 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
84.73.113.58
In Reply to: Good points posted by EGeddes on December 29, 2005 at 09:44:37:
I read an article recently that mentioned an equation you helped develop regarding the correlation of subjective tests with THD. There was also an older attempt at making a better correlation by a guy named Shorter. As his equation was much simpler to use than yours (one did not need to know the transfer function of the amp) I have applied this to 1 Khz distortion spectra that are published on the Soundstage.com archives. What I found was rather interesting. There was a vast difference between most solid state amps and tube amps when the THD was weighted using Shorter's equation (basically the order of the harmonic squared/4 then times the amplitude of that nth harmonic).For sure the SS amps still had a lower overall total distortion...even with the weighting factor. But what was most interesting was when I looked at what percentage of that total was to be found in the harmonics greater than the 5th. In this regard, the SS amps had much more of their weighted THD in the 6th to 20th harmonic (the limit of the measurements used). There was a large disparity between the SS amps and most PP tube amps and then a gap again between them and SET amps (or hybrids and PP tube amps with no negative feedback). I can tell you which amps I find to sound the best (I have heard about half of the amps that I did this calculation on) and they are all ones that scored a lower higher harmonic:total ratio.
Will one be able to tell the difference blind between various well made SS amps? Perhaps not. Will they be able to tell the difference between an amp that has 80% of its weighted harmonics in the higher order range (> 5th) and one that has only 5% of its total weighted harmonics greater than the 5th harmonic? I should hope so or else there is no need to correlate subjective hearing and measurements at all because they all sound the same. My guess is the difference will not be subtle and could be reliably detected under blind or DBT conditions.
If you would like I can email you a copy of the spreadsheet I have compiled. If it is possible to apply your GedLee equation to an FFT spectrum of a 1Khz sine wave let me know how and I will use the same data with your equation.
Follow Ups:
You would find that with the GedLee metric, the tube amps would most likely rank better than the SS ones and that the correlation to perception was actually quite good. That is also what Keith Howard found, hence the article that you read.As far as my applying my metric to your data, I not really interested.
First, no one cares about truth in audio and when you try and offer it up you only get aggravation. So I'll stay out of that argument thanks.
And second, I am not very interested in anything that one has to actually do a DBT test on to find differences. These are miniscule and unimportant in the bigger picture of audio reproduction - otherwise they would be so obviuos that there would be no need for the DBT argument.
What makes a difference - three things - loudspeakers, rooms, and loudspeakers in rooms. Everything else is rather unimportant, unless its just plain broke.
Earl Geddes
I would add "recording and mixing techniques". ;)
"You would find that with the GedLee metric, the tube amps would most likely rank better than the SS ones and that the correlation to perception was actually quite good."Earl, I have no doubt that I would find this to be true. The fact that you bothered to come up with this metric strongly suggests that you have no doubt of the audible differences between amps, regardless of the paucity of data to "prove" it. I would argue that you believe that the DBT is not necessary to tell differences in amps and that in some cases the differences are obvious. I find it difficult then to understand the following quote:
"And second, I am not very interested in anything that one has to actually do a DBT test on to find differences. These are miniscule and unimportant in the bigger picture of audio reproduction - otherwise they would be so obviuos that there would be no need for the DBT argument.
What makes a difference - three things - loudspeakers, rooms, and loudspeakers in rooms. Everything else is rather unimportant, unless its just plain broke."
Because I find these comments curious. If you in fact believe the differences are not worth worrying about then why bother to take the time to develop a metric for the trivial?
The differences for a short audition may seem "miniscule", upon longer auditioning it can in fact be a make or break aspect to the sound. An amp that has a good score with your metric should be more satisfying long term than one that scores poorly, should it not? If the end result of using an amp that scores poorly is listener fatigue and a desire to stop listening to music this can hardly be considered "miniscule" or "unimportant".
In fact, I have often found that loudspeaker colorations are often easier to ignore than amplifier distortions over a long time period. The reason is probably that the coloration is everpresent and probably a relatively low order phenomenom (harmonically). The amplifier distortions that cause problems are almost always higher order and often signal and level dependent and therefore standout and annoy. The simple fact is that people often "get used" to speaker/room colorations and after a while simply don't hear them anymore (unless they are really serious).
An example: I went to go audition some amps in France not long ago. I took a friend who has good hearing along with me. We got their and found that the guy had a pair of really old (late 1950s) JBL Olympus (or some such name) horn speakers (tweeter and mids were horn bass was a big paper cone with passive radiator). Upon first listening, both my friend and though the speakers were horribly colored and I thought that there was no way to evaluate an amp with these speakers. A funny thing happened though, after about 30 minutes of listening to these things we "got into" the sound and the colorations, while still audible were not interferring with the pleasurable sound they made. In fact with regards to the dynamics of music they were outstanding and it was quite easy to hear musically relevant low level details of music. So in spite of the colorations they were pretty good sounding speakers.
Now, I hear live unamplified music every day and if I had to go back to those speakers afterwards I would need time to "adapt" again, where the colorations are less noticeable. My system now needs very little adaptation and is I think quite accurate as far as these things go(I make my own recordings as well). However, if I almost never heard live music and therefore did not need adaptation, I could see why he loved and lived with those particular speakers.
The point is this: I can now understand how someone can stay with colored speakers more easily than a distorted or fatiguing sounding amplifier.
I do not underestimate the importance of the speaker and the speaker/room interface, not at all. I take my speaker choices very seriously and what I have found is that as I improved the speaker and speaker/room interface the relative importance of amplifiers (and preamps) became more and more important. As colorations were removed by speaker choice (or design as I also design and build speakers as a hobby) I found that I could much better hear what contributions were made by the electronics. Thus the increase in importance. If high frequency distortions are swamped by a poorly designed tweeter that makes more problems of its own then of course this problem in the amp is not as audible. If the treble is accurate then the problem is very audible.
Let me pose a thought experiment for you. If you give people a series of amps to listen to sighted or unsighted it doesn't matter and ask them to rank the amps in terms of their preferrence (making sure that the speaker in question doesn't do anything funny with all the amps tested) and then compare this to your metric for weighting of distortion then this should be a "blind" test, should it not? I mean as long as the listeners don't know the test results before they listened then they don't know which amps "should" sound the best. In this respect they don't know what the expected result should be and they are free to choose based on sound looks or whatever. If the result comes back that the amps that fared the best with your metric also got the best rated sound then this "proves" two things: 1) that in fact the amps sound different and 2) that preferrence can be correlated with your metric. The test could even be carried out over several weeks to allow listeners to develop a long term preferrence and overall ranking. It would also destory the myth that the results are unimportant to the overall sound quality. What do you think?
First, the metric was developed for speakers, not amps. It was only after I had time to ponder the implications of what we had learned that I realized where a tube amp might sound better than a SS amp. It's the feedback that is the problem not whether its sand or vacuum. At any rate, I did this work for speakers and now I know that nonlinear distortion in speakers is irrelavent, so my interest in the metric is pretty low.Regarding
"If you give people a series of amps to listen to sighted or unsighted it doesn't matter and ask them to rank the amps in terms of their preferrence (making sure that the speaker in question doesn't do anything funny with all the amps tested) and then compare this to your metric for weighting of distortion then this should be a "blind" test, should it not?"
No, not unless its actually blind. People will rate the preference on a whole range of things unless the test is blind, but sound quality is probably not one of them. Blind, and I have acrtually done this exact test, the preference ratings will be random with no correlation to anything let alone a metric. Unless, of course as I have said, one of the amps is broken, in tersm of design or performance.
"What do you think?"
I think that you would be wasting your time. My metric says nothing about preference, only audibility. Preference is another question altogether.
Your anecdotal evidence might be interesting, but does prove anything. But I do agree that some forms of crossover distortion caould be quite audible even on "poor" colored loudspeakers and yield a fatigueing situation. But I would describe this amp as "broken" in that case. In any decent amp this would more than likely not occur.
I use a $150 Pioneer receiver for both my systems. Sounds just fine.
Very low "cool" factor however. Really turns off the audiophiles.
Earl Geddes
" I know that nonlinear distortion in speakers is irrelavent"
Why is that? Is it because most of the distortion is largely 2nd or 3rd order harmonics? My guess is that most processes don't produce much in the way of high order harmonics like you find in electronics.
"It was only after I had time to ponder the implications of what we had learned that I realized where a tube amp might sound better than a SS amp. It's the feedback that is the problem not whether its sand or vacuum"Hmmm...its interesting that as a result of the implications from using your metric that you realized that a tube amp might sound better than a typical SS amp but didn't your ears also tell you this?
BTW. I tend to agree that feedback is the problem because I own and have heard some hybrid amps with no feedback that have truly excellent sound quality (far better than most pure tube amps in fact that I have heard). One of the ones that I know to have no feedback scored very well using the Shorter metric and it sounds superb. I have also heard pure tube amps with limited feedback that sounded extremely good as well (I own an OTL amp that is in this category).
Interestingly, the hybrids and the OTL don't sound so different as topologically they are very similar. One thing I could imagine is that all amps that score very well or very poorly using your metric (or the one from Shorter) would perhaps have similar characteristics.
"My metric says nothing about preference, only audibility. Preference is another question altogether."
Hmmm...not directly no but indirectly I think it must. The whole point if I am not mistaken is to better correlate measurements with audibility of distortion. It provides a factor that relates how distortion looks when weighted properly. Apparently the lower this value the better it should correlate with subjective sound quality because it means lower audibility of distortion products.
Therefore, I was taking audibility and "different" as a given in that someone thinks a correlation needs to be improved.
"but I do agree that some forms of crossover distortion caould be quite audible even on "poor" colored loudspeakers and yield a fatigueing situation. But I would describe this amp as "broken" in that case"
As nearly as I can tell from the measurements I have looked at nearly all complementary Class AB amps would be "broken". I would not be surprised if you looked at the measurements of these $150 pioneer receivers you use that you would find crossover distortion in their measurements and that it is likely to be audible.
"Blind, and I have acrtually done this exact test, the preference ratings will be random with no correlation to anything let alone a metric"
You have? Is it published somewhere? How many amps and what types? How many listeners? ABX or some other design? I would be interested in seeing those results.
What I find really strange though, Earl, is that you are not denying the difference in audibility of amps. You even admit that there is a possibility that perhaps amps with particular design characteristics would likely sound better, precisely because the audibility of their distortion is less according to improved metrics for correlating distortion audibility and THD. Yet you still insist that it is not important when most listeners would disagree with you.
Again, I agree with you regarding the importance of the speaker and speaker/room interface, however, if this is a constant then I would pose to you that changing the electronics will interject audible variations from that constant. These variations can be significant enough to severely affect the listening experience and the subjective "realism" of the reproduction. I have heard this too many times with too many systems over too long of time periods to believe that it is all in my head and not due to real environmental factors.
It can't be all psychological self-delusion. Pavlov's dog sure salivated when the bell rang but they also found that if once in a while he didn't get food the effect of the bell to stimulate the salivation response weakened greatly. Same thing with a pretty amp sounding better. The person may be able to convince themselves it sounds better for a while but after a while the amp either has to produce sound up to its looks or it will be replaced.
"Why is that? Is it because most of the distortion is largely 2nd or 3rd order harmonics? My guess is that most processes don't produce much in the way of high order harmonics like you find in electronics."Correct. The mechanical nature of the major distortion mechanisms prevents higher order nonlinearities and low orde ones are not audible. What is audible is diffrcation in varying degress with level making it sound like nonlinear distortion when in fact it is actualy nonlinear perception.
"Hmmm...its interesting that as a result of the implications from using your metric that you realized that a tube amp might sound better than a typical SS amp but didn't your ears also tell you this?"Nope - cause I don't spend any time worry about amps. Its a small effect.
"You have? Is it published somewhere? How many amps and what types? How many listeners? ABX or some other design? I would be interested in seeing those results."
It was never published it was an in house study and it was on total systems. I talk about it in my new book - chapters availabel on line.
"What I find really strange though, Earl, is that you are not denying the difference in audibility of amps. You even admit that there is a possibility that perhaps amps with particular design characteristics would likely sound better, precisely because the audibility of their distortion is less according to improved metrics for correlating distortion audibility and THD. Yet you still insist that it is not important when most listeners would disagree with you."
Yes, I believe that SOME amps CAN sound different. I don't believe that ALL amps DO and I still insist that amps can be obtained which have a negligable factor on sound quality. So in the big picture amps are not worth worrying about because loudspeakers can not be obtained which have a negligable influenece, or even a small influence, on sound quality (such that it needs to be ABX'd) so this is where I put my energies. Biggest bang for the buck sort of thing.
"if this is a constant then I would pose to you that changing the electronics will interject audible variations from that constant"
Correct for the most part except "will" should be changed to "can" because you cannot say that all electronics "interject audible variations", some do, some don't. And the presumption of the loudspeakers being a constant is like arguing about pennies when buying a car but not worying about the brand that your going to buy. It may make a difference but the difference isn't worth arguing about if your fixed on a crummy brand.
"It can't be all psychological self-delusion. "
Well actually it can and for the most part is.
You seem to be a classic case. (I don't mean to be derogitory - you have been polite and I appreciate that.) I present evidence and you present conjecture while continually holding to your previous views. The self-delusion comes from not being open minded to the uncomfortable reality of things. I'm not open minded because "I've been there and done that".
Earl Geddes
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
