SEPTEMBER 1939 267

AN ELECTRON SWITCH

by C. DORSMAN and S. L. de BRUIN.

An apparatus is described with which the time function of two different magnitudes canbe observed simultaneouslyon the fluorescent screen of a cathode-ray oscillograph. Anumber of practical examples are discussed where such investigation is of value, inter alia,the simultaneous registration of vibrations at different points of a mechanical system orof the current and voltage conditions in a transformer.

With the aid of a cathode-ray oscillograph, suchas the GM 3152 recently described in this Review 1)the image of an electrical tension can be representedon the fluorescent screen of a cathode-ray tube inrelation to a second voltage. If a linear time baseis used, the curve of the variation of the voltagewith time can then be projected on the screen. Inaddition, a third variable is available, viz., the in-tensity of the cathode beam which can be modulatedin relation to a third magnitude. In this way twomagnitudes can be simultaneously represented as afunction of the time, displacements in phase in themagnitudes under measurement being revealedin the projected images. But this method does notreveal any small variations in the third magnitude,so that it becomes essential to devise an apparatusfor registering two curves simultaneously showingthe variation with time of two different voltages.Apparatus designed for this purpose have often

Fig. 1. Measuring arrangement comprising an electron switch,Type GM 4196, and the associated cathode-ray oscillograph,Type GM 3 152.

1) Philips techno Rev. 4,210, 1939.

621.317.755.06

been described III the literature, and their under-lying principle has also been discussed in thisReview 2). The present article gives a description ofan apparatus, the socalIed electron switch GM 4196,which has been devised for this purpose, togetherwith details of certain of its applications. A viewof the measuring outfit, consisting of the electronswitch and its associated cathode-ray oscillographGM 3152, is shown in fig. 1.

Method of operatien of the electron switch

A circuit is shown in fig. 2 with which the timefunctions of two electrical tensions VA and VB canbe represented simultaneously. These two voltagesare impressed on the control grids of the two pen-todcs LA and LB, and an alternating voltage of

Ra

o

SAJZi---+-VA 0-1~---11--

32354

Fig. 2. The two voltages VA and VB to he observed simul-taneously by the cathode-ray oscillograph 0 are applied tothe control grids of the pentodes LA and LB. The switchingvoltages SA and SB varying rectangularly with time are ap-plied to the screen grids. The anodes of LA and LB are connect-ed through a common anode resistance Ra to a constantpositive voltage.

10000 cycles frequency varying rectangularly withtime is applied between the two screen grids. Thisvoltage is of such a magnitude that the two valvesLA and LB alternately either function in thenormal way or are in such a state that no anodecurrent passes through the valve. In consequenceof this, a current passes through the common anoderesistance Ra which flows alternatively through LAand LB and is hence controlled in turn by the ten-sions VA and VB at the control grids of LA and LB.The voltage through Ra thus fluctuates with a fre-

2) Cf also Philips. techno Rev., 3, 154, 1938.

. 268 PIIILIPS TECHNICAL REVIEW Vol. 4, No. 9

quency of 10000 els between two values whichare determined by VA and VB respectively.lf, now,the anode voltage is applied to the first pair ofdeflecting plates of the cathode-ray oscillograph 0,while a linear time base is applied to the otherpair of plates, an image of the type shown in fig. 3will be projected on the screen.

Fig. 3. Oscillograph image obtained when two oscillogramsare, registered simultaneously with the aid of the electronswitch. Actually the number of impulses is, of course, muchgreater than sketched here.

To ensure that the light spot travels sufficientlyswiftly between the projected curves (fig. 3) ofVA and VB, the amplifier must be suitable for am-plifying a rectangular oscillation of 10000 cyclesper sec. It must, therefore, be able to amplify notonly the fundamental wave of 10 000 cycles butalso'have roughly the same gain for a large numberof har~onics of this fundamental wave. In thecathode oscillograph GM 3152 the gain remains

, I 'practically constant up to 106 cycles, so that thehundredth harmonic is ~otyet severely attenuated.

of'

,3"23"S:!r

Fig'. 4. Circuit for generating the switchingvoltages SA and Snvarying rectangularly with time, at the anodes of the pentodesLe and LD, whose control and screen grids are eapacitivelycoupled with each other.

Fig.4 shows the circuit used for obtaining thealternating voltage varying' rectangularly withtime and which is required for the screen grids ofLA and LB. In principle this circuit representsa multivibrator as designed .by Abraham andBlo ch. Between the "two correspouding pentodesLe and LD two condensers are connected which linkthe control grid' of each of these valves with thescreen grid of the other valve of the pair. The stateof equilibrium, which would be obtained if thevoltages of the control grids for the two valveswere equal, and likewise those of th~ screen gridsfor the two valves, is found not to be stable: If, forinstance, at' a certain moment the' voltage at the

control grid of Le is too high, the current through Lewill increase, so, that a greater amount of currentwill have to be dissipated by the screen grid resist-ance of Le, which will reduce the screen grid voltage,of Le and owing to the capacity coupling also cutdown the control grid voltage of LD. A lowercurrent will then flow through LD and the screengrid current of LD will fall, and hence the voltageat the screen grid of LD will rise. The presence ofthe. capacity coupling will then also raise thevoltage at the screen grid of Le. The condition of. the circuit hence will be labile, since we startedwith the assumption that the control grid voltageat Le was already too high. The increase in thevoltage at th~ screen grid of LD is limited by theanode voltage to which the screen grid is connectedover a resistance. If the screen grid voltage of LDcannot increase further, the control grid of Le will 'discharge itself through its resistance leak, with theresult that the whole process will be repeated in theopposite direction. The relaxation time of this os-cillation is hence determined by the capacity andthe leak resistance of the control grid. .If now the anode voltages of Le and LD, which

vary roughly rectangularly with the time, are ap-plied to the screen grids of LA and LB, these valveswill be working or idle in turn, so that the amplifiedvoltage VA' and VB are' applied alternately to th~deflecting plates of the cathode-ray tube. Since theoscillator circuit lies between the screen grid~ ~iof Le and LD, while the reversing voltages are takenfrom the an 0 des of these valves, the oscillatoris practically unaffected by the voltages which maybe applied to the valves LA and LB and which areto be registered by the oscillograph.

Assume that the voltages VA and VB to beregistered have constant values and are equal toone another and that changing-over takes placeinstantaneously by means of pure rectangular anodevoltages of Le and LD,' the, total anode voltage of

LA+L8tr-~--~+---~~--r--t 3"23"28

Fig. 5. Anode currents 'of the pentodes LA and Ln if theyare increased and decreased linearly with time.

SEPTEMBER 1939 AN ELECTRON SWITCH 269

LA and LB will then also be constant and therewill be no indication on the screen of the changing-

___ t3'23'2f!

Fig. 6. Actual shape of the time functions of the anode cur-rents of LA ami LB and of LA + LB.

over operation. This would still be the case if theanode currents of LA and L B during changing-overvaried linearly with the time as shown in fig. 5.

smooth curves become visible on the screen. Toobtain this result it is not enough to make theswitching signals purely rectangular, so as to obtaina swift change-over from one image to the other,but, in addition, the frequency of reversal persecond must be made so great that the large numberof small dashes appear to the eye to merge into asmooth and continuous curve. If a switching fre-quency of 10 000 cycles per sec. is employed, theoscillogram of a function with a periodicity of 50will show practically no breaks, as may be seen fromfig. 8. Furthermore, the tenth harmonic of 50cycles, whose frequency is 1/20 of the switchingfrequency, will also still have a satisfactory outline.If a switching frequency lower than 10 000 cycleswere used, the subdivision of the oscillation of500 cycles would become too coarse, while if theperiodicity were made very much greater than10 000 cis difficulties would arise in obtainingswitching signals with a sufficiently rectangular

a

S2S'6

b c

Fig. 7. The shape of the switching voltages SA and SB is shown in fig. a). The shape of thesorillogram for VA = Ve is shown in fig. b), and for VA ~ Vn in fig. c).

Actually, however, the flanks of the anode currentimpulses of LA and LB are always curved (fig. 6), sothat the total anode current which passes throughRa, and hence the tension across Ra, will showsmall troughs. The voltage between anode andearth, which is passed to the oscillograph, willtherefore reveal small peaks at the moments ofchanging over from one of the voltages to be reg-istered to the other, both when the two voltagesare equal and when they differ (fig: 7). These una-voidable irregularities can, however, be minimisedby making the current impulses as steep as possible,when the peaks are barely visible, as shown infig. 8.

Shape of projected curves

The curves registered simultaneously by the os-cillograph when using the electron switch are madeup of short dashes which must follow each otherIII such close succession that fairly sharp and

time function, since the capacity of the current-carrying parts gives too low an impedance forthese high frequencies.

Fig. 8. With the electron switch two curves are obtained whichto the eye appear to have a continuous outline.

Registration of mechanical oscillations

To register mechanical oscillations with a cathode-ray oscillograph, these must first be converted intoelectrical voltages, for which e.g. an electrodynamicsystem can be employed: an electrical coil moves

270 PHILIPS TECHNICAL REVIEW Vol. 4, No. 9

in the field of a permanent magnet (fig. 9), the coilbeing attached to the object whose vibrations areunder investigation.

Fig. 9. A coil oscillates in the field of a stationary magnet withnorth-pole N and south-pole S. In the moving coil electricaltensions E are induced which are proportional to the velocityof deflection v. •

In the moving coil an electrical voltage is in-duced which is proportional to the velocity providedthe deflections sustained are not too great. If theoscillations at two different points of a mechanicalsystem are imparted to two different coils, bothoscillations can be registered simultaneously by acathode-ray oscillograph with the aid of the elec-tron switch, thus permitting direct visual com-parison with respect to amplitude and phase.

S2S19

Fig. 10. The large amplitude represents the mechanicalvibration of an electric motor, while the small amplitudesreproduce an alternating voltage with a standard frequencyof 500 cycles per sec.

The voltage induced by a mechanical vibrationcan also be registered in conjunction with anothermagnitude on the screen of the cathode-ray tube;thus in fig. 10 the vibrations of an electric motor(with a large amplitude) are compared with analternating voltage of 500 cycles as normal fre-quency (small amplitude). This comparison showsthat the frequency of the fundamental wave ofthe vibration is 53 cycles, so that the motor isrunning at a speed of about 3200 r.p.m. The fun-damental wave of the oscillation also contains anumber of notches; this oscillation of much higherfrequency has been produced by one of the movingparts rubbing at some point or other. The frequencyof this disturbance is 530 cycles.

Phase displacements between currents and voltages

It is frequently important in the examinationof electrical plant to determine not only the timefunctions of various currents and voltages butalso to know the phase relations of these magni-tudes. With the aid of the electron switch, bothcurrent and voltage may be projected simultane-ouslyon to the screen of an oscillograph, so that thephase relations are directly visible. In an issue ofthis Review which appeared last year, the com-bined oscillograms of currents and voltages of agas discharge lamp were reproduced and discussed.Another exemple may be mentioned here, viz. thecurrent and voltage conditions in a circuit contain-ing a transformer with an iron core so highlysaturated that the secondary voltage remainspractically constant when large variations occurin the primary voltage. The principle of operationof a stabiliser of this type (fig. 11) has already been

,,2$27

Fig. 11. Circuit of a transformer with a highly-saturated ironcore, as a result of which the secondary voltage remainspractically constant, although the primary voltage may fluc-tuate widely. VI and V2 are the primary and secondary volt-ages respectively, and il and iz the primary and secondarycurrents. Z2 is the load, which is principally capacitative (anapparatus for anode-volts).

described in this Review 4). If a pure sinusoidalvoltage is applied to the primary side, a voltagewhich is heavily flattened (fig. l2a), i.e. with aconsiderable amount of third harmonic, is obtainedon the secondary side owing to the high saturationof the iron. The curve of the primary current alsoindicates a pronounced third harmonic, which,

a bFig. 12. The sinusoidal curve represents the primary voltageimpressed on tbe transformer; the flattened curve in a) is thesecondary voltage, and the peaked curve in b) the primarycurrent, which both have a pronounced third harmonic.

