Amplifiers are really very simple. There is an input voltage and an output voltage, and at any instant of time these are both just numbers (with units of volts). If the input voltage is 1 Volt at a certain instant of time and the output voltage is then 20 Volts, then the amplifier has a gain of 20. In an ideal amplifier this number, 20, would be a complete description of the performance, and the output voltage would always be simply 20 times the input voltage. In a real amplifier with an input voltage varying with time the output voltage has a maximum level, and a maximum rate of change, and the gain falls at high frequencies, but within these limits it is still possible to approach the ideal behaviour.
There is a very simple way to find out how close to the ideal a given amplifier comes. If the gain is supposed to be 20, then all we need to do is divide the output voltage by 20 using an accurate resistive divider, and in an ideal amplifier the result will be identical to the input voltage. (Resistors add some distortion of their own, and metal film types should be used in the divider to keep this low). The most informative way to observe any differences is to subtract the input voltage from the attenuated output signal and display the result on an oscilloscope. For useful results careful balancing of amplitudes and phase shifts is needed so that only distortion remains. More details of this method can be found elsewhere (Design Note No.6, Distortion Measurement). It is posible to use this test method with music signals and loudspeaker loads, so the result is then certain to include all distortion mechanisms present in normal use.
This method is certainly not new, and an example appeared in Wireless World in 1953 (E.R.Wigan, "Diagnosis of Distortion" June 1953 pp261-266). The method was used by Peter Walker of Quad to demonstrate the performance of the Quad amplifiers. It was found that the distortion signal extracted from the Quad 303 was inaudible even when listened to on its own without the masking effect of the full signal. (Hi-Fi News and Record Review, Nov 1970 pp1655-1657). It is possible to increase the percentage distortion level, and when increased many times to the point at which an effect on sound quality was just audible, then it was found that the distortion listened to alone was clearly audible. It was said, with some justification, that the Quad amplifier is many times better than it needs to be, and if another amplifier sounds different then it is wrong.
Since then the number of 'wrong' amplifiers has multiplied. There are highly priced amplifiers being promoted which have performance so poor that the term hi-fi is no longer appropriate. Some people claim that some of these devices are in some way 'more musical', which may or may not be true. If an amplifier changes a signal, even in such a way that many listeners prefer the result to the unchanged original, then this has little to do with hi-fi, which by definition seeks to reproduce the original sound with 'high fidelity', i.e. with the greatest possible accuracy. The distortion extraction technique described gives a precise indication of the level of accuracy achieved. If the output voltage at a certain instant of time is supposed to be 10 volts and it is actually 10.001 volts, then the 0.001 volts error can be extracted and measured. If the error is only 0.0001 volts then the level of accuracy is obviously greater. Having extracted the distortion signal over a period of time, if it is found to be at an audible level then there may still be a genuine subjective question of how unpleasant different harmonics are or whether distortion associated with transients is worse than sine-wave distortion and so on. It is widely agreed that second harmonic distortion is the least unpleasant, and a certain level may even appear to improve sound quality. My own design aim is, however, to reduce distortion to below audibility rather than to get involved in disputes about what sort of distortion people prefer to listen to.
Most audio enthusiasts quite reasonably care more about their own subjective judgements of sound quality than any absolute standard of accuracy, and I have no argument with this. I am happy to just accept that accuracy and 'subjective quality' are two different things. Appearance, name and reputation sometimes appear to play a crucial role in subjective judgements, as demonstrated in a cable test reported in which big differences were described by listeners to what were believed to be known and visible different cables, when in fact the actual cable being used was never changed.
The Quad 303 amplifier mentioned earlier was certainly not perfect, and some aspects of performance can easily be improved on. The noise level was relatively high because of the use of a large input series resistor, and the maximum output current limited the ability to drive low impedance loads, so in fact it was only good for driving medium to high impedance speakers, and high sensitivity speakers, e.g. horn-loaded types, could make the noise audible. One of the problems of designing a general purpose amplifier is that speakers have an inconveniently wide range of characteristics. There exist speakers with highly reactive impedances falling to 2 ohms or less at some frequencies, and although it is easy to dismiss these as bad designs it seems a good idea to at least ensure no damage will be done by using them, and that even if the maximum output level is then restricted it is still reasonably undistorted up to that level.