Can I ask a question:
Are we going to have any listening impressions in this thread?
I've been searching in my head for quite a while to find some kind of suitable analogy to what is going on here. I think I've found it in a way that some here will understand.
The assertion here is that if a set of measurements show no difference control to DUT, then there is no difference in sound. So physical measurements are taken to describe the sound as perceived by one of the human senses. This is an absurd argument. The depth of perception may vary from person to person, but take a large enough sample of people and very quickly a bias towards "good" or "bad" can be quite reliably formed.
Sy, I'm curious. Would you apply these same standards of measurement vs another human sense? How about to a bottle of wine? So if 2 bottles measured the same, would they taste the same? What would you measure? Alcohol content? Would you look to ppm levels or ppt? tannin levels? headspace analysis of aromatics?
I believe your notion of "the measurements show the same therefore it sounds the same" is naive and misleading, and to be honest I'm surprised. Do you feel the same way about capacitors?
Fran
Are we going to have any listening impressions in this thread?
I've been searching in my head for quite a while to find some kind of suitable analogy to what is going on here. I think I've found it in a way that some here will understand.
The assertion here is that if a set of measurements show no difference control to DUT, then there is no difference in sound. So physical measurements are taken to describe the sound as perceived by one of the human senses. This is an absurd argument. The depth of perception may vary from person to person, but take a large enough sample of people and very quickly a bias towards "good" or "bad" can be quite reliably formed.
Sy, I'm curious. Would you apply these same standards of measurement vs another human sense? How about to a bottle of wine? So if 2 bottles measured the same, would they taste the same? What would you measure? Alcohol content? Would you look to ppm levels or ppt? tannin levels? headspace analysis of aromatics?
I believe your notion of "the measurements show the same therefore it sounds the same" is naive and misleading, and to be honest I'm surprised. Do you feel the same way about capacitors?
Fran
We have had listening impressions in this thread, what we haven't had is listening tests. That is, listening conducted in conditions contrived to eliminate bias.
Because of a well known phenomenon, the placebo effect, it is commonplace to conduct tests involving human reactions to stimuli under conditions known as 'double blind'. This involves the application of a 'null stimulus' to some or all test subjects, depending on the exact nature of the test. Neither researchers nor test subjects are aware which stimulus the subjects are being exposed to at the time of test. Such tests are notoriously difficult to arrange, certainly by comparison with uncontrolled tests.
I for one would welcome a controlled listening test.
I do, however, strenuously object to the introduction of uncontrolled listening impressions as evidence of the efficacy of any modification, particularly where unsupported by instrumented test, and more particularly where there is no real theoretical underpinning, and where in fact theory suggests that there is a genuine risk that many users might experience a degradation in performance.
Unfortunately there is a second consequential risk, that they may hear the difference and take it for an improvement.
When, for example, a poster claims an improvement in noise performance as a result of improved grounding, I do not hasten to disagree with him, because there is at least some reason to think that he may be correct.
If somebody comes on here and makes a post wherein he says that he has modified a piece of equipment, he has observed a reduction in distortion or other artifacts in it's output, and his subsequent listening tests bear out his measurements, I will be happy to try the modification myself.
If, however, someone comes on here and asserts on the basis of neither instrumented or listening test that he has discovered a beneficial modification of an unconventional nature he can expect to be met with scepticism, since it is well known that enthusiasts are prone to delude themselves.
When the assertions are accompanied by supporting statements such as this, from jkenys first post in this thread,
...which is quite obviously no more than the working of an overactive imagination on limited experience and evidence, then derision is the only acceptable response, lest less well informed and qualified readers are misled.
To prefer instrumented test over observer impression is not, as you suggest, absurd. Can you see the moons of Jupiter? I very much doubt it. Do you dispute their existence on the basis that you cannot see them? I very much doubt it. Their existence is revealed to human senses only with the aid of instrumentation. This is why we use instruments; because they amplify the range of phenomena which we can detect and the accuracy with which we can determine their magnitudes. Otherwise devices such as radios would be designed and tested by people with acute senses at greatly reduced cost.
Hi-fi would not exist were it not for instruments and instrumented test.
w
Because of a well known phenomenon, the placebo effect, it is commonplace to conduct tests involving human reactions to stimuli under conditions known as 'double blind'. This involves the application of a 'null stimulus' to some or all test subjects, depending on the exact nature of the test. Neither researchers nor test subjects are aware which stimulus the subjects are being exposed to at the time of test. Such tests are notoriously difficult to arrange, certainly by comparison with uncontrolled tests.
I for one would welcome a controlled listening test.
I do, however, strenuously object to the introduction of uncontrolled listening impressions as evidence of the efficacy of any modification, particularly where unsupported by instrumented test, and more particularly where there is no real theoretical underpinning, and where in fact theory suggests that there is a genuine risk that many users might experience a degradation in performance.
Unfortunately there is a second consequential risk, that they may hear the difference and take it for an improvement.
When, for example, a poster claims an improvement in noise performance as a result of improved grounding, I do not hasten to disagree with him, because there is at least some reason to think that he may be correct.
If somebody comes on here and makes a post wherein he says that he has modified a piece of equipment, he has observed a reduction in distortion or other artifacts in it's output, and his subsequent listening tests bear out his measurements, I will be happy to try the modification myself.
If, however, someone comes on here and asserts on the basis of neither instrumented or listening test that he has discovered a beneficial modification of an unconventional nature he can expect to be met with scepticism, since it is well known that enthusiasts are prone to delude themselves.
When the assertions are accompanied by supporting statements such as this, from jkenys first post in this thread,
...which is quite obviously no more than the working of an overactive imagination on limited experience and evidence, then derision is the only acceptable response, lest less well informed and qualified readers are misled.
To prefer instrumented test over observer impression is not, as you suggest, absurd. Can you see the moons of Jupiter? I very much doubt it. Do you dispute their existence on the basis that you cannot see them? I very much doubt it. Their existence is revealed to human senses only with the aid of instrumentation. This is why we use instruments; because they amplify the range of phenomena which we can detect and the accuracy with which we can determine their magnitudes. Otherwise devices such as radios would be designed and tested by people with acute senses at greatly reduced cost.
Hi-fi would not exist were it not for instruments and instrumented test.
w
I do apply the same standards. Tasters distinguish differences routinely among wines in controlled tests. If a winemaker claims that a new process is beneficial, there will be controlled tests to back it up. If there aren't, no-one takes them seriously.
Electronics is a lot easier to measure than wine since the outputs are one dimensional. Speakers are a lot tougher and closer to wine in that respect- the outputs are NOT single-valued.
I can't answer for SY (and we often disagree) but I do not find this to be "naive and misleading" at all.
If there is jitter, it will show up in an FFT of a pure tone. Period. Jitter creates inharmonic distortion in music or in pure tones. If there is a change in jitter, that will be seen in the FFT. The only question here is "Is the FFT noise floor low enough?" So far, there has been no evidence that it is not.
I am not one who believes that all amps sound the same, or that passive components live in a Gertrude Stein word (a cap is a cap is a cap). Not at all.
But show me a good FFT of an amp or a passive component that shows the same spectrum thru a usable power range? I'll say "Yes, they sound the same."
No, not all audio gadgets sound the same - but some do. That's life.
|
Pano, your seemingly utmost confidence in 'good FFT's piques my curiosity. Does it also apply to digital products? Would a digital source player sound the same as a different player with an identical FFT? What counts as a 'good' FFT btw? Is there a correct window and no. of bins to be used for instance?
After you mentioned Gary Pimm's system, I went to his website and found he's showing FFTs of various of his amps (or perhaps amp components). Seeing as you said you aspire to his system, do you also aspire to his FFTs? I must say I thought they looked fairly dreadful but I don't doubt for one moment his system sounds as good as you said it does.
After you mentioned Gary Pimm's system, I went to his website and found he's showing FFTs of various of his amps (or perhaps amp components). Seeing as you said you aspire to his system, do you also aspire to his FFTs? I must say I thought they looked fairly dreadful but I don't doubt for one moment his system sounds as good as you said it does.
You don't see the sidebands on your FFT because it is lumped in the 1khz peak. Look at the area under the 1khz peak, see how thick it is, on a device that can measure jitter accurately it would be a single line at 1khz, not a thick mountain peak?
I don't see how you can measure jitter in the pS range with a 192khz (5 uS samples) resolution setup?
If you really believe that both these devices sound the same then I challenge you to listen to them - I will pay for the shipping from SY's & buy the Chinese unit for you.
This "controlled testing" in wine processes improvement that SY mentions is hardly just chemical analysis alone - is a part of it tasting?
Again, SY, I ask - why not produce a plot of a device with known low jitter using your equipment & we can all evaluate that your equipment is capable of resolving this in the FFT? Talking about the low noise floor is not the same as showing an actual PS jitter measurement on your equipment.
Yes, and always under double blind conditions. I eagerly await your controlled listening test results.
What I am saying is the Hanning windowing of the FFT may be masking large sidebands of jitter that are close to the test frequency.
Are you assuming pure sinusoidal jitter that would be predicted to give sidebands quite far from the test frequency? Unforunately the jitter we encounter cannot be assumed sinusidal and not always follow the Rj = 20 log(Jw/4) dB equation.
A very good question. I'll try to give a good answer - it may be a bit long and woolly. ;(
I aspire to the sound of Gary's system, or better. The FFTs are just a tool to get there. Not sure, but I think that the electronics are about 1/3 of Gary's sound, the speakers 2/3rds. Certainly the electronics do not seem to get in the way of the signal. Detail is great, without being "etched". It's a very natural sounding system.
Looking at the FFTs of Gary's Tabor Amp, I see what looks like a clean P-P amp. A nice drop off of harmonics but with the odd order a little high.
http://www.pimmlabs.com/web/solid_state.htm The ideal harmonic drop off would be: (relative to H2) H3 -3dB, H4 -8dB, H5 -15dB, H6 -24dB and so on*. H2 could be as high as about -20dB below the fundamental and this harmonic structure will still sound like a pure tone. The ear works in strange ways. Harmonics that drop off like this are masked, you don't hear them.
Of course nothing is quite so simple. No amp or speaker is perfect, and even low THD can be "out of balance" which will sound strange to the ear. A good system will not only have a harmonic structure that is balanced, and therefore masked, but can maintain that balance into a complex load across its usable power and frequency range. Not an easy task. We don't often see the harmonic structure in manufactures lit. We might see THD at 1Khz into a static load. Some reviews do better tests.
Seeing what the harmonics are doing will give you a good idea of what a particular component sounds like. (Speakers are more complex). Certain harmonic structures give a certain "sonic signature" to a component - just like musical instruments. Warm, cold, harsh, lush, edgy, smooth - most can be traced back to the harmonic structure. Of course the lower the overall distortion, the less you hear the signature unless things are really out of whack.
Gary's posted FFT show me an amp that is well behaved and has the typical P-P suppression (tho slight) of even order harmonics. We listened to his system loud, but it was always clean. You knew it was loud, but it just didn't bother you, no lurch for the volume control, ever. I doubt we ever got over 1/2 watt RMS, though. Maybe we did, it was rockin'.
So for me at least, FFT is a great tool for seeing the harmonic structure of the signal and how the system changes it. Looking at a single frequency at a fixed level will tell you something, but looking at different frequencies at different signal levels will tell you much more.
I hope that helps. Whatever is not clear, feel free to ask, maybe I can do better.
* there is some disagreement on the scale in some studies, but the trend is similar.
Well then, you're postulating that the jitter frequency is less than the linewidth, on the order of a couple of hertz. I find that difficult to believe. I'm not near my lab at the moment, but I don't think the apodization was Hanning.
- Status
- Not open for further replies.