This month I thought I might investigate the different types of output stage employed in power amplifiers. This subject is one of some contention among informed audiophiles. Some hold that only the expensive and hot running class 'A' provides audio perfection and others, probably the majority, are quite happy with the ubiquitous class 'B' stage.
If you're not up in electronic theory you may wonder what all the fuss is about. To understand the problems involved we have to go back to basics. Figure 1 shows a single transistor with the emitter (e) connected to ground. The circuit to the positive rail is completed by RL. Now any amplifier operates by using a small signal, the input, to modulate an external power source. The power source is the power supply to the equipment.
It follows then that the laws of physics are not violated, you still don't get 'owt for nowt'. Looking at Figure 1 you will notice that the base is connected across a potentiometer so that adjusting it will alter the voltage at the base. Starting with the base at the negative supply rail, a multimeter will show no voltage drop across RL. As the pot is adjusted to positively bias the transistor a point will be found where the voltage across RL suddenly increases. If the pot is carefully adjusted until half the supply voltage appears across the load the base voltage will be found to be about 0.65V.
We now have a class 'A' amplifier. This may be defined as any amplifier in which current flows through the load at all points in the output signal's cycle. To illustrate this we need an oscilloscope and a signal generator. The signal is fed into the base via C1 and is thus superimposed upon the base voltage. In consequence when the signal goes positive the transistor is turned harder on so the voltage drop across RL increases. When the signal goes negative the converse occurs. If the input signal is too large the output waveform, as observed on a scope, will be clipped. The resulting sound, if R1 were a speaker is the familiar fuzz sound beloved of guitarists.
The point though is that if the transistor is initially biased in class 'A' the output stage will reproduce an essentially perfect replica of the input at its output.
If we now carefully adjust the bias until the transistor is just conducting and observe the output we will get a halfwave rectified signal. The transistor is said to be operating in class 'B'.
If we now take a PNP and NPN transistor and arrange them as shown in Figure 2 we can make an amplifier output that theoretically requires no bias at all but can reproduce the input signal Both transistor bases receive the input signal simultaneously, R1 and R2 set the DC voltage at the output at half supply volts. RV1 sets the bias. The illustration shows the result of zero bias. Note the distortion at the zero crossing points of the waveform. This distortion, not unnaturally, is known as crossover distortion.
Crossover distortion is the main problem with this type of stage and is usually countered by applying a small bias voltage via RV1. The signal now looks okay on a scope. What we have just examined is the conventional amplifier output stage used in 99% of commercial amps. It's easy to see why; quiescent current need only be a few tens of mA to reduce the crossover. Further distortion reduction is obtained by the overall feed back loop.
The matter might well rest there if it were not for the problems inherent in transistor design. In particular thermal runaway. As a transistor heats up the amount of bias it requires for a given current to flow reduces. Power transistors particularly tend to get hot because of the power they dissipate. As they get hotter the more current will flow making them hotter still, so even more current flows. If this effect is unchecked the final result is the destruction of the devices.
The cure is to mount the transistors on a hefty heatsink and to make the bias voltage temperature sensitive. This usually means diodes or a transistor bolted onto the heatsink for thermal feedback.
As speakers become less efficient and amplifier power rises to cope the problems become more pressing. This leads us on to another form of output stage, the so called 'current dumper'.
Figure 3 shows this. Note that a high gain amp is used to feed the output pair which are used without bias. Notice the use of R1 which feeds from the amp to the output. The amp itself is designed to deliver an output current of 100mA. When an input signal is fed into the input, current flows into the load via R1. As this happens TR1 or TR2 is biased on providing the majority of the current to the load. However, in the idling mode neither TR1 or TR2 are conducting. They cannot, therefore, suffer from thermal runaway. Distortion is reduced by overall feedback. This technique is known as current dumping. The most well known application of this is the Quad 405 amplifier.
Feature by Jeff Macaulay
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!