Thing is, and the reason I don't mess with Digital stuff is you aren't comparing apples to apples there. Even a little bit of extra LRC can affect slew rates, so getting to a point where you are sure both clocks are connected optimally. Many confounders to trip people up who don't have suitable test equipment (like me).
Bill,
Yes, I know. If I was going to a serious listening test I would put both clocks on the same board, power them the same, etc. (Can't use Accusilicon either, since I don't want to use 45MHz - 49MHz clocks if I can help it. That's okay, there are lots of junk 22MHz - 24MHz clocks that should be easy enough to distinguish from a good clock.)
I would probably also build a test box with some buttons, LEDs, and at least one switch. All would interface to a microcontroller that would run training and testing sessions, randomize clock selection, and tally scores.
However, before I queue that on my list of soon to work on projects, I need confirmation that if I can show one or a few people who have no trouble blind testing clock differences accurately, that should be enough to prove it is humanly possible.
If no agreement on that, then I think trying to provide some proof is something maybe to consider working on in the more distant future.
Yes, I know. If I was going to a serious listening test I would put both clocks on the same board, power them the same, etc. (Can't use Accusilicon either, since I don't want to use 45MHz - 49MHz clocks if I can help it. That's okay, there are lots of junk 22MHz - 24MHz clocks that should be easy enough to distinguish from a good clock.)
I would probably also build a test box with some buttons, LEDs, and at least one switch. All would interface to a microcontroller that would run training and testing sessions, randomize clock selection, and tally scores.
However, before I queue that on my list of soon to work on projects, I need confirmation that if I can show one or a few people who have no trouble blind testing clock differences accurately, that should be enough to prove it is humanly possible.
If no agreement on that, then I think trying to provide some proof is something maybe to consider working on in the more distant future.
Last edited:
It seems to me, if I remember correctly, at least two people saying words to that effect here very recently.
Indeed they have not validated or denied anything.
The measurements simply cannot explain what happens in the conversion process due to timing errors.
I think in your right mind you know that every DAC sounds different from the others, sometimes shades sometimes huge difference.
Therefore anyone can distinguish one DAC from another, although they perform similar THD and SNR.
130dB THD or 100dB SNR are not more than standalone numbers that don't help to explain how our brain perceive the sound.
Can you hear -130dB of THD?
No, you cannot.
Does this mean that the device is perfect?
Again no, merely THD measurements does not help to explain what we perceive.
If this were not the case, not only would the DACs or amplifiers sounded the same (we cannot hear such low level of distortion), but would also be perfectly capable to reproduce the musical event in an absolutely realistic way.
This is not the real world.
Don't forget that you are measuring electrical signals at the output of a DAC or an amplifier.
Not the electrical signals that the ear sends to the brain which processes them to represent what we hear.
There is a big difference.
And the temporal aspect is one of the most unknown.
There is Temporal Auditory Acuity, but I'm guessing that's not what you mean? You also need to ask yourself how these temporal aspects are getting to your brain in the first place.
Last edited:
I believe it is possible hear hear IMD in music caused by a pretty low level nonlinearity. Experiments with ES9038Q2M harmonic distortion correction registers showed it could a surprisingly small change in linearity. FFT measurements gave me an estimate down around -120dB when measured as HD.
Shouldn't be surprising but people keep forgetting I once sorted unity gain non-inverting audio opamp buffers in order of distortion by ear, double blind. I also said it was damn hard, which is was.
In addition, some people don't have trouble hearing the difference between dither/undithered CDs. ESS found that some people could hear very low level noise floor modulation in dacs, etc. Some people seem rather dogmatic about ignoring or dismissing such things.
Opinion: Maybe the fundamental problem with deniers is that hearing threshold measurements with single sine wave tones suggest all the above must be impossible. Some people seem to want to believe that human hearing in linear enough and stationary enough to be accurately modeled as no more than weakly nonlinear. If linearity holds, then those sine wave measurements should apply to all other listening experiments. Clearly, hearing is more than weakly nonlinear in a way that is conceptually not the same as with nonlinear electronics. That is to say, much more nonlinear processing occurs in the brain than occurs in analog audio electronics. Bottom line, sine wave thresholds don't apply to all situations.
Shouldn't be surprising but people keep forgetting I once sorted unity gain non-inverting audio opamp buffers in order of distortion by ear, double blind. I also said it was damn hard, which is was.
In addition, some people don't have trouble hearing the difference between dither/undithered CDs. ESS found that some people could hear very low level noise floor modulation in dacs, etc. Some people seem rather dogmatic about ignoring or dismissing such things.
Opinion: Maybe the fundamental problem with deniers is that hearing threshold measurements with single sine wave tones suggest all the above must be impossible. Some people seem to want to believe that human hearing in linear enough and stationary enough to be accurately modeled as no more than weakly nonlinear. If linearity holds, then those sine wave measurements should apply to all other listening experiments. Clearly, hearing is more than weakly nonlinear in a way that is conceptually not the same as with nonlinear electronics. That is to say, much more nonlinear processing occurs in the brain than occurs in analog audio electronics. Bottom line, sine wave thresholds don't apply to all situations.
Last edited:
This is precisely the point.
I don't know how these temporal aspects are getting to my brain.
Then I don't know how to measure these aspects.
So I can only rely on what I perceive by comparing it with the live musical event.
Now now Mark we've had this before. You scored the same opamp in 2 different positions.
But if that was damn hard it was still a significantly higher distortion level than the clock issues we are talking about here.
Not only, since you can get different sound quality with the same speakers and replacing the DAC.
So IMHO it's the timing errors in the digital to analog conversion that affects what I'm perceiving.
Do not mix up THD dB with phase noise dBc, there is no relation.
You cannot convert phase noise in THD or SNR.
Unfortunately I don't know.
I can only hear, measure and then speculate about a relation between what I have perceived and measured.
The measurements help to speculate but they do not clarify the problem.
I have said infinite times I cannot demonstrate the relation by measurements or mathematical formulas.
Disagree. I gave up on the last one. My process required more or less bubble sorting each opamp up by comparing it with each of the other opamps already sorted. At that point the duplicate opamp was on the bottom of the list. I said enough is enough at that point and called it quits. My ears were ringing enough, and the mental effort was exhausting. I thought someone else would nail it so wouldn't matter if I quit. Anyway, might be easier now with a much better dac, and I still have the files
Regarding which is harder, by far the opamps were harder. What happens in dacs is that RF stuff gets aliased and or convolved down into the audio band where the dac presents it in a much more obvious way. Nobody claims to be directly hearing femtosecond clock disturbances, the only claim is about the eventual audible consequences. Those are pretty easy to hear. Some of them require looking at interchannel correlation though, that's where imaging effects are occuring. Don't hear those in mono.
Last edited: