JOURNAL

OF

THE

BRITISH

INSTITUTION

OF

RADIO

ENGINEERS

UNBALANCE EFFECTS IN MODULATORS*

by

D. G. Tucker, D.Sc, Ph.D. (Member)]

SUMMARY The various groups of output frequency-components which can be produced in rectifier modulators of the shunt (or "Cowan") and ring types are separated into those which are inherent in the operation of the modulator and those which are absent when the modulator is perfectly balanced. It is shown that in the shunt modulator only one balance control potentiometer is needed, and only one unbalance component can, in general, be brought to a real minimum by a particular adjustment; nevertheless, all the unbalance components tend to vary (and balance) together. In the ring modulator, two independent balance controls can be provided, and two (though not any two) unbalance components can be simultaneously brought to a minimum. The effect of the signal voltage on the magnitude of unbalance output-components is examined, and shown to be zero in a modulator which is perfectly balanced in the absence of signal, and to be noticeable only near the overload point even when an initial unbalance exists. LIST OF SYMBOLS

fc fs m r n r v f( )

= carrier frequency. = signal frequency. may be zero or any positive integer. may be zero or any positive integer. may be any positive integer. = forward resistance of rectifiers as a function of voltage. E = signal e.m.f. = Eo cos 2irfst. Vc = carrier voltage across each rectifier. R = resistance of external circuit (see Figs. 1 and 2). VCL = carrier-leak voltage. Vai(t) carrier-leak voltage as a time-function. a = coefficient of square term in rectifier resistance/voltage law. j8 = coefficient of cube term in rectifier resistance/voltage law.
1. Introduction

vs v0

= signal voltage. displacement of rectifier characteristic along voltage axis, causing unbalance in modulator. \ x carrier current in Appendix 1, and equals carrier voltage in Appendix 2. = \ x signal current in Appendix 1, and equals \ X signal voltage in Appendix 2. = coefficient of nth power term in power-series expansion of rectifier voltage in terms of current. coefficient of nth power term in power-series expansion of rectifier current in terms of voltage. = "signature." (sign x = + 1 when x > 0 and 1 when x<0.)

a b cn

dn

sign

Balanced modulators are in common use in a variety of applications, and no doubt reliance is usually put on the partial suppression of certain output components as a result of the nominal balance of the circuit. However, the number of output components which are produced in an unbalanced modulator in addition to those
* Manuscript received October 8th, 1954. (Paper No. 308.) |H.M. Underwater Detection Establishment. U.D.C. No. 621.376.223.

produced in a perfectly balanced one is very large, and the conditions for attaining a high degree of balance in a practical modulator are usually rather complicated. It is therefore rather surprising that so little has been published on the subject of unbalance effects. References 1 to 5 cover the existing literature known to the author. There is a particular lack of discussion of any aspect of the subject other than that of "carrier-leak," i.e., the direct leakage of the local, or ""carrier" or "switching," oscillation. In practice, however, other unbalance effects 199

JOURNAL

OF

THE

BRITISH

INSTITUTION

OF

RADIO

ENGINEERS

are often important, e.g. even-order intermodulation and harmonic distortion, inputsignal leak, the worsening of the carrier-leak as the input signal voltage is raised, etc. The discussion below attempts to clarify the overall picture of the subject by showing how the various groups of unbalance output components are produced and how they can be balanced out by simple circuit adjustments in the well-known shunt (or "Cowan") and ring rectifier modulators. The influence of signal voltage on carrier-leak is included. The discussion is illustrated by three simple mathematical appendixes and by experimental results. 2. Balancing the Shunt Modulator The shunt modulator, shown in Fig. 1, is the simpler one to deal with, as only one balance control can be provided. The control potentiometer can be fitted anywhere in the bridge, but always affects the same group of output components. Ideally, i.e. with perfect-switching

