Bob Cordell Interview: Negative Feedback

Graham Maynard said:
Hi All,


We are a full 40 years down the road since Nelson Pass published his class-A NFB amplifier which (memory guess) also had an output impedance around -65dB at 10kHz, but at a Sine angle of only ~10deg. This 10 degree figure confers an additional 15dB advantage in loudspeaker control at 10kHz (high AF) compared to the so often seen 90deg, and the Pass amp had NO crossover distortion, so forgive me for being cynical, but I often wonder what has really been achieved during a period so long that todays men and women were not even twinkles in their parents eyes.


The problem is that not all of us agree with your point; I think you are simply wrong (if I understand you correctly). As I have said before, yes, if you look at the output impedance of a feedback amplifier it will often tend to be a bit inductive; presumably this is the phase angle you are referring to (what else could it be, if viewed from the outside world?). However, the fact that it is somewhat inductive is, in and of itself, not particularly important or problematic; it is a matter of degree.

As I have pointed out before, in those feedback amplifiers that employ an output inductor for stability, it is often the case that the 3-5 uH of inductance of this passive device is on the order of, or more than, the inherent inductance of the feedback amplifier.

My EC amplifier had a damping factor of 5000 out to 20 khz (as measured without an output inductor), and yet it used plenty of NFB with dominant pole compensation. How's that for loudspeaker control?

Cheers,
Bob
 
Hi Bob,

I do not know what EC stands for, but PMA passed me a link and his circuit turned out to be an IC driven D2S amplifier similar to that published by Ian Hegelun in EW.

Take two of these amplifiers. Load one with an equivalent LS and one with 8R in order to equalise group delay. Then X-Y the difference. This is 1V input at 10kHz.

Now this error is far greater than anyone would expect for an amplifier having a simulated DF of -66dB as I illustrated earlier in this thread. Amplifiers that appear inductive when looking into the output terminal interact in unexpected ways when loaded by a LS.

Cheers ........ Graham.

PS. Resizing from 1024 to 1000 damaged the image.
 

Attachments

  • pma.gif
    pma.gif
    47.8 KB · Views: 659
Graham Maynard said:

PS. Resizing from 1024 to 1000 damaged the image.

I believe we are both old enough to remember the 8-bit images on the first Windows PC's. Some people behind the scenes are currently working very hard on the new software which will eventually further improve the quality of the graphical possibilities. 🙂 Sorry for OT everyone.

/Hugo
 
Graham Maynard said:
Hi Bob,

I do not know what EC stands for, but PMA passed me a link and his circuit turned out to be an IC driven D2S amplifier similar to that published by Ian Hegelun in EW.

Take two of these amplifiers. Load one with an equivalent LS and one with 8R in order to equalise group delay. Then X-Y the difference. This is 1V input at 10kHz.

Now this error is far greater than anyone would expect for an amplifier having a simulated DF of -66dB as I illustrated earlier in this thread. Amplifiers that appear inductive when looking into the output terminal interact in unexpected ways when loaded by a LS.

Cheers ........ Graham.

PS. Resizing from 1024 to 1000 damaged the image.

Hi Graham,

you shown my circuit here (inspired by Nelson Pass). Anyway, slight differences exist. Mainly, CCS current must be higher.

Regarding Ian Heggelun, his circuit is a bit different in principle. But you seem to read my mind, as I am working on a new amp.

Regards,
Pavel
 
Nelson Pass said:


Not to be argumentative, but isn't -65 dB approximately the
same figure as DF = 5000?

😎


Hi Nelson,

I'd guess -65 dB would correspond to a DF of about 1800, but that's certainly close enough among friends.

My point was that with a DF that high, if it is slightly inductive it doen't matter a hill of beans in the big picture. Even a pure 0.1 uH inductor would have a DF of about 700 at 20 kHz, and it seems hard to imagine that a speaker would care about an extra 0.1 uH in the source impedance (especially in the case where the amp might already have an output inductor on the order of a couple uH).

Cheers,
Bob
 
Hi Bob,

Personally I would not want to use an amplifier using an output choke or lead inductance greater than 500uH, nor would many who use a ribbon tweeter.

You appear to be discounting that output phase shift could be a problem for the amplifier itself though, for DF is entirely active, and when load requirements render the amplifier non-linear then it is the amplifiers having phase shifted damping which momentarily jump higher off the NFB loop rails.

Damping factor is about loudspeaker control, not amplifier control. Yes I know that damping factor confers amplifier immunity, but only as long as the amplifier remains linearly active, and simple tests do not necessarily reveal problems. Maybe Jan's suggestion would.

Cheers ........ Graham.
 
Graham Maynard said:
Hi Bob,

Personally I would not want to use an amplifier using an output choke or lead inductance greater than 500uH, nor would many who use a ribbon tweeter.

You appear to be discounting that output phase shift could be a problem for the amplifier itself though, for DF is entirely active, and when load requirements render the amplifier non-linear then it is the amplifiers having phase shifted damping which momentarily jump higher off the NFB loop rails.

Damping factor is about loudspeaker control, not amplifier control. Yes I know that damping factor confers amplifier immunity, but only as long as the amplifier remains linearly active, and simple tests do not necessarily reveal problems. Maybe Jan's suggestion would.

Cheers ........ Graham.

Hi Graham,

I said output inductors on the order of less than 5 uH, not 500 uH. I totally agree that a 500 uH output inductance would be excessive.

I think you are generally confusing linear and nonlinear effects.

Cheers,
Bob
 
john curl said:
Stay the course, Graham. Inductors are not good. Don't ask me why, just listen.


I agree that the use of stabilizing output inductors remains a matter of controversy. I could just as eaily choose to make my amplifier without one, but that would not change the fact that the output is slightly inductive (i.e., impedance rising slightly with frequency). Your amplifiers are also at least as slightly inductive, even if they do not employ an output inductor.

My point is that Graham's concern about the phase angle of a microscopic inductance is not well-founded. Even the outputs of amplifiers with no negative feedback are slightly inductive, as you well know.

Bob
 
Hi Bob,

Sorry that was a typo; I meant 500nH as being my limit for acceptance.

Why do expert folk try to tell me that I am confusing linear and non-linear effects ?

Loading can make an amplifier non-linear = amplitude distortion in time.
Linear amplifier + load induced variation of overall propagation delay through internal inductance = time distortion of amplitude.
The greater the number of active stages, and the higher the capacitance of devices the greater the effect upon asymmetrical waveform drive, which sine investigation cannot reveal.

Okay the latter is small, but can still be much higher than specified amplitude THD.


Hi John,

I agree, and I did check it out by listening in my younger years. There are many sources/amps/speakers where you could not tell whether say a 5uH choke is being used or not, this because they are simply not good enough.

Whether I'd be able to hear the difference now I do not know.
If I could hear it when young, others still could today, so standards should be upheld.

Cheers .......... Graham.
 
Bob wrote;-
"Even the outputs of amplifiers with no negative feedback are slightly inductive, as you well know."

Yes, and thus they do not generate an output terminal error potential wrt the NFB amplifier's ultra low resistance source of amplified input.

NFB amplifiers intensify current flow and increase loudspeaker circuit Q, and when the amplifier output stage is not near flat phased it can react with the loudspeaker at some instants thus generating reproduction peaks. eg. tizzieness, upper mid peaks etc.

Cheers ....... Graham.
 
Graham Maynard said:
Bob wrote;-
"Even the outputs of amplifiers with no negative feedback are slightly inductive, as you well know."

Yes, and thus they do not generate an output terminal error potential wrt the NFB amplifier's ultra low resistance source of amplified input.

NFB amplifiers intensify current flow and increase loudspeaker circuit Q, and when the amplifier output stage is not near flat phased it can react with the loudspeaker at some instants thus generating reproduction peaks. eg. tizzieness, upper mid peaks etc.

Cheers ....... Graham.


Gobble-dee-gook.

Cheers,
Bob
 
BOB and powerbecker,

Stating "Gobble-de-gook" shows either an inability to discuss reasonably, or an unwillingness to !!!
Rubbishing anyone's comment stands alonside witnessing (for those not involved) unnecessary argument, and awful (for those involved) personal insult.


Now I do not know whether either of you have heard NFB which reduces THD yet induces reproduction distortion or not, many here have, so if you have then please explain this phenomenon in YOUR own words in case you have a different way of thinking about what I have so far clearly failed to communicate to your differently thinking minds.

And I say this without accusing you too of writing Gobble-de-gook or of being wrong.
(This is how incorrect audio related aspects become published by 'experts', because so often they control/ignore the publishing of feedback, and mainstream readers never get to hear about other possibilities, which in general tend to be due to reactive phase relationships and the lowering of NFB loop generated output impedance so often stated as being a prime desirable aim!)

I have heard this immeasurable change in reproduction and I know that many here have too, so I await your explanation please !!!

If my explanation is wrong then please deconstruct so that everyone of us might learn.

Next please explain in your own words the generation of the red and green traces which were part of my illustration above in post#396, for this *is* the distortion due to internal amplifier capacitance rendering a NFB amplifier output terminal inductive and thus generating an entirely new 'interface induced error' voltage wrt amplified input at the low impedance output node, as compared to other arrangements on the same SS topology giving the yellow and mauve traces.


Cheers ........... Graham.