Post by John WilliamsonPost by Liz Tuddenham[...]
Post by WoodyIIRC correctly the main reason that a valve amp sounded 'better' is that
it tended to produce most distortion at second harmonic of which the
human ear is very tolerant. 5-10% distortion is barely noticed, but,
say, 0.5% third harmonic will quickly want you to give up!
Third harmonic sounds much more excruciating than second, but the
intermodulation that accompanies any type of harmonic distortion is
horrible. It produces a muddy background.of complex tones unrelated to
the music. It's one of those things that listeners with older systems
became used to, but there is a startling increase in clarity when it is
removed.
I had this "second harmonic is more musical" argument put to me by a
triode pre-amp enthusiast, so I loaned him my intermodulation meter. He
measured all sorts of things with it, but he never told me the outcome
of his pre-amp measurements.
The way I heard it was that with the valve amps, the first harmonic in
the distortion that was not musically related to the fundamental was
above most people's hearing range, while with transistors in class B,
the first unrelated harmonic was well within the audible range.
In effect, valve distortion showed up as an extra note in the chord, and
so "fitted in". Transistor distortion clashed, so your ear noticed it a
lot more.
There are several effects which get muddled up:
Even harmonics appear as the octaves of the fundamental frequency and so
they sound as though they could have been part of the music. Odd
harmonics do not fall on the musical scale, so they stand out by
clashing with the music. On single tones, the intermodulation
distortion which accompanies this distortion will not be noticeable -
but with complex music it generates sum and difference tones between the
fundamentals and between the harmonics caused by the distortion. This
results in a mushy background which varies with the pitch and amplitude
of the music.
Without feedback, single ended stages give predominantly even harmonic
distortion because of the shape of the transfer curve, whereas, in a
properly balanced push-pull stage, the even harmonics cancel and you are
left with the odd harmonics, which are usually lower in level.
Designers of push-pull amplifiers are then tempted to drive the devices
harder until the distortion figures read the same as previous
amplifiers, so the effect is that now the third harmonic appears at the
same level as the second did previously. Thus a 3-Watt single-ended
amplifier will not sound as bad as a 3-Watt push-pull amplifier if they
are both rated to give 5% distortion at full output.
Triodes have internal feedback because part of the voltage gradient
between anode and cathode appears in the space between grid and cathode
where the electrons emitted from the cathode are gradient-controlled.
Claims that triode amplifiers work without feedback are nonsense because
the feedback is already built-in and is not as linear as external
feedback with resistors would be. A cascode connection, where the
cathode-follower action of the upper valve prevents voltage change on
the anode of the lower valve is one way of overcoming this internal
feedback because the anode-to-cathode voltage gradient of the lower
valve doesn't change with the signal. Tetrodes and pentodes achieve the
same effect with a screen between the anode and the other electrodes
Tetrodes, pentodes and transistors have virtually no internal feedback,
so the non-linearity (and production variability) of their transfer
characteristics have to be corrected by external feedback.
The phase shifts occurring in the output transformer of a valve
amplifier make the feedback become positive outside the audio range. If
there is a lot of gain and a lot of feedback at these frequencies, the
overall loop becomes unstable and oscillates. This means that valve
amplifiers are restricted in the amount of feedback they can use,
whereas transistor amplifiers with no signal transformers can use much
more feedback (and tend to rely on it for linearisation). With heavy
feedback, the linearity of the transfer characteristic is improved, but
when the output 'hits the stops' the kink is much sharper and the
harmonics this generates are much greater and extend to higher
frequencies. This is why, in general, valve amplifier overload
gracefully but transistor amplifiers overload disgracefully.
Feedback cannot overcome slew-rate-limiting, which was the cause of the
so-called 'Transient Intermodulation Distortion' of some early
transistor designs. Class-B stages with inadequate power supplies,
signal-dependent rectification artefacts and unintended feedback through
power connections can all give rise to audible effects but these are
caused by faults in the design, not by whether the amplifier uses valves
or transistors.
--
~ Liz Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk