If I might chime in here.....
There has been a lot of animosity between the various camps of high end audio over the years. Much of it is academic grandstanding, but essentially the argument that amps should be 'straight wires with gain' is entirely plausible and has justly spurred many very clever people, Edmond Stuart being one of them, to create ever better amps with lower and lower THD. This is a very worthwhile intellectual challenge in my view and I salute their efforts.
However, amongst the non-technical, there is an abiding love of tube amps, which from a THD viewpoint are very poor, with lashes of H2, H3 and H4 - and maybe beyond. Yet again and again they are chosen by the cognoscenti, who value the subjective listening experience above all else.
Many people, Geddes et al, have attempted to correlate measured distortion, notably THD, with the subjective listening experience. This is fraught with difficulty, however, because tastes differ from one listener to another, mood changes, the music is often different and uncontrolled, etc. It is, of course, very difficult to be objective about the subjective, but that is no reason to decry the double blind test, which like wine tasting, often reveals some curious contradictions.
Perhaps the question that should be asked, since the technology is now well understood, is this: 'What is it about human hearing that prefers highly colored, highly distorted sound reproduction?' If we can answer this question, then we might have some clue as to the psychoacoustic cues which drive the subjective listening experience, at least over a large sample.
Finally, Tief raises a very interesting technology, the Lavardin IT amplifier, which addresses the hitherto little known 'memory distortion'. Is this amp so wonderful? If yes, why? If no, why not? Is this technology worth a look.
My 2c: Much work needs to be done in the area of distortion profile. Jean Hiraga identified the linear reducing harmonic profile in the late sixties; yet few have taken this up. I have found it works, and such profiles do sound good. With simulation and careful measurement we can break down the various harmonics, so this conundrum could well be solved quite soon. Furthermore, is there a quality of global negative feedback which is inimical to the ear? If so, and many anecdotal reports would indicate there is something to it, what exactly is this phenomenon? And if a zero gfb amp sounds better, is there an explanation?
The smart money is on people like Lavardin, Charles Hansen, Nelson Pass - the list goes on. I look forward to the day someone can draw a strong correlation with high validity between a double blind listening test and the published specs. I'll wager they are not THD....
Cheers,
Hugh
There has been a lot of animosity between the various camps of high end audio over the years. Much of it is academic grandstanding, but essentially the argument that amps should be 'straight wires with gain' is entirely plausible and has justly spurred many very clever people, Edmond Stuart being one of them, to create ever better amps with lower and lower THD. This is a very worthwhile intellectual challenge in my view and I salute their efforts.
However, amongst the non-technical, there is an abiding love of tube amps, which from a THD viewpoint are very poor, with lashes of H2, H3 and H4 - and maybe beyond. Yet again and again they are chosen by the cognoscenti, who value the subjective listening experience above all else.
Many people, Geddes et al, have attempted to correlate measured distortion, notably THD, with the subjective listening experience. This is fraught with difficulty, however, because tastes differ from one listener to another, mood changes, the music is often different and uncontrolled, etc. It is, of course, very difficult to be objective about the subjective, but that is no reason to decry the double blind test, which like wine tasting, often reveals some curious contradictions.
Perhaps the question that should be asked, since the technology is now well understood, is this: 'What is it about human hearing that prefers highly colored, highly distorted sound reproduction?' If we can answer this question, then we might have some clue as to the psychoacoustic cues which drive the subjective listening experience, at least over a large sample.
Finally, Tief raises a very interesting technology, the Lavardin IT amplifier, which addresses the hitherto little known 'memory distortion'. Is this amp so wonderful? If yes, why? If no, why not? Is this technology worth a look.
My 2c: Much work needs to be done in the area of distortion profile. Jean Hiraga identified the linear reducing harmonic profile in the late sixties; yet few have taken this up. I have found it works, and such profiles do sound good. With simulation and careful measurement we can break down the various harmonics, so this conundrum could well be solved quite soon. Furthermore, is there a quality of global negative feedback which is inimical to the ear? If so, and many anecdotal reports would indicate there is something to it, what exactly is this phenomenon? And if a zero gfb amp sounds better, is there an explanation?
The smart money is on people like Lavardin, Charles Hansen, Nelson Pass - the list goes on. I look forward to the day someone can draw a strong correlation with high validity between a double blind listening test and the published specs. I'll wager they are not THD....
Cheers,
Hugh
Last edited:
I am shure that good correlation with subjective listening impressions will be given by a complex metric, in which THD contributes with around one percent weight or similar.
Experienced audiophiles pay most attention to such sound features, like "transparent", "opened", "live-like", "involving" sound. I propose to look for objective scientific grounds for these terms in the information theory, not in simple analog measurements.
I would substitute all the above mentioned sound characteristics by one term - "information rich" sound. Live audio is plenty of information, especially at very low dB levels, and our hearing is capable to deal with it. The wider information steam comes from audio system, that overlaps with information stream from live performance, the better reproduction quality is (for audiophiles).
So, we must look at how systems handles most information reach components of audio signal, rather than how it will affect 1-bit information sine wave at maximum amplitudes. Human audio perception seems to be "trained" by evolution in recognizing and fixing the facts that definite tiny audio effects do happen, even at noisy surroundings, and we possibly able to extract useful info even from bottoms of the audio system noise floor.
Experienced audiophiles pay most attention to such sound features, like "transparent", "opened", "live-like", "involving" sound. I propose to look for objective scientific grounds for these terms in the information theory, not in simple analog measurements.
I would substitute all the above mentioned sound characteristics by one term - "information rich" sound. Live audio is plenty of information, especially at very low dB levels, and our hearing is capable to deal with it. The wider information steam comes from audio system, that overlaps with information stream from live performance, the better reproduction quality is (for audiophiles).
So, we must look at how systems handles most information reach components of audio signal, rather than how it will affect 1-bit information sine wave at maximum amplitudes. Human audio perception seems to be "trained" by evolution in recognizing and fixing the facts that definite tiny audio effects do happen, even at noisy surroundings, and we possibly able to extract useful info even from bottoms of the audio system noise floor.
Last edited:
Some text from that site:
"The time alignment of an amplified music signal and its distortion components has a profound effect on perceived sound quality. Nakamichi researchers discovered that the human ear is much more tolerant of harmonic distortion if the distortion components are time-aligned with respect to the primary signal. Nakamichi Harmonic Time Alignment (HTA) amplifiers adopt a wideband, low open-loop gain design. A minimal amount of negative feedback is used, but, more important, it is kept constant over the entire audio spectrum. This assures the proper timing between the primary signal and any amplifier distortion components."
Seems reasonable, from information point of view. While harmonics are proper time-aligned, human hearing does not recognize them as independent events, wrong information is not added. This is quite measurable thing.
However, some hardly measurable kind of distortions steel remain, like those from parts quality, connectors, cables etc.
I doubt it's the complete answer but something worth keeping in mind. High OLG amplifiers that rely heavily on compensation are generally more prone to sound harsh & fatiguing. Strangely, lowering thd from already very low levels
seems to improve them.
Hugh Buddy, about two years or more ago, you wrote me some paragraphs in your upper-class English that basically says the same as described here.
The problem I believe is not necessarily the order of the artefact that the amplifier generates, but to how this artefact correlates with the same order artefact that the instrument produced, it will interfere in some unpredicted way (which can be measured) and clearly heard.
If the amplifier produced harmonic autocorrelates with that from the signal it plays, the sound will be the best regardless of the level of the harmonic.
Now one must be very aware that the signal being played consists of a fundamental and say some harmonics. The amplifier amplifies this exact fundamentals and adds its own harmonics onto the actual natural harmonics. Besides the amplifier amplify the natural harmonics being reproduced and adds new harmonics to these that actually never existed before and what we are listening to, and that is were things go horribly wrong and caused mainly by interference.
funny thing about this distortion in amplifiers, were talking low low levels, and yet we use paper-cone and horn-loaded systems to get the most life out of the recordings....there distortion is measured more than a 100 times higher sometimes even a 1000 times higher.....and yet the music is swinging with life.
Iam quite confident that what we measure is not what we hear, amplifiers are far more different than the small distortion numbers suggests....This ability to play live or live like is for the few...and absolutely founded somewhere else.
Iam quite confident that what we measure is not what we hear, amplifiers are far more different than the small distortion numbers suggests....This ability to play live or live like is for the few...and absolutely founded somewhere else.
Nico,
I don't recall that, your memory is better than mine. I have wondered a lot about the phase alignment of the artefacts as they come out of LTSpice, that's definitely related.
I found myself chuckling out loud to Vladimir's '...with around 1 percent weight or similar'. Absolutely hilarious, Vladimir, I agree with you emphatically, whooping from the stalls!
MB, I agree, but we should not expect the amp to detract from the original recording either, and given that some amps, notably some tube amps, seem somehow to 'improve' the sound, rather than assuming they are doing it by corrupting the original recording, is not also valid to assume that other, lesser amps might well be detracting somehow from the original recording and 'damaging' the music?
