🚀 go-pugleaf

I just read an article by  francis@cs.adelaide.edu.au (Francis Vaughan):

>Another alternative is possible.  Rather than use passive
>attenuation to set the output level, why not run the signals
>to the power amp at the line levels of the rest of the
>system and then vary the gain of the amplifier?  This
>presupposes an amplifier designed using a fairly normal
>feedback system.  

Yes, one usually thinks of long wires ("interconnects") as an
opportune place for noise to find its way in, hence larger singals
there are appealing.

>The gain is set as the ratio of the two feed back resistors.
>By replacing these with a simple potentiometer the gain
>could be varied from anything between zero and whatever the
>maximum useful gain of the system is.

>Disadvantages must include a potential loss of stability as
>bits of the feedback loop have long wires attached to them.
>Also if the potentiometer became worn there would be a real
>danger of the feeback loop opening suddenly, leaving the amp
>producing its open loop gain straight into the speakers.
>Very nasty.

These are indeed disadvantages.  To insure stability, the 
amplifier's open-loop gain is intentionally rolled off so that
the product of its open loop gain and its feedback attenuation
is less than 1 when calculated at the frequency of 180 degree
phase shift in the open loop gain.  To lower the gain, the
attenuation of the feedback is reduced and the amp requires 
more compensation (which results in less high-frequency gain
and a slower slew rate).  Designers would have to compensate
for the lowest possible gain, and then live with the fact that
at higher gains the amp has more compensation than needed to
keep it stable.  

Non-power circuits like op-amps may be compensated to be
stable at unity gain (the lowest possible gain under normal
circumstances) and have specifications that are so good that
one does not regret the fact that they are over-compensated
when operated at a higher gain.  I think that power amp designers
are more challenged by large voltage swings, reactive loads, high
currents, etc. so that they want to select just the right amount of
compensation for a particular gain.

I do not think that one stands to gain much signal-to-noise ratio.
Lowering the gain of the power amp would only decrease the noise of the
power amp itself and the noise picked up by the wires.  Unless one is using
a rather high-powered amplifier that is designed with correspondingly high
voltage gain (so that it can be driven to deliver its higher power) it seems
unlikely that the amplifier and cable noise would dominate.

One way to get some of the advantages listed without a variable gain amplifier
is to use attenuation at the input of the power amp.  For instance, 
on Haffler 220 amps that have no input potentiometers, I have installed
resistive dividers designed such that the optimum output level of the
preamplifier causes about the right listeneing level.

A good idea, but it has some practical problems with implementation.  

Tom Matthews