[Don Pearce[_3_]]

On Sun, 04 Sep 2011 12:20:16 -0500, John Byrns
wrote:

In article ,
(Don Pearce) wrote:

On Sun, 04 Sep 2011 11:24:20 -0500, John Byrns
wrote:

In article ,
(Don Pearce) wrote:

On Sun, 04 Sep 2011 00:16:25 -0400, Don
wrote:


If I had two KT88's (I wish!) I'd build the Williamson design, but
stabilize it with the "low frequency shelf" discussed here at length a
short while ago.
Cheers,
Roger

Regarding the "low frequency shelf" and high frequency
roll-off filters:
why are they usually after the voltage amp?
Why not a band-pass filter before the V1 tube?

o-----||---/\/\/\/\--------------------o
| |
===
|
|
| |
| |
o---------------------------------------o

-Don

I think it is a sort of masochism - a craving for low and high
frequency instability. Large phase shifting mechanisms within a loop
are never a great idea. There should be ideally a single pole that
doesn't let up until beyond the unity gain frequency.

The use of a dominant pole to stabilize the feedback loop is transistor
thinking, not always practical in tube circuits.

Just good engineering practice and pretty much standard in op amps.
And the first op amps were valve, not transistor.

That's all pretty much true, however we aren't talking op amps in this thread,
were talking valve audio amps, and more specifically the particular design Don
posted.

Can you give us a worked example of how you would use a dominant pole to
stabilize Don's amplifier design, specifying all relevant time constants in the
amplifier. I would be especially interested in how you would do the low
frequency stabilization were a number of irritating tradeoffs often come into
play, and any assumptions about the effect of the OPT on phase shift are likely
to be more accurate than at high frequencies. The object is for the amplifier
to remain stable with a speaker load, no load, and any other load it might be
likely to encounter during its lifetime.

Short answer - no, it isn't possible with this design. Capacitive and
transformer AC coupling combine to demand low frequency compensation
to get anything reasonably stable. Motor boating has always been a
plague to this kind if amp. As for high frequencies, there are too
many uncontrolled sources of phase shift for that to work. And of
course the severe limitation in open loop gain puts the cherry on the
top.

Dominant pole is what you use when you have everything under control.

As for different loads, if the amp had decently low output impedance
(which demands high OL gain and feedback), there would be issues with
stability into odd loads.

The design is a compromise, but that is fine - it is clearly an
experiment with something a little historic.

d