128 PHILIPS TECHNICAL ,REVIEW VOLUM.E 18

COLD-CATHODE TRIGGER rrUBES

by C. H. TOSSWILL *).

The triode cold-cathode glow-discharge tube made its first appeara':lce about twenty years agoas an alternative to the hot-cathode thyratron: it consumed no stand-by power and was immuneto heater or filament breakage, but could only be used where the current demand was small andwhere the 'high maintaining poteruiol and slow decay of the cold-cathode discharge were accept-able. These triodes have found many applications, notably as selectors in remote party-linetelephone-switching; as counting elements in rings and other arrays; and as informationstoragedevices, where their irreversible triggering action has been exploited. This article describessome of the improvements made in recent years in the design of cold-cathode trigger tubes.

Introduction

The work to be described here began in responseto certain criticisms of existing cold-cathode tubes.It has had two.distinct stages: first, the design of ageneral purpose experimental cold-cathode tubewith more closely-defined and stable characteristics,and second, the development of two specialisedforms suited to service in radiation monitors 1).

"

Principle of operation

The triode cold-cathode tube is founded upon twofundamental properties of the low-pressure glow-discharge: first, the potential difference across agap needed to initiate a glow-discharge is greaterthan that subsequently required to maintain it, and,second, the initiating or breakdown potential can-be reduced by the insertion of a subsidiary ortrigger discharge, Therefore, if a potential lyingbetween the breakdown (Vb) and maintaining (Vm)values is applied to the main 'gap, no discharge willpass until a suitable trigger discharge is supplied.The main discharge will then continue until thevoltage across the main gap is reduced to below the

,.maintaining potential Vm-As the current carried bythe main discharge may greatly exceed that carriedin the triggèr discharge, a power gain is available.

In a triode, the 'trigger discharge passes betweenone main electrode and a third, "trigger",' electrode,and has an effect dependent upon its duration andthe tube geometry. The relation -hetween minimumsteady trigger current (It) and main gap or anodepotential (Va) shown in jig. 1 applies to tbe casewhere both discharges have a common cathode, andis often known as the transfer characteristic. Certain

*)' The work described in this article was carried out atMullard Radio Valve Co., Mitcham, Surrey.

1) E. Franklin and J. Hardwiek, The design of portablegamma and beta radiation measuring instruments, J. Brit.Inst. Radio Engrs. 11, 417-434, 1951. D. Taylor, Radiacinstrumentation, J. sci. Instr. ;;!9,315·322, 1952.

621.373.444 :621.387

time delays, the statistical and formative lags (see 'below), affect the development of the trigger dis-charge itself. Always, however, the influence of the'trigger is restricted to the beginning of the maindischarge; after this, the trigger, has no effect and theonly means of interrupting the discharge is to lowerthe main gap potential beneath Vm. Such reduction

Vam -----------------

-It 88542

Fig. 1. Transfer characteristic of a trigger tube. This curvegives the minimum steady trigger current It necessary at agiven anode voltage Va to initiate breakdown. Once break-down occurs (at the line dividing regions A and B) the'tube continues to conduct so long as the anode voltageremains above the maintaining potential Vam.

causes the gas to revert to the condition prior to thepassage of the discharge, to be "de-ionized", in acertain characteristic time. In the method of ex-tinction and in the irreversible áction of the controlelectrode the cold-cathode triode thus resembles thehot-cathode thyratron.

The breakdown characteristics of cold-cathodetrigger tubes may be represented by a closed loop

"

1956/57, ~o. 4/5 COLD-CATHODE TRIGGER TUBES

such as that shown infig. 2 which refers to a triode,the Mullard Z 300T (fig.3). Points lying inside thisloop correspond to anode and trigger voltages whichdo not cause a discharge. In order to fire the tube,the anode voltage andlor trigger voltage must changein such a way that the loop is crossed somewhere.

b

i100

a

-100 -50-Vt

d

-100

e88543

Fig. 2. Breakdown characteristic of Z 300T tube. For the tubeto fire the closed loop must be crossed somewhere. The config-uration of the trigger discharge depends on where the loopis crossed (regions a-f, see text). '

. I

Once the line has been crossed and the dischargeinitiated, the anode and trigger voltages may oncemore correspond to a point inside the loop. Thesection a of the loop refers to the firing of the tubeby a discharge from trigger to cathode (directionof positive "current flow). It is evident from thegraph that the triggering voltage required is here.independent of the anode voltage. A discharge fromanode to cathode occurs when section b of the loopis crossed. Section c refers to a discharge from anodeto trigger, section d from cathode to trigger, sectione from cathode to anode and section f from triggèrto anode.The barium oxide layer on the cathode of the Z300T

tube gives it a work function lower than that of theanode, so that the tube has rectifying properties.For given supply voltage and series resistance thecurrent flow from anode to cathode and trigger to ,cathode is therefore larger than in the reversedirection. For this reason, this tube is' operated in

quadrant, 1 of fig. 2, where anode and triggervoltages are positive.

Mechanism of breakdown; statistical and [ornuuiuelags

When a potenrial difference of a· few volts isapplied between cold electrodes in a gas, the latterbehaves substantially as a perfect insulator 2).Ionization due to cosmic radiation will give rise toa very small current, which will increase to a satura-tion value as the field is increased to a point whereall the free electrons and ions are drawn to theelectrodes. This saturation current is not constantwith time, 'however, being subject to wide sta-tistical variations owing to the random nature ofthe ionization. The saturation current will be greaterand largely free of fluctuations if the cathodeis-exposed to light so as to emit a copious supplyof photo-electrons.As the applied field is increased, the free electrons

become' capable of imparting energy to the outerelectrons of the gas atoms: the latter may be excitedin the normal way and. emit radiation, or they maybe raised to metastablestates, or, when the collid-ing electron has sufficientenergy, some atoms maybe ionized. With the gaspressures .and electrodeseparation commonly usedin trigger tubes, the meanfree path of the free elec-trons will be only a.fraction of the inter-electrode distance: henceionization by collision willoccur only when the ap-plied potential is severaltimes the 'ionization po-tentialof the gas.

The positive ions in thegas acquire much the sameenergy as the free electronsbut, because their mass isequal to that of the gasatoms, they give up on 'anaverage about half theirkinetic cnergy at each col-lision. Since, also, elasticcollisions are more prob-.able than ID the electron

129

T

K

K

A

T

Fig, 3. Electrode structu~e(schematic) of the Z 300Tcold-cathode tube, A anode,T trigger electrode, K ca-thode, M mica spacers. Thetube is filled·with argon toa pressure of 8 mm Hg, Theelectrodes are of nickel, thecathode having a coatingóf barium oxide upon its, Inner face. '

2) For a general review ofthe field of gas discharge phenomenasee F. M. Penning and M.J. Druyvesteyn, Rev. mod. Phys.12,1940.

130 PHILlPS TECHNICAL REVIEW VOLUME 18

case and charge-exchange may occur, the positiveions contribute little to ionization 'by collision.

With' the onset of ionization due to electroncollision, the anode current increases rapidly withthe applied potential. If the number of new freeelectrons created by a single electron moving throughunit potential is 17 (the gas ionization coefficient),the number .of new free .electrons created -by nelectrons ..traversing an element dl between theelectredes is

dn = u17E dl.

Assuming that no electrons leave the region of thecathode per second and that the field E is uniform,the number of electrons arriving at the anode per'second is:

where L is' the electrode separation -and- V theapplied potential.Th'e anode current may th~n be written

10 being the cathode emission current correspondingto some electron-supplying mechanism, e.g.- photo-emission.

So long as the anode current is determined by (1)the discharge cannot become self-sustaining:' if10 ceases, (by shutting off the light in the case ofphoto-emission) the anode current will vanish.Physically, this means th~t when the free electronsinitially present and those created by i~nizationhave been drawn to the anode, the anode currentmust cease unless·further free electrons are supplied.Hence, f~r the development of a self-sustainingdischarge a secondary electron-supplying processis required. This may be effected by the positiveions; which on striking the cathotle, give rise tofurther electron emission ~hich enables the dis-charge to become self-sustaining. The number ofnew cathode electrons emitted per ion-pair formedbetween .the ele~trodës is termed 'the secondary'ionization coefficient y. It is closely related to thework function cp of the cathode; for a give~ gas, y islarger the smaller cp.

Taking this secondary process into account, the~ relation between I 'and 10 becomes:

e'1V1=10 V'1- y[e'1 -1]

The characteristics respresented by this expres-sion are plotted in fig.·4 for various values of theinitial cathode' current 10'

When the voltage V across the electrodes has thevalue Vb defined by

y[e'1Vb_I]=I, ..... (3)

the conclusion from (1) that' I --+ 0 as 10 --+ 0 isno longer necessarily valid, for equation (2) thenleaves I indeterminate. What actually happens canbe seen from fig. 4. The voltage V across the tubeand current I are determined for a given value of10 by the intersection of the relevant curve with theload line R (curved owing to the logarithmic scale)

vI,1£'

~~'

~

a

(1)89414 _/og1

b

(2)

89415 _1og1Fig. 4. Voltage-current characteristics (schematic) for a dis-charge between two electrodes in a gas, for various values of theinitial current 10' For clarity the current is plotted logarithmic- .ally. The numerical value of 10 for each curve is given by itsintercept on the current axis. 'à) Normal self-breakdown, with 10 < Iocr. Thisrefersto break-downin the trigger gap. Rl are load lines (curved owing to logscale) corresponding. to supply voltages Vs, Vs' and Vs". Whenthe supply voltage exceeds Vb (e.g. Vs') breakdown occurs andthe current which then flows for an initial current of 101 (say)is given by the point A. If 10 then -+ 0, the discharge stahilizesat B. For a higher supply voltage (Vs") the discharge stahilizesat B'. The 'trigger discharge may i also , burn in the glow-discharge region E if the series resistor Rl is small enough.b) Triggered breakdown (Io> Iller) of the anode gap. When10 > Ióc., space charge effects beginto appear which cause thebreakdown voltage to be lowered. The 'dotted curve' from thepoint C represents the locus of the points at which spacecharge begins to have an effect, For an initial current of 102,

the breakdown voltage is lowered to the point D. R2 is the loadline corresponding to a supply voltage VS1 and series resistanceR2 and is tangentlal to thè curve 102, In the absence of 102 thetube will be non-conducting but when 10 is raised to the value102by the trigger discharge, breakdown occurs and the tubecurrent stabilizes at the point E.

_, -l-

" .

1956/57, No. 4/5 èOLD-CATHODE TRIGGER TUBES '131

corresponding to the supply, voltage Vs and theresistance in series with the source. I

As the supply voltage Vs'is raised (fig. 4a), theload line is displaced parallel to the ordinate and at avoltage somewhat higher' than Vb a self-sustainingdischarge becomes possible; the voltage and current.between the electrodes are then represented byu, point 'such as A (fig. 4a). The discharge is self-sustaining for, as 10 -+ 0 the working point on theload line moves to the left until intersectiori occursat a voltage V = Vb (e.g. point B in fig. 4a).

With further increase in Vs the current I in-creases (B -+ B') but Vremains at Vb. Ata certainvalue' of I (point C) space charge effects begin todistort the hitherto uniform field; further increase inI is accompanied by a fall in the voltage V acrossthe electredes and by the development of the glowdischarge (region E in fig. 4a).

For values of the initial cathode current 10 lessthan a certain critical value Iocn the breakdownvoltage (i.e. the maximum voltage Vmax reached inthe establishment of a self-sustaining discharge) hasa constant value Vb determined solely by the materi-als, geometry and gas filling of the tube. Above thiscritical value, however, the breakdown voltageVmuxbecomes a function of 10' Space charge effectsnow appear at an early stage and prevent V fromreaching Vb. Consider the characteristic 102(> Iocr)and a series resistor R2 giving the load line R2 infig. 4b. As the supply voltage Vs is increased to thevalue VSl' V passes its maximum value near D andbreakdown occurs. The voltage _v across the electr-odes drops sharply and, for a not too large value ofthe load (i.e. :::;;R2 in this case), current and voltagestabilize at a 'point such as E éorrésponding to theglow discharge regime.

Normal self-breakdown between two electrodcs ina gas (fig. 4a) is the procees taking place in thetrigger g'ap of a cold-cathode tubè. 10 is normally much'less than 'Iocr. A pulse with a voltage greater than, that corresponding to the gap breakdown voltage Vb(hereafter designated Vtb) causes a self-sustainingdischarge to be set up in the trigger gap for theduration of the trigger pulse. (A condenser is oftenconnectad between trigger and cathode to intensifyor prolong the trigger discharge.) A fraction of the"charge carriers of this discharge are -transported.to the main anode-cathode gap aud form the "10"

to initiate the .main discharge, as outlined below.(Wnether the trigger discharge burns in the Towns-end region _:_the current region up to the point C,fig. 4 - or dn the glow discharge region near E,

. depends on the circuit conditions. These must, ofcourse, be so chosen that the trigger discharge is of

sufficient intensity to provide the "10" necessaryto ignite the main gap - see below.)

.Triggered breakdown is the proces~ by which themain anode-cathode discharge is started. The voltageat which a self-sustaining discharge, across this gapcan start is temporarily lowered by- providing aninitial current' Jo that exceeds the value of Iocr -forthis gap (fig. 4b ).: As mentioned above, this 10 isderived from the already-established trigger dis-charge, which can easily he "made to supply chargecarriers at such a rate' that Iocr is exceeded. (Thelowering of anode breakdown voltage as a functionof trigger current It was shown schematically infig. I). If, then, across the main gap, a steadypotential is applied somewhat below the normal self-breakdown value Vb {hereafter designated Vub), thetube will remain quiescent until triggered .by thepresence of an adequate initial current 10 due toa discharge set up in the trigger gap.

There is always a éertain delay after the applica-,tion of the, trigger pulse hefore ,the s~lf-sustaiuingdischarge is set up in the trigger ,gap. The delay ismade up of two components. Th~ first, which is byno means constant, is known as .Ócthe statistical lagand arises from the necessity for the presence of atleast one favourably-placed electron to enable thedischarge to start. In practice, if. 10 ~ Iocr an un-certain and frequently lengthy dèlay occurs. Withincrease in 10 the statistical lag is gradually elimi-nated. It' is desirable that the initial cu;rrent Jo liesbetween the value just able to remove the statisticallag and the value Iocr. The inter-electrode currentwhich thenflowsis called apriming current. Withmanycathodes a suitable level of ambient ilhimination willgive rise to such a priming current; other means forproviding a priming current will be described below.

The other delay is the time interval re~uired tobuild up the self-sustaining discharge from themoment the first favourably-placed electron appearsuntil the discharge has reached .its equilibriumvalue. This delay is called the formative lag; it doesnot vary randomly but it depends sharply, on .theexcess voltage. of the trigger pulse above the break-down voltage Vb of .the ,trigger gap. To keep -thisdelay small, the trigger pulse voltage should exceedthe trigger breakdown voltage by, some tens of volts(e.g. Vs" infig. 4a) .

Desirable. characteristics of cold-cathode tubes.

To widen the field of application of trigger tubesthe following should be aimed at:a) Good reproducibility' of breakdown and main-

taining potentials, both from tube to tube andin individual tubes during life.

132 PHILIPS TECHNICAL REVIEW VOLUME 18

b) Faster operation, if possible approaching 'the"response of hot-cathode devices.

c) Improvement of input .sensitivity.d) Reduction or elimination of random operating

de~ays, even at low levels of ambient ~umina-tion.

e) -Reduction of incremental impedance and noiselevel; for .some uses, e.g. for speech-channel. applications,. the noise level of most tubes is toohigh.,

f) Small dimensions.It would not be possible to improve on all these

points simultaneously, as some of them are' partlycontradictory. The initial aim, therefore, was toevolve a design and a manufacturing technique to.give a cold-cathode. tube in which. some of. theexisting characteristics were improved, viz. pointsa, d and j: uniformity of characteristics andstability during life, freedom from random delays(statisticallags) and robust miniature construction.These particular properties were selected for im- .provement because it appeared that they werenot incompatible and also because they arecollectively of importance in many applications 3).It will be noted that all three features are ingre-

dieuts of reliability, the improvement of which wasindeed the primary aim. It will further be notedthat while the raising of the intrinsic input sensitivitywas not' considered at this stage, the effectivesensitivity of the whole circuit is of course improvedwhen the tube characteristics are accurately knownand maintained, for the input signals may then safe-ly be smaller without fear of the tube not respondingor, worse, spurious breakdown when not signalled.

Improvement of reliability and performanee

The breakdown and maintaining potentials ofcold-cathode tubes areinversely related to the valuesof the gas ionizatio~ coefficient 'YJ and the secondary'ionization coefficient y. It seems obvious thereforeto choose gas fillings and tube materials which yieldlarge and stable values of both coefficients: in fact,the inert argon filling and barium-on-nickel cathodeof the Z 300T tube were selected on this basis.However, the fragile nature of the bariumlayer (which is susceptible to change even 'with

3) Others have aimed at improvements in directions notattempted in the present work. An experimental tubesuitable for speech channel applications, the Z SOOT,hasbeen developed in the Philips Laboratories in Eindhoven;see J. Domburg and W. Six, Philips tech. Rev. 15,265-280, 1953/54. Other work in this field is reported byM. A. Townsend and W. A. Depp, Bell. Syst. tech. J. 32,1371-1391,1953. Work on improving the speed of response

. is described [by G.H. Hough and D. S. Ridler, ElectronicEng. 24,152-157,1952. - ,

normal discharges and is removed entirely bysputtering if temporarily overloaded) together withgaseous contaminants from the wall of the. tube andelsewhere, lead to changes in 'YJ and y which areobserved by the user as the objectionable variationsmentioned above. Th~ steps taken to stabilize thesequantities are outlined below.Another matter requiring attention is the elimina-

tion of the s~atis~ical lag. As we have seen, to dothis, a priming current must flow in the otherwisedormant tube,

Molybdenum sputtering technique

In 1946 Penning and others 4) described means ofstabilizing the basic glow discharge quantities forcertain materials prior to the successful measure-ment of the "normal cathode fall" (the constantdrop in potential near the cathode surface, whichoccurs when the current through the tube is betweencertain limits). The sustained cathode sputteringtechnique employed in this work 'had obviousapplication in any glow-discharge device dependentupon stability of the operating potentials, and wassoon employed in the voltage reference tubes85A1 and 85A2, where a molybdenum cathode isused in conjunction with a filling ofneon and argon 5).Similar arguments clearly favoured the introduo-tion of this system in the class of trigger-tube underconsideration. There are the following objectionsto the molybdenum-sputtering technique. Workingpotentials are high compared with those of tubesusing barium or, say, potassium cathodes. Photo-emission arising from normal ambient illuminationis ineffective in supplying the priming currentnecessary for the eliminatien of the stntistical lag.Lastly, the manufacturing technique is not withoutits difficulties and the price of the materials is afactor of some significance. However, the advantagesgreatly outweigh the objections.

The successful application of the cathode sput-'tering technique to trigger-tube manufacture de-mands an electrode structure quite different tothat of Z 300T (fig. 3). Thus, in the first place, micaspacers must be abandoned. They act as barriersto the transport of sputtered material to the wallsof the tube, and when coated with this material,act as a source of inter-electrode leakage. Also, micais a laminar material, and often secretes contaminants. which can be slowly released during the operation ofthe tube. Secondly, the cylindrical form of the

4) F. M. Penning and J. H. A. Moubis, Philips Res. Rep. 1,119-128, 1946; T. Jurriaanse, F. M. Penning and J. H. A.Moubis, Philips Res. Rep. 1, 225-230, 1946; and T. Jurri-aanse, Philips Res. Rep. 1, 407-418, 1946.

5) T. Jurriaanse, Philips tech. Rev. 8, 272, 1946.

1956/57, No. 4/5 COLD-CATHODE TRIGGER TUBES 133

cathode is not suited' to' sputtering. The cathodemust have a shape which allows the unimpeded-escape of sputtered material, and must he so placedthat while the walls are covered witha uniform film,the lead-out area remains bar~. Thirdly, the tubeenvelope must be small enough to be covered in areasonable 'time, and also he free from re-entrantportions such as exist in the old "pinch" construe-tion.

Experiments showed that a small oblong cathodeplate mounted with its plane parallel to the axisof the tube, in a standard 7-pin or 9-pin miniatureenvelope, met all these needs very well. This struc-ture is shown in jig. 5. Subsequent field experiencehas shown that this rather frail-looking structureis well able to withstand the shocks of normalservice.

Fig. 5. Electrode structure (simplified) of experimental Mo-sputtered tetrode trigger tube. K cathode, T trigger electrode,AK tubular auxiliary cathode surrounding the anode A. SSare shields which leave a region near the lead-out wires freeof sputtered material (see below).

