This would be nice if it were true, and make digital audio design a LOT easier.
Jitter is added to the Master Clock by every gate and flip-flop that selects, buffers and divides-down the Master Clock. These add jitter dependent on the di/dt of the power subsystem and the rise-time of each gate. Jitter is added by crosstalk and ground-plane noise from other circuits. Jitter is added by losses and dielectric absorption in conductors and cables, including FR4. Jitter is added by reflections on transmission-lines that arrive at the source driver before the edges have completed transitioning. And on and on......
Steve N.
Empirical Audio
The typical jitter in a flip flop (unlelss it is a very messy slow technology) is totally nil WRT audio. All of the actual jitter is caused by the last flip flop, because if there was a sampling problem with an edge triggered or latch flip flop before the end, then it would be an error -- and that is worse than jitter. The jitter in-between is hidden by the sampling process of the clock. The only real jitter is WRT the timing clock at the very end of the chain. The digital audio signal itself ion the disk is self-clocked and clock is tied directly to the signal with nothing but an error as an alternative. Tha dead pefect timing is just resynced wit a FIFO, and reclocked to the system clock rate. If the drive rotational timing deviates too far, then you get an error -- not jitter. This same kind of thing has been done in primitiave video recorders since color came in to being. Since digital techniques came in -- the signal is so clean that there is a choice between the signal being 100% accurate or an error (I am speaking of digial video decks.) Digital video cannot withstand much jitter at all when compared with audio because the frequencies and timing is so fast in comparison. Even an HDMI connection is more stringent than CD stuff.
* A simple existence proof about my position on the jitter matter where the last timing jitter in a functioning digital system (where the input is self-clocked) is the only important jitter: My very sophisticated compressor/expander system, where the connections between the threads are effectively fifos that can asynchronously fill up and empty in very crazy ways, but still produce the same timing as if they were a stationary, repeatable delay. The reason is that the output is queued and clocked -- and all of the compoenents of the signal going into the processor do come out at the same time. Even the time delay between the input and output is always a constant fixed time -- having nothing to do with the crazy interactions between the inter-thread queuing and the operating system scheduling. This crazy queueing behavior is infinitely worse than any normal hardware that we might be talking about on a CD rom or music CD. Essentially, the input signal is self-clocked, and the output clock is driven by the program. The only difference on a CD is in the need to speed up or slow down the rotating medium -- but in that case, the rotation is a kind of fifo in its own right, and as long as that weird HW buffer isn't overrun (essentially that is an error), then the signal will be 100% intact with no jitter resulting from the rotating media. All of the buffering is known as 'elastic.' If there is an error reading the medium where the clock and the data are too noisy to use, then theat wouldn't be jitter, per se -- but would be an error that might be corrected by ECC. On things like CDroms, the data is the clock and the clock is the data -- in that they are not seperate things. This is (in a way) similar to a modulated RF signal -- you really cannot logically have the modulated data witihout the RF signal -- you either have both or neither. The same is true of the clock and data on CDs, laser disks, and most other rotating media nowadays.
If there was that many jitter problems in current technology (anything near current), we probably wouldn't even have 486 processors, let alone what we have today. Internally, such processors typically have massive resychronization and crazy clocking schemes to minimize power/etc. Geesh, we make radio receivers at over 100MHz with 14dB dynamic range (before performing tricks to get amazing SNR) out of the same kind of technology -- and our HDTVs have similar kinds of technology also -- and they are far, far more stringent than audio stuff. In fact, even the old NTSC color system was so fragile that even a degree of error at 3.58MHz produced bad color noise (that is why cheap VHS VCRs had bad troubles with color -- yet similar technology gave me my D9 decks which were little different than a digital frame store in accuracy. If we cannot do audio without significant jitter, then any kind of quality video would have been too much high tech (which it wasn't.)
Nowadays, the average consumer can have studio quality HDTV -- and lots of reclocking at hideous speeds (e.g. memory modules.) Jitter is NOT an issue. True data errrors ARE the issue.
Of course, there is such a thing as jitter, but it doesn't manifest in the way that some audio people might think. Again -- the only real jitter that is audible(except for true data errors in between) is the last jitter source in the chain, and it is NOT likely to even be detectible at that point.
IMO, most of the differences are in the DAC quality, different sample rates, different op-amp designs (op amps are really big sources of all kinds of odd timing effects), and too many things to count. A digital cable makes no TECHNICAL differences execpt when it is so bad that it causes errors. An analog cable can make some difference -- but sometimes the fancy designs cause more trouble than even zipcord for speakers (some cables used to have too much capacitive loading for some amplifiers, just for example -- probably/hopefully fixed today.)
On the other hand -- as I wrote above, I believe that if someone really believes that something sounds better -- then it really does sound better to them. I cannot argue that point, and will always support people when they make claims about their perception. Frankly, I have SOMETIMES found that signal defects sound good -- but of course, such situations cannot always be sustained when playinjg different material. IMO, it is always best to have the simplest flattest/lowest noise/least distortion design, and then change it from there for personal taste...
You know, that food sometimes tastes better in a restaurant than at home, even if the food is the same. For me, if I am at a fish market -- NOTHING tastes good, even a wonderful steak. We have both our environments and psychological needs, but the technical world is different than that. I live in the technical world, and sometimes have troubles realizing what people are really saying about how they feel. Once I realize that, then I am happy -- becuase I have made the technical argument known, and the REAL reality is how the music makes us all feel. Audio often has both the 'feel' of a salon and the real technical stuff. Conflating the two is not a good thing (IN MY OPINION.)
John