Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Tubes are objectively best! Feedback is bad! (Read 31652 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Tubes are objectively best! Feedback is bad!

Reply #25 –
Quote
To reiterate, many valve amplifiers create 'THD' measuments which, if produced by transistor amps in their characteristic fashion, would render them completely (and I mean completely) unusable for music reproduction.
[a href="index.php?act=findpost&pid=307967"][{POST_SNAPBACK}][/a]


Of course, because they wouldn't measure the same. They would have a totally different distortion spectrum. I'm talking about detailed measurements, such as distortion spectra with different test signals, for example. THD is a single figure which is not what I'd consider a detailed measurement.

Edit: fixes.

Tubes are objectively best! Feedback is bad!

Reply #26 –
Quote
What exactly is being measured in the first place - and how meaningful these measurements are,  was the subject of the paper Axon provided a link to at the top of the thread.


SNR =  ( sum (g*input_signal(t))^2 )/( sum (g*input_signal(t)-output_signal(t))^2)

Where 'g' is the amplifier gain, of course.

Quote
To reiterate, many valve amplifiers create 'THD' measuments which, if produced by transistor amps in their characteristic fashion, would render them completely (and I mean completely) unusable for music reproduction.

Why is this relevant?  A tube amplifier is not a transistor amplifier. A tube amplifier incontrovertably has different characteristics than a transistor amplifier, so why is your hypothesis that they would have the same values meaningful? I think we all agree that they have different characteristics, so why is your statement even marginally interesting?  Do you have some reason to suggest that your comparison is in any fashion meaningful? Any listening tests? Any validated measurements that suggest your comparison is meaningful?
Quote
Whatever one's views on vacuum-tube amplifiers and the whole-digit THD figures they produce, they are certainly not 'unusable', and this alone should give one pause for thought.


Can you show me where someone is asserting that vacuum-tube amplifiers are unusable? I'm speaking as someone who has designed and built several of my own, fixed 100's of others (not any recently, thankfully, zap! ow!), and who now uses lots of solid-state amplifiers with much greater satisfaction, since I'm working in reproduction, not in artistic creation or special distortion effects.

I don't have to pause for any thought, and I doubt that anyone else will need to if they realize, as we all seem to in advance, that the characteristics of tubes and transistors, the circuit designs they are used in, and their types of distortion are in fact quite different.

You're trying to compare an apple to a pomegranite for their avocado-like flavour, it seems to me.
-----
J. D. (jj) Johnston

Tubes are objectively best! Feedback is bad!

Reply #27 –
Quote
Quote
What exactly is being measured in the first place - and how meaningful these measurements are,  was the subject of the paper Axon provided a link to at the top of the thread.


SNR =  ( sum (g*input_signal(t))^2 )/( sum (g*input_signal(t)-output_signal(t))^2)

Where 'g' is the amplifier gain, of course.

Quote
To reiterate, many valve amplifiers create 'THD' measuments which, if produced by transistor amps in their characteristic fashion, would render them completely (and I mean completely) unusable for music reproduction.

Why is this relevant?  A tube amplifier is not a transistor amplifier. A tube amplifier incontrovertably has different characteristics than a transistor amplifier, so why is your hypothesis that they would have the same values meaningful? I think we all agree that they have different characteristics, so why is your statement even marginally interesting?  Do you have some reason to suggest that your comparison is in any fashion meaningful? Any listening tests? Any validated measurements that suggest your comparison is meaningful?
Quote
Whatever one's views on vacuum-tube amplifiers and the whole-digit THD figures they produce, they are certainly not 'unusable', and this alone should give one pause for thought.


Can you show me where someone is asserting that vacuum-tube amplifiers are unusable? I'm speaking as someone who has designed and built several of my own, fixed 100's of others (not any recently, thankfully, zap! ow!), and who now uses lots of solid-state amplifiers with much greater satisfaction, since I'm working in reproduction, not in artistic creation or special distortion effects.

