Distortion in power amplifiers, Part I: the sources of distortion

Very little reliable information appeared to exist on amplifier distortion, and I found myself embarked on a major effort to track down all the sources of distortion in the typical solid-state power amplifier. The initial investigations were not very illuminating, until I realised that changing a component value in the typical power amplifier circuit very often alters two or more distortion mechanisms simultaneously, making the results hard to interpret.

I was sitting in my armchair late one Spring night when the full force of this struck home, and thereafter I devised ways to simulate or measure the distortion mechanisms in isolation. This approach is well-illustrated in Parts 2 and 3 of the series, dealing with the input pair and the Voltage-Amplifier Stage, respectively.

Things then began to fall into place, and one day I put together all the various minor and, apparently insignificant, improvements I had made to the utterly conventional amplifier circuit I had started with. The distortion was exhilaratingly low, stability was good, and I soon felt that I could write a pretty comprehensive guide to distortion and power amplifier design. Electronics World were good enough to give me all the space I asked for, and I believe I succeeded; here, in eight chapters, is the result.

Introduction
It seems surprising that in a world which can build the Space Shuttle and detect the echoes of the birth of the universe, we still have to tolerate distortion in power amplifiers. Leafing through recent reviews and specifications shows claims for full-power total harmonic distortion ranging more than three orders of magnitude between individual designs, a wider range than any other parameter.

Admittedly the higher end of this range is represented by subjectivist equipment that displays dire linearity, presumably with the intention of implying that other nameless audio properties have been given priority over the mundane business of getting the signal from input to output without bending it.

Given the juggernaut rate of progress in most branches of electronics this seems to me anomalous, and especially notable in view of the many advanced analogue techniques used in op-amp design; after all power amps are only op-amps with boots on. One conclusion seems inescapable: a lot of power amplifiers generate much more distortion than they need to.

This series attempts to show exactly why amplifiers distort, and how to stop them doing it, culminating in a practical design for an ultra-linear amplifier. It should perhaps be said at the outset that none of this depends on excessively high levels of negative feedback. Many of the techniques described here are also entirely applicable to discrete op-amps, headphone drivers, and similar circuit blocks. Since we are almost in the twenty-first century I have ignored valve amplifiers.

Since mis-statements and confusions are endemic to audio, I have based these articles almost entirely on my own experimental work backed up with spice circuit simulation; much of the material relates specifically to bipolar transistor output stages, though a good deal is also relevant to mosfet amplifiers. Some of the statements made may seem controversial, but I believe they are all correct. If you think not, please tell me, but only if you have some real evidence to offer.

The fundamental reason why amplifier distortion persists is, of course, because it is a difficult technical problem to solve. A Science proverbially becomes an Art when there are more than seven variables, and since it will emerge that there are seven major distortion mechanisms to the average amplifier, we would seem to be nicely balanced on the boundary of the two cultures. Given so many significant sources of unwanted harmonics, overlaid and sometimes partially cancelling, sorting them out is a nontrivial task.

Make your amplifier as linear as possible before applying NFB has long been a cliche, (one that conveniently ignores the difficulty of running a high gain amp without any feedback) but virtually no dependable advice on how to perform this desirable linearisation has been published. The two factors are the basic linearity of the forward path, and the amount of negative feedback applied to further straighten it out.

The latter cannot be increased beyond certain limits or high-frequency stability is put in peril, whereas there seems no reason why open-loop linearity could not, in principle, be improved without limit, leading us to the Holy Grail of the distortionless amplifier. This series therefore takes as its prime aim the understanding and improvement of open-loop linearity. As it proceeds we will accrete circuit blocks to culminate in two practical amplifier designs that exploit the techniques presented here.