is used with real rectifiers, with their gradual transition from high to low resistance and vice versa, even orders of modulation are introduced since the modulating function is no longer symmetrical about its mean value. 1 Thus output components of the type 2mfc (2n \)fs appear, generally at a fairly low level compared with the others. These components are not due to unbalance, and cannot therefore be eliminated by adjustment of resistances. If the four real rectifiers are not identical, other output components of the following types appear due to unbalance: (a) Carrier-leak and its harmonics, nfc. (b) Even-order distortion products of evenorder modulation, 2mfc 2nfs, which are produced in each individual rectifier even when these are perfect switches, but ideally cancel out at the output terminals. (c) Even-order distortion products of oddorder modulation, (2m -f \)fc 2nfs, which are not produced at all in perfect-switch rectifiers, and are probably always negligible except with large signal or low carrier voltages. (d) A d.c. component. All these components can be grouped into one main class:mfc 2rfs, where r, like m, may be zero or any positive integer. All members of this class can be reduced to a minimum by adjustment of the potentiometer RV. Generally, at low frequencies where capacitive unbalances are insignificant, any one component may be brought near to zero, but the optimum balance point is usually different for the different components. Appendix 1 gives a more detailed explanation of how all these various frequency components arise and how balancing controls them. Some experimental results on a shunt modulator, using four germanium rectifiers (not specially selected) at audio-frequencies, are shown in Fig. 2, which is a correlation diagram in which the output of 2/ c has been arbitrarily chosen as abscissa, and where the various points were obtained by varying the setting of RV. It will be seen that the products fc, 2/c and 2/ c 2/ s have balance points fairly close together and vary together. When the potentiometer was placed at the junction of rectifiers 1 and 2 instead of as in Fig. 1, rather worse results were obtained; the balance points agreed much less closely, and the lines in the

INPUT SIGNAL

CARRIER Fig. 1.Shunt (or "Cowan") modulator.

(or "linear") identical rectifiers and a squarewave carrier, and with the signal level below overload point, the output contains only the input signal (frequency/ s ) and the sidebands of odd orders of modulation, (2m -j- l ) / c / s , where m may be zero or any positive integer. Tf a sine-wave carrier is used (or any non-square waveform), but still with perfect-switching rectifiers, then non-linear distortion6 occurs due to the interference of the input signal with the switching of the rectifiers, which should be, ideally, entirely controlled by the carrier. These distortion products are of the type (2m + l)/c (In + \)fs, where n takes all integral values from unity upwards. When a sine-wave carrier
200

Aprtt 1955

D.

G. TUCKER

UNBALANCE

EFFECTS IN

MODULATORS

graph did not approximate to straight lines. In these experiments, the output of 2/c fs was 1-15 mV and did not vary by more than 0 - 5 db over the range of potentiometer settings involved in Fig. 2.

B O

40

1
fc/ /

o30
Q. t

rs

o 20

10

1 1

1 I

02?

1/
10

y
y

2rc-2fs i >

01

20 OUTPUT O F 2f c (mV)

30

3. Balancing the Ring Modulator The ring modulator, shown in Fig. 3, is much more complex than the shunt modulator, as the four positions of balancing potentiometers shown in the diagram all have different effects. With identical linear rectifiers and square-wave carrier, the only output components are (2m + l)/c / s ; and with identical linear rectifiers and sine-wave carrier, the nonlinear distortion products (2m+l)/e(2n+l)/s are added. Identical real rectifiers would not alter this situation, since the modulating function would be symmetrical about its mean level (of zero). All other components which can be introduced are due to unbalance. Thus, the unbalance output components comprise: (a) Carrier-leak, nfc. (b) Input signal leak, nfs (although amplitudes for n > 1 are usually negligible). (c) Sidebands of even-order modulation, 2mfcfs. (d) Even-order distortion products of evenorder modulation, 2mfc 2nfs. (e) Odd-order distortion products of evenorder modulation, 2mfc (2n + l ) / s . (/) Even-order distortion products of oddorder modulation, (2m + l ) / c 2nfs. (g) A d.c. component, although this is lost unless a variant of the ring modulator which has no output transformer3 is used.
RV4

Fig. 2. Shunt modulator. Effect of balancing potentiometer in position shown in Fig.l.

Experimental results: germanium rectifiers; R = 1300Q; f c. = 3 kc/s ; fs = 300 c/s ; Vc = 0-6 V approx. ; output of fcfs = 70 mV. INPUT SIGNAL CARRIER

it should be noted that a d.c. component is produced by unbalance, and the fact that the balance adjustment for this usually gives a fairly good (though not optimum) balance for the other components leads to a simple means of adjusting and maintaining carrier-leak by a d.c. meter.2 This is discussed in detail, together with an account of the phenomena associated with capacitive unbalance, in a forthcoming paper by Connon.7

Fig. 3.Ring modulator.

