I'm thinking of an amplifier whose gain varies only under certain circumstances, with steady sine tests giving constant gain. Anything that will simulate an amplifier that gives great THD figures, but will sound bad on transients. And I'd then like to work backwards, and see what tests we could do to detect the problem.
The best device for this test was designed by Richard Heyser about 50 years ago, as a demonstration of the limitations of steady state testing. It involved a relay that would open and close ONLY with asymmetrical signals, such as music, and stay closed with constant tones. I don't have a schematic, but it might be around, somewhere.
I think thermal effects were mentioned as a possible mechanism for problems that evade conventional testing, or are not repeatable. Didn't posters then discuss just how quickly a transistor core can heat up and cool down? Is it milliseconds, or seconds? Either way, I'd call it a potentially transient mechanism that wouldn't show up when testing with a steady signal level.
Hi,
As Earl Geddes how he produced the one file in his set that had a measured THD of 0.1% and sounded so grossly distorted, it beggared belief...
Ciao T
As Earl Geddes how he produced the one file in his set that had a measured THD of 0.1% and sounded so grossly distorted, it beggared belief...
Ciao T
a digital recording is a "measurement" - a list of numbers
IMD tests can be structured to tease out lots of this information - Cordell suggests dropping the SMTE 50/60 Hz down to 10-20 Hz to look for these hypothetical effects, you can also do dynamic "load soak" test - monitor distortion during cool down after the soak or interrupt every few seconds to do measure as the amp heats up
Self claims to have not seen any thermal effects with low audio frequency - he is still exploring bias stability/compensation as his "Optimum Class B" requires to prevent large changes in crossover distortions during turn on warm up or "load soak" magnitude music power excursions and ambient T variation
I think most here prefer the idea of using much higher AB bias, 100 mA or more per output Q so there is much less sensitivity, even simple thermal compensation, considerable "over compensation" won't reduce such an output stage to Class B or C operation under any thermal profile possible within the amp I,V limits
I do get tired of the limited "conventional measurement rhetoric - if you ask an engineer to setup a audio power amp development lab today he will go out and buy a recent AP analyzer - so right now "conventional audio measurement" - means everything in the menus, users manual of a recent AP analyzer
some here don't seem to even be up to speed on Cabot's 15 year old paper surveying, comparing audio distortion measurements
(waiting for John's "Cabot had it in for Otla” ad hominem attack - don't expect to ever see a technical criticism though)
IMD tests can be structured to tease out lots of this information - Cordell suggests dropping the SMTE 50/60 Hz down to 10-20 Hz to look for these hypothetical effects, you can also do dynamic "load soak" test - monitor distortion during cool down after the soak or interrupt every few seconds to do measure as the amp heats up
Self claims to have not seen any thermal effects with low audio frequency - he is still exploring bias stability/compensation as his "Optimum Class B" requires to prevent large changes in crossover distortions during turn on warm up or "load soak" magnitude music power excursions and ambient T variation
I think most here prefer the idea of using much higher AB bias, 100 mA or more per output Q so there is much less sensitivity, even simple thermal compensation, considerable "over compensation" won't reduce such an output stage to Class B or C operation under any thermal profile possible within the amp I,V limits
I do get tired of the limited "conventional measurement rhetoric - if you ask an engineer to setup a audio power amp development lab today he will go out and buy a recent AP analyzer - so right now "conventional audio measurement" - means everything in the menus, users manual of a recent AP analyzer
some here don't seem to even be up to speed on Cabot's 15 year old paper surveying, comparing audio distortion measurements
(waiting for John's "Cabot had it in for Otla” ad hominem attack - don't expect to ever see a technical criticism though)
Last edited:
So how is it that otherwise sensible-sounding engineers are still going on about certain amplifiers sounding bad, and others sounding good, and never even enquiring as to what this arsenal of 'conventional' measurements might reveal? What is the point of this thread if it's all so perfectly straightforward?
I'll admit that it gets me down, a bit. I'd love to see all the good technical stuff on these posts backed up with real 'proof' but as soon as it gets interesting (e.g. the discussion of thermal transients), it all goes a bit mystical, and the ultimate test is deemed to be a listening test! And if you happen to be of the persuasion that has no faith in anyone's ability to overcome psychological bias and/or simply their mood, then it's very frustrating! In previous threads, a real double blind test with any relevance to real listening has been shown to be so difficult to arrange, that it's probably never been done, so I just can't take it as read when someone talks about how an amplifier sounds good or bad. I'd love to see the measurements!
Where does he discuss this one? Since no one could hear a marching band at -60dB I'm curious.
Well JCX, I was not going to mention it, but I did once criticize Cabot for 'knocking Otala' when Cabot failed to get identical measured results to Otala, BUT Cabot used a different circuit to analyze. I just happened to be there in the audience when he took the 'cheap shot' at Otala, so I confronted him with it. A young man's 'overreach' to promote his point-of-view. Now Cabot and I are professionally on the best of terms, have been for decades.
Scott,
A while back, I think mainly to promote his distortion metric he had three files on-line to download that had been distorted. One with a process that produced > 10% THD, one with slightly below 10% THD and one with 0.1% THD, plus the undistorted file.
Both of the heavily distorted (measurement) files sounded essentially undistorted, but I could tell the more distorted one as being different from the original. The "low measured distortion" file was just awful..
Ciao T
A while back, I think mainly to promote his distortion metric he had three files on-line to download that had been distorted. One with a process that produced > 10% THD, one with slightly below 10% THD and one with 0.1% THD, plus the undistorted file.
Both of the heavily distorted (measurement) files sounded essentially undistorted, but I could tell the more distorted one as being different from the original. The "low measured distortion" file was just awful..
Ciao T
Which thread do you mean, and which sensible-sounding engineers do you mean? All sensible-sounding engineers including my humble person explained many times, looking from all possible sides, that measurements that we perform depend on topology of the amp we design, and on many other factors, so for one topology one kind of measurements is needed, another kind is useless, third kind is obvious from schematic and particular parts selected, and so on.
In short , different nonlinearities are differently perceived by the auditor,
or said otherwise , not all kind of distorsions have the same effect on
sound perception....
Is that new ?....
T, I have not closely followed the Geddes thread or papers, but he makes a point IF he can show something that measures 'bad' sounds OK, and something that sounds much 'better' sounds awful.
We were confronted by this by ABX, years ago, where what he called 'grunge' was virtually all 2'nd harmonic, and nobody could hear it easy. That is the way, often, with ABX testing. You always have to be 'on your guard' with it.
We were confronted by this by ABX, years ago, where what he called 'grunge' was virtually all 2'nd harmonic, and nobody could hear it easy. That is the way, often, with ABX testing. You always have to be 'on your guard' with it.
we do have a collision of world views going on in several different domains
hopefully most engaged in the hardware end of audio at least start from the "Scientific Rationalism" corner of global world views
but its not possible to build up from 1st principles complete descriptions of any technological device of any utility
some will then seize on this “gap” – claiming that since you don’t “really understand” then “anything “ could be possible
there are much bigger “gaps” in Physical Acoustics than in EE understanding of audio amplifiers – the transducers, mics, loudspeakers have much more complicated interaction with the sound fields they sense, reproduce than we have to deal with in the electrical signals they use in between
then to get to “sound” you have to add rooms, and the musical event recording
It is probably fair to say the Science has a very fuzzy “knowledge” at all of human perception
why is it that anyone could expect “simple” answers to such an ill posed question as (amplifier) measurements vs sound?
but do seem satisfied with simple "cheap shots" that agree with their prejudices
hopefully most engaged in the hardware end of audio at least start from the "Scientific Rationalism" corner of global world views
but its not possible to build up from 1st principles complete descriptions of any technological device of any utility
some will then seize on this “gap” – claiming that since you don’t “really understand” then “anything “ could be possible
there are much bigger “gaps” in Physical Acoustics than in EE understanding of audio amplifiers – the transducers, mics, loudspeakers have much more complicated interaction with the sound fields they sense, reproduce than we have to deal with in the electrical signals they use in between
then to get to “sound” you have to add rooms, and the musical event recording
It is probably fair to say the Science has a very fuzzy “knowledge” at all of human perception
why is it that anyone could expect “simple” answers to such an ill posed question as (amplifier) measurements vs sound?
but do seem satisfied with simple "cheap shots" that agree with their prejudices
here in Coppertops post is the worded contradiction of measurement vs listening.
Although not quite terrible a Quad 405 is a good candidate, it measures close to perfect
but sounds, very much less so.... why ?
Cheers / Chris
I do not. My point is that our ears/brain system is sensitive to something other than just amplitude and frequency that can assist us with at least localization. That leaves time and phase. Ohm's acoustic law is frequently and improperly applied to a broader set of situations than he was talking about.
Funny about the Quad 405. It and the 303 were part of one of my first amp A/B tests. Up to that point I really thought amps didn't make much difference. I was amazed to hear much difference there was between amps that basically "measured the same".
The 405 and 303 aren't much different, tho.
The 405 and 303 aren't much different, tho.
JCX,
Yes, which is why I pushed in this thread why on earth something as "simple" as the final amplifier, considering all the terrible things that have negative effects on the signal from mic to the speaker, would have such a significant effect on the sound. Enough to make my wife not tolerate some amplifiers playing horns, and tolerate others, never having an issue with live horns at much higher volumes.
We have had some insightful discussion of designers views on how to make a good amp. But not yet have I seen a sliver of a hint on what test I could run against an amp that fails my wife's listening test, and one that passes. This would allow one to point to the problem and fix it for all amps. In this specific case, it does not require exotic or expensive design as a mere Rotel 840 or 951 passes her subjective yes/no test, where a lot of others costing quite a bit more do not. I cam make similar subjective pass/fail on some classical guitar recordings that make the bass strings sound metallic. Very non-scientific, subjective description, but I hear it reliably. Again my original statement: If two amps sound different, at LEAST one is wrong. So, what is the difference? What distortion can cause my wife's teeth to grate and strings sound metallic? I don't know, I am just a computer systems architect, not an analog designer.
Of course, maybe subconsciously those with the talent to design very good amps may not want the world to know how theirs do less wrong.
Yes, which is why I pushed in this thread why on earth something as "simple" as the final amplifier, considering all the terrible things that have negative effects on the signal from mic to the speaker, would have such a significant effect on the sound. Enough to make my wife not tolerate some amplifiers playing horns, and tolerate others, never having an issue with live horns at much higher volumes.
We have had some insightful discussion of designers views on how to make a good amp. But not yet have I seen a sliver of a hint on what test I could run against an amp that fails my wife's listening test, and one that passes. This would allow one to point to the problem and fix it for all amps. In this specific case, it does not require exotic or expensive design as a mere Rotel 840 or 951 passes her subjective yes/no test, where a lot of others costing quite a bit more do not. I cam make similar subjective pass/fail on some classical guitar recordings that make the bass strings sound metallic. Very non-scientific, subjective description, but I hear it reliably. Again my original statement: If two amps sound different, at LEAST one is wrong. So, what is the difference? What distortion can cause my wife's teeth to grate and strings sound metallic? I don't know, I am just a computer systems architect, not an analog designer.
Of course, maybe subconsciously those with the talent to design very good amps may not want the world to know how theirs do less wrong.
In short , different nonlinearities are differently perceived by the auditor,
or said otherwise , not all kind of distorsions have the same effect on
sound perception....
Is that new ?....
Basically what he was saying is that a lumped THD spec isn't a terribly meaningful metric.
se
- Status
- Not open for further replies.