What no-one seems to have mentioned in this discussion is
the subjective weighting of the harmonic distortion. Douglas
Self seems pretty hung up on the total harmonic distortion
(THD) figure, without worrying for a moment whether the
ninth harmonic might be more or less bothersome than the
second harmonic. It seems (I forget the reference) that a
study published decades ago showed that the irritating
effect of a given harmonic went more or less as the square
of its order - that is, the ninth harmonic is nine times as
obnoxious as the third, the eleventh 13 times as nasty, and
so on, even if these higher harmonics are all very much
lower in amplitude than the second or third.
I think that the fallacy in Self's preference for Class B
over Class AB is that music is very different from test
tones. Music has a high "crest factor" - in other words,
most of the "information" in a musical recording is
at the low levels. If you can delay the onset of crossover
distortion until the output is a volt or more, you gain a
significant improvement in low-level detail and "presence"
without the heat penalties of full Class A. You hear many
times that "the first watt is the most important".
The problem with loop feedback is that it pushes the non-
linearities of an amplifier towards higher-order harmonic
content (again, I forget the reference). This is one
possible reason for the more listenable character of single-
ended amplifiers with low feedback, like the SETs and some
of the Pass designs. Crossover distortion, which is
eliminated in a properly-designed Class A output stage, is
intrinsically high-order.
Another issue is the amplitude of the crossover distortion
of a Class B or AB amplifier. Ben Duncan a few months ago
wrote an article for Hi-Fi News showing that MOSFETs had a
lower instantaneaous peak open-loop distortion than a
bipolar output stage, even though the THD and even the
harmonic spectrum of the distortion were comparable. This
may or may not be audible.
Any comments?
Alex
the subjective weighting of the harmonic distortion. Douglas
Self seems pretty hung up on the total harmonic distortion
(THD) figure, without worrying for a moment whether the
ninth harmonic might be more or less bothersome than the
second harmonic. It seems (I forget the reference) that a
study published decades ago showed that the irritating
effect of a given harmonic went more or less as the square
of its order - that is, the ninth harmonic is nine times as
obnoxious as the third, the eleventh 13 times as nasty, and
so on, even if these higher harmonics are all very much
lower in amplitude than the second or third.
I think that the fallacy in Self's preference for Class B
over Class AB is that music is very different from test
tones. Music has a high "crest factor" - in other words,
most of the "information" in a musical recording is
at the low levels. If you can delay the onset of crossover
distortion until the output is a volt or more, you gain a
significant improvement in low-level detail and "presence"
without the heat penalties of full Class A. You hear many
times that "the first watt is the most important".
The problem with loop feedback is that it pushes the non-
linearities of an amplifier towards higher-order harmonic
content (again, I forget the reference). This is one
possible reason for the more listenable character of single-
ended amplifiers with low feedback, like the SETs and some
of the Pass designs. Crossover distortion, which is
eliminated in a properly-designed Class A output stage, is
intrinsically high-order.
Another issue is the amplitude of the crossover distortion
of a Class B or AB amplifier. Ben Duncan a few months ago
wrote an article for Hi-Fi News showing that MOSFETs had a
lower instantaneaous peak open-loop distortion than a
bipolar output stage, even though the THD and even the
harmonic spectrum of the distortion were comparable. This
may or may not be audible.
Any comments?
Alex
Harmonic structure
Alex,
An interesting point, and one I can subscribe to.
One of the reasons early CD players sounded so bad was that they produced increasing distortion at lower levels (as do all digital media), but although levels at 'normal' music levels were still quite low, the harmonic structure extended right to the upper bandwidth limits of the medium.
A phono cartridge will often produce quite high (compared to digital) levels of distortion, but it is low order. When a cartridge mistracks the distortion levels do not rise anywhere near as much as expected, but the harmonic structure extends higher, into areas the ear is more sensitive.
As to the Self designs, although he doesn't specifically state the harmonic structure, we are talking small (fractions of 1 percent) differences in distortion, looking at the results on his website I would be surprised if the harmonic structure of the optimally biased class B is worse than class AB.
One problem I can forsee is that it may be difficult to maintain optimal biasing under all environmental changes, if the amp moves into an under-biased state it will definitely sound worse, whereas by starting in class AB it's unlikely to move into an under-biased state.
Andy.
Alex,
An interesting point, and one I can subscribe to.
One of the reasons early CD players sounded so bad was that they produced increasing distortion at lower levels (as do all digital media), but although levels at 'normal' music levels were still quite low, the harmonic structure extended right to the upper bandwidth limits of the medium.
A phono cartridge will often produce quite high (compared to digital) levels of distortion, but it is low order. When a cartridge mistracks the distortion levels do not rise anywhere near as much as expected, but the harmonic structure extends higher, into areas the ear is more sensitive.
As to the Self designs, although he doesn't specifically state the harmonic structure, we are talking small (fractions of 1 percent) differences in distortion, looking at the results on his website I would be surprised if the harmonic structure of the optimally biased class B is worse than class AB.
One problem I can forsee is that it may be difficult to maintain optimal biasing under all environmental changes, if the amp moves into an under-biased state it will definitely sound worse, whereas by starting in class AB it's unlikely to move into an under-biased state.
Andy.
A philosophical question:
Assume for the moment that there was no heat/efficiency penalty for going pure class A. Why on Earth would anyone go class B, or even AB? It's acknowledged by everyone, Self & Slone included, that class A produces lower distortion right out of the gate.
It is only because of the efficiency penalty that people even have discussions such as this, as otherwise the superiority of class A would relegate other classes to the 'sure it works, but why would anyone want to?' category.
Now for reality. There *is* a heat penalty, and for a given amount of hardware, you're going to get less power out of the circuit, so it's more expensive in parts (and I'm not just talking heat sinks, here). But if you come at it from the admittedly hypothetical stance that I outlined above, justifications for class B start to sound more like rationalizations.
Yes, I'm prejudiced, as I'm sure most folks have figured out by now. I'll take class AB over B, and class A over AB...when I can stand the efficiency penalty. (And no, before someone starts talking about circuit topology vs. gain devices vs. whatever...operating class isn't the only determinant of sound quality.)
Given my druthers, I'll take low order distortion products over high order, any day. Even if there's a bit more of the lower order, as I don't subscribe to the viewpoint that distortion is the only thing effecting sound quality. But the spectral content of the distortion--that's a big component of the problem.
Class B does have advantages for pro sound, where sound quality isn't paramount. Lotsa power in a lightweight package is gratefully accepted by those of us who have lugged Ampeg SVTs one or two too many times. Ditto for PA equipment, etc. But for home use? It depends on which sets of compromises you're willing to make. I pay with heat and efficiency. Someone else might choose to pay that last degree of purity in order to get more power. Life's full of choices.
Grey
Assume for the moment that there was no heat/efficiency penalty for going pure class A. Why on Earth would anyone go class B, or even AB? It's acknowledged by everyone, Self & Slone included, that class A produces lower distortion right out of the gate.
It is only because of the efficiency penalty that people even have discussions such as this, as otherwise the superiority of class A would relegate other classes to the 'sure it works, but why would anyone want to?' category.
Now for reality. There *is* a heat penalty, and for a given amount of hardware, you're going to get less power out of the circuit, so it's more expensive in parts (and I'm not just talking heat sinks, here). But if you come at it from the admittedly hypothetical stance that I outlined above, justifications for class B start to sound more like rationalizations.
Yes, I'm prejudiced, as I'm sure most folks have figured out by now. I'll take class AB over B, and class A over AB...when I can stand the efficiency penalty. (And no, before someone starts talking about circuit topology vs. gain devices vs. whatever...operating class isn't the only determinant of sound quality.)
Given my druthers, I'll take low order distortion products over high order, any day. Even if there's a bit more of the lower order, as I don't subscribe to the viewpoint that distortion is the only thing effecting sound quality. But the spectral content of the distortion--that's a big component of the problem.
Class B does have advantages for pro sound, where sound quality isn't paramount. Lotsa power in a lightweight package is gratefully accepted by those of us who have lugged Ampeg SVTs one or two too many times. Ditto for PA equipment, etc. But for home use? It depends on which sets of compromises you're willing to make. I pay with heat and efficiency. Someone else might choose to pay that last degree of purity in order to get more power. Life's full of choices.
Grey
Is Douglas Self the Pope?
In my opinion Douglas Self has done more to hurt the evolution of audio amplifier design than most.
Most people would think that the known criteria (distortion, frequency response and slew rate) are all that are required to gauge the performance of an amplifier and Mr.Self appeals to their closed minded notions. The printed word, rubbish or not, is pretty powerful.
I put to you, that there are other factors that you have mentioned that are not even considered by Mr.Self. Static measurements do not mean a thing when music takes a dynamic non-repetative form.
After all, we are trying to recreate an illusion and the device that bring us closer to this illusion should be judged as superior regardless of how it measurers.
Maybe oneday we will discover all the criteria that makes an amplifer accurate but till then we should not be closed to different ideas.
I don't mean that what I have said is an outright condemnation of Mr.Self's work, as he has brought a few good ideas to our attention, but he seems to have run into a brick wall regarding newer non-traditional ideas.
Jam
P.S. I wonder if Mr.Self has listened to an amplifier while adjusting it's bias from class B into class A and at the same time resisting the temptation to look at his distortion analyzer.
[Edited by jam on 07-05-2001 at 11:10 PM]
In my opinion Douglas Self has done more to hurt the evolution of audio amplifier design than most.
Most people would think that the known criteria (distortion, frequency response and slew rate) are all that are required to gauge the performance of an amplifier and Mr.Self appeals to their closed minded notions. The printed word, rubbish or not, is pretty powerful.
I put to you, that there are other factors that you have mentioned that are not even considered by Mr.Self. Static measurements do not mean a thing when music takes a dynamic non-repetative form.
After all, we are trying to recreate an illusion and the device that bring us closer to this illusion should be judged as superior regardless of how it measurers.
Maybe oneday we will discover all the criteria that makes an amplifer accurate but till then we should not be closed to different ideas.
I don't mean that what I have said is an outright condemnation of Mr.Self's work, as he has brought a few good ideas to our attention, but he seems to have run into a brick wall regarding newer non-traditional ideas.
Jam
P.S. I wonder if Mr.Self has listened to an amplifier while adjusting it's bias from class B into class A and at the same time resisting the temptation to look at his distortion analyzer.
[Edited by jam on 07-05-2001 at 11:10 PM]
Jam,
You and I are in agreement on this issue. Or, as I'm prone to saying, music is the reference, not meters.
Theory is a wonderful *starting* point. Clearly, one must pay attention to Ohm's law, for instance, or you'll get nowhere. But to use theory as the end point? We'd all still be listening to 70's era amplifiers, as they were already 'perfect' by 'low distortion' standards. But did they sound good?
No.
That's why they're all in landfills across the nation.
I remember selling a line of solid state electronics in the late 70's (Nikko) that claimed three or four zeros in their distortion specs. Yet they sounded ghastly. I owe them a belated thank you for being a part of the wakeup call; distortion specs are virtually meaningless in the real world. (If they meant anything, two amplifiers with identical specs would sound the same, yet they don't. Case closed.)
Steady state measurements are not the answer, although they can point the way to the occasional good question...
Grey
You and I are in agreement on this issue. Or, as I'm prone to saying, music is the reference, not meters.
Theory is a wonderful *starting* point. Clearly, one must pay attention to Ohm's law, for instance, or you'll get nowhere. But to use theory as the end point? We'd all still be listening to 70's era amplifiers, as they were already 'perfect' by 'low distortion' standards. But did they sound good?
No.
That's why they're all in landfills across the nation.
I remember selling a line of solid state electronics in the late 70's (Nikko) that claimed three or four zeros in their distortion specs. Yet they sounded ghastly. I owe them a belated thank you for being a part of the wakeup call; distortion specs are virtually meaningless in the real world. (If they meant anything, two amplifiers with identical specs would sound the same, yet they don't. Case closed.)
Steady state measurements are not the answer, although they can point the way to the occasional good question...
Grey
I just love stirring the pot......
Regarding someone who said that music is different from tones....
Well, music IS tones. Several of them, so if an amplifier's able to faithfully make almost indistinguishable copies of its input at its output, isn't that what you're after? Or do you actually want the amplifier to ADD something to the waveforms that it's amplifying? If so, then get into the valves and other distortion producing stuff. If not, then go for accuracy.
I tend to Slone's argument that, if you want all the distortion added then use external processors that add that distortion in a controllable manner.