Of these, (e) and ( / ) are thought to be insignificant in practice, except with large signal or low carrier voltage; they do not occur with perfect-switch rectifiers. 201

JOURNAL

OF

THE

BRITISH

INSTITUTION

OF

RADIO

ENGINEERS

These products, (a) to (g), can be grouped in three main classes: (A) (2m + l ) / c 2rfs
2mfc2rfs (B) 2mfc (2r + l ) / s (C) where m, as before, takes all integral values from zero upwards, and r does the same. These three classes have different balancing properties. For components of class (A), the fundamental carrier-leak (m = 0, r = 0) may be taken as typical, and this is produced when the carrier current divides unequally into the two halves of the primary of the output transformer, in such a way that there is a resultant flux which is different in magnitude and/or polarity on the positive and negative carrier half-cycles.

11-0
> E "2
_ i

3 I>

>0S

--

1
/

7y '/ /

0-2>

2fc-fs
2f -2f / .4c

-0 - 1 ;

i A

^/ /

20

40 60 80 100 OUTPUT OF 2 f c ( m V )

is not equal and opposite on both positive and negative carrier half-cycles. This unbalance can therefore be corrected by adjustment of either RV2 or RV3. For components of class (C), the inputsignal leak (m = 0, r = 0) may be taken as typical, and this is produced when the signal current flowing through the output transformer is not equal and opposite on both positive and negative carrier half-cycles. This unbalance can therefore be corrected by adjustment of either RV2 or RV4. All the higher-order components included in classes (A) to (C) flow in the individual rectifiers, and the argument regarding balancing is the same as for the typical components chosen. A detailed illustration of the various effects discussed above is given in Appendix 2. Components of all classes can thus be simultaneously adjusted by the use of either RV1 and RV2, or RV3 and RV4.* It will be exceptional, of course, for the optimum settings of each potentiometer to be identical for all components controlled by it, since this would only occur if the system were non-reactive and the only difference between one rectifier and another were a fixed resistance which did not vary with the voltage across the rectifiers. But usually, in practice, one pair of settings gives a reasonable suppression of all unbalance output components, as is illustrated by measured results shown in Fig. 4, made on a typical ring modulator using germanium rectifiers chosen at random from a large batch. It will be seen that four important components controlled by RV2 give optimum balances fairly close together, and tend to vary together. The various points on the graph were obtained by adjusting RV2. In practice it will usually prove convenient to fit a fixed resistor in the connection to the bottom side of the primary of the output transformer in order to compensate for the resistance of RV2 in the other sideif this is not done, RV1 will need to be larger than otherwise necessary in order to cope with the longitudinal unbalance introduced. A similar argument applies to RV4 and RV3.
*It should be noted that adjustment of RV1, by slightly redistributing the voltages across the rectifiers, does slightly change their resistances and so has a second-order effect on the adjustment of RV2. Similarly RV2 reacts slightly on RV1, and RV3 on RV4 and vice versa. These interactions show up in practice only when a high degree of balance is sought.

Fig. 4.Ring modulator. Effect of balancing potentiometer RV2.

Experimental results: germanium rectifiers; R = 5 k f 2 ; f c 3 kc/s ; fs = 300 c/s; E/Vc = 0-3 approx; Vc = 0-5 V peak; output of fc f = 47 mV.

This unbalance can therefore be corrected by adjustment of either RV1 or RV4. For components of class (B), the second harmonic of carrier-leak (m = 1, r = 0) may be taken as typical, and this is produced when the carrier current divides in the output transformer in such a way that there is a resultant flux which
202

April 1955

D.

