https://www.temporalcoherence.nl/images/docs/AmplifierIssues.pdf
Why do amplifiers sound different?
By Hans Van Maanen
«Power amplifiers are an essential part in the sound reproduction chain. And although semiconductor amplifiers have been around for over 60 years, there is still a lot of development going on. Nowadays, the distortion figures of high-end power amplifiers are very impressive (e.g. < 0.001% harmonic distortion) and easily outshine those of microphones and loudspeakers. Yet, when it comes to listening, differences are noticed between amplifiers and their distortions can be heard in spite of the use of loudspeakers with much higher distortion figures. In this note I will discuss some aspects which play a role in this –at first sight incomprehensible- phenomenon, albeit that I will address a part of the puzzle, not all noticeable differences can be explained by the points I will bring up, partly
because I don’t know everything there is to know and partly because not all causes have yet been identified, I think. So please see this as a contribution to the discussion, not as the final word on it.
Therefore, I welcome contributions of others as “two know more than one”, as an age-old Dutch expression says.
One of the basic problems is that we try to “catch” distortion in a single number. But
one could pose the question whether this is feasible. To take a simple example: would the audible effect of say 1% harmonic distortion of only the second harmonic be just as noticeable or annoying as 0.1% of harmonic distortion of each of the second to the eleventh harmonic? Or be equivalent to 1% harmonic distortion of the tenth harmonic only? I don’t know the answer (because I never tried such a comparison as it is rather hard to do) but there is another example: valve (tube) amplifiers are often highly rated for their musical quality,
even though their distortion figures are horrible, compared to those of semiconductor amplifiers. Could there be a similarity between loudspeaker properties and valve amplifiers, distortion wise? Well, there is: both produce mostly lower harmonics (up to the fifth) with virtually no harmonics above that as is illustrated in fig. 1. Semiconductor amplifiers, however, tend to generate harmonics up to very high numbers as can be seen in fig. 2. In literature, there is agreement that our hearing tends to mask frequencies close(r) to the
exciting tone than those further away. Or, in other words, the lower harmonics are easily masked by the exciting tone whereas the high harmonics are not, as is shown in fig. 3. On top
of that most mechanical musical instruments generate only harmonics up to the fifth of the basic frequency, so distortion products introduce only a small change in the ratio of the harmonics, usually less than is caused by the linear distortion of loudspeakers. So the disSo it is not
really surprising that components which generate only lower harmonics are not so much experienced as annoying than components which generate more higher harmonics, even at a lower level. So the distortion figure of an amplifier is in itself of little use. A spectral specification would be more useful, but is rarely given.
...............................................
for the rest of the article go to the adress above