Provision of priming currentMethods of providing a priming current include:

a) The admission of normal daylight - the simplestsolution, and that 'used in Z 300T. However, thisis ineffective with the molybdenum cathode, whosephoto-emission is negligible for the spectral rangetransmitted by normal glasses. Moreover, relianceon the p'hoto-electric effect imposes undesiredrestrictions upon the conditions of use of the tube.b) .The use 'of radio-active materials., This wouldhave been a more satisfactory solution, but wasrejected because of lack of .experience with "soft"emitters whose influence would not be' objectionableoutside the tube, .and because the third system hasadditional advantages quite apart from the priming,c) The use of an auxiliary discharge. Such a dis-charge may be direct, passing the trigger and catho-de, or indirect, achieving its effect by the transportof some of its ions into the trigger-cathode space,possibly over a considerable distance.

The direct auxiliary discharge imposes certainrestrictions on the trigger circuit, but is adopted inone ofthe tubes described below (Z 801 U). An upperlimit somewhat below 1 micro-ampere is set on thepermissible value of such a direct priming current bythe need to avoid distorting the applied fields withinthe tube by the accumulation of space-charge.The indirect discharge, which may be larger, is used

in the stabilizer tube Z 800 U _:_with the dischargeflowing to the anode from an auxiliary cathode. The_latter is of tubular form, and surrounds the wire anode(fig. 5). This tubular auxiliary cathode provides aneconomical solution to two problems: how to obtain'within a miniature envelope an effective spacing largeenough to create a useful difference between Vab andVam, and how to screen a discharge between anodeand auxiliary cathode from the trigger-cathode space.By varying the recession. of the anode tip within theauxiliary cathode, Vab. may be altered withoutaffecting the remainder of the trigger-tube, whilethe presence of the auxiliary discharge stabilizes thepotentialof the auxiliary cathode relative to theanode. The 'auxiliary discharge may now reach10 !l,A, and so enter' the current range controllableby standard resistors, without exerting an excessiveinfluence on the field in the trigger-cathode gap..

With the introduetion of the auxiliary discharge,the maximum operating delay is reduced from about1 sec to the order of 100 fl. sec, which is essentiallythe formative lag.

."'.

wl

Experimental trigger tube. The experimental tubé (fig. 5) incorporating the .features outlined above showed satisfactory behavi-

.,

. ,

134 PHILIPS TECHNICAL REVIEW VOLUME 18

our. In order to keep the triggering' voltage Vtb toa minimum, the trigger-cathode separation was keptsmall (1-2 mm). In addition, a trace of argon (about1%) was added to, the gas filling of neon, at a press-ure of 40 mm Hg. Under these conditions Vtb ~US,V. .The recession of the anode tip within the tubular

auxiliary cathode is effectively limited by the factthat, howevér much the recession, there' is a limitto the attainable anode-cathodehreakdown'voltage(Vab) fixed by the breakdown between auxiliarycathode and cathode. Beyond a certain point, there-fore, further recession will not enhance the value ofVl\b., For the gas filling used, Vab has its maximumat 250 y. Under these conditions, the discharge-maintaining voltage Vam - 100 V.

. The triggering voltage stability of early 'experimental,tubes was not entirely up to expectation. This was found to bedue to contamination of the trigger electrode, which was thenof nickel. So long as different materials were used for triggerand cathode, it proved impracticable to clean both by sputter-ing. The use of a molybdenum trigger gives a satisfactorystability, but it has the disadvantage that in circuits in whichthe trigger may become earthed, the maximum operatingvoltage may have to he smaller to preclude spurious trigger-anode breakdown.

Variations in the dischargc-maintaining potentialVam are similar to those of the voltage referencetube (diode) 8SA2 and are satisfactory.

Self-quenching operation of trigger tubes

The achievement of better stability and thevirtual elimination of statistical lags cleared theway for the design of two trigger tubes for certainspecific applications, viz. for use in a stabilizercircuit (Z 800U) and for use in a rate meter circuit(Z 801U) 6). The former has to respond to very smallcontinuous currents and the latter to very smallpulses of current.

While both the Z'800U and the Z 801U may heused in certain conventional cold-cathode tubecircuits, they are primarily designed for the oscilla-tory self-quenching circuit. In this type of circuit(fig. 6), the anode current is derived from the dis-charge of a condenser and is therefore .intermittent,passing as a single pulse each time the tube is trig-gered. The self-quenching circuit has the advantage.that the trigger electrode is capable of exercising aquasi-continuous and reversible controlover themean anode current. This gives,the circuit a numberof important applications ..Trigger tubes working insuch circuits face conditions differing markedly fromthose prevailing in steady current circuits. Of

6) These two special-purpose tubes are not available commer-cially from the Philips Electronic Tube. Division.

particular ~mportance are the factors governingextinction of the discharge after each current pulse.The self-quenching action is obtained with an anodeload Ra which is considerably less than might besuggested from the fact that to extinguish an

Vc> s ---------.

Ca

Fig. 6. Trigger tube in self-quenching circuit.

established steady discharge (the supply voltageremaining constant) Ra would have to be increasedto at least lOB n. The self-quenching circuit canoperate with Ra only a hundredth of this value.

A study of the self-quenching discharge leads to the followingexplanation. The absence of a resistor limiting the condenserdischarge enables anode breakdown to be followed by anexceptionally large tube current; this has passed its peak andlargely decayed by the time that Ca has discharged sufficientlyfor Va to fall to the normal maintaining voltage Vam, and has;in consequence, created an abnormal positive ion cloud aroundthe anode.Asthis cloud moves away, the anode is able to descendweIl below Vam in an "overswing"; at the end ofthe overswingthe flowof current through Ra carries Va up again to Va,m, butnot before the process of de-ionization has gone far enoughto lead to the collapse of the discharge. The overswing, whichgives the tube the appearance of having inductive properties,commonly carries Va down to 3/4 of Vam. In the case of theZ 800U tube other considerations led to a design havingsatisfactory self-quenching properties; with the Z 801U, how-ever, the main problem was the re'conciliation of the self-quenching requirements with the remainder of the specification.

Stabilizer trigger tube

The trigger tube Z 800U 7) (fig. 7) was originallydeveloped for use in a stabilized H.T. supply forGeiger counters in radiation monitors 1),viz. the low-current stabilizer circuit introduced by Gouldiri"gB)(fig.8). The tube has to meet the following threeconditions:1) Vab to be approximately 300 V.2) The tube te;' operate in a self-quenching circuit.3) The trigger input current demand to be less than

10-7 A.

7) C.H. Tosswill, British Patent No. 709103.8) F. S. Goulding, British Patent No. 732776; and, A variable

voltage stabilizer employing a cold-cathode triode, Elec-tronie Eng. 24, 493-497, 1952 (the Z 800U is referred toas VX BI07, and. the Z BOIU as VX 80B6); G. O. Crow-ther, Cold-cathode voltage stabilizer, Electronic Eng.25, 127, 1953.

1956/57, No. 4/5 COLD-CATHODE TRIGGER TUBES 135

89062

Fig. 7. Photograph of the Z 800 U trigger tube (centre, see alsofig. 5) showing its size in relation to the Z 300 T tube mentionedearlier (fig. 3). (The scale on the rightis in cm.) The tubularauxiliary cathode and the sputtering shields round the lead-inwires can just be seen through the envelope of the Z 800 U. Theratemeter trigger tube Z 801 U described later (see fig. 14)and the general purpose tube Z 803 U are identical in sizeand appearence.

With the same electrode structure as that used inthe experimental tube (fig. 5), the required break-down voltage was obtained by using a gas filling ofneon with 3% argon at a pressure of about 40 mmHg. This percentage of argon is somewhat greaterthan that used in the experimental tube (1%). Thechange also raises the maintaining voltage Vamto 105 V and the triggering voltage Vtb to 125 V.The tube has to operate under circuit conditions

which create self-quenching discharges in both thetrigger-cathode and the anode-cathode gaps (jig.9a,in which the priming electrode is omitted for sim-plicity). Both anode and trigger currents therefore

Cout

88548

Fig. 8. Coulding's stabilizer circuit B) using the trigger tubeZ 800U. n, = 68 MD, R2 = 100 MD, Ct = 270 pF, Ra = 1MO, C. = 0.005 fLF, Rak = 100 MO, Cout = 0.25 fLF. In thisstabilizing circuit a fraction R2/(Rl + R2) of Vout is appliedto the trigger electrode. Therefore if Vout > Vtb(Rl + R2)/R2the tube will be triggered and the pulse train through the tubewill cause COll' to discharge until Vout has fallen sufficiently todispose of the inequality. Any further fall in Vout will cut offthe tube current because the discharges are self-quenching.In this way a stabilizing action is secured (provided that[;.,> Iout) with Vou' slightly Îr1 excess of Vtb (Rl + R2)/R2•

take the form of trains of pulses and the interactionbetween them complicates the general analysis ofthe problem. We m~y confine ourselves to theimportant special case when the anode supplyvoltage Vas< Va,b and RaCa ~RtCt. Each anodepulse is then necessarily the immediate conse-quence of a trigger pulse, though individual anodewaveforms are independent of the actuating triggerdischarges (fig. 9b).

The mean anode current bears a constantrelation to the mean trigger current, since both aredetermined by condensers charged to the (fixed)supply potential Vas and to the trigger breakdownpotential Vtb. Since the anode current pulse traincan be stopped as well as started by the action ofthe trigger, i.e. by variation of the trigger supplypotential Vts, the mean anode current la can becontinuously controlled over a considerable rangeby the trigger supply voltage Vts' The circuit offig. 9a may there fore be used as a D.e. amplifier.lts amplification factor is Qa/Qt, i.e. the ratio ofthe charges on the anode and trigger capacitors.

VabVos

Va

t

QFig. 9. a) Circuit giving self-quenching discharges in bothanode-cathode and trigger-cathode gaps. Priming electrode(auxiliary cathode) omitted for simplicity. R, = 100 MO,Ct = 470 pF, Ra = 1 MO and Ca = 2000 pF.b) Anode and trigger waveforms for the circuit shown in(a). When RaCa < R,Ct, a trigger pulse above a certain ampli-tude gives rise to an immediate anode pulse. The anode wave-forms, however, are independent of the form of the actuatingtrigger discharge. (The breakdown and extinction processesoccupy an interval that is short compared with RaC. andRtCt, and therefore appear in the graph as a vertical line.)

Operatien under the above conditions impliesthat both discharges are self-quenching (which canbe achieved without using the very high componentvalues which would be necessary under steadyoperating conditions) and that the charge passingin a single trigger pulse must be sufficient to causeanode breakdown. This requirement is analogousto the It! Va relationship (fig. 1) for triggering inthe case of the steady current circuit. In the self-quenching case one of the variables is thus a compo-nent value (Ct); however it is more fruitful toconsider another approach, viz. the effect of thetrigger supply potential. (The variation of Vabwith Ct can be deduced from fig. 12.)

136 PHILlPS TECHNICAL REVIEW . VOLUME 18

Critical trigger supply potential (Vtsc)

Fig. 10 shows the anode supply potential Vasplotted against the trigger 'supply potential Vts'.The curve represents the conditions under which. aflow of mean anode current la can just be detected.The value of Vts corresponding to the nearly verti-cal part XY is called the critical trigger supply

lBA

x~, ............. -i 7"--

iVts c (x<Vas<Y)

88550 -\ltsFig. 10. Vos plotted against the value of Vts required to justgive a mean anode current, for a self-quenchingcircuit. Withina considerable range of anode supply voltages (X- Y) therequired trigger supply potential is independent of Vos. Thisvalue is termed the critical trigger supply potential, Vtsc.

potential; the tube is preferably operated in thisregion since Vts is here nearly independent of theanode supply voltage Vas.Further penetration of region B (fig.l0) leads to a

growth of la; on return to region A., la is cut off.This provides an interesting comparison with theirreversible action of fig. 2, where the current can-assume only a certain fixed value ..

The critical trigger supply potential Vtsc can be, resolved into three components:a) The breakdown voltage Vtb necessary acrossthe actual trigger-cathode space to initiate break-down. This .has been .discussed in the previoussub-section.b) .The potential difference developed across Rt(fig. 9a) due to ohmic leakage over the tube surfaces.This has been reduced to negligible proportions byreducing the leakage current to 10-10 A. Two simplemeasures were sufficient to achieve this: first, themounting of small shields upon each lead-out wireclose to the point of entry into the glass, so as tocreate a small bare zone within which no sputteredmaterial could fall; second the spraying of the ex-·terior of the envelope in the neighbourhood of thelead-out wires with a silicone solution to inhibit theformation of conducting films of moisture.

c) The potential difference developed across Rt dueto the pre-breakdown trigger-discharge current Itb(the trigger condenser does not contribute to thiscurrent at this early stage). The maximum value ofItb determines the current drain on the trigger.supply which, as stated, must not exceed 10-7 A.Itb is dependent on a number of circuit and tubeparameters; these will now be discussed.

Trigger supply current drain at breakdown (Itb)

.The influence on [tb of four variables was examin-.. ed as outlined below.1) Effect of anode supply potential Vas. Fig. 11shows the effect of Rt on Vtsc' Since the changes inapplied' potentials were slow, the separation of thetwo curves can only he due to the voltage drop acrossRt, due to surface leakage current together with Itb'The leakage current (10-10 A) accounts for a separa-tion of only 0.006 V for the two resistances used(10MO and 68MQ). Hence the observed separation ofapproximately 1 volt must be almost entirely due toItb; the value of Itb is thus 10-10 X 1/0.006, i.e.2 X 10-8 A. Itwill be noted that the separation ofthe two curves is constant within a considerablerange of anode supply voltages: this implies thatItb is independent of Vas. Fig. 11 also shows thatthe critical trigger supply voltage is substantiallyindependent of Vas (cf. fig. 2 for the steady dis-charge case).2) Effect of auxiliary cathode current lak. Fig. 11is based on a value of lak = 3 (LA. Little change in

~Or---.----.----.----.----.----.---'\IOlts

280~--4----4--~rt---_+----~---+----1V. I. Rt = IOM.n.

ta~0~---+----+----H~---+_2_R_t_=,5_8_M._~--r---~/2

8

240~--~----r---~----+----+----'_--~

220r---4----4----*---_+----+_--_+--__,A

200r---4----4----~--_+----+_--_+--__,

mO~--~----~--_fl----+---_+----,_--~

mOr----+----,_----H~~~±_--~----~----,..----r.-_jj_.<1V ....