G. TUCKER

UNBALANCE

EFFECTS

IN

MODULATORS

4. The Influence of the Signal Voltage on Unbalance There appears to have been no published discussion of this matter, although it is probably well known that carrier-leak, which is usually

adjusted in the absence of signal, can be altered by the application of the input signal. If a modulator is perfectly balanced (i.e. has all rectifiers identical) in the absence of a signal, then the application of a signal of perfectly symmetrical waveform (i.e. positive and negative excursions identical) cannot cause unbalance products to be formed whatever the shape of the rectifier characteristics. This conclusion, although perhaps an intuitive one, is probably readily acceptable as "obvious." However, proof for two very different cases is available. Firstly, the ideal ring modulator with perfect-switching rectifiers but with a sinusoidal carrier voltage, discussed in Reference 6, has a total output given by equation (11) of the paper cited; this equation includes the effect of signal voltages on the switching of the rectifiers, but can easily be shown to contain no components at the carrier frequency or its harmonics, nor at the other frequencies which appear due to unbalance. Secondly, Appendix 3 shows that a simple shunt modulator with rectifiers very different from perfect switches has no carrier-leak if the rectifiers are identical, irrespective of the signal voltage. No general treatment of the effect of signal voltages on the unbalance of a modulator which is already unbalanced in the absence of the signal can be attempted here, but Appendix 3 shows that in a simple case which is easily analysed the effect definitely exists but is small, only becoming noticeable as the overload point is approached. Theory such as this cannot be expected to be adequate for prediction of practical carrierleak magnitudes, and serves primarily to elucidate the principles. Nevertheless, it is clear that the effect is very dependent on the actual shape of the rectifier characteristics and on the nature of the initial unbalance, and will, in practice, be very hard to predict. Figure 5 shows the results of numerous measurements of the effect of signal voltage on the fundamental component of carrier-leak in shunt modulators using germanium and copperoxide rectifiers, with square-wave and with sine-wave carriers, and, in the case of copperoxide rectifiers, with a higher carrier frequency so that reactive effects might be included with a low backward resistance. Measurements on a ring modulator are also included. In all cases the effect is of the nature predicted above, and the order of magnitude does not conflict too
203

10

-40

-30

-20

-10 0 E/Vc IN DECIBELS

+10

Fig. 5.Effect of Signal Level.

Experimental results o o = shunt, germanium, R = 2-5kf2, - v square carrier x x = shunt, germanium, R = 2-5kf2, 3 k c / s sine carrier ffs = 200 c/s shunt, copper oxide,R = 2-Skil, =sine carrier ) _ A - - A - = shunt, copper oxide, R = 2-5kfi, sine carrier, fc = 15 kc/s - - e - = ring, germanium, R = 5kD, fc = 3 kc/s sine carrier,

[fc =

JOURNAL

OF

THE BRITISH

INSTITUTION

OF R A D I O

ENGINEERS

seriously with that suggested by equations (6) and (7) of Appendix 3. Equation (7) can account only for an increase in leak, but since increases and decreases occur in practice, it is clear that equation (6) may be more realistic; in this equation the effect of the signal depends, both in magnitude and sign, on the relative magnitudes and sign of the coefficients a and b of the rectifier resistance/voltage law. One particularly interesting feature in Fig. 5 is the result using the shunt modulator with copper-oxide rectifiers and fc = 3 kc/s, the initial leak being 8 mV. Here the balancing potentiometer was adjusted first to one side of optimum balance, and then to the other side, the leak being set to 8 mV in each case. The effect of the signal voltage was opposite, however, as one side caused a decrease due to signal and the other side caused an increase. Although this type of result was often obtained, it was by no means the rule. 5. Conclusions The effects of unbalance in rectifier modulators of the shunt (or "Cowan") and ring types have been discussed in relation to all output components which can be produced, and it has been shown that the use of balancing potentiometers can lead to substantial suppression of the following types of product: Shunt Modulator: mfc 2rfs Ring Modulator: (2m + l)/ c 2rfs 2mfc 2rfs 2mfc (2r + l)/s where m and r take any positive integral value or may be zero. In the shunt modulator only one balance control is possible, and thus, in general, only one of the products can be brought to a real minimum by a particular adjustment; nevertheless, all the unbalance products tend to vary (and balance) together. In the ring, modulator, two independent balance controls can be provided, and one product from each of two of the three classes shown above can be simultaneously balanced to a true minimum, and'a fair overall balance of all three classes of product can usually be achieved. The types of output product which cannot be balanced out at all are: Shunt Modulator: mfc (2r + l)/ s Ring Modulator: (2m + l)/ c (2r + l)/ s 204

