Hi John,
I'd forgotten about that post; I'd hoped it would draw comment, but not even my critics said anything, a pity.
The monotonic distortion comment is not original, it came from Jean Hiraga, a Franco-Japanese designer raised in Paris but living in Japan since the seventies, and the designer of 'Le Monstre', a famous amp originally described in 'La Revue du Son' back in the eighties. There is a thread on this amp in DIYaudio; this guy is at least as significant to the history of audio as John Linsley Hood or Nelson Pass.
The thinking behind the comment seems connected to the psychoacoustic masking effects of human hearing. A well documented effect holds that more lower order distortions, particularly H2, H3 and H4, will substantially mask the higher, more objectionable artefacts, leading to a more harmonious subjective listening experience. I have found this to be generally true; if we accept that amplifiers all produce distortion, it makes good sense to work with rather than against this phenomenon, in the manner perhaps of a pretty girl who works with makeup to mask skin imperfections and thus create the most pleasing appearance.
Phase shift is interesting. Phase relationships from distributed sound (such as an orchestra) are picked up differentially by varying path lengths to the two ears, and this, along with relative amplitude, is used to discern direction. We are talking very small phase differences here since path lengths may only be a few inches in thirty or forty feet, so any amp which introduces no phase shifts with increasing frequency should image more accurately. At the high end of audio the image depth, accuracy and integrity are sought-after qualities, but not easy to measure, certainly not found in specifications. Thus these issues are important, particularly where they relate to large distributed sources like orchestras. Classical music buffs like to be able to position the various instruments; it's important, and adds to the credibility of the listening experience. Drum sounds, for example, are so much more plausible if delivered from a point subjectively assessed to be right at the rear of the orchestra, just where they should be!! Image depth is notoriously difficult to achieve with solid state electronics, and is very well handled by most tube amplifiers. I strive for a credible image depth in my amplifiers, and the Lifeforce, Maya, Soraya and NAKSA all exhibit very believable image depth. It is definitely something one has to design in, and it is related to phase distortion.
Lastly, capacitor distortion is real and insidious. Caps store electrical energy as mechanical energy - the powerful attraction between the plates squeezes the dielectric tightly, straining it like a taut spring. When discharged, the attraction is dissipated, the dielectric 'relaxes', and the stored energy is transferred electrically to the load. The initial compression of most plastic materials is very non-linear; low pressures create quite a bit of strain, which soon linearises with increasing force. Logically the low voltage charging of a cap corresponds to this non-linear region, so that energy transfer is less than ideal. If the cap is 'primed' with a bias voltage at least exceeding the peak AC voltage moving through it, then the zero volt bias condition is thus avoided, and the transfer is much more linear. In some ways this is akin to the crossover region of a Class AB amp, as the signal traverses the zero point during the switching of the output devices as they pass current to the load. As a general rule for audio amps, this means around 2V DC should bias the input blocking cap, permitting it to transfer the audio signal with considerably less distortion than if the dielectric passes to and fro across the zero bias condition, as it does in most amplifiers, which generally place less than 0.1 volt across the input blocking cap.
When you do the comparison between biased and unbiased caps, you find that the 'quality' of the cap is not too important with a strong DC bias, and that boutique 'audiophile' caps can be eschewed as with DC biasing it is very difficult to hear any sonic difference. I have verified this with cheap, good quality industrial caps versus high value teflon foil caps.
I use this principle now in the Soraya, the Maya and the NAKSA, and the result is more 'n better music.
Hope this answers your points, John, thank you for posting,
Cheers,
Hugh
