Stereophile | January 2017 Issue

[opus112]

You have to read the paper to realize the point of disagreement.

This looks to be a baseless claim.

In a nutshell the paper says psychoacoustics supports a natural order and levels of distortion which tube amps have and SS doesn't. And hence the SET Tube amp is transparent.

To me this talk about 'harmonic profiles' is Cargo Cult-ery and I've no desire to go there. I've already disagreed with morricab on it so its uninteresting to me unless there's some real science behind it.

Clearly SET amp with high output impedance is not transparent into many loads as we are discussing.

Agreed.

And there is no basis for the psychoacoustics conclusions he is drawing. The goal here should be to get the distortions below audibility, period. Then it doesn't matter what their order is.

You're saying 'distortions' are the only issue in terms of SQ? In which case seems to me you're barking up the wrong tree, the issue AFAIA is noise. And its modulation with signal.

 

[opus112]

To show that you were attempting to say that it is typical for headphones to have impedances of hundreds of ohms which is not correct.

But there was no such attempt.

 

[morricab]

That is not what makes the audibly different. Impedance can be high or low. What makes a SET amp an equalizer is the *variations* in impedance, not its nominal marketing number. Here is the author himself again:

"The output impedance of these amplifiers is in the ohms range, hundreds of times worse
than solid-state push-pull amplifiers. This requires careful mating with loudspeakers that
do not have great impedance variation with frequency. "

So what I am saying is 100% the same as what he is saying. I welcome you to find speakers with flat line impedance curve.

Just like THD we do NOT want to use single number impedance values.


How would you prove it?

If the impedance on a speaker is above 8ohms then variations in impedance will have negligible impact on FR. Use 30+Ohm headphones if you must...there are some that can hook directly to the amp output. You will here differences easily between SS and SS, which you are claiming most have distortion products below audibility unless clipping....nonsense but you said it.

Even if pure tone harmonics are in audible the IM products and modulated noise floor are still there. Also, there are many soundstage phenomena, imaging etc that have nothing to do with FR variations that are also obvious to hear. Tone can be affected by variations in FR, I can agree with this.

 

[morricab]

Sure. When more than the ear is used to evaluate audio gear, anything goes. And rarely do those opinions agree with each other as evidenced by this thread.

Also, there can be secondary reasons SET amps sound different than SS+resistor, one of which could be it soft clipping. Or having higher distortion to being with.

You are ignoring that you introduced that the distortion is inaudible for two equally low distortion low impedance amps but no one here but you believes that the two amps will sound identical.

 

[morricab]

This looks to be a baseless claim.



To me this talk about 'harmonic profiles' is Cargo Cult-ery and I've no desire to go there. I've already disagreed with morricab on it so its uninteresting to me unless there's some real science behind it.



Agreed.



You're saying 'distortions' are the only issue in terms of SQ? In which case seems to me you're barking up the wrong tree, the issue AFAIA is noise. And its modulation with signal.

There is nothing "cargo cult" like coming from the BBC. Shorter realized over 60 years ago that proper weighting of the influence of harmonics in distortion was far more important than the total amount. This is true whether it is pure tone harmonics or IMD.

I once made an excel sheet with data from Soundstage and weighted it with Shorter's equation...it was intersting which amps rose to the top and sank to the bottom. I also compensated for the power output that the measurment was done by using the THD vs. power plot. I also looked at different frequencies (assuming the nature of the distortion stayed the same...not always a good assumption but had no choice) because many amps vary the distortion with Frequency. Weighting up the harmonics makes a big difference in the outcome and I am sure it would be far from an R^2 of 1 with a listening test but I bet it would have a positive correlation with Shorter corrected distortion figures.

Geddes published two conference papers on the subject and found no correlation between THD or IMD as a bulk number and sound quality with a panel of listeners. Once his metric was applied then he had a pretty decent fit. His metric though is not user friendly for the data available on the internet.

So, you all can dismiss Cheever if you like but there is other evidence out there that the pattern of the distortion trumps the amount and a much better metric for predicting general sound quality (listener preference of course) is needed. IMO, Cheever just took things a bit farther in trying to define the SLOPE of the exponential decay (Shorter's equation is also an exponential just of a somewhat shallower slope) by using the ear/brains own harmonics...maybe its not right but it has a compelling logic to it. He also rightly made it SPL dependent. That is the main thrust of his work not so much the actual testing.

Jean Hiraga also noted this kind of pattern in amps he found to be the best. While his opinion only, he was one of the most respected hifi critics and he had strong observational skills and some design chops as well.

Nelson Pass has shown in his white paper how distortions, while measuring low averaged over the whole spectrum, can build up in ceratin frequencies of quite high values and this was confirmed to some extent by Boyk and Sussmann through simulation.

What I can say is that I am a bit shocked that Amir is coming on here and basically saying that all SS amps below clipping will sound the same because their distortion (according to him) is below audibility (how he has confirmed this I do not know). Nevermind the differences one will get between tube and SS with an appropriate, playing field leveling, load.

 

[morricab]

"And while some people will argue that even order distortion (second, fourth, sixth harmonics, etc.) is more pleasing than than odd order distortion (third, fifth, seventh harmonics, etc.), this isn't quite correct. If odd ordered harmonics were so objectionable, we simply wouldn't be able to bear listening to most musical instruments.

It is not an issue of even order versus odd order, but rather high order versus low order (seventh, eighth and ninth harmonics, etc. versus second, third and fourth harmonics, etc.). This is because as the harmonics get further and further away from the fundamental, the more sensitive we are to perceiving them as second tones, or in the case of a piece of audio equipment, distortion. And this is what led Norman Crowhurst and others in the 1950s to suggest weighting distortion products depending on how far they are from the fundamental.


So the Holy Grail for distortion is twofold. First, they must be low order (second, third, fourth) and second their levels should reduce monotonically from the fundamental, i.e. third lower than second, fourth lower than third, etc. SET amps can achieve this in good part due to their single-ended nature and helps explain why they can sound so good even if their THD figure might suggest otherwise. But the less than optimal air-gapped output transformer takes its toll and produces significant amounts of high order distortion products.

The 535 however admirably achieves this goal. Figure 2 shows the 535's distortion spectra measured while driving a 32 ohm load (an actual headphone in this case and not just a resistor) at nearly full output swing (the headphones were incredibly loud, and that was with them sitting on the test bench).

As you can see, the 535 only exhibits second, third and forth order distortion products and they drop rapidly and monotonically. "

 

[morricab]

http://www.dolphin-hsl.com/articles/distortion.html

"If the THD is several percent, but it is all h2, you probably won't notice it at all. It is this type of distortion that low-u triodes tend to produce. On the other hand, even very small amounts of high order harmonics can be all but unlistenable"

"The least harmful is h2. Since the second harmonic is musically correlated, being one octave higher, it tends to make for a "richer" sound. Excessive amounts tend towards a "darker" sound, as has been the complaint against the very nonlinear 12AV7. From a design point of view, this would not look like too much of a problem, since even order harmonics are nulled with balanced circuitry. As for h3, this, too, is musically correlated, being nearly an octave-and-a-half. The effect of h3 is to lend a sense of "detail". More h3 lends "edge" or "brightness" to the sound. For some types of music, this can be desirable. As for higher order harmonics, these aren't musically correlated, and sound very dissonant and unpleasant. The higher order the harmonic, the less of it you can tolerate"

"Fifty or so years ago, Norman Crowhurst proposed a weighting system of distortion measurement that would account for this: undervaluing the less destructive low order harmonics, and emphasizing the higher order harmonics. Simple THD measurements don't do this. Of course, the industry wanted nothing to do with it."

"All IMD is quite bad since its frequency components are rarely, if ever, musically correlated. This is the reason to stay away from nonlinear devices and nonlinear operating points. "

"The simplistic answer is to add as much NFB as you can manage while maintaining a reasonable level of stability. Unfortunately, it's not that easy. Over fifty years ago, Norman Crowhurst looked into the problem, and discovered that the main effect of NFB is the reduction of h2 and h3. It doesn't do much, if anything, for removing low levels of higher order harmonics. NFB can make the situation even worse by filling the noise floor with lots of low level, high order harmonics. This can, and does, ruin any pretense to sound stage. "

 

[amirm]

You are ignoring that you introduced that the distortion is inaudible for two equally low distortion low impedance amps but no one here but you believes that the two amps will sound identical.

I am not here to discuss anyone's belief. Heaven knows that would be a useless exercise . Instead I was asked to comment on validity of the paper you have put forward. That paper at least superficially relies on two factors:

1. A single blind ABX test showing preference for a SET amplifier over solid state.

2. Psychoacoustic evidence that fundamentally SET amplifiers are transparent and solid state are not.

If we stick to the plot, then anyone who says amplifiers sound different needs to show #1. I don't think there are any takers on that . All I see are sighted evaluations that don't rely on ear alone for their conclusions. Mind you, amplifiers may sound different but again, that is another discussion altogether.

On the second one I have shown evidence that masking which the author himself acknowledges shows that negative feedback based solid state amplifiers generate distortions that are masked by it.

These are the facts. That one or many think or have some idea about amplifiers sounding different is the old tired "I am right you are wrong" which doesn't interest me or is within the scope of what you put forward.

If we are done with the paper not proving its point, then we have made progress. If we are not done, then let's see that evidence based on premise of #1 and #2.

 

[amirm]

http://www.dolphin-hsl.com/articles/distortion.html

"If the THD is several percent, but it is all h2, you probably won't notice it at all. It is this type of distortion that low-u triodes tend to produce. On the other hand, even very small amounts of high order harmonics can be all but unlistenable"

"The least harmful is h2. Since the second harmonic is musically correlated, being one octave higher, it tends to make for a "richer" sound. Excessive amounts tend towards a "darker" sound, as has been the complaint against the very nonlinear 12AV7. From a design point of view, this would not look like too much of a problem, since even order harmonics are nulled with balanced circuitry. As for h3, this, too, is musically correlated, being nearly an octave-and-a-half. The effect of h3 is to lend a sense of "detail". More h3 lends "edge" or "brightness" to the sound. For some types of music, this can be desirable. As for higher order harmonics, these aren't musically correlated, and sound very dissonant and unpleasant. The higher order the harmonic, the less of it you can tolerate"

"Fifty or so years ago, Norman Crowhurst proposed a weighting system of distortion measurement that would account for this: undervaluing the less destructive low order harmonics, and emphasizing the higher order harmonics. Simple THD measurements don't do this. Of course, the industry wanted nothing to do with it."

"All IMD is quite bad since its frequency components are rarely, if ever, musically correlated. This is the reason to stay away from nonlinear devices and nonlinear operating points. "

"The simplistic answer is to add as much NFB as you can manage while maintaining a reasonable level of stability. Unfortunately, it's not that easy. Over fifty years ago, Norman Crowhurst looked into the problem, and discovered that the main effect of NFB is the reduction of h2 and h3. It doesn't do much, if anything, for removing low levels of higher order harmonics. NFB can make the situation even worse by filling the noise floor with lots of low level, high order harmonics. This can, and does, ruin any pretense to sound stage. "

I have seen no one that is defending THD being a metric of fidelity. Why are we/you still trying to prove that? Subjectivists could care less about it and from objectivity point of view, it runs foul of psychoacoustics due to shape of critical bands.

 

[amirm]

"And while some people will argue that even order distortion (second, fourth, sixth harmonics, etc.) is more pleasing than than odd order distortion (third, fifth, seventh harmonics, etc.), this isn't quite correct. If odd ordered harmonics were so objectionable, we simply wouldn't be able to bear listening to most musical instruments.

That is an odd and in my opinion improper argument. What is in the music is in the music. It is the sound that we like or don't like. If an electronic device piles on more distortions, then at some point we hear that extra distortion. It has nothing to do with instruments having or not having that type of profile in their content. They are the original art. We are trying to preserve them when we play them.

It is like saying fish out of ocean has salt so adding more salt to them no matter how much doesn't have a negative effect.

So the Holy Grail for distortion is twofold. First, they must be low order (second, third, fourth) and second their levels should reduce monotonically from the fundamental, i.e. third lower than second, fourth lower than third, etc. SET amps can achieve this in good part due to their single-ended nature and helps explain why they can sound so good even if their THD figure might suggest otherwise. But the less than optimal air-gapped output transformer takes its toll and produces significant amounts of high order distortion products.

Again, this is trumped by masking and threshold of hearing. The masking window doesn't care what is in it. Here is what it looks like again:

If the distortion products are so high as to peak above the graph, then yes, the lower order ones are less audible due to shading of masking having that non-linear drop off.

 

[amirm]

You're saying 'distortions' are the only issue in terms of SQ? In which case seems to me you're barking up the wrong tree, the issue AFAIA is noise. And its modulation with signal.

Well, put forward the psychoacoustic science and we can discuss it. Until then we are discussing the current thesis which is the distortion pattern leading one to believe one amplifier topology is better than the other.

 

[morricab]

I am not here to discuss anyone's belief. Heaven knows that would be a useless exercise . Instead I was asked to comment on validity of the paper you have put forward. That paper at least superficially relies on two factors:

1. A single blind ABX test showing preference for a SET amplifier over solid state.

2. Psychoacoustic evidence that fundamentally SET amplifiers are transparent and solid state are not.

If we stick to the plot, then anyone who says amplifiers sound different needs to show #1. I don't think there are any takers on that . All I see are sighted evaluations that don't rely on ear alone for their conclusions. Mind you, amplifiers may sound different but again, that is another discussion altogether.

On the second one I have shown evidence that masking which the author himself acknowledges shows that negative feedback based solid state amplifiers generate distortions that are masked by it.

These are the facts. That one or many think or have some idea about amplifiers sounding different is the old tired "I am right you are wrong" which doesn't interest me or is within the scope of what you put forward.

If we are done with the paper not proving its point, then we have made progress. If we are not done, then let's see that evidence based on premise of #1 and #2.

You are grossly oversimplifying the analysis when you say : 1) A single blind ABX test and 2) SET amplfiers are transparent and SS are not

He states instead that there is a pattern he has derived from aural harmonics that he considers to be appropriate for the amplifier's distortion to "hide" and be essentially inaudible. Amplifiers that conform to this pattern then sound essentially transparent. Those that do not will have some audible consequences. It is clear that even SETs do not meet perfectly the pattern that Cheever has set up; however, they do meet it better than amplifiers with obvious deviations based that are ultimately consequences of their design choices.

The basis for his theory is well established in the audio literature from the likes of Shorter and Crowhurst among others. You may not agree with the slope of the curve genertaed by his "aural harmonics" but it has been known for quite some time now that weighting of harmonic distortion is necessary to get some semblance of a correlation with listening perception.

The ABX single blind test was not meant to be definitive or exhaustive and he makes this clear...why do you harp on this and not the bigger picture? Sure data has to be generated but a lot was done in the past that supports the same supposition.

"If we stick to the plot, then anyone who says amplifiers sound different needs to show #1. I don't think there are any takers on that . All I see are sighted evaluations that don't rely on ear alone for their conclusions. Mind you, amplifiers may sound different but again, that is another discussion altogether."

I would take you up on this. We have done our own single blind tests and found them to be rather conclusive. IMO, double blind is too cumbersome to execute in a domestic situation but single blind works and is suitable for a first approximation. Cheever gives you a single blind test and you sneer at the result.

"negative feedback based solid state amplifiers generate distortions that are masked by it."

not so. High order harmonics generated by SS feedback amps are not masked at normal listening levels. Essentially anything over about 9th harmonic is not masked no matter what.

I have recently found a Ph.D thesis covering this subject:

https://books.google.ch/books?id=G3...onepage&q=Norman Crowhurst distortion&f=false

You would do well to read this and wonder about your blind faith in measurments.

 

[amirm]

You would do well to read this and wonder about your blind faith in measurments.

I have said nothing to lead you to this comment. Indeed my faith is in using ears and only ears to determine fidelity. Failing that, then I look at product design, architecture, proper research, measurements, etc. to determine fitment for my use and my audio beliefs. Only using that triangulation can I have high trust in what I am believing.

On that basis and again, as the author of your own paper put forward, high impedance SET amplifiers can absolutely color the sound. This is easily shown and demonstrated. To say that they provide transparency is simply wrong. The author says that. I say it. And so do countless other sources.

 

[amirm]

You are grossly oversimplifying the analysis when you say : 1) A single blind ABX test and 2) SET amplfiers are transparent and SS are not

No, that is the net of it. Here is the part that puts forward the ABX:

"4. Conclusion. A call for a new methodology
The previous standard measurements clearly show the perfection of the high
power solid- state design and the horrible “performance” of the zero feedback. Yet, in
listening tests [15], the single ended tube amplifier was unanimously judged as sounding
closer to the truth. It seemed more dynamic. It had less “grain”16, especially in the midrange.
It seemingly had higher resolution as it presents a better “imaging” [17] sound field. "



And [15] is this:

"[15] Although a strict scientific experiment was conducted comparing the amplifiers I do not present an
detailed analysis here. In summary, the type 45 amplifier was chosen as preferable 100% of the time by all
the different listeners (5) in a “single-blind environment”, meaning, the listener toggled a remote push
button and either the amplifiers were swapped or not. A numeric display was incremented at each selection
and the listener noted if the amplifier changed or not, and if the change was to the one preferred. I attended
all sessions and verified that levels where matched, that there was no clipping, and the program material
kept within the flat pass-band of the type 45 amplifier-speaker combination. The source was a live piano
microphone feed."

This is core to his thesis: that "strict scientific" listening test shows measurements don't correlate with listening results. If we throw the listening test out then the rest becomes academic and likely would have led to a failed thesis if his advisors have any knowledge of the field.

For the second part, he says this in the conclusions:

"The public has limited access to participate in demonstrations of single ended
audio amplifiers. Very few mainstream Hi-Fi shops have this type of
equipment set-up. I know of none in the Boston area. If the reason for this is the need
for suppliers and salespeople to stick to what they know- mainly the specifications
race then the specifications need to change. These amplifiers are clearly superior in
many of the most important areas of sound reproduction. Unfortunately the design
has some drawbacks. "

If all he wanted to say was the harmonic order and shape there would have been no reason to bless SET amplifiers this way. He does this throughout the text and hence my strong impression that he came into this with this bias and then worked backward to draw a path to it. This is also consistent with the view you are putting forward. So not sure why it is bad to state it as such.

 

[morricab]

That is an odd and in my opinion improper argument. What is in the music is in the music. It is the sound that we like or don't like. If an electronic device piles on more distortions, then at some point we hear that extra distortion. It has nothing to do with instruments having or not having that type of profile in their content. They are the original art. We are trying to preserve them when we play them.

It is like saying fish out of ocean has salt so adding more salt to them no matter how much doesn't have a negative effect.



Again, this is trumped by masking and threshold of hearing. The masking window doesn't care what is in it. Here is what it looks like again:

If the distortion products are so high as to peak above the graph, then yes, the lower order ones are less audible due to shading of masking having that non-linear drop off.

You do see that the 250hz masked is only masking 2nd a bit of 3rd and a tiny bit of 4th harmonic, right? No higher order harmonics are covered. They will be exposed and therefore potentially an audible.

Your insist also on showing a single tone and it is clear that IMD products are going to be everywhere.

 

[morricab]

Frequency response errors do not inherently affect transparency. Not sure where you get this from.

I put Cheever forward as A paper in a hypothesis not THE only paper. I have also mentioned and linked work from Geddes, a PhD thesis, Boyk and Sussmann, Pass etc. Also commentary regarding Shorter, Crowhurst etc.

If you want to dwell on the one paper that is your problem, I put forward that the combined information points strongly that there is a preferred pattern and that yes SET generally fits it best.

You countered with low distortion SS amps should have inaudible distortion and since they also don't have FR errors then one can conclude that they probably sound the same and that all differences between tube and transistor amps is down to linear FR errors.

I said try that with headphones, which have high enough impedance to make response errors from just about any amp negligible.

I dare you to try and tell me your "trained" ears don't hear a difference...even between "same" SS amps.

 

[morricab]

No, that is the net of it. Here is the part that puts forward the ABX:

"4. Conclusion. A call for a new methodology
The previous standard measurements clearly show the perfection of the high
power solid- state design and the horrible “performance” of the zero feedback. Yet, in
listening tests [15], the single ended tube amplifier was unanimously judged as sounding
closer to the truth. It seemed more dynamic. It had less “grain”16, especially in the midrange.
It seemingly had higher resolution as it presents a better “imaging” [17] sound field. "



And [15] is this:

"[15] Although a strict scientific experiment was conducted comparing the amplifiers I do not present an
detailed analysis here. In summary, the type 45 amplifier was chosen as preferable 100% of the time by all
the different listeners (5) in a “single-blind environment”, meaning, the listener toggled a remote push
button and either the amplifiers were swapped or not. A numeric display was incremented at each selection
and the listener noted if the amplifier changed or not, and if the change was to the one preferred. I attended
all sessions and verified that levels where matched, that there was no clipping, and the program material
kept within the flat pass-band of the type 45 amplifier-speaker combination. The source was a live piano
microphone feed."

This is core to his thesis: that "strict scientific" listening test shows measurements don't correlate with listening results. If we throw the listening test out then the rest becomes academic and likely would have led to a failed thesis if his advisors have any knowledge of the field.

For the second part, he says this in the conclusions:

"The public has limited access to participate in demonstrations of single ended
audio amplifiers. Very few mainstream Hi-Fi shops have this type of
equipment set-up. I know of none in the Boston area. If the reason for this is the need
for suppliers and salespeople to stick to what they know- mainly the specifications
race then the specifications need to change. These amplifiers are clearly superior in
many of the most important areas of sound reproduction. Unfortunately the design
has some drawbacks. "

If all he wanted to say was the harmonic order and shape there would have been no reason to bless SET amplifiers this way. He does this throughout the text and hence my strong impression that he came into this with this bias and then worked backward to draw a path to it. This is also consistent with the view you are putting forward. So not sure why it is bad to state it as such.

Amir, Cheever is hardly the first one to come to the conclusion in listening trials that distortion level does not correlate well with listening results. Geddes found the same damn thing! Obviously guys from the past found the same thing or why bother finding a better metric??

You keep overlooking this obvious fact and that Cheever didn't really need a listening test for showing there is a difference he did it to show his metric was worth something. Same with Geddes, who used synthetic distortion instead of amp types.


The industry didn't want to touch because it opens Pandora's box and a nice low number is much easier to sell and to have people convincing themselves it must sound perfect because that distortion number is so low it MUST be inaudible...as you have sagely pointed out.