3) Philips techno Rev. 3, 156, 1938.4) Philips techno Rev., 2, 279, 1937.

SEPTEMBER 1939 AN ELECTRON SWITCH 271

AN ELECTRICAL MEGAPHONE

however, augments the maxima, as may be seenclearly in fig. 12b. Furthermore the figure revealsthat both primary current and secondary voltageare somewhat delayed with respect to the primaryvoltage.

Registration of pulsating direct voltages

As ordinary electrical amplifiers only amplifyalternating voltages, an oscillograph will usuallyreveal only the A.C. component of pulsating directvoltage. One of the main advantages of the electronswitch is that it enables us to study also the direct-voltage component with the oscillograph, since itconverts the direct voltage into an alternatingvoltage of the switching frequency used, and forwhich the amplifier of the oscillator happens to beparticularly suitable.

In the oscillograms in fig. 13 the two curves showthe zero line of the voltage and the pulsating directvoltage of a source of anode voltage. With a low

load of 100 milliamps the ripple of this directvoltage is obviously only small (fig. l3a), while apronounced pulsation is shown with a heavy loadof 500 milliamps (fig. 13b).

a b

Fig. 13. The oscillogram of a pulsating direct voltage as wellas the zero line can be registered with the electron switch.a) The voltage of an anode-volts supply reveals a small

ripple at a load of 100 milliamps.b) At a load of 500 milliamps the curve shown in a) acquires

a heavy pulsation.

by J. de BOER. 621.395.61

A portable electrical voice amplifier, the "Portaphone" Type No. 2831, has been devisedto obtain a greater range of the human voice than when speaking normally without direc-tive aids or when using an ordinary megaphone. This apparatus increases the intensityof the sound energy 30 to 100 times the gain realised with an ordinary megaphone. Hence,the range of the voice is 5 to 10 times larger with the "Portaphone" than with an ordinary

megaphone.

In ordinary speech the range of the human voiceis not very great, and various means have thereforebeen devised for increasing the range to which thevoice will carry. The oldest of these aids is thespeaking tube or megaphone which was inventedabout the middle of the seventeenth century bythe German divine K i r Cher and the EnglishmanMorland. Its action consists in imparting a direc-tivity to the sound waves so that they are concen-trated into a narrow beam with a small solid angle.Recent developments in amplifying technologyhave, however, raised the question whether betterresults in the transmission of the human voiceover great distances could not be realised withthe aid of a simple electrical amplifier. While re-taining the concentrating effect of the megaphonehorn, an apparatus of this type would moreoverprovide a source of sound which is more powerfulthan the human voice.

In constructing such a simple VOIce amplifier,the first requirement is to arrive at maximum con-venience in use. The loudspeaker, horn, and micro-phone should form a compact unit which can beconveniently carried in the hand, while the elec-trical amplifier may be accommodated in a separatecase. The "Portaphone", Type No. 2831, whichhas been designed on these general lines, is shownin fig. 1; the carbon microphone is mounted in theloudspeaker which also carries a horn and a handlewith switch. The flat case containing the electricalamplifier is carried by a strap; its total weight is6.8 kg. and in addition to the amplifier also holdsa 2-volt accumulator to furnish the filament currentand to feed the microphone, as weU as dry batteriesfor the grid and anode voltages up to 150 volts. Theelectrical gain is approximately 40 decibels. Theacoustical gain is somewhat lower owing to lossesin microphone and loudspeaker (cf. at the end of