I also go along with Baxandall and Hafler in their methods to demonstrate that all competently designed amplifiers do sound the same at the same volume levels.
There are too many variables when people do comparison tests between amplifiers. These variables make the small differences between amplifiers really academic. For example:
1. Small differences in volume have an effect on perceived subjective quality.
2. The environmental conditions (temperature, humidity) also have an effect particularly on speakers.
And on and on. So, how do you know that it's that amplifier producing the differences or is it any of the variables in the equation?
The biggest variable in all of this is our ear/brain combination. That combination is in no way precision (see any good psychoacoustic textbook) and is full of things like the expectancy effect and other phenomena.
Several years ago I went through all the subjective bollocks and then started to really LOOK at the results of what was supposed to be audiophile grade stuff. I found in particular that many so called audiophile designs had high levels of distortion of various types and were marginally stable to boot, with one design oscillating at 100kHz or thereabouts.
Others couldn't drive even a moderate load and so on and so on.
I repeat THERE IS NOTHING SPECIAL ABOUT MUSIC. IT'S JUST ANOTHER WAVEFORM TO BE AMPLIFIED.
The theory of amplification has been well understood for a long time. Why does it take so long for the audio field to catch up?
Ah well, I've rabbited on for long enough. I now open the floor for the replies.
regards,
Keith
Regarding someone who said that music is different from tones....
Well, music IS tones. Several of them, so if an amplifier's able to faithfully make almost indistinguishable copies of its input at its output, isn't that what you're after? Or do you actually want the amplifier to ADD something to the waveforms that it's amplifying? If so, then get into the valves and other distortion producing stuff. If not, then go for accuracy.
I tend to Slone's argument that, if you want all the distortion added then use external processors that add that distortion in a controllable manner.
I also go along with Baxandall and Hafler in their methods to demonstrate that all competently designed amplifiers do sound the same at the same volume levels.
There are too many variables when people do comparison tests between amplifiers. These variables make the small differences between amplifiers really academic. For example:
1. Small differences in volume have an effect on perceived subjective quality.
2. The environmental conditions (temperature, humidity) also have an effect particularly on speakers.
And on and on. So, how do you know that it's that amplifier producing the differences or is it any of the variables in the equation?
The biggest variable in all of this is our ear/brain combination. That combination is in no way precision (see any good psychoacoustic textbook) and is full of things like the expectancy effect and other phenomena.
Several years ago I went through all the subjective bollocks and then started to really LOOK at the results of what was supposed to be audiophile grade stuff. I found in particular that many so called audiophile designs had high levels of distortion of various types and were marginally stable to boot, with one design oscillating at 100kHz or thereabouts.
Others couldn't drive even a moderate load and so on and so on.
I repeat THERE IS NOTHING SPECIAL ABOUT MUSIC. IT'S JUST ANOTHER WAVEFORM TO BE AMPLIFIED.
The theory of amplification has been well understood for a long time. Why does it take so long for the audio field to catch up?
Ah well, I've rabbited on for long enough. I now open the floor for the replies.
regards,
Keith
Harmonics
Keith, you are of course quite right, however precisely amplifying a signal to relatively high power levels is easier said than done. More so when one considers that the precise phase relationship between various frequencies must be preserved.
Harking back to an earlier point about harmonics, I have recently been modelling some of Nelson’s pre-amps was interested to notice a correlation between the incidence of odd order harmonics and what I recall of people’s accounts of their performance. Most significant in my mind was how adding constant current devices dramatically reduces the second harmonic, significant in the output of the Balanced Zen, yet replaces it with odd harmonics which remain significant at high multiples. While I am waiting for my boards to arrive before I can listen to the differences, by some accounts the original BZ sounds “better” than its modified brother.
Cheers,
Pete
Keith, you are of course quite right, however precisely amplifying a signal to relatively high power levels is easier said than done. More so when one considers that the precise phase relationship between various frequencies must be preserved.
Harking back to an earlier point about harmonics, I have recently been modelling some of Nelson’s pre-amps was interested to notice a correlation between the incidence of odd order harmonics and what I recall of people’s accounts of their performance. Most significant in my mind was how adding constant current devices dramatically reduces the second harmonic, significant in the output of the Balanced Zen, yet replaces it with odd harmonics which remain significant at high multiples. While I am waiting for my boards to arrive before I can listen to the differences, by some accounts the original BZ sounds “better” than its modified brother.
Cheers,
Pete
Amplifier Design
>Well, music IS tones. Several of them, so if an >amplifier's able to faithfully make almost >indistinguishable copies of its input at its output, isn't >that what you're after?
Well, maybe larger copies
Actually what I'm after is a musically enjoyable sound, I don't actually care how many zero's are in the distortion figures, just that I can enjoy and be engaged in music.
Many amplifiers whose specifications would imply they can produce exact copies of their inputs do not engage me. They make musicians sound amateur and that is definitely NOT what I want.
The point is a single pure tone bears no relationship to the harmonically related tones produced by musical instruments, hence music is always going to be the ultimate test (assuming you are using the amp to listen to music!).
>I also go along with Baxandall and Hafler in their methods >to demonstrate that all competently designed amplifiers do >sound the same at the same volume levels.
So by implication you are saying that 99% of all audio amplifiers on the market are not competently designed?
Anyone with ears can tell that there are huge differences between most amps, and since all the variables you quote can be controlled in any given comparison this is the only conclusion I can reach.
>The biggest variable in all of this is our ear/brain >combination. That combination is in no way precision (see >any good psychoacoustic textbook) and is full of things >like the expectancy effect and other phenomena.
I wouldn't disagree that the human interface is not a considerable variable, but again it all comes down to your viewpoint.
If one listens to MUSIC, in terms of Pace, Rythm, Pitch and Timing (the fundamental elements of music) one will reach similar conclusions about any given product, even if on some days the same product excites and stirs the soul (or not!) mor than others.
If one listens to Hi-Fi details (Bass / Mids / 'sweet' Treble) etc. you are likely to draw the conclusions you mention.
These concepts are often cited as 'subjective bollocks' by those that don't understand or are not prepared to listen, but frequent demonstrations have proved their worth. I hav many friends who listen to sounds when assessing Hi-Fi, they are constantly fiddling and never happy with their systems. Those colleagues that listen to music do not. They have assessed what they need to satisfy their personal expectations and use the system as it should be used - a means to an end (music).
