|
Resistance-coupled push-pull amplification is commonly employed in cases where the highest possible standard of reproduction is required. Some difficulty is often experienced, however, in obtaining a satisfactory method of phase reversal in the input circuits, and in this article the chief methods are discussed in detail and a new system is described.
In low-frequency amplification the push-pull system is widely recognised as giving a smaller degree of amplitude distortion than the more usual single-sided amplifier. It is consequently becoming more and more commonly employed in all cases where the attainment of the highest standard of reproduction is given serious attention. At one time its use was confined to transformer-coupled amplifiers, but it is now often used with resistance coupling. If the best results are to be secured particular attention must be paid to the input circuit. No difficulty arises when using transformer coupling, of course, for the usual push-pull transformer provides the necessary phase reversal in a straightforward manner. When resistance coupling is employed, however, matters are not quite so straightforward and there are many alternative arrangements possible.
Fig. 1. - The input of a typical push-pull amplifier is shown here.
The input circuit of a typical resistance-coupled push-pull amplifier is shown in Fig. 1, and in practice C1, C2 and R1, R2 are given identical capacities and resistances respectively. In feeding an amplifier of this nature the problem resolves itself, into obtaining an input voltage between A and E which is at any instant of the same value and of opposite sign to that which exists between B and E. In other words, if the point A be 1 Volt positive with respect to E, the point B must be 1 Volt negative also with respect to E. When neither terminal of the input need be earthed, this result can be secured very simply indeed. Suppose, for instance, that the amplifier of Fig. 1 is to be fed from an AC generator, which might well be a gramophone pick-up, the phase reversal can be secured by connecting two equal resistances in series across it and earthing their centre point. This is shown in Fig, 2 Where the resistances are denoted, by R1 and R2.
Fig. 2. - The simplest method of phase-reversal for the input consists of a centre-tapped resistance.
The operation of a circuit of this type is very easy to understand. Assuming for the moment that the arrangement of Fig. 2 is not connected to anything else, the generator drives a current through the two resistances. As these have the same value and as there is the same current through them, the voltage drop across each is the same and is one-half the generator voltage when the generator has no internal impedance. The voltage is less when the generator impedance is appreciable, but the voltage across R1 is still equal to that across R2. Now the question of phase arises, and is equally easy to settle. Suppose that at any instant A is 2 Volts positive with respect to B. Since the voltage drop across R1 equals that across R2 it is obvious that A must be 1 Volt more positive than E, and 1 Volt more positive than B. But if E is 1 Volt positive with respect to B, then B must be 1 Volt negative with respect to E; Hence A and B are at any instant at equal and opposite potentials with respect to E and the required phase reversal has been obtained. This holds also when another circuit is connected to that of Fig. 2, provided that the external impedance between AE has the same value and is of the same nature as that be-tween BE.
It follows that an amplifier of the type of Fig. 1, which is really the inputs of The Wireless World Push-Pull Quality Amplifier [★] First described May 11 and 18, 1934, and reprinted February 22, 1935., can be fed from a gramophone pick-up by connecting it to the terminals A and B, leaving E blank, for the grid leaks R1 and R2 provide the necessary centre-tap to the input. In general, however, this method is not very satisfactory, for a pick-up having a very large output is usually needed; the Quality Amplifier, for instance, requires an input of 7 Volts peak. When required, a volume control can be connected as shown in Fig. 3.
Fig. 3. - When a gramophone pickup is used, this circuit can feed an amplifier; the input is split by R1 and R2 of Fig. 1.
This method of phase reversal may still be, used when it is desired to couple a diode detector directly to an amplifier as shown in Fig, 4.
Fig, 4. - The arrangement of Fig. 2, as applied to a diode detector, is shown at (a), and at (b), the connections necessary when volume control is required.
When a volume control is not wanted, it is best to split the diode load resistance into two equal parts, R1 and R2; as shown at (a); no part of the circuit except the centre-tap must be earthed. If volume control be required, however the arrangement of (b) can be used, and here the grid leaks in the amplifier are relied upon for splitting the input to the amplifier.
Fig. 5. - One very satisfactory method of feeding an amplifier is obtained by placing equal resistances R1 and R2 in the anode and cathode circuits of a triode.
