[Chris Hornbeck]

Continuing related drivel in previous threads:

To date, we've generated three pairs of small signal
response plots, with each pair comprising a "Bode plot".

The first pair describes our amplifier's "forward"
or "open-loop" response. We've measured this by removing
the feedback scaling network from the amplifier output
and connecting that end of the network to ground. We've
then roughly (we don't need a lot of detail) plotted
output, both amplitude and phase, compared to a fixed input.

Our second pair of amplitude and phase response plots
is of the feedback scaling network itself. We've
measured this by driving the network at the point just
previously "removed from the amplifier output and
connected to ground" for the first pair plot and observing
at the input summing junction, what in the conventional
case Patrick calls V1 cathode. Good name.

The third Bode plot is just a combination of the first two,
but it's what describes our feedback amplifier's
response, including both forward and feedback response
together (duh.) We all know about how gain must fall
below unity before phase shift approaches 180 degrees,
yada-yada, and *this* is the place where that applies.


Enough recap. Now let's do three new pairs of response
plots at large signal and with a Cursedly difficult load.

Definitions of "difficult" vary with amplifier topology,
expected loudspeaker type and personal threat aversion.
Hopefully, useful discussion here will broaden perspectives
and help narrow choices.

"Large signal" means voltage swings near the maximum
possible, and can most conservatively include an attempt
to measure response when in clipping. Can't be trusted,
but might be interesting.

The new second pair of response plots (of feedback scaling
network response) doesn't normally need to be redone from
small-signal conditions (shouldn't change enough to matter),
so new The Curse plots are just for the forward path and a
new summed third pair representing our amplifier in (what we
define as) worst-case conditions.


And we're STILL not to a "what to do with this bullcrap"
point. Sheesh. But we do have a nominally complete picture
of our amplifier's performance under both idling and
wailing conditions. To understand stability, we literally
have *everything* that we need. Everything.

Hoping Patrick will jump in near this point with some of
the next stuff... (he can hardly be comparatively too
long-winded!)

Chris Hornbeck

[Patrick Turner]

Chris Hornbeck wrote:

Continuing related drivel in previous threads:

To date, we've generated three pairs of small signal
response plots, with each pair comprising a "Bode plot".

The first pair describes our amplifier's "forward"
or "open-loop" response. We've measured this by removing
the feedback scaling network from the amplifier output
and connecting that end of the network to ground. We've
then roughly (we don't need a lot of detail) plotted
output, both amplitude and phase, compared to a fixed input.

Our second pair of amplitude and phase response plots
is of the feedback scaling network itself. We've
measured this by driving the network at the point just
previously "removed from the amplifier output and
connected to ground" for the first pair plot and observing
at the input summing junction, what in the conventional
case Patrick calls V1 cathode. Good name.

The third Bode plot is just a combination of the first two,
but it's what describes our feedback amplifier's
response, including both forward and feedback response
together (duh.) We all know about how gain must fall
below unity before phase shift approaches 180 degrees,
yada-yada, and *this* is the place where that applies.

Enough recap. Now let's do three new pairs of response
plots at large signal and with a Cursedly difficult load.

Definitions of "difficult" vary with amplifier topology,
expected loudspeaker type and personal threat aversion.
Hopefully, useful discussion here will broaden perspectives
and help narrow choices.

"Large signal" means voltage swings near the maximum
possible, and can most conservatively include an attempt
to measure response when in clipping. Can't be trusted,
but might be interesting.

The new second pair of response plots (of feedback scaling
network response) doesn't normally need to be redone from
small-signal conditions (shouldn't change enough to matter),
so new The Curse plots are just for the forward path and a
new summed third pair representing our amplifier in (what we
define as) worst-case conditions.

And we're STILL not to a "what to do with this bullcrap"
point. Sheesh. But we do have a nominally complete picture
of our amplifier's performance under both idling and
wailing conditions. To understand stability, we literally
have *everything* that we need. Everything.

Hoping Patrick will jump in near this point with some of
the next stuff... (he can hardly be comparatively too
long-winded!)

Chris Hornbeck

I am just trying to work out exactly what is expected of me.

Patrick Turner.

[Iain Churches]

"Patrick Turner" wrote in message
...

Hoping Patrick will jump in near this point with some of
the next stuff... (he can hardly be comparatively too
long-winded!)

Chris Hornbeck

I am just trying to work out exactly what is expected of me.

Patrick Turner.

Some diagrams? Of typical examples.
Excellent thread: Thank you Chris


Iain