>The theory of amplification has been well understood for a >long time. Why does it take so long for the audio field to >catch up?
What is often overlooked by so many 'experts' in this field is that audio is one of, if not the, hardest disciplines of all.
One has to design for 10 octaves and 130dB dynamic range at the same time. There is no other area of electronics endeavour where this is necessary.
That's why it's so hard to get it right, and why there are big musical differences between products.
Andy.
>Well, music IS tones. Several of them, so if an >amplifier's able to faithfully make almost >indistinguishable copies of its input at its output, isn't >that what you're after?
Well, maybe larger copies
Actually what I'm after is a musically enjoyable sound, I don't actually care how many zero's are in the distortion figures, just that I can enjoy and be engaged in music.
Many amplifiers whose specifications would imply they can produce exact copies of their inputs do not engage me. They make musicians sound amateur and that is definitely NOT what I want.
The point is a single pure tone bears no relationship to the harmonically related tones produced by musical instruments, hence music is always going to be the ultimate test (assuming you are using the amp to listen to music!).
>I also go along with Baxandall and Hafler in their methods >to demonstrate that all competently designed amplifiers do >sound the same at the same volume levels.
So by implication you are saying that 99% of all audio amplifiers on the market are not competently designed?
Anyone with ears can tell that there are huge differences between most amps, and since all the variables you quote can be controlled in any given comparison this is the only conclusion I can reach.
>The biggest variable in all of this is our ear/brain >combination. That combination is in no way precision (see >any good psychoacoustic textbook) and is full of things >like the expectancy effect and other phenomena.
I wouldn't disagree that the human interface is not a considerable variable, but again it all comes down to your viewpoint.
If one listens to MUSIC, in terms of Pace, Rythm, Pitch and Timing (the fundamental elements of music) one will reach similar conclusions about any given product, even if on some days the same product excites and stirs the soul (or not!) mor than others.
If one listens to Hi-Fi details (Bass / Mids / 'sweet' Treble) etc. you are likely to draw the conclusions you mention.
These concepts are often cited as 'subjective bollocks' by those that don't understand or are not prepared to listen, but frequent demonstrations have proved their worth. I hav many friends who listen to sounds when assessing Hi-Fi, they are constantly fiddling and never happy with their systems. Those colleagues that listen to music do not. They have assessed what they need to satisfy their personal expectations and use the system as it should be used - a means to an end (music).
>The theory of amplification has been well understood for a >long time. Why does it take so long for the audio field to >catch up?
What is often overlooked by so many 'experts' in this field is that audio is one of, if not the, hardest disciplines of all.
One has to design for 10 octaves and 130dB dynamic range at the same time. There is no other area of electronics endeavour where this is necessary.
That's why it's so hard to get it right, and why there are big musical differences between products.
Andy.
Stirring the pot!
Keith,
What you say is correct for the most part but you forget one important thing. Your assumption is based on an amplifier driving a known resistive load.Every real world speaker I know of has a reactive component to it's make up, and all amplifiers are going to react differently to a complex load.
For example, a Yamaha reciever might measure as good or better than a Levinson amplifier on the bench but then hook each up to a pair of Martin Logans and you will see (hear) what I mean.
It is how the amplifier reacts to this load, that for the most part, determines the sound of the amplifier.This is why the Baxandell and Hafler methodology are severely flawed.
Connect an amplifier to a real world load and all bets are off. You cannot solve a three dimensional problem with two dimensional math.
Jam
Keith,
What you say is correct for the most part but you forget one important thing. Your assumption is based on an amplifier driving a known resistive load.Every real world speaker I know of has a reactive component to it's make up, and all amplifiers are going to react differently to a complex load.
For example, a Yamaha reciever might measure as good or better than a Levinson amplifier on the bench but then hook each up to a pair of Martin Logans and you will see (hear) what I mean.
It is how the amplifier reacts to this load, that for the most part, determines the sound of the amplifier.This is why the Baxandell and Hafler methodology are severely flawed.
Connect an amplifier to a real world load and all bets are off. You cannot solve a three dimensional problem with two dimensional math.
Jam
Keep stirring the pot!
Keith,
I will be the first person to defend you when you say that there is too much hype and B.S. in the audio industry. There are far too many con artists manufacturing products out there.
It is easy to say that a Martin Logan is weird load but to many people it is not, maybe a Yamaha reciever is stretching it a bit but lets take two competently designed amplifiers. A Mark Levinson and a Krell, about same power ratings and specs and able to double power into 4ohms and again into 2ohms, yet they sound different driving the same speaker. I have tried this experiment, maybe you should.
Both the above amplifiers should be able to drive almost any speaker but sound different at all power levels into the same speaker. Why?
There are possibly a hundred reasons why they don't. We don't have all the answers.
Trying to fit the cart to the horse is certainly not the answer.
Some posssible questions you might want to ask yourself.
1. How is the feedback loop of the amplifier reacting to the signal?
2.Is there some non-linear phonomea occuring when the amplifier is driven by a non-steady state signal?
3.Is the load causing the amplifier to break into oscillation on peaks?
4.Why does my head hurt when I listen to this amplifier?
5. Is the Earth flat?
6.Is Julian Hirsch alive and well and living with Elvis?
Regards,
Jam
P.S. It is hot as hell out here and I am sweating like a hog.
[Edited by jam on 07-06-2001 at 07:48 PM]
Keith,
I will be the first person to defend you when you say that there is too much hype and B.S. in the audio industry. There are far too many con artists manufacturing products out there.
It is easy to say that a Martin Logan is weird load but to many people it is not, maybe a Yamaha reciever is stretching it a bit but lets take two competently designed amplifiers. A Mark Levinson and a Krell, about same power ratings and specs and able to double power into 4ohms and again into 2ohms, yet they sound different driving the same speaker. I have tried this experiment, maybe you should.
Both the above amplifiers should be able to drive almost any speaker but sound different at all power levels into the same speaker. Why?
There are possibly a hundred reasons why they don't. We don't have all the answers.
Trying to fit the cart to the horse is certainly not the answer.
Some posssible questions you might want to ask yourself.
1. How is the feedback loop of the amplifier reacting to the signal?
2.Is there some non-linear phonomea occuring when the amplifier is driven by a non-steady state signal?
3.Is the load causing the amplifier to break into oscillation on peaks?
4.Why does my head hurt when I listen to this amplifier?
5. Is the Earth flat?
6.Is Julian Hirsch alive and well and living with Elvis?
Regards,
Jam
P.S. It is hot as hell out here and I am sweating like a hog.
[Edited by jam on 07-06-2001 at 07:48 PM]
Keith, ok I think we know we are being “wound up here”, but I’ll play the game one last time. What you say is correct for a stable, sine wave, situation, which is fine if you wish to listen to sine waves all day. Personally I’ve heard enough of them in my career so prefer to listen to music.
It appears we diverge in our opinion of a “competently designed amplifier”, again for me it’s one that produces music in a pleasurable manner. Whilst you may enjoy spec sheets, I much prefer reading about the artist I’m listening at the time.
Which brings us back to a question already asked of you, are you saying the majority of amplifiers are not “competently designed”? Virtually without exception, even a relatively budget Japanese receiver will “out spec” more esoteric examples, such as Pass Lab’s. So by your definition the Japanese units will always sound better, right?
As for cables … not going there. Put it this way I argued BOTH sides of the equation with equal passion. Coming from a technical background I believed cables would make virtually no difference, at least none that could be heard. I finally set up an experiment to prove it … and that’s when I was forced to “change sides”. Rather embarrassingly (some of my arguments were quite forceful), but I could not deny the difference I was hearing. I will confess however that I don’t agree with many of the reasons given as to why differences are heard; there is a lot of misinformation out there in that regard.
While I was once a “spec sheet junkie” (hey I have to justify 6 years at college somehow!!), experience has shown me that this can be only a rough guide to how pleasurable the final product will sound. Since then there has been a lot of listening, experimenting, and comparison; forced to abandon my cherished test equipment, the most telling was done by my UN-trained ears.
Cheers,
Pete
It appears we diverge in our opinion of a “competently designed amplifier”, again for me it’s one that produces music in a pleasurable manner. Whilst you may enjoy spec sheets, I much prefer reading about the artist I’m listening at the time.
Which brings us back to a question already asked of you, are you saying the majority of amplifiers are not “competently designed”? Virtually without exception, even a relatively budget Japanese receiver will “out spec” more esoteric examples, such as Pass Lab’s. So by your definition the Japanese units will always sound better, right?
As for cables … not going there. Put it this way I argued BOTH sides of the equation with equal passion. Coming from a technical background I believed cables would make virtually no difference, at least none that could be heard. I finally set up an experiment to prove it … and that’s when I was forced to “change sides”. Rather embarrassingly (some of my arguments were quite forceful), but I could not deny the difference I was hearing. I will confess however that I don’t agree with many of the reasons given as to why differences are heard; there is a lot of misinformation out there in that regard.
While I was once a “spec sheet junkie” (hey I have to justify 6 years at college somehow!!), experience has shown me that this can be only a rough guide to how pleasurable the final product will sound. Since then there has been a lot of listening, experimenting, and comparison; forced to abandon my cherished test equipment, the most telling was done by my UN-trained ears.
Cheers,
Pete
The real world
At the outset I'd agree with Jam - there is to much BS and pseudo-science in audio circles, but...
/Quote
Frequency ranges of 10 octaves are known and dealt with in the RF world (what about an amplifier with a bandwidth form 2MHz to 2GHz?). Large dynamic ranges are also known (although 130dB is a bit much even for the audio world). How about 70 - 80 dB being more realistic? Instrumentation amplifiers are commonly used over such dynamic ranges.
/Quote end
But I did say at the same time!
As an RF engineer this is a very rare circumstance indeed, even if an amplifier did work over this frequency range, the chance of it maintaining the phase relationship (amongst other things) of the full 10 octave frequency range is nil. The widest bandwidth / amplitude range design I've ever been involved in was a radar altimeter design, it didn't approach the criteria I suggest as necessary for a competent audio amp. To work with 10 octaves, 130dB and maintain phase relationships, along with consistent dynamic behaviour is something not encountered in any other electronics endeavour.
I'd be happy to be proved wrong with specifics, but I've never encountered any similar circumstance in all of my engineering experience.
130dB is not an unrealistic range for real-world audio, I assume by this response you have never tried any live recording of real world sounds. The dynamic range of real life is definitely a realistic 130dB, and an ideal audio system would be able to reproduce it. The large proportion of audio equipment, competently or incompetently designed, would not be able to, owing to cost related design limitations. It is a fair and reasonable goal to aim for though, if our goal is to achieve realistic musical performances.
/Quote
Instrumentation amplifiers are commonly used over such dynamic ranges.
/Quote end
But not at the same time!
It is again very rare for an amplifier, in a lab environment, to be reproducing a complex signals, spanning 10 octaves between 0dB and -130dB, whilst maintaining perfect phase / timing relationships between the signals amplified. That is the difference between music and artificial lab environments. And don't forget we are talking about complex, harmonically related signals here, not pure tones. I'm not for one second suggesting that there is any equipment that can fulfil these roles, but I suggest the better it can, the better it sounds.
Also consider the load under lab conditions - it's unlikely to be a complex dynamically-changing impedance.
/Quote
My ear/brain combination, to me, is NOT part of the initial equation simply because they can be fooled so easily. They are also a reason why so many auditions are so invalid. You go to the audio shop and audition sets of speakers. Did you put each pair in an identical position? Did you set up the amplifier at the same volume for each pair?
/Quote end
You cannot put each 'speaker in an identical position, since the bass alignment, amongst other things, of different 'speakers necessitates they be placed in different positions for optimal performance. All one would achieve by this technique is to find the 'speakers that work best in that position, not the best 'speakers.
My definition of a good amplifer is one that can cope with dynamic impedance variations, within it's design envelope.
Anyone knows that volume level will be more greatly affected by 'speaker sensitivity (which could differ by several dB's) than amplifier power. 3dB change in speaker sensitivity is the same as a doubling of amplifier power. Volume level is only relevant in as much as it can provide necessary power and overload margin for the set of 'speakers in question (which I believe is what you are alluding to).
Of course one has to choose amplifiers that are capable of dealing with the load a 'speaker represents, but ALL 'speakers will have a varying dynamic load. Ignoring absolute volume levels it is still necessary for any given amplifier to behave consistently under any given load, within it's operating envelope.