The effect of the signal voltage on the magnitude of the carrier-leak (i.e., m = 1, r 0 in the class mfc 2rfs or m = 0, r = 0 in the class (2m + l)/ c 2rfs) has been examined, and shown to be zero in a modulator which is perfectly balanced in the absence of signal, and to be noticeable only near the overload point even when an initial unbalance exists. 6. Acknowledgments This paper is published by permission of the Admiralty. 7. References 1. D. G. Tucker. "Some aspects of the design of balanced rectifier modulators for precision applications." / . Instn Elect. Engrs, 95, Part III, p. 161, 1948. 2. D. G. Tucker. "Carrier-leak control in rectifier modulators by the use of a d.c. meter." Proc. Instn Elect. Engrs, 99, Part III, p. 402, 1952. 3. D. G. Tucker. "Modulators and Frequency Changers." (Macdonald, London, 1953.) 4. Y. Degawa. "On the distortion outputs of ring modulators." Nippon Electrical Communication Engineering, July 1940, p. 16. 5. O. Henkler. "Anwendung der Modulation auf beim Tragerfrequenzfernsprechen Leitungen." (Hirzel, Leipzig, 1948.) 6. D. G. Tucker. "Intermodulation distortion in rectifier modulators." Wireless Engineer, 31, p. 145, 1954. 7. G. L. Connon, "Carrier leak in shunt modulators." Proc. Instn Elect. Engrs, 102, Part B, 1955. To be published. 8. Appendix 1: Types of Output Product in the Shunt Modulator It may not be immediately clear why the output products of the shunt modulator with real rectifiers and non-square carrier waveform fall into the two classes: tnfc 2rfs which can be balanced out, and mfc (2r + l)/s which cannot be balanced out. A brief discussion of a simple case should readily clarify this matter. Consider a non-reactive shunt modulator which has "constant-current" signal and carrier

April 1955

D.

G. TUCKER

UNBALANCE

EFFECTS IN

MODULATORS

sources, and takes its output as the voltage developed across the rectifier bridge. The voltage across any individual rectifier can be related to the current through it by a power series of the type

v=2 c ^ M
/I = 0

00

giving rise to the frequencies 0, fc, fc 2/ s , fc 4/ s , 2fc, 2fe 2fs, 3/c, 3/ e 2/s, 4/ f , 5fc, 2/ s , 4/s, i.e. the group mfc 2rfs. These can be balanced out. The condition for simultaneous balancing-out of all such products by means of a resistance adjustment is evidently Cni Cn2 = constant for all values of n, where it is assumed that cn\ applies to rectifiers 1 and 4 and cn2 to rectifiers 2 and 3. If all four rectifiers are different the condition becomes
Cn2

(!)

and it is assumed that the rectifier resistance does not become infinite since the "constantcurrent" drive must always be maintained. Assume for simplicity at this stage that coefficients Cni apply to rectifiers 1 and 4, and Cn2 to rectifiers 2 and 3. This justifies the assumption of equal division of current between the two branches of the carrier circuit and between the two branches of the signal circuit. Then if Ic cos 2Trfct 2a is the carrier current and Is cos 2rrfst = 2b is the signal current, then the current in rectifiers 1 and 4 is (a b) and that in rectifiers 2 and 3 is (a -\- b), taking the direction of best conduction of the rectifiers as the positive flow. The output voltage of the modulator is therefore

These conditions practice.

= constant. + are unlikely to be met

in

by-Jcmia-by

....(2) ]

= i 2 ( * + + 2 (CM2 - K

9. Appendix 2: Types of Output Product in the Ring Modulator Following the lines of Appendix 1, the working out of a theoretically simple case of the ring modulator for real rectifiers and non-square carrier waveform should help .the understanding of the general statements of Section 3. Consider a non-reactive ring modulator which has "constant-voltage" (or zero-resistance) signal and carrier sources, and takes its output as the current through a terminating resistance of zero. If Vc cos 2-nfd = a is the carrier
voltage and Vs cos 2TT/S/ = 2b is the signal

(3) When the modulator is perfectly balanced, cn% Cnx is zero for all values of n, and only the first term in (3) remains. Thus the output products which cannot be balanced out are those contained in while the products which can be eliminated by balancing are those contained in

voltage, then the voltages across the four rectifiers are (a + b), (a b), (a b) and {a + b) for rectifiers 1 to 4 respectively. As in Appendix 1, polarity is taken with respect to the direction of best" conduction of the rectifiers. Let the current through a rectifier be related to the voltage across it by the power series

