|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
72.52.96.59
In Reply to: RE: Do they? posted by RPMac on January 29, 2015 at 07:11:49
I built the classic Pass amp a few years ago. Was not all that impressed. Then I tried a class B amp design running hard in class A, similar to the ESP designs. Much better.
I know the "less is better" idea is very popular in high performance audio. The SET is a great example. But I have always found in my work with audio amps that the extra components some feel compromise a design really do offer an improvement.
Few will argue a current sink in the tail of a tube diff amp phase splitter makes a considerable improvement in distortion. The reasonthis was not done in 1960 was technology. But still I remember there was one tube amp that did use a tube current sink.
Follow Ups:
"I know the "less is better" idea is very popular in high performance audio. The SET is a great example."
It is popular to characterize SET as less but that is only as far as the actual amplification circuit. The power supplies in the best are quite complex and of course there has to be far greater care in the choice of passive parts. Some go to the extreme of making their own caps, transformers and resistors...so less circuit but more in terms of what goes into the amp itself.
That being the idea that an SET has more critical component quality needs than a PP amp? A bad cap is a bad cap for example.Are you saying the SET amp being simpler electrically is more transparent and therefore more does not mask the failings of poor components? That is a popular audiophile theory btw.
But it doesn't make any sense. A good PP amp has much lower measured distortion than an SET. So it would seem the PP amp would be far more critical of bad components than an SET.
This old idea that the simpler an active circuit the less it colors the signal is flawed audiophile logic. All active electronic components and some passive have non-linear operating areas. Using additional components to correct that non-linearity improves the signal fidelity and adds less distortion to the path. There's a balance though. Noise for example is an enemy here.
It is the advance of electronics knowledge that got us to more complex amplifier circuits. And what drove that advancement is the quest for better and better quality. So IMO, this idea that the pinnacle of amplifier design was the 1930 SET is ludicrous.
How many recording studio are using SET amps or the solid state single transistor version? I know there might be one or two small shops but lets face it, this is long dead technology in the professional world.
Edits: 01/29/15
"But it doesn't make any sense. A good PP amp has much lower measured distortion than an SET. So it would seem the PP amp would be far more critical of bad components than an SET.
"
You have to ask yourself WHY does a PP amp have much lower distortion...it is only through the use of a lot of feedback. Remove the loop and you will get MUCH higher distortion...I don't think this aspect has been properly explored as to it's sonic implications...although Cheever did some work in this regard.
"This old idea that the simpler an active circuit the less it colors the signal is flawed audiophile logic. All active electronic components and some passive have non-linear operating areas. Using additional components to correct that non-linearity improves the signal fidelity and adds less distortion to the path"
This is simply false. It is only by wrapping it in feedback that you get that result. A multistage amplifier without feedback will have very high distortion. Adding more non-linear components gets a more and more non-linear system...to think otherwise is silly.
"How many recording studio are using SET amps or the solid state single transistor version? "
Irrelevant because they are drinking the orthodoxy coolaid.
A push-pull amp without feedback does not have inherently more distortion than a single-ended amplifier. In fact, it has much less.
A multi-stage amplifier needn't have excessively high levels of distortion. Really, you only need as many stages as required to get the desired gain. In this respect, push-pull amps are no different than single-ended amps.
The phase splitter is an added complication in a push-pull amp, but a good phase splitter need not add significant distortion.
-Henry
"A push-pull amp without feedback does not have inherently more distortion than a single-ended amplifier. In fact, it has much less."
For a Class AB amp this is not true...for a Class A amp it could be true.
"A multi-stage amplifier needn't have excessively high levels of distortion. Really, you only need as many stages as required to get the desired gain. In this respect, push-pull amps are no different than single-ended amps.
The phase splitter is an added complication in a push-pull amp, but a good phase splitter need not add significant distortion."
You can make plenty of gain with a single stage...no need for multiple stages...unless you want to use a lot of feedback.
Phase splitters are notoriously difficult to make with low distortion.
