 
|
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 |
69.181.149.250
In Reply to: What difference does it make if it's right or wrong, the guy bullshitted his way to a masters degree posted by Soundmind on June 13, 2006 at 13:50:29:
I can only do 1 million to one in distortion. Can you do better?
Follow Ups:
My hunch is that even if all the sound and distortion were lumped at a single frequency, it would be inaudible at that level. That's what, 60 db down? Here's a DBT experiment for you to test actual sound acuity. Try a single test tone and add varying percentages of harmonics to it to determine at what level the threshold of audibility occurs for both single related harmonics, combinations of harmonics, and non harmonically related spurious sound such as white noise. Then you will know the threshold of audibility of IM, and of various distributions of harmonic distortion you can use as a design criteria for a recording/playback chain. Dr. Bose said in the 1970s for harmonically related sounds, it's 6%. Maybe he was wrong. So far all other discussion has been qualitative. Without setting parameters, you can claim anything and there is nothing to talk about. How convenient for those who want to spend the rest of their lives designing endless variants of equipment to perform the same function, there will never come a time when the performance is good enough to quit and say we have achieved our goal beyond the best performance necessary, the only further improvement possible is to make it cheaper and more reliable. I hope you take this challenge seriously as it would entail real research and provide those practicing the art with real knowledge that is of value. I regret that you have not taken up my suggestions of this kind of pursuit in the past. Good luck in teaching all of us something of real value.
Try Keith Howards software to add whatever distortion you like (up to 20th harmonic) at any level you like to whatever music you want. Even quite low levels are audible, no single test tone needed. It is audible with music.
I lose track of who said what sometimes. A lot of people who dismiss DBTs often cite DIY home brew informal testing as invalidating their conclusions. What invalidates them for me is when not only the opinions but the substance comes from someone I regard as having a consisant point of view biased in favor of those whose contribution of advertising revenues supports their own business enterprise. I see it as an unacceptable conflict of interest. I also give no credibility to those who trot out scientific evidence, statistics, and data when it supports that point of view and dismiss it when it doesn't.
Not sure about that Soundmind. Personally, if I had the time I would conduct properly designed DBTs with cds using distorted and undistorted data. I made files using Keith Howard's software. I took a 10 second segment of solo violin (My rationale for this software was that it is a single instrument with a high harmonic content so added distortion may have a good chance to be audible).Instead of using some regular orderly patterns like Mr. Howard did, I took actual distortion data taken at 1Khz from Soundstage magazine. I took the data from 5 different amps (of different topologies), some recommended from listening, some not. I then encoded the original undistorted cd with each distortion making a total of 6 10 second files on the cd (one for each amp and the undistorted signal). I can hear the difference but I didn't try it yet blind or double blind. Still, with such a cd it is much easier than trying to assemble all the amps; However, it is unlikely that this character for each amp will hold over all speakers etc. so in a way the distorted cd is BETTER for getting consistent results from a variety of systems because it focuses only on the distortion aspect of the argument.
You might consider that the solid state amplifier being more accurate reveals shortcomings in the speaker the less accurate tube amplifier masks. Is this possible? Can a better amplifier result in worse sound? Here's an example you might recall. I think it was PatD who first posted it on another site but I'm sure it was discussed here too. This was the experiment the Japanese professor of electrical engineering set up to demonstrate to his students that untrasonic frequencies were inaudible. He made a recording having very extended response into the ultrasonic range, set up a very wideband sound system, and could selectively filter the ultrasonic portion in and out of the composite signal. To his amazement, he found his students could readily identify the difference and knew when the ultrasonic components were present. When he went back and looked at the experiment, he realized that the ultrasonic frequencies were causing intermodulation distortion in the audible range of the tweeter. (He reconfigured the speaker to have a separate tweeter for the ultrasonic range and got the expected results.) But in this identifiable configuration, had he substituted one amplifier identical to another except the better (more accuate) one having bandwidth flat to beyond audibility and the other cut off at 20 khz, the better amplifier would have resulted in the poorer sound because it would have created the same distortion while the poorer amplifier would not have. When looking for the failure in a chain, the first and best place to look is at the weakest link, in this case the speaker, not one of the strongest links, the amplifier. This is an example of why I am very dubious about buying an amplifier on listening alone or even giving much value to that factor unless I am testing it with the very speakers and other equipment I intend to use it with. And I still want to know all of the specifications.
Yes Soundmind I do think about it from this direction, which is why I am very interested in speaker technology as well. THere are actually many things in speaker design that you and I agree about (a closed box loading for woofers for example). However, I think they are two different problems because one can often still clearly hear differences in electronics quality with speakers that are very flawed. The distortions that amplifiers make are electronic in nature whereas speaker distortions are largely mechanical and it seems that one doesn't readily mask the other. I know you think all well designed amps sound the same, I wish I could assure you they don't but I see the futility of making that argument at the moment. Let's just leave it at I have heard from countless examples that often things like box coloration, cone breakup, phase errors, etc. don't seem to mask the distortions of the electronics, perhaps because these distortions ride on the signal already when it reaches the speaker?
" I know you think all well designed amps sound the same"I would like you to point to even one instance in the years I have been posting here where I ever said that. Quite the opposite, I learned a long time ago that they don't. But I attribute most of their differences to differences in real world (connected to a loudspeaker instead of a risistor) frequency response which IMO is correctable using active equalization. I have no problem using this method although I must admit that the casual user who merely twiddles the controls endlessly and aimlessly has no hope of using this powerful tool to its real potential or even to any real advantage. As I've said, it usually takes me two years to successfully adjust one (I have one system going on 2 1/2 years now.) And how do I know that amplifiers sound different? Because when I substitute one in a system for another as when one blew up on me, I had to start re-equalizing the system all over again. And it took just about 2 years to get it right.
"IMO is correctable using active equalization. I have no problem using this method although I must admit that the casual user who merely twiddles the controls endlessly and aimlessly has no hope of using this powerful tool to its real potential or even to any real advantage"Soundmind, I have a digital equalizer (Behringer DEQ 2496 only in digital in/out mode) and I know how to use it very well. My system is equalized and I have different curves for my two amps for just the reason you have stated. You know what? They still sound significantly different.
Then IMO, in all likelihood, they are not adjusted to compensate properly. The only way I can see to do that is to connect each of them to identical loudspeaker loads and simultaneously to a differential amplifier which will facillitate nulling their difference. (You didn't think it would be easy did you?)
"they are not adjusted to compensate properly"Sorry I don't know what in the hell that means.
I don't think your prescription is necessary. You said FR response differences and I have corrected for that. What exactly am I nulling with your proposed setup? FR, distortion?? It is not clear to me what your proposed "solution" solves. If FR is the same then within the limits of the two amps the differences in sound must be due to distortions.
HowdyHow about two amps with essentially the same design but one being build for more power output? (In my case, say a SimAudio W-5 stereo amp vs. a SimAudio W-10 mono block.)
Even tho I never get near the power limits of either the W-10 has a lot more finesse than the W-5. I hear more details at the same listening level with the W-10...
(Yah, I know this isn't the best example, stereo separation, etc. come to mind, but still I don't think the differences come from things like that. I'm just positing that a loafing amp might sound better than a less loafing amp and that the difference has nothing to do with equalization...)
-Ted
I have no way of knowing but if it had to take a wild guess, I'd say that if everything else was pretty much the same, the difference would most likely be due to the power supply. IMO, the most important element in any power amplifier is the power supply. No amplifier of any design can be better than its power supply. The best are very conservately designed, are line and load regulated, and have rapid recovery from overload. And BTW, almost surely have solid state rectifiers. The most important element in the power supply is the power transformer followed by the filter capacitors. That's why if I had only one specification to judge an amplifier by, I'd want to know how much it weighs. I've gotten a lot of reports from friends who told me they made substantial improvements in their old Dynaco Stereo 70s just by replacing the power supply with a much more robust one of the same bias voltages. It's discouraging to see very expensive amplifiers even though they might have modest rated power output capbilities with puny power supplies you could build yourself for a hundred dollars or less. The least a $10,000 amplifier can give you is a good power supply.
HowdyI was pushing you about the over simplification about freq response being able to explain amp differences. I just chose another (obvious to most) difference in amps.
I guess my overall point is that there are so many (reasonable) differences in amps, especially if you consider price points, that it doesn't really do a lot of good to say "all reasonably designed ...'s sound the same", (not that I'm accusing you of oversimplifying that much.)
-Ted
Then would your also conclude that to isolate one difference such as the difference in the relative amplitude of different harmonic distortion components as a definitive explanation of why some people prefer one amplifier or type of amplifier over another (not even talking about whether it is "more accurate" whatever that means) without exhaustive testing eliminating so many other possibilies and variables is ludicrous?
It is not ludicrous if you make a prior the assumption that the end result of all these variables in an amplifier is various distortions. Measuring the end result and simply correlating this to the response of the end user is a way to see if the design is doing what it should.If you are a designer and want to eliminate as much distortion as possible it is first important to know which distortion is the most important to eliminate first. The point of the correlation is not to find root causes as to the sources of the distortions (like weak or poorly regulated power supplies, bad circuit design etc.) but to give a holisitic view as to its performance where it counts, with the listener.
HowdyLudicrous is perhaps too strong :)
I do agree that there are a lot of things that matter and that the thesis in question is but a start on one of them. Also I agree that controlling more variables in an experiment is a good idea and that one would sure want more things controlled in this case.
But it is accurate IMO that many of the measures quoted by manufactures don't tell us enough about things that matter when listening in practice and perhaps something like the thesis's proposed measure would be a good step in the right direction.
-Ted
We have known about distortion weighting for close to 50 years, it appears the rest of the industry and the majority of consumers are just not interested.
d.b.
HowdyIn all seriousness, do you or did you ever find a set of distortion wights and use them? If so were they of value to you in your work?
-Ted
And as far as I can tell none of the previously developed weighting schemes were particularly effective, which means its not a "solved" problem so why should it be implemented if it is not significantly better at predicting sound quality?
As I said before, with a little work it could be done, and it's certainly better at this point then the present mess. Not to worry; even if they did come up with something near perfect, the industry would ignore it, and the consumers aren't interested. It's an intellectual curiousity as far as the rest of the world is concerned.
d.b.
Neither are the people who design, manufacture, sell and buy measuring equipment. That tells me that they have considered and rejected this entire idea as being of no value to them. Interestingly, it would be a very simple matter today to incorporate such weighing into measuring equipment, just a minor software addition yet it still isn't done. That says it all AFAIAC
If you bothered to read the intro to Cheever's thesis you would see that none of the previously attempted metrics ended up being sufficiently predictive. Which metrics would you incorporate?
Sorry but I missed it.
"The Audio Asylum is a free,
independent resource"
I am addressing Soundmind about distortion measurement, since I have many 10's of thousands of dollars in test equipment (at least when it was new) and I use it virtually every day. How about the rest of you?
...I'm still unclear of your point. BTW if your point is that you have experience measuring equipment and still cannot correlate it to listening then I would ask why you have the measuring equipment in the first place. It is also possible that you personally may have trouble correlating but others don't. But in any event I am still unclear but interested in the point you are making.BTW I have the greatest respect of your ability to design and produce amps that deliver clean high power and measure (and sound) very good, and all at a reasonable price. That is the mark of great engineering, however you achieve it.
"The Audio Asylum is a free,
independent resource"
I CAN correlate some of my measurements with subjective performance. However, I cannot measure FM distortion in amps with my test equipment, and you probably don't know why I should be concerned. One or two good measurements is not enough. Learn and grow.
Progress is always good, but it is also important to understand where progress is most needed. You design amplifiers for a living, I do something else. I probably know more about your profession, as an amateur, than you know about mine. Consumers are the final arbiters of taste, but they can never of course be as informed as professionals. Still, I hope I never said that a couple of measurements were enough, that would indeed be a mistake. Information is not necesarily equivalent to knowledge because it can frequently be misguided, even for the most enlightened designers. When you have a screwdriver you see screws everywhere... This is true of doctors, lawyers, engineers...you name it.FM distortion is nonlinear distortion of the signal in the time domain. I'm sure it's a worthwhile goal to reduce it. But how much more phase distortion is being induced by speakers and rooms? We must keep things in perspective. Measurements are useful for that too.
Right now as a consumer I am a lot more interested in progress (and the dissemination and commercial application of exisiting knowledge base instead of all this nonsense) in the fields of loudspeaker and room correction than amplification or digital.
Learn and grow indeed.
"The Audio Asylum is a free,
independent resource"
I don't think that you understand FM distortion in linear amplifiers. I recommend that you read Barrie Gilbert's thesis on the subject.
nt
"The Audio Asylum is a free,
independent resource"
There's a nice collection of articles on analog design here . I believe the article John is talking about is this one .
You are correct, Andy C, thanks for finding the links.
To the amazement of many here, I'm sure, ;-) this is the sort of article that excites audio designers and shows a previously unpredicted problem with negative feedback, except for Otala's early paper, of course, that never got into the AES Journal, because it was considered too controversial by the Lipshitz crowd. This is for real. Matti told me this, himself. That's why he dropped out of the AES publishing game.
To the amazement of many here, I'm sure, ;-) this is the sort of article that excites audio designers and shows a previously unpredicted problem with negative feedback, except for Otala's early paper, of course, that never got into the AES Journal, because it was considered too controversial by the Lipshitz crowd.Hi John,
Yes, I seem to remember a pretty long thread about the Gilbert paper on DIYA :-). I think your interpretation of this paper is somewhat different from mine though. The conclusion I reached was that the closed-loop distortion was as usual much less than the open-loop distortion. However, the open-loop amp had no AM-to-PM conversion, while such distortion was present in the closed-loop amp. It seems to me that referring to this as a "problem" depends on the assumption that the subjective effect of the AM-to-PM distortion is far worse than that of AM-to-AM. I'm not aware of any studies that have taken place regarding the relative audibility of AM-to-PM distortion compared to AM-to-AM when the magnitude of the distortion components are equal. It would be interesting to find out though. But regardless of our differing interpretations, I think we both agree that it's an interesting paper that's well worth spending time on.
And speaking of interesting publications, there's a classic text on the Volterra series that for a long time was out of print and unavailable except for very expensive used copies. It is now back in print again, with new material added. I didn't realize until recently that for time-invariant nonlinear systems, the input-output relationship was completely described by the Volterra series, which includes the linear system theory as a special case. Very thought-provoking stuff. The book is The Volterra and Wiener Theories of Nonlinear Systems .
HowdyI ordered it, it sounds like a fun read :)
-Ted
...with the thread title taken from the Beavis and Butthead quote, "It's 10 PM. Do you know where your weiner is?". But I figured people would think I was weird - and I didn't want them to know the truth! :-)
HowdyIt's been a while since I did any optimal Wiener filtering, but my wife seems to like...
-Ted
this still does not make it clear to me what the order of magnitude of this problem is in a well designed circuit.
"The Audio Asylum is a free,
independent resource"
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 ]