It is not unreasonable for a low powered amplifier to drive a 'difficult' load, but only to certain amplitude levels. Many amplifiers could not do this, 'good' behaviour under dynamic conditions is essential for all amplifiers not driving purely resistive loads.
I know of no 'speaker that can claim a purely resistive constant load, therefore a competently designed audio amplifier must be capable of dealing with this.
/Quote
My advice to people who are buying equipment then, is to start from the speakers and work back. Once you have the speakers (check that they'll work in your home) and know the load that they put on amplifiers (any speaker manufacturer worth his/her salt should publish such information) you can then determine the characteristics of the amplifier you want to use.
/Quote end
When was the last time you saw an amplifier manufacturer quote dynamic output impedance figures. How often have you seen speaker impedance plots against frequency as part of the manufacturers specifications?
Without these the whole process is fraught with difficulty. How does 'Mr. Average' on the street deteremine whether Amp X or Amp Y can drive his 'speakers. The answer is he cannot, unless he is assured the fundamental design can cope with a range of dynamically changing loads. You can't expect any person in the street to be calculating stability margins etc., that's just unrealistic. The amp has to have certain criteria that are taken as read, the customer chooses on simple, easily understandable variables, such as power / noise etc. The techy stiff has to be right, in most cases it's not!
As for starting with 'speakers that would be a classic mistake that I thought had been rendered faulty years ago. One cannot, under any circumstances, improve the quality of the signal being reproduced. Garbage In, Garbage out as the computer guys still say. Start with a competent source, and compromise further down the chain, depending upon available funds.
/Quote
As far as weird speaker cables are concerned, don't waste my time. I use heavy multistranded vehicle wire for three reasons. One it's reasonably priced. Two, its flexible and three it has low resistance
/Quote end
I'm genuinely pleased these work adequately for you, and would also agree, by implication, that there are many suppliers only too glad to rip-off punters for 'high-end' super cable junk, which is just standard cables in fancy heatshrink etc.
But, to imply that cables have no effect on sound, and that one speaker cable of suitable dimensions / impedance is that same as another is something I cannot subscribe to.
I am a cheapskate and object to spending money where not necessary, but I do use cable designed for my amplifiers, which compared to many is still inexpensive, but sounds better than other similarly dimensioned, cheaper, alternatives.
Don't forget that some amplifier designs omit the common output inductor, in favour of the speaker cable performing this task. This means that the cable must have certain, definable, electrical characteristics for the amplifier to work correctly.
Of course one could add the inductor internally, but it is a lowest common denominator compromise, suited to mass market products, but not if one's goal is the ultimate performance for a given set of circumstances.
Andy.
At the outset I'd agree with Jam - there is to much BS and pseudo-science in audio circles, but...
/Quote
Frequency ranges of 10 octaves are known and dealt with in the RF world (what about an amplifier with a bandwidth form 2MHz to 2GHz?). Large dynamic ranges are also known (although 130dB is a bit much even for the audio world). How about 70 - 80 dB being more realistic? Instrumentation amplifiers are commonly used over such dynamic ranges.
/Quote end
But I did say at the same time!
As an RF engineer this is a very rare circumstance indeed, even if an amplifier did work over this frequency range, the chance of it maintaining the phase relationship (amongst other things) of the full 10 octave frequency range is nil. The widest bandwidth / amplitude range design I've ever been involved in was a radar altimeter design, it didn't approach the criteria I suggest as necessary for a competent audio amp. To work with 10 octaves, 130dB and maintain phase relationships, along with consistent dynamic behaviour is something not encountered in any other electronics endeavour.
I'd be happy to be proved wrong with specifics, but I've never encountered any similar circumstance in all of my engineering experience.
130dB is not an unrealistic range for real-world audio, I assume by this response you have never tried any live recording of real world sounds. The dynamic range of real life is definitely a realistic 130dB, and an ideal audio system would be able to reproduce it. The large proportion of audio equipment, competently or incompetently designed, would not be able to, owing to cost related design limitations. It is a fair and reasonable goal to aim for though, if our goal is to achieve realistic musical performances.
/Quote
Instrumentation amplifiers are commonly used over such dynamic ranges.
/Quote end
But not at the same time!
It is again very rare for an amplifier, in a lab environment, to be reproducing a complex signals, spanning 10 octaves between 0dB and -130dB, whilst maintaining perfect phase / timing relationships between the signals amplified. That is the difference between music and artificial lab environments. And don't forget we are talking about complex, harmonically related signals here, not pure tones. I'm not for one second suggesting that there is any equipment that can fulfil these roles, but I suggest the better it can, the better it sounds.
Also consider the load under lab conditions - it's unlikely to be a complex dynamically-changing impedance.
/Quote
My ear/brain combination, to me, is NOT part of the initial equation simply because they can be fooled so easily. They are also a reason why so many auditions are so invalid. You go to the audio shop and audition sets of speakers. Did you put each pair in an identical position? Did you set up the amplifier at the same volume for each pair?
/Quote end
You cannot put each 'speaker in an identical position, since the bass alignment, amongst other things, of different 'speakers necessitates they be placed in different positions for optimal performance. All one would achieve by this technique is to find the 'speakers that work best in that position, not the best 'speakers.
My definition of a good amplifer is one that can cope with dynamic impedance variations, within it's design envelope.
Anyone knows that volume level will be more greatly affected by 'speaker sensitivity (which could differ by several dB's) than amplifier power. 3dB change in speaker sensitivity is the same as a doubling of amplifier power. Volume level is only relevant in as much as it can provide necessary power and overload margin for the set of 'speakers in question (which I believe is what you are alluding to).
Of course one has to choose amplifiers that are capable of dealing with the load a 'speaker represents, but ALL 'speakers will have a varying dynamic load. Ignoring absolute volume levels it is still necessary for any given amplifier to behave consistently under any given load, within it's operating envelope.
It is not unreasonable for a low powered amplifier to drive a 'difficult' load, but only to certain amplitude levels. Many amplifiers could not do this, 'good' behaviour under dynamic conditions is essential for all amplifiers not driving purely resistive loads.
I know of no 'speaker that can claim a purely resistive constant load, therefore a competently designed audio amplifier must be capable of dealing with this.
/Quote
My advice to people who are buying equipment then, is to start from the speakers and work back. Once you have the speakers (check that they'll work in your home) and know the load that they put on amplifiers (any speaker manufacturer worth his/her salt should publish such information) you can then determine the characteristics of the amplifier you want to use.
