part of it if I increase negative feedback to 'force' a 'better' signal
through the output transformers.
I put 'better' in '' because that's part of the issue: how one defines
'better'.
People talk a lot about clipping but it's not as easy to visualize from
text as it is by looking at the waveforms, like in spice (or 'the real
thing'). This is, IMO, also related to the less than scientific comment
you'll hear folks say that "tube watts sound bigger than solid state
watts" meaning a tube amplifier of x watts seems to be 'louder' than a SS
amp of the same power.
Basically, everything distorts and solid state amps 'solve' this by
employing gaggles of negative feedback but tube amps don't use nearly so
much. That affects clipping: what the amp does when it can't follow the
input signal.
It seems that what one might consider a 'draw back' of tubes is what makes
clipping less of a problem. Mainly that tube gain drops off as plate
voltage decreases, as in a large signal swing near the tube's maximum
power capability. That tends to flatten the signal and does, of course,
introduce distortion but it's a gradual distortion that seems to be less
offensive to the ears.
However, solid state amps, typically with heavy feedback, attempt to keep
the signal 'correct' right up to the point where it immediately goes to
hell in a hand basket introducing catastrophic, discordant, distortion.
And it isn't just a matter of putting the feedback in. It seems to be the
'nature' of solid state devices because their gain doesn't vary as much
with the signal itself. I.E. they tend to try amplifying right up to where
there's 'nothing left'.
Now comes the tricky part. It would seem that an easy solution to this
would be to simply not overdrive a solid state amplifier, so there's no
clipping, and the supposed 'problem' becomes moot.
My theory is that, in essence, you can't do it. That while the 'average'
material may seem to be well within clipping it still contains transient
peaks that do drive into the 'danger zone', so to speak, and that you will
get clipping distortion, hence a 'harsh' sound, introduced on those peaks.
It's like an elephant attacked by gnats. It's a small, seemingly
inconsequential, portion of the whole but enough to be 'noticed'.
The tube amp, however, will 'smooth' those peaks, and also distort, but
it's a more pleasing, 2'nd order, harmonic that tends to sound 'natural'
rather than harsh. Perhaps even 'complimentary'. (There may be some human
perception effects as well since the 2'nd order harmonic is a 'natural'
one so your mind might subconsciously 'fill in' missing elements as it
'expects' them to 'naturally' be, similar to how your eyes fill in missing
information with a TV picture, whereas the un-natural SS distortion is
'confusing' to the ear and discordant)
Which, IMO, is why it seems that 'tube watts are bigger than SS watts."
Trying to crank each to the same watt level will result in the SS amp
introducing harsh harmonics on the transient peaks so you want to turn it
back down. It simply 'sounds bad', even before the 'obvious' clipping,
whereas the tube amp's 2'nd order harmonics sound pleasing so it seems to
be just fine with the same watts. So you have the non-scientific 'human
perception' that you can crank a 1 watt tube amp as 'loud' as a 5 watt SS
amp because it still 'sounds good' even though, in reality, it's
distorting.
Or, put simply, tubes seem to operate like a natural, though crude,
compander. You can make the base, average, material 'louder' and they'll
naturally compand the peaks in a pleasing, albeit distorted, manner
whereas the typical SS amp doesn't.
However, wrap a boat load of negative feedback around a tube amp to
compensate for poor components (like my less than stellar output
transformers) and you run the risk of making it sound more like a SS amp
because it's also 'compensating out' the natural companding action of the
tubes. With enough feedback you can make them clip like a knife blade
slicing off the wave tops, just as in a SS amp.
There's also the issue that feedback, while 'correcting' for distortion,
also introduces new distortion of it's own and it may not be 'better' to
reduce the more pleasing 2'nd harmonics to infinitesimally small levels
while adding even an infinitesimal amount of discordant distortion. I.E.
Even a minuscule amount of bad distortion may be a whole lot worse than
large levels of 'pleasing' distortion, the elephant and gnats syndrome, so
'numbers', as real and 'scientific' as they are, don't necessarily reflect
'perception'.
Anyway, lots of things to look at