|
Obtaining high quality reproduction with Class B.
Quiescent output systems are now widely used in battery-operated receivers in which economy in current consumption is of great importance. They are also often used with a mains drive in cases where very large output is required. If satisfactory results are to be secured with a Class B stage very careful design of the components is necessary, and in this article the requirements are discussed in some detail.
The economy of battery Class B is well known, but the extension of this principle to mains-operated amplifiers has received little publicity. The scheme results in a substantial saving in the initial cost of valves, mains equipment, etc., with a considerable reduction in running cost and overall weight of the apparatus, an important feature for PA work. Owing to the poor design of components in the early days, Class B became regarded as incapable of good quality reproduction. This, however, is not true, and the purpose of this article is to outline the important points of design with special reference to the transformer technique concerning which several very erroneous notions are widespread.
Fig. 1. - The fundamental circuit of a mains-driven Class B stage.
A skeleton circuit is illustrated in Fig. 1 Two similar power valves form the output stage, and these are biased back to the bottom bend of their characteristics. For small signals it is clear that these valves will function in QPP, an arrangement that has been fully described in past issues of The Wireless World However, in QPP the grids of the output pair are never allowed to become positive, whereas in the application described the grids are made to swing over such a wide range that they do become positive and therefore grid current flows. Hence, for small signals, the output can be regarded as QPP, but for large signals part of the cycle is handled as in QPP, while the peaks are handled as in Class B. A study of Fig. 2 will clarify the foregoing.
Fig. 2. - As shown in this diagram, grid current only flows when the signal input exceeds the grid bias, so that with small voltages the operation is more like QPP than Class B.
Valve Rating
Since the power valves are over-biased, very little current will flow in their anode circuits with no signal, it is therefore permissible for the peak currents to be larger than would be allowable in Class A so long as the average anode dissipation does not exceed the manufacturer's rating for the valves in question. A suitably chosen pair of output valves operated as described can be made to give a speech output several times greater than the output obtainable using the same valves in straight push-pull.
At this juncture it is interesting to note that variations of the circuit illustrated are at present widely used for Wireless Relay and PA, work, A good commercial example is the 'Sareco' amplifier illustrated in Fig. 3.
Fig. 3. - A large output Class B amplifier.
This amplifier gives a speech output of 90 Watts with 5% distortion, and the frequency response is within ±2 dB from 30 to 10,000 Hz.
The general theory about to be examined applies equally to the present circuit and to all forms of Class B whether mains or battery operated. Returning to Fig. 1, it will be noted that the driver valve has a varying impedance in its anode circuit. while no current is flowing in the grid circuits of the output pair (with small signals) the impedance across the secondary of the driver transformer will be very high. As soon as grid current flows (with large signals) the effective impedance will drop and will become a minimum when the signal voltage reaches a maximum.
Fig. 4. - The operation of the driver stage is illustrated here: at (a), only the load resistance RL appears, but (b) includes the transformer resistance RT. In (c) the valve is replaced by its internal resistance RV..
The stage can be simplified as shown in Figs. 4, a, b and c. Fig. 4 (a) shows the anode load as a variable resistance RL Fig. 4 (b) includes the total resistance of the transformer transferred to the primary RT Fig. 4. (c) completes the simplification by representing the AG resistance, of the valve as a pure resistance RV.
It is obviously the purpose of the driver stage to produce across AB (Fig. 4) an undistorted voltage wave form exactly similar to that applied across XY. If this is to be successfully achieved with a varying anode impedance, the regulation of the whole driver stage must be of a high order. In other words, the signal voltage lost across RT and RV must always be small compared with the signal voltage produced across RL even when RL is at its minimum value It at once follows that a satisfactory state will exist as long as the minimum value of RL is larger compared with the sum of RT and RV. Reverting to the practical circuit, this means that the minimum value of the effective impedance of the output grids multiplied by the square of the transformer ratio must be high compared with the AC resistance of the driver valve plus the total resistance of the driver transformer referred to the. primary.
The Driver Transformer
The required circuit conditions being established, the design of the driver transformer can be considered in greater detail. A step-down ratio, primary to secondary, will ensure a high impedance in the driver anode circuit, but a limit is soon reached in this direction, since the transformer secondary must provide sufficient voltage to load the output stage. It is apparent that the best ratio is that which will cause the output stage to be fully loaded just before the driver valve becomes fully loaded. This condition gives the greatest possible step-down. ratio and therefore the highest workable anode impedance. Having calculated this optimum ratio, it does not follow that it will prove satisfactory in every case. Great care has to be exercised in the selection of the driver valve or it will be found that the ratio, as determined above will still not provide a sufficiently high primary impedance. The remedy is to resort to the use of two driver valves in push-pull.
To obtain a good frequency response the primary inductance must be such that the reactance at the lowest important frequency is at least twice the load resistance of the driver valve. The leakage inductance must be minimised to avoid attenuation of the extreme upper frequencies and also to reduce the tendency which this inductance has to resonate with the input grid capacity of the output stage and so produce parasitic oscillation.
Good regulation is obtained by keeping the total copper losses of the driver transformer as low as possible. It can be shown theoretically that for a normal transformer this condition is attained when the primary copper losses are equal to the secondary copper losses. However, a driver transformer has virtually two secondaries that are used alternately, since, only, one-half of the total secondary carries current at any instant. Hence, to obtain maximum efficiency, the total secondary must occupy twice the space that would be required if the transformer had only one secondary.
The Transformer Windings
The practical result is that in a well designed driver transformer approximately one-third of the winding space is filled by the primary, while the total secondary occupies the remaining two-thirds. This has given rise to the idea that the value of the secondary resistance is more important than that of the primary, a supposition that is quite without foundation, it being the correct proportioning of the resistances that is important.
The final problem is to wind a low resistance primary that shall occupy only one third of the available winding space and yet possess a sufficiently high inductance. The polarising current from the plate of the driver valve complicates matters and necessitates an air gap in the core. The solution is to employ a much larger section of iron than is usually associated with an inter-valve transformer. Further, the writer has found a substantial advantage from the use of a comparatively new magnetic alloy known as 'Vicor', produced by Messrs Magnetic and Electrical Alloys, Ltd, of Wembley. This material combines low losses with a very high permeability that is less affected by a polarising field than the better known transformer materials. A smaller air-gap can be used with this alloy than would otherwise be necessary, thus not only is the inductance improved but the magnetic leakage is reduced, with beneficial results to the frequency response.
The Output Transformer
Before leaving the driver stage, it may be noted that Fig. 1 shows the bias voltage applied to the output valves as being derived from batteries. This has been done to simplify the main issue. In practice this voltage is obtained from the power supply, but the stringent conditions set by the circuit demand a technique that cannot be described in this article.
Having examined the driver transformer in some detail, the output transformer will present less difficulty. The essential feature of the output stage is that only one of the output valves is in operation at a time, since they alternately handle each half cycle (see Fig. 2). It follows that the necessary output ratio must be calculated with reference only to one-half of the primary, and also, since the two halt-primaries are really alternative whole-primaries, they will together occupy approximately twice the winding space taken by the secondary for a condition of maximum efficiency.
Since plate current flows alternately first in one valve and then in the other, the two valves do not provide equal and opposite DC in the two halves of transformer primary, as is the case with normal push-pull. Does this mean that the core will be liable to saturation and that it should be provided with an appropriate air gap?. It seems to be generally accepted that the core is polarised by the plate current, which is unfortunate since, truth to tell, the core of a Class B output transformer is no more polarised by the plate current than that of a normal push-pull transformer. This statement is directly opposed to the public writings of more than one technician and must, therefore, be supported by further reasoning.
For the primary of a transformer to lose inductance owing to the polarisation of the core it is necessary for the mean magnetisation throughout one complete cycle to be some value other than zero.
Fig. 5. - A single cycle of alternating current is represented by the familiar sine curve.
Consider Fig. 5, which represents one cycle of alternating current passing round the windings of a choke or transformer primary. The flux in the core will follow the same curve (approximately) and since there is no suggestion of DC nobody will hint at a state of polarisation. However, throughout the half-cycle marked AAA the current is unidirectional; further, at any single instant the winding is traversed by pure DC. Nevertheless, the core is not polarised because the average current (or flux) for the complete cycle is zero.
Fig. 6. - When DC passes through a transformer winding, the sine curve is displaced as shown.
Fig. 6 shows a similar curve for an inter-valve transformer carrying the full anode current of its associated valve. Here the core is polarised because the mean current (or flux) is not zero for the complete cycle.
Returning now to the Class B output transformer, it will be seen that one valve provides a half-cycle of current (or flux) in one direction while the other valve completes the cycle by providing an identical half-cycle in the reverse direction. Hence the average primary current (or flux) is zero for the complete cycle and the core is therefore not polarised.
Fig. 7. - The connections of a typical full-wave rectifier circuit, illustrating how the DC is balanced and leaves the core unaffected.
These conditions are almost exactly reproduced in the circuit of a full wave rectifier as illustrated in Fig. 7. The rectifier anodes pass current alternately, as do the output valves in Class B. Hence the mains transformer has a DC flowing in only one-half of its HT winding at any instant, exactly similar to the primary of the Class B output transformer. The electro-magnetic conditions existing in the two cases are strictly comparable. It has always been recognised as one of the advantages of full wave rectification that the mains transformer is not polarised. Why not so with Class B?
To conclude, a word about the HT supply to the output stage is essential. With no signal the HT current flowing is small, but it increases in proportion to the grid input voltage. If the HT supply be taken from the mains the usual rectifier and smoothing circuits have far too poor a regulation to work satisfactorily under these circumstances. A special technique has to be employed. The mercury vapour rectifier, in conjunction with a rather unusual choke, the inductance of which must be an inverse function of the DC passing through it, forms the basis of the HT supply circuit, but the design of such equipment is really a separate subject in itself.
| 