Feedback
Lost'n Found wrote
Do you guys like or dislike feedback in amplifiers?
I know it is a question that . . . . -_- sigh
But still, what do you think?
I think we can't live without feedback, so what is ur feedback?
Ah, well, erm...we seem to like arguing about it, anyway.
Firstly, on the definition of feedback. Every real dynamic system
contains feedback in the sense of regulation due to load, or dynamic
equilibrium. This sense is trivial for our purposes.
Engineers have a particular meaning, however, which comes from and is
precisely defined by control system theory. One part of this
definition is the "canonical system diagram", which represents a
system as a forward function, whose output is summed with (or
subtracted from) the system input, and whose input is that sum. There
is no function in the feedback path. The implication is that the
forward and feedback paths must be separate and distinct. If there is
in reality only one path from input to output, then the analysis
becomes trivial, and there is no point in considering it as a system
with feedback.
Feedback in the trivial sense cannot be a bad thing because it is in
everything. Hence if it is bad, there is no such thing as good, in
which case bad becomes as trivial as the feedback: it doesn't exist.
Feedback in the engineering sense is a bad thing, IMHO. Often,
however, it is less bad than its absence in a particular real circuit.
Think of it like medicine. Never a good thing, but better than being
sick.
Bad because it is a complication. Because it is problematic and the
problems require solutions at the expense of further complication.
Because the complication is not euphonic and so, because the solutions
are never perfect, the inevitable faults aren't musical.
The issue is strongly linked to another. Some believe that domestic
audio systems should be excluded from the category of musical
instrument, on the grounds that their function is merely to reproduce.
It follows from this premise that performance can be measured in terms
of deviations from some original music that happened somewhere else,
or at a different time, or both. Performance measured by such criteria
is always improved by any reduction in the sum of deviations.
Perfection is guaranteed if every deviation is zero, and hence the sum
of deviations is also zero. Such perfection cannot be achieved without
feedback, and in reality so far feedback has been found necessary to
get even close.
Several problems arise from that view. Such perfection is not actually
possible: deviation is never quite zero. On the simple face of this,
there is no agreed method of combining various deviations, or kinds of
distortion, into a single measure. Hence there is no single measure of
quality. As a simple example, if I can reduce 2H distortion by 20dB at
the expense of creating 3dB of extra 7H and a spot of crossover
distortion, is that an improvement? Or a matter of taste...perhaps
even statistically average taste? Second, rather less simply, it is
close to self-evident to say that the idea of reproducing the sound of
a symphony orchestra or a rock concert in my room is impossible. Two
speakers in this room is just never going to do that. Now, it is
possible to see this in terms of error, as an engineer might, and end
up with another matter of taste.
I believe there is a serious philosophical question behind all this. I
don't actually *want* a reproduction. Music is music...it is
indubitably here in my room...my system is making actual real music.
Judging it by the sum of differences is just not appropriate. Fidelity
is not the same thing as precision. I want my music to be coherent *in
its own right*, to have its own spirit and life. I want to hear the
music the musicians play, not what they measure. That is not a matter
of taste, but one that begs philosophical interpretation.
The same question arises in a different form if I ask whether there is
a single, perfectly right and proper way of playing, say, Beethoven's
5th? Obviously not, so is every performance, regardless of any measure
of quality, of equal merit? Obviously not. How do we define fidelity?
How do we judge performance?
When it seems just right, just now. It is a single feeling, not a sum
of differences. It is unlikely to come from a system with lots of
feedback, which is aimed at the sum of differences. Some report it is
possible, more or less just about nearly, with none. Or very little.
You'll know when, or not.
cheers, Ian
"in message . ..
|