/Quote end
When was the last time you saw an amplifier manufacturer quote dynamic output impedance figures. How often have you seen speaker impedance plots against frequency as part of the manufacturers specifications?
Without these the whole process is fraught with difficulty. How does 'Mr. Average' on the street deteremine whether Amp X or Amp Y can drive his 'speakers. The answer is he cannot, unless he is assured the fundamental design can cope with a range of dynamically changing loads. You can't expect any person in the street to be calculating stability margins etc., that's just unrealistic. The amp has to have certain criteria that are taken as read, the customer chooses on simple, easily understandable variables, such as power / noise etc. The techy stiff has to be right, in most cases it's not!
As for starting with 'speakers that would be a classic mistake that I thought had been rendered faulty years ago. One cannot, under any circumstances, improve the quality of the signal being reproduced. Garbage In, Garbage out as the computer guys still say. Start with a competent source, and compromise further down the chain, depending upon available funds.
/Quote
As far as weird speaker cables are concerned, don't waste my time. I use heavy multistranded vehicle wire for three reasons. One it's reasonably priced. Two, its flexible and three it has low resistance
/Quote end
I'm genuinely pleased these work adequately for you, and would also agree, by implication, that there are many suppliers only too glad to rip-off punters for 'high-end' super cable junk, which is just standard cables in fancy heatshrink etc.
But, to imply that cables have no effect on sound, and that one speaker cable of suitable dimensions / impedance is that same as another is something I cannot subscribe to.
I am a cheapskate and object to spending money where not necessary, but I do use cable designed for my amplifiers, which compared to many is still inexpensive, but sounds better than other similarly dimensioned, cheaper, alternatives.
Don't forget that some amplifier designs omit the common output inductor, in favour of the speaker cable performing this task. This means that the cable must have certain, definable, electrical characteristics for the amplifier to work correctly.
Of course one could add the inductor internally, but it is a lowest common denominator compromise, suited to mass market products, but not if one's goal is the ultimate performance for a given set of circumstances.
Andy.
Re: Keep stirring the pot!
jam said:Keith,
/Funny bone removed (except for parts)
> I will be the first person to defend you when you say that
> there is too much hype and B.S. in the audio industry.
> There are far too many con artists manufacturing products
> out there.
> It is easy to say that a Martin Logan is weird load but
> to many people it is not, maybe a Yamaha reciever is
> stretching it a bit but lets take two competently designed
> amplifiers. A Mark Levinson and a Krell, about same power
> ratings and specs and able to double power into 4ohms and
> again into 2ohms, yet they sound different driving the
> same speaker. I have tried this experiment, maybe you
> should.
Fair comment. I have done that. They, to me, sound the same when the speakers are in the same position in the same listening environment set to the same volume, listened to from the same position, using the same material. In other words, everything must be the same except the amplifier. Also, the memory of the exact intonation etc of a piece of music is known to be short, so even the time taken to change out the amplifier is too long for a person to retain exact details of the musical nuances. See any good reference on the ears and the brain. It seems also that you're relying on your ear/brain combination to tell you that there is a difference between two pieces of equipment. You haven't addressed my comments about position or other such like items.
My main beef really is with those in reviews who put what really is their opinion as absolutes. The problem they have is that there is no baseline for them to compare against. If someone says "It is my opinion that.....", that's fine. If they say "The amplifier has wide deep soundstage etc etc" then that's horsepucky. What do they compare it with and how do they/we know?
> Both the above amplifiers should be able to drive almost
> any speaker but sound different at all power levels into
> the same speaker. Why? There are possibly a hundred
> reasons why they don't. We don't have all the answers.
There are many technical reasons that they might not sound the same. Did the specs include the current available from the outputs? At what frequency and power was the distortion measured? And on and on. Further, was the testing done using ABX comparison or under double blind condition?
> Trying to fit the cart to the horse is certainly not the
> answer.
Ah, but it is. You don't make a single horse pull a stagecoach. It might do it for a while but it'll soon become knackered and quit.
> Some posssible questions you might want to ask yourself.
> 1. How is the feedback loop of the amplifier reacting to
> the signal?
In a well designed amplifier, fine thank you. Why all the concentration on feedback? Is it the latest fad to knock it? Everywhere but in audio it's used to great effect to improve linearity and so on. Why is audio so different?
> 2.Is there some non-linear phonomea occuring when the
> amplifier is driven by a non-steady state signal?
This problem has been dealt to in most other fields, including instrumentation and RF, where the signal's dynamic ranges are enormous. Why not audio?
> 3.Is the load causing the amplifier to break into
> oscillation on peaks?
This often happens. If it does, then it's a sign of incompetent design. No problem there and no magic.
> 4.Why does my head hurt when I listen to this amplifier?
Let me see... You had too much to drink last night?
> 5. Is the Earth flat?
Heh! According to the latest proof, yes. There is some dissenting opinion particularly from those devils the scientists, but we all discount a lot of what they say, don't we?
> 6.Is Julian Hirsch alive and well and living with Elvis?
Ah now, I thought Elvis was working at a McDonalds is Des Moines. Come to thisnk of it Julian Hirsch does too doesn't he?
> Regards,
> Jam
> P.S. It is hot as hell out here and I am sweating like a
> hog.
Oh dear.
It's my contention that:
1. Either the things you hear are psychological effects or
2. They can very easily be explained by measurement.
Heh!
regards, Keith
Jeez, go out to dig a hole in the yard for some herbs, and miss a whole bunch of stuff...
Keith,
Honestly, I lost track of all the points I was going to pick apart. Probably just as well, as I'd be putting in a post here that would be long even by my standards.
However...
One or two points do stick in my mind--for starters, if you believe that the human mind/ear combination is incapable of remembering sound, how do you recognize your mother's voice? Don't try the psychoacoustics angle on me, as one of my degrees is in psychology, and I've got a few books on psychoacoustics that I can pull out.
My second point is this: If you don't move the speakers when auditioning amps, how is speaker position a variable? It's easy to set level--measure the level of a 1kHz test tone going into the speaker terminals. It's a lot easier than trying to use a dB meter, and more accurate. You're acting as though there are fifty wildly uncontrolled variables at play. Curious. I only see the one...the amplifier. The rest can be easily controlled.
