Chris Hornbeck wrote

Negative feedback always reduces phase shift.

Except, of course, when it doesn't. Feedback of *any* kind
always reduces the stability of the system. Small signal
open loop amplitude and timing errors are reduced in
proportion to the gain reduction; beyond the limits
where gain is reduced...

Could have done with perhaps another word or two here. It's not the most common
way of putting it...

In general I was impressed with Arnie's succinct and, dare I say, sane
contribution. Of course you are right to point out the constraints. Arnie
doesn't always cross his tees.

In fact I would be happy to adopt a combination of your contributions as a
working definition of "feedback". Anything I can't connect a wire to is
metaphysics.


I bring this up because we all have a mindset that
feedback is an unavoidable necessity in electronics,
especially in larger stages driving loudspeakers.
But, strangely, we have no such mindset concerning
loudspeakers themselves, which have one or two
orders of magnitude larger distortions of all kinds,
(and introduce a few new ones unique to themselves),
yet we never insist on feedback around our loudspeakers.

Some do, surely?

If misapplied it can
cause an amplifier to become less stable, but in fact when properly applied,
negative feedback increases stability.

Sophistry...

What, from Arnie? It's just his definition of stability is a little unstable.

Feedback of any kind always decreases stability.

Thanks. And thanks again.

In the sense that everyone always means the term.

True of "stability", more or less, but sadly not of "feedback".

And, there is no such thing as unconditional stability.
There is only conditional stability, and that only by
constraining circumstances. Life's a bitch; like that.

And again. My life's kind of OK though, I have to admit.

For example, the
gain of both tubes and transistors can be very sensitive to temperature.
With negative feedback, the important parameters of the amplifier are set
by a pair of resistors, whose properties can be made to be very stable and
independent of temperature.

Perhaps not the best example, because vacuum tubes operate
in their own very high temperature environment and were made
to tolerances impossible for semiconductors, buy yeah, a
feedback loop of a pair of resistors is a pretty durned
good reference.

Wot, no compensation? Is that possible?

cheers, Ian