These precise arrangements, however, are not widely used, because some additional amplification is usually required for gramophone use and this is generally retained on radio also. One very satisfactory arrangement is shown in Fig. 5, and the method of phase reversal is basically the same as in the case of the circuits just discussed. The resistances R1 and R2 must have the same value, usually 20,000- 50,000 Ω and the decoupling capacitor C1 must be of large capacity, some 8 μF. Thorough decoupling is usually necessary, and it is wise, even when C1 has a capacity of 8 μF, to make R4 as high as 50,000 Ω. Grid bias is obtained by the voltage drop across R3 and a value of 1,000-2,000 Ω is normally suitable; the by-pass capacitor C2 should be large, at least 25 μF.
This system may be used on both radio and gramophone and a gain of about 10 times from the input terminals to each half of the output, or 20 times between the input and AB may be expected when suitable components are used. The greatest disadvantage of the arrangement is that neither terminal of, the input may be earthed. This does not greatly matter on gramophone, for there is rarely any difficulty in isolating each pick-up lead, but it is often inconvenient on radio. In the first place, it prohibits the use of some detector circuits, and secondly it greatly increases the difficulty of filtering the detector output properly.
Fig. 6. - The paraphase system embodies an extra valve giving a stage gain of unity.
There are many cases where it is essential tor one terminal of the input to be earthed, and it is then usual to employ the paraphase system. There are many variants of this, but there is actually little to choose between them and one of the simplest arrangements is that of Fig. 6. It will be seen that whereas the B output terminal is fed directly from the input, the A terminal is fed through the valve which is connected as a conventional resistance-coupled amplifier. Since there is a complete phase reversal in one stage of resistance amplification, it is clear that in this way the output terminals A and B are always at opposing potentials. In order that the magnitude of the potentials be the same, however, the gain of the stage must be unity. The valve grid, therefore, is fed with only a portion of the input from the potentiometer R1; in practice, R1 is adjustable and is set so that the amplifier as a whole is balanced and each output valve gives the same output. Incidentally, any of the other arrangements may be exactly balanced if desired by making the appropriate resistances adjustable.
The Paraphase System
Although the valve used in paraphase gives no effective amplification the phase-reversing stage as a whole gives a gain of 2 times. This is easily seen when it is remembered that the whole of the input appears between the B and E terminals, while the potential between the A and E terminals is derived through the valve. With the direct input split by a tapped resistance (Fig. 2) only one-half the input appears between each pair of output terminals.
A New Phase Changer
Fig. 7. - The basic connections of a new method. It is similar to the arrangement of Fig. 5, but one side of the input can be earthed.
A modification of the circuit of Fig. 5 has recently been put forward [★] Electronics, October, 1935. which removes the objection of requiring an unearthed input. The basic circuit without decoupling and biasing arrangements is shown in Fig. 7, and at first sight it would seem that the operation is the same as that of Fig. 5. It is, however, very different save in the method of phase reversal. The voltage which operates the valve is not the true input voltage e1 (Fig. 7.), but the voltage eg developed between grid and cathode, and this is less than e1 by the output voltage E, which appears across R2. In other words, there is severe anti-phase feed-back because R2 is common to both input and output circuits. This does not cause distortion as long as R2 is not by-passed by a capacitor, but any attempt to reduce the feed-back by shunting R2 by a capacitor is foredoomed to failure, for it will not only affect the frequency-response characteristic but will also upset the balance between the two halves of the output.
The true input voltage to the valve eg = e1 - E, and as E is the output to one side of the push-pull amplifier, eg must be less than E. The gain of the stage, measured between the input terminals and one pair of output terminals (AE or BE) must be less than unity, so that the stage attenuates instead of amplifying. Actually, the gain closely approaches unity and is given by E/e1 = μR2/Ra + 2R2 + μR2) where μ = valve amplification factor, Ra = valve AC resistance. The other symbols are as shown in Fig. 7, R1 being equal to R2. Using a typical valve having a resistance of 10,000 Ω with an amplification factor of 20 it can be seen that when both R1 and R2 are 25,000 Ω the gain E/e1 = 0.893. The total gain of e1 to the voltage between the AB output terminals is, of course, just double this, or 1.786 times, so that if the push-pull amplifier requires a total input of 7 Volts peak this feeder stage needs only 3.92 Volts peak or 2.78 Volts RMS. The gain is, in fact, nearly 90% of that given by the paraphase system.
The precise arrangement of Fig. 7. can rarely be used, because the grid cannot normally be returned to the earth line. The steady anode current passing through R2 makes the cathode considerably positive with respect to the earth line - if the anode current be 1 mA. and R2 be 25,000 Ω, the cathode is 25 Volts above earth. If the grid were returned to the earth line it would be negative with respect to cathode by the drop across R2 - with the figures just mentioned it would be negative by 25 Volts which is usually far too great a bias.
Practical Circuit Values
Fig. 8. - In practice, the new system is arranged in the manner shown in this diagram.
There are several ways in which the circuit can be modified to obtain the correct grid bias, and one of the simplest is shown in Fig. 8. It will be seen that a bias resistance of 1,000-2,000 Ω, R2, with a by-pass capacitor C2 of at least 25 μF, is introduced in the cathode circuit. The grid is returned to the negative side of this through the grid leak R1, so that the bias on the valve is merely, the voltage drop across R2. If these modifications are not to upset the operation of the circuit it is necessary for the reactance of C2 to be small compared with R2 at the lowest frequency, and R1 must be very large compared with R3. In the anode circuit the reactance of C3 at the lowest frequency, should be small in comparison with R4, but this depends to some extent upon the value of R5. In practice, satisfactory values for the components are C1 - 0.1 μF, C2 - 50 μF, C3 - 8 μF, R1 - 2 MΩ, R2 - 2,000 Ω, R3 = R4 - 25,000 Ω, R5 = 50,000 Ω. The valve selected does not greatly influence the results, and the MH4 and MHL4 classes are entirely suitable. The HT supply can be from 200 Volts to 300 Volts without necessitating any change in the values of the components.
Perhaps one of the greatest advantages of this system of feeding t a push-pull amplifier is the way in which a tone control can be devised. If both R3 and R4 be shunted by capacitors of suitable value - 0.002 μF to -0.01 μF - the higher audible frequencies are greatly attenuated, as one would expect. If only R3 be shunted, however, the total output of the push-pull amplifier is increased at high frequencies! What happens is this. When R3 and R4 are both shunted the amplification of the valve falls at high frequencies owing to the by-pass effect of the capacitors; the feed-back to the grid circuit is reduced, however, so that the effect of the by-pass capacitors is not as great as with an ordinary circuit. Now when only R3 is shunted the voltage developed by the signal between cathode and earth is reduced, at high frequencies, the feed-back is less, and a larger proportion of the input is effective in operating the valve. In other words, the effect of shunting the resistance R3 by a capacitor is to increase the voltage between grid and cathode at high, frequencies. Consequently a larger voltage is developed across R4. It can thus be seen that at, high-frequencies the voltage at the AE terminals rises, while that at the BE terminals falls. The rise at the one pair, however, is greater than the fall at the other, so that there is a gain in the total output.
A very useful accentuation of the upper register can be obtained in this way, but it is, of course, accompanied by a loss of balance in the-push-pull amplifier, for at high frequencies one-half of the chain does most of the work. This does not always matter, however, for two reasons - first, the signal amplitudes at high frequencies are usually small, and secondly, the harmonics introduced by any non-linearity are likely to be above audibility. Considering this point first, if the method be used to accentuate frequencies above 5,000 Hz only, there will be no lack of balance at lower frequencies, and all the advantages of push-pull will be obtained for such frequencies. Now if the lack of balance at higher frequencies does introduce harmonic distortion, it is unlikely to be present on any frequency lower than 7,000 Hz, for if the amplifier be balanced at 5,000 Hz the lack of balance between 5,000 Hz and 7,000 Hz will be quite small. The second harmonic of 7,000 Hz is - 14,000 Hz, which is getting near the upper limit of audibility. It is fairly certain that such a frequency could not be heard unless it were of considerable intensity, and in any case it is unlikely to be reproduced by any but the very best loud speaker. The other point which helps to make the lack of balance unlikely to result in distortion is that the amplitudes at frequencies over 5,000 Hz are normally quite small compared with those at low frequencies. The large amplitudes occur at low frequencies, and it is at these frequencies that the amplifier must be balanced if distortion is to be avoided.
It can thus be seen that this method of tone control, although at first sight in admissible, is actually quite permissible, and useful increase in the high-frequency response to compensate for sideband cutting for other deficiencies can be obtained very simply and without reducing the gain of the amplifier. This last is the most important point, for most other effective methods of tone-correction only give a relative increase in the high-frequency response, and it is actually obtained by reducing the amplification at low and medium frequencies. This is undoubtedly rather wasteful.
|
