Wrapping my mind around stability (or instability)

[CascoSieg]

Recently, PDF64 wrote "margin of stability would be improved if its presence control was moved to the other end of the front panel. [...]" (in this thread: https://el34world.com/Forum/index.php?topic=27723.msg304840#msg304840 ). Shortly after I wondered why he thought that (and I'd certainly welcome PDF64's thoughts about it), I realized that while I have heard about parasitic oscillation, positive feedback, good vs. unwanted instability, I probably don't have a good enough grip on it to understand an explanation.  One doesn't have to read long about Ken Fisher to find something referencing his artful management of feedback and instability in his circuits, but I've never found a good resource to help me understand it better: what are the causes, conditions, controllable dynamics, etc., what are all the ways it can sound like?
I'd love to hear about builders' experiences with it, and certainly any technical references to guide my learning about it.
If I wanted to breadboard a test circuit to play around with it, what should that look like?
Thanks!

[PRR]

The Fender style Presence pot is quite late in the amplifier, near speaker level. The build in question had it near (I forget) first VOL pot? Input jack? At first glance it suggests output-near-input, the classic way to get a squeal. But it may not be as "hot" as it looks.

[p2pAmps]

The Fender style Presence pot is quite late in the amplifier, near speaker level. The build in question had it near (I forget) first VOL pot? Input jack? At first glance it suggests output-near-input, the classic way to get a squeal. But it may not be as "hot" as it looks.

Yes indeed but after testing this in all the ways I could think of it just wasn't an issue.  In real world playing I cannot make it an issue.  The first one of these amps I built the buyer insisted on this arrangement so I tried it and it worked out just fine.  Not saying in all cases this is correct or wouldn't cause an issue but in my particular case its fine.  Believe me fellas, if this was in any way giving me grief it would be moved immediately.

Cheers

[pdf64]

I can’t think of why there would er be good instability?
Or why it would be attractive to push an amp’s gain / bandwidth such that it was on the verge of instability (ie margin of stability was zero). As the slightest change (eg different valve / pickup / speaker / atmospheric conditions) could result in instability.

An audible squeal (free running oscillation) etc is only one subset of how instability might manifest.
eg may be free running oscillation above audio bandwidth, parasitic oscillation (ie amp is stable at idle but blips of astable oscillation manifest under certain signal level conditions). 

When taking my EE degree, I found the ‘classical control theory’ module incredibly enlightening, and was massively helpful in getting my head around this stuff. So you may want to try dipping your toes into that.

Theoretically, an amp could be modelled using control theory, all its poles and zeros identified / quantified, and a result be calculated for its margin of stability. In reality that’s awfully complex to attempt, though it’s feasible to do it for its power amp.

Testing for stability is tricky; need a sig gen, scope, resistive dummy load and ideally a reactive dummy load, though a real speaker could be utilised.

[p2pAmps]

I can’t think of why there would er be good instability?
Or why it would be attractive to push an amp’s gain / bandwidth such that it was on the verge of instability (ie margin of stability was zero). As the slightest change (eg different valve / pickup / speaker / atmospheric conditions) could result in instability.

An audible squeal (free running oscillation) etc is only one subset of how instability might manifest.
eg may be free running oscillation above audio bandwidth, parasitic oscillation (ie amp is stable at idle but blips of astable oscillation manifest under certain signal level conditions). 

When taking my EE degree, I found the ‘classical control theory’ module incredibly enlightening, and was massively helpful in getting my head around this stuff. So you may want to try dipping your toes into that.

Theoretically, an amp could be modelled using control theory, all its poles and zeros identified / quantified, and a result be calculated for its margin of stability. In reality that’s awfully complex to attempt, though it’s feasible to do it for its power amp.

Testing for stability is tricky; need a sig gen, scope, resistive dummy load and ideally a reactive dummy load, though a real speaker could be utilised.

Thanks for all the great input and information. 

[shooter]

Testing for stability is tricky

especially when "breaking" into a un-designed OR designed feedback loop.  Akin to looking into the box, seeing if the Cats alive or dead 

[p2pAmps]

Testing for stability is tricky

especially when "breaking" into a un-designed OR designed feedback loop.  Akin to looking into the box, seeing if the Cats alive or dead 

Just not sure what else to do.  I have 20 of these in the field with working pros and ZERO complaints of any kind.  Did I dodge a bullet I wonder.

[acheld]

Did I dodge a bullet I wonder.

Of course!   That's the beauty of it.  It's easy to predict oscillation, but it doesn't always happen the way we think it should.

So when you find a design that works, that is amp heaven.

[pdf64]

Sorry, don’t want you to be unnecessarily concerned, if any nasties do emerge, it may / probably only be at extreme control settings, into a load mismatch etc. Something outside the norm. And even then, it may not even be noticeable.

eg JTM45 can be problematic, as they have such a high feedback ratio. I happened to plug one I’d just built into a cab whilst the scope was still on its output. With the presence at min it started to oscillate, but it was at 40kHz, so was inaudible, and it didn’t affect the amp’s capability at all.
Often free running oscillation can throw the bias of one or another stages off, but not in this case. A couple of 100pF caps brought it and my OCD under control. In reality it wouldn’t have been an issue, for me at least, as presence never goes below 75%.

[shooter]

an epic FAIL on a feedback loop happened on a radar director, 2 axis, couple ton unit.  It slammed the stops 3 or 4 times before I could hit the kill switch, the entire ship felt the slams!!
you do ALOT of testing the repairs BEFORE you have any confidence in success   

[CascoSieg]

Ok, that helps. I'll see what I can find about "classical control theory" but so far I'm gathering that to the extent it is possible to identify and control all the factors that may contribute to stability/instability, as with other complex non-linear systems, it may not be worth trying to understand it in complete and precise terms. And, at the same time it also sounds like there are distinct strategies for keeping things "linear enough", harmonically complex (as in KF), and otherwise able to perform with consistency.
Thanks for chiming in!

[jjasilli]

Recently, PDF64 wrote "margin of stability would be improved if its presence control was moved to the other end of the front panel. [...]" (in this thread: https://el34world.com/Forum/index.php?topic=27723.msg304840#msg304840 ). Shortly after I wondered why he thought that (and I'd certainly welcome PDF64's thoughts about it), I realized that while I have heard about parasitic oscillation, positive feedback, good vs. unwanted instability, I probably don't have a good enough grip on it to understand an explanation.  One doesn't have to read long about Ken Fisher to find something referencing his artful management of feedback and instability in his circuits, but I've never found a good resource to help me understand it better: what are the causes, conditions, controllable dynamics, etc., what are all the ways it can sound like?
I'd love to hear about builders' experiences with it, and certainly any technical references to guide my learning about it.
If I wanted to breadboard a test circuit to play around with it, what should that look like?
Thanks!

I think there is a semantics problem lurking in the ambiguous meaning of the word instability.  Perhaps even overdrive tone can be considered by someone to be a form or degree of instability.  But more technically it means the amp is malfunctioning.  Hence the reaction above that there can be no good instability.

Feedback can lead to instability as PRR said.  Even negative feedback may not actually be as negative as we expect, but be partially in phase with the main signal.  This could lead to runaway positive feedback and loud squealing.

[shooter]

it may not be worth trying to understand it in complete and precise terms.

I had to take all the classes AND pass the tests, then I went work in the real world 
the simplest answer is keep large signals from small signals, including ground and Power.  that simple practice will avoid 93.72% of the problems, the rest you gotta read/study to avoid 

[tubeswell]

JTM45 can be problematic, as they have such a high feedback ratio. I happened to plug one I’d just built into a cab whilst the scope was still on its output. With the presence at min it started to oscillate, but it was at 40kHz, so was inaudible, and it didn’t affect the amp’s capability at all.
Often free running oscillation can throw the bias of one or another stages off, but not in this case. A couple of 100pF caps brought it and my OCD under control. In reality it wouldn’t have been an issue, for me at least, as presence never goes below 75%.

Yes I find 100-250pF g1-k on both output tubes can knock this oscillation out (but I don't trust mica caps for this because when they start to short, it knocks the bias out). 1-3kV 100-500pF between the LTP plates is less troublesome, but equally as effective. 100-220pF 'speed up' cap in parallel with the NFB resistor also works.