Well, I don't want to get into a technical pissing contest, but there's nothing inherent about Class AB push-pull that makes it high-distortion. Granted, you tend to get higher-order kinks in the transfer function, but it's all a matter of choosing your operating point carefully. The distortion spectrum is very different with push-pull, and you may very well prefer the sound of single-ended (or push-pull Class A). But on an absolute magnitude scale, you need a hell of a lot of crossover distortion to make up for the reduction in even-order distortion that push-pull affords you. Somewhere in my files I have a couple of articles on this.Whether or not you get enough gain from a single stage depends on the sensitivity of your output tubes and how much power you want to get out for a given input voltage. I like my amps to have 30dB of gain, which coincidentally corresponds to a voltage gain of 30. This is borderline attainable in a zero-feedback design, depending on what you're trying to achieve.
With respect to phase splitters, the current-sourced long-tail pair is pretty damned good. IIRC, Allen Wright had a zero-feedback 300B push-pull design with a cascade differential driver, and I assume it was very good. I've seen other push-pull zero-feedback amp designs as well.
I think the general consensus is that tube distortion is benign enough that you can pass a signal through several stages without issue. Most phono preamps are open-loop and run three stages, yet still sound good if designed properly.
Overall, I have to agree with the statement made here recently by someone else, that the biggest problem with negative feedback is instability. My experience with tube amps having modest amounts of feedback is that they sound like crap if they aren't stable, but absolutely amazing once the loop is stabilized properly.
I really don't see this "orthodoxy" thing. All tube amp designs are orthodox (or at least the ones that get talked about here). I don't have a personal beef with you so don't get me wrong. I just think, in a way, that feedback-bashing has kind of become the new hi-fi orthodoxy. Feedback is a tool, and one that most people really don't understand.
-Henry
Edits: 01/31/15
I agree there is no reason to take a piss on the forum...I assure I am having a serious debate here because I while I am well aware of what different circuits do technically to a circuit I am also pretty convinced that most engineers don't really know what the sonic consequences are of a pursuit of technical perfection.
"But on an absolute magnitude scale, you need a hell of a lot of crossover distortion to make up for the reduction in even-order distortion that push-pull affords you. Somewhere in my files I have a couple of articles on this."
That is the crux of the problem...you don't need much crossover distortion to sound very objectionable...and it gets worse, relatively as the level goes down. The perception problem is strongly non-linear and it even seems that changing the distortion from monotonic to strongly odd order is also a perceptual no no. D.E.L Shorter proposed n(squared)/4 (where n is the harmonic order) as a weighting factor for the perceptual affect...later studies found that was not severe enough...read Cheever for a good overview on the history or the papers outlining the GedLee number by Earl Geddes.
"I like my amps to have 30dB of gain, which coincidentally corresponds to a voltage gain of 30. This is borderline attainable in a zero-feedback design, depending on what you're trying to achieve."
I find that most amps have too much gain...but feedback is not a good way to reduce it IMO.
"Allen Wright had a zero-feedback 300B push-pull design with a cascade differential driver, and I assume it was very good. I've seen other push-pull zero-feedback amp designs as well."
That amp was pretty speaker dependent. When the speaker was right though it was WHOA good.
"Overall, I have to agree with the statement made here recently by someone else, that the biggest problem with negative feedback is instability. My experience with tube amps having modest amounts of feedback is that they sound like crap if they aren't stable, but absolutely amazing once the loop is stabilized properly."
Well having had a couple of amps that had adjustable feedback I can say without a shadow of a doubt that they sounded better with the feedback switch to the minimum. Especially my VAC 30/30, which completely shrunk in sound even going from zero to the min feedback (2db I think)...going higher did even more damage to the sound. Now, maybe it was becoming unstable but there was no indication of this. I have yet to hear feedback make a real sound quality improvement.
"I really don't see this "orthodoxy" thing. All tube amp designs are orthodox (or at least the ones that get talked about here). I don't have a personal beef with you so don't get me wrong. I just think, in a way, that feedback-bashing has kind of become the new hi-fi orthodoxy. Feedback is a tool, and one that most people really don't understand."
It is not for me either. I do this in the purest scientific way...I first make observations...lots and lots of them...then start to ask "why do I prefer the sounds of these over these". Then start to read literature about what has been done to correlate sound quality and design...etc.
Thanks for the reply. I think we're on the same page here.
I have experienced expansion of the soundstage with decrease in feedback, but to me this was an artifact of added distortion, not an improvement due to reduced distortion. But, of course, this wasn't in an amp designed for low open-loop distortion in the first place.
To my ears, good solid-state amps are basically perfect (though it's hard to know for sure because there are so many unaccounted-for variables, and what is "perfect" sound anyway?). I hear differences between amps, and they are damned hard to explain with basic measurements. I believe the most complex and poorly-understood component in the chain is the listener. It doesn't follow at all that a "perfect" amplifier necessarily equates to "happiest listener."
I think from a system design perspective, the simplest approach is to build an amplifier that has the best possible measured performance, based on a suite of appropriate measurements; then go looking elsewhere to tailor the sound to your preferences. I think the "perfect" amplifier problem has been solved for all intents and purposes. It doesn't mean an "imperfect" amplifier won't sound better under some circumstances.
The audiophile belief that a good ear is in infallible measuring instrument strikes me as plain ass-backwards. It's a very complicated question.
Happily (for me), I've backed away from hard-core audiophile values after spending many frustrating years at it. Overall I'm happier now, but that's just my compromise, and I certainly respect people who want to keep drilling down in their quest for "The Absolute Sound..."
..As long as they don't tell me I'm a jerk for holding fast to certain beliefs about the physical nature of things that, so far, are working out very well for me in other areas of my life.
Regards,
Henry
"I have experienced expansion of the soundstage with decrease in feedback, but to me this was an artifact of added distortion, "
This is a pure (and incorrect) assumption on your part. In fact it has been noted that high order harmonics are associated with perceived loudness. What this means is that when these harmonics are present the sense of depth is reduced (as well as 3d imaging) because high frequencies, which are crucial to depth perception are exaggerated. This is well known from psychoacoustic studies.
"To my ears, good solid-state amps are basically perfect"
Not even remotely...read on for links discussing measurements and perception.
"I hear differences between amps, and they are damned hard to explain with basic measurements"
Exactly, which is in contradiction to your previous statement.
"It doesn't follow at all that a "perfect" amplifier necessarily equates to "happiest listener.""
Since all this stuff is only to make happy listeners I find this statement to be silly at best and truly missing the point at worst.
"I think from a system design perspective, the simplest approach is to build an amplifier that has the best possible measured performance, "
Simplest perhaps, the right way doubtful...at least with current tools at hand.
"based on a suite of appropriate measurements;"
It is not about appropriate measurements per se, it is using them along with psychoacoustic data and then CHANGING the design to get measurements that meet the right requirements. This is fundamentally different than designing as low as possible THD and IMD in the hopes that it sounds better...this has been the misguided way for the last 60 years. Read the links below where THD and IMD actually have weak NEGATIVE correlations with sound quality.
" I think the "perfect" amplifier problem has been solved for all intents and purposes"
Obviously this is your opinion and not at all anything like a consensus. It is clear from research that until a truly linear amplification device is invented then a perfect amplifier is far from available. These kinds of statements seriously make me doubt your understanding of just how complex the problem of perception and correlation to design truly is.
"The audiophile belief that a good ear is in infallible measuring instrument strikes me as plain ass-backwards. It's a very complicated question."
It is complicated that I will heartily agree. I don't think anyone is claiming an infallible ear but in the end, the messy and contradictory human is the final arbiter...not the scope. This means that there WILL be contradictions. The machine can only tell what is there but not its impact...I will say this over and over until it sinks in.
"As long as they don't tell me I'm a jerk for holding fast to certain beliefs about the physical nature of things that"
I don't think that but I think you will benefit from stepping out from your paradigm a bit and thinking about it more. Rather than saying "there is no way humans can hear that", instead approach it that they might be able to and if they can then there are serious implications with regard to different amplifier designs.
In the end, it is a hobby to enjoy music and I only take this seriously in as far as I like to know WHY things affect other things. I still sit down and enjoy music like most on this forum.
Now, here are some links...read them and tell me what you think.
http://gedlee.azurewebsites.net/Papers/THD_.pdf
http://gedlee.azurewebsites.net/Papers/Distortion_AES_I.pdf
http://gedlee.azurewebsites.net/Papers/Distortion_AES_II.pdf
http://www.dancheever.com/main/cheever_thesis_final.pdf
There is much more information out there (from Pass, Bernd Gottinger, Otala, Norman Crowhurst and others) but these summarize a lot of what is out there.
I think maybe you are taking this a bit more seriously than I am.
I kind of resent the fact that you seem to be calling me an idiot, but really, it's not that important to me. So if we have
to agree that I'm an idiot to avoid having a fight, I guess I'm cool with it.
I responded to your initial post just to comment on your claim that a multi-stage amplifier without feedback will have
very high distortion. I disagree with that. Everything after that seems to be sliding down the slippery slope into a
technical-argument-slash-male-chest-beating contest in which I am a only very reluctant participant.
The Geddes-Lee papers talk about metrics that better correlate between objective measures of distortion and listener
perceptions of distortion. This is good work. What I don't see in your citations is any data regarding the audibility of
distortions in state-of-the-art solid-state amplifiers. I agree that traditional THD and SMPTE IMD measurements are
pretty poor. Borbely talked about this in detail in his series of articles on a three-tone IMD analyzer published years
ago in "Audio Amateur." I personally believe that if you can engineer an amplifier that has vanishingly low static and
dynamic distortions, wide frequency response, low noise, low susceptibility to EMI, is stable, has a very low output
impedance, and flawless time-domain response, then you'd be very hard-pressed to find anything about it to fault,
listening-wise.
My anecdotal comments about what I've heard are scientifically meaningless and were not intended to be taken as
scientific generalizations of fact. I don't have the patience to go into all this in detail. If we were to get together to
put some of these ideas to a test, I think we're both smart enough that we would agree on the protocols and on the
resulting findings.
I'm particularly miffed that you dismiss my comment that a "perfect" amplifier may not the the one that makes listeners
the happiest. I suppose if you equate perfection and happiness then my statement is nonsensical. But I do believe
in the idea of euphonic distortions. Perception is complex enough, and the whole two-channel stereo concept is so
flawed anyway that it makes perfect sense to me that mathematically "perfect" amplification may not be preferable
to some listeners.
I'm surprised that you cite the Cheever paper. This is a master's thesis, not serious science. I've read it before and
I think it's embarrassing fluff. I always laugh at the very first statement: "Humans respond emotionally to complex
musical messages which contain no real survival value." He is so clearly talking out of his ass, this is so obviously
refutable, that it's hard to take anything seriously that follows in the paper. I would get into a more detailed critique
of the thesis, but to what end?
Finally, falling back again to subjective perception, when I try to listen seriously to a stereo system, I ask myself
whether or not the system gives me a seemingly transparent window on the music. I listen for grain or noise
"between" the notes. I listen for anything unpleasant or painful especially in the high frequencies. I try to assess
whether or not the experience is subjectively "involving." I can honestly say that good solid-state amps, including
the 100W MOSFET amp I built a few years ago using boards from a DiyAudio group buy, float my boat very nicely.
I hear what I want to hear, and I have nothing to complain about.
Anyway, I'd really like to get out of this argument. As I said, I have no real negative feelings toward you, even though
some of your comments have kind of struck me the wrong way. I'm disappointed that you have come away with a
negative impression of me, but as I said, I can live with it.
Go in peace.
-Henry
First, I don't think you are an idiot. Now that we have that out of the way I would like to address a few points...
It is clear that you are coming from an engineer's point of view regarding design, testing, specs etc. That's ok, but you have to think about something (something that I have to think about all the time in my line of work so maybe why I dwell on it with audio) are these specifications meaningfully describing how the product behaves for it's INTENDED purpose. If they are not for that purpose then at best they are defining some sort of minimum quality control on the basic function of a product but not how it actually performs.
"multi-stage amplifier without feedback will have very high distortion. I disagree with that. "
unless it is pure Class A it is mostly true. Take the feedback loop off any given amp with 0.001% or whatever distortion and see what you end up measuring...it might even be unstable, which is essentially 100% distortion.
"What I don't see in your citations is any data regarding the audibility of distortions in state-of-the-art solid-state amplifiers."
The types of distortions mentioned by Geddes mathematically are of the sort that will always be present to some degree in many push/pull feedback amps that are not pure Class A so specific examples are not really necessary. I can give you lots of links to measurements of what are considered SOTA solid state amps and the distortion patterns you will see are consistent with the mechanisms described by Geddes. There are many push/pull tube amps, running Class AB and with feedback that have the same issues even if it won't sound quite the same (Crowhurst writing in the 50s was talking about tube amps after all).
I find these two statements from you particularly contradictory:
"I agree that traditional THD and SMPTE IMD measurements are pretty poor."
"I personally believe that if you can engineer an amplifier that has vanishingly low static and
dynamic distortions, wide frequency response, low noise, low susceptibility to EMI, is stable, has a very low output impedance, and flawless time-domain response, then you'd be very hard-pressed to find anything about it to fault, listening-wise."
And yet that is exactly what Geddes found out...in fact there was a slightly NEGATIVE (but marginally insignificant...at least at 95% confidence) correlation! Now, how could this be that flawless measuring amps sound at best no better and at worst worse? Well, based on my studies it is the WAY they are "reducing" the distortion...it looks great with static tests but the reality is something different. Crowhurst basically showed that you are mostly just pushing the distortion around to different harmonics and making a multitude of distortion components such that it actually resembles a noise floor. Pass has shown that distortion "concentrates" into particular frequencies that can be many times higher than the static distortion levels (see his paper on the First Watt website). You can believe what you want but it doesn't seem to hold up to scrutiny.
"But I do believe in the idea of euphonic distortions. "
Keith Howard basically put this old chestnut to rest by adding distortion of particular patterns to a digital recording. He explored many variations and found the following: ALL degraded the sound as compared to the unadulterated track. So there was no addition that was "Euphonic" all distortion added was bad; however, he also found that the monotonic distortion pattern (even and odd in a exponential decay in level with increasing order) was the least damaging of those he tried. Since there is no such thing as a distortion free electronic device this is an important finding, IMO. I have downloaded his software (he made it freely available) and done this myself with a recording I found to be revealing and got about the same results as Keith.
So, I DO NOT believe in euphonic distortion as it has been suitably proven to me that all distortion is damaging and it is a question of which is worse and which is better and level is only a small part of that (as would be obvious from the Gedlee number, Cheever's analysis and/or Shorter's equations for weighting harmonics).
"This is a master's thesis, not serious science. "
WHAT!!?? How is a Master's thesis not serious science?? I have a Ph.D. in Analytical Chemistry so is my Ph.D. Thesis also not serious science?? It is in fact one of the definitions by which I get to call myself a doctor of Chemistry and a scientist. The only difference in a master's thesis and a Ph.D. thesis is the scope of the work...a master's thesis is based on a more narrowly focused line of research and may have more direct support from the supervisor. I would be really curious about how you arrive at such a preposterous claim.
"He is so clearly talking out of his ass, this is so obviously refutable, that it's hard to take anything seriously that follows in the paper. "
A guy makes one questionable comment and you dismiss the whole work...seems pretty harsh to me...a lot of great men would have been discredited by this reasoning. His review is quite good and his conclusions match well with Geddes and others.
"Finally, falling back again to subjective perception, when I try to listen seriously to a stereo system, I ask myself whether or not the system gives me a seemingly transparent window on the music. I listen for grain or noise "between" the notes. I listen for anything unpleasant or painful especially in the high frequencies. I try to assess whether or not the experience is subjectively "involving." I can honestly say that good solid-state amps, including the 100W MOSFET amp I built a few years ago using boards from a DiyAudio group buy, float my boat very nicely. I hear what I want to hear, and I have nothing to complain about."
All fine with me and if that is what you hear it is what you hear...I would probably find something to be dissatisfied with in the sound of those designs...I have heard too many that didn't really work to be confident in the outcome.
I would like to respond, and we could go on for a few days, I think. I have a bunch of what I believe to be compelling rebuttals to the points you have made. I also think you make some good points of your own and would like to acknowledge them.
Unfortunately, I am just having trouble working up the motivation for the discussion. I'm really sorry because you and I could evidently have a pretty substantial debate here, and that would be a rare thing on this forum.
It's really my fault, and I beg your indulgence, but all of a sudden I'm thinking of the hours, days, weeks, and months I used spend in this kind of activity. And I realize I'm happier when I just let it go.
Again, please accept my apologies. Hopefully someone else will come along and keep this discussion going. I would be interested to read the follow-ups.
-Henry
"This is simply false. It is only by wrapping it in feedback that you get that result. A multistage amplifier without feedback will have very high distortion. Adding more non-linear components gets a more and more non-linear system...to think otherwise is silly."Well is adding a feedback circuit not the same as "adding additional components to correct a non-linearity"? It sure is!
This negative view of feedback is spread by these audiophile rags written by sales people. Any trained EE knows the power of feedback if applied properly and that extends to many other engineering disciplines as well. Cruise control, autopilot, these too are feedback systems and the exact same rules of math and physics apply. I know, I studied servo design!
The whole feedback smear campaign actually started with Matti Otala in the mid 1970s. But the contributions of his work are largely unknown and/or misunderstood by the audiophile community. Most audiophiles jump on the old time travel paradox where you can't correct a signal that has already passed through the amplifier. And that is quite correct and a real problem with wide band amplifier design like analog video.
But the time travel problem is totally irrelevant to audio amplifiers!
Simply put, no musical instrument can produce a rise time fast enough in our 15psi atmosphere to induce the popular audio amp feedback problem myth. An audio signal is many magnitudes slower than the propagation delay of the amplifier circuit. This whole leading edge distortion theory with audio amp feedback simply does not occur with music. You can easily produce the fault with a simple square wave generator but real music, speech, and sound effects cannot.
What Matti Otala really did was to bring about the focus on amplifier stability. While the feedback loop can correct non-linearity very well in an audio amplifier, if not properly compensated it will enhance instability. And it is/was that ringing and aggravated instability that produced the unpleasant 70s transistor sound as well as the destructive runaway oscillation which destroyed may amplifiers and speakers. Early solid state designs piled on huge amounts of feedback because they were finally freed from the limitation of the tube OPT phase shift. Yes they overlooked the stability issue in greater detail.
Feedback in audio amplifiers is a godsend if properly managed. And today due to the contributions of enginerrs like Matti Otala we now know how to do that.
Edits: 01/30/15 01/30/15 01/30/15
Gusser, when you say class B amp, do you mean an amp running in class B or AB? In either case, how do you run a class B or AB in class A?
Thanks.
A class B amp modified for class A bias.
Maybe I have a different definition of class B. The ESP site lists the NAD 3020, Cyrus 1 and Sugden A48-II as class B amps but from my perspective these are all class AB as the output devices don't completely switch off at the zero crossing as in a typical class B design. Is that how you interpret it or have I missed something?
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: