VOL. 5 N~. 3 .

PhilipsDEALING WITH TECHNICAL PROBLEMS.

RELATING TO THE PRODUCTS, PROCESSES AND. INVESTIGATIONS OFN.V. PHILIPS' GLOEILAMPENFABRIEKEN

EDITED BY THE RESEARCH LABORATORY OF N.V. PHILIPS' GLOEILAMPENFABRIEKEN, EINDHOVEN, HOLLAND

TESTING A~PLIFIER OUTPUT VALVES BY MEANS OF THE 'CAT~ODE RAYTUBE

hy A. J. HEINS van der VEN. 621.317.755: 621.396.645

In testing amplifier output valves, the most important data are contained in the In' v..diagram if one knows over which part of the diagram the values of voltage and currentprevailing during operatien range, i;e. if the position of the load line is known. The la- Vn

'diagrams as well as the load lines can very easily b~ obtained with the help of a 'cathoderay tube. The necessary apparatus is described in this article. A number of auxiliary ar-rangernents are also studied, by which the axes and the necessary calibration lines in thediagram can be traced on the fluorescent screen, and which make it possible to. causethe diagrams of two .output valves which are to be compared to appear simultancouslyon the screen. In order to obtain the load line in the correct place in the diagram, use musthe made of dir'ect current push-pull amplifiers for the deflection voltages of the cathoderay tube. The position of the load line upon inductive loading is discussed and explainedby a number of examples. In conclusion one,npplication of the instaflation in the develop-ment of output pentodes is dealt with. .

If one wishes to characterize briefly the perfor-mance of an output valve of !tn amplifier or radioset, it is enough to give the sensitivity, the distor-tion as a function of the power output, and in some

, cases the maximum power which can he deliveredwithout grid current flowing.. ~or a more carefulexamination, however, particularly for the dis-covery of caus,es of deviations or errors, such forexample as differences in distortion in differentvalves, one m~st in general have reference to theIa- Va diagram, which. forms the actual basis fo~'the judgement of the properties of an outputvalve. This diagram (fig. 1) gives the variation ofthe anode current Ia as a function of the anodevoltage Va with the negative control grid voltageVg as parameter 1). From the series of curves theprimary quantities of the valve: slope, internalresistance and amplification factor, may be readoff immediately for every operating point, as isexplained in the text under the figure. A connectionmay, however, also be found between the shape ofthe curve and' the above-mentioned quantitiessuch as distortion and power output. Since these

1) The screen grid voltage, .which forms a second parameterin the 'case of pentodes, is kept constant for the whole·diagram.

quantrties are to some degree dependent _on theloading impedance which is included in the anodecircuit, it is first necessary to find out how the loa'd-ing of the ~alve is e~pl'esse~ in the Ia- V~ curve.'

mA120

100 19.0

-4V

-6V

-8V---tOY

-t2V100 200 300 -Va 400 SOOV

3'476:11'

Fig. 1. The la-V.. diagram of the output valve EL3' W is theoperating point; in this valve it ordinarily lies at v., = 250volts, 1., = 36 mA, v:~= -6 volts. The screen grid voltageis permanently fixed at 250 volts. The slope S is given by thevertical distance between two successive curves, the ampli-fication factor g by the corresponding horizontal distance,the internal resistance by the slope of the la- VIZ curve at theoperating point. AB is the load line for a pure resistance of70PO ohms.

If the valve is loaded with a resistance R whichis connected to the anode circuit through a trans-former in order to have no D.e. voltage drop in R(fig. 2), an alternating current IR will begin to

'62 PHILIPS TECHNICAL "REVIEW Vol. 5, No. 3

flow in R when an A.C. voltage is applied to thegrid of the valve. The A.C. voltage on the loadingresistance is then VR = R . I l'/' The total anodevoltage Va and anode current la of the valve aregiven by the sum of VR or IR and the values ofthe anode D.C."voltage or current indicated by theeperating point. The-relation between I a and Va ofthe loaded valve, the socalled load line, is thereforein this case represented by a straight line throughthe operating point (AB in fig. I), with the slopetan a = R with respect to the la axis.

(}-s-

Vg2 Va3'4764

Fig. 2. A loading resistance R is included in the anode circuit .of the output valve via a tr".nsformer 1:,1.

If one considers the grid voltage to be varying,the values of la and Va at any moment are alwaysdetermined by the point of intersection of the loadline with the la- Va curve corresponding to the value·of the grid voltage ~t that moment. If,for instancè,the operating point lies On the curve for Vg = -6volts .and the grid excitation voltage is sinusoidalwith ~. peak value of 5 volts, then the' points ofinter~ecti?n ofthe line Al!with thecurves Vg.=-1·volt and Vg - -ll volts give the extreme valuesof la and Va. The distance AC is therefore twicethe amplitude of the altemating current occurring,the distance BC twice the amplitude of the A.e.voltage. The power output to the loading resistanceat the grid excitation voltage in question is thu'sgiven by 1/4 of the area of the triangle ABC. :Insofar as the successive la- Va curves cut off

equal segments of the line AB, a proportionalchange in anode voltage and' current correspondsto any change in the grid voltage. There is therefore

- no distortion. With large amplitudes along the load.line, however, the segments cut off on AB becomegradually smaller (fig. I). The variation öf .anodevoltage and current with a sinusoidal grid excitation'voltage is then no longer sinusoidal, but exhibitsa flattening, The distortion occurring could hecalculated by careful measurement of the segmentscut off on the load line.,If the loading impedance is not a pure resistance, .

·the ioad Iine is not straight In the la- Va diagram.Depending on the grid excitation voltage a larger orsmaller portion of the whole diagram is the~ 'c?ve~ed

by the load ·lines. A knowledge of this region makesit possible to estimate the importance for the reo'production, of any irregularities occurring in thela- Va diagram.

In the following a method will be described bywhich not only the Ia-Va diagram, but also theload line can very easily be obtained. The principleof this method, which makes use, of the cathoderay tube, has already been outlined in this period-icalê), at least as far as the recording of I.a-Vadiagrams is conéerned. We' shall here go moredeeply into the practical execution of the methodand also give several examples of its application.

Instaflation for the 'recording of la- Va diagrams.

An la- Va cl,taracteristic can be registered on thescreen of a cathode ray tube in the following way .A given D.C. voltage is applied to the control gridof the output valve to be investigated, while theanode voltage is made to vary periodically 'fromzero to a maximum value. This anode voltage isapplied, via a potentiometer and an amplifier, tothe plates for horizontal d~flection of the fluorescentspot. The voltage on a small measuring resistance,through which the anode current flows, causes thevertical deflection, which is therefore proportionalto the anode current at every moment.

In order to trace different la- Va curves on thescreen successively, the grid bias must he givendifferent values successively for short times. Avoltage which variès in steps must therefore beapplied to the grid, and in order to obtain a lastingimage on, the screen of the cathode ray tube, thisvoltage varying in steps mu~t be run throughcompletely several times per second.The stepwise varying voltage is obtained by

means of a rotating switch SI which passes over28 contacts (fig. 3). Sixteen of these are connectedto the taps. of a potentiometer which correspondto the 15 steps of the. desired voltage series, seefig. 4. On the following 6 contacts (17 to 22) thereis a high negative voltage, so that the valve to beexamined passes no anode current at all duringthis time. By this means, and by the great rapidityat which the switch rotates (1000 to 1500 r.p.m.]it is possible to record the whole la- Va diagramwithout the valve becoming overloaded ..

For the sake of orientation in the diagram, thetwo axes and the corresponding scale are necessaryin addition to the la- Va curves. The remaining

2) H. van Su ch t el en; Applications of cathode ray tubes,Philips techno Rev. ,3, 339, 1939. See also Philips technoRev. 4, 56, 1939, where à related problem had to he solvedin a different way •.

