Bob Cordell Interview: Negative Feedback

Re: Global NFB or not?

Johan Potgieter said:
This thread goes so fast that I may be repeating - in that case apologies.

Depending (very much!) on the design - so in general: Even with better consistency in the characteristics of modern transistors, if the necessary measures are not taken with enough local NFB, there may be such a variation in worst case global gain/phase characteristics that it is well nigh impossible to get consistent/stable results with global NFB without trimming. This is a strong point in favour of local NFB, with or without extra global NFB.


Johan,

You are very much right on target. Whether it is by emitter degeneration or by local loops (including feedback compensation loops like Miller compensation), local feedback to achieve a consistent, well-behaved forward path is important, whether the amplifier will be then enclosed in a global loop or not.

The use of global NFB does not give license to ignore the quality of the open loop. This is the mistake that some made in applying NFB in the early days, and which resulted in bad-sounding amplifiers that gave NFB a bad rap.

Cheers,
Bob
 
Re: Re: Re: Re: Re: Re: Re: The Effects of Psychology on Evaluation

Bob Cordell said:
Hi Mike,

Never give up Mike. I hope it wasn't anything I said that is pushing you away. I didn't think I was disagreeing with you, and would like to better understand your point and where you are coming from. This is definitely the place to discuss the pros and cons of NFB.

Whatever it is that the zero FB cowd loves and misses, we should try to better understand IT. The IT may or may not have to due with absolute faithfulness to the origibal signal. For example, If one were to build a zero-FB amp with extremely low distortion and extremely high DF, would it sould more like an NFB amplifier or like a typical no-FB amplifier in reasonably controlled tests (i.e., the listener doesn't know which uses NFB and which does not)?

Cheers,
Bob

Hi Bob,

I'm a grownup (sometimes) no one is pushing me away.

My experience and the conclusions drawn are not a good fit here. There is a high level of technical expertise to which I tend to oversimpify by comparison, which is fine. It's the opposite of preaching to the choir.

This internet fish bowl is to much for me, there's not much positive in it, so... I'll send you an e-mail.
 
Re: Re: Re: Re: Re: Re: Re: Re: The Effects of Psychology on Evaluation

I’m going to say one more thing and then move on.

The reason for yet another one of my attempts at conversation on this thread / site is that I honestly believe in and have experimented extensively with the topics I chose to comment on. Does that make me right or wrong? No, but it does fuel the need to try to discuss my experiences with people that might provide thoughtful feedback. Two years ago I found this site and, after reading a large chunk of the Blowtorch thread I had learned a lot and, thought that there might be that interaction that I now see I was looking for.

After many attempts, on many topics, I can say that there is a well defined body of knowledge that circulates here and anything outside of this, which I will say is examined in minute detail, is not appreciated nor welcome.

As for my latest foray into this feedback discussion, a few weeks back I made an observation about Bob’s interesting series of simulations examining the old wives tales involving the audible effects of feedback versus non-global feedback (my interpretation). I was impressed with the foresight to do the project and present the results. I felt that the results supported my personal contention that a properly implemented design does not suffer from its use and even brought up the possibility of moving from simulation to an extremely interesting concept of a test amplifier. The conversation died there. I wrestled with the thought of proposing this to Bob directly, since I have most the materials and experience to build a couple of sets quite easily, but decided not to because without enthusiastic support of the idea it would just turn into a science fair project and prove nothing.

My last series of posts were simply to see if there was any acknowledgment of the value of physical testing. It never even got around to that part, just a black and white, “your wrong” about a one liner. For my part I tried to reiterate that there is no proof that any of the “known to be true factors” such as the final possible resolution of both the CD and vinyl formats were anything other than people sitting around postulating. Any scientific follow-through to test these assumptions were obviously flawed. In other words, the LP would be dumped before its full realization, and the attempts to replace the “low-res CD” with DVD-A and SACD would also follow years later, all based on words. Both of their performances today supports the thought that their possible performance was not topped out.

I assumed my meaning was clear but forgot to take into account the lack of interest factor that occurs once the writer is assumed to be threading outside of the known topic list and set of key beliefs. Also responsible is my attempts to not be to wordy. Actually it was a pointless exercise from the outset, but I’m such a blind optimist I just keep trying.

No matter, I’ve spent a little bit of time over the past couple of days doing some homework and have come to understand that most of the key players on this site have been here for many years. Many of the topics seem to circle back around after a while and amusingly a quite similar conversation repeats itself. What’s technically right and wrong is clearly delineated (I wish I had realized this before stepped on a few of these topics early on, only to find myself twisted around on some side issue and ending up embarrassed, and a bit more unsure the next time I tried to participate).

Anyway, because this thread has frozen on my last posting and I felt a bit of clarification for my part was warranted (mostly for me). I want to add this and let the discussion return to the technical aspects of feedback, which is already in progress.

Regards, Mike.
 
Re: Re: Re: Re: Re: Re: Re: Re: Re: The Effects of Psychology on Evaluation

MikeBettinger said:
I’m going to say one more thing and then move on.

The reason for yet another one of my attempts at conversation on this thread / site is that I honestly believe in and have experimented extensively with the topics I chose to comment on. Does that make me right or wrong? No, but it does fuel the need to try to discuss my experiences with people that might provide thoughtful feedback. Two years ago I found this site and, after reading a large chunk of the Blowtorch thread I had learned a lot and, thought that there might be that interaction that I now see I was looking for.

After many attempts, on many topics, I can say that there is a well defined body of knowledge that circulates here and anything outside of this, which I will say is examined in minute detail, is not appreciated nor welcome.

As for my latest foray into this feedback discussion, a few weeks back I made an observation about Bob’s interesting series of simulations examining the old wives tales involving the audible effects of feedback versus non-global feedback (my interpretation). I was impressed with the foresight to do the project and present the results. I felt that the results supported my personal contention that a properly implemented design does not suffer from its use and even brought up the possibility of moving from simulation to an extremely interesting concept of a test amplifier. The conversation died there. I wrestled with the thought of proposing this to Bob directly, since I have most the materials and experience to build a couple of sets quite easily, but decided not to because without enthusiastic support of the idea it would just turn into a science fair project and prove nothing.

My last series of posts were simply to see if there was any acknowledgment of the value of physical testing. It never even got around to that part, just a black and white, “your wrong” about a one liner. For my part I tried to reiterate that there is no proof that any of the “known to be true factors” such as the final possible resolution of both the CD and vinyl formats were anything other than people sitting around postulating. Any scientific follow-through to test these assumptions were obviously flawed. In other words, the LP would be dumped before its full realization, and the attempts to replace the “low-res CD” with DVD-A and SACD would also follow years later, all based on words. Both of their performances today supports the thought that their possible performance was not topped out.

I assumed my meaning was clear but forgot to take into account the lack of interest factor that occurs once the writer is assumed to be threading outside of the known topic list and set of key beliefs. Also responsible is my attempts to not be to wordy. Actually it was a pointless exercise from the outset, but I’m such a blind optimist I just keep trying.

No matter, I’ve spent a little bit of time over the past couple of days doing some homework and have come to understand that most of the key players on this site have been here for many years. Many of the topics seem to circle back around after a while and amusingly a quite similar conversation repeats itself. What’s technically right and wrong is clearly delineated (I wish I had realized this before stepped on a few of these topics early on, only to find myself twisted around on some side issue and ending up embarrassed, and a bit more unsure the next time I tried to participate).

Anyway, because this thread has frozen on my last posting and I felt a bit of clarification for my part was warranted (mostly for me). I want to add this and let the discussion return to the technical aspects of feedback, which is already in progress.

Regards, Mike.


Thanks for your observations, Mike. I think there may be a lot of truth to some of your observations, although I don't think that the participants in these threads are trying to keep certain area of topics from being discussed.

I have also at times put out an idea and it either did not get picked up on, or it took off on a distracted path. In some ways, these threads are a microcosm of the industry. Some of the really hard things to do are not followed through on, only to be replaced by speculation or people citing years of experience and this is how it has always been.

We all need to be able to challenge conventional thoughts out there without being hammered by conventionalism.

A long time ago I actually did build an amplifier that could easily be reconfigured between no feedback and high feedback and did some tests on it. It was in the context of the IIM debate started by Otala. By doing so, and by using Otala's own recommeneded test approach, I showed that the amplifier actually had lower IIM when it used feedback.

But we must also not throw out the baby with the bathwater. While my experiments showed that Otala was wrong in blaming NFB for IIM, they did not assert that there was not something in the nature of IIM that was worth looking at. Related to this, for example, was the need for good output current capability in power amplifiers. Blaming the problem on NFB was simply an unfortunate distraction. I think a lot of that sort of thing still happens to this day.

But these kinds of experiments and detailed measurements can be hard to do, and sometimes it is easier for people to speculate.

As a very curious person, I really do wish there were more controlled follow-ups on debated issues. Follow-ups that attempt to eliminate non-relevant variables, for example.

I was fascinated to read the descriptions by Charles Hansen in another thread on his resistor listening tests. It raised a thousand follow-up questions in my head. He just rightly cares about what is the best-sounding resistor to use in his amplifier, while I probably tend to care more about why that resistor sounds better than the others; e.g., is there some relevant measurement that we can isolate so, going forward, we might know what characteristics to look for. Maybe it has something to do with the metallurgy of the interface between the resistive element and the external leads. Maybe there is some kind of rectification or resistance modulation going on there.

But if there was, and that was responsible for the sound difference, wouldn't we be able to measure it. Would it result in a measurable distortion in the equipment in which it is used? These kinds of questions fascinate me.

By the way, I have very definitely seen very measurable nonlinearity in power resistors used as dummy loads for power amplifiers. I have also seen measurable distortion at low frequencies in power amplifiers when the feedback resistor is not well-over-sized due to its temperature variations with signal.

Cheers,
Bob
 
The Effects of Psychology on Evaluation

Positive as always, Bob. You must be younger than I am... (oops! don't reply! :smash: )

Just a general observation:

Bob Cordell said:
But these kinds of experiments and detailed measurements can be hard to do, and sometimes it is easier for people to speculate.

It is very often easier to speculate, and that would be OK, if only not so many basic scientific facts were strangled in the process....

As a very curious person, I really do wish there were more controlled follow-ups on debated issues. Follow-ups that attempt to eliminate non-relevant variables, for example.

Like me. If only one had the necessary equipment (which I was once privileged to have had access to), and having that, if one had the time...

I was fascinated to read the descriptions by Charles Hansen in another thread on his resistor listening tests. Maybe it has something to do with the metallurgy of the interface between the resistive element and the external leads. Maybe there is some kind of rectification or resistance modulation going on there.

I searched but could not find the actual description among the host of relevant posts. But this is not new. If it was an observation only ... With all due deference, with what amount of credibility can one still be expected to view such in this day and age - but, as said, I did not read.

But if there was, and that was responsible for the sound difference, wouldn't we be able to measure it. Would it result in a measurable distortion in the equipment in which it is used? These kinds of questions fascinate me.

Yes, Bob, by now one would be able to measure it - audible effects, that is (with instruments capable of measuring > 40dB below threshold-of-hearing). And many people tried if only to humour the question, and found nothing of nearly a relevant magnitude. It seems like the once popular "diodic effect" in wire crystals. (Diodes don't start to conduct much below 0,5V nominally, etc. - not to resurrect that cyclic debate again. But others keep letting the draught in.)

To support you and all others of like inclination, I have the same malady (actually once earned a salary that way). I also am puzzled by what honest and knowledgable friends report to hear, and I also am still looking for answers. (In the process I had the dubious "success" to show what many things are not.) But my departing point is that we are not going to advance by re-inventing the wheel, or just make very sure that 3+5 is actually still 7.99999 (general remark; not saying you or others do). We certainly don't know nearly everything, but we do know certain things. And this is the brambly kind of undergrowth an honest rersearcher too often has to struggle through - more's the pity. But it is encouraging to find company.

Edit: Getting rid of some of the re's in the title!
 
I might chime in here in wholesome support of Mike Bettinger, whose recent post rings all too true.....

EEs have given us audio, video, and a host of other monolithic technologies of yore. They have sought to substantiate and mass produce every new development, and argued unstintingly for objectivity and math insights. They have assisted the industrialists in the economies of manufacture, and often enhanced technical performance in the process. This is all good.

However, the nature of EE is doing it better for less. The advances which most of the electronic technologies - the discovery of the transistor, the laser, the integrated circuit and a host of others - tend to be more in the line of physics, originally discovered by scientists. On that basis alone, unless we are looking at the truly visionary EEs (who presumably would hold doctorates), we might be looking at the wrong group for pushing the limits and showing an open mind.

With due respect to Bob, and a few others here who are genuinely creative people, the majority of discussion has been circular, focussed on math, particularly PSpice, modelling, with the implicit assumption that absolute distortion measurements correlate harmoniously (?) with the listening experience. The empiricists are dismissed as cowboys, and when an impressive design which has been built and auditioned comes along, like Edmond/Ovidiu's PGP, rather than fostering a rash of frenetic building all over the world, they seem only to spawn heated arguments for or against.

Then there are those who pooh pooh the subjective judgments related to wire, cable, lids on/off, etc, as the product of feverish, uneducated minds. Why is this so?

Well, it seems to me that EEs are members of a priestly class, and class distinctions must be drawn so that dissenters may be easily be discredited. All professions must do this, not just to promote intellectual 'clarity', but also to protect the incomes of the members. To preserve such elitism it is necessary to create intellectual divisions, and decry the golden eared set who clearly have more money than sense. And yet, only a very few EEs are dedicated audiophiles, and fewer still seek full engagement with beautiful music. Perhaps these guys have the good sense to keep away from such disputes!!

I hold Robert Cordell in the highest esteem because he is prepared to open his mind and listen. Robert will posit the question, why does it sound good? But there are many others here who do not. Consequently, the true audiophile, one who arguably loves his/her music and will spend any amount necessary to achieve 'engagement' with it, is ignored in these weighty prognostications. This encapsulates the entire problem, the lack of a real connection between the ultra technical and the market driven realities. The classic example of this syndrome occured recently when the Dartzeel NHB108 was rubbished mercillessly, presumably without the critics having heard it, entirely on the basic of a conceptual schematic I published on this very list.

John Curl understands all this, and can keep up with the best of the technocrats. Because he has reluctantly accepted, after years of experience, that there are just some design aspects he does not fully understand but implements for performance reasons, he too comes in for trenchant criticism. It seems all techniques must have mathematically precise reasons; yet no one expects such an approach from an instrument maker! This is the same sort of nonsense Nicola Tesla faced at the turn of last century, and I believe most of his heresy has been vindicated.

OK, here's the question.

WHAT IS THE PRECISE MATHS CORRELATION BETWEEN DISTORTION MEASUREMENTS AND LISTENING PLEASURE?

IOW, how do we correlate between low measured distortion, and that which is appreciated by the listener/consumer? And how can we reconcile amplifiers of identical THD which sound quite different?

It is often said that everything in audio electronics has been discovered and well documented in the last sixty years. But the psychoacoustic effects, notably the man/machine interface, are not so well known, and perhaps this additional discipline needs more investigation by those here suitably qualified. There are many who continue to purvey circular, often sneering, arguments to the cowboys who mindlessly assemble circuits of which they understand nothing, yet these cowboys are still touting to the unsuspecting audiophiles out there who clearly need an education.

Until the question in block letters above is answered, this tragic corruption of trade will continue, but I for one would love to know why the odd, unexplained things I do in my circuits sound so damn good.......

Flame suit on.... and enjoying the heat already.....

Hugh
 
AKSA said:
[snip
OK, here's the question.

WHAT IS THE PRECISE MATHS CORRELATION BETWEEN DISTORTION MEASUREMENTS AND LISTENING PLEASURE?

IOW, how do we correlate between low measured distortion, and that which is appreciated by the listener/consumer? And how can we reconcile amplifiers of identical THD which sound quite different?
[snip
Hugh

Hi Hugh,

Amplifiers of identical THD which sound quite different?
Is that so? Please, read this amusing story:

http://www.carveraudio.com/CarverChallenge.pdf

Cheers, Edmond.
 
I got something from my recent experiment. I make an experiment on overcurrent detection (voltage accross RE). With ordinary music with 4ohm speaker, it trips. With test tone to 4ohm dummy load it doesn't. Single pair output transistors.
John Curl suggested about the importance of high current capability of output stage. About speaker impedance dropping somewhere to very low ohm, for very short time. It don't show up with test tone+dummy resistor load.
I narrowed the suspicion about feedback sound to 2 points. 1 is crossover distortion+output current capability of output stage. 2 is about clipping behavior. Square waves (where inherently have high order components involved in global feedback=nasty. The more clipping, the more front end pushes). I really suspect these 2 factors makes the feedback sound. I compared test tone, pure sinusoidal, pure square wave, clipped sinusoidal. The clipped and square wave has exactly the same sound as the fatiguing sound.
Edmonds CMCL is something. You should look at this and the concept behind it further. Exploring clipping problem + good output stage will be the right way in addressing the "transistor", "feedback" sound. ClassD has less of this problem (although it uses transistor and quite alot of feedback) because it basically works with square waves, and the LC filter will turn everything square to sinusoid-like (not square anymore).
 
Interesting article, Edmond, thank you for the reference.......

The man is clearly very bright, and he must have done a lot of tinkering (and listening, as it turns out) to divine how parts/dimensioning/topology influence the sound, or 'the transfer function'. Nowhere, however, does the article actually mention distortion figures per se, and neither does Carver explain their relevance either, regrettably. I guess this knowledge remains in his head......

I note too that the marketers have remained in the ascendant; with the same old mystique selling the same old products - high end has indeed remained the same after all.

But did Bob ever talk about negative feedback, to be or not to be, that is the question?


Cheers

Hugh
 
Hugh,

If you like that kind of articles, try this one:

http://www.bostonaudiosociety.org/bas_speaker/abx_testing2.htm

Not about correlating THD to sound, but about how difficult it actually is to be sure that two different chains actually do sound differently. Which, I posit, is a sine qua non before you start to think about where the diference could come from.

Enjoy your Subday,

Jan Didden
 
Jan,

Wonderful!! I love it!!

This would appear to show that the principal determinant in assessing high end audio gear is prejudice, random chance, and the obsessive desire to tout one's own gear!!

QED, WTF are we all arguing about then? The marketers should just get on with it, with little or no input from the engineering 'truths'.

Hugh
 
Hugh,

I've been reading this thread since it started. I don't contribute because I don't have the technical chops. But perhaps there's something here,

IOW, how do we correlate between low measured distortion, and that which is appreciated by the listener/consumer? And how can we reconcile amplifiers of identical THD which sound quite different?

Think it's the "way" anomalies like THD track the intensity of music signal in and around the amp's noise floor where we're not supposed to hear anything.

Our hearing "clamps" onto anything correlated with musical signal. It can be a gazillion dBs down in the "noise floor." Any noise that's correlated with the music is something our hearing "attaches" to a nearby sound we recognize.

Anomalies can be really quite sneaky. For instance, a violin playing A 440 has its timbre defined by 9 - 10 harmonics plus the sort of unharmonic noise the violin body emits, plus the sound of the attack and effect of the bow sticking and unsticking as it goes across the strings. And given the spectrum's energy budget, it doesn't take much in the way of spuriae for our hearing to modify these various components. Amp A doesn't sound quite like AmpB - not saying anything about better or worse.

If you're playing a piece of music on your system how many dBs down would another piece of music have to be before you wouldn't have a just noticeable difference in the sound? I suspect so far down you could get the bends. But if, down at that level, there is just a very small anomaly in the the amp's output that tracks the music, then it's just a part of the music - it's not necessarily annoying, or noticeable as an artifact it just colours the sound in some way most of us can't quite put our finger on

In the bad old days we weren't disturbed in our enjoyment of music by tape hiss and groove noise because it wasn't correlated to the music. It didn't even quite drive us mad when it just about drowned out the music! We were listening to the music, not the noise. That's the consumer - some are more discriminating about how the music is reproduced than others.

Earl Geddes and Lydia Lee have an AES article on a different metric for rating equipment you might find interesting. THD and IMD don't seem be things that are terribly critical in his subjects' listening experience.

I've been thinking many folk who design audio gear have been leaving really important stuff to psychologists and the hired guns who write codecs..

If you Google psychoacoustics you get 220,000 hits

masking "Psychoacoustics" - 7400
codecs "Pyschoacoustics" - 10500

These people mostly aren't interested in what you do but they surely write about what they've done!

I'm just adding the following because I think it might be the best website I've ever seen.

There's a German guy building violins who probably knows more about how people hear the instrument than anyone who builds audio gear. Go to his site, look up table of contents, go to"our research", click "sound analysis" and then click "Sound analysis examples for bowed string instruments."

He can actually give objective correlates for "warm" and "cold" sound.

http://www.schleske.de/06geigenbauer/en_akustik3schall6beispiele.shtml
 
janneman said:
Hugh,

If you like that kind of articles, try this one:

http://www.bostonaudiosociety.org/bas_speaker/abx_testing2.htm

Not about correlating THD to sound, but about how difficult it actually is to be sure that two different chains actually do sound differently. Which, I posit, is a sine qua non before you start to think about where the diference could come from.

Enjoy your Subday,

Jan Didden


If you guys liked the Bob Carver / Linn tests, your gonna love this:

Very well respected engineer, siting an interesting experience wrt
cables.

http://www.stevehoffman.tv/forums/showthread.php?p=1773290&highlight=directional#post1773290

cheers

Terry
 
AKSA said:
Jan,

Wonderful!! I love it!!

This would appear to show that the principal determinant in assessing high end audio gear is prejudice, random chance, and the obsessive desire to tout one's own gear!!

QED, WTF are we all arguing about then? The marketers should just get on with it, with little or no input from the engineering 'truths'.

Hugh


Hugh,

Its more than that. Don't let me start, but I've been reading about how we form our world view in a general sense, and once you start on that track, you roll from one horror into another. We think that being consious beings helps to be logical, objective and all that. Not so. Conciousness' main goal in life is to protect and safeguard your and my ego. It will twist any truth, fudge any facts, make up any memories, to get to that goal.

Our only chance to get a glimpse of anything objective is to try to circumvent that natural inclination to rig things our way. In science, it is rigurous peer review, repeatability of experiments, what is called 'the scientific method'. And even then, no guarantees.

In audio, we live by anecdotes. We humans are story tellers. We love stories and hate statistics. Everybody loves the story about grandpa who made his neighbours wife pregnant at 94 after a life full of smoking and drinking. Nobody cares about statistics saying that your life expectance goes down by 6 years if you smoke from your 16th birthday.
People read about someone who has glowing anecdotes about his new speakers, and we love it, we want to buy them also.

Look at the reaction of most people to a proposal for DB testing. Knowing what I wrote above, its quite predictable: the stress of testing prevents accurate listening, the switch used to switch between systems hides the differences, the listening sessions are too short, the listening sessions are too long, on and on. At the same time, we unquestioningly accept a report from an unknown person here who, in his listening room of which we know nothing, on amplifiers we never heard about, using music we have no clue about, that changing capacitor A for capacitor B "opens up the sound stage, takes away a veil, changes day into night".
We're story-tellers. We LOVE a good story!

Jan Didden
 
lumanauw said:
I got something from my recent experiment. I make an experiment on overcurrent detection (voltage accross RE). With ordinary music with 4ohm speaker, it trips. With test tone to 4ohm dummy load it doesn't. Single pair output transistors.
John Curl suggested about the importance of high current capability of output stage. About speaker impedance dropping somewhere to very low ohm, for very short time. It don't show up with test tone+dummy resistor load.
I narrowed the suspicion about feedback sound to 2 points. 1 is crossover distortion+output current capability of output stage. 2 is about clipping behavior. Square waves (where inherently have high order components involved in global feedback=nasty. The more clipping, the more front end pushes). I really suspect these 2 factors makes the feedback sound. I compared test tone, pure sinusoidal, pure square wave, clipped sinusoidal. The clipped and square wave has exactly the same sound as the fatiguing sound.
Edmonds CMCL is something. You should look at this and the concept behind it further. Exploring clipping problem + good output stage will be the right way in addressing the "transistor", "feedback" sound. ClassD has less of this problem (although it uses transistor and quite alot of feedback) because it basically works with square waves, and the LC filter will turn everything square to sinusoid-like (not square anymore).

If the protection trips on real load and not on resistive dummy nominal load this comes very often from the phase shift of the load more than nominal impedance drop. It boils down anyway to enough current for the voltage used, not voltage clipping

JPV
 
My $0.02

AKSA said:
The empiricists are dismissed as cowboys, and when an impressive design which has been built and auditioned comes along, like Edmond/Ovidiu's PGP, rather than fostering a rash of frenetic building all over the world, they seem only to spawn heated arguments for or against.

Hugh,

People that glorify the THD numbers only are as inept as those who preach psychoacoustic results only. There are a gazillion of other metrics to evaluate and compare amps. And they are not all independent (e.g. SR is correlated to TIM).

I personally don't have any issues with people listening and liking the sound of whatever amp. My problems start when:

a) Science is replaced by voodoo ("Diamond shaped cristallized silver wires with reflown teflon insulation sound better").
b) Subjective results are extrapolated ("NFB sounds bad").
c) Obtusity replaces open mindness ("A/B blind tests are irrelevant").
d) Marketing replacing scientific facts.

BTW1: I started the PGP amp auditions tests yesterday only. I took copious notes and I'll post as soon as I'll put them together. There are good news and some not so good news and I was focusing mostly on the latter. I was mostly interested in criticism rather than praises. I was able to find a few very nice (although not new) correlations between sound quality and some implementation details. These tests will continue along 2008 and the sites will be periodically updated.

BTW2: I can't speak for others, but certainly I was not expecting a "a rash of frenetic [PGP amp] building". I am aware of a few constructions going on, that's flattering, but ultimately the PGP amp was build because it was fun to.
 
This is what I can see on scope, about the difference between feedback and non global feedback amp.

A is the wanted sinusoidal from an amp reproduction.

B is the waveform from non-global negative feedback amp near rail voltages. It flatten a bit below the supply rail. If the gain raised from this point, it forms trapezoidal and square wave, with normal chopped edges.

C is the behavior of a global feedback amp. No baker clamp or anti clipping installed. Near voltage rails, suddenly there are "sticking", making square waves. Or, it can make D, trapezoidal with extended edges/spikes on edges.

If the gain raised from this point, it makes trapezoidal to square wave with sharp edges (unlike B).

B and D are different. On the flat section (clipped signal), B is natural (only chopped), while D have additional spikes besides chopping.

While in scope viewing it makes a little difference, in FFT analysis (using 192/24 soundcard), the difference is quite large. D exhibit much more high harmonic content than B. I think this is because of global feedback.

If the output stage cannot provide sufficient current (to real loudspeaker load, which demands more current than dummy resistive load, like JPV said above), or have no clipping management involved, the sound from that particular amp will be harsh.
 

Attachments

  • sticking.jpg
    sticking.jpg
    33.5 KB · Views: 445