2(a+br+%(a-br

(5)

/ =2

0)

The expansion of (4) and (5) for n = 4 and 5 will sufficiently demonstrate the effect. Thus (a + by -(a- bY = 836 + 8ab* (a + b)5 -(a- b)5 = lOa^b + 20a2b3 + 2b5 giving rise to the frequencies fs, fc fs, 2fcfs, 3fcfs, 4fcf,, 3fs,fc3fs, 2fc 3/s, 5fs, i.e. the group mfc (2r + l)/ s . These cannot be balanced out. Also (a + by + (a - by = 2a* + I2a2b2 + 2b* la + b)5 + (a - b)5 = 2fl5 + 20a3b2 + lOab*

In general the four rectifiers have different sets of coefficients, dni, dn%, dn3 and dn, and in these circumstances the assumption made above that half the signal voltage appears across each rectifier is not strictly justified. But it greatly simplifies the calculation and leads to an error only in the amplitude of unbalance components; it does not affect the nature of the result nor the conditions for balance, discussed .later. The assumption is completely justified if balancing condition (9a)see belowapplies, whether the other conditions (7) and (11) are met or not.
205

JOURNAL

OF

THE

B R I T I S H I N S T I T U T I O N OF RADIO E N G I N E E R S
by means of potentiometer RV2. The output in class (C) is (2) ^(dnx - dni - dn2 + dns) [(a + b)n - (a - by], n odd (10) and this can be balanced out by making
dnX dnl dn2 dnZ dnX dni = dnl dn3
(1 la) (1 l b )

The output current is ix i2 i3 + i

= 2 dni(a + by - J dn*( - l) (a - b)

- 2 [( - l)n<fti2 + dnz](a -b)n

(3)

If all dn are equal, clearly all terms in which n is odd cancel out, and the even terms leave only

by means of potentiometer RV4, or by making by means of potentiometer RV2. 10. Appendix 3: The Influence of Signal Voltage on UnbalanceAn Example Consider a shunt (or "Cowan") modulator, as shown in Fig. 1, which has a square-wave carrier voltage of amplitude 2VC with zero source impedance, and a circuit resistance R which is very large compared with the forward resistance, /y(v), of the rectifiers, which is considered as a function of voltage. The signal e.m.f. is E = Eo COS 2vfst. Consider only the forward half-cycle of carrier; we shall assume that there is no conduction on the backward half-cycle. The signal voltage developed across each rectifier is (1) vs=Eff(v)l4R and we assume E<VC so that switching is not affected by the signal. Rectifiers 1 and 4 have resistance r/(Vc vs) and 2 and 3 have resistance r/(Vc + vs), if all four rectifiers are identical in the absence of signal. But, for generality, assume that the rectifiers are initially unbalanced by having their characteristics displaced in opposite directions along the voltage axis, so that 1 and 4 have resistance rj(Vc v0 vs) and 2 and 3 y (Kc + v0 -j- vs). have A It is easily shown that the carrier-leak voltage is T , rr/(F c - v0 - v.) - r/(Vc + v0 + v,)! VcL~Vc\_ J rf{yc) (2) on the assumption of a relatively small unbalance. Now the resistance/voltage characteristic of // can be represented approximately and arbitrarily by a law which gives a falling curve thus >7(v) = r0 + a(2Kc - v)2 + fS(2Vc - v)3 . .(3)

2 2 dnia + b)n - 2 2 dn(a - b)n, n e v e n . . . . (4)

which is therefore the output which cannot be balanced out. By expanding a few terms as in Appendix 1, it is quickly seen that this contains only frequencies of the type (2m + l)/ c (2r + l)/ s . The remaining terms when the dn are unequal are those which can be balanced out, and can similarly be seen to fall in the three classes

(2m + \)fc2rfs (A) 2mfc2rfs (B) and 2mfc (2r + l)/ s (C) The output in these classes can be obtained by writing (3) .as \ 2 [dm + ( - 1)V M4 + ( - Xydni +

+ i 2 idnl + ( - M ) * - (- )
- dM] [(a + 6) + (a - by]

+ dm] [(a + b)n -(a1 w 14 1 Vn2

b)n]

(5)

From this it can be seen that the output in class (A) is d dn dn d a bn

\ 2 ( >* ~ *-+ * ~ *> [( + ) +

+ (a - by], n odd (6)

and this can be balanced out by making (7a) dn\ + dn% = dn3 + dni by means of potentiometer RV1, or by making dnX dnl = dnZ dni (7b) by means of potentiometer RV4. The output in class (B) is \ 2 (dnx + dvA - dm - dni) [(a + b)n + + (a - b)n], n even (8) and this can be balanced out by making dnx + dm = dnz + ^2 (9a) by means of potentiometer RV3, or by making dnX dni = dnZ dni (9b) 206

April 1955

D.

G. TUCKER

UNBALANCE EFFECTS IN MODULATORS If the initial unbalance is due to rectifiers 1 and 4 having a coefficient fix in equation (3) and 2 and 3 having a different coefficient j82, instead of being due to differences in contact potential, then the carrier-leak is easily shown to be \ + t sign (cos 2*fet)] X .(7) which corresponds very closely to equation (6). From this work it is clear that (a) the leak is zero, irrespective of the value of Eo, if v0 = 0 and fix = /32, i.e. if the initial balance is perfect. (b) the effect of the signal voltage is in any case quite smallarising only from the cube term in (3) since Eo < Vc and rf{Vc)<R. It appears, too, that these conclusions are unaffected if the carrier is sinusoidal instead of square-wave.

On substituting in (2) we obtain [4aKc(v0 + vs) + (vs 3v 0 2 v s )

}]

(4) This is the value of the leak at any instant during the forward half-cycle of carrier. To obtain the full time-function of the leak we must introduce the switching function, i.e., if fc is the fundamental frequency of the carrier, Vaiit) = VCL[\ + i Sign (COS2TT/C0] . . . -(5) and it is immediately clear that since vs is alternating, the only contribution to leak at the carrier frequency and its harmonics comes from the "d.c." component of (4), i.e., on putting E EQ coslnfst and using equation (1), the carrier-leak is M O = [Vcjrf{Vc)\ [ i + i sign (cos 2nfet)] X
4ocF c v o

3VC\) .(6)

GRADUATESHIP EXAMINATIONNOVEMBER 1954

FINAL PASS LIST
This list contains the results of the remaining oversea candidates not included in the lists published in the * February and March issues of the Journal. A total of 553 candidates entered for the Examination The following candidates have completed the requirements of the Graduateship Examination and are eligible for transfer or election to graduateship or higher grade of membership
DUTTA, Asim Kumar. (S) Barrackpore. GOUDAS, Nicolaos. (S) Alexandropolis. HAIRETAKIS, Emmanuel. (S) Piraeus. MATHIOUDIS, Miltiades. (S) Athens. SERVETAS, Evangelos. (S) Aegaleo. VASSILIOU, Athena. (S) Athens.

The following candidates passed the parts indicated against their names
BALASUBRAMANIAN, Venkatram. (S) Delhi. (II). NEGREPONTIS, Eleutherios. (S) Athens. (II, Ilia). CARLIS, Georges. (S) Athens. (Ilia). CHATTERJEE, Bhabatosh. Calcutta. (I). CHOPRA, Jamak Kumar. (S) Delhi. (Illb). CHRISTODOULOU, Christos. (S) Athens. (II). DIMAS, Dimitrios. (S) Athens. (II, Ilia). (II, Ilia). (Illb).

PANDAZIS, Georgios. (S) Athens. (II). PAPADOPOULOS, Emmanuel. (S) Athens. (II, Ilia). PITSINIGOS, Savas. (S) Athens. (II). POUTIS, Athanasius. (S) Athens. (II). POLYMENEAS, George. <S) Piraeus. (Ilia). ROMANIDIS, Andrew. (S) Athens. (Ilia, Illb).

FATSIS, Nicholas. (S) Piraeus. GEORGIOU, Gregory. (S) HARCHARAN SINGH. (S) Mhow.

Athens. (II). Delhi. (II).

SODHI, Pashori Lai. (S) Simla. (II). SPYROPOCJLOS, Nicholas. (S) Athens. (II). ZAFIROPOULOS, Peter. (S) (S) denotes a Registered Student Athens. (Illb).

KANWAR, Randhir Singh. (S)

207