< 0.5ps jitter for digital audio: nonsense, or am I missing something?

[nctnico]

That jitter is not even related to the samples themselves but to the bits. Now, you have a 24 bit 192khz stereo system, all the data goes through I2S, that is about 9 megabit/s. That 0.5 ps jitter per bit suddenly becomes 5ppm of your bit. If you take the noise analogy, -106dB, almost audible range, so I would say, he is right.
It is actually not as simple. I have the impression, that engineers oversimplify audio all the time. Jitter is changing the frequancy of the signal, so you cannot make that simple calculations.

Now you are going off the rails. There is a simple formula to determine the amount of jitter you can tolerate on a sampling clock to achieve a certain signal to noise ratio (dynamic range). The more bits (dynamic range) and the higher the frequency the less jitter you can tolerate.

[dmills]

Who cares bout jitter on the bclk, as long as worst case it meets setup and hold timing everything is fine (Digital is nice like that).
 
In an R2R or similar simple minded DAC it is jitter on the LRClk that makes it to the output, but that signal is fairly slow so it is clearly going to be mainly additive jitter from the divider chain here assuming a suitable VCXO (Phase noise drops by 6dB with every divider stage).

In a delta sigma design (Most of which are not in fact 1 bit in any meaningful sense) it is the modulator clock that is critical and this is actually harder because it is much faster, 12Mhz or so typically against 48KHz or so for the LRclk.

All of this is standard stuff and the equations are well defined for any given performance target, the debate is not one of engineering but of the human physiology and what the targets for system performance should be (A FAR more interesting question, and one warranting actual research). 

The time domain is getting some play recently as there is some evidence that the ear may have a distinct set of hardware that detects transients and this may actually have better temporal resolution that the ~20KHz bandwidth tone detection capability would imply, the jury is out, we will see as some more experiments are done.

Regards, Dan.

[Someone]

http://thewelltemperedcomputer.com/KB/BitPerfectJitter.htm seems good,
about halfway down actual peer reviewed human listening test paper's results are shown 10s-100s of ns with random jitter, music

maybe few ns with structured jitter and test tones

a common confusion is that Adams, Dunn calculate jitter degrading S/N at the 16 bit lsb level for 20 kHz full amplitude sine - pure engineering calc - not a listening test verified number

Good collection of information, listening tests comparing unfamiliar music lend toward less critical listening since its hard for the listener to have any baseline reference and recorded music is so dense with information. There will remain a big gap between what can be audible in extreme corner cases and what is noticeable in real world use unless the tests are trying to provoke the known limitations.

All of this is standard stuff and the equations are well defined for any given performance target, the debate is not one of engineering but of the human physiology and what the targets for system performance should be (A FAR more interesting question, and one warranting actual research).

Exciting, fun, and expensive.

[amspire]

When I was playing around with the Magic Sinewaves PWM, I needed nanosecond accuracy of the pulse widths (including jitter) to get below the 0.01% distortion, but that was different - that was needed to make the harmonic cancellation work. That was with a 1 bit DAC. But even with my lousy 1 bit DAC, 0.5ps jitter and pulse width accuracy would get me well below 0.000001% theoretical distortion. I couldn't make the hardware to get close. You would probably need to run the audio preamplifiers at around 100VAC signal level just to get far enough above the noise floor.

Anyway, wouldn't you want a spread spectrum clock to minimize the effect of any intermodulation artifacts?