Reducing gain is not the criterion for feedback. There are ways of reducing gain which do not involve feedback, and ways of doing feedback which do not reduce gain (e.g. positive feedback!). The criterion for feedback is that a sample of the output is added to or subtracted from the input.simon7000 said:
It is not an expansion of the definition. Those who claim that degeneration is not feedback are restricting the definition. Degeneration is one method for local feedback.
Last edited:
There is one (that I know of) difference between the usual degeneration and the usual loop feedback and that is open loop bandwidth. Usually the open loop bandwidth of a single degenerated stage is much greater than our passband of interest. A loop around a whole properly stabilized amplifier usually has an open loop bandwidth less than our band of interest.
Why does this matter? Dunno, but the Baxandall/Putzeys feedback transfer function *is* different between the region where loop gain is flat and the region above open loop where the forward path is integrating. But what does this mean?
There probably isn't a cookbook general answer, but I wonder if a gedanken model could be worked up incorporating this difference that could help explain the common experiences of many WRT feedback, intrinsic linearity, yada yada.
Thanks,
Chris
many SOI, CFA op amps are now "faster" from input to output than many discrete Q used in audio designs - do they get a free pass on this "issue"?
feedback control around a delay has been studied in several contexts
"ultimate" limit of feedback in electronic amplification of course
in industrial process control for "transport delays"
in mixed signal digital control loops for ADC, DAC latency, calculation delay, 1st order holds
why doen't the crowd so suspicious of negative feedback try actually looking at the extensive literature?
feedback control around a delay has been studied in several contexts
"ultimate" limit of feedback in electronic amplification of course
in industrial process control for "transport delays"
in mixed signal digital control loops for ADC, DAC latency, calculation delay, 1st order holds
why doen't the crowd so suspicious of negative feedback try actually looking at the extensive literature?
I'd hardly consider myself "suspicious of negative feedback" since I'm currently working on ways to increase it for my new phono equalizer. I do however note that there is a fundamental, calculable and measurable difference between higher-than and not-higher-than open loop bandwidths, and that there is an extensive body of literature by experts, including the guy who name appears in the title of this thread, giving us several rough rules of thumb.
My interest is in refining the rules of thumb, not in rewriting quantuum physics.
Thanks,
Chris
good John, does that mean you have Cherry and Hooper?, oh I forgot - you dismiss them for not singling out slew rate, anticipating Otala - clearly any subsequent work can be ignored...
anything by B J Lurie? - tough going, I haven't understood everyting, text are "buggy" too which doesn't help - but some interesting insights I thought
if its just a ******* contest about whether such literature exists at all you could just look at http://www.eng.newcastle.edu.au/~jhb519/limits.html to get a flavor
the "standard" for this thread seems to quite a bit lower than peer reviewed journal's for "publishing" subjective listening results - more common seems to be to refer to buddies' off-stage "ears I trust" and admitting, stipulating in fact that blinding protocols can't be used to reproduce their "hearing" results
and I generally avoid claiming direct links between circuit theory and audible results - I almost exclusively "attack" assertions that move into circuit theory proper and can be answered there as best the EE community currently knows how
by that standard John's continual regurgitation that Otala's "flat open loop bandwidth" prescription is necessary for low PIM is about as "dead" on technical grounds as any idea gets to the EE professional community
note I am not linking this criticism to any listening result - "PIM" is a measurable signal distortion, caused by multiple circuit mechanisms - I've shown in "Classical Feedback" framework, using Bode's sensitivity analysis, and full circuit sim that John's claim that the JC3 inner feedback "works" to, in his words, reduce PIM is simply not true
the inner feedback R doesn’t make one class of PIM any worse, but doesn’t reduce others as much as simply using global feedback – and global feedabck gives the same performance on the distortions John claims the inner loop fixed
anything by B J Lurie? - tough going, I haven't understood everyting, text are "buggy" too which doesn't help - but some interesting insights I thought
if its just a ******* contest about whether such literature exists at all you could just look at http://www.eng.newcastle.edu.au/~jhb519/limits.html to get a flavor
the "standard" for this thread seems to quite a bit lower than peer reviewed journal's for "publishing" subjective listening results - more common seems to be to refer to buddies' off-stage "ears I trust" and admitting, stipulating in fact that blinding protocols can't be used to reproduce their "hearing" results
and I generally avoid claiming direct links between circuit theory and audible results - I almost exclusively "attack" assertions that move into circuit theory proper and can be answered there as best the EE community currently knows how
by that standard John's continual regurgitation that Otala's "flat open loop bandwidth" prescription is necessary for low PIM is about as "dead" on technical grounds as any idea gets to the EE professional community
note I am not linking this criticism to any listening result - "PIM" is a measurable signal distortion, caused by multiple circuit mechanisms - I've shown in "Classical Feedback" framework, using Bode's sensitivity analysis, and full circuit sim that John's claim that the JC3 inner feedback "works" to, in his words, reduce PIM is simply not true
the inner feedback R doesn’t make one class of PIM any worse, but doesn’t reduce others as much as simply using global feedback – and global feedabck gives the same performance on the distortions John claims the inner loop fixed
Last edited:
by negative feedback, i think that to mean "Global Negative Feedback"....
local or degenerative feedback in a single stage within the circuit doesn't count....
having said that, i am reminded of a tube amp wherein they put a cathode follower stage in front of the real input tube where the global negative feedback was fed to the cathode.........so there they could claim no global negative feedback........
As Ed points out, we have more ways to apply feedback than we have descriptive terms for them. I think in the interests of keeping things simple that his three broad categories make sense and cover all cases, except for "error correction". (In my opinion error correction, which relies on adding gain in the feedback loop, should be classified differently.)
And there will still be many circuits that are not so easy to classify -- such as the Jadis JA-80 tube amps that had a positive feedback loop inside the overall feedback loop. The idea was to increase the gain of the circuit so as to increase the overall negative feedback. (Which was likely one of the main reasons for its distinctive sound quality.)
The difficulty is that there will never be any agreement on what to call something. Some people will insist on calling a single tube with a cathode resistor (as found in a Conrad-Johnson preamp) "local feedback" instead of Ed's proposed "degeneration". And anybody that makes any circuit using degeneration does not want to use that in any of the descriptive material meant for the general consumer -- it sounds too much like "degenerate". (For non-US English speakers, "degenerate" is a serious insult, roughly equivalent to "pervert" which implies abnormal sexual practices. Not too appealing...)
This problem is quite handily illustrated by this post:
by negative feedback, i think that to mean "Global Negative Feedback"....
local or degenerative feedback in a single stage within the circuit doesn't count....
having said that, i am reminded of a tube amp wherein they put a cathode follower stage in front of the real input tube where the global negative feedback was fed to the cathode.........so there they could claim no global negative feedback........
The amp Tony is referring to is undoubtedly the Marantz 9 monoblock. A great amp and one I could happily live with for as long as I could get fresh tubes for it.
But the real point is that there is no way to describe this type of design easily. It doesn't really fall into one of Ed's proposed three categories. But back then, people all thought that negative feedback was a good thing, and if anything, the Marantz marketing people tried to hide the fact that the input tube was outside of the feedback loop. (I'm not even sure why they included the tube at all, but that is another story altogether.)Other examples include the Boulder power amplifiers. They are essentially two discrete op-amps in series, each with a lot of feedback around them. But there is no feedback around the composite. So technically they could claim "no global feedback". But that would be not only disingenuous, but outright deceptive to the lay consumer. However in this particular case it is not a problem. For one thing Jeff Nelson has far too much integrity to stoop so low (although many others do not). And besides, he likes feedback and sincerely believes that it improves a product.
Dunno what it means, but I do know that it affects the sound of the circuit.
Thank you for including my name in such august company. And something to note -- the first three names mentioned are designers whose products literally dominate the "Class A" ratings of Stereophile magazine. (Allen Wright also made outstanding products, but Stereophile did not review them due to their "five dealer" rule.)
So either:
a) The system is corrupt and the rankings are achieved through bribery or some other duplicitous means.
b) We have strange hypnotic powers to convince them of something that isn't actually true, even though (to the best of my knowledge) neither JC nor NP actually visit reviewers. I used to, from time-to-time, but have been unable to for the past five years due to an accident that paralyzed me.
c) Maybe, just maybe, we are on to something.
Of course, none of this will change anybody's mind, any more than a forum posting will change a person from a Democrat to a Republican (sorry for the US-centric reference).
Obviously a cheap attempt to denigrate John Curl, Thorsten Loesch, and myself. But why separate Nelson from us? What is it in your mind that makes Nelson a great designer and we three others such bad designers?
(This will be especially interesting, seeing as how Nelson himself just posted a couple of pages ago how if he didn't have stiff competition [from us], he wouldn't get anything done.)
- Status
- Not open for further replies.