Voltage-Regulated Power Supplies*Design and Construction ConsiderationsALEXANDER B. BERESKINt, ASSOCIATE, I.R.E.

Summary-Power supplies with low regulation have always beenan important consideration in the electronic and allied fields. Thisnaturally dictates the use of voltage-regulated power supplies of thetypes described previously by Hunt and Hickman' and other investi-gators.

It is the purpose of this paper to discuss the problems involved andto develop an orderly procedure for designing and constructing thesevoltage regulated power supplies for specific applications. The correla-tion between design data and actual tests on a finished model will alsobe shown.

OPERATION OF REGULATOR CIRCUIT

EFERENCE to previous literature will show1t(i that many variations of the basic degenerative

regulator circuit are available for use but theparticular version on which this article is based isshown in Fig. 1.

keep the grid voltage of V3 close to the cutoff value.Under this condition the grid current of V3 is negligiblysmall and the following relation is established:

To trace the operation of the circuit, assume first of allthat the circuit has been previously balanced and thatthen, for any reason whatsoever, there is a tendencyfor Eo to increase. The increase in Eo will produce acorresponding increase in Es and the grid of V3 will be-come more positive, thus increasing the plate currentin V3. This increase in current produces a greatly am-plified increase in voltage across R5 and therefore

Fig. 1

The section of the diagram, in Fig. 1, shown to theleft of the dotted line is a conventional single-phase-full-wave rectifier with choke input to the filter. Thecondenser C3 is a condenser of very low capacitancewhich may be used to minimize "hash" if a gaseousrectifier is used and RB together with R7, R6, and V2serves the purpose of bleeding whatever current maybe necessary to obtain operation in the "flat" region ofthe rectifier output characteristic.The section of the diagram to the right of the dotted

line comprises the voltage-regulator portion of the cir-cuit and operates in the manner indicated below. Thecathode of the pentode V3 is kept at a constant poten-tial above the negative side of the line by means of theregulator tube V2 or any other constant-potential de-vice. If the resistance R5 is sufficiently large, then theover-all tendency of the regulator circuit is to keep thevoltage F8 approximately constant at a value that will

* Decimal classification: R356.2. Original manuscript receivedby the Institute, September 24, 1942.

t University of Cincinnati, Cincinnati, Ohio.I F. V. Hunt and R. W. Hickman, Rev. Sci. Instr., vol. 10, p. 6;

1939.

drives the grid of 14 more negative. This tends to de-crease the load current and therefore to bring Eo backto its original value. If the voltage Eo had decreasedinstead, a reverse of the sequence of events mentionedwould have taken place.

DESIGN FACTORS

In order to simplify the discussion of the factors in-volved in the design of the regulator circuit, it will bebroken down into its component parts and each partwill be discussed independently.The fundamental function of the V2 tube is to main-

tain a constant potential (ER) between the cathode ofV3 and the negative side of the line. Since a certainamount of current must be bled from the power supplyanyway, it is convenient to use one of the VR75-30,VR105-30, or VR150-30 tubes or a series combinationof any number of these tubes depending on the voltageER that is required for V2. Ordinary neon or argon"glow tubes" may also be used although there isusually considerably more variation in their character-istics than in those of the standard voltage-regulator

Proceedings of the I.R.E. 47Februa ry, 1943

Proceedings of the I.R.E.

line of tubes. The use of batteries, while perfectly feasi-ble, is usually not convenient due to the necessity ofoccasional replacement. Under certain conditions os-cillations may be started by the regulator tube V2 butthese can usually be stopped by a condenser C5. Thiscondenser is ordinarily not necessary.The R6-R7 combination of resistors should be de-

signed to limit the current through V2 to approximatelythe maximum rated value for the maximum value ofEi that is expected. At the same time, R6 and R7 shouldbe proportioned so as to produce the correct value ofscreen voltage for 173. In the cases where Es is liableto vary between wide limits, it may be desirable touse a VR105-30 tube in place of resistor R6 so as tomaintain a constant screen voltage on V3.The tube V3 and the resistor R5 form a simple direct

