The Great Cable and Interconnect Swindle: An Etiology

[4est]

I guess a better question is where did the idea come from that feedback is bad? Using work by Nyquist, Harry Black worked out what was needed for amplifier stability with feedback. TIM is bad and again it has been well worked out how to avoid it. Feedback is a non-issue.

I am not sure if I dare say this, but the idea of it being "bad" came from how it sounded, not how it measured. In general, IIRC people are referring to global feedback as bad more so than local feedback. A non issue to you maybe, but often it sounds constrained to the point of castrated IME.

[Jud]

I guess a better question is where did the idea come from that feedback is bad? Using work by Nyquist, Harry Black worked out what was needed for amplifier stability with feedback. TIM is bad and again it has been well worked out how to avoid it. Feedback is a non-issue.

Dennis, since I recall the distinction that others are pointing to re local vs. global feedback being made by designers of amps like Spectral and Pass (and of course the Pass DIY designs are open to close, detailed examination re how much feedback is applied and where), I wonder what the reason for the distinction might be. Perhaps the actual vs. ideal behavior of global feedback, being applied across the entire circuitry of a component, is harder to predict/control than that of local feedback? Don't know if that's the case, since I am not an audio engineer; just throwing it out as speculation in hopes someone knowledgeable will give a more specific, correct answer.

[barrows]

esl: "Feedback is a non-issue."

I suggest that you do the listening test I suggested, Ayre MX-R/VX-R vs Classe, using a few very good recordings of natural music, you might re-asses your opinion. Or perhaps you would not, but it is worth a listen...

[The Computer Audiophile]

Isn't the whole feedback versus no feedback issue just like discussing tube versus solid state? There are several roads to Rome. Feedback ave and no feedback street both get us there.

[esldude]

I am no circuit designer, but here is my take from my best understanding: solid state devices are inherently non-linear in some ways, and global feedback is a way to linearise a circuit. According to the math, there should be no problem to using feedback, but, some believe that global feedback does cause problems in terms of phase which are audible. Now the measurements folks are going to be all over this, 'cause there is very little in measured performance to suggest that global feedback really is bad (but there are some measured downsides).
I think anyone who is interested in this should head to their nearest friendly Ayre dealer, and listen to the MX-R or VX-R amplifiers in comparison to a comparably priced high feedback design (say, Classe for instance). The Ayres achieve very good measured performance without a global feedback loop. See what sounds better to you...

All amplifying devices are inherently non-linear. Feedback was developed for tubes originally.

Yes, I will 'be all over this' I suppose. If you make something and it sounds different, and you attribute this to lower or greater feedback then where is the jump to feedback causes problems with phase? Most of the time this idea has been pitched without much evidence just a supposition that it might be so.

Haven't heard any Ayre amps since the very earliest offerings. Those sounded good, but had audible roll off of the treble due to the low feedback design. They also had distortion that might have been mildly audible with some loads and a higher output impedance rather like a tube amp. As I was using Quads at the time it depressed the treble further due to this. Not bad design by any means, but the reason it sounded different was many of the same reasons tubes sound different. Looking at measurements of his newer designs he has worked around all of this without using global feedback. But I haven't heard one myself to see if it sounds different.

[barrows]

Isn't the whole feedback versus no feedback issue just like discussing tube versus solid state? There are several roads to Rome. Feedback ave and no feedback street both get us there.

That would be one perspective, but, I think the analogy is a little different. Tube amplifiers always measure significantly worse than their solid state brothers. With amplifiers like Ayre's MX-R and VX-R it is proven that solid state amplifiers, with state of the art measured performance, can be built without relying on a global feedback loop, so those who claim feedback is necessary to achieve "proper" results have no basis for that point of view anymore. Whereas, it is pretty easy to assert that all tube amplifiers have distinct "problems" when it comes to measured performance.

[esldude]

I am not sure if I dare say this, but the idea of it being "bad" came from how it sounded, not how it measured. In general, IIRC people are referring to global feedback as bad more so than local feedback. A non issue to you maybe, but often it sounds constrained to the point of castrated IME.

So would you say there are no feedback amps that sound good?

[esldude]

That would be one perspective, but, I think the analogy is a little different. Tube amplifiers always measure significantly worse than their solid state brothers. With amplifiers like Ayre's MX-R and VX-R it is proven that solid state amplifiers, with state of the art measured performance, can be built without relying on a global feedback loop, so those who claim feedback is necessary to achieve "proper" results have no basis for that point of view anymore. Whereas, it is pretty easy to assert that all tube amplifiers have distinct "problems" when it comes to measured performance.

Should we invoke the straw man. In this case, the straw man is those claiming feedback is necessary to achieve proper results. There aren't any of those. Feedback is useful to achieve proper results. No one said it was the only path. Balanced circuitry helps, high voltage devices help, there is more than one way and I don't recall people saying feedback is a necessity. Ayre's latest offerings appear to get there without feedback, their earlier offerings didn't quite get there as they had performance maybe a bit beyond good tube designs.

[esldude]

Dennis, since I recall the distinction that others are pointing to re local vs. global feedback being made by designers of amps like Spectral and Pass (and of course the Pass DIY designs are open to close, detailed examination re how much feedback is applied and where), I wonder what the reason for the distinction might be. Perhaps the actual vs. ideal behavior of global feedback, being applied across the entire circuitry of a component, is harder to predict/control than that of local feedback? Don't know if that's the case, since I am not an audio engineer; just throwing it out as speculation in hopes someone knowledgeable will give a more specific, correct answer.

I think Spectral uses some global feedback though not sure. Pass usually likes to use local feedback. Though not common among high end companies, some EE's have the opinion well done global feedback results in better design than breaking it up into local feedback at each stage. Feedback whether local or global isn't a mystery. The requirements for it to work correctly are well known and understood.

[barrows]

esl: Right on!

"Ayre's latest offerings appear to get there without feedback, their earlier offerings didn't quite get there as they had performance maybe a bit beyond good tube designs."

And this specifically why I referenced the MX-R and VX-R, as I feel they are breakthrough designs which have all the advantages of a high feedback design, with none of the disadvantages.

Now, I look at global feedback loops as a crutch, considering all their disadvantges. To apply high levels of global feedback one needs additional gain, additional gain means more devices, more devices means more noise/distortion and more complexity, to reduce the additional noise we need to add more feedback (the "Halcro" effect), etc... From a technical standpoint the simpler approach of linearising the circuit itself to the point where the additional complexity of feedback is no longer necessary is a much more elegant solution-high levels of feedback seem like bandaid approach in comparison.
As to the sound, I will leave that to those that listen to decide, but it is worth doing the comparisons.

And, please note, I am not including class d in this point of view, as class d is an entirely different subject which relies entirely on high feedback to work at all. In fact, I am loking forward to hearing the nCore in my system in the next few weeks.

[4est]

So would you say there are no feedback amps that sound good?

I did say often, not always, and this was in response to your saying feedback was a "non issue".

[Jud]

I think Spectral uses some global feedback though not sure. Pass usually likes to use local feedback. Though not common among high end companies, some EE's have the opinion well done global feedback results in better design than breaking it up into local feedback at each stage. Feedback whether local or global isn't a mystery. The requirements for it to work correctly are well known and understood.

Seems Keith Johnson thinks use of feedback has disadvantages:

Johnson: A common problem in circuits is "thermal tails." If you shock the circuit with a transient audio event, then the internal parts are stressed. This stress creates heat which remains after the event, which in turn creates a memory in the circuit. It takes a finite amount of time for things to cool off. When the next transient comes along, the circuit behaves differently because it is warmer. This problem is quite audible. The faster the circuit, the less energy stored, and the less the circuit changes from transients that have just occurred.

A similar phenomenon happens with magnetic fields, particularly in power amplifiers. Even though the circuit inside might be very fast, the end result is not speed because you have to wait for the event. The feedback and compensation parts inside amplifiers store charge in a non-linear way when hit with a transient event. The charge then bleeds off after the event, creating a settling problem. You may not hear it as a sound, but you perceive that the sound is not particularly clear.

High speed isn't to get from point A to B fast. It's that you don't want anything to remain once you've gotten there.

(Underlining added.)

[Jud]

Here we go - I think this from Nelson Pass sets the right tone:

Audiophiles seem to revel in minor controversies – vinyl vs CD's, tubes versus solid state,capacitor, wires, magic dots... and negative feedback.

At one extreme, the position is that “feedback makes amplifiers perfect”. At the otherextreme, “feedback is a menacing succubus that sucks the life out of the music, leaving adried husk, devoid of soul”.

The former viewpoint usually belongs to so-called “objectivists” who have a fine appreciationfor electronic theory and measurements. Their opposites would be the “subjectivists” whoemphasize the listening experience and often own tube amplifiers. Accusations areoccasionally made that objectivists can't hear, and conversely that subjectivists hear thingsthat aren't there. This being the entertainment industry, I hope everyone is having a good

time.

Read the entire article here: http://www.passdiy.com/pdf/distortion_feedback.pdf

[esldude]

I did say often, not always, and this was in response to your saying feedback was a "non issue".

This happens over and over again. When I say something is a non-issue that doesn't mean there are never any circumstances in the known universe that it cannot have an effect or make a difference. It does mean, the principles involved are well understood technically, what is needed for something to be used transparently is known, and available technology can use this or do this without it being any kind of problem.

Feedback is just such an issue. People love to pipe up about TIM. Fine, but that has also been understood and figured out. Is it impossible to have TIM cause trouble? Nope, just listen to the first Dynaco amps. But any competent designer in the last 30 or more years knows how to avoid it and design properly for it making it a non-issue. If you have incompetent designers that is another thing. TIM isn't an issue even with bargain basement AV receivers.

So yes, I repeat, feedback is a non-issue.

[4est]

This happens over and over again. When I say something is a non-issue that doesn't mean there are never any circumstances in the known universe that it cannot have an effect or make a difference. It does mean, the principles involved are well understood technically, what is needed for something to be used transparently is known, and available technology can use this or do this without it being any kind of problem.

Feedback is just such an issue. People love to pipe up about TIM. Fine, but that has also been understood and figured out. Is it impossible to have TIM cause trouble? Nope, just listen to the first Dynaco amps. But any competent designer in the last 30 or more years knows how to avoid it and design properly for it making it a non-issue. If you have incompetent designers that is another thing. TIM isn't an issue even with bargain basement AV receivers.

So yes, I repeat, feedback is a non-issue.

Forgive me Dennis, but I disagree. The aforementioned quotes via Jud speak more to the point than I am capable of myself. The terms "non issue" and "transparent" lead one to the impression that there are no trade offs and a clear path forward. That simply is not the case, or the likes of N. Pass would not feel the need to address it whilst making different circuits for differing purposes. I agree with you in general, but not in specific and feel that you are over simplifying it and underestimating the net impact to the (this) end user. I am not qualified to speak much further than this.

[50000]

Forgive me Dennis, but I disagree. The aforementioned quotes via Jud speak more to the point than I am capable of myself. The terms "non issue" and "transparent" lead one to the impression that there are no trade offs and a clear path forward. That simply is not the case, or the likes of N. Pass would not feel the need to address it whilst making different circuits for differing purposes. I agree with you in general, but not in specific and feel that you are over simplifying it and underestimating the net impact to the (this) end user. I am not qualified to speak much further than this.

Pass is quite obviously correct, imo. Feedback creates ugly forms of distortion that replace simpler, lower-order forms. I also like to think the latter sound better.

Negative feedback can reduce the total quantity of distortion, but it adds new components on its own, and tempts the designer to use more cascaded gain stages in search of better numbers, accompanied by greater feedback frequency stability issues.


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.

[esldude]

Forgive me Dennis, but I disagree. The aforementioned quotes via Jud speak more to the point than I am capable of myself. The terms "non issue" and "transparent" lead one to the impression that there are no trade offs and a clear path forward. That simply is not the case, or the likes of N. Pass would not feel the need to address it whilst making different circuits for differing purposes. I agree with you in general, but not in specific and feel that you are over simplifying it and underestimating the net impact to the (this) end user. I am not qualified to speak much further than this.

The stuff in the Nelson Pass link is good stuff. It also reads like a chapter out of a basic electronics textbook. Nothing new or earth shattering or unusual about it. And his examples are just that simplified examples of what can go wrong. Designs abound that take all the normal stuff into account and work fine. Doing the exact same measurements he provides in the link don't turn up amps that show those relatively high and obvious higher order harmonics. Instead you see the first couple barely visible and the rest lost in noise. What Pass says is true, it is basic, and yes it is pretty much a non-issue. The implied step beyond what he says to say high feedback increases higher order harmonics and it doesn't sound good is conjecture with little evidence to support it if we are talking about competently done commercial designs which are plentiful.

[esldude]

Pass is quite obviously correct, imo. Feedback creates ugly forms of distortion that replace simpler, lower-order forms. I also like to think the latter sound better.

All distortion can be ugly. I once used a single transistor pre-amp I made up myself. One active bipolar per channel. The first iteration was truly ugly in sound. It would have had only one harmonic of distortion, but with enough distortion that is ugly and then some. With some effort, improved power supply etc. it sounded okay. Using two biploar transistors per channel, in a Darlington configuration it was better. Sounded much better, measured much better was much better. Still just simple harmonic levels involved. Sometimes extra is better as it accomplishes more.

Pass is not incorrect it just isn't the whole story. His measurements are of simplified examples of how feedback can go wrong. Amps with sometimes lots of feedback don't give results like he shows in most cases. Again those were examples to show what can happen. Simple minded opinions can be misleading. Designers have no incentive to go out and complicate circuitry. They generally do so for good reason.

[4est]

Simple minded opinions can be misleading.

Exactly!

[Paul.Raulerson]

Isn't the whole feedback versus no feedback issue just like discussing tube versus solid state? There are several roads to Rome. Feedback ave and no feedback street both get us there.

Whoo boy, I had no idea the subject was controversial in the least. I was just curious.

Thanks for the great answers guys, I am avidly reading your debate.

-Paul

[50000]

Pass is not incorrect it just isn't the whole story. His measurements are of simplified examples of how feedback can go wrong. Amps with sometimes lots of feedback don't give results like he shows in most cases. Again those were examples to show what can happen. Simple minded opinions can be misleading. Designers have no incentive to go out and complicate circuitry. They generally do so for good reason.

You're bucking the laws of physics to think feedback doesn't create its own sonic effects, if that's what you're saying. Yes, feedback reduces levels of distortion, but it changes the overall distortion signature, of necessity because it introduces forms of distortion not otherwise produced by the circuit if operated without feedback. I meant 'ugly' in a technical sense. For instance, it's well documented, if not broadly understood, that in a smoothly declining 2nd-to-nth harmonic distortion spectrum, sonic effects of odd-order distortions are masked where the previous-higher even-order component is higher in amplitude (higher-amplitude 2nd thus masks lower-amplitude 3rd, etc.). And fwiw, there's considerable evidence the human ear can hear into the noise floor. This makes sense, as noise is generally chaotic. One would expect that the ear can identify an orderly voltage signal appearing in a sea of chaotic noise. We do this all the time.

Designers have several incentives to complicate circuitry. One important incentive is to reduce quoted, on-paper distortion figures, an important marketing concern that, from what I can tell, is driven less by a priority for sonic fidelity than a simple numbers competition.

And you're oversimplifying to think I'm saying feedback is necessarily undesirable. It desirably reduces certain forms of distortion. I'm merely adding that it does so for a price. Some people are unwilling to pay that price. And for good, articulable reasons that cannot be brushed away as unscientific.

[barrows]

esl:

"It does mean, the principles involved are well understood technically, what is needed for something to be used transparently is known, and available technology can use this or do this without it being any kind of problem."

I believe the above statement confuses two different things: "well understood technically" and "used transparently" are not the same. "Technically" generally describes something which can be shown in standard measurements, the concept of "transparency" though, is subjective in nature and requires listening to evaluate.

Again, I suggest a listening test of the Ayre MX-R/VX-R versus a comparably priced, high loop feedback design such as Classe, for a comparison in "transparency".

There is a wonderful thread over at DIYaudio.com dealing with John Curl's CTC, Blowtorch preamplifier (it is very, very long). Many things are discussed, including feedback; Mr. Curl is one of the most experienced audio engineers around, and has designed both high feedback, and no feedback products-good reading for anyone who would like to broaden their knowledge.

[barrows]

"Designers have several incentives to complicate circuitry. One important incentive is to reduce quoted, on-paper distortion figures, an important marketing concern that, from what I can tell, is driven less by a priority for sonic fidelity than a simple numbers competition."

The incentive to use global feedback is great. Especially amongst those designers not clever enough to linearise their circuits without it. This (and the resulting sound quality) is why I am so impressed with the Ayre Reference Series amps.

I am not going to say "feedback is bad", but I do think people should listen to these amps closely in a known context befre they dismiss the possibility out of hand.

[50000]

Also compare listening reports of people making the L'Amp SIT amplifier (L'Amp: A simple SIT Amp - diyAudio) or reviews of Pass' new SIT line of amps. These amplifiers are single-stage, one-active-component amplifiers, and are as simple as an amplifier can be. They thus represent the furthest end of the simple-complex continuum, and are an ideal comparison to complex, feedback amplifiers.

[50000]

The incentive to use global feedback is great. Especially amogst those designers not clever enough to linearise their circuits without it. This (and the resulting sound quality) is why I am so impressed with the Ayre Reference Series amps.

I am not going to say "feedback is bad", but I do think people should listen to these amps closely in a known context befre they dismiss the possibility out of hand.

I'm with you, Barrows. I have tremendous respect for Charles Hansen's design techniques.

I have personally never preferred class A/B feedback amplifiers. I'll be interested to hear the Ncores, as feedback would seem to work differently (have different sonic effects) with class D.

Oh, one other incentive manufacturers have to create complex circuits: complexity can allow greater tolerance for looser-tolerance components. Every hour a designer spends matching output devices is an hour not spent by competition relying on feedback and complexity to absorb component mistolerances.

As to which sounds better, a simpler, zero-(other than naturally occurring)-feedback amp, or a more complex, feedback amp, one can only determine that from listening.