1'V

~o.~ __~ __~--~i~i·--~~~~~~~~110 120 130 140 150 150 170volts

-Vts 88551

Fig. 11. Variation of Vtsc with Vos for two different valuesof the trigger resistor Rt. Ct = 1000 pF. From the separationof the two curves the trigger supply current drain at break-down can be deduced (see text).

. '

1956/57, No. 4/5 COLD-CATHODE TRIGGER TUBES 137

Itb can be detected for variations of Iak between1 and 10 (LA.Thi.s somewhat unexpected result isprobably due to the fact that the auxiliary dis-charge starts near the end-of the cylindrical auxiliarycathode and expands inward with growth.3) Effect oftrigger condenser Ct. Fig.I2 shows again'the variation of Vtsc with Vas but with Rt' kept

28volt

0s

0 t. Ct = 1000pF2. Ct= 430pF

03 Ct= 100pF

01--3,~

0 .

0Iv2

26

22

20

18

160

14110 120 130 140 150 160 170volts

88552 -Vts

Fig. 12. Effect of trigger capacitance Ct on the Va. - Vt•characteristic. Trigger resistor Rt = 68 MD.

constant and three different values of Ct. The lowerthe value of the latter, the higher is the criticaltrigger supply voltage: this implies that Itb isgreater. Hence there is a lower limit to the capacit-ance of Ct, if Itb is to be kept below 10-7 A.4) Effect of trigger-cathode spacing: In the absenceof mica spacers, a certain spread in the electrodespacings is inevitable. Ïu tubes with smaller spacingslarge valties of Itb (10-6 A) were liable to occurintermittently, an effect visible as a corona sur-

. rounding the trigger wire instead of the normal glowadjacent to the cathode. This trouble was.eliminatedby increasing the trigger-cathode separation from1 mm to 3 mm, whereby the trigger breakdo~voltage Vtb is also increased, from 125 to 135V.

Performance in stahilieer circuit

The investigations described above showed thatonly very min~r modifications were necessary tothe original' experimental tube in order to getsatisfactory operation in a self-quenching circuitwith a very small trigger current drain.

In the restricted mode of eperation defined earlier(Vas < Vab, Ra Ca~RtCt), increase of Vts above:Vtsc gives rise to a finite mean anode current I~'whos'e value for a given set of supply potentials

and component values depends on Vtsc; the .anodéextinction. potential and the trigger extinctionpotential. Since Vtsc and the two latter quantitiesare constant for a given tube, the tube can be usedfor stabilizing a supply potential a~ in the Gouldingcircuit (fig. 8). The stabilizing action of the circuit.will be clear from the figure and the details given inthe caption.

This stabilizer, using the Z 800D tube, has thefollowing important properties ..a) It works in a voltage range (200-300 V) which'cannot be easily accomodated by either the glowdischarge or the corona discharge diode.b) Its current range is 5-150 ~A. Diode glowdischarge tubes do not normally work successfullyin this range. The corona stabilizer possesses asimilar current range but cannot compete in lifestability with the trigger tube circuit embodying theZ 800D.è) The output voltage can be adjusted, for eachtube, by altering the resistor ratio R2/(R1+R2)

(see fig. 8); the output voltage can in this way bevaried over a consideràble range to allow for thespread of Geiger-Mü1ler tube characteristics.. The properties of the Z.800D tube may also beexploited in more conventional circuits in which thecurrent flows continuously. Retention of the triggercondenser will still permit operation with smallsignal currents down to 10-8 A. If the auxiliary -discharge be dispensed with and the occurrenceof statistical lags accepted, operation with inputcurrents down to 10-10 A is possible.

Rate meter ~rigger tube

The Z 801D 9) was developed for use in ratemeter circuits, that is to say, circuits whose outputcurrent is a measure of the rate of arrival of inputpulses. The pulse source relevant to the developmentof this tube was a Geiger-Müller tube, ·but the resultsobtained are. applicable to any system handlingpulses of similar character.

The basic cold-cathode tube rate meter circuitdevised by Franklin 10) is shown in fig. 13a. Theadvantages of this circuit are described in full in thepatent specification andwill only be summarised here.

Each time the tube is triggered, the anode circuitgoes through a single self-quenching cycle. Providedthat the anode has sufficient time following thedownward swing to regain the full Vus value beforethe arrival of the next input pulse, then la will-beproportional to the .rate of triggering. The connect-.ion of the trigger to the anode through a very large

D) C. H. Tosswill, British Patent No. 696077.10) E. Franklin, British P~tent .No. 638833.

138 PHILIPS TECHNICAL REVIEW VOLUME 18

resistance has the effect of poising the trigger close tothe breakdown potential. Two purpo~es are servedin this way: a direct priming current is, pro~dedfor the trigger-cathode space without the need, fora fourth electrode and even the smallest in~ut

Q

Fig, 13, a) Basic rate meter circuit according to Franklin t'').b) Franklin's rate meter circuit using a tetrode trigger tube.Typical component values: Rt = 100 Mn, Ct = 50 pF, Rn =200 kn, Cn = 0.01 fLF,Rnk =,1 Mn, Cak = 50,pF.

pulse applied to the trigger will cause it to exceedthe breakdown potential. Forthe successful workingof this circuit three conditions must be fulfilled:a) The trigger priming current must not carry thetrigger circuit to the point of instability, i.e. thepriming current must be less than Itb. Whetherthe trigger priming current> or <Itb depends onthe values of Rt and Ct (fig. 13a).b) In this circuit; the input pulse will be able todeliver the whole of its charge (Qin) into the trigger-tube. For a given value of Vas there is a certainmin~um value of Qin for anode breakdown.c) The anode and trigger discharges must be self-quenching,An experimental triode was constructed to study

the rate meter circuit requirements. The molyh-denum-sputtering technique was again used andthe same neon-argon' filling was used as 'in thestabilizer tube., The leakage-counteracting screensof the stabilizer tube - essenrial to keep the leakageresistances high compared with Rt (Rt R::! lOB ohms)- werc also fitted. In order to meet the radiationmonitor requirements, it is additionally necessarythat:a) The "extinction current" should be at least 210(LAto give an adequate value o£Ia at the anticipatedoperating frequency. By the term "extinctioncurrent" is meant the current passing through Raat the lower end of the anode' swing, with Ra set atthe minimum value compatible with self-quenchingoperation. The "extinction current" in the case ofthe triode was only 70 [LA.,b) The input charge (Qin) necessary for reliahleresponse to Ceiger Müller tube signals, should be notmore than about 5X 10-u coulomb with the anode

88553

supply voltage Vas safely less than the breakdownvoltage Vab. The value for the experimental triode'was 5 X 10-10 coulomb. .c) The frequency range should preferably be ahout10 kc/so The triode had a range of only 1 kc/soThese aims are difficult to reconcile, improvement

of one being at the expense of the others. However,by considering how to increase the extinction currentwithout loss elsewhere, we are led to the idea of.separating the function of triggered breakdown fromthat of extinction, so as to permit of manipulatingthe two independently. A tetrode tube is thereforeimplied, though for reasons. other than thoseobtaining in the case of the stabilizer tube. Thecorresponding rate meter circuit is shown in fig. 13b.

Fig. 14. Simplified diagram of electrode structure of rate meter'trigger tube Z 801U. K cathode, AK auxiliary cathode, Ttrigger electrode, A anode, G glass tube, SS, shields. '

1956/57, No: 4-/5 COLD-CATHODE TRIGGER TUBES 139

, The first rate meter tetrode had as' its' fourth sequence of breakdown: the input, charge flowselectrode ("auxiliary cathode") a perforatcd disc across the' trigger auxiliary cathode - g~p; break-'situated in the path of the discharge to the main down then occurs between trigger and main cathode;caihode(fig.14).Aninputsignalwflsnowfollowedby "finally, breakdown occurs between anode andbreakdown between anode and auxiliary cathode;then by a moderate discharge leading to a risé inauxiliary cathode potential; thcn by anode tomain cathode breakdown; and finally by the normaldownward swing of anode potential. At. this pointit was found, as expected" that the auxiliarycathode barrier successfully interfered with thetransition to a steady glow-discharge between anodeand main cathode and led to the collapse of the.discharge at higher levels of extinction current thanbefore. Trigger action remained identical with that, of a triode, with the auxiliary cathode acting ascathode.

The contribution of the auxiliary cathode toextinction is that of widening the gap between thenormal maintaining potential for a steady dischargeand the minimum potential reached in the anode"overawing": both as a physical obstacle and as aparasitic current drain, the auxiliary cathode hasa greater effect on the former potential.

The tetrode Z 801U has an extinction current of240 (.lAagainst 70 for the triode; the input sensitivityand frequency response are unchanged. In 'theproduction version the perforated disc was changedto a wire grille. The tube has a filling of neon with5% argon, and is shown in fig. 14.

The negative pulse circuit

The cold-cathode triode is most frequentlyswitched by a positive signal applied to the trigger;so also were the first tetrode Z 801U tubes. TheGeiger-Müller .tube however is normally better usedwith its eathode grounded, and thus delivers anegative pulse. It was found that the tetrode eouldbe triggered by a negative pulse 11) applied to theauxiliary cathode (fig. lSa), involving a different

Vas

Rt

.QFig. IS. a) Negative pulse rate meter circuit.b) Negative pulse circuit with improved frequency response,Typical component values: Rt = 100 Mn, Ct= 100 pF, Ra =200 kn, Ca = 0.01 (J.F,Rak = 1Mn, Cak = 15 pF.

11) F.S. Goulding and C.H. Tosswill, British Patent No. 723892.

main cathode. Äpart from its convenience- inGeiger-Müller tube service, this negative pulsecircuit has three substantial advantages. Firstlythere is a remarkable gain in sensitivity: an inputcharge Qin of 3X10-11 coulomb will suffice, asagainst the 5 X 10-10 coulomb needed with a positivepulse. Secondly, the input charge Qin is constant overa wide range of Vas: this makes the' sensitivitycomparison in practice even more favourable tothe negative-pulse circuit. Thirdly, since the func-tion of the auxiliary cathode in initiating break-down now ends with the passage of the signal charge,there is no objection to inserting an impedanee inthe auxiliary cathode lead which will prevent in-stability even if the priming current is equal to Itb'The better sensitivity of the negative-pulse circuit

follows from reducing the function allotted to theinput signal. The essential difference between thetwo modes of operation is that with the negativepulse applied to the .auxiliary cathode, the triggerdiseharge has to be extended only over the smallauxiliary-to-main eathode gap instead of over themueh larger anode-to-trigger gap.'This account of the action is supported by the

absence of a variation of Qin with Vas, unlike thepositive pulse circuit. Breakdown between triggerand main cathode should be largely unaffected byVas; while once this stage has been passed thetrigger excursion (about 40 volts) will release4.X10-9 coulomb, suffic~ent to cause anode break-down for quite small anode supply voltages Vas·

Freque,;c)' response

The relation between Ra and anode-circuit timeconstant RaCafor values of auxiliary-cathode resistorbetween 4 and 20 MD is such that, for the normalanode swing of 70 volts, an adequate extinctioncurrent is obtained even at Ra Ca = 300 f1.sec.Against this, the trigger-circuit time-constant is10 m sec for the val~es of fig. 15a. This makes thetrigger circuit the limiting factor; the value of 10.rn sec is too long for a practical radiation monitor.This difficulty was overcome by connecting Ct to theanode instead of to ground (fig. 15h). If Ca~ Ct,then the anode waveform is little affected, while thetrigger - after a depression :_ is restored throughpositive ion collection by the end of the anode swingto a potential normal. for this stage of the cycle.Thereafter anode and trigger potentials rise togetherand the whole circuit is re-set.

140 PHILlPS TECHNICAL REVIEW VOLUME 18

Until the above circuit change :was made it had.Ó» been assumed- that a return to the normal -input

sensitivity was dependent only upon completion of, 'cir~uit recovery together with the re-establishmentof the priming current. Moreover; it seemed likelythat incomplete de-ionization would make the circuitexceptionally responsive for a period. These expecta-tions, however, were not borne out by the results., Fig. 16 shows the v~riation of the required Qin withtime elapsed since the Jast input pulse. Circuitequilibrium' is restored after about 1 msec, buteyen after 10 msec sensitivity is stili rising slightly.

_~Olr---.---~---r--~----r---~---r--~X 10 , coulomb

60r-+-r---r---+---+---+---~--~~

\40'r---~--'_---+----r---+----r---+--~r-i.201r---r---r---+---+---+-~~--~~

5 10 15~t

Fig. 16. Recovery of input charge sensitivity. About 1 millisecafter the I?ain discharge has passed, the charge sensitivity ofthe tube has reached a value of about 5X 10-11• At 10millisecthe sensitivity, is 3X10-11 and still improving slightly.

20msec

"

The conclusion seems to be that the correctaction of the Z 801U tube depends on the develop-ment of a very sm~ll unst'able discharge betweentrigger mid main cathode, mid th~t this unstabledischarge cannot occur until' the general level ofionization in the tube h~s fallen too far. for thetrigger-resistor current Itb to be provided by ioncolleetien alone. De-ionization in this sense is likely.to be protracted, being delayed by the persisteneeof metastable atoms and by a measure of ipnreplacement once ciréuit recovetf is' complete.

