Yes, random jitter isn't really of much interest. But thanks for providing the link.
Then your playback system is really no better than those used in recording studios for monitoring? From the way you've described yours, I'd hazard its considerably worse. On my el-cheapo system I can hear which recordings are pants and which aren't. One example of a pants recording is:
Amazon.com: The Very Best of MTV Unplugged: Various Artists: Music
Fortunately that's the exception not the rule.
Yes. What you write about that forum is true of life in general - for the large part people's lives are ego-driven. Of course that's not invalidating people's lives, any more than it invalidates the evidence and arguments presented on that forum to say its ego-driven.
So those of us who have worked in AF design are down to just the one braincell then?
Flawed premise leads to flawed conclusion.
You evidently don't understand what is meant by data-dependent jitter.
Data-dependent jitter is non-periodic. Trivially, if the signal is a square wave then there can be no data-dependent jitter. If the transitions are pulled out of place by a burst of ones or a burst of zeros, this is called data-dependent jitter. It might have frequency content in the case of a continuous recorded tone of constant amplitude, but not in the case of music. It's not described as random because of it's relationship to the data, but because the data itself is effectively random in terms of bursts of ones and zeros, it's random in effect.
You have described my argument as 'straw man', but as far as I can see you have no argument to offer at all. JosephK himself was at a loss to describe a mechanism whereby this jitter could be audible. The fact that it is there does not correlate it with any audible effect. You are the one suggesting that it is audible, it is incumbent on you to demonstrate that it is.
Please bear in mind that this is your obsession, not mine. I have taken the time to research at least some evidence which is relevant. Before you start accusing me of trolling, produce for me at least one piece of evidence which shows that all the effort on your part is not going into producing hot air.
w
Data-dependent jitter is non-periodic. Trivially, if the signal is a square wave then there can be no data-dependent jitter. If the transitions are pulled out of place by a burst of ones or a burst of zeros, this is called data-dependent jitter. It might have frequency content in the case of a continuous recorded tone of constant amplitude, but not in the case of music. It's not described as random because of it's relationship to the data, but because the data itself is effectively random in terms of bursts of ones and zeros, it's random in effect.
You have described my argument as 'straw man', but as far as I can see you have no argument to offer at all. JosephK himself was at a loss to describe a mechanism whereby this jitter could be audible. The fact that it is there does not correlate it with any audible effect. You are the one suggesting that it is audible, it is incumbent on you to demonstrate that it is.
Please bear in mind that this is your obsession, not mine. I have taken the time to research at least some evidence which is relevant. Before you start accusing me of trolling, produce for me at least one piece of evidence which shows that all the effort on your part is not going into producing hot air.
w
Last edited:
Any reference you have for this? I'd appreciate a headsup - sonics of jitter are less of an issue IME than common-mode RF.
It would be related - via the transfer function of the PLL. Waki's paper points out that random jitter is inaudible way beyond the performance of the average SPDIF receiver IC. So that leaves ISI-induced jitter to be explored.
Inapt isn't quite right, it would just indicate that jitter on the SPDIF would have to be higher still to have more than 250nS residual. Given that the jitter we're talking about is tens of pS, one could reasonably hope it would be less than this at the DAC, and additionally as JK points out, it is high frequency, where the filtering effect of the PLL is most effective.
w
w
Yes which leads him to seriously question whether this whole jitter thing is perhaps a red-herring and that the sonic differences are something else. Like common-mode RF for example. This (CMRF) would also explain reports of USB cables sounding different when async USB is used - no means there for jitter to affect the sound.
I have learned a lot from this thread.
Tubelab, I don't personally know anybody who would do that but I am sure there are 11 meter guys who would do anything if you gave them the idea that they could tear up more equipment over a longer distance by doing it. I have seen car trunks full of batteries and big capacitors just to give them the power they need to ruin everything electrical for everybody they possibly can and go mobile with it.
Tubelab, I don't personally know anybody who would do that but I am sure there are 11 meter guys who would do anything if you gave them the idea that they could tear up more equipment over a longer distance by doing it. I have seen car trunks full of batteries and big capacitors just to give them the power they need to ruin everything electrical for everybody they possibly can and go mobile with it.
Old stuff. I've read it in the past (dont feel like reading it again).
Do they say which was the MINIMUM jitter to which the listeners compared?
I bet it was 50ps or higher. In short, the listeners lacked a good reference.
Newer test have showed that jitter in the 5/10s ns range is indeed audible.
BTW RF and other noise creates RANDOM jitter, which is harder to measure but easier to spot by the ear (= worse).
I don't know how you'd know it'd be random. That's certainly one possibility but its not the only one. Also common-mode RF's manner of affecting the sound is not necessarily via corrupting the clock. Its just as likely to get into output stage opamps and drive them into non-linearity.
Just to clear something up, the correct way to control ringing overshoot when doing signal integrity work etc is to use terminations, paralell, series or thevorin, end, middle etc. It may be a minor point but using attenuation is an incorrect term
You terminate to match impedences, you terminate to absorb excess nenergy put into the line, blah blah blah.
An excellent reference on how to terminate properly is Chapter 6, High Speed Digital Design, Howard Johnson.
A great way of reducing these problems is to use a signal rise time that is comesurate with the clock speed, ie one that is not to fast, in fact the slower the better.
USB cables sound different! I'd like to see some shots of waveforms from both the transmitor and reciever end.
You terminate to match impedences, you terminate to absorb excess nenergy put into the line, blah blah blah.
An excellent reference on how to terminate properly is Chapter 6, High Speed Digital Design, Howard Johnson.
A great way of reducing these problems is to use a signal rise time that is comesurate with the clock speed, ie one that is not to fast, in fact the slower the better.
USB cables sound different! I'd like to see some shots of waveforms from both the transmitor and reciever end.
In regard to SPDIF its not the case that 'the slower, the better'. Reducing the rise time means greater ISI and it follows that the data-dependent jitter will increase. Hawksford wrote in quite some detail about that in his paper in collaboration with Julian Dunn.
As with most things, there's a compromise to be struck between the bandwidth and the jitter.
You'll presumably be only inspecting the differential mode noise if looking with a scope and hence miss the probable root cause of the sound degradation.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.