MAÎfCH 194,0 TESTING OUTPUT VÄI..YES

Fîg, '3. The anode voltage of the output valve L1 to be investigated is applied to the platesfor horizontal deflection of the .cathode ray tube K via the amplifier VI; the voltage overthe measuring resistance Rl\f, through which the anode current flows" is applied via theamplifier V; to the plates for vertical deflection. By means of the rotating switch SI thegrid.bias of the output valve is varied in steps. The auxiliary switch 82 which is fastenedto the aids of SI short circuits the pairs' of brushes A, .B, C at certain moments, and thuscauses the various auxiliary lines of the diagram to be traced on the fluorescent screen. ...-The switches S:I and S.1 which rotate with half the velocity of SJ serve to record the dia-grams of two output valves L1 ánd L. at the same time, while the voltage on the grid g/,of the cathode ray tube is lowered for the diagram of one of the valves. With the helpof the potentiometer Pl the current through the potentiometer P can be regulated and thegrid bias stepwise variation can be made more or less .steep.

contacts änd .the auxiliary switch Sr which' isfastened to the axis of the rotating switch andpasses over the pairs of brushes A, B, C (fig. 3)serve these purposes.The horizontal axis is produced automatically

due' to the fact that during the currentless periods(cdntacts 17-22) the fluorescent spot moving backand forth is not deflected in a vertical direction.The vertical axis is traced in the diagram when theswitch passes' over contacts 23 and 24. At this

3"4766

Fig. 4. The variation of the grid bias of the output valve underexamination during one revolution of the rotating switch(SI i~ fig. 3).

19 21

moment the input of the amplifier £~i: the voltagefor horizontal deflection is short circuited via thè'brushes A. On the contactsPd and 24, which nowfeed the grid, an A.C. voltage now acts só that theanode current may vary sufficiently to describe

, I

the whole la axis. The same is true for the followingtwo contacts, 25 and 26.·At the moment at whichthese contacts are ;connected to the ~rid, thesecond pair of brushes B bringsabout a substitutionof the anode A.C. voltägè by the normal anode D.C.voltage of 250 volts, for .instance. The vertical lineVa ' 250 volts is therefote now drawn in the dia-gram, which facilitates finding the operating point,and at the same time provides the scale for thehorizontal axis. 'I'he scale for the vertical axi~ isobtained by passing a direct current of knownmagnitude through the measuring resistance R1!lduring the resting period of the valve (contacts. 17 and 18). This is done by means of the two brushesC, and gives in the diagram a horizontal line at adefinite height above the Va axis. .The remaining two contacts 27 and 28 are at a

. "negative voltage of a given magnitude, -6 voltsfor instance. The Ia- Va curve with the parameterVg == -6 volts is traced via these contacts .. By.regulating the potentiometer .eurrent and th~sthe voltage on the contacts 1 to 16, one of the

63

64, PHILIPS TECHNICAL REVIEW Vol. 5, No. 3

15 Ia- Va curves can be made to coincide with theestablished curve for Vg = -6 volts, and the param-eter value for all the other curves is thus alsoknown.It is, however, seldom necessary, and with regard

to clarity, it is sometimes even undesirable torecord 15 1(1- V(I curves of the diagram. In thatcase the stepped voltage in fig. 4 is made so steepby increasing the potentiometer current, that onpart of the contacts 1 to 16 the voltage is highenough to completely suppress the anode currentof the output valve. The valve then has a longerresting period.

The comparison of two output valves

In addition to the rotating switch S1 and theauxiliary switch S2 there is a rotating contact S3which rotates at one half the speed of the firstswitch (fig. 3 and fig. 5). By means of this contactthe grid bias according to fig. 4 may be appliedalternately to two output valves. On the screenof the cathode ray tube the 1(1- V(I diagrams offirst one and then the other valve are drawnalternately, and since this takes place for eachvalve about 8 to 12 times per second, the twodiagrams appear simultaneouslyon the screen.This makes it easy to detect quickly small dif-ferences between two valves of the same type.In order to be able to distinguish the two sets ofcurves from each other, the current of the cathoderay is diminished slightly with the help of the

Fig. 5. The combination of rotating switches (S, to 54 in fig. 3).For structural reasons, the switch arm of S, (a brush) isstationary while the 28 contacts (lamellae of a collector,middle axis on the right) rotate beneath it. The 15 resistancesof the potentiometer P turn with the eollector, the necessaryvoltages are applied via five slip rings. The auxiliary switch 52'also for practical reasons, is divided into three contact makers(middle axis, left). The front axis turns at half speed andmoves to:two switches S~ and 54'

contact S4 during the time when one of the valvesis in circuit, so that the set of curves for this valveappears on the screen with a lower light intensity.

Fig. 6. Photograph of the Ia- Va diagrams of two valves takenat the same time. One diagram is traced with lower intensityfor the sake of distincion.

Fig. 6 gives an example of such an applicationof the apparatus. One valve (curves with lowerintensity) has a smaller slope and a higher maximumanode current at Vg = 0 (highest curve). Further-more the Ia- Va curves of this valve do not run asclose to the la axis as those of the other curve,a fact which has an unfavourable effect on thedistortion at high values of the grid excitationcurrent. Without going more deeply into this matterit may be noted that these deviations are causedby the fact that the cathode of this valve was notaccurately centred with respect to the control grid.

Tracing the load line

When the 1(1- V(I diagram has been photographedwith a camera set up in front of the fluorescentscreen, the load line mayalso be recorded on thesame negative. For this purpose the output valveis first hrought into the correct operating condi-tions, i.e. by means of suitable D.e. voltages ongrid and anode the operating point is arrived at,and the loading impedance is put into the anodecircuit via a transformer. If the anode current isnow projected on the screen as a function of theanode voltage once more, the load line is obtained.There is however still a complication: withoutspecial precautions the load line is not in the correctposition in the diagram. Ordinary amplifiers, suchfor example as those built into cathode ray oscil-lographs, do not pass D.e. voltages. When a voltage,whose mean value is not equal to zero, and whichtherefore has a certain D.e. component, is appliedto the cathode ray tube, the image on the screen

MARCH 1940 TESTING OUTPUT VALVES 65

always adjusts itself so that this average value ofthe voltage lies at the centre of the screen. Whensuch amplifiers are used, therefore, the Ia- Vu diagramis projected 'Oli the screen in such a way that thecentre of the screen is the point of greatest densityof all the values of voltage and current occurring.The same is true in the case of the load lin~. Ac-tually, however, the. cen!re of the load line mustcoincide with the operating .point of the diagram,which, means that amplifiers used for this purposemust not suppress the D.C. v~ltage component.

A'B

Fig. 7. First stage, of the D.e. push-pull amplifiers for thedeflection voltages. By giving' the resistances Rl-R4 suitabledimensions, upon application of an A.C. voltage to AB, thepotential at a is made to increase by the same amount asthat at b falls. The D.e. voltage potential at both pointsis adjusted to the same value by regulation of the anode direct

, .current,

In order. to obtain an accurately focussed image,op the screen of the cathode _ray tube, the potentialin the middle of each set of defle~tion plates mustremain constant (the focussing of the electronbeamdepends upon this). This means that the voltages~lfst be applied to the, deflection plates in a push-pull conp-ection. One thus arrives at the somewhatunusual requirements of D.C. push-pull amplifiers.

Fig. 7 shows diagrammatically how the first,stage of these amplifiers is arranged. The 'valueof the resistance is for, example the following:R3 = 1RRI' R4 = 2 Rl' R2 = R3 + R4 = 20 Rl'If an A.C. voltage is applied to the input side AB,an A.C. voltage occurs' over Ra + R4 which isten times as great as that on RI' The A.C. voltage

over R4 is thus equal to that over Rt, but in op-posite phase. Since the potentialof a at everymoment rises as much as that of b falls, the tensionbetween a and b is the desired push-pull voltage,provided a and b have the same D.C. voltage po-tential. This may he: realized by regulation of theanode direct current, for instance by means of anadjustable resistance in series with the' cathode,or by a suitable adjustment of the screen, gridvoltage. The' further amplification of the symmet-rical halves of the push-pull voltage is carried outin the mann~r usual for D.C. amplifiers. A diagramof the complete circuit is' given infig. 8. The pointsa and b correspond to the points a and b .in fig. 7.It may be seen that in the practical applicationno special battery has heen used for the screen grid.In addition to the anode current the screen gridcurrent now also flow~through the cathode resist-ance. Therefore the ratios between the resistancesRI to R.l (fig. 7) are slightly altered.. Fig. 9 is a photograph ?f the whole apparatus.

Load lines for different cases

In the simples t case in which the loading of theoutput valve consists of a pure resistance, the loadline is a straight line through the operating point,'~s was indicated in 'fig. 1. Fig. 10 i~;a photographof such a case. The loading resistance in this' caséwas equal to the so-called optimum resistance withwhich the output valve delivers the maximum powérat a definite distortion (10% in' this case]. Foroutput pentodes the optimum" value is usuallyequal to the quotient óf anode D .'C. :;oltag~ andcurrent. In that case the _a:rllpliiud~s of the ,~:nodèA.C. voltage'~~d' current can simultaneously reacht~eir maximuni <vahiés, which ..are approximatelyequal to ,thc' anode :D.C., voltage' and' current,respeciiveÎy.'···, , . . .': " . ~,:

In practical' cases .tl~e load OIi an output valveusually consists of one or more loud speakers. The. . .. .. .-.

i

"

, .

Fig. 8. Diagram of the"complete circuit of the D.e.~push.-pull. amplifiers a and b correspondto the points a and b in fig. 4. '

+270V +180V+700V--~----------~-----------+-

3"4768

66 PHILIPS TECHNICAL REVIEW Vol. 5, No. 3

Fig. 9. View of the complete installation. L is the output valve under investigation, RM themeasuring resistance for the anode current, V the two amplifiers for the deflection voltages,K the cathode ray tube, C the camera set up in front of the screen. On the meter A thecalibration current can be read off which is sent through the measuring resistance RMduring the resting period of the output valve (contacts 19-20 in fig. 4). \""\1iththe poten-tiometer PI' the steepness of the grid bias stepwise variation is regulated.

impedance of these is not a pure resistance but isinductive for a large part of the frequency range,due to the self-induction of the loud-speaker coil.The variation of the absolute value and the phaseangle of the impedance of an ordinary loud speakerof good quality with appropriate transformer is

Fig. 10. Load line for the case in which the output valve isloaded with a pure resistance. The resistance here has theoptimum value: the operating point lies in the middle of thesection which is cut off on the straight load lines by the axes.

represented infig. 11. As a measure of the magnitudeof the impedance the value at 1 000 cjsec is generallytaken. In our case this amounts to about 7 000 ohms,which corresponds to thc optimum resistance forthe 9 watt pentode EL3. The phase anglc at thefrequency mentioned is abou t 450•As a result of the phase shift between anode

A.C. voltage and current a straight line is ingeneral not ohtained in the Ia- Va diagram of theoutput valve. With a grid A.C. voltage of givenfrequency and amplitude, a more or less ellipticalload line is obtained around the operating point.In fig. 12 a number of such ellipses are shown whichare obtained at different amplitudes of the grid A.C.voltage. If the grid A.C. voltage contains differentfrequencies more complicated figures occur. Exam-ples are shown in figs. 13 and 14 in which twofrequencies, having the ratios 1 : 4 and 1 : 15,respectively, were applied to the grid, and in fig. 15in which three frequencies were combined. Fi-nally, infig. 16 the image is given which is ohtainedwhen the output valve, loaded with a loud speaker,is allowed to amplify music for some time. Onecan no longer speak of a load line, but of a loadfield which occupies a more or less extensive portionof the Ia- Va diagram.

fl25000

TESTING OUTPUT VALVES 67MARCH 1940

20000Z

115000

I",\

7 :\Zf '\-, 6

//. "//

>---r-.._ .>: \

.,:.::. \ J

~~~ \qJ/

/~

/

s- ,0

70 200 300 500 1000 2000 JOOO,5000 10000\ -y,,,\ 31,"-

10000

5000

o

$4769

Fig. 11. Variation of the absolu te value Z and the phase angle rpof the impedance of a normal loud speaker of good qualityas a function of the frequency v in cjsec. In this measurementa condenser of 2 000 [J.F was connected in parallel with theprimary winding of the loud-speaker transformer, as is thecase in many radio receivers. At 1000 cjsec, Z = 7000 ohms.At lower frequencies Z changes only slightly: the minimumlies at' 200 c/sec and amounts to about 5 000 ohms. At higherFrequencies Z first increases sharply to about 21 000 ohms at3 000 c/sec, it then decreases again due to the influence of thecondenser. Cheaper types of loud speaker show in general agreater variation of Zand sp with the frequency.

The form of the la" Va diagram

After an -idea has been obtained III this way ofthe size of the region used in the la" Va diagram,the requirements can be more precisely definedwhich the la" Va diagram must satisfy in order toobtain as little distortion as possible in the repro"duetion and as large an output as possible.

Since the anode A.C. voltage and with it theoutput is limited by the crowding of the la" Vacurves near the vertical axis, it is desirable that thecurves should run as closely as possible along thevertical axis at low anode voltages or in otherwords that in this region the anode current shouldincrease very rapidly with the anode voltage. Bythe introduction of the screen grid it was possibleto satisfy this requirement, without being corn"pelled to use positive grid voltages, involving aconsumption of energy by the grid as in triodes.At the same time, however, in the tetrode thusformed the phenomenon occurs that secondaryelectrons, formed at the -anode, may pass to thescreen grid, which causes a kink in the la" Va.curve (fig. 17a). In order to avoid these kinkswhich are accompanied by great distortion, a thirdgrid (suppressor grid) is introduced, which is atcathode potential, and which suppresses the sec"ondary emission from the anode to the screen grid.In the older types of pentodes, however, this sup"

a'

0'

a'Fig. 12. With inductive loading of the output valve (connectionto loud speakers) the load line becomes a more or lessdistorted ellipse around the operating point when the gridexcitation voltage is sinusoidal. In this recording the am-plitude of the grid excitation voltage was varied in steps withthe frequency constant. -

Fig. 13. Two A.C. voltages with a frequency ratio of 1 ; 4 areapplied to thc grid of the inductively- loaded output valve.

Fig. 14. Like fig. 13, but with a frequency ratio of 1 ; 15.

68 PHILIPS TECHNICAL REVIEW Vol. 5, No. 3

Fig. IS. Three frequencies are supplied to tbe griel.

pres sion was often so thorough that at low anodevoltage part of the primary electrons emitted bythe cathode were also forced to reverse their direc-tion in front of the third grid. These electrons werethen lost for the anode current, and the result wasthat the anode current increased less rapidly withthe anode voltage.

By a correct choice of position and dimensionsof the suppressor grid the decrease in power outputwhich hereby occurs can be kept very small. Inthe work involved in this development the methoddescribed in this article for the recording of Ja- Vadiagrams has proved very useful. A change in po-sition and dimensions of the suppressor grid is toa certain extent equivalent to a definite changein its potential. By applying different negative orpositive voltages to the suppressor grid - whichin contrast to the ordinary construction must haveleads to the outside - and by inspecting theIa- Va diagram in each case, it is possible to discoverthe voltage at which the most favourable form ofthe diagram is obtained. Fig. 17 shows three recordsfrom such an investigation. In a the kinks arevery pronounced, in b they have entirely dis-appeared, but at the same time the anode currentincreases much more slowly with the anode voltage,in c the correct oompromise has been found inwhich the curves mount as steeply as possiblewhile the kinks are only present in a region of thediagram which, according to the investigation ofthe load field (see above), is not used in practicalcases.

Fig. 16. The A.C. voltage of music is applied for some lime lothe grid of the output valve. A certain "load field" is nowcovered in the Ia- Va diagram.

CL b cFig. 17. The 1,,- V;, diagram of a pentode at different potantials of the suppressor griel.a ) The suppressor grid is too strongly positive, the kinks in the characteristics due to

secondary emission from the anode are not sufficiently suppressed.b) The suppressor grid is too strongly negative. The kinks in the characteristic have dis-

appeared, but at low anode voltages the curves rise less steeply.c) The suppressor grid has the optimum potential. This value of the potential gives an

indication of the sense in which position and dimensions of the grid must be changedin order to obtain the same result with this grid at cathode potential.