coupled-voltage amplifier whose function it is to am-plify the voltage applied to the grid of 13 as much aspossible. Triodes of the 6F5 type can be used for thispurpose, but their voltage amplification is usually quitesmall compared to that obtainable with the 57-6J7family of pentodes. The 57 tubes have been found towork quite satisfactorily with Ei voltages of 3 700 volts,using 0.25 megohm and higher values for R5. (As manyas eight 2A3 tubes in parallel have been used for V4with 1 megohm for R5 and 3700 volts for Ei.) With low-voltage operation it is convenient to use the 6J7 or6J7G tubes, but with high-voltage operation it is moreconvenient to use the 57 tube because filament trans-formers with the proper insulation are more readilyavailable for this tube. The possibility of using the802 transmitting pentode for this purpose, especiallyat the higher voltages, should not be overlooked.As explained before, the stable operating voltage Eo

is determined by the values of the voltage E8 and theresistors R1 and R2.

Eo (R1+ R2)ER1

If it is desirable to have a variable voltage Eo, R1 andR2can be broken up into two fixed resistors R3 and R4and a potentiometer R,. If the maximum and mini-mum values of Eo are chosen together with any con-venient value of RP, then the values of R3 and R4 maybe calculated in the following manner:Since

(R3 + Rp + R4)Eomax -= E

R3

and(R3 + RP + R4)

EOtie=-E8

therefore,RPR3 =

FOmax

Eom in

andR4= R3 -- 1-Rp.

EsThe value of ES required in the equations above maybe estimated very closely by subtracting from ERslightly less than the cutoff voltage of V3. As an ex-ample, if a VR105-30 tube is used for V2, ER is equalapproximately to 107 volts with 25 milliamperes flow-ing through it. If a 6J7 tube is used for 1V3 with a screenvoltage of about 100 volts, 6 volts on the grid just failsto produce cutoff in V3. From these values it can beestimated that E8 should be about 101 volts. The cor-respondence between design data using this value ofE8 and the actual voltage ranges obtained, as will beexplained later, shows that this is a very good ap-proximation to the correct value.

If it is desired to have several ranges over which thevoltage may be varied with the one potentiometer,then calculations of R3 and R4 can be made as indi-cated before for each individual range, and the com-posite values of R3 and R4 may be obtained by meansof a multigang selector switch.The exact amount of change of Eo that is applied to

the grid of V3 depends upon the ratio R1/(R1+R2), andif this ratio is very small, unreasonably poor regulationwill result. One good way of increasing this ratio forany particular value of Eo is to increase the value of EReither by using a different type of tube or else by usingtwo or more of the voltage-regulator tubes in series forV2. One of the disadvantages of increasing ER is thatit raises the value of the minimum Eo that it is possibleto obtain. This limitation will be explained later on.The sacrifice in the minimum value of Eo obtainableis usually well compensated for by the improved regu-lation that results.

Practically all of the variation in Eo due to ripplevoltage can be applied to the grid of V3 by shunting R2with the condenser C4. The capacitive reactance of thiscondenser should naturally be small compared to theresistance R1 at the ripple frequency. If condenser C4is used, the regulator circuit will become considerablymore sensitive to periodic variations of Eo at ripple andhigher frequencies than to other variations.While there are several types of tubes that can be

used for V4, the most satisfactory one appears to bethe 2A3 due to its large current-handling capacityand exceedingly low plate resistance. The operation ofthis tube in voltage-regulator circuits usually carries itinto the region beyond that required for ordinary am-plifier operation and for this reason a set of platecharacteristics for one of these tubes, up to 950 voltson the plate, is shown in Fig. 2. Since the permissibleplate dissipation on the 2A3 tubes is not ordinarilyspecified by the tube manufacturers, a test was run onone of the modern types of this tube. It was found thatthe plates began to show a visible glow in a dark roomwhen the plate dissipation attained a value between 35and 40 watts. It is concluded from this test that it

February48

13ereskin: V'oltage-Regulated Power Supplies

should be permissible to design the circuit, for voltage-regulator operation, using maximum plate dissipationson the 2A3 tubes of about 25 watts.

It is very often found that the full desirable rangeof the voltage regulator cannot be attained withoutexceeding the maximum permissible plate dissipationof the 14 tube. This problem can be very easily over-come by connecting two or more of these tubes, as maybe required, in parallel. In order to avoid parasiticoscillations that may develop between the paralleltubes, it may be desirable to connect a small amountof resistance in series with the individual plates orgrids of the tubes.

If the use of several tubes in parallel for V4 iS stillnot sufficient to cover the desired voltage variation, orif a closer control of the voltage is desired, then it maybe convenient to split the over-all voltage range intominor ranges and to change the transformer input volt-age suitably when the ranges are changed.

For any combination of this regulator circuit thereis a definite minimum output voltage which can be ob-tained and which depends to a large extent upon thevalues of ER, Es, and the plate current for the individ-ual 14 tubes. The manner of estimating this voltagecan best be shown by the following example: assumeER to be 100 volts and Ei to be of such a magnitudethat the plate voltage of the tube 14 is 600 volts, andfind the minimum value of Eo for a current of about 1milliampere through the individual 2A3 tubes used for14. Reference to Fig. 2 shows that the voltage of thegrid of V4 must be 160 volts negative with respect toits cathode. Assuming as an extreme case that theplate voltage of V3 iS only 20 volts, this means that thecathode of 4 must be 100+ 20+ 160 = 280 volts abovethe negative side of the line and, therefore, any furtheradjustment of the R,-R2 combination tending to re-duce Eo will only drive the grid of 13 positive withoutreducing Eo. In choosing the minimum value of Eo re-quired on the lowest range, therefore, care should betaken that this voltage is not less than the minimumvoltage permitted by the circuit. As mentioned before,it is usually desirable to sacrifice some of the minimumvalue of Eo obtainable in favor of improved regulationby increasing the value of ER.

If the minimum current involved in the above prob-lem were greater than 1 milliampere, there would be acorresponding decrease in Es, the plate voltage of theV4 tube, and the grid voltage required on this tube. Allof these changes would tend to reduce, slightly, theminimum voltage obtainable. For this reason it wouldbe desirable, in practice, to bleed a small amount ofcurrent on the output side of the regulator circuit inaddition to that bled by the R1-R2 combination. Avoltmeter across the output terminals would help fromthis point of view and also would indicate the correctoutput voltage at all times.A discussion of the rectifier circuit would not nor-

mally be in the province of this paper, but since the

ability of the regulator to maintain a constant outputvoltage depends on the burden placed on it, a shortreview of the factors that improve rectifier regulationwill be given here.

It is a well-known fact that the regulation of a rec-tifier with a choke input filter is superior to that ob-tained when a condenser input filter is used. In orderto obtain these benefits, however, it is essential that

/201-190 ,Ev 117111I/00- u

80- ~ ~ ~ ~-8

60-

o0 /00 200 300 Yoo 500 600ED - /o/fs

Fig. 2

700 800 900

operation of the circuit be limited to the "flat" por-tion of the output characteristic. Operation alongthis portion of the curve can be insured by a cor-rect proportioning of the first choke and the over-all bleeder current (minimum current drawn fromthe rectifier) so that the inductance of the chokeL=Ei/IbleederX 1/1131 henries. It might be well topoint out here that this is the inductance required atthe actual operating conditions and not the inductanceat zero direct current as rated by some of the manu-facturers. As mentioned before, the bleeder current ismade up of the currents through RB, the R1-R2 com-bination, the R7-R6-V2 combination, and any otherpermanent load that may be connected across the line.

In the case of high-voltage operation the relativevalues of L and Ci may be such that the ripple voltageappearing at Es is several hundred volts and this natu-rally would place a large burden on the regulator cir-cuit. In these cases it is usually convenient to use atwo-stage choke-input filter with a swinging choke inthe first stage and a smoothing choke in the secondstage.The selection of rectifier tubes, of course, depends on

the various currents and voltages desired. In the low-voltage range high-vacuum tubes find high favor whilein the high-voltage range tubes of the 866A type areindicated. Whenever gaseous tubes are used it is de-sirable to use a condenser C3 of about 0.01 to 0.05microfarad to eliminate the majority of the "hash"originating when the gas tubes "break down."

It is highly desirable, especially in high-voltage ap-plications, that the plate transformer used have asgood a regulation as possible and that the filter chokeshave a low direct-current resistance.

I.iz

1943 49

Proceedings of the I.R.E.

The voltage required on the secondary of the platetransformer may be readily estimated in the followingmanner:

direct output voltage of rectifier tubes = maximumoutput voltage required

+plate voltage of V4 at maximum current pertube (assuming E, =-20 volts as a safety fac-tor)

+direct voltage drop in filter at maximum cur-rent value.

DESIGN AND CONSTRUCTION COMPARISONSLow-, medium-, and high-voltage power supplies

have been designed by the author on the basis of theanalysis presented in this paper and a very satisfactorycorrelation has been found between the design data andthe actual performance in all cases. Fig. 3 is the circuitdiagram of a medium-voltage-regulated power supplydesigned to deliver up to 200 milliamperes in a voltagerange from 250 to 1350 volts.The complete range was designed to be covered in

Fig. 3

If gaseous rectifiers are used, the voltage drop acrossthese tubes can be neglected and, correcting for formfactor, the full-load secondary voltage of the platetransformer to center tap should be 1.11 X (direct out-put voltage of rectifier tubes). In order to compensatefor the regulation of the transformer, the no-load valueof the secondary voltage to the center tap should betaken as being about 5 per cent higher than the full-load value unless more definite information is avail-able.

If several ranges of voltage are required, a tappedautotransformer may be used to provide the correcttransformer secondary voltage for each range and thetaps of the autotransformer can be changed on themultigang selector switch used to change voltageranges. In order to catch unavoidable circuit changesthat may occur, it is usually desirable to check thedesign values of the voltage taps with the actual valuesrequired, before winding the autotransformer.

four uniform overlapping ranges as(values taken at no load):

shown in Table I

TABLE I

Design ActualRange REamin E.m.x Eomin EOmax

1 250 600 295 6032 500 850 503 8353 750 1100 750 10804 1000 1350 991 1320

It can be seen from Fig. 3 that the range switchingis obtained by means of a three-gang-four-positionselector switch and that a separate input voltage issupplied to the plate transformer from a tapped auto-transformer on each range. It was essential to changethe input voltage on each range in order to avoid ex-cessively large plate dissipations on the 2A3 tubeswhen the power supply was used on the lower ranges.The design and actual values of voltage of the auto-transformer taps are given in Table II. The design

Jiel)r,uary50

Bereskin: Voltage-Regulated Power Supplies

values for the voltage of the autotransformer taps wereperfectly satisfactory in all cases, but the values ac-tually used were increased slightly to compensate forline-voltage fluctuations. This increase was permissiblein all cases because the number of 2A3 tubes used wasconservatively chosen for a maximum plate dissipa-tion of only 18 watts on each tube, and the slight addi-tional plate dissipation could cause no trouble.

TABLE IIVOLTAGE OF AUTOTRANSFORMER TAPS

Range Preliminary ActualDesign Values Used

1 65 682 85 883 105 1084 125 128

The actual proof of the effectiveness of the regu-

lated power supply depends, of course, on how constantthe output voltage can be kept with variations in loadcurrent. The values for the maximum and minimumsettings of the third range, shown in Table III, are

typical of all the ranges and were considered to besatisfactory for all practical purposes. The output rip-ple voltage obtained in the above tests was in all cases

negligibly small.