I don't have to pause for any thought, and I doubt that anyone else will need to if they realize, as we all seem to in advance, that the characteristics of tubes and transistors, the circuit designs they are used in, and their types of distortion are in fact quite different.

You're trying to compare an apple to a pomegranite for their avocado-like flavour, it seems to me.
[a href="index.php?act=findpost&pid=307973"][{POST_SNAPBACK}][/a]



As is often the case on marginally 'skeptical' forums such as here at HA, everything is 'arguable' and so is argued to death.

Another characteristic I've found seems to be that almost every statement one makes is 'construed' in whatever way the 'arguee' wishes in order to to carry on arguing.

It's really pretty tiresome, and I am actually quite tired, so I'll leave it there.

goodnight,
R.

Tubes are objectively best! Feedback is bad!

Reply #28 –
Science, and verified and unrefuted theory, has only one way, I'm afraid. There's even some science that is mathematically true, same as 2+2=4 is true in an irrefutable way. This "science" are called theorems (BTW, one of which is Nyquist/Shannon sampling theorem).

edit: forgot calculation result

Tubes are objectively best! Feedback is bad!

Reply #29 –
[a href="index.php?act=findpost&pid=307937"][{POST_SNAPBACK}][/a]
[/quote]

So, what do you listen to music on?

And how much, in your wisdom, do you think someone should spend on an amplifier before any further expense is a waste of money and marks one out as a sucker?

For example, was the Creek 4330 I mentioned too expensive, and an obvious rip-off (£500)?

Just so we can nail this down to specifics, rather than idiotic pejoratives.

R.
[a href="index.php?act=findpost&pid=307947"][{POST_SNAPBACK}][/a]
[/quote]


To answer your question:  I have two mono-bridged halfer 500's, each producing ~1,200 watts, a halfer pre-amp and a hafler tuner.  A ReVox CD player and a Technics turntable with a Signet TK8LCp phono cartridge.  I have two sets of speakers: KEF 104/2's and a pair of Sound Lab Pristines (electrostatics).  I also have a SONY STRED725 for day-to-day TV, radio and movies.  I have a couple of more recent CD players and an MD player (SONY MZ-NH1)

I have no idea what this has to do with an ABX test between Stereophile and Bob Carver.  This is factual, as distressing as it may be.  For your further education, in a blind ABX test by Stereo Review many years ago a Mark Levinson amp/pre-amp combo was beaten out by a cheap Japanese receiver.  To say that tubes are better than transistors only requires a demonstration of how better.  To also believe that more expensive equates always to better is to invite challenges.   
Nov schmoz kapop.

Tubes are objectively best! Feedback is bad!

Reply #30 –
Getting this halfway back on topic....

I've been crunching on Moir's paper (thanks Chris!), and it indeed is a very valuable contribution. But Moir's smoking crack. I think his 2% JND is far too conservative, and in the strictest sense, is meaningless in the context of modern music listening.

Moir himself states that with careful practice THDs of 0.01% should be detectable for pure sine wave inputs. He then immediately dismisses it as being completely artificial. Unfortunately for Moir, such sine waves are far more common in music than they were in 1981, and it would not surprise me at all if a CD exists with a perfect noise floor and an extended pure sine wave. Assuming a 0.01% THD is detectable with such a music selection, you can't say anymore that 2% is JND for "all real music". If you did you've shifted the goalposts from "absolutely transparent" to "transparent most of the time for certain kinds of music", which is blatant discrimination against certain genres. This is the same argument that a lot of people tried to make in the vorbis ABX test done with pink noise (where did that thread go anyways?) and they fell flat on their faces with it.

As a quick sanity check of these numbers, I created a 44.1 .wav file with a 44.1hz (note hertz) tone, and zeroed out the first nonzero sample of each period. (That is, one out of every 1000 samples is zero in a sine wave, not including the zero crossings.) This represents a THD of 0.028% and is a close approximation of low-level crossover distortion. It's so detectable I didn't even bother to ABX it.