I've heard this argument again and again, often accompanied by tub thumping and purist notions of straight wire with gain, but ideology won't solve it. Careful listening and lots of experimentation and measurement is the answer. Examine VladimirK's preamp. It is open loop throughout, uses very carefully selected semiconductors, and was designed to improve upon the best tube preamps. There is little attempt to reduce low order harmonic distortion, too - that should be a signal lesson (sic!) for all of us......
I do think Mike is right, I think the best amps have the right topology and a few good caps in them.
Cheers,
Hugh
I don't recall that, your memory is better than mine. I have wondered a lot about the phase alignment of the artefacts as they come out of LTSpice, that's definitely related.
I found myself chuckling out loud to Vladimir's '...with around 1 percent weight or similar'. Absolutely hilarious, Vladimir, I agree with you emphatically, whooping from the stalls!
MB, I agree, but we should not expect the amp to detract from the original recording either, and given that some amps, notably some tube amps, seem somehow to 'improve' the sound, rather than assuming they are doing it by corrupting the original recording, is not also valid to assume that other, lesser amps might well be detracting somehow from the original recording and 'damaging' the music?
I've heard this argument again and again, often accompanied by tub thumping and purist notions of straight wire with gain, but ideology won't solve it. Careful listening and lots of experimentation and measurement is the answer. Examine VladimirK's preamp. It is open loop throughout, uses very carefully selected semiconductors, and was designed to improve upon the best tube preamps. There is little attempt to reduce low order harmonic distortion, too - that should be a signal lesson (sic!) for all of us......
I do think Mike is right, I think the best amps have the right topology and a few good caps in them.
Cheers,
Hugh
Last edited:
Yes, that is a really good sounding peace of gear, but in that thread, there was only one person who revealed some interest to the schematics, and his interest was limited mainly to "what harmonics spectrum shall we get".
In education of audio engineers, they are trained good for dealing with public address systems. High-End aspects are not accounted for, as usual - no market - no care. The fate of Nakamichi is a typical example, one among many others.
Vladimir,
As it happens I agree with all of your comments, your design decisions, and your conclusions. I am not so versed as you in the engineering maths, but like you I share a conviction that psychoacoustics is the key.
You unselfishly shared your design, but you should not assume that because you had little response people took no attention. I noted it very carefully, but have nothing to add to an already polished, accomplished and impressive body of work. I'm confident other designers here studied it with interest. It is largely how I would have done it, and indeed I'm working on a preamp right now which shares many concepts.
But there is no money in it, just as you say, look at Nakamichi, which was a byword of innovation and excellence when I was a young man. When the marketers believe their own hype, you know an industry is in serious trouble.....
And I'm not so sure PA is a suitable arena for modern audio designers, and furthermore, drawing attention to this situation is probably counter productive too. Perhaps I'd better ease off before I rouse the 'monster in the river'.
Cheers,
Hugh
As it happens I agree with all of your comments, your design decisions, and your conclusions. I am not so versed as you in the engineering maths, but like you I share a conviction that psychoacoustics is the key.
You unselfishly shared your design, but you should not assume that because you had little response people took no attention. I noted it very carefully, but have nothing to add to an already polished, accomplished and impressive body of work. I'm confident other designers here studied it with interest. It is largely how I would have done it, and indeed I'm working on a preamp right now which shares many concepts.
But there is no money in it, just as you say, look at Nakamichi, which was a byword of innovation and excellence when I was a young man. When the marketers believe their own hype, you know an industry is in serious trouble.....
And I'm not so sure PA is a suitable arena for modern audio designers, and furthermore, drawing attention to this situation is probably counter productive too. Perhaps I'd better ease off before I rouse the 'monster in the river'.
Cheers,
Hugh
Last edited:
Vladimir,
If I would try to summarize your ideas in a single word (dangerous, I know!), would 'transparent' be the word?
And then I mean it in the basic sense: add nothing to the input signal, take nothing away from the input signal.
Would you agree to that goal?
jan
I would proceed from an axiom, that an amp ALWAYS affects audio signal (even a single resistor does affect, how the whole amp can not do that?).
So, one must formulate, what will be the less among various devils. The solely THD and sine wave distortions spectrum (they say very small part of the whole story) is not the target to be addressed at first turn.
What I propose, first to formulate, what parts of audio signal curve present the most important information (in audiophile sence), and design a measurement setups for this goal. I believe, the most important information is low-level one, that near the noise floor and below it. These kind of signals not only distorted, they most frequently become partially or completely shaded, or lost, at amp's output. Important to understand, what are the conditions when such an information becomes "lost" for hearing. Most simple case is the not sufficient slew rate of an amp, when reproduction of sharp and high amplitude transients shade low-level information. But other mechanisms do exist.
By analysing low-level information, we distinguish live audio from reproduced one.
