juma,
you seem to have missed that I was not directly addressing you - If I hadn't offered any criticism of your position, how would I "soften it"
you however show a less excusable reading comprehension problem when you speak for Mr Pass after I quoted his post in another thread that directly contradicts your words
you may be well educated but so far you do not seem to be grasping the terms of the debate or are simply wrong - as can be determined from skimming a few of the papers/presentations that the Google search I recommended brings up
when talking about "University" accepted EE understanding/teaching of feedback control theory there is "right" and "wrong" - unlike Literary Criticism:
xkcd: Impostor
Ila may wish to construct a personal definition of "linearization... or bootstrapping technique" when the rest of the engineering world calls it "feedback" but I was hopeful that his post of a small signal circuit schematic of the sort used academically in these analysis indicated an openness to education
you seem to have missed that I was not directly addressing you - If I hadn't offered any criticism of your position, how would I "soften it"
you however show a less excusable reading comprehension problem when you speak for Mr Pass after I quoted his post in another thread that directly contradicts your words
you may be well educated but so far you do not seem to be grasping the terms of the debate or are simply wrong - as can be determined from skimming a few of the papers/presentations that the Google search I recommended brings up
when talking about "University" accepted EE understanding/teaching of feedback control theory there is "right" and "wrong" - unlike Literary Criticism:
xkcd: Impostor
Ila may wish to construct a personal definition of "linearization... or bootstrapping technique" when the rest of the engineering world calls it "feedback" but I was hopeful that his post of a small signal circuit schematic of the sort used academically in these analysis indicated an openness to education
I'm in deep waters here, but I just read about NFB in Bob Cordell's book this morning.
He says that feedback does arrive nanoseconds after the original signal, but its just
another way of saying that it is less effective at higher frequencies. He goes on to
say that this effect is very much present in the audible frequencies and is the reason
why distortion increases at 20 kHz compared to 1 kHz.
I assume you are talking about the article which should
be the subject of this thread. If there is some factual
inaccuracy in the piece, I would be happy to be informed
of it. I took another look at it, and there does not appear
to be any opinion offered on the subject of feedback.
I build amplifiers with and without feedback loops,
preferring whichever does the job best for the intended
application, or simply as an exploration.
As to degeneration, it was already understood when Black
invented negative feedback. Horowitz and Hill refer to it
as a "form of negative feedback" but there is a specific
distinction - a degenerated common emitter/source
/cathode circuit experiences a higher output impedance
with degeneration and a lower output impedance with
loop negative feedback.
I am not arguing the merits of either - I make too much
money from feedback to be negative about it.
I do think it is helpful to speak with more clarity, and I
always make the distinction between degeneration and
loop feedback. Given that a finite resistance always exists
in the emitter/source/cathode, and a finite output
impedance is always intrinsic to the collector/drain/plate,
you can argue that there is no such thing as "no feedback".
Thanks for that.
I never quite understood what people/books meant when they said a source follower has current gain but not voltage gain.
I think current gain then is a misnomer. There is no such thing as current gain. There is only Ohms law which transposes to I=V/R
If I have misunderstood what you are saying then please correct me.
Instead of "has," substitute "can have." It depends on input impedance and load.
I can feel a paradox coming on.
It does seem then, that there is no such thing as 0 negative feedback.
However, if you can have positive feedback (or is it feed forward), and negative feedback, then a lay person would logically conclude that 0 feedback should be possible.
However this does not appear to be the case. I hope the planet does not stop spinning
Without 0, positive and negative do not exist
So in conclusion lets call it 42back
Don't worry about me I have been reading too much of Hitch Hikers Guide to the Galaxy.
It does seem then, that there is no such thing as 0 negative feedback.
However, if you can have positive feedback (or is it feed forward), and negative feedback, then a lay person would logically conclude that 0 feedback should be possible.
However this does not appear to be the case. I hope the planet does not stop spinning
Without 0, positive and negative do not exist
So in conclusion lets call it 42back
Don't worry about me I have been reading too much of Hitch Hikers Guide to the Galaxy.
Last edited:
Hi Nelson,
I think I said “tone” in regard to your paper, the final page contains:
"
...you can visualize instances of feedback pyramid schemes...
We have seen that nonlinear distortion becomes larger and more complex depending on the nonlinear characteristic of the stages, the number of cascaded stages, and the number of spectral elements in the music...
The resulting complexity creates distortion which is unlike the simple harmonics associated with musical instruments, and we see that these complex waves can gather to create the occasional tsunami of distortion, peaking at values far above those imagined by the distortion specifications.
If you want the peak distortion of the circuit of figure 13 to remain below .1% with a complex signal, then you need to reduce it by a factor of about 3000. 70 dB of feedback would do it, but that does seems like a lot.
By contrast, it appears that if you can make a single stage operate at .01% 2nd harmonic with a single tone without feedback, you could also achieve the .1% peak in the complex IM test.
I like to think the latter would sound better.”
Hardly a “ringing endorsement” of negative feedback in audio amplifier design?
It is sad commentary on the state of Audiophile “knowledge” that you have to introduce Intermodulation Distortion as the “unknown” Elephant in the room – it kind of blows rather large holes in any theory of “consonant distortion” – and the way it is introduced could mislead some into perhaps thinking it is a new idea to electrical engineering also – despite its recognition in early radio and audio engineering – some would say the difficulty of achieving low IM in “no feedback” tube circuits drove Black's invention of negative feedback
Your presentation of cascaded stage distortion while mathematically correct is either naive or disingenuous (in part due to the unstated assumption of equal signal levels) to my “ears” in any feedback amplifier discussion – really read Cherry’s Last AES paper – the “input referred” distortion analysis is not original with Cherry but he does put several concepts together in a powerfully explanatory way
Cherry, “Estimates of Nonlinear Distortion in Feedback Amplifiers” JAES V48#4 2000
"A method is proposed for characterizing nonlinear distortion in feedback amplifiers by means of two generators at the input - one voltage and one current. These generators are the negatives of the voltage and current antidistortion which, added to voltage-source and current-source input, respectively, would reduce the output distortion to zero. Their values can be calculated easily from known variations of small-signal parameters such as transistor gm, ß, and the capacitances. The distortion-to-signal ratio is given by formulas similar to the offset-to-signal ratio or noise-to-signal ratio, and depends only on the distortion generators, the signal source, and the source impedance. The method correctly predicts harmonic and intermodulation distortion for practical amplifiers. It points toward techniques for reducing distortion, and it explains certain experimental results which hitherto have been problematical.
"
Briefly your simple cascading of identical nonlinearities to create that “tsunami of distortion” ignores that the signal and distortion-to-signal ratio decreases as you move towards the earlier/input stages of a high gain feedback amplifier’s cascaded gain stages – and all but the last stage of most audio amplifiers operate in Class A – so the generated distortion is very low in those early gain stages anyway - it is quite typical for the distortion generated in the diff pair tanh nonlinearity to be sub ppm for audio signals
And 70+ dB of combined local and global feedback is already present in typical “Lin” topology variation Cordell, Self designs to beyond 20KHz (remember that gain is "tied up" inside the VAS Cdom Miller feedback loop)
To defend calling degeneration something other than “negative feedback” is like insisting that pholgiston theory’s “successful phenomenological explanations” (if there were any) would have to be referenced as preferred due to their historical priority in any modern chemistry class
I think I said “tone” in regard to your paper, the final page contains:
"
...you can visualize instances of feedback pyramid schemes...
We have seen that nonlinear distortion becomes larger and more complex depending on the nonlinear characteristic of the stages, the number of cascaded stages, and the number of spectral elements in the music...
The resulting complexity creates distortion which is unlike the simple harmonics associated with musical instruments, and we see that these complex waves can gather to create the occasional tsunami of distortion, peaking at values far above those imagined by the distortion specifications.
If you want the peak distortion of the circuit of figure 13 to remain below .1% with a complex signal, then you need to reduce it by a factor of about 3000. 70 dB of feedback would do it, but that does seems like a lot.
By contrast, it appears that if you can make a single stage operate at .01% 2nd harmonic with a single tone without feedback, you could also achieve the .1% peak in the complex IM test.
I like to think the latter would sound better.”
Hardly a “ringing endorsement” of negative feedback in audio amplifier design?
It is sad commentary on the state of Audiophile “knowledge” that you have to introduce Intermodulation Distortion as the “unknown” Elephant in the room – it kind of blows rather large holes in any theory of “consonant distortion” – and the way it is introduced could mislead some into perhaps thinking it is a new idea to electrical engineering also – despite its recognition in early radio and audio engineering – some would say the difficulty of achieving low IM in “no feedback” tube circuits drove Black's invention of negative feedback
Your presentation of cascaded stage distortion while mathematically correct is either naive or disingenuous (in part due to the unstated assumption of equal signal levels) to my “ears” in any feedback amplifier discussion – really read Cherry’s Last AES paper – the “input referred” distortion analysis is not original with Cherry but he does put several concepts together in a powerfully explanatory way
Cherry, “Estimates of Nonlinear Distortion in Feedback Amplifiers” JAES V48#4 2000
"A method is proposed for characterizing nonlinear distortion in feedback amplifiers by means of two generators at the input - one voltage and one current. These generators are the negatives of the voltage and current antidistortion which, added to voltage-source and current-source input, respectively, would reduce the output distortion to zero. Their values can be calculated easily from known variations of small-signal parameters such as transistor gm, ß, and the capacitances. The distortion-to-signal ratio is given by formulas similar to the offset-to-signal ratio or noise-to-signal ratio, and depends only on the distortion generators, the signal source, and the source impedance. The method correctly predicts harmonic and intermodulation distortion for practical amplifiers. It points toward techniques for reducing distortion, and it explains certain experimental results which hitherto have been problematical.
"
Briefly your simple cascading of identical nonlinearities to create that “tsunami of distortion” ignores that the signal and distortion-to-signal ratio decreases as you move towards the earlier/input stages of a high gain feedback amplifier’s cascaded gain stages – and all but the last stage of most audio amplifiers operate in Class A – so the generated distortion is very low in those early gain stages anyway - it is quite typical for the distortion generated in the diff pair tanh nonlinearity to be sub ppm for audio signals
And 70+ dB of combined local and global feedback is already present in typical “Lin” topology variation Cordell, Self designs to beyond 20KHz (remember that gain is "tied up" inside the VAS Cdom Miller feedback loop)
To defend calling degeneration something other than “negative feedback” is like insisting that pholgiston theory’s “successful phenomenological explanations” (if there were any) would have to be referenced as preferred due to their historical priority in any modern chemistry class
Last edited:
Some feedback, or Lots of feedback.
At the end of the day the ear has the final say, especially if you are interested in selling amps to people who don't give a rats **** about feedback but care about what the amp does for their favourite music collection.
The music should be enjoyable to listen too, how you get there does not matter.
At the end of the day the ear has the final say, especially if you are interested in selling amps to people who don't give a rats **** about feedback but care about what the amp does for their favourite music collection.
The music should be enjoyable to listen too, how you get there does not matter.
I largely agree with that. As you go higher in frequencies, the phaseshift between the input signal and the fraction of the output signal that is fed back, increases. The result is that the subtraction at the input stage is less complete, there remains a residual, that is higher at higher frequencies.
The whole idea is that the original signal components would cancel in the subtraction and that the distortion (which is only in the output signal fraction) remains and acts as a kind of 'pre-distortion' to make the output less distorting.
So, if at higher frequencies the cancellation is less good, the distortion components remain somewhat 'hidden' and the 'pre-distortion' works less well.
jan didden
I think it's simpler than it looks. Evidently, a source follower cannot have voltage gain.
Now look at the input gate, suppose that gate has a 1meg ohms input Z, and suppose that at the source you have a 1k load.
If you connect a 1V input signal, that takes 1uA from the signal. That 1V coming out on the source causes 1mA into your 1k load. So it seems clear there is a current gain of 1000.
jan didden
Last edited:
I can feel a paradox coming on.
It does seem then, that there is no such thing as 0 negative feedback.
However, if you can have positive feedback (or is it feed forward), and negative feedback, then a lay person would logically conclude that 0 feedback should be possible.
However this does not appear to be the case. I hope the planet does not stop spinning
Without 0, positive and negative do not exist
So in conclusion lets call it 42back
Don't worry about me I have been reading too much of Hitch Hikers Guide to the Galaxy.
Very astute observation. I fully agree with Nelson that you can't have any active component in a circuit and have zero feedback. BUT I agree with you that you can connect a positive feedback loop to cancel the neg feedback. In fact, I've played with such circuits, they have some interesting properties but are a [female dog] to stabilize and to keep the pos and neg fb in exact balance.
jan didden
Well, yes, but 'gain' basically relates two values like input to output. That's why 'gain' is dimensionless. Ohms' law acts locally, isolated, on a component with a voltage across it and has the dimension, err, ohms. So they're two different things. I don't see how one has anything to do with the other.
jan didden
Attention Moderators
Please block Zen Mod from seeing this thread.
This could get me into deep trouble at the Pass Pub
You're one of the very few people here who have the mental power to do that. I think you should change your moniker
That's it.
jan didden
seems that jcx is on thin ice , asking Papa about his paper , without considering that majority of Papapers are written for usual DiyA greedy boy , not for vane JAES members court .
MH
regarding all things called feedback - I personally can't care less , because these things are way over my head , considering that I'm (repeatedly demonstrated ) much dumber than usual DiyA greedy boy .
so -
is triode just pentode with intrinsic NFB ........
or is pentode just triode with intrinsic PFB ?
( I can't care less even for that much more important question , which is in root of all things and universe - as long I have xformer between so ideologically questioned part , and speaker )
, without considering that majority of Papapers are written for usual DiyA greedy boy , not for vane JAES members court .MH
regarding all things called feedback - I personally can't care less , because these things are way over my head , considering that I'm (repeatedly demonstrated ) much dumber than usual DiyA greedy boy .
so -
is triode just pentode with intrinsic NFB ........
or is pentode just triode with intrinsic PFB ?
( I can't care less even for that much more important question , which is in root of all things and universe - as long I have xformer between so ideologically questioned part , and speaker )
Last edited:
A little experiment
Once again, it was not my intention to open up the never ending feedback story again, nor do I want to hijack this thread, which is originally about Nelson’s paper from 2008.
From my point of view, his paper is very accurate and well written as ever. It sums up what is common knowledge and good practise in audio engineering. I always enjoy reading his articles. I hope it was not an infringement of the majesty by pointing to an article I wrote mainly for customers.
Now to the controversial story:
Most, if not all available textbooks call the emitter degeneration circuit “current feedback”. There is no doubt about it. By repeating something again and again it is not getting better. The name has historic reasons and we should be a little more cautious using it. Current feedback operational amplifiers like LT1223 carry the same name, but work with a real feedback loop.
The name “current feedback” for this circuit originates from the following considerations:
A rise of the input voltage causes Ube to rise; Ic increases now; at the same time the voltage drop over Re increases too. The rising voltage on Re counter acts with input voltage, leading to a decreasing collector current Ic. That’s what is meant by feedback here, I am pretty clear about it. But this is not a classic feedback loop, it is just the action of the transistor trying to satisfy the equation
Ic = Is x exp(Ube/Ut).
You may call it the inherent feedback of active devices. Because it is inherent, it is very fast in contrast to loops which comparably slow.
Perhaps a little experiment can help us here. Please take a look at the three circuits a – c) All circuits work with quiescent current of 6mA, a collector (plate) resistance of 1kOhm. The output voltage of each of the three circuits is “disturbed” by injecting an error current of +/- 1mA or 707uA rms. In the real world, this error current may originate from an altering base current of a heavily load BJT stage connected to the output. A feedback loop would try to counter act this disturbance. Circuit a) does the expected; we notice a correction current of 456uA (rms). Circuit b) is doing nothing to counter act. The resulting emitter current is 345nA (rms), 1300 times lower than in circuit a). The very small fluctuation is due to the Early effect. Circuit c) is a vacuum tube ECC82 with an inherent feedback. This feedback effect is called Plate resistance. From the numbers it is easy to calculate an Rp of 7.3kOhm which is roughly what you find in the datasheet of this tube.
Once again, circuit b) shows no reaction; it is not a classic negative feedback loop.
Q.E.D.
A last word to the degeneration / transfer curve story. You won’t find a statement in my paper, that degeneration is not altering the transfer curve of the amplifier. This would be a silly implication.
Once again, it was not my intention to open up the never ending feedback story again, nor do I want to hijack this thread, which is originally about Nelson’s paper from 2008.
From my point of view, his paper is very accurate and well written as ever. It sums up what is common knowledge and good practise in audio engineering. I always enjoy reading his articles. I hope it was not an infringement of the majesty by pointing to an article I wrote mainly for customers.
Now to the controversial story:
Most, if not all available textbooks call the emitter degeneration circuit “current feedback”. There is no doubt about it. By repeating something again and again it is not getting better. The name has historic reasons and we should be a little more cautious using it. Current feedback operational amplifiers like LT1223 carry the same name, but work with a real feedback loop.
The name “current feedback” for this circuit originates from the following considerations:
A rise of the input voltage causes Ube to rise; Ic increases now; at the same time the voltage drop over Re increases too. The rising voltage on Re counter acts with input voltage, leading to a decreasing collector current Ic. That’s what is meant by feedback here, I am pretty clear about it. But this is not a classic feedback loop, it is just the action of the transistor trying to satisfy the equation
Ic = Is x exp(Ube/Ut).
You may call it the inherent feedback of active devices. Because it is inherent, it is very fast in contrast to loops which comparably slow.
Perhaps a little experiment can help us here. Please take a look at the three circuits a – c) All circuits work with quiescent current of 6mA, a collector (plate) resistance of 1kOhm. The output voltage of each of the three circuits is “disturbed” by injecting an error current of +/- 1mA or 707uA rms. In the real world, this error current may originate from an altering base current of a heavily load BJT stage connected to the output. A feedback loop would try to counter act this disturbance. Circuit a) does the expected; we notice a correction current of 456uA (rms). Circuit b) is doing nothing to counter act. The resulting emitter current is 345nA (rms), 1300 times lower than in circuit a). The very small fluctuation is due to the Early effect. Circuit c) is a vacuum tube ECC82 with an inherent feedback. This feedback effect is called Plate resistance. From the numbers it is easy to calculate an Rp of 7.3kOhm which is roughly what you find in the datasheet of this tube.
Once again, circuit b) shows no reaction; it is not a classic negative feedback loop.
Q.E.D.
A last word to the degeneration / transfer curve story. You won’t find a statement in my paper, that degeneration is not altering the transfer curve of the amplifier. This would be a silly implication.
Attachments
Of course b) doesn't react as you are injecting outside the feedback circuit.
If you would inject in the feedback circuit, like in the emitter, you WILL have a reaction because you are now disturbing the feedback circuit. If you do that, you will see all the classical events of a reacting feedback circuit.
Anyway, I think you make an artificial distinction between a 'classical fb loop' (whatever THAT is) and the local feedback provided by an Re.
jan didden
If you would inject in the feedback circuit, like in the emitter, you WILL have a reaction because you are now disturbing the feedback circuit. If you do that, you will see all the classical events of a reacting feedback circuit.
Anyway, I think you make an artificial distinction between a 'classical fb loop' (whatever THAT is) and the local feedback provided by an Re.
jan didden