Steve, that was an example appealing to creative, not an arithmetic thinking.
What many of us relates with live-like reproduction, sits in low-level microdetails, even at the levels relevant to distortions of passive components.
You are afraid of a low level signal suppressed by large steep transient. You have already said that. The only reason for this to happen would be low slew rate. And slew rate is best disclosed with large level signal.
You know that human perception of "low level details" may be improved by addition of noise?
What do low-level microdetails in music have to do with 10uS pulses?
And what exactly is a "microdetail"?
Sorry, but sounds like gibberish to me.
se
I find it not possible to object you in this conclusion, possibly excessive slew rate removes main trobles related with the mechanism of microdetailes suppression by fast big transients. But all listenable effects of parts, wires, insulation, connectors, PCB and case materials, noise, etc., they do remain. A way to measuring of these effects was shown by Ed Simon. One can not state, that fastest unit is the best one.
Resistors are listenable, wires are listenable, Ed Simon measured effects caused by them, he could give a hint to your question. I would not take this responsibility for me.
John Curl said:
I know sine waves get a bad rap these days as a test signal, but are they all that easy and predictable? They are constantly changing, can go thru all the levels if you want and cross zero twice per period. The good old swept sine isn't even consistent on timing. Seems like a pretty good test signal. There are other useful signals, for sure - but are sines all that easy and useless?
Hi,
My Sanyo Washing machine says "Fuzzy Logic Control" on the control panel and is from the oughties. Given how erranious it tends to interpret my intents I feel the description is apt...
Fuzzy Distortion in an audio sense is a distortion that is transient in nature; that is it is not caused by a transient, but may be, the key is that changes with time and signal; to be clear the transfer function is modulated in a significant way by a mechanism that changes it's nature in a semi-random way.
In some way the classic "thermal tails" in badly designed solid state stuff are another similar distortion (also popularised by Héphaïtos as Memory Distortion), yet in this case quite distinct from it, as the cause is not thermal.
To me it qualifies under this heading, as with many (most) converters this "fuzziness" is actually signal (history) dependent and pseudo-stochastic.
Really? Interesting you would say that.
You may be right. So far switching amp's I encountered were pretty dire. I have a strong commercial incentive in finding something that can at least beat a decent Class AB Solid State Amp, so far I have come up empty handed. Some are not as bad as others, but the best still fail to beat my "real-world check" Amp, which is a Gainclone (so by all acounts one of the lowest lifeforms in fidelity...
Of course, what do you expect from a two or four level Delta Sigma modulator running at maybe 4 to 8 times the sample rate of CD. Come, next to that that DSD is pure gold, dire as it is (next to 24/176.4 PCM).
Ciao T
My Sanyo Washing machine says "Fuzzy Logic Control" on the control panel and is from the oughties. Given how erranious it tends to interpret my intents I feel the description is apt...
Fuzzy Distortion in an audio sense is a distortion that is transient in nature; that is it is not caused by a transient, but may be, the key is that changes with time and signal; to be clear the transfer function is modulated in a significant way by a mechanism that changes it's nature in a semi-random way.
In some way the classic "thermal tails" in badly designed solid state stuff are another similar distortion (also popularised by Héphaïtos as Memory Distortion), yet in this case quite distinct from it, as the cause is not thermal.
To me it qualifies under this heading, as with many (most) converters this "fuzziness" is actually signal (history) dependent and pseudo-stochastic.
Really? Interesting you would say that.
You may be right. So far switching amp's I encountered were pretty dire. I have a strong commercial incentive in finding something that can at least beat a decent Class AB Solid State Amp, so far I have come up empty handed. Some are not as bad as others, but the best still fail to beat my "real-world check" Amp, which is a Gainclone (so by all acounts one of the lowest lifeforms in fidelity...
Of course, what do you expect from a two or four level Delta Sigma modulator running at maybe 4 to 8 times the sample rate of CD. Come, next to that that DSD is pure gold, dire as it is (next to 24/176.4 PCM).
Ciao T
The first problem is generating a perfect sine wave. You can come quite close but you will never get there.
Then when you go to analyze what comes out you again have a limit to precision comparing what went in to what comes out.
You can do this to quite an impressive level, but the issue is if that level is surpassed by what a "Golden Ear" can really discern.
The idea is to create a test that makes it easier to see some particular deviation from perfect.
A complex waveform may yield a simple and easy to understand answer. The best example of this is using Bessel functions to determine FM modulation. Which also suggests a possible test for phase modulation due to amplitude changes.
Hi,
Sines in themselves are fine. The issue these days is that we average ton's of cycles of them so any small stochastic or pseudo-stochastic signal changes are in fact averaged out.
Ciao T
Sines in themselves are fine. The issue these days is that we average ton's of cycles of them so any small stochastic or pseudo-stochastic signal changes are in fact averaged out.
Ciao T
I am very happy with this remark, as I have posted before, also the human sensory system adds copious amounts of noise at the threshold of detectability. When it is almost dark, you see noise, when it is almost silent, you hear noise. This pushes small signals over the threshold.
A reason I post this again is to go a bit more into my earlier remark why class D's fuzzy switching remnants may be beneficial to sound. All drivers have hysteris, 'stickiness'. By constantly feeding a high frequency left-over through them, this stickiness may be alleviated, since the drivers are always wriggled around ever so minutely. No way for me to measure, but it could explain the excellent small signal reproduction I heard from class D. First Watt and so on.
In short, noise in itself maybe not a bad thing, depending on the spectrum and level of course.
To me, this is quite clear at qualitative level, and I realize how much it is more difficult to provide some quantitative measurements and proofs.
But, unfortunately, most probably, nothing will change in education and common practice, until some extraordinary person will spend all his efforts to resolving the problem down to everybody's understanding. Until this has not happened, thousands and thousands of professionals and DIYers will be happy with simulations and spectra measurements.
Last edited:
Let me speculate a bit:
1) sine is much easier for reproduction at high frequencies, than any other kind of periodic signal (compare how much square and sine are distorted at 1 MHz or 20kHz, depending on amp), it brings minimum information, and in this respect, has almost nothing in common with information reach musical signals
2) information percepted from musical signal is not lost (only a bit colorated) if some harmonics of high-level signal components are added, but sine-based tests give no chance for judging about possible losses of information
3) ??? anybody else
Last edited:
Either that, or more people will realize that listening evaluation is more revealing than conventional measurements.
- Status
- Not open for further replies.