Pooh Bear said:

To illustrate that a tube produces distortion, all you need to do is to look at the
transfer characteristic. No tube has a truly linear 'curve' .... ahem - hence they
are called curves it seems - lol.

Ahem........let's put some nuances into this picture.
Look at the "curves"of a BJT, a JFET, a MOSFET, a triode and a
pentode.
ALL of them are non-linear.
Tubes are somewhat quadratic, BJTs have an exponential curve.

The best part is that triodes have the most linear and long Ia/-Vg
curve of all devices.
Besides, no one is so naive to use a bare device, there's always some
feedback involved, be it locally or globally.

The fun part is that tubes with little LOCAL feedback don't
necessarily need global feedback to rely on for relatively
distortion-free transfer of audio signals.
Transistors, and especially BJTs, do need heavy doses of it to just
function as proper amplifiers .

As you probably know, the use of SS and GNFB can add distortions that
have little or no relation to the input signal. The spectrum of the
distortion alters.
Of course, local feedback in SS stages (e.g. emitter degeneration in
differential amplifiers) may solve most of these problems, but the
mostly very high common mode signal may still pose problems.
HF interference is another subject. FTZ and NP0 don't solve this
altogether. EMC is a serious problem ( and business) these days.

This is something I rarely see in consumer audio applications however.

Then there's something like "thermal distortion", variable junction
capacitances etc.

Of course, this is a very simple approach of a subject where one could
fill thousands of MBs with and still not be complete.

May I refer you to the work of Matti Otala, Artur Seibt, Jean Hiraga,
Kaneda-san and Peter Garde for instance? A very interesting read,
note.

If you're in for an in-depth discussion of these matters, wait until
Sunday, then I'll have an entire day to discuss this and related
subjects.

I'd be delighted to do so.
Just let me know, and I'll start a separate thread with this subject
here in RAO. Maybe RATech too, as there are some sharp minds wandering
around there as well.

--
Sander deWaal
"SOA of a KT88? Sufficient."