BTW, it is precisely because of the linearity of the tube that we can design 'romance' into it. You do that by manipulating operating points and plate loading, and when you design for the lowest possible distortion with tubes, you finish with a sound very like good SS, at least in terms of detail and accuracy.
...
To amplify this - if you think back to various discussions here regarding extra biasing of solid state amps, this is what Hugh has called "building in some romance".
I am willing to bet that by taking an amp and increasing its output stage bias, which pushes it more towards class A operation, we do several things at once: we subjectively speed up the output stage (because its class A to B crossover is now higher up, it can react in class A fashion to more of the signal), we linearize it somewhat (how much depends on several factors, and in particular on the transfer characteristics of the output devices) and effectively make it sound "sweeter".
However, we also increase, yes INCREASE, distortion, because it had in all probability been biased at its lowest distortion point. Once we leave that point in either direction, we start having more distortion. What type, how much, where and how is impossible to generalize because there are too many variables, but I find that by and large THD goes up somewhat (essentially immaterial, as even so it will still stay way below the hearsing threshold), but also the intermodulation factors, which are not to be taken lightly.
I completely agree with Hugh's assessment that the lowest distortion point will yield best "detail", and if you add bias to your amp, you will probably notice that you have lost some extra fine detail, but that the whole is better integrated into a more seamless entirety. Less bass, mid and high, more of the organic whole.
Cheers,
DVV