, ('

The cure for the delay is evidently to acceleratethe process of de-ionization. This has been' done inlaboratory models having an electrode "labyrinth"in which the discharge passes through narrowpassages rather tha~ through' k' void. However,

as the radiation monitor for which the Z 801p wa~developed can acccpt the frequency responseprovided hy the ~tandard structure (up to 1000 cis),'it has proved to. be unnecessary to increase thefrequency response up to 10 kc/so. ,

Charge amplification

In the .rate meter circuit for which the Z 801U wasdeveloped, .the function of the tube is to deliver alarge fixed .output charge every time the tube isswitched by the arrival of a small input chargenot less than 3X10-11 coulomb. So far as we' areaware, this represents an input sensitivity far higherthan that obtainable with any other type of cold-cathode tube. The tube provides, at low frequencies,an amplification factor' of not less than 104, sinceQout may b~ ás much as 5 X 10-7 .coulomb.If it is qesired to make use ,of input pulses of one

volt or less, then the effective input' capacitancemust exceed 30 pF. A capacitance of this size wo~ldupset the stability of the trigger circuit, and musttherefore be isolated from the tube by means ofselenium diodes 12).

The radiation monitor which uses the Z 801U tubeas' a rate meter tube now also employs another in avariant of the Goulding stabilizer circuit to providean output of about 155 volts.

The two tubes described above were developedspecifically for their intended applications, viz. thestabilizer and rate meter circuits of a radiationmonitor. A third tube, the, Z803U, has, however,~lso been developed for more general application. Inthis tube the priming electrode is an anode, a primingdischarge of sO,me10 !LA flowing between it and thecathode in a region shielded from the trigger-cathodespace, The tube has the same stable triggering char-açteristics as its predecessors; the nominal triggerbreakdown voltage Vtb is 132 V. Anode breakdownvoltage is 290 V, while the maintaining voltage Vamis 105 V. The tube can pass a mean current of 8mA.

Another general purpose tube, type Z70U, of sub-miniature dimensions is in course of' developmentat Eindhoven. This tube can be-used with a highersupply voltage than the Z 803U but its meancurrent capacity is less. A further article will appearin due course in this. Review describing thesedevelopments.

Acknowledgement should be made of the contri-irib~tion to this work of Messrs E. Franklin,

1?) See the article by F. S, Goulding referred to in By.'

~956f57, No .. 4/5 COLD-CATHODE TRIGGER TUBES 141

K. Kandiah and F. S. Goulding of the ElectronicsDivision of A.E.R.È., Harwell, dire~ted by r».D. Taylor.· .

Summary. After a briefintroduction dealing with the operationand properties of cold-cathode trigger tubes, work is describedon the development of two special trigger tubes for use in aradiation monitor. Two of the measures taken to improve.reliability and performance are dealt with: the use of molyb-denum sputtering to hchieve better stability and the provisionof a priming current to' eliminate the statisticallag. Both tubesare designed to work in self-quenching circuits. The 'stabilizer,

tube, a tetrode with an auxiliary cathode, is designed forproviding a staliilising supply for Geiger tubes. The' circuitconditions for-satisfactory operatien are described. The ad-vantage~ .of this stabilizer are its voltage range (200-300 y),its current range (5-150 jJ.A)and the fact that the stabilizedoutput- voltage can readily be adjusted by changing the valuesof two resistors. The rate meter tube is designed to handleGeiger tube pulses. It is also' a tetrode but differs from thestabilizer tube in that its priming electrode, again an auxiliarycathode, lies in the path of the main discharge and provides a"direct" priming discharge. The special characteristic of thistube is its extremely high sensitivity; in "the negative' pulsecircuit, the input charge sensitivity is 3X 10-11 coulomb. Thistribe can also be used for voltage stabilization.

, ,

~N AUTOMATIC NOISE, FIGURE INDlCATOR,.

.The output sig~ai from any amplifier is inevitably

, accompaniedby a certain kind of disturbanèe in thesame .frequency range as the signal. Amongst thecauses of this kind of disturbance are the thermalmotion of the electrons in resistors and irregularitiesin the transit time of electrons in valves. Suchdisturbances are given the general name of noise,irrespective of their origin; their amplitudes have arandom, distribution, that is, in accordance with aGaussian curve.The extent to which an amplifier produces noise

is expressed by the noise factor F. For a linear.amplifier [having an impedance Zi between itsinput terminals) F is defined, for a frequency f,as theratio of the noise-output: power LIfn in the frequencyintervalf ± îLlfto tbat part of LlPÎl caused by thethermal agitation noise '~f Zi when Zj is at a tempe-rature of 290 OK (17 Oe). '

The noise factor is a function of frequency. To~ake it possible none the less to characteri'ze thenoise properties of an amplifier by a single quantity,the mean noise factor Fm has been introduced. Thisis by definition 1) the ratio of the total output-noisepower Pn to that pàrt of Pn caused by the thermalagitation noise of Zi, Zi again being at a temperatureof 290 OK.Whenever mention is made of the noisefactor in what follows,it is the mean. noise fact;or Fmthat is meant.

It is often useful to be able to measure the noise,factor of, for example, amplifiers in the relaystations of short-wave radio' links, or that of,equipment f07 radio-astronomy 2). One method ~f

1) Thls definition is in accordance with American Standard53 IRE 7 SI. Other definitions may be found in 'the

, Iiterature; see for example Philipstech. Rev. 2,329,1937; 6,. 133,1941; 11, 84, 1949/50.

2) See for example C.A.Muller, Philips tech, Rev. 17; 305-315and 351-361, 1955f56.

, , I

621.317.34 :621.396.822

measuring the noise factor is based on the fact thatthe power of the thermal noise produced by a resistoris proportional to the, absolute temperature of theresistor. In this method the output noise poWCl;ismeasured (with a thermocouple) with two input,resistors: first with a resistor thai:/has a suitableresistance at room temperature, and then with aresistor that has the same resistance at anotherknown temperature, e.g. that of liquid nitrogen.From the two temperatures, the resistance, and thetwo output noise powers, the noise factor can thenbe readily evaluated.Another well-known method is based on the

principle of passing a known noise current throughthe input resistor, e.g. the ourrerrtflowing througha saturated diode 3). The r.m.s. value of i, thefluctuations in the current, is determined by

i2 = 2eILlj, (1)

e representing the charge on the electron and I themean value of the diode current; If we connect aresistor Ri, whose resistance is small compared tothe internal resistance of the noise diode, betweenthe input terminals. of the amplifier, and th~n pass'through it a noise current of such intensity that thenoise power .at the output is exactly doubled, thcnoise figure .is given by:. , • I

, , eFm =--IRi,., 2kT

where k is Boltzmann's constant and T the absolute'. ~ - .

temperature of the resisto~Ri,Inserting the numeric-al values of the constants e and k (e ' 1.60 X 10-19

coulomb, k = 1.38 X 10-23' joules;oK) and 290 oK. . .'

3) See for example Philips tech. ~ev, 2, 330, 19117.