TABLE III

Idc (ma)Eo (volts)

Idc (ma)Eo (volts)

Maximum Setting of Range 3

7 18 28 821080 1080 1080 1080

Minimum Setting of Range 3

5 13 22 59750- 750- 750- 749

1321079

131 202748 748

Eight 2A3 tubes were used in parallel for controlpurposes and, as mentioned before, the maximum platedissipation per tube was conservatively designed to beonly 18 watts. The resistors R12- Rig, connected in theplate circuits of the individual 2A3 tubes, are each a

100-ohm, 1-watt carbon resistor and are for the pur-

pose of eliminating any tendency towards parasiticoscillations that may exist in the circuit.The condenser C1, connected from the center tap of

the V1 - V2 filament transformer to the center tap ofthe plate transformer, is a 0.015-microfarad condenserand is intended to reduce the "hash" that is presentwhen rectifier tubes of the 866A type are used. Theeffect of this condenser is negligibly small as far as the"choke input" filter is concerned.

In all high-voltage applications, and especially wheregaseous rectifiers are used, it is desirable to bring thefilaments up to normal operating temperatures beforethe plate voltage is applied. It is also desirable to turnoff the plate voltage before the filament power is re-

moved. This point is taken care of by the steppedswitch S which closes the filament circuit first and theplate circuit next, reversing the process when theswitch is opened. A time-delay relay could be used withequal success for this purpose.

It can be seen from the circuit diagram that the posi-tive side of the line has been grounded. This is the nor-mal operating condition for many of the high-voltageapplications, but the negative side of the line couldhave been grounded instead if the operation had re-quired it.

In order to allow static charges that may build upbetween the alternating-current line and the regulatedpower supply to leak off, and to minimize slight linedisturbances, it is desirable to have a resistance-capacitance combination as indicated by C6, C7, R21,and R22.

CONSTRUCTIONAL CONSIDERATIONSIn the case of low-voltage low-current regulated-

power supplies, it is quite convenient to place thewhole rectifier and regulator on one chassis. Whenhigher voltage and current combinations are required,the use of a single chassis becomes inconvenient andthe use of a rack with two decks offers definite advan-tages. In this case the rectifier circuit and most of theheavy filtering equipment can be placed on the lowerdeck, while the comparatively lightweight regulatorequipment can be placed on the upper deck. Connec-tions between the two decks can be obtained by meansof a multiple-conductor cable and a plug and socket.

CONCLUSIONSIn summarizing, a sequence of design steps may be

pointed out although the order of the sequence neednot be followed rigorously. Referring to Fig. 1:

1. Choose maximum and minimum output voltagerequired and subdivide into several ranges, if desired,with a generous amount of overlap.

2. Considering the voltage ranges chosen in step 1,decide upon the type and number of control tubes V4required in parallel within their permissible plate dis-sipation. If the number of tubes required is unreason-ably large, a revision of the choice in step 1 may benecessary.

3. Upon the choice of suitable tubes V2 and V3 andpotentiometer Rp, calculations can be made to deter-mine the required values of R3 and R4 for each of theranges chosen in 1.

4. Choose the values of the resistors R5, R6, and R7with special attention to the possibility of using avoltage-regulator tube in place of R6 if Ei is variedappreciably for the various ranges.

5. Choose appropriate values of L and RB to insureoperation on the "flat" region of the rectifier outputcharacteristic for all ranges.

6. Choose appropriate values of C1, C2, and C4 de-pending on the maximum voltages expected, and ifthe ripple-voltage input to the regulator circuit is toohigh, an additional filter stage should be added.

7. Upon choosing appropriate tubes for V1 calculateI the secondary voltage required on the plate trans-

former for the particular range involved and, if

I - - - . .1 I --

1943 51

necessary, supply the primary of the plate transformer 10. Provide appropriate physical arrangement offrom a tapped autotransformer. parts on a chassis being careful to furnish adequate in-

8. If gaseous rectifiers are used, choose a suitable sulation where required.value for C3. The above procedure may be revised appropriately

9. Add any refinements to the circuit that the par- if a specific plate transformer is available and it isticular application may demand. desired to obtain optimum performance from it.

The Measurement of Transcription-TurntableSpeed Variation*H. E. ROYSt, ASSOCIATE, I.R.E.

Summary-Speed constancy or freedom from speed fluctuation("wows") is becoming more important due to the widespread use ofrecords in radio broadcasting. Equipment of a simplified nature whichwill evaluate the wow content as a single figure is needed for standard-ization purposes. Some of the existing equipment is reviewed, and theimportance of having a meter with proper ballistic constants for measur-ing the wow content is shown.

B ECAUSE of the increase in the use of transcrip-tion and home-phonograph records in the radiobroadcasting field, it is becoming more impor-

tant to maintain high standards of disk reproduction;one of the requirements of faithful reproduction isthat of maintaining a constant speed of the reproduc-ing turntable. It is not only necessary that the averagespeed as indicated by an ordinary speed-measuringdevice be the same as that of the recording turntablein order to maintain the correct pitch, but it is impor-tant that any instantaneous deviations from this aver-age speed be kept at a minimum, for it is these devia-tions, heard as "wows," that are particularly objection-able.As perceived by ear, the effect of these deviations

from average speed depends upon the percentagechange and the rate of variation. Published data' haveshown that pitch variations as low as 0.3 per cent aredetectable by ear in a 1000-cycle tone when the varia-tions occur at rates of 1 to 3 cycles per second. If thecycle of pitch variation is either faster or slower, theear becomes less sensitive to pitch changes. The abovetests were conducted with earphones and so independ-ent of the acoustic characteristics of the room. Al-though a dead room has little effect, a live roomincreases the apparent sensitivity to pitch changes,because of an amplitude pulsation created by shiftingof a standing-wave pattern.2 3 This amplitude pulsa-

* Decimal classification: 621.385.97. Original manuscript re-ceived by the Institute, June 26, 1942. Presented, Summer Con-vention, Cleveland, Ohio, June 29, 1942.

t RCA Manufacturing Company, Indianapolis, Indiana.1 E. G. Shower and R. Biddulph, "Differential pitch sensitivity

of the ear," Jour. Acous. Soc. Amer., vol. 3, pp. 275-287; October,1931.

2 T. E. Shea, W. A. MacNair, and V. Subrizi, "Flutter in soundrecords," Jour. Soc. Mot. Pic. Eng., vol. 25, pp. 403-415; Novem-ber, 1935.

3 W. J. Albersheim and D. MacKenzie, "Analysis of sound filmdrive," Jour. Soc. Mot. Pic. Eng., vol. 37, pp. 452 454; November,1941.

tion, however, is not considered too objectionable asit is often purposely introduced by the musician, andknown as tremolo, and so it is believed justifiable touse the earphone data in considering pitch variationsonly. At low rates of variation the change is perceiveddirectly as a pitch variation, and at rates such as oncearound of the turntable at 78 revolutions per minute,the variation resembles so much the spoken sounds"wow-wow," that the word "wow"' is generally used todenote such speed variations. As the rate of variationis increased it is heard as a flutter and, finally, all senseof pitch variation is lost; but if a constant-note recordis reproduced, sidebands not harmonically related willbe heard. The presence of this distortion can be ex-

AMPLIKIER

CON.AT NOTERfECORD PICKUP

7/I,9N4SBLf

GAL VANYOMETLR

V - A--

,rv__Eoocr,l 1,- FMEr&ETVAR/ArIaNFig. 1-Method of making wow measurements.

plained theoretically as resulting from frequency mod-ulation produced by the variation in speed."2'4

Realizing the importance of wow-free turntables forrecord reproduction, the National Association ofBroadcasters in its standardization program is under-taking the task of setting wow limits on disk recordingand reproducing equipment. Since equipment andmethods of measurement must be established beforelimits can be set and maintained, the writer, as a mem-ber of the standards subcommittee, has undertaken a

4 E. W. Kellogg and H. Belar, "Analysis of the distortion re-sulting from sprocket hole modulation.," Jour. Soc. Mot. Pic. Eng.,vol. 25, pp. 492-502; December, 1935.

Proceedings of the I.R.E.52 February, 1943