Therefore, I find it reasonable to consider an amp as "transmission channel". If it able to transmit everything, what is considered as an information by hearing, the better it is.
Last edited:
I agree that it is very important. Whether it is the most important, I don't know.
But it is certainly an important area to start. And I am convinced that whatever we find to be of importance there, we will be able to measure or else be able to develop a measurement method/equipment. Technology is very seldom the limitation.
I think here the limitation is to find the limits to our hearing and these limits seem to vary depending on circumstances and conditions. I don't think it can be done without some massive listening tests program, which will be a limitation. Another option is a program to identify and analyse documented listening tests that have been done over the years and try to extract a generalised datum. But I don't see anyone doing that sort of thing soon. Especially since there is no money in it
jan
Janneman, very reasonable things you are saying.
Somewhere I have written, that if high-end audio would be as important as, lets say, researches in nuclear energy or information technology, complex and almost final amps testing techniques would be already developed.
In my present vision, this technique will be like testing of special kinds of transmission lines, with special kinds of test signals. Test signal could be just extremely precisely digitized non-periodic fragment of intense symphonic orchestra sound, and highest quality audio computer card calculates difference between input and output fragments according to well elaborated metrics (the last will be the major problem, I guess). And linear distortions, related to temporal shifts etc., will play much more important role than nowdays.
Somewhere I have written, that if high-end audio would be as important as, lets say, researches in nuclear energy or information technology, complex and almost final amps testing techniques would be already developed.
In my present vision, this technique will be like testing of special kinds of transmission lines, with special kinds of test signals. Test signal could be just extremely precisely digitized non-periodic fragment of intense symphonic orchestra sound, and highest quality audio computer card calculates difference between input and output fragments according to well elaborated metrics (the last will be the major problem, I guess). And linear distortions, related to temporal shifts etc., will play much more important role than nowdays.
Last edited:
When $800 speaker cables are "reasonably priced" theres money in it. The problem is that sound quality has become fashion, based only on personal preferance (often of some reviewer in the manufacturers pocket). Most people (even the audiophiles who can hear)) dont want wire with gain, they like colour. Your type of testing would give us a bell curve with a wide skirt, that showed what colour most people want, not whats most transparent. DBT almost always test peoples tastes (the ones that dont are the ones that only ask if there is a differance, but these are very limiting). Look at the questions they ask.
Which sounds better? Which sounds cleaner, more transparent, etc.
They are all subjective answers.
This statement seems to be a root of very long lasting wrong beliefs. Nobody likes colour. And if people do not like SS amps with close to zero THD (measured at high level simplest sine test signal) they claim that THD is not the most important property. We must educate from this situation and create new measurement methods.
Even the relativity theory seems to be not lasting forever, why somebody must be sure that THD and other measurements with simple periodic signals are the last words in the history of audio?
You see, that's where I think quality equipment comes in to provide that function - the ability to color in sound through means of equalizers. There you'll have different camps as well - analog vs digital, though if you want to stretch it into more complex sound processing, a DSP would be a fashion. But, for some reason, this kind of hardware conflicts with an ingrained dicipline with audiophiles, which is, that less is usually better, less components for the signal to traverse through.
I think what is important is that we sort of want there to be a difference between amps in sound quality. If there would be no difference, there would be little reason to discuss amps sound quality wise. "A wire with gain" does not allow for difference to exist and as such, is not a very appealing vision for those that want to be able to distinguish between sound quality in the broadest sense, including natural harmonic distortion of tubes. On a same note, it's what a community like this forum allows to exist, amps are a great way to create something that would carry your personal 'stamp', amps clearly are distinguishable from each other by sound.
I don't see why the two camps could not coexist; both with a mutual understanding and respect. The tho camps are basically those that chase the straight wire with gain, while others chase the uniqueness and personal stamps of designers that comes with various circuits.
That my 2 FETs
Hello
I agree that low-level information are very important.
I will say that there is also the ability of an amp to deal with complex musical signals, most IM thd tests are done with just two signals, that's not enough. And the amp need to deal with complex load, and let not forget the importance of the harmonic profile of the amp thd.
Bye
Gaetan
I agree that low-level information are very important.
I will say that there is also the ability of an amp to deal with complex musical signals, most IM thd tests are done with just two signals, that's not enough. And the amp need to deal with complex load, and let not forget the importance of the harmonic profile of the amp thd.
Bye
Gaetan
Hello
It's not alway that they like colour, yes many amps will ad some colors but some of them will sound much more alive and engaging than many of those "wire with gain" amps.
So, of two evils one must choose the lesser, and many will chose an amp with color because that this amp won't kill the life coming from a good recording.
Bye
Gaetan
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.