Tubes are objectively best! Feedback is bad!

Reply #31 –
EDIT: woops.

Tubes are objectively best! Feedback is bad!

Reply #32 –
Quote
Getting this halfway back on topic....

Moir himself states that with careful practice THDs of 0.01% should be detectable for pure sine wave inputs. He then immediately dismisses it as being completely artificial. Unfortunately for Moir, such sine waves are far more common in music than they were in 1981, and it would not surprise me at all if a CD exists with a perfect noise floor and an extended pure sine wave.


Certainly, if such a component is found in music(pure sine wave, non-masked by other spectral components, and for sufficient duration to be detectable[also covered in the paper]), then it would be a problem. But, I've never come across such a musical piece. I have some old Telarc synthized classical music, that while it has a strong amount of pure tones, they are not isolated sufficiently for any duration during the music. Not that a few real examples(among millions) don't exist, but the summary of the paper was for typical music program. It would obviously be erroneous to state that "all music all of the time" would be transparently reproduced. Such a statement, taken literally, would imply that one was all-knowing[a God, if you will]. The paper, as you stated(is very valuable), so it is only appropriate to look at it in the present context of applicable music. If John Doe listens to music that is composed of pure sine wave tones, then  <0.01% THD is required for transparent playback. But just as this type of music can(and probably does exist), remember, you can also win the pick 6 lotto, given enough random picks. Don't forget that almost all speakers(plasma speakers being a probable exception), have far greater THD contribution than the electronics driving them. But also, the crossover type distortion that you refer to is not the same spectral distribution as normal amplifier THD -- so it's not represenative of the typical harmonic distribution(except for in very poorly designed amplifiers or amplifiers that are operating out of specification -- bias out of whack), which the Moire article specifically covered.

-Chris

Tubes are objectively best! Feedback is bad!

Reply #33 –
Quote
This represents a THD of 0.028% and is a close approximation of low-level crossover distortion. It's so detectable I didn't even bother to ABX it.
[a href="index.php?act=findpost&pid=307986"][{POST_SNAPBACK}][/a]


Exactly. It's a kind of distortion that creates far-removed harmonics, and it is quite generally agreed that it's an "incredibly bad, horrible, awful sort of thing".
Your example is 71 dB down from max, but is very well selected to ensure that there is no masking at all of artifacts in the ear canal resonance range.

Ergo, I'll bet it sounds awwwwful.

But now try running the same tone, without the distortion, through a process that does  x(t)+alpha*x(t)^2, set alpha so that you have the same distortion, and see what you can hear, ok?
-----
J. D. (jj) Johnston

Tubes are objectively best! Feedback is bad!

Reply #34 –
Quote
As is often the case on marginally 'skeptical' forums such as here at HA, everything is 'arguable' and so is argued to death.
Ah, we're too skeptical! OK guys, we need to be more credulus and just believe what he says. That will make life much easier since we won't have to think for ourselves. Everyone can just go out and buy a high end tube amp -- it's so easy!

Quote
Another characteristic I've found seems to be that almost every statement one makes is 'construed' in whatever way the 'arguee' wishes in order to to carry on arguing.
I think that there are two possible explanations for this phenomenon. 1) You are being unclear. 2) You are wrong.

Quote
It's really pretty tiresome, and I am actually quite tired, so I'll leave it there.
Imagine how tiring it is for the people who have to refute various audiophile myths week after week.

Tubes are objectively best! Feedback is bad!

Reply #35 –
Quote
It would obviously be erroneous to state that "all music all of the time" would be transparently reproduced. Such a statement, taken literally, would imply that one was all-knowing[a God, if you will].

Not at all. There's a very large difference between "everthing in the universe" and "everything we know". In terms of what we objectively observe, and design for, we can only take into account the latter - but our knowledge grows over time. Besides, it's perfectly valid to design towards complete transparency, and given that we're in a relatively audiophile discussion, that's sort of the point.

Quote
But also, the crossover type distortion that you refer to is not the same spectral distribution as normal amplifier THD -- so it's not represenative of the typical harmonic distribution(except for in very poorly designed amplifiers or amplifiers that are operating out of specification -- bias out of whack), which the Moire article specifically covered.

Yeah, I was not aware just how bad crossover distortion really was until I looked at it myself. Eeeaaugh. The sample I tested with, being one impulse different from the ideal sine, had the exact same harmonic amplitudes from the 2nd to the 500th.

Nevertheless, said amplitude turned out to be.... about -96db. Now, I'm just hypothesizing at the moment, but I suspect that the audibility of different harmonics is related to the equal-loudness curves. If that is true, then it's plausible that only a few high-order harmonics at -100db are enough to be audible. And judging by the sample RMAA results at RightMark, where even pro cards have harmonics only 100db down, such a result would be very relevant to amplifier testing.

BTW, if you want to be scared out of your wits, look at a harmonic spectrum for an Etymotic ER-4S (my headphones, incidentally). Headroom had one posted but I can't find it right now. I would swear they fudged something up, because it seemed like the THD value for their results would be in the multiple percentage points, but I can't think of any particularly good way a dummy-head testing apparatus could inject that many harmonics without obviously clipping.

Tubes are objectively best! Feedback is bad!

Reply #36 –
Quote
Not at all. There's a very large difference between "everthing in the universe" and "everything we know". In terms of what we objectively observe, and design for, we can only take into account the latter - but our knowledge grows over time. Besides, it's perfectly valid to design towards complete transparency, and given that we're in a relatively audiophile discussion, that's sort of the point.


That was my point: that you can only qualify the statement within the context of what is currently known. My statement of all knowing was intended to illustrate the non-feasability of making such a statement as absolute(as opposed to tenative).

So far as this being an audiophile discussion and being about transparency: I am remaining within a scope of practical application(s), not any and all applications/circumstnaces. That's all that is feasible. The transducers will still be responsible for far more harmonic distortion than the electronics. The 0.02% contribution of THD by an average[properly functioning and designed] amplifier is insignificant in contrast to the loudspeaker distortion(s). If absolute transparency under any and all conditions is desired(not just the overwhelming majority of real musical program), serious consideration of what to do about the transducers is required.

Quote
Nevertheless, said amplitude turned out to be.... about -96db. Now, I'm just hypothesizing at the moment, but I suspect that the audibility of different harmonics is related to the equal-loudness curves. If that is true, then it's plausible that only a few high-order harmonics at -100db are enough to be audible.


Again, since the spectral distribution of crossover distortion harmonics and relative levels of these harmonics represented in your sample file/experiment as compared to what is  produced by normally operating amplifier circuits, it's not fair to consider this within the scope of standard THD discussions, if this is what you intended. Crossover distortion should not be any issue in properly designed modern equipment, used within it's intended applications(for example, don't use a 1000 watt/channel amp on 100 dB/1meter sensitive horn speakers, the gain is severely mismatched and will make noisefloor very audible, along with any other very low level effects that would never be normally apparent).

Quote
BTW, if you want to be scared out of your wits, look at a harmonic spectrum for an Etymotic ER-4S (my headphones, incidentally). Headroom had one posted but I can't find it right now. I would swear they fudged something up, because it seemed like the THD value for their results would be in the multiple percentage points, but I can't think of any particularly good way a dummy-head testing apparatus could inject that


I have always doubted the value of measurements from that site. However, if the measurement was made on a dummy head, the harmonics that occured within the same band(s) as the ear canal resonance(s) would be exaggerted in measurement of the device. So, this must be accounted for when cross-referencing with a harmonics audibility chart.

-Chris

Tubes are objectively best! Feedback is bad!

Reply #37 –
Quote
That was my point: that you can only qualify the statement within the context of what is currently known. My statement of all knowing was intended to illustrate the non-feasability of making such a statement as absolute(as opposed to tenative).

So far as this being an audiophile discussion and being about transparency: I am remaining within a scope of practical application(s), not any and all applications/circumstnaces. That's all that is feasible. The transducers will still be responsible for far more harmonic distortion than the electronics. The 0.02% contribution of THD by an average[properly functioning and designed] amplifier is insignificant in contrast to the loudspeaker distortion(s). If absolute transparency under any and all conditions is desired(not just the overwhelming majority of real musical program), serious consideration of what to do about the transducers is required.

Agreed. All this is sort of a moot point compared to the transducers.

Quote
Again, since the spectral distribution of crossover distortion harmonics and relative levels of these harmonics represented in your sample file/experiment as compared to what is  produced by normally operating amplifier circuits, it's not fair to consider this within the scope of standard THD discussions, if this is what you intended. Crossover distortion should not be any issue in properly designed modern equipment, used within it's intended applications(for example, don't use a 1000 watt/channel amp on 100 dB/1meter sensitive horn speakers, the gain is severely mismatched and will make noisefloor very audible, along with any other very low level effects that would never be normally apparent).

I agree that crossover distortion is a pathological case. This is all hypothetical at this point, there isn't really any supporting evidence, and in this particular context, it would need to be applied to NFB amplifiers to be considered valid.

Quote
I have always doubted the value of measurements from that site. However, if the measurement was made on a dummy head, the harmonics that occured within the same band(s) as the ear canal resonance(s) would be exaggerted in measurement of the device. So, this must be accounted for when cross-referencing with a harmonics audibility chart.

That could very easily be it. IIRC, they used THD tests at 1khz.

I'm too tired to look closer into this - I have some code that implements the sin^2 distortion that Wood suggested, but actually listening to the output will need to wait until tomorrow.

However, in the process of screwing around I wanted to see what would happen with a pathologically small distortion. Instead of zeroing out the first nonzero sample in a cycle, now I'm multiplying it by 0.8. By my calculations (which admittedly could be wrong - I had a bit of a scare with THD earlier tonight), this turns out to have a THD of 0.0056%. I can ABX this 26/34, and 13/16 after a reset. There are a lot of variables floating around that could crud this up - this was using my normal foobar playback chain, which includes an equalizer and a resampler (SSRC normal mode) going from 44.1 to 96khz. But still - one sample! And my headphone amp is VERY noisy, too! (Admittedly I had to crank it up considerably to succeed at this, but still...)

More research is definitely needed here. Does anybody else want the samples I'm using?

Tubes are objectively best! Feedback is bad!

Reply #38 –
Quote
There are a lot of variables floating around that could crud this up - this was using my normal foobar playback chain, which includes an equalizer and a resampler (SSRC normal mode) going from 44.1 to 96khz. But still - one sample! And my headphone amp is VERY noisy, too! (Admittedly I had to crank it up considerably to succeed at this, but still...)

More research is definitely needed here. Does anybody else want the samples I'm using?


I am inclinded to suspect that you are mucking something up, if this is supposed to be represenative of the harmonic distribution of an average amplifier. I will try to get around to preparing some samples with known harmonic components added, and then e-mailing those samples to you. For example: emulation of the standard harmonic distribution of a common SS amplifier on music program(with strong pure tone components). If you can provide a music sample of which you believe would be most revealing of added harmonic distortion, please send via e-mail. I can't think of anything that I might have in my collection that would be especially sensitive to harmonic distortion(s), other than the afformentioned Telarc CD(of which I'm not certain of the current location).

-Chris

Tubes are objectively best! Feedback is bad!

Reply #39 –
Quote
You're just another in a long line of idiots shouting TOS violations when you have nothing better to add yourself.
[a href="index.php?act=findpost&pid=307918"][{POST_SNAPBACK}][/a]

Sorry to butt in, but isn't this what you might call a TOS #2 violation?

Tubes are objectively best! Feedback is bad!

Reply #40 –
Quote
I am inclinded to suspect that you are mucking something up, if this is supposed to be represenative of the harmonic distribution of an average amplifier. I will try to get around to preparing some samples with known harmonic components added, and then e-mailing those samples to you. For example: emulation of the standard harmonic distribution of a common SS amplifier on music program(with strong pure tone components). If you can provide a music sample of which you believe would be most revealing of added harmonic distortion, please send via e-mail. I can't think of anything that I might have in my collection that would be especially sensitive to harmonic distortion(s), other than the afformentioned Telarc CD(of which I'm not certain of the current location).

Yeah, if I limit the test to primarily polyphonic and/or studio recordings it gets a lot harder. If I wanted to avoid electronic examples entirely, and were able to guess the best type of music for detection, the best example I can think of is a double bass solo that is *very* well mastered. The point being that the instrument needs to drop low enough that ultra high harmonics are still audible.

I'll look to see if any existing music might work tomorrow.

Tubes are objectively best! Feedback is bad!

Reply #41 –
Quote
To answer your question:  I have two mono-bridged halfer 500's, each producing ~1,200 watts, a halfer pre-amp and a hafler tuner.  A ReVox CD player and a Technics turntable with a Signet TK8LCp phono cartridge.  I have two sets of speakers: KEF 104/2's and a pair of Sound Lab Pristines (electrostatics).  I also have a SONY STRED725 for day-to-day TV, radio and movies.  I have a couple of more recent CD players and an MD player (SONY MZ-NH1)

I have no idea what this has to do with an ABX test between Stereophile and Bob Carver.  This is factual, as distressing as it may be.  For your further education, in a blind ABX test by Stereo Review many years ago a Mark Levinson amp/pre-amp combo was beaten out by a cheap Japanese receiver.  To say that tubes are better than transistors only requires a demonstration of how better.  To also believe that more expensive equates always to better is to invite challenges.   
[a href="index.php?act=findpost&pid=307981"][{POST_SNAPBACK}][/a]


Very impressive rig.  but what on Earth made you decide you needed that kind of power? You do weddings and barmitzvah's, presumaby?

To me that sounds a mad as any 1.5 watt SET.

I use a Townshend Rock BTW, a proper turntable.

Bob Carver and Stereophile? I've never read Stereophile (I'm in the UK, so even I did read much of the hifi press, which I don't , it's relevence is tangential to most Brit's). The anecdote is  is entertaining, but it's just that, an anecdote, and doesn't really say or prove anything.

Are tubes 'better than transistors'?. That's how you choose to construe what I said, but it isn't actually. Are men taller than women?

Transisitors - low-voltage/high current-density. Tubes - high voltage/low current-density. Very different.

R.

Tubes are objectively best! Feedback is bad!

Reply #42 –
Quote
Getting this halfway back on topic....

I've been crunching on Moir's paper (thanks Chris!), and it indeed is a very valuable contribution. But Moir's smoking crack. I think his 2% JND is far too conservative, and in the strictest sense, is meaningless in the context of modern music listening.

Moir himself states that with careful practice THDs of 0.01% should be detectable for pure sine wave inputs. He then immediately dismisses it as being completely artificial. Unfortunately for Moir, such sine waves are far more common in music than they were in 1981, and it would not surprise me at all if a CD exists with a perfect noise floor and an extended pure sine wave. Assuming a 0.01% THD is detectable with such a music selection, you can't say anymore that 2% is JND for "all real music". If you did you've shifted the goalposts from "absolutely transparent" to "transparent most of the time for certain kinds of music", which is blatant discrimination against certain genres. This is the same argument that a lot of people tried to make in the vorbis ABX test done with pink noise (where did that thread go anyways?) and they fell flat on their faces with it.

As a quick sanity check of these numbers, I created a 44.1 .wav file with a 44.1hz (note hertz) tone, and zeroed out the first nonzero sample of each period. (That is, one out of every 1000 samples is zero in a sine wave, not including the zero crossings.) This represents a THD of 0.028% and is a close approximation of low-level crossover distortion. It's so detectable I didn't even bother to ABX it.
[a href="index.php?act=findpost&pid=307986"][{POST_SNAPBACK}][/a]



I'm not sure it ever got completely off topic. The merits or otherwise of tube amps features prominantly in the paper you linked to.

Audibilty of harmonic (and intermodulation) distortion (or loss of 'transparency) is only part of the issue here, it's also a question of whether this audiblity corresponds in severity to the THD figures arrived at using current test protocols.

It would seem pretty clear that they don't, so looking for 'threshholds' of audiblity of THD seems futile.

Which is precisley the point of Cheever's thesis.

It's worth adding that 'linearity' in the response of audio amps doesn't relate to distortion figures of any kind, and the most linear device that exists is .... a single ended triode.

R.

Tubes are objectively best! Feedback is bad!

Reply #43 –
Quote
You're trying to compare an apple to a pomegranate for their avocado-like flavour.
[a href="index.php?act=findpost&pid=307973"][{POST_SNAPBACK}][/a]


If I could be bothered with a .signature, that quote would be going straight in there!


Anyway, I don't understand what there is to argue about here. (Unless you're in the wrong forum and don't accept the terms of reference for discussion here on HA). If exactly the same signal reaches a loudspeaker from two different amplifiers (as best as we can measure), they'll sound the same. There is no magic. If the signals are actually different, then one or the other may sound better, but if you use silly simple measurements that hide the important features (e.g. measure the total amount of distortion, rather than its spectrum!) then you'll completely miss the reason why.

Which sounds better? That's a matter of taste.

Which is more accurate? The one which changes the signal the least (aside from gain!).

Which is more perceptually accurate and transparent? Difficult to answer! Measurements and double blind subjective testing can help, but only if you have a "reference" which is measurably "perfect" - e.g. (OUT/G)-IN=0 (or something ~120dB down). Without this, or with finite differences which are ABXable, this is almost impossible to answer.

Cheers,
David.

Tubes are objectively best! Feedback is bad!

Reply #44 –
Quote
It's worth adding that 'linearity' in the response of audio amps doesn't relate to distortion figures of any kind, and the most linear device that exists is .... a single ended triode.
[a href="index.php?act=findpost&pid=308063"][{POST_SNAPBACK}][/a]

Well, if you go to the strict meaning of linearity, then, it has everything to do with distortion. In fact, the proper name of the distortion we are talking about is "nonlinear distortion". The more the linearity, the less the distortion, and viceversa. Perfectly linear devices, such as cables, create no distortion at all.

Still, some people mistake 'linearity' for 'frequency response', at least in my country (Spain), but in fact they are totally different things.

Tubes are objectively best! Feedback is bad!

Reply #45 –
Quote
Still, some people mistake 'linearity' for 'frequency response', at least in my country (Spain), but in fact they are totally different things.
[a href="index.php?act=findpost&pid=308071"][{POST_SNAPBACK}][/a]


Absolutely, it concerns things like 'slew-rate', and rise/decay times.

R.

Tubes are objectively best! Feedback is bad!

Reply #46 –
Quote
Yeah, if I limit the test to primarily polyphonic and/or studio recordings it gets a lot harder. If I wanted to avoid electronic examples entirely, and were able to guess the best type of music for detection, the best example I can think of is a double bass solo that is *very* well mastered. The point being that the instrument needs to drop low enough that ultra high harmonics are still audible.

I'll look to see if any existing music might work tomorrow.
[a href="index.php?act=findpost&pid=308047"][{POST_SNAPBACK}][/a]

What do you define as "ultra-high" harmonics? At what relative level? Any properly designed SS amp used within it's intended specification(s) will have rapidly decading harmonics, that if above the noisefloor at all(at orders that could be considered 'ulta-high'), will be at very low level, relative to such.

I have a very well recorded cello solo album. If you wish, I can rip and send a sample clip from track, in order for you to analyze, to see if it meets your expectations of what should be most revealing.

-Chris

Tubes are objectively best! Feedback is bad!

Reply #47 –
Quote
What do you define as "ultra-high" harmonics? At what relative level? Any properly designed SS amp used within it's intended specification(s) will have rapidly decading harmonics, that if above the noisefloor at all(at orders that could be considered 'ulta-high'), will be at very low level, relative to such.

Well, I'm still thinking under hypothesis that Cheever's thesis isn't totally bunk, and some harmonics matter more than others for audibility. With that in mind, I can split up harmonics into three categories:
  • Those emitted by normal instruments and voices (1st to 10th and probably beyond a good ways).

  • Those that are not a part of an instrument response

  • Those that are not a part of an instrument response, and furthermore hit the peak of the equal-loudness curve.

This third category is what I'm suspecting is the most audible under low distortion conditions. For a 200hz fundamental, the 18th fundamental would lay at 3600hz which is pretty close to ideal, and the lower it gets the higher the harmonic is necessary.

Of course, I haven't really figured out which harmonics were the most audible in my earlier tests, and that would help narrow the search a bit.

Quote
I have a very well recorded cello solo album. If you wish, I can rip and send a sample clip from track, in order for you to analyze, to see if it meets your expectations of what should be most revealing.

I'd gladly take that, although I'm not sure I'd call this an expectation. More of a guess  Would you take my samples in return?

EDIT: I'll admit to not knowing a whole lot about real-life instrument harmonics, and if real harmonics can extend out to the 30th or 40th then this could all be moot.

Tubes are objectively best! Feedback is bad!

Reply #48 –
I'm not sure that anybody responded on this particular point...

Quote
Quote

  • He is clearly not arguing from an objective position. The first chapter of the paper states that his SET amp is universally considered as subjectively better than a SOTA solid state amp, even though it measures uniformly worse. The rest of the thesis continues off the conclusion that the existing measurements are crap. To put it mildly, He Is Not On Our Side.[a href="index.php?act=findpost&pid=307719"][{POST_SNAPBACK}][/a]


I'm not sure I agree with you here.

Surely he's proposing a measurement protocol - which if implemented is then defacto, and therefore 'objective'?

That wasn't quite what I was having issue with.

I'm not doubting that his figure of merit is objectively calculated with clearly defined "best" and "worst" numbers. But the whole point of the derivation is based entirely off of subjective criteria. Moreover, he provides a number of points of attack on the existing objective measurements, but he treats the subjective ratings as absolute and infallible.

Tubes are objectively best! Feedback is bad!

Reply #49 –
Quote
Quote
In regards to the high-end discrete amp he first listens to, the author declares that the proper way to measure the amp is by castrating its feedback stage and then computing a figure of merit based on the open-loop distortion spectrum.


Oh.   

Do I understand this properly?  Are you saying that this wonderful gentleman argues that one must first break an amplifier before measuring its performance?
           
[a href="index.php?act=findpost&pid=307900"][{POST_SNAPBACK}][/a]

Right. He goes through this long-winded exercise to show that NFB on a nonlinear gain stage results in suppressed low harmonics and extended high harmonics. He then merely postulates that these high harmonics are audible, and that the only way to get rid of them is to avoid feedback - so that the only way to avoid potentially audible lower order harmonics is to make as linear an open-loop response as possible. He doesn't actually back up this assertion with any real evidence. And his measurements of the properly-functioning SS amp show almost immesuable amounts of distortion, so he can't actually pin down the badness of that amp to anything that's measurable. He can't apply his TAD to it either, because the TAD will report maximum distortion for a perfectly nondistorting amplifier, and the SS amp is pretty close to one.