,000 ohms.

Amongst the range of " Cossor" valves should bementioned three of the type which are of recentconstruction. These are 2 -volt valves, designedas detector, H.F. amplifier, and for power use

The Atlas full -wave rectifier for obtaining H.T.supply from alternating current mains. Grid biasing

potentials are also provided.

respectively. The power valve filament takes.15 ampere and the detector and H.F. valvesx ampere. A point of interest in the construction

of these valves is the shock -proof filament suspen-sion system. The filament is arched and is held inposition by a fine wire but is not under tension asis the case with other types. It is claimed that asa result the filament will withstand shocks withoutrisk of breakage.

B.S.A. Radio, Ltd., have now entered into thefield of valve manufacture. A number of newvalves with heavy strip filaments operating at lowtemperatures were shown. The type P425 is apower valve with a 4 -volt filament consuming

October, 1926

0.25 ampere and capable of handling large inputvoltages with a moderate high tension potential.The impedance is given as 5,000 to 7,000 ohmswith an amplification factor of 6.4 to 6.8. Anothernew valve, the PE612, has very similar character-istics, and is fitted with a filament consuming only0.12 to 0.14 ampere at approximately 6 volts.Like the new type Mullard valves, the B.S.A.Standard valves (produced jointly by B.S.A.Radio, Ltd., and Standard Telephones & Cables,

G.E.C. battery eliminator fitted with gas dischargefull -wave rectifier and neon lamp.

Ltd.) are fitted with pins of unusual construction.A slot is cut through the pin not quite reaching tothe ends so as to give a slight expansion about themiddle and produce a reliable spring contact withthe socket.

Among Marconi and Osram valves those ofspecial interest are designed for use as rectifiers inH.T. battery eliminators. For single-phase rectifi-cation the U4 has been produced, which is a two -electrode dull emitter, operated at an applied A.C.

Burndept Ethopower A .C. unit.

voltage to the anode of 220 volts maximum (R.M.S.).For biphase rectification the U5 is available,which is fitted with a dull emitting filament ratedat 5 volts 1.6 amperes. This valve will withstandan applied anode voltage of 250+250 (R.M.S.)

C2


maximum, and will provide a direct current up to5o milliamperes.

A new single -wave rectifying valve has beenproduced by Burndept intended particularly foruse in charging high tension batteries from alter-nating supply mains. It is styled the P695, passing

Single -wave rectifier for charging H.T. batteries fromA.C. supply.

a current of 0.95 ampere at 6 volts and is capableof being used for charging accumulator batteriesup to 120 volts with a maximum charging rate of6omA.

Full Wave Gas Discharge Rectifiers.Gas discharge rectifying valves have been pro-

duced by both Osram and Burndept for use inapparatus for obtaining D.C. supply from A.C.mains. They are full -wave rectifiers and possessthe outstanding merit that they can he overrunand having no filament will give prolonged service.

Fellophone A.C. battery eliminator.

Valves of this type become heated in operation,while compared with the thermionic rectifier givea wave form that necessitates the use of largereservoir condensers in the smoothing equipment.

Conspicuous at the Exhibition was the varietyof battery eliminating sets for use on A.C. andD.C. supply.

A.C. Battery Eliminators.The Atlas H.T. battery eliminator produced by

H. Clarke & Co., of Manchester, is designed toprovide full -wave rectification with the use ofeither half or full -wave rectifying valves. Itspower transformer is of large size with obviouslygood insulation between the sections, mid -point

The half -wave single output rectifier of TudoradioCo., Ltd.

tappings being provided at both output windings.Mullard DU5 or DUro valves are used as recti-fiers and smoothing is accomplished by the usualchokes and condensers though it is interestingto note that both open and closed core chokes are

Climax power transformer for use in the constructionof battery eliminators.


fitted and series connected. Several voltageoutputs are produced by shunting the rectifiedoutput across a potentiometer and two fixedtappings are made to give approximately 6o and15o volts. A third output voltage is controlled

The Marconiphone power transformer with tappedprimary so that it can be used on 100/120 volt or

200/240 volt supply.

by a multi -contact switch which gives any potentialin small steps up to the maximum of 15oFour taps are also provided to give grid biasingpotentials of 4, 8, 12 and 16 volts. The entireoutfit is accommodated in a steel box.

The General Electric Company also producebattery eliminators for use on A.C. and D.C.

Gambrell set operated entirely from A.C. supply.The filaments which pass a current of o.o6 ampere

are series connected.

623 October, 1926

mains, though it is the A.C. instrument that is ofparticular interest as it makes use of a full -wavegas discharge rectifying valve. Owing to theparticular wave -form obtainable with this type ofrectifier a very liberal smoothing equipment isnecessary which is augmented in this instance bythe use of a neon tube the non -uniform conductivity,properties of which assist in providing a smoothoutput.

The Ethopower H.T. unit of Burndept also makesuse of a gas discharge rectifier and includes a very

Cable H.T. unit for use on D.C. supply. A singlevariable output is obtainable.

liberal smoothing equipment. To provide more thanone voltage output a lead is taken from the maxi-mum voltage terminal and the potential is droppedthrough a resistance shunted with a by-passcondenser.

Although the use of rectifying equipment suggeststhe entire elimination of H.T. primary or secondarybatteries there are many who would prefer toretain their accumulator source of H.T. supply.For use on A.C. mains Burndept Wireless, Ltd.,offer a useful unit which will provide the necessaryslow charging rate making use of a thermionicvalve rectifier. A pilot lamp according to itsbrightness indicates the value of the chargingcurrent, a visible glow being obtained at 35 mA.

Among the many battery eliminators makinguse of half -wave rectifiers are the units made bythe Fellows Magneto Company and the TudoradioCompany. The particular merit of the Fello-phone high tension unit is that in overall shapeand dimensions it can replace a dry cell high tensionbattery.

W ates, Bros., D.C. battery eliminator.

October, 1926

A number of parts are available for the con-struction of battery eliminators working fromA.G. supply, the power transformer being thecomponent which it is not always convenient toconstruct. Both Climax Radio Electric, Ltd., aswell as the Marconiphone Co., Ltd., have producedsuitable transformers for use in full -wave rectifying

The Rectalloy electrolytic battery chu; r for2, 4 or 6 -volt L.T. accumulators.


circuits supplying both filament and anode poten-tials. The anode circuit winding of the Marconi -phone transformer is intended to give 160 +160volts, whilst the transformer is tapped so that itcan be used on voltages between too and 240.

Other auxiliary apparatus for use in batteryeliminator construction includes the " Climax "smoothing choke and potential divider, the latterbeing wire wound to a resistance of 20,000 ohmsand tapped in ten sections.

Ecko mains unit giving one fixed and two adjustableoutput potentials.

Deriving Filament Supply from A.C.The incorporation of a battery eliminator in a

receiver is carried out by Gambrell Bros., the setbeing of special interest, for both filament heatingas well as anode current supply are obtained fromA.C. mains. The rectifier makes use of an Osram

U5 type valve, the filaments of the receiving valvesbeing series -connected and heated from the rectifiedoutput. This arrangement is entirely new and isof special interest particularly as it is used withsets fitted with I, 2, 3 or 4 valves.

D.C. Potential Dividers.Battery eliminators for use with D.C. supply

have been produced by many manufacturers, andconsist essentially of a resistance bridging themain leads as a potential divider, the outputbeing smoothed with the aid of condensers 'aswell as chokes in some instances.

Ethodyne seven -valve superheterodyne receiver. Thedouble wound frame is to eliminate long wave

interference.

In several of the models provision is also madefor obtaining grid biasing potentials, though insuch cases it is essential in order to eliminatemains noise to entirely free the grid circuit ofripple. In a well -designed D.C. eliminator, how-ever, fluctuations of potential in the grid circuitmay be used to compensate for fluctuations ofanode potential. Voltage regulation by means ofseries resistances are to be found in some of theD.C. sets, and variable high resistances are some-times fitted to provide a critical control of H.T.potential, so that the voltage drop can be regulatedaccording to the resistance of the anode circuit.


Almost every manufacturer of complete setshas produced a portable receiver, in many casesincorporating batteries and loud -speaker. Themajority of these sets are superheterodynes, andit is interesting to consider the circuit arrangementsadopted.

Superheterodyne Designs.There were at least twenty superheterodyne

sets shown, and only in a single instance was thedetector valve arranged to function also as anoscillator. The use of a separate oscillator valveseems to have become standard practice.

As to the intermediate amplifier, there is atendency to employ only two high -frequencystages, while in many of the sets transformershaving iron cores are used. It was not easy toexamine the internal construction of these trans-formers, though it would appear that in no single

Interchangeable oscillator coupling unit fitted to theB.S.A. standard superheterodyne.

instance was the iron core completely closed.Several of the couplings were wound in groovedspools and fitted with cores of iron wire. It wouldseem that transformers are matched not byindividual adjustment, but by manufacturing sothat the windings are physically identical.Although it was suggested in some instancesthat matching was effected by adjusting thewindings, and in others by altering the amount ofiron used for the core, it would seem that both ofthese methods are rather inapplicable whenthe windings are subsequently enclosed in ametal screening box. In order that the spoolsmay be of identical inductance, machine methodsof winding have been adopted in one or twoinstances in preference to running on the requirednumber of turns into a given space. No infor-mation is available as tr the winding ratios, thoughit would seem from -I space occupied by thewindings that step-ui .tios, perhaps as high asfour to one, had been pted.

Many new feati- outstanding merit will

October, 1926

be found in the Ethodyne intervalve couplings.In this set, two stages provide the requiredamplification, and the so-called filter circuit isabandoned, both stages being designed to tune overthe required, though limited, frequency band.

2o -volt Hart H.T. accumulator unit.

Stabilising the intermediate amplifier is invariablycarried out by means of a grid potentiometer,excepting in the well-known Igranic set wherestability is produced by neutralising stray capacities.By using suitable valves and correctly designedtransformers, higher amplification per stage isnow being obtained, with the result that in manyof the sets only two intermediate stages have beenemployed. The majority of the sets made use offrame reaction, and as regards L.F. amplificationthere is a trend towards using a resistance -coupled stage followed by a second, which is trans-former coupled, a switch being sometimes providedfor cutting out the resistance amplifier.

Ediswan " Hylo " 12 -volt H.T. section.

Many of the superheterodynes were tunableonly over the broadcast band, though in the setsshown by L. McMichael Ltd., and the B.S.A.Radio, interchangeable oscillator coupling units


were fitted. As to tuning the frame, the G.E.C.set includes a multi -contact switch which adds thenecessary loading and at the same time changesover the connections to the oscillator so that thewavelength of 30o to 3,000 metres is continuouslycovered.

The new nickel steel alkaline battery for H.T.supply by Batteries Ltd.

Among receiving sets employing other circuitsystems the Pye 5 -valve set is of special interest,for being designed to receive from the Daventrystation only the tuning of the two high -frequencystages once being accurately adjusted operateswithout the provision of control knobs. Thetwo H.F. stages are stabilised by neutralisingthe effects of stray capacity and the detectorvalve which is an anode bend rectifier, is followedby two L.F. stages. It is interesting to note herethe decline in the use of the leaky grid condenserfor rectification, particularly in sets embodyingone or more stages of H.F. amplification. ThePye set, which is entirely self-contained, is fittedwith three simple controls, one of which disconnectsan L.F. amplifier, the second is a continuouslyadjustable volume control and the third is avernier condenser connected across the frame andused to correct for small differences of tuning.The majority of the portable sets are fitted withcentre tapped frames.

Hart H.T. Secondary Batteries.The increased use of secondary batteries for H.T.

supply is evidenced by the considerable number ofH.T. accumulators shown during the Exhibition.The Hart battery, which is of new design, is madeup in 20 -volt units, the individual cells being incylindrical glass containers with sealed tops andrubber plugs fitted with glass vents. A woodenrack is arranged so that the sections can be as-sembled one on top of the other, thus effecting aconsiderable economy in space.

A novel feature was to be found in the D.P.battery, inasmuch as each individual cell wascompletely sealed with a thick rectangular rubberplug which permits of the plates being withdrawnand the cell easily cleaned out without the need forbreaking open the cells as in the case when bitu-mastic or other material is used for sealing. The

D.P. Battery Co., Ltd., have also introduced anentirely new form of high tension battery, theprincipal feature of which is the provision of alarge barrel type switch extending the full lengthof each bank of 20 cells which when rotated gives24, 6 or 2 volts at the output terminals.

The use of glass containers is now standardpractice at least for H.T. battery constructionand the Edison Swan Electric Company, in theirnew " Hymeg " battery, in addition to the use ofglass containers, fit glass lids. The glass boxesare ribbed externally to provide spacing betweenthe cells, and in general the battery, which is carriedin an acid proof wood case, will be found to be cleanand comparatively free from acid spray in use.

The introduction of the steel plate battery forH.T. supply by Batteries Ltd., of Redditch, willundoubtedly develop. The cells are constructedon the lines of the well-known " Nife " nickelsteel alkaline cell which has proved so successfulin a number of applications where skilled andconstant attention are not available. The disad-vantage of using cells of this type is that on dis-charge the voltage slowly falls, whereas with thelead plate battery the voltage remains practicallyconstant during discharge. This is perhaps noserious disadvantage as regards H.T. supply whilemany advantages are gained. The cells arepractically foolproof and require no attentionbeyond the occasional addition of distilled waterand will not deteriorate if left unattended ascompared with the sulphating that sets in in thecase of the lead battery. The steel plates cannotbuckle and the steel containers cannot crackwhile the battery can be short circuited or evencharged in the reverse direction without damage.The charging rate for a given ampere -hour capacityis higher than the lead battery but the mostimportant merit is its long life in spite of indifferenttreatment.

Dubilier Univane condenser in which the movingplates are propelled singly from the position of

minimum to maximum capacity.

Components.Among components mention might be made of

the new Cosmos resistance -coupled L.F. ampli-fying stage in which the condenser and resistancesare enclosed in a moulded sub -base which carriesthe valve holder.


As regards variable condensers progress hasessentially been made in the methods adoptedfor supporting the plates, the provision of goodbearings and a general improvement in finish.The Dubilier " Univane " condenser by an ex-ceedingly ingenious arrangement gives a criticaladjustment of capacity without the use ofgearing. Each moving plate is separately con-trolled and by rotating the dial the plates are

Exterior of the B.T.-H. loud -speaker. It has amoving coil movement driving " free edge " cone.

picked up in turn and carried from the positionof minimum to maximum capacity. Apart fromthe scale on the dial an auxiliary indicator showsthe number of plates carried over to the positionof maximum capacity.

A useful buzzer wavemeter was shown byGambrell. The type " D " instrument covers awavelength range of from 5o to 500 metres and isfitted with a reliable form of buzzer giving aconstant note and it is claimed a constant ampli-tude in the trains of oscillations.

The buzzer contact, instead. of being rigidlymounted, is a spring and this form of constructionprobably accounts for the high note which is adistinct feature of the instrument. Two Gambrellcoils are supplied together with the necessarycalibration charts.

Changes in Loud -Speaker Design.The decline in the popularity of the horn type

loud -speaker, which is being replaced by models ofthe cone type, was evidenced at the Exhibition,some of the newer, models being fitted with coneshaving " free edges." The principle of using amoving coil in a magnetic field is not new in loud-speaker construction, but the objection to thissystem has been the need for providing the polarisa-tion current for the permanent field thus creatingan extra drain on the current supply from thebatteries.

The new B.T.-H. loud -speaker model R.K. is amoving coil instrument with a conical diaphragmpractically unsupported at its edges. It is under-stood that an air -tight barrier is, however, fittedround the edge to prevent interaction betweenthe sound waves set up from opposite sides of thediaphragm, and with this same purpose in view the

October, 1926

diaphragm is mounted in the centre of a largecabinet so that the entire front board assists inpreventing interaction, which would have theresult of considerably reducing the volume ofsound produced. A polarising field is set up byfour pairs of permanent magnets, roughly 8 in. inlength by 1i in. wide and nearly in. in thick-ness, and arranged in pairs. An original featurealso is that of incorporating a power amplifier inthe loud -speaker cabinet.

Simple Testing Equipment.

At several of the stands testing sets were exhibitedprincipally for examining valves, making con-tinuity tests in high and low resistance circuits,and for resistance and capacity measurements.

Peter Curtis, Ltd., exhibited a testing set suitablefor making all the usual tests and measurementsmet with by the retailer or small manufacturer.

Cleartron Radio, Ltd., now produce a simplevalve testing set which they supply to retailers whohandle their valves, roughly revealing filament andanode characteristics.

The Liberty Radio tester manufactured byRadi-Arc Electric Co., Ltd., is fitted with a two -range voltmeter and change -over switch for makingbattery tests, as well as a neon lamp which, whenconnected to the lighting mains, may be used forcontinuity and insulation tests.

The Liberty testing set for voltage. measurements,continuity and insulation tests.

To the transmitting amateur probably the mostinteresting exhibit at the Exhibition was theexperimental quartz -controlled oscillator shown byMr. A. Hinderlich, M.A., who has apparentlyconducted considerable investigation work on theproperties of quartz crystal by means of a valveoscillator. He demonstrated in a very simpleway the use of a crystal for controlling fre-quency.


An Examination of the Properties ofAudio -Frequency Amplifiers by Means

of the Duddell Oscillograph.By W. Baggally.

S there appear to be very little if anydata available on the nature andmagnitude of waveform-as distinct

from amplitude -distortion produced by audio -frequency triode amplifiers as commonlyutilised in the reproduction of broadcasttelephony and also for public address purposes,-an investigation was undertaken into theproperties of such an amplifier.

The diagram of the instrument is shownin Fig. i and is quite orthodox ; the couplingis by means of " Radio Instruments" standardpattern audio -frequency intervalve trans-formers, Marconi-Osram LS5 valves beingnormally used.

00000

C>C>0 0 0

T

Fig. 1.

The Duddell oscillograph suggested itselfas a possible instrument for use in the presentinvestigation, and although widely knownand used, a short description may not beout of place here.

The basic principle involved is that of themoving -coil galvanometer, but in theoscillograph the coil is replaced by a singletightly stretched loop of phosphor -bronzestrip, the length of the vibrating portionof the loop being defined by bridge pieces ina similar manner to the strings of a violin.

[R132

Bridging this loop midway between thetwo bridge pieces is a very small mirrorwhich reflects a beam of light from an arclamp on to a photographic plate which isfalling freely under the influence of gravity.

This assemblage is immersed in oil andplaced between the poles of a powerfulmagnet, the object of the oil being to dampthe vibrations and render the apparatusaperiodic.

On passing an alternating current throughthe loop, a couple is set up which causesthe mirror and its reflected beam to oscillate,so recording the time graph of the A.C. onthe falling plate.

The whole vibrating system is so smalland tightly stretched that the period un-damped is of the order of .000i second, sothat no difficulty from resonance effectsis experienced at moderate frequencies,especially in view of the oil immersion.

In the actual oscillograph, two independentvibrators and a fixed mirror to give a zeroline are immersed in the same oil bath andsupported between the poles of the samemagnet.

Having described in brief the action of theoscillograph, an account may be given of theactual work undertaken.

The apparatus of Fig. 2 was set up, afterhaving calibrated the vibrators on D.C.;a previous experiment had shown that theimpedance was equal to the D.C. resistanceat the frequency used in the experiments,viz., approximately 480.

It will be realised that the amplitude ofswing of the reflected beam gives the peakvalue of the current passing through theoscillograph vibrator.

Owing to various visual difficulties, suchas the width of the spots of light, readingscould only be estimated to 0.5 divisions onthe scale used when calibrating the oscillo-graph, so that numerical results probablydo not attain a greater accuracy than ioper cent.


The method of examining amplifiers tohe described is capable of furnishing thefollowing information, all referring, of course,to the particular frequency chosen :-

I. An indication of the nature and extentof the waveform distortion produced bythe amplifier under working conditions.

2. The total voltage amplification obtainedwith specified output conditions.

3. The peak voltage on the first grid,and hence the maximum permissible voltageconsistent with absence of overloading andconsequent distortion.

4. The best working adjustments for theamplifier.

Let us consider the methods of obtainingthese results in the order given.

October, 1926

PS, a stepdown transformer ; and A, theamplifier under observation.

Let the sensitivity of the vibrators beadjusted to equality and be S divisions of thescale per ampere.

Let V1 and V, be the amplitudes of swingof V, and V, respectively, measured inscale divisions.

Let I be the peak value of the current inthe circuit GL R V,.

Let E be the peak value of the amplifierinput E.M.F.

With A removed, let the primary terminalsof PS be connected across R, also let R,be removed and R, short-circuited, andwith the alternator running let R be adjusteduntil Vi=V2; let this value of R be denotedby K, the shunt effect of PS being in generalnegligible.

Let N be the ratio of the peak value of thecurrent in the circuit S R, R, to the peakvalue of the current in the circuit R, V,.

Let M be the voltage amplification of Awith PS in the output circuit, the secondaryof PS being loaded with a resistance V,where V is the resistance of V,.

Then with the amplifier, removed as above,M=i, also N=i, Vi=V, and R.K.

Now keeping all other conditions un-changed, let A be introduced into the circuit,then we have

Ep=MEwhere Ep is the peak voltage across the

Fig. 2. primary terminals of PS.i. For this experiment, it is merely

necessary to adjust the amplifier to theconditions under which it is desired to takegraphs, at the same time noting that V,and V, are giving convenient deflections.

A photographic plate is allowed to fall infront of the vibrating light spots and ondevelopment the input and output graphswill be found superimposed and may beexamined at leisure, when the nature andoften also the cause of any distortion presentwill become evident.

2. Referring to Fig. 2, G is the alternatingcurrent source ; L, a choke coil for thepurpose of purifying, as far as possible, thewaveform of the current delivered by G;V, and V the vibrators of the oscillograph ;R, R, and R2, non -inductive resistances ;

Therefore

Vor M= -2V,

Now assume K changed to some othervalue R, then since E changes in the sameratio without affecting V, we have

KM -V 'RV,

Now let R, and R, be introduced and we

obtain M=NKV 2RV,

Now in order to obtain N Is! can proceedas follows :-

It is necessary that the load on PS bekept constant throughout the experiments

(1)

October, 1926

and equal to the resistance of V, alone,so that we have

R 2 +1/R, ± I/ V-v

which becomesV2=/N.Ri+V) (2)

also from the theory of shunt circuits we haveV

N --I substituting for R1 in (2) we obtain

V2 = R,(V+ N-Vi)

R1=

transposing we have

R,= I'

. (3)

(4)

Oscillograph 1.

It will thus be seen that by makingmeasurements of the amplitudes cf swingof the oscillograph vibrators, at the sametime noting the values of the various re-sistances, the amplification obtained at thefrequency used is at once obtained by theabove simple formula.

Of course, when the amplifier is removedfrom the circuit, an oscillogram may be takenand will show up any waveform distortionoccurring in PS.

3. The peak voltage impressed upon thefirst grid is obtained from the relation

- V1R . (5)

which is obvious and requires no proof.4. If a rotating mirror is available, the

reflected beams from the oscillograph maybe allowed to fall upon a small white screen,the screen being viewed by reflection in therevolving mirrors, when the waveforms will

63o EXPERIMENTAL WIRELESS &

be visible ; the adjustments of the amplifierare then varied until the distortion is reducedto a minimum.

Oscillograph 2.

The following tests were conducted on theamplifier in order to illustrate the method :-

E was adjusted to a value of f volt peak,as probably representing average con-ditions, the frequency of the generatorbeing, as previously mentioned, about 480cycles per second.

Oscillograms 1-4 were then taken on theamplifier with one, two and three stages,using LS5 valves, and with three stagesusing R4 valves respectively ; the operatingdetails are shown in the table.

Oscillograph 3.

Oscillograms 5 and 6 are included asexamples of what may happen to the outputwaveform if the amplifier is wrongly used ;No. 5 shows the effect of excessive over-loading, while No. 6 was taken while thevalves were oscillating at a frequency of the


order of 104 cycles per second and is atribute to the facility with which theoscillograph follows rapid and complexcurrent fluctuations.

Oscillograph 4.

No precise details are available concerningthese last two graphs, as they were obtainedin the course of preliminary work.

It can easily be shown that a triodepossessing the constants of the LS5 re-quires an anode voltage of about 30o anda negative grid potential of about 3o voltsin order to take full advantage of itscharacteristics. (It will be noticed fromoscillogram 3 that the last valve is over-loaded when the anode and grid voltages are120 and 10 respectively.)

In order to determine the maximumpermissible peak voltage on the first gridunder the best working conditions, apotential of 30o volts was applied to the

Oscillograph .5 -

anodes, the immediate result being thebreakdown of one of the intervalve trans-formers, which were of the old (non -sectional)type manufactured by Messrs. RadioInstruments, Ltd.

These were replaced by the new (section-ally wound) type by the same firm and theamplifier again connected up.

The instrument was now found to beoscillating at a frequency of about 16 periodsper second, and in order to stabilise it,it was necessary to connect a (nominal)0.1 megohm leak across the secondaryof the first transformer and reduce the anodepotential to 23o volts.

The output wave was then examined in arotating mirror while the input voltage wasincreased until the first signs of distortionappeared, and then decreased until thedistortion just disappeared.

The value of input voltage was calculatedand found to be 0.49 volt, the amplificationbeing 216, the steady grid potential 3ovolts and the anode voltage 230.

Oscillograph 6.

Keeping the grid and anode voltagesconstant, the amplification with two stageswas 4o and with one stage 4.

These figures, and those in the table,giving the amplification produced by the

TABLE.E.M.F. on first grid = r volt peak.Anode volts =12o.Grid volts =lo.LS5 valves, filament current normal.

Oscillo- No. of Loud- Voltagegram No. stages. speaker. amplification.

Very weak 3.22 2 Moderate 403 3 Loud 150

R valves, "anode volts 65, grid volts 1.5, filamentcurrent excessive

4 3 I Weak I 15

October, 1926 632

various combinations of stages, call forsome comment, and further research alongthese lines is probably indicated.

It may be said at the outset, that thereare no mechanical faults in the amplifier, as thevalves and transformers have been changedand all tested separately, also the individualcomplete stages, in their places on theebonite panel.

The low amplification given by the firststage is explained by the fact that thisfigure does not include the step-up of a trans-former, which is included in the values forthe other stages.

It will, however, be observed that thetotal amplification given by the three stagesworking together should be 4x fox 10=400, whereas that actually obtained is aboutone-half this number.

Pending further experiments, the writeris inclined to think that the cause may bereverse reaction effects between the stages,as it is found that on reversing the con-nections to primary or secondary of eitherof the intervalve transformers, the instrumentoscillates violently, thus demonstrating thepresence of reaction effects which, whenshifted in phase by 18o degrees, may lowerthe amplification.

It was also found on one occasion thatmerely reversing the connections to theprimary of the output step-down transformercaused the amplification to change from 75to 172, thus showing that large variationsin the amplification attained under slightlydiffering conditions are to be expected.

The loud -speaker results given wereobtained by changing over the amplifieroutput from the oscillograph and itsauxiliary apparatus to a large ' Magnavox "loud -speaker and are merely intended to actas a very rough guide as to what may beexpected under the conditions quoted.

It is to be noted that the frequency (about480) is somewhat low for a loud -speaker,so that it is probably not working to thebest advantage ; however, the loud -speaker


tests are interesting when comparing theoscillograms.

The loud -speaker strength when usingthree stages with 23o volts on the anodesand an input of 0.49 peak volt, was aboutthe same as with 120 volts on the anodesand an input of i peak volt, although in theformer case distortion was absent and. in thelatter, quite marked ; this is an excellentexample of the enormous advantage, bothin purity and volume, to be derived from theuse of sufficient anode potential to bring thestraight portion of the characteristic to thenegative side of the axis of anode current.

The writer draws the following conclusionsfrom the experiments, but in what followsdue allowance must be made for the preferencegenerally evinced by loud -speakers for thehigher frequencies.

1. An average general purpose receivingvalve operating a loud -speaker, howeverweakly, using average values of anodevoltage (e.g., 65) and grid potential (e.g.,1.5), will in general be overloaded.

2. A power amplifier valve of the LS5class, using the values of anode and gridpotential usually to be found in loud -speakingbroadcast receivers (e.g., 120 and I0, say),will produce a moderate volume of soundsufficient to fill a fair-sized room withoutoverloading, but no more.

3. A similar valve, whose anode and gridpotentials have been so raised as to makeuse of the whole of the linear part of thecharacteristic curve, is, with a suitable loud-speaker, capable of producing a considerablygreater volume.

Statements such as the above are ofnecessity of a vague general nature, owingto the difficulties of estimating soundintensities, etc.

It is almost certain that quite a largeamount of distortion may pass unnoticedowing to the tolerance of the human ear,but it nevertheless may be perceived sub-consciously, so that every effort should bemade to reduce it to the absolute minimum.


Calibration of Ultra Short Wavelengths.F. Aughtie, B.Sc. (G6AT). [R213

THE usual way of calibrating a circuitfor short wavelengths is to employ acalibrated oscillator, which is rich in

harmonics, and by taking readings of succes-sive " chirps " determine the order of thewavelength and hence determine whichharmonic of the oscillator is beating with thereceiver.

An example will make this clear : Supposewe obtain chirps at 120, 150, and 18o metres,the differences between readings are both3o metres and we see that the receiver isoscillating at 3o metres, beating with the4th, 5th and 6th harmonics of the wave -meter respectively.

If, however, we try to calibrate on about5 to io metres from a wavemeter whosefundamental range is, say, roo to 200 metres,some considerable difficulty is experienced indetermining which harmonic of the wave -meter is causing the chirp. The reason forthis is as follows :-

An ordinary condenser dial cannot be readto within less than about half a degree, thatis roughly to within one-third per cent.This, combined with the fact that the calibra-tion curve will not actually be perfectlysmooth -although it will be drawn so -causes the average wavemeter to have anaccuracy not better than half of one per cent.Now suppose we get chirps at roo and 105metres by wavemeter. Our ioo metres maybe anywhere between 100.5 and 99.5 ; simi-larly the 105 metres is somewhere between105.5 and 104.5 roughly. Hence the differ-ence between these may be anything between4 and 6 metres. Are we, therefore, at the25th harmonic of ioo metres or at the 17th,or which one in between these is the rightone to choose ? If we go higher up the scaleof the wavemeter the latitude of successivedifferences will increase so that even byaveraging we are still left to make the choicebetween four or five possible values.

After some thought the writer has evolveda method of deciding which harmonic iscausing the chirp, and is writing this article

trusting that the method may be of somehelp to other experimenters interested in theultra short wavelengths. The method isbest explained by an example :-

In an actual calibration (of the minimumwavelength for a particular coil) chirps wereheard at settings of the calibrated oscillatorcorresponding to 150.7, 156.1, 161.4, 167,and 172.5 metres. The successive differencesare : 5.4, 5.3, 5.6, 5.5 metres. Set out fivecolumns and at the head of each place thereadings 150.7, 156.1, etc. (see Table I.).In each column write the value of thedifferent harmonics of the particular wave-length which may be correct and also one ortwo above and below these, i.e., under 150.7we have written the readings 6.03, 5.8o, 5.58,5.38, 5.2o, 5.03 ; being respectively the 25th,26th, 27th, 28th, 29th, 3oth harmonics of150.7 metres. Under 156.1 we have 6.02. . . 5.04 being the 26th . . . 31stharmonic. And so on for the other readings.

TABLE I.

150.7 156.1 161.4 167 172.5

6.03 6.02 5.98 5.97 5.955.80 5.79 5.77 5.77 5.755.58 5.58 5.57 5.57 5.565.38 5.38 5.38 5.39 5.395.2o 5.21 5.21 5.22 5.235.03 5.04 5.05 5.o6 5.07

Differences are : 5.4, 5.3, 5.6, and 5.5.TABLE II.

147.8 153 158.4 163.8 I 169

5.68 5.67 5.67 5.65 5.645.47 5.47 5.46 5.47 5.455.28 5.28 5.28 5.28 5.285.10 5.11 5.12 5.22 5.125

4.83 4.94 4.95i

4.96 4.97

Differences are : 5.2, 5.4, 5.4 and 5.2.

Now look along the first line ; the readingsare : 6.03, 6.oz, 5.98, 5.97, 5.95 : there is


a steady fall from left, to right. Along thebottom line the readings are 5.03, 5.04, 5.05,5.06, 5.07-a steady rise from left to right.On looking at the whole table, it will be seenthat in all the lines above the one 5.38,5.38, . . . 5.39, there is a fall from left toright, whereas in all the lines below this onethere is a steady rise from left to right. Itis therefore clearly seen that the particularwavelength measured lies between 5.38 and5.39. Our probable error, therefore, is of

the order of 0.2 per cent. despite the fact thatthe successive differences differed by quitea considerable amount.

To make the method quite clear a furtherexample is set out in Table II. As before,the readings of the wavemeter are given atthe head of the columns ; successive differ-ences give the clue as to which harmonics towrite down and these are entered in thecolumns. It will be seen that in this casethe wavelength was 5.28 metres.

A Very Sensitive Valve Galvanometer.By Prof. G. W. 0. Howe, D.Sc. [R251

RJAEGER and H. Scheffers, of theSiemens & Halske Co., have recently

'described* a very sensitive valvegalvanometer employing a double grid valvewith resistance back -coupling. The diagramof connections is shown in the figure. The

space charge current is of the inner gridand the anode current is are always togetherequal to the saturation current, so thatis -1-i s=i5=constant. The current is passesthrough the back -coupling resistance R.and tends thus to lower the potential of theouter grid. The current is to be measuredpasses through the resistance R, and thustends to raise the potential of the outergrid ; hence its resultant potential Vg=i,R1-1.R3+B. The bias voltage B maybe positive or negative. In the anodecircuit, is= (Vg +a Vs)/R, where a and R1are constants of the valve, assuming that one

* See Elektrofechnik and Maschinenbau, p. 1038,27th December, 1926.

is working on the straight part of the.characteristic.

From these three equations it can beshown that

dVg R,di, R3

As R3 approaches R, in magnitude thisexpression approaches infinity, and a verysmall change in the current is causes a verylarge change in Vg and therefore also in theanode current which is read on a sensitivegalvanometer. The effect of the back -coupling is equivalent to an increase in thesteepness of the characteristic of the valve ;it can be explained very simply as follows :The current is increases the potential of theouter grid ; this causes an increase of theanode current, but since the inner grid is sopositive that its current together with theanode current is equal to the constant emissioncurrent of the filament, the increase of anodecurrent causes a decrease in the current13 and therefore a reduced voltage drop inthe resistance R3. This tends to raisethe potential of the outer grid still furtherand the whole cycle is gone through again,the effects being cumulative.

One cannot push the sensitivity too far,however, owing to the difficulty of keepingthe filament temperature constant, but withcare the value of R. can be so adjusted thatthe effective steepness of the characteristicis ten times its original steepness.


Abstracts and References.

October, 1926

Compiled by the Radio Research Board and reproduced by arrangement with the Departmentof Scientific and Industrial Research.

R000.-WIRELESS AND GENERAL.R008.-UNITED STATES-RADIO PATENTS. (Elec-

trical Review, 13th August, 1926, p. 264.)" The rise of radio has resulted in the swamping

of the United States Patent Office with applications."Although the number of examiners handling radioapplications has been almost trebled, 1,85o petitionsare pending, and the radio division is five monthsbehind the applications.

R008.-SAFEGUARDS FOR THE RADIO INVENTOR.--E. Curtis. (Proc. Inst. Radio Engineers,Vol. 14, August, 1926, pp. 471-477.)

Precautions to be taken by the radio inventorwho is not associated with an organisation whichincludes a patent department, with explanationwhy such precautions are necessary.

R070.-COLLEGIATE TRAINING FOR THE RADIOENGINEERING FIELD.-C. Jansky, Jr. (Proc.Inst. Radio Engineers, Vol. 14, August, 1926,PP. 431-445.)

A paper presented before the Chicago section,I.R.E., in March, and at the New York meetingin June, together with the discussion following it.

R080.-SHORT-WAVE TRANSMISSIONS. ( WirelessWorld, 18th August, 1926, pp. 221-222.)

A list of some 14o stations in all parts of the worldworking on wavelengths between 13 and 15o metres.

R100.-GENERAL PRINCIPLES AND THEORY.RII3.-NOTES ON WIRELESS MATTERS. (Electrician,

23rd July, 1926, p. roo.)Two letters referring to Mr. Turner's article on

wave propagation in the issue of 9th July, p. 42(these abstracts, E.W. & W.E., September, 1926,p. 567). The first is from the author himself cor-recting certain misprints in formula (d), and thesecond is from Mr. T. L. Eckersley, showing thatformula (b) is incorrect, and that the field strengthis not proportional to z/ "H, but to t/H, the powerradiated not being independent of H.

In the Electrician of i3th August, p. 176, Mr.Turner replies to this objection.

Ri13.-WIRELESS TRANSMISSION I. & II.-CharlesE. Snell. (Electrical Review, 23rd July,p. 138 and 6th August, 1926, p. 207.)

The first article discusses the subject particularlyfrom a theoretical aspect and suggests a way ofbringing theory and practice into agreement.

The second article deals with the phenomena ofthe daylight effect and atmospherics showing thatlight is thrown upon these by the views put forwardin the first article.

The following conclusions are reached :-I. Wireless transmission is dependent partly

on the electrical conductivity of the earth's surfaceand partly on the dielectric properties of the earth'satmosphere.

2. The electrical conductivity of the earth'ssurface is invariable and is the contributory causeof the constancy of efficiency shown in daylighttransmission.

3. The variations in transmission efficiency atnight are occasioned by variations in the airdielectric brought about by night radiation fromthe earth's surface.

4. The night radiation from the earth's surfaceis a function of the sun's mean altitude so far asthis radiation influences the electrical efficiencyof the air dielectric ; other causes, however,occasion transient variations in the night radiationsufficient to obscure this relationship in isolatedobservations.

5. The transient variations in the earth's nightradiation are present to a greater degree in casesof overland transmission and in the vicinity ofland masses.

With regard to "atmospherics," distutbances atnight are stated to be frequently due to earthcurrents circulating in the neighbourhood of thereceiving aerial. It is suggested that these currentsarise from the equalisation of potential differencesbetween the materials of the earth's surface.

A letter referring to this article, by T. Barton, inthe issue of loth August, p. 301, suggests that theall-important factor in wireless transmission is itswavelength of propagation. Evidence is given toshow that greater distances of transmission can beobtained during the day than during the night,using short wavelengths.

RI I 3. I .-WI RELESS SIGNAL FADING.-E. V. Apple-ton. (Wireless World, nth August, 1926,pp. 181-182.)

A simple method of measurement and some ofthe results obtained.

RI 3. .-ON THE RELATION BETWEEN SHORTWAVELENGTHS AND POSSIBLE COMMUNI-CATION HOURS TOGETHER WITH THE COM-MUNICATION DISTANCE.-T. Nakayama, T.Ono, and C. Anazawa. (Journ. Inst. Elect.Engineers of Japan, July, 1926, pp. 695-711.)

Description of a year's investigation which leadsto the following conclusions :-

1. The wavelength of about 24 metres iscritical.

z. Waves of 25 metres or more are bestutilised for night long distance communication.

October, 1926

3. Waves of 22 metres or less are good fordaylight long distance communication.

4. The phenomenon of skipped distanceexists especially for waves below 20 metres.

13.2.-SPECIFIC STUDY OF CONDITIONS AFFEC-TING RANGE OF RADIO STATIONS.-C.Jansky, Jr. (Bulletin No. 297, U.S. Bureauof Standards.)

A paper describing the organisation of a groupof observers by the bureau, the methods employedfor making observations, and the forms used forrecording 8,50o observations made over a period ofa year (1922-23). The data obtained were analysedon automatic machines, and charts were givenshowing the variation of strength of atmospheric,variation of fading, relative magnitude of obstaclesto reception, variation of interference from receivingsets, relative magnitude of obstacles to receptiongrouped in bi-monthly periods, and mean re-liability of reception as a function of distance.

RiI3.4.-UPPER AIR PHENOMENA. (Nature, 14thAugust, 1926, p. 22x.)

An account of Dr. Dobson's paper : " TheUppermost Regions of the Earth's Atmosphere,"being the Halley lecture given on 5th May, 1926.The paper reviews the present state of our know-ledge of the constitution of the upper atmosphereand of the chief natural phenomena occurringabove the isothermal layer.

RI r3.4.-METEORS AND THE CONSTITUTION OFTHE UPPER AIR.-Prof. Lindemann. (Nature,7th August, 1926, pp. 195-198.)

The view is expressed that the persistence of theconducting layer at night is due to solar radiationforming some unstable substance, such as ozone,which gradually reacts or breaks up forming ionsduring the hours of darkness, and that the layerof maximum ionisation after sunset will move upas the ozone becomes used up during the night toregions of lower density where the rate of reactionis smaller. It is explained that it is above 5o km.that the concentration of ozone compared withthat of water -vapour and carbonic acid becomesimportant and that the amount of ozone formedpet cm.3 reaches a maximum somewhere in theneighbourhood of 6o km.

This hypothesis will account for the height ofthe Heaviside layer varying from 5o km. at sunsetto some 8o km. at sunrise which the facts observedin radio seem to require.

Rf13.4.-RELATION BETWEEN THE HEIGHT OFTHE KENNELLY-HEAVISIDE LAYER ANDHIGH FREQUENCY RADIO TRANSMISSIONPHENOMENA. A. Hoyt Taylor.-Proc. Inst.Radio Engineers, Vol. 14, August, 1926,PP. 522-540.)

The theoretical basis of this paper was publishedin the Physical Review for January, 1926. Thatpaper was in turn based upon certain experimentaldata which have been considerably extended sincethe date it was submitted for publication. Thepresent paper describes further investigation ofthe skip distance effect, and the indication that itgives as to the height of the Heaviside layer, and


the lower limit of wavelength which will be usefulin long-distance communication. The results arenot summarised. An interesting case is mentionedof a station that appears to show no skip at all.

R113.7.- ON THE ATTENUATION OF WIRELESSSIGNALS IN SHORT DISTANCE OVERLANDTRANSMISSION.-J. Ratcliffe and M. Bar-nett. (Proc. Cam. Phil. Soc., July, 1926,pp. 228-303.)

The following summary is given :-I. It is pointed out that the signal strength

of a wireless wave at a distant point dependson :-

(a) the electrical constants of the ground,(b) the curvature of the earth,(c) the existence of an " atmospheric " ray

coming downwards from the Heavisidelayer.

2. We can eliminate the effect of (b) and (c),and so obtain direct evidence about the electricalconstants of the ground, by making measurementsnear the transmitter.

3. Measurements on wavelengths of 30o metresand upwards give information about the resis-tivity of the ground, and on shorter wavelengths(15 metres) give the dielectric constant of theground.

4. Attenuation measurements have been madeover short distances for wavelengths of 1,600metres and 36o metres.

5. The results are compared with those calcu-lated from Sommerfeld's theory and show closecorrelation for distances beyond ro wavelengthsbut show deviations from the theory for shorterdistances.

They give as values for the resistivity of theground

p = 1.8 x Io18 E.M.U.for the Daventry signals, and

p = 0.6 x Iolafor the London signals.

A note is added saying that since the paper waswritten further measurements have been madewithin 20 miles of the London transmitter, whichshow that calculations based on a single value of" p" will not give as close an agreement with theexperimental results as can be obtained by assumingthat p changes with the nature of the ground.For example, the results up to 20 miles agreewith the value p=2 x roll E.M.U., whereas theresults for greater distances, as shown above,indicate the value p=-- o.6 x rola E.M.U. It isinteresting to note that such a change in p wouldbe anticipated from the known geological natureof the ground, which consists of London claysup to about 20 miles from the transmitter and ofchalk at greater distances.

RII3.8.-RELATIONS ENTRE LES PERTURBATIONSELECTROMAGNETIQUES ET LES TROUBLESSOLAIRES.-A. Nodon. (L'Onde Electrique,Vol. 5, July, 2926, pp. 359-364.)

The following summary is given :-The propagation of electromagnetic waves

THE WIRELESS ENGINEER 637 October, 192o

appears to be closely linked with the state ofionisation and conduction of the upper layers ofthe atmosphere. The ionisation seems due toultra penetrating radiation emitted by the sun andstars. This emission undergoes profound modifi-cations during solar and electromagnetic distur-bances, entailing corresponding effects in radiopropagation.

- POLARISATION CHANGES CAUSED BYGROUND ABSORPTION.-E. F. W. Alexander -son. (General Electric Review, August, 1926,PP. 553-554.)

In the reception of short waves, the signal oftenappears to come in with equal strength from alldirections. Measurements with a loop indicatethe presence of a horizontal and a vertical wavecomponent with different velocities of propagation,resulting in differences of polarisation from pointto point. The author discusses the phenomenon,using as an analogue sound waves proceeding froma loud -speaker horn, and suggests that the earthmay absorb the horizontal polarised waves andreflect the vertical ones, thus giving rise to a twistingwave spiral.

RI 4. - PERTURBATIONS ATMOSPHERIQUES ETLONGUEURS D'ONDES.-H. de Bellescize.(L'Onde. Electrique, Vol. 5, July, 1926, pp.347-358.)

The following summary is givenPropagation formula, connecting the oscilla-

tions in transmitting and receiving antenna, canonly be of practical utility when the correspondencethat exists between the electromotive forcesreceived and the degree of security obtained isdefinitely known. Now atmospherics constitute theprincipal obstacle to be surmounted, and it istherefore upon them that the values to give theseelectromotive forces depend.

It is examined whether the hypothesis on whichatmospherics vary arbitrarily is able to accountfor the results found by experiment.

A first verification based on the measurement ofelectromotive forces equivalent to the disturbancestends to prove that, on the average, these latterincrease slightly with the wavelength, while asecond method consisting of combining the hypo-thesis with Austin's propagation formula leadsto the opposite result.

It seems then, on the one hand, that the hypo-thesis can be retained, as a first approximation,and on the other hand, that current propagationformula should undergo slight alteration so as toless favour decreasing waves.

It is insisted that where the hypothesis is con-cerned, as also the verifications attempted, referonly to a mean value around which the instantaneousphenomena vary without doubt within wide limits.

RI i4. - LES ATMOSPHERIQUES. - R. Bureau.(L'Onde Electrique, Vol. 5, July, 1926, pp.301-346.)

" Conference de documentation faite it la Societedes Amis de la T.S.F., le 19 mai, 1926."

The paper, whose main theme is that atmos-pherics are not caused by storms, but ascendingcurrents of air, is divided into five sections :-

1. Methods of observation and measuringinstruments.

2. Results.3. Meteorological influences.4. Theories.5. Conclusion.

Observations on the frequency of atmosphericsare discussed, and on their intensity, directivity,wavelength and nature. The results are found toshow three distinct types of variation :-

(a) With a maximum at night.(b) With a maximum in the afternoon.(c) With any distribution of the periods of

violent atmospherics.These three types point to three distinct causes

for atmospherics and can serve as a basis for theirclassification.

The first type, which is the least important ofthe three, is thought to originate in layers ofanticyclonic inversion ; its disappearance atsunrise corresponds tgdescending currents extendingdown to the ground ; the second type is attributedto the heating of the ground by the sun and theresulting ascending currents of air ; and the thirdto the effect of atmospheric discontinuities. Allthree refer back to the vertical temperaturegradient.

The properties of atmospherics of the secondtype (grinders) are shown to follow from theirorigin (air ascending owing to solar heating ofground), for they

(a) Appear at a definite time (afternoon) ;(b) Are much more intense in summer than in

winter ;(c) Are much more violent over the continents

than oceans ;(d) Are observed chiefly in low altitudes ;(e) Diminish rapidly in intensity with altitude ;(f) Are specially violent when the meteorolo-

gical situation favours ascending currents due tosolar heating of the ground (in France pseudopolar front).The properties of the third type of atmospherics

(clicks) are also linked with their origin (convectioncurrents caused by the movement of masses of air),for they

(a) Appear at any hour of the day or night :(b) Are not confined to the warm season, but

are also observed in winter ;(c) Are characteristic of medium latitudes ;(d) Appear indifferently over sea or land,

though they can be reinforced by mountainchains which have an effect on meteorologicalperturbations;

(e) Vary in intensity with altitude, the direc-tion of variation depending upon the meteoro-logical situation ;

(f) Develop parallel with the meteorologicalperturbations (appear with cold fronts anddisappear with warm).The atmospherics coming through at any station

almost always belong to the three categories, andif they are studied en bloc it is impossible to avoidobserving a mixture of the properties of the threevarieties.

October, 1926 638

As concerns other theories on the origin of atmos-pherics, cosmic ones are not considered justifiable,and the theory of local meteorological action is setover against that of a tropical origin for atmos-pherics.

An exhaustive bibliography of the subject isappended, also a communication from Mr. WatsonWatt, made at the end of the lecture, where hebriefly indicates the newer methods of the RadioResearch Board, which lead to other conclusionsthan those arrived at by M. Bureau.

Ri14.-LIGHTNING. (Nature, 7th August, 1926,p. i9o.)

A letter by Dr. Dorsey referring to Dr. Simpson'sarticle (Proc. Roy. Soc., May, 1926), intending toshow that the positive tongue theory advancedby Dr. Simpson and the electronic dart theoryproposed by himself are not conflicting, butmutually complementary (see these abstracts,E.W. & W.E., July, 1926, p. 454.)

A reply from Dr. Simson follows in which heindicates fundamental difference between his pointof view and that of Dr. Dorsey.

RI20.-STRAIGHTENING OUT THE ANTENNA.-B.Melton. (Q.S.T., Vol. To, August, 1926,pp. 30-32)

An article " intended to straighten out ourideas on radiating systems in general."

R125. - PRACTICAL DIRECTION - FINDING . -R. L.Smith -Rose and R. H. Barfield. (WirelessWorld, irth August, 1926, pp. 193-197)

Construction and calibration of an accuratedirection -finder for amateur use.

RI 25 . I .-THE CAUSE AND ELIMINATION OF NIGHTERRORS IN RADIO DIRECTION-FINDING.-Dr. Smith -Rose and R. Barfield. ( Journ.Inst. Elect. Engineers, Vol. 64, August,1926, pp. 831-843)

A paper read before the Wireless Section, 5thMay, 1926, summarised as follows

The paper describes experiments which havebeen carried out with a view to obtaining moreconclusive evidence as to the causes of the apparentvariations in bearings observed under certainconditions on wireless direction -finders. In thecourse of the experiments the Adcock " four -aerial " direction -finder has been developed, andwith its aid it has been shown that the actualdeviation in azimuth of wireless waves is practicallynegligible. These experiments thus constitutea proof that the variable errors observed on closed -coil direction -finders at night are caused by down -coming waves arriving from the upper atmosphereand polarised with the electric force in a horizontalplane. The investigation also indicates the possi-bility of the Adcock system being developed intoa practical direction -finder which is free fromnight errors, and those errors associated withobservations on aircraft transmissions made at aground direction -finding station.

The discussion following the reading of the paperis also given.


RI 25 .6 . - DIRECTIVE DIAGRAMS OF ANTENNAARRAYS.-R. Foster. (Bell System TechnicalJournal, VOL 2, pp. 292-307.)

Two systematic collections of directive amplitudediagrams are shown for arrays of two and of sixteenidentical antenna spaced at equal distances alonga straight line with equal phase differences intro-duced between the currents in adjacent antennae,assuming that each antenna radiates equally inall directions in the plane of the diagram. Thediagrams may be used to obtain the directivediagram in any plane though an array made up ofantenna which do not radiate equally in all direc-tions in this plane but which satisfy the otherconditions. The total directive effect is given asthe product of the individual effect multiplied bythe group effect.

RI 30 . -DEM ONSTRATION OF SOME SIMPLE EXPERI-MENTS WITH THERMIONIC VALVES.-Dr.Rayner. (Proc. Phys. Soc. Lond., Vol. 38,November 4, 1926, pp. 335-336.)

R130.-OPERATION OF THERMIONIC VACUUM TUBECIRCUITS.-F. B. Llewellyn. (Bell SystemTechnical Journal, July, 1926, pp. 433-462.)

The following synopsis is given :-Given the static characteristic of grid current-

grid potential, and plate current -plate potential,for any three -element vacuum tube, the generalexact equations for the output current when thetube is connected in circuits of any impedancewhatsoever, and excited by any variable voltage,are here derived. The method of derivation isillustrated in the special case where resistancesonly are considered, and the adaptation of compleximpedance to use in non-linear equations is shown.Approximations that are allowable in variouspractical applications are indicated, and theequations are applied in some detail to grid -leakdetectors, and in brief to other types of detectors,modulators, amplifiers and oscillators.

RI3I. - VALVE CHARACTERISTIC SURFACES. - E.Harwood. (Wireless World, iith August,1926, pp. 185-187.)

Investigation of the best working conditions fora valve oscillator.

R . I 3 I .-USE OF PLATE CURRENT -PLATE VOLTAGECHARACTERISTICS IN STUDYING THE ACTIONOF VALVE C IRCU ITS . -E. Green. (E.W. &W.E., July, 1926, pp. 402-406 ; August,1926, pp 469-476.)

R138.-EQUATIONS FOR THERMION IC E MISSION.-P. Freedman. (Nature, 7th August, 1926,p. 193.)

R144.-EFFECTIVE RESISTANCE OF INDUCTANCECOILS AT RADIO FREQUENCY-PART IV.-S. Butterworth. (E.W. & W.E., August,1926, Pp 483-492.)

R i4.9. -SUR LA DETECTION PAR LES CONTACTSMETALLIQUES DETECTEUR S YMETRIQUE.-H. Pelabon. (Comptes Rendus, 28th June,1926, pp. 1605-1607.)


RI 62 .-AARIAL FILTER CIRCUITS. - P. Tyers.(Wireless World, 4th August, 1926, pp.169-x71.)

Experiments with a simple method of improvingunselective receivers.

R200.-MEASUREMENTS AND STANDARDS.R20I.-EMPLOI DE L'ELECTROMETRE A QUADRANTS

DANS LES MESURES DE PRECISION EN HAUTEFREQUENCE.-L. Cagniard. (Comptes Rendus,2 I St June, 1926, pp. 1528-I530.)

R235.-THE PROPERTIES OF MUTUAL INDUCTANCESTANDARDS AT TELEPHONIC FREQUENCIES.-L. Hartshorn. (Proc. Phys. Soc. Lond.,35, 4, pp. 302-320.)

The following abstract is given :-It is well known that when a mutual inductance

carries alternating current, the value of the effectivemutual inductance varies with the frequency, andfurther that the secondary P.D. is not exactlyin quadrature with the primary current. Thuscertain frequency corrections are introduced intoall alternating -current bridges in which a mutualinductance standard is used. An experimentalinvestigation has been made of the actual mag-nitude of such corrections for mutual inductancestandards of three types :-

I. A stranded wire fixed standard, with spacedwinding to minimise capacity effects.

2. The Campbell variable mutual inductancestandard.

3. The Tinsley variable mutual inductancestandard made in accordance with Butterworth'srecommendations. Values of the corrections aregiven for all these instruments under variousconditions. It is shown that whereas the variationof mutual inductance with frequency is mainlydue to capacity effects ; dielectric losses in theinsulation, or alternating current conductance,play an important part in the determination ofthe phase defect, the impurity being found to beproportional to a power of, the frequency higherthan the second, ever the telephonic range offrequency.

R240.-EIN NEUES RoHRENGERAT ZUR MESSUNGSEHR HONER WIDERSTANDE MIT SEINERSONDERANWENDUNGEN. (A new valvehigh -resistance ohmmeter with particularapplications.)-S. Strauss. (Elehtrotechnikand Maschinenban, 44, 19, pp. 348--355)

If the grid of a vacuum tube is charged negativelyno current can flow in the anode circuit. If a highresistance is connected beween the grid and thecathode, a negative charge placed upon the gridwill slowly leak across this high resistance. Assoon as the grid voltage approaches a certainminimum value an anode current will begin toflow, increasing proportionally to the decrease ofthe grid voltage. Parallel to the high resistanceis a fixed condenser, which will be charged by theanode current and later discharged across theresistance, the charging and discharging takingplace at definite intervals. The number of dis-charges of the condenser in a given time interval

October, 1926

"is a direct measure of the unknown resistance.A complete portable equipment based upon thisprinciple is described.

R25 I .2.-MEASURING SMALL ALTERNATING-CURRENTS.-E. Banner. (Electrician, 2othAugust, 1926, pp. 202-203.)

Description of some modifications of the Duddellthermo-galvanometer for H.T. work.

R251.3.-TRANSFORMERS FOR THE MEASUREMENTOF LARGE CURRENTS AT RADIO FREQUENCIES.-I. Maloff. (General Electric Review, Vol.29, August, 1926, pp. 555-558.)

An account of the theoretical and experimentalconsiderations leading to the development of twotypes of radio -frequency current transformer.The first type is intended for a primary currentrange of 5 to roo amps, at frequencies from to to5ookC and z,000 volts at radio frequency betweenthe windings. The core consists of a number ofvery thin ring -type enamelled laminations of highgrade iron. The secondary is wound aroundthis ring and placed in a box container, made fromspecially treated maple wood. The primarywinding is placed in the groove in the outer surfaceof the box. The second type of instrument coversradio -frequency current transformers of maximumprimary current rating from 4o to 40o amps atfrequencies from ro to 2ookC and with ro,000volts at radio frequency between the windings.The primary consists of a copper tube conductorpassing through the centre of the ring -type core,while the secondary is wound around this core.The secondary is intended to operate at groundpotential, while the primary may carry current atradio frequency with a potential as high as ro,000volts effective value with respect to ground. Alltransformers are designed for 5 amps maximumsecondary current. Illustrations of the two typesdescribed are shown.

R272.-A RADIO FIELD -STRENGTH MEASURINGSYSTEM FOR FREQUENCIES UP TO FORTYMEGACYCLES.-H. Friis and E. Bruce.(Proc. Inst. Radio Engineers, Vol. 14,August, 1926, pp. 507-519)

The apparatus is a double detection receiving setwhich is equipped with a calibrated intermediatefrequency attenuator and a local signal comparisonoscillator. The local signal is measured by meansof the intermediate frequency detector which iscalibrated as a tube voltmeter.

R800.-APPARATUS AND EQUIPMENT.R337.-MERCURY ARC RECTIFIERS.-A. B. Goodall.

(Q.S.T., Vol. io, August, 1926, pp. 8-i I.)An account of some reliable arrangements

recently developed at Washington.

R340.-THE POWER SUPPLY OF WIRELESS SETS.-P. Tyers. (Electrical Review, 2oth August,1926, p. 299.)

Present methods for obtaining filament and high-tension supply from the electric lighting mainsare stated to be complicated, not universally appli-cable to every type of receiver, and not fool proof.


A device is required which, when plugged into themains, can be substituted immediately for theordinary filament accumulator and high-tensionbattery, without necessitating any change in thewiring of a set or the use of special valves. " Any-one who succeeds in producing some arrangementat a competitive price should be assured of a veryready market."

R340. - SENSITIVE VALVE RELAY. - G. Blake.(Wireless World, nth August, 1926, p. 188.)

R340.-COMPLETE SUPPRESSION OF A SINGLEFREQUENCY BY MEANS OF RESONANT CIR-CUITS and REGENERATION. J. Stratton.(Journal of the Optical Society of America,Vol. 13, July, 1926, pp. 95-105)

Discussion of the possibility of suppressing agiven frequency by both tuned grid and tunedplate circuits. Curves are shown which indicatea greater sharpness of tuning for the tuned gridthan for the tuned plate circuit. This may beaccounted for by the fact that whereas the con-denser of the tuned grid circuit is shunted only bythe very high input resistance of the valve fromgrid to filament, the condenser of the tuned platecircuit is shunted by the internal plate resistancewhich is relatively low compared with the resistanceof the condenser. From this point of view thetuned grid is preferable to the tuned plate circuit,

R343.-PROBLEMS IN BROADCAST RECEIVERDESIGN.-P. P. Eckersley. (E.W. & W.E.,August, 1926, PP 499-507.)

A lecture delivered before the Radio Societyon 26th May, together with the discussion thatfollowed.

R343.7.-OPERATING RECEIVING FILAMENTS WITH-OUT BATTERIES.-R. Kruse. (Q.S.T., Vol.to, August, 1926, p. 25.)

R344.9-LE QUARTZ OSC ILLAN T ET SES APPLI-CATIONS 3 LA T.S.F.-J. Quinet. (RadioRevue, Vol. 5, August, 1926, pp. 114-127.)

A reprint of a paper read before the Radio Clubde France in May, 1926. The first part of the paperis devoted to the theory of piezo-electricity ingeneral and to the case of quartz in particular.The second part of the paper describes an importantapplication of the quartz oscillator (due to Prof.Langevin) to submarine signalling, and the locationof obstacles under water, depth of the sea, etc.The principle is to send out a wave towards theobstacle and receive the echoed or reflected wave,when the distance can be calculated. The methoddescribed utilises the vibrations of a quartz oscillatorof suitable dimensions as a means of creating, inthe water, waves of compression as of sound wavesbut of ultra audible frequency. The oscillator alsoacts as a receiver of the reflected wave, so thatthe time of emission and return can be studied.Illustrations and diagrams of the apparatus aregiven.

The third part of the paper first describes theapplication of the quartz crystal to various

measurements, e.g., as a manometer, electrometer,oscillograph, etc. The use of the crystal for thegeneration of oscillations is then considered. Thestandard forms of quartz oscillator circuit areillustrated and discussed, and a complete trans-mitter shown, using quartz as master control, withradio -frequency, power amplification to a tunedaerial. The application of the crystal to correctionof a wavemeter is also discussed, and the paperconcludes with notes on the preparation of thequartz for use as an oscillator.

R344.9-USES AND POSSIBILITIES OF PIEZO-ELECTRIC OSCILLATORS.-A. Hund. (Proc.Inst. Radio Engineers, Vol. 14, August,1926, PP. 447-469.)

A survey of what is known about piezo-electricoscillators, the following conclusions being drawn :-

I. Experiments with quartz plates show thatthey can be used in a valve circuit for producingradio -frequency currents of fixed frequenciesbearing a definite relation to the dimensions ofthe plate.

2. The piezo oscillator can be used togetherwith an auxiliary generator for standardising afrequency meter.

3. A single piezo-electric plate can be employedas a standard for the entire range of frequenciesused in radio communication.

4. By using special arrangements a small platecan be employed for producing audio -frequencycurrents.

5. Methods are given for grinding a plateaccurately to a given frequency.

6. Formula are given in designing plates to adirect frequency to a fair degree of accuracy.

7. Other miscellaneous applications are des-cribed.

R384.i.-A SHORT WAVE WAVEMETER.-A. E.Tubbs. (E.W. & W.E., August, 1926, pp.479-482.)

R400.-SYSTEMS OF WORKING.

R4z3.-A 20-40-80 METRE CRYSTAL -CONTROLLEDTRANSMITTER. -L. Root. (Q.S.T., Vol. 1o,August, 1926, pp. 33-35.)

R430.-BEHAVIOUR OF RADIO RECEIVING SYSTEMSTO SIGNALS AND TO INTERFERENCE(Abridged).-L. J. Peters. (Journ. Amer.Inst. Elect. Engineers, August, 1926, pp.707-716.)

The following synopsis is given :-This paper develops a point of view and method

by means of which it is possible to arrive at manyof the transient effects occurring in radio systemsby a consideration of steady-state propertiesalone. The scheme is to replace the voltages inradio receiving systems due to interference andsignals by a group of generators having the correctvoltages and frequencies. These generators canbe thought of as having been in the circuit for anindefinitely long time, so that only the steady-state response of the system need be considered.

' THE WIRELESS ENGINEER 641

The generators which replace the voltages inducedin an antenna by interrupted continuous wavestations, by spark telegraph stations, by telephonestations and by static are worked out. Thedesirable properties of radio receiving systems forreceiving various types of signals through inter-ference are arrived at and an ideal system isdescribed which may be used as a standard ofreference for judging the merits of any actualfrequency selecting system. It is shown that thisideal system reduces the interference from allsources to the smallest possible value which canbe obtained in a system which makes use of fre-quency selection to reduce interference. Thepaper thus arrives at the degree to which inter-ference can be mitigated by frequency -selectionmethods. In order to illustrate the method oftreating actual systems, calculations are given fora simple series receiver. The interference causedby transmitting stations of various types and bystatic is discussed and the factors determiningsuch interference are pointed out.

R435.-CIPHER PRINTING TELEGRAPH SYSTEMSFOR SECRET WIRE AND RADIO TELEGRAPHICCOMMUNICATIONS.-G. S. Vernam. (BulletinReprint B-198 Bell Telephone Laboratories,June, 1926.)

The paper describes a printing -telegraph ciphersystem developed during the war for the use ofthe Signal Corps, U.S. Army. The system is sodesigned that the messages are in secret form fromthe time they leave the sender until they aredeciphered automatically at the office of theaddressee. The operation of the equipment isdescribed, as well as the method of using it forsending messages by wire, mail or radio. The paperalso discusses the practical impossibility of pre-venting the copying of messages, and the relativeadvantages of various codes and ciphers as regardsspeed, accuracy and the secrecy of their messages.

11500.-APPLICATIONS AND USES.

R500.-ELECTRICAL COMMUNICATION.-H. Charles -worth. (Journ. of the Amer. Inst. Elec-trical Engineers, August, 1926, pp. 737-742.)

Report of Committee on Communication pre-sented at the annual convention of the A.I.E.E., atWhite Sulphur Springs last June.

As concerns radio telegraphy, mention is made ofa new service between the Dutch East Indies andSan Francisco and also from New York to Hollandfor handling through traffic from Java.

The work of decreasing the use of spark transmit-ters with their corresponding interferences has pro-gressed to such an extent that over 15o Americanmerchant ships are now equipped with valvetransmitters.

The use of the radio direction finder for navigationis rapidly increasing and has passed the experimentalstage. The Great Lakes are leading in this respect,due to the demonstration of the particular value ofthis apparatus in the difficult navigation problemsinherent in the Great Lakes.

With regard to radio telephony, the two-wayconversation tests between England and the United

October, 1926

States are of course referred to, of which we havehad full accounts.

The research of radio propagation is stated tohave confirmed the existence of a positive ionisationgradient in the upper atmosphere, postulatedindependently by Kennelly in America and Heavisidein England. Short wave experiments show thatthe " skipped " distance, in general, increases withincrease of transmission frequency and is greatera' night-time than by day. These phenomena areexplained by the waves received at a distance beingdeflected downwards by an upper ionised regionof relatively low index of refraction and lowattenuation.

The observed indications of rotation in the planeof polarisation and directional errors find a qualita-tive explanation when due account is taken of thefact that the earth's magnetic field, acting upon thefree electrons, will change the velocity of the waveand produce rotation. This effect is especiallymarked in the vicinity of 214 metres, which corres-ponds to the resonant frequency of an electron inthe earth's magnetic field.

Investigations of fading and signal distortionshow that they are closely related to the high degreeof frequency selectivity exhibited by the radiomedium.

An observed variation of signal strength duringmagnetic storms is significant, the effect being areduction of the normal signal strength by nightand an increase above normal by day.

That increasing use is being made of the shortwave range is evidenced by the continued increasein the number of short wave licences issued tocommercial concerns. Experimental work indi-cates the possibility of twenty -four-hour telegraphservice with short waves over distances of severalthousands of miles. The transmission varieslargely with the time of day, however, one frequencytransmitting much better for one period of the dayand another frequency for another period, so thatthe use of several wave frequencies appears to berequired to give uniform operation.

Reports are also given on radio broadcasting andthe electrical transmission of pictures.

R51o. -WIRELESS AT SEA.-J. Slee. (Electrician,3oth July, 1926, pp. 124 and 132.)

An account of recent development in apparatusemployed on board merchant ships.

R514.-A UNICONTROL HIGH -FREQUENCY RADIODIRECTION FINDER.-F. Dunmore. (Bul-letin No. 525, U.S. Bureau of Standards.)

The development of such a direction -finderenabling a ship to readily locate another ship, isdescribed, and its installation on a coastguardpatrol boat. The direction -finder coil consistsof four turns of ignition cable wound on a 20 -in.frame. A tuning unit and coupling transformerhave been designed so that the coil may be used onthe ship's receiving set without changing its tuningadjustments, which are locked in the two xookCposition. The controls necessary when taking abearing are reduced to one-that of rotating thedirection -finder coil to obtain the minimum signal.

October, 1926 642

R514.-WIRELESS POSITION -FINDING Ow SHIPS.-L. B. Turner. (Electrician, 3oth July, 1926,pp. 127 and 128.)

A discussion of the results of recent investigationby the Radio Research Board.R514.-MARINE DIRECTION-FINDING-L. Bain-

bridge -Bell. (Electrician, 3oth July, 1926,pp. 125-126.)

Description of the latest development of theRadio Communication Company's apparatus, whichworks on the Robinson principle, and the methodsof operation.

R540.-SHORT-WAVE RADIO IN THE ANTARCTIC.-L. Jenssen. (Q.S.T., Vol. io, August,1926, pp. 12-14.)

It is related how, with comparatively low poweron short waves, the Ross Sea Whaling Expeditionwas able to maintain nightly contact with therest of the world, whereas before, when the longerwaves had been relied on, the Expedition was cutoff most of the time.

R545.-LA PRATIQUE ET L'EXPERIENCE DEL'EMISSION.-M. Sacazes. (Radio Revue,Vol. 5, August, 1926, pp. 128-134.)

A note on the author's work during the pastthree or four years on small power transmitters.Descriptions are given of the results obtained withseveral different circuits, the systems shown beinga Mesny symmetrical circuit, a Colpitt's circuit,a master oscillator with power amplifier, etc.Remarks on adjustment and operation are given.R582. - RADIO PICTURE TRANSMISSION. - F.

Schroter. (Electrical World, 17th July, 1926,p. 129.)

Abstract of article in Zeitschrift des VereinesDeutsche? Ingenieure, 70, 22, pp. 725-732.

A new process of transmitting line drawings orphotographs over wires or by radio is described.Judging from the six samples reproduced, theprocess is much superior to that employed inAmerica. The perfect results obtained are ascribedto a highly efficient photo -electric cell and areceiving apparatus free from inertia. The formeris a potassium cell in a toroid-shaped glass bulbwhich reacts on reflected light instead of lightpassing through a transparency. The receivingelement is a Kerr cell, based upon the hitherto notutilised physical fact that polarised light passingthrough nitro-beniol between electrically chargedcondenser electrodes suffers double refraction, thedegree of which is influenced by the electric charge.The photo -electric cell responds to frequencies ashigh as ro5 cycles, while the Kerr receiver doesnot lag appreciably for 108 cycles. It is thereforepossible to transmit a picture 4 in. x 4 in. in sizewithin 5 seconds. It is further possible to transmita motion picture with Io frames per second, whichmakes television readily possible.

R586.-TELEVISION AN ACCOMPLISHED FACT.-A.Dinsdale. (Radio News, September, 1926,pp. 206, 207, 28o and 282.)

Description of the apparatus invented by J. L.Baird.


R600.-STATIONS : DESIGN, OPERATION ANDMANAGEMENT.

R624.-SERVICING OF BROADCAST RECEIVERS.-L. Manley and W. Garity. (Proc. Inst.Radio Engineers, Vol. 14, August, 1926,PP. 541-566.)

A paper attempting to group service problemsunder general classifications, prescribing methods.of diagnosing them followed by a prescription forcorrecting them.

R616.5.-" KDKA."-D. Little and R. Davis.(Proc. Inst. Radio Engineers, 14, 4, August,1926, PP. 479-506.)

Description of the latest equipment in use atthis station, both for regular broadcasting and for -short -wave international broadcasting and relaywork. An account is also given of the short-wavetransmitter employed for interworks telegraphservice.

R622.-WAVELENGTH REORGANISATION. (Electrical'Review, 13th August, 1926, p. 264.)

The Office International de Radiophonie, Geneva,.announces that after 18 months' work, an agree-ment has been ratified concerning the repartitionof wavelengths in Europe. The problem was to -fit 200 stations (erected and projected) in a wave-band wherein there would be only room for 99 -

if each station were given a different wavelength.It has been decided to employ 83 exclusive and i6 -

common waves. The former are divided amongthe different countries of Europe according to-

their requirements, Great Britain having 9, andthe latter will be shared by 117 low power stationsserving limited areas.

R800.-NON-RADIO SUBJECTS.534.-METHODS OF HIGH QUALITY RECORDING AND-

REPRODUCING OF MUSIC AND SPEECH BASEDON TELEPHONE RESEARCH.-J. Maxfield andH. Harrison. (Bell System Technical Journal,v, 3, July, 1926, PP 493-525.)

A detailed analysis of the general requirements of -recording and reproducing sound without appre-ciable distortion. It is pointed out, at length, how -many of the hitherto unsolved problems of soundrecording and reproduction have been readilysolved by the application of a detailed knowledgeof telephone transmission theory.

534.-THE POWER OF FUNDAMENTAL SPEECHSOUNDS. - C. Sacia and C. Beck. (Bell'System Technical Journal, July, 1926, pp.393-403.)

A continuation of work on speech power bymeans of oscillographic studies of vowels, semi-vowels and consonants. A previous paper con-sidered the characteristics of a few individualsounds from the power standpoint, but the principal'emphasis was placed upon speech as a whole. Inthis later analysis, sounds are considered individuallyon the basis of instantaneous and mean power. A.practical application of the results is suggested.


535.3 DIE PHOTOZELLE NACH ZWORYKIN. (Zeit-schrift des Vereines Deutscher Ingenieure, 70,22, p. 732.)

Description of a new photo -cell, discovered by theRussian physicist Zworykin, which is a combinationof a valve and a light cell, and can be employedto start machines, etc., since it controls relativelystrong currents.

535.3.-THE ALKALI METAL PHOTO -ELECTRIC CELL.-H. Ives. (Bell System Technical Journal,V, 2, pp. 320-335.)

537.-THE COSMIC HARNESS OF MOVING ELEC-TRICITY.-M. Pupin. (Journ. Amer. Inst.Elect. Engineers. August, 1926, pp. 758-761.)

643 October, 1926

Presidential address delivered at the annualconvention of the A.I.E.E., White Sulphur Springs,last June.537.-CONTEMPORARY ADVANCES IN PHYSICS-XI.

IONISATION.- K. Darrow. (Bell SystemTechnical Journal, v. 5, July, 1926, pp.463-492.)

621.355.02.-A NEW THEORY OF THE LEAD AC-CUMULATOR ?--Prof. Howe. (E.W. & W.E.,August, 1926, pp. 466-468.)

621.353.-Dav CELLS FOR WIRELESS PURPOSES.-R. L. Smith -Rose. (Wireless World, 28thJuly, 1926, pp. 117-119.)

Length of life that may be expected from H.T.and L.T. batteries.

Esperanto Section.Abstracts of the Technical Articles in our last Issue.

Esperanto - Sekcio.Resumoj de la Teknikaj Artikoloj en nia lasta Numero.

R000.-RADIO oENERALE.Ro5o.-RESUMOJ KAJ ALUDOJ.

Kompilita de la Radio Research Board (Radio-Esplorada Komitato), kaj publikigita lati arangokun la Brita Registara Fako de Scienca kaj Indus-tria Esplorado.

R100.-6 ENERALAd PRINCIPOJ KM TEORIO.RI27.2.-OFTAJ ERAROJ EN KONDENSATORAJ

KALKULADOJ.-E. H. W. Banner.Oni. montras, ke la kutimaj esprimoj por la

kalkulado de la kapacito de kondensatoro laic

giaj korpaj dimensioj ne kunkonsideras la modi-figon kafize de la " randaj " efektoj. Ci tiujestas nur nekonsiderindaj kiam la proporcio dela " plata surfaco " rilate al " distanco aparta "estas tre granda. Por la korekto, kiam di tio neestas, esprimoj estas donitaj, kiuj traktas prikondensatoraj de rondaj kaj rektangulaj platoj.Oni ankaii rimarkigas pri la efikoj de la vera dikecode dieleltriko kaj de variado je la valoro de dielek-trika konstanto.RI61 & R162.-LA AMPLIFADO KAJ SELEKTIVECO

DE NEOTRALIGITA AGORDITA CIRKVITOANODA.-D-ro. N. W. McLachlan.

Larga revuo de la nefitraligita agordita cirkvitoanoda lau la duala vidpunkto pri la faktoroj kiujregas Stupan amplifadon kaj selektivecon respek-tive. Post enkonduka revuo pri la nefitraligaprincipo, la afitoro pasas al amplifo de la kom-pleta atupo. Ekvivalentaj cirkvitoj estas diskutitaj

matematike kaj rezultoj montritaj por efektivajvalvaj amplifigoj dum kapacito trans fiksa bobenoestas altigita por plilongigi la ondolongon, kaj dumkapacito estas altigita kaj indukteco malgrandigitapor teni la agordecon je konstanta ondolongo.La rezultoj de valva rezisteco p kaj volta faktoro mestas ankafi diskutitaj, kaj oni montras, ke amplifomalgrandigas de pligrandigo de p, se m restaskonstanta, kaj grandigas de pligrandigo de m, se prestas konstanta. Konceme selektivecon, onimontras rezonancajn kurvojn por ilustri la diver-sajn statojn diskutitajn. Lau nuna okazo, onikonkludas, ke la rezultoj de l'faktoroj cititaj estasguste kontrafiaj de iliaj rezultoj je l'okazo deamplifado.

R200.-MEZUROd KAJ NORMOZ.R218-SENDADOJ DE SENFADENAJ ONDOJ DE

NORMAJ FREKVENCOJ EL LA NACIA FIZIKALABOREJO (Brita).

Komunikajo de la Brita Registara Fako deScienca kaj Industria Esplorado.

Pro la rekomendo de la Radio-Esplorada Komi-tato de la Departemento, plioftaj sendadoj deondoj de normaj frekvencoj estas farotak el de laNacia Fizika Laborejo, Teddington, Anglujo -(Voksignalo 5HW).

La sendadoj estas lafijene.:-(A) Mallongonda programo dum la unua

mardo en diu monato, de 1,500 horo gis 1,600horo, Tempo de Greenwich, la frekvencoj sendo-taj estanta 26o, 300, 36o, 500, 58o, 700, 84o, kaj96o kilocikloj.

October, 1926 644

(B) Longonda programo dum la tria jafido enCiu monato, la frekvencoj sendotaj estantaj 3o,40, 5o, 66, 86, 115, 160 kaj 200, kC.Oni faras rimarkigojn pri la ricevado de normaj

ondoj kaj ilia utileco dum normigado.

R251.-KONTINUKURENTAJ INSTRUMENTOJ EN SEN-FADENAJ APARATOJ.-J. F. Herd.

La afitoro unue konsideras la bezonon porpreciza scio pri kontinukurentaj valoroj kajalgustigoj te senfadena ricevilo, speciale se kvantalaboro estas necesigita. Li tiam montras kiel,per arango de jakoj (de la ordinara telefona modelo),unu K.K. instrumento, kiel ekzemple, tafiga mili-ampermetro, estas uzebla por apliko al diversajcirkvitoj. La arangoj montritaj enhavas mezuronde anoda kurento, de filamenta of potenciometratensio, filamenta kurento, k.t.p. Oni ankatialdonas generalajn rimarkigojn pri l'arangoj.

R.251.-AMPERMETRA PANELO. E. C. Atkinson.La artikolo priskribas panelon por utiligi amper-

metron por diversaj celoj. Pere de 'Stopiloj (lafimodelo de Wheatstone'a Ponto), plenskalaj gradarojde 3.o, o.6, kaj 0.024 amperoj estas haveblaj, dumla instrumento estas ankafi utiligebla por legi 3kaj 12 voltojn. La §arga kurento por akumulatoroestas ankafi montrebla sur unu skalo, kaj fila-menta kurento sur alia, sen Aango de konektajoj.

R300.--APARATO KM EKIPAJO.R384.-APARATO POR KONTINUKURENTA PROVADO.

-T. S. Skeet.La artikolo pritraktas la utiligon de movbobena

instrumento (Weston'a Modelo 301, plena skalomontranta gis 5 m.a.) kiel voltmetron montrantanr000 v., zoo v., zo v., 5 v., 0.5 v., kaj o.r v., kaj kielampermetron montrantan roA., 5A, 2.5A, 5oomA,kaj 5omA. La instrumento estas muntita enskatolo kun bomoj kaj komutatoroj por la reguligode la bezonitaj seriaj kaj Auntaj rezistancoj, kajla kompleta instrumento estas ilustrita kaj pris-kribita.

R.385.-LA KLAVARANo0J DE VALVAJ SENDILOJ.-W. T. Ditcham.

Bonega priskribo estas donita pri modernapraktiko rilate al la enkonduko de klavaj interup-toroj en valvaj sendiloj. Tiaj sendiloj estasdividitaj en du klasojn, (A) Memekscitaj oscilatoroj,(B) Amplifikatoroj ekscititaj de mastra oscilatoro,kaj oni donas ilustrajojn de diversaj klavaj motodojpor Ciu. En la unua klaso, la metodoj montritajenhavas arangojn por utiligo de aparatoj kis 5 K.V.,


kaj por potencoj gis 20 K.V., je rapideco de 6ovortoj eiuminute, kaj de Pao v.e.m. En la klasode mastra oscilatoro, oni montras metodojn porklavi longtrafajn rapidegajn komercajn staciojnuzantajn rektifitan Alternkurenton por potenco-provizo, ekzemple, la Marconi'aj stacioj Ce Ongar,Essex, Anglujo. Fine, oni diskutas klavajn meto-dojn por altpotencaj stacioj provizitaj per Kon-tinukurenta Alta Tensio. Klaraj ilustrajoj pri ladiversaj metodoj estas presitaj tra l'artikolo.

R500.-APLIKOJ KM UZ0d.R545.009.2.-AMATORA LONGDISTANCA FUNKCIADO.

-H. N. Ryan. Perioda kontribuajo pri latemo de longdistanca funkciado.

R600.-STACIOJ : DESIGNA, FUNKCIADO,KM ADMINISTRADO.

R616.5.-RADio-VIENo.-Prof. G. W. 0. Howe.Tre kompleta priskribo pri la nova altpotenca

Brodkasta Stacio ee Vieno. Oni donas planojnde la antena kaj kontrapeza arangoj, dum dia-gramoj montras la konektajojn de la sendilo,amplifikatoroj, modulaj metroj, k.t.p. Fotografajilustrajoj montras la staciajn konstruajohn, lasendilon, valvojn de la diversaj tipoj uzataj,reguligan panelon, amplifikatorojn, modulajnmetrojn, k.t.p.

La stacio uzas la mastro-oscilatoran funkcime-todon, kun potenca amplifado per akve malvar-migitaj valvoj. Rektifita Altemkurento provizasla altan tension. La senilo estas ses mejloj mal-proksime de la mikrofonejo en Vieno, al kiu giestas ligita per speciala kablo. La aparato Ce lamikrofoneja konstruajo konsistas el, krom lameikrofono, trigtupa amplifikatoro, plua unu§tupaamplifikatoro, amplifikatora modula metro kon-trolanta la parolan kurenton senditan tra la kablo.Estas ankafi sendila modula metro funkciigita dekurentoj senditaj tra la kablo el de la senda stacio,tiel ke la antena modulado estas observebla Ce lamikrofonejo.

R800.-NE-RADIAL TEMO7.510.-MATEMATIKO POR SENFADENAJ AMATOROJ.

-F. M. Colebrook.Datirigo de la serio el antatiaj numeroj. La

nuna numero traktas pri Indicoj, kun la Difinode Indico, Indica Lego, Negativaj Indicoj, RipetitajProdutoj de Potencoj, Produtoj de Potencoj kajkvocientoj, Radikoj, Produto de Radikoj, la PlenaGeneralizo de Indica Formularo, k.t.p:


Correspondence.

October, 1926

Letters of interest to experimenters are always welcome. In publishing such communicationsthe Editors do not necessarily endorse any technical or general statements which they may contain.

Plate -Current, Plate -voltage Characteristics.To the Editor, E.W. & W.E.

SIR,-Your contributor Mr. E. Green makes anerror on pp. 370, 475 of the August issue of E.W.&W.E., which is a very common one. The valveis a voltage operated device, and we spend ourexistence trying to get the biggest voltage amplica-tion from valve to valve.

No power is present in the secondary of a trans-former or tuned coupling unless there is gridcurrent or charging current and in either of thesecases the grid filament voltage is cut down by theI R voltage drop.

No power is taken from the primary and so the,transformation is a pure voltage transformationand the greater the voltage across the primarythe greater the overall amplification. The greaterthe primary impedance the greater the fraction of

that is utilised.Theoretically then, infinite impedance and an

infinite step up would be ideal. In practice oneis limited by intercapacities and space considera-tions, but still the main law holds good that thetransformer should be of the utmost possibleimpedance and not of equal impedance with thevalve as it would have to be for maximum poweramplification.

The matter is totally different when the lastvalve driving the loud -speaker is under consideration.Here power is required to move the air and thebest power output is obtained when the loadimpedance equals that of the valve.

Here again, practical details upset any attemptat maximum efficiency. The matching ofimpedances only holds good at one frequency, theimpedance is changed at every alteration of theair -gap in the loud speaker, and the makers giveus only the D.C. resistance in ohms

In support of my views of the above, Iwould refer you to articles which have appearedin the Wireless World, " The Resistance -CoupledAmplifier," by Dr. H. Kroncke, 23rd September,1925. " Everyman's four -valve," by W. James,28th July and 4th August, 1926. They are usingresistance amplification with a resistance ofmegohm, grid -leak of 5 megohms, couplingcondenser of .0003µF, and a plate current ofa few microamps. They are in fact employingthe principles of pure voltage amplification.

I do not know who started the error originally,but it is certainly time it was laid to rest..

I. A. J. DUFF, B.A., A.M.Inst.C.E.

To the Editor, E.W. & W.E.SIR.-With reference to the remarks of your

correspondent, Mr. I. A. J. Duff, we are both agreedthat the valve is a voltage operated device, and the

biggest voltage amplification from valve to valveis what is normally aimed at. I note that the twoarticles he mentions as supporting his views ontuned circuit and transformer amplifiers, bothdeal with resistance amplification.

In the case of a tuned circuit amplifier, as shownin my figure 6 (b) (c) and (d), used for amplificationat one definite frequency, the power output ofthe valve is used to maintain the oscillating currentin the tuned circuit, and obviously maximumpower output will give the maximum voltageacross the condenser.

The case of the transformer amplifier, as Mr.Duff says, is not so simple, and the theory I gavewas reproduced from memory from an article byMr. Catterson Smith published in the Radio Reviewof July, 1920. I find on referring to the articlethat he did not consider the case where gridcurrent was eliminated.

The simple theory that the transformer loadis made equal to the anode A.C. resistance of thevalve must be abandoned, nevertheless Mr. Duff'sstatements require some correction. If they weretrue we could fix on a standard transformer whichwould be best for all types of valve. In fact,however, the step-up ratio of the modern type oftransformer is arranged so that the transformerload has a definite ratio to the anode A.C. re-sistance of the valve in use. In proof of this Iwould refer Mr. Duff to an abstract of a paper byMr. P. W. Willans, on " Low Frequency Inter -valve Transformers," published in E.W. & W.E.for July, 1926. The transformers described weremade with a fixed secondary winding, and as regardsthe primary winding, I quote the following frompage 437-" Four different ratios were specified,i.e., 2.7 : I for a 40,000 ohm valve, 4 : 1 for a 16,000-18,000 ohm valve, 6 : I for a 7,000 ohm valve,and 8 : r for a 4,000 or 5,000 ohm valve."

The remarks on the valve driving the loud -speakerdo not call for any comment.

E. GREEN.

Valve Nomenclature.To the Editor, E.W. & W.E.

SIR,-With reference to your paragraph on" Valve Nomenclature " (differential resistance)in your Editorial of the current issue, I supposeit is too late, now, to raise objections to the term" Anode A.C. resistance." But it appears to methat, though this term is better than " impedance,"it is still misleading, and ambiguous.

If we consider the case (which frequently occurs)of a valve with anode differential resistance 12,having an inductance L in its anode circuit-thestatic differential resistance of the whole circuitwill =Ra, but the dynamic dif. res. will =R,

The expressions "dynamic" and "static differential


resistance " seem to me to be reasonable ; but ifwe substitute " A.C. resistance " for " differentialresistance," we have " static A.C. resistance,"which is a contradiction in terms.

This objection seems to me to be real and notmerely superficial. The quantity " differentialresistance " is not dependent on A.C.

If " differential resistance " is too long or obscure,would not " slope resistance " be better than A.C.resistance ?

Surrey. R. ST. 9. LENG.[Our correspondent is incorrect in assuming that

the whole circuit would have a " dynamic differ-ential resistance " of Ra 0Z. This is theimpedance of the circuit at the frequency w/za ;there is no such thing as a dynamic differentialresistance, it is essenially static, being determinedfrom the slope of the static characteristic curve.It is for this reason that we think that either" differential resistance " or " slope resistance "preferable to the adopted term " A.C. resistance."If a valve had any property resembling skin effect,the A.C. resistance would not be exactly the samething as the differential resistance, but so far aswe know the two things are identical.

It is not too late, however, to raise objectionsto the contents of the Glossary of Terms used inElectrical Engineering, recently issued by theBritish Engineering Standards Association. Thisfirst edition is regarded as tentative and we our-selves have already objected to many of the termsand definitions.-G. W. 0. H.]

Power Losses in Insulating Materials.To the Editor, E.W. & W.E.

SIR,-In the Editorial in the current number ofthe E.W. & W.E., there is a reference to a paper ofmine entitled " Parasitic Losses in InductanceCoils at Radio Frequencies." You mention thepaper of Mr. E. T. Hoch on the subject.

I would like to draw your attention to the sequelof my paper, which appeared in the followingnumber of E.W. & W.E. (June, 1925) in which Ispecially mentioned Mr. E. T. Hoch's paper (page543) and I quoted the first part of the paragraphthat was quoted in your Editorial.

I remember being astonished at the time that soobvious a criterion of the suitability of dielectricsas the product of power factor and dielectric con-stant was not more generally used and I considerthat it would be advantageous if this should becomestandard practice.

London, S.W.io.RAYMOND M. WILMOTTE.

A New Theory of the Lead Accumulator.To the Editor, E.W. & W.E.

SIR,-In your issue of August, on page 468, Dr.Howe describes a " cell which cannot sulphate,"which seems likely to be useful for H.T. supply.This appears to be another case of lost knowledge

rediscovered, for I remember seeing many yearsago an antique " Rochefort " station at New-haven, Sussex-with a spark like a brick in a bucketand a log. dec. like X's -where they had a battery,not unlike that described.

The plates were in the form of saucers and wereabout a yard across, and were stacked up oneinside another, slightly separated by porcelainpieces, with the under side of one in a pool ofacid held by the next lower one. This battery wasnot of negligible capacity, however, as it was usedto run a motor alternator.

It appears that the arrangement-which was,I think, of French origin-would fulfil the require-ments of M. Fery's new theory as regards freedomfrom sulphation.

It would be perfeCtly easy to substitute acidsoaked kieselguhr for the earlier free acid sincethis silicious skeletony stuff has long been usedas an absorbent for sulphuric acid, nitro-glycerineand other tasty fluids.

I have much pleasure in sending this recollectionfor publication in the hope that the Patents Officewill give a suitable reception to the inevitablemillionaire corporation when it arrives from theland of "bounce "'to conduct a zoo per cent. cam-paign against the pockets of our wireless enthusiasts.

Ashford, Kent. WM. A. RICHARDSON.

Fading and Mountains.To the Editor, E.W. & W.E.

SIR,-I would like to explain to Mr. Maddisonthat I introduced the subject of D.C. conductivitybecause it is a property of real importance whenconsidering a reflecting medium. Moreover it canbe and has been measured. For this reason myargument was hardly " futile."

I am sorry to say this manner of so lightly dis-missing others' statements pervades Mr. Maddison'ssubsequent remarks. He makes, in the most casualway, the following exceptionable statements :-

" The only limiting factor of this risein pres-sure is the rate at which the charge will leak-"(The factors limiting the extent of coalescencelimit the potential rise.) " The air around thecloud will be very highly ionised-" (Why ?)" The extent of this ionisation can rise far higherthan in the Heaviside layer for this reason."(What reason ? Not the following sentence,which is incorrect). "-the Heaviside layeris only ionised by electronic bombardment fromthe sun." (Only, indeed 1).