Three: Double blind studies aren't necessary for audio (although some people will argue this until they're blue in the face, mainly because they don't understand the scientific process). Why? Because all we need to do is satisfy ourselves that amplifiers *do* sound different. That's purely observational. Galileo didn't need to do double blind studies to demonstrate the reality of the moons revolving about Jupiter. All he needed was to open his eyes and look. Science begins (much to some peoples' consternation) with observation. Period. Step one is to demonstrate the fact that amps (and all other pieces of electronics) do sound different. We may not immediately know why. In fact, it may take quite some time before we know why. For instance, it's a fact that cancer exists--but it's clearly premature to say that we understand all the factors involved in causing cancer. To demand that someone come up with a 'why' before the observations can begin distorts the scientific process. Establish factual observations first. Theory comes later.
Interestingly, there's a long history of studies proving that what the subjective people said was true. Audibility of passive components is a classic. The audibility of absolute phase seems to be coming of age these days. Funny. People who actually bothered to *listen* have been saying it was audible for years. But it wasn't "legitimate" until it was "proved" in a study. Did absolute phase suddenly become audible on the day the study was published? No. It was audible all along. But suddenly it was legit, so now Self/Slone et. al. act all pompous, saying,"Well, harrumph, harrumph, uh, yes...er, this is something that needs to be looked into." *Sigh* Needless to say, they didn't admit that that others had been saying it for forever and a day. Perish the thought!
Once upon a time, I felt as you felt; that specs mattered--that they told me something about the sound of a piece of equipment. How wrong I was.
Let me ask you three interrelated questions:
1) What kind of music do you listen to?
2) When's the last time you heard live, unamplified music?
3) How often do you listen to live, unamplified music?
It doesn't take much time at a concert hall listening to classical music before you start figuring out that something is wrong, somewhere. Let me give you a hint--it isn't the concert hall that wrong--it's your stereo. Jazz, bluegrass, folk, and opera will work too. As long as it's unamplified. I didn't realize just how bad my stereo sounded until I started listening to something other than rock, then went to hear classical (in my case) live.
Music, not meters, is the reference.
Grey
Keith,
Honestly, I lost track of all the points I was going to pick apart. Probably just as well, as I'd be putting in a post here that would be long even by my standards.
However...
One or two points do stick in my mind--for starters, if you believe that the human mind/ear combination is incapable of remembering sound, how do you recognize your mother's voice? Don't try the psychoacoustics angle on me, as one of my degrees is in psychology, and I've got a few books on psychoacoustics that I can pull out.
My second point is this: If you don't move the speakers when auditioning amps, how is speaker position a variable? It's easy to set level--measure the level of a 1kHz test tone going into the speaker terminals. It's a lot easier than trying to use a dB meter, and more accurate. You're acting as though there are fifty wildly uncontrolled variables at play. Curious. I only see the one...the amplifier. The rest can be easily controlled.
Three: Double blind studies aren't necessary for audio (although some people will argue this until they're blue in the face, mainly because they don't understand the scientific process). Why? Because all we need to do is satisfy ourselves that amplifiers *do* sound different. That's purely observational. Galileo didn't need to do double blind studies to demonstrate the reality of the moons revolving about Jupiter. All he needed was to open his eyes and look. Science begins (much to some peoples' consternation) with observation. Period. Step one is to demonstrate the fact that amps (and all other pieces of electronics) do sound different. We may not immediately know why. In fact, it may take quite some time before we know why. For instance, it's a fact that cancer exists--but it's clearly premature to say that we understand all the factors involved in causing cancer. To demand that someone come up with a 'why' before the observations can begin distorts the scientific process. Establish factual observations first. Theory comes later.
Interestingly, there's a long history of studies proving that what the subjective people said was true. Audibility of passive components is a classic. The audibility of absolute phase seems to be coming of age these days. Funny. People who actually bothered to *listen* have been saying it was audible for years. But it wasn't "legitimate" until it was "proved" in a study. Did absolute phase suddenly become audible on the day the study was published? No. It was audible all along. But suddenly it was legit, so now Self/Slone et. al. act all pompous, saying,"Well, harrumph, harrumph, uh, yes...er, this is something that needs to be looked into." *Sigh* Needless to say, they didn't admit that that others had been saying it for forever and a day. Perish the thought!
Once upon a time, I felt as you felt; that specs mattered--that they told me something about the sound of a piece of equipment. How wrong I was.
Let me ask you three interrelated questions:
1) What kind of music do you listen to?
2) When's the last time you heard live, unamplified music?
3) How often do you listen to live, unamplified music?
It doesn't take much time at a concert hall listening to classical music before you start figuring out that something is wrong, somewhere. Let me give you a hint--it isn't the concert hall that wrong--it's your stereo. Jazz, bluegrass, folk, and opera will work too. As long as it's unamplified. I didn't realize just how bad my stereo sounded until I started listening to something other than rock, then went to hear classical (in my case) live.
Music, not meters, is the reference.
Grey
More pot stirring!
Keith,
Great comeback.
By your reasoning what is a competently designed amplifier? If you exclude almost everything for some reason or another we are back to the fact that all amplifiers do not sound the same.
I have conducted the experiment as you described and everyone involved was able to tell the difference between the two amplifiers under test.
I would like to know what amplifiers (brands and models) you consider competently designed and what speakers you used as a reference?
Regards,
Jam
P.S. Remember TIM distortion that Matti Ottola discovered (or invented), that was supposed to be the definitive test for amplifier performance, has fallen by the wayside. I still maintain we have not discovered all the measurement criteria that an amplifier should be judged by.
Grey,
Where were you when we needed you?
[Edited by jam on 07-06-2001 at 10:22 PM]
Keith,
Great comeback.
By your reasoning what is a competently designed amplifier? If you exclude almost everything for some reason or another we are back to the fact that all amplifiers do not sound the same.
I have conducted the experiment as you described and everyone involved was able to tell the difference between the two amplifiers under test.
I would like to know what amplifiers (brands and models) you consider competently designed and what speakers you used as a reference?
Regards,
Jam
P.S. Remember TIM distortion that Matti Ottola discovered (or invented), that was supposed to be the definitive test for amplifier performance, has fallen by the wayside. I still maintain we have not discovered all the measurement criteria that an amplifier should be judged by.
Grey,
Where were you when we needed you?
[Edited by jam on 07-06-2001 at 10:22 PM]
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.