Class D or AB power amplifiers?

Updated: 10-12-2017

For long the most obvious amplification method was to increase the amplitude and current capability of the analog source music signal directly, without conversion. With time, and the possibility to integrate a plurality of transistors on a single chip, very well performing class AB amplifier circuits were invented to satisfy even demanding audiophile use. The main problem with class AB amplifiers were the inherent power efficiency of maximum 70%. For stationary use the costs of the energy lost in the amplifier is not significant and the heating can be handled with heat-sinks. But, for portable use any saving on the battery is important. Also, if little heat loss would mean that heat-sinks became obsolete and the power supply could be reduced in capacity, money could be saved in commercial production. The development of the class D amplifier had its reasons.

Let it for a start be admitted that even very cheap class D power amplifier boards perform amazingly well, even for audiophile use. They are surprisingly immune to external noise and can be implemented to provide THD levels as good as class AB amplifiers. That class D amplifier chips can be used without heat-sink is in many cases not true and often their implementation with a minimum of energy-storage in the form of supply-line decoupling capacitors seems under-dimensioned. Class D amplifiers typically need a power output filter that also often seems to be implemented without proper rating. For commercial purposes the class D amplifiers are an important gain that allow the amplifiers to be fully enclosed in for instance sound bars. Technically, the class D amplifiers are much more complex than class AB amplifiers and required sophisticated circuit integration techniques before becoming attractive. Today that problem is solved.

A class D amplifier for analog input signals includes three stages: an analog-to-PWM converter stage, a push-pull power stage and a power output filter that converts the PWM signal back to an analog signal.
The analog-to-PWM converter may be implemented with a fast, high-resolution A/D converter that samples the analog signal according to the Nyquist criterion and where the repeated sampling values are converted into PWM cycles. But, such a fast, high-resolution A/D converter is not cheap and the further logic may also be costly. Further, doing better than 0.01% distortion is not evident, So, it is probably better to let experienced manufacturers, having invested much time in finding the best solutions, supply that functionality on a chip.
It sounds easy to design a push-pull power stage for a PWM signal but it actually is not. If there is a dead-time between the switches it increases the distortion and catch-diodes or a correctly dimensioned snubber circuit may be needed for protection. If there is an overlap in the conduction of the high- and low-side switches the result is current-spikes from “ shoot-through” that may cause distortion but at least increase power losses and stress the switches. Complex circuits are known that detect if one switch is turned OFF before turning the other ON, but the switches and control circuit is better implemented on a single chip because the temperature dependence is important. So, also for the push-pull power stage it is often better to let experienced manufacturers supply that functionality on a chip.
The power output filter may be adapted by the person building the amplifier. But again, the chip manufacturers have invested quite some time in deciding on very suited filter designs such that they can sell their chips. Therefore, in most cases it is better to follow their advise and ensure that the quality of the filter components are right.
All in all, you are quite bound in your design when deciding for a class D amplifier.

For audiophile purposes using loudspeakers, the situation is a little different than for portable use. The speakers used are typically stationary and the power efficiency not essential.

High performance integrated circuits exist that allow implementation of an audiophile level class AB amplifier with a minimum of external components. Take the TDA7294 /TDA7293 as examples. These integrated circuits are cheap, suited for DIY purposes and widely available for small quantity purchase. With such integrated circuits and few ordinary, external components an amplifier can quickly be made that would have been state-of-the-art two decades back. An audiophile class AB amplifier may even be build with discrete components only such that the designer has a maximum of choice. Or, a class AB amplifier may be based on a high-quality OP-AMP in the input stage. The heating issue remains for class AB amplifiers but can be solved with heat-sinks. The price of the energy lost as heat is very moderate in a household and will not upset an audiophile. So, it is not obvious that class AB amplifiers are outdated for audiophile purposes.

An advantage of class AB amplifiers, compared to class D amplifiers, is that no need for a power output filter exists. Class D amplifiers may cause problems with EMI (ElectroMagnetic Interference), the class AB amplifiers not. Class AB amplifiers with a complementary push-pull stage at the output offer, through the emitters, intrinsic low impedance for clamping of inductive loudspeaker impedances. For the class D amplifiers this is less obvious. The class AB amplifiers still offer some important features.

Therefore, a comparison of sound quality between class D and class AB amplifiers is relevant for audiophile purposes. Commercially the class D technology may have taken the lead, but audiophiles may still build their own class AB amplifiers if the sound quality is better.