There are actually many technical reasons that tube audio should sound better than solid state devices. These will be briefly discussed here in a manner which is, hopefully, not too intimidating to the casual browser.
There are two types of distortion which are
usually quoted in regard to high fidelity performance - harmonic distortion
and intermodulation distortion. Usually, harmonic distortion is quoted as X
% THD at a certain power output and frequency. This is a little deceptive
and the better, more reputable manufacturers will quote something like maximum
of X % THD below maximum rated power from 20 to 20000 Hz.
Intermodulation distortion is much more objectionable and is the result of two
frequencies beating together to make a third, unwanted sound. This should be
quoted as the combination of two frequencies at a particular power level. I
have noticed that manufacturers are not quoting intermodulation distortion as
frequently nowadays probably because it is the best way of predicting
what a piece of equipment will actually sound like.
The distortion measurements are of
interest, but they are not the whole story. A piece of equipment can be
designed to measure well but end up not sounding particularly good. There are
many factors not measured which can affect the sound quality and these may be
even more important than those which are. Since these factors are intangible
and not measurable they are controversial in the audiophile world. My two
cents on the issue is that I would rather listen to a tube system with
mediocre measurements than to a mass market receiver that has impressive
statistics. My judgement is reserved on high end solid state equipment
particularly since I have not had much opportunity to listen to any of it.
Tubes, especially triodes, are inherently
more linear than solid state devices. This means, in effect,
that what goes in is closer to what comes out, all things being equal (the
exact explanation would be rather complicated). The main mechanism to combat
the effects of non - linearity in an electronic device is called
negative feedback. Negative feedback returns a portion of the output
signal opposite in phase to the input; what this does is to cancel out
differences between the input and the output. The magnitude of the output is
somewhat reduced, but differences between the output and input are reduced
even more. It is possible to construct a perfectly acceptable tube type
amplifier without negative feedback, but I am not aware that this is currently
done with any solid state circuit since the distortion level would be
unacceptably high.
A tube type push - pull amplifier
without feedback can be made with harmonic distortion levels of less
than 3 % which is considered reasonably high fidelity. Since a solid
state amplifier requires much more feedback to reduce distortion it needs more
gain which entails more components and greater complexity. Any increase in
complexity tends to increase distortion as well; this increases the
number of engineering compromises which must be made. One result of this
is that solid state amplifiers tend to have slightly increasing distortion at
lower listening levels, and also have a sudden, huge increase in
distortion (clipping) when the signal level reaches the maximum
output levels (this is a frequent cause of burnt out tweeters, by the
way). In contrast, tubes tend to clip much more gradually and
have less distortion as listening level decreases. It is also thought by
many audiophiles that too much feedback actually reduces the sound
quality of a circuit and that it is preferable to use as little as possible.
Transistors are much more electrically
fragile than tubes. They must be protected from harm by special circuits which
are quite complex. A very great deal of the circuitry in a solid state
amplifier does nothing but protect the transistors from self destruction;
probably the majority of the devices in fact. This makes it even more
difficult to design an amplifier which sounds good.
Tubes are produced with much tighter tolerances than solid state devices (laws of physics) and solid state circuits must be designed to accept wide variations in device characteristics (or alternatively, the devices must be screened to ensure they are within specification). Tube circuits can be set up to take advantage of this and are consequently much simpler in scope. I am harping on simplicity for the simple reason that, in audio equipment, less is more. The fewer components hung on an amplifier the better as far as I am concerned. Complexity introduces issues like phase shift, rise delay, and a host of other poorly understood and unquantifiable changes to the sound as it goes through the equipment.