One may be excused for hazarding the opinion thatunsupported statements like these are responsiblefor Mr. Maddison's " first and foremost difficulty."

Nevertheless, I still agree with Mr. Maddisonthat cloud may have some effect which should beinvestigated. His implied conception of a cloudas (may we say) a multitude of minute condensersis new. We must ask the mathematicians.

A. WOODMANSEY.Harrogate.

ik


Some Recent Patents. [R008The following abstracts are prepared, with the permission of the Controller of H. M. Stationery Office, fromSpecifications obtainable at the Patent Office, 25, Southampton Buildings, London, W.C.2, price II- each.

A SHORT-WAVE OSCILLATOR.(Convention date (U.S.A.), 17th February, 1925.

No. 247,942).C. F. Burgess Laboratories and W. H. Hoffman

describe in the above British Patent the construc-tion of a rather interesting short-wave oscillator,in which the grid and anode circuits are con-nected in bridge formation. The invention is ofeven more interest as it states that very highfrequency oscillations, corresponding to a wave-length of a few metres, can be generated with avalve of the normal type, the inter -electrodecapacity not being of any great importance. Theaccompanying illustration shows the generalarrangement of the circuit, and it will be seen thatthere are two similar oscillatory circuits L1 C1 andL2 C2. The two capacities C1 and C2 may bevariable condensers mounted on a common shaftso arranged that rotation of the shaft gives sub-stantially equal variation of the capacity of thetwo condensers. The circuit L1 C1 is included inthe grid -filament circuit of the valve V, and thecircuit L2 C2 is included in the anode circuit ofthe valve V. The stopping condenser K of abouto.003ILF is employed to isolate the positivepotential of the high tension battery, which isconnected between the filament F and one end ofthe coil L2. A resistance R, having a value ofabout 5,000 ohms, is connected between the mid-point of the condensers C1 and C2, and one end ofthe coil L1 and the far side of the stopping con-denser. Thus it will be seen that the resistance R

is connected across points of substantially equalpotential. The patent specification is fairly de-tailed, and gives considerable information regardingthe arrangement of the circuit and the values ofthe components. For example, it is stated thatthe inductances L1 and L2 may consist of half aturn of No. 10 s.w.G. copper wire, the radius ofthe turn being about two inches. An inductanceL3 is shown coupled to the inductance L1, thisinductance L3 being used to transfer the outputof the generator to an aerial or other load circuit.

A LOW -CURRENT RESISTANCE AMPLIFIER.(Application dates, 7th April, 9th July, 13th July,

and 5th November, 1925. No. 256,299.)Several resistance amplifiers have recently been

devised in which the anode circuits contain resis-tances of the order of 5 megohms, and the aboveBritish Patent, which is the subject of four pro-visional patent specifications, and is granted toA. H. Midgeley, discloses the arrangement of anamplifier which is the result of experimental workon the lines of the provisional specifications.

While the invention can be carried out by the useof ordinary valves, it is preferred to use a valvein which the anode is about lir inch long, and

inch in diameter, the valve being providedwith a cathode designed to take a current ofhalf an ampere at o.8 of a volt. The accom-panying illustration shows one form of circuit, inwhich the input circuit of the first special valvecontains an inductance L in the form of a frametuned by a variable condenser C. The first threevalves, V1, V2 and V3 are of the special typereferred to, and contain anode resistance R of theorder of 5 megohms. The anodes are coupled tothe succeeding grids by means of coupling con-denser K, no grid -leaks being employed. Inoperation the filament emission is reduced to apoint such that saturation occurs at zero grid volts.The potential of the high tension battery may be ofthe order of 5o volts. The last valve, V4, whichoperates the telephones or loud -speaker, is of theordinary type, and is provided with a grid -leak R,which is connected to the grid of the precedingvalve. The specification states the manner inwhich the amplifier may function. It is believedthat the first three valves all act as detectoramplifiers. It is further suggested that there isa high -frequency component throughout theamplifier which has a ripple upon it of progressivelydiminishing amplitude in each stage, while there is

October, 1926 648

also a low frequency component which increasescorrespondingly in amplitude. The specificationis extremely lengthy and detailed, and givesconsiderable practical information and suggestedmethods of working.

AN ELLIPTICAL DIAPHRAGM LOUD -SPEAKER.(Convention date (U.S.A.), 3oth July, 2924.

No. 237,878.)The Western Electric Company Limited and

C. E. Lane describe in the above British Patent arather interesting type of cone diaphragm loud-

speaker. Readers are, nodoubt, familiar with thepatent specification of theWestern Electric Company's" Kone " loud -speaker, whichconsists essentially of twocones joined at their bases,one cone being truncated andfixed to a support, a tele-phone driving mechanismbeing located within twocones and driving the apexof the cone by means of alink, a substantially free edgebeing obtained. It has beenfound that with a symmet-rical arrangement of thisdescription, although re-sponding very uniformly toboth the upper and lowerfrequencies of the musicalband, a certain amount ofresonance is apt to be set upunless the symmetrical dia-

M phragm is of very consider-able area. This difficulty isovercome by the use of adiaphragm of elliptical for-mation. Thus the generalarrangement should be clearby reference to the accom-panying illustration. Heretwo cones C1 and C2 arejoined at X, X, the conesbeing of elliptical formation.The cone C2 is truncated,

and is fixed to a supporting member M providedwith a base B. The telephone driving mechanismT, such as the well-known balanced armature,drives the apex of the cone C1 by means of alink L. It is claimed that by an arrangement ofthis description any resonance effects due to sym-metrical vibration are overcome, and thus a morelinear characteristic and substantially undistortedreproduction can be obtained.

A RADIO -FREQUENCY AMPLIFIER.(Convention date (Belgium), 28th October, 1924.

No. 241,530.)A rather interesting form of two -valve receiver

which should be of interest to experimenters isdescribed in the above British Patent, which hasbeen granted to J. Abele and J. H. Berens, Theaccompanying illustration should make the in-vention clear. The first valve V1 acts as a radio -


frequency amplifier, and has connected betweenthe grid and filament a tuned circuit L1 C1. Thesecond valve V2 is used as a rectifier and regenera-tive amplifier. The anode circuits of the twovalves contain respectively radio -frequency chokesZ1 and Z2. Incoming potentials applied betweenthe grid and filament of the first valve cause ampli-fied potential to be produced across the impedanceZ1, from which they are transferred by a con-denser C3. Another tuned circuit L2 C2 is provided,and is arranged so as to act as an auto transformer,the far side of the condenser C3 being connectedto a point E, while the filament is connected to apoint F, one end of the tuned circuit L2 C2 beingconnected directly to the grid of the valve V2,and the other end taken through a variable con-denser C2 to the top of the choke Z2. The amplifiedpotentials occurring across Z1 are then transferredto the primary of the auto -coupled transformer,these potentials being detected, and regenerativelyamplified by the valve V2. It will be noticed thatthe contact point F of the filament is variable,

and may constitute a slider working over theinductance L2. Reaction or self oscillation canbe controlled by the reaction condenser and theposition of the filament slider F. The specificationstates that the circuit possesses many advantages,and the degree of reaction is not varied by alteringthe position of the two tuning condensers. Therectified currents are transferred by means of alow frequency transformer T included in the anodecircuit of the second valve.

MULTIPLE EARTH CONNECTIONS.(Application date, 4th May, 1925. No. 256,317.)

It is well known that the use of a multi -earthconnection to a radio -frequency receiving systemwill bring about considerable loss of efficiencyunless the earth system is balanced. (It is usuallypreferred to employ one earth connection.) Whena broadcast receiver derives its anode current fromelectric light supply mains trouble of this nature


may arise due to the fact that one side of themains may be earthed. The above British Patent,granted to Fullers United Electric Works Limitedand A. P. Welch, overcomes this difficulty by theintroduction of radio -frequency chokes, whichtend to keep any radio -frequency currents fromthe earth connection of the supply mains. Theaccompanying diagram illustrates a circuit in

-../VVVV

October, 1926

mounted on a glass stem or foot S, and a glasstube G surrounds the lower portion of the anodewire so as to prevent any discharge taking placebetween the bottom of the anode wire and theedge of the plate. The leads from the two anodesand the cathode are brought out to fins on the baseof the tube in the usual manner. The tube isfilled with a gas which, at the operating temperature

,

xorrO-OW

which a number of radio -frequency chokes areinserted in various parts of the circuit. Anordinary two -valve receiver is shown providedwith an aerial circuit consisting of a variable con-denser C1, an inductance L1 and a normal earthconnection at E through a stopping condenser,the earth on the mains supply being shown at F.Two radio -frequency chokes Z, Z are shown in themain high-tension leads. Owing to appreciableself capacity across the primary of the first inter -valve transformer T, or the existence of a by-passcondenser across it another choke is shown insertedat Y. Chokes are again shown inserted at Xbetween the filament leads to the two valves.Again another choke is shown at W, and is used inconjunction with a parallel wave -trap circuitL2 C2.

A GAS DISCHARGE RECTIFIER.(Convention date (U.S.A.), 17th July, 1924.

No. 237,236.)An exceedingly interesting form of full -wave

gas discharge rectifier is described in the aboveBritish Patent by the British Thomson -HoustonCompany Limited and E. E. Charlton. Thenovelty of the invention lies partly in the shapeand arrangement of the electrodes, and partly inthe filling of the tube. The accompanying illus-tration shows a very simple form of the rectifyingtube, in which the cathode consists of a plate Pof nickel, molybdenum, or tantalum. The anodesconsist of two wires W, only one of which is visible,of tungsten or nickel, tungsten being preferableowing to its suitability for sealing into the glassfoot. The cathode P and the anodes W are

F

of the device, has a density sufficiently high toproduce the necessary conductivity by ionisation.In one embodiment of the invention a mixture ofhelium and neon is employed, but it is stated thathelium alone gives better results. The best pres-sure of the gas depends upon the geometricalfeatures of the device and the voltage at whichit is desired to work. It is stated that from 5 to

8o mms. of mercury are pressures which are foundto be suitable, but helium alone at a pressure of19 mms. appears to be preferable. The specifi-cation states that it is advisable to place in thetube a material which has chemical reactiontowards gaseous impurities. Metallic magnesiumand calcium are mentioned as being suitable. Itis stated that the voltage drop is lowered by about20 per cent. by observing this precaution, and it

October, 1926 65o EXPERIMENTAL WIRELESS ec

may be lowered still further by admitting an alkalinemetal such as caesium or rubidium before themagnesium getter is vapourised. The alkalinemetal just mentioned is oxidised by admittingoxygen and warming the tube. The oxide thenappears to combine with any hydrogen which maybe liberated during operation of that tube. Thespecification, which is very extensive, gives detailsfor the construction of high power tubes in whichthe cathode is made in the form of two cylindersmounted side by side enclosing the two anodes.

AN INTERESTING CRYSTAL PATENT.(Application date, 26th March, 1925.

No. 251,078.)A rather interesting modification of the carbo-

rundum crystal detector associated with a valvereceiver is described in the above British Patentby the Carborundum Company. If a crystaldetector is used for a powerful valve amplifier itfrequently happens that the magnitude of thepotentials is such that the detector is not capableof rectifying them properly. It has been found,however, that by suitable treatment of the crystalit can be made to rectify efficiently voltages ofan appreciable magnitude. A piece of carborun-dum or silicon carbide is freed on its surface fromany impurities, and one end is coated or sprayedin some known manner with a fine metallic layerof copper or silver, the layer being further builtup and soldered into some convenient holder.

H 8

F

C

B

The rectifying end of the crystal is then used inconjunction with a steel plate with which it isheld in contact by considerable pressure. Asuitable arrangement of the detector is shown inthe accompanying illustration, in which it will beseen that the carborundum C is attached as men-tioned to a block of metal or solder F, while theother end of the crystal is in contact with a steelplate S, controlled by a helical spring H, the wholebeing contained within a tube T provided withterminals B.

A FOUR -ELECTRODE VALVE SUPERSONICCIRCUIT.

(Application date, 5th March, 1925.No. 252,789.)

A rather interesting arrangement of two four-electrode valves as a frequency changer is describedby H. Andrews and the Dubilier Condenser Com-pany, Limited, in the above British Patent. Theinvention should be quite clear by reference to theaccompanying illustration. The second valve V1is arranged as a generator of continuous oscillationsto provide the local source and an inductance L1and capacity C1 tuned to the desired local frequency

are connected between the anode A, and the innergrid GI. The oscillations thus produced are trans-ferred to the outer grid G4 of the first valve V,.The incoming oscillations are introduced to theinner grid circuit of the valve V1 by means of thetuned circuit L3 L3 Cs, a grid condenser and leakC, R being used for rectification purposes. Theoscillations introduced to the first valve from thesecond modulate the anode current of the firstvalve, this anode current further being varied bythe applied grid potentials from the circuit L2 C.Thus a beat frequency is obtained in the anodecircuit of the first valve, potentials at the beatfrequency being produced across the tuned circuit./.494, from which they are passed on to an inter-mediate amplifier at L3. The specification alsodescribed a somewhat similar arrangement in whichonly one valve is utilised, one grid being used forthe generation of oscillations, and other for dealingwith the input voltages.

Readers of E.W. & W.E. would greatly assist us in our Publishing arrangementsif they would PLACE AN ORDER WITH THEIR NEWSAGENTS for theregular supply of the Journal, instead of purchasing their copies casually.

worldradiohistory.com€¦ · awnt le &the wireless engineer vol.. iii. october, 1926. no. 37....

Documents