PhilNYC;153822 Wrote: > Unless a digital system is using a master/slave clock architecture, it > needs to use a PLL to synchronize the transport signal to the DAC. A > PLL can reduce jitter in the incoming signal, but has some of its own > inherent jitter if the clock in the transport has any difference from > the receiving clock. A FIFO buffer can help reduce jitter in this kind > of design, but it cannot eliminate it. Do a Yahoo/Google search on > "asynchronous jitter" and see what comes up...this topic has been > researched at length by the telecom industry, military, and yes, the > audio industry. Pretty much all the credible research says that > re-clocking does not "eliminate jitter", and in some cases it says that > the act of re-clocking can sometimes ADD jitter. >
I think we are talking about two different things. If you read above in this thread, there was a claim (by ezkcdude) that ASRC could significantly reject jitter (or something, I'm not going to go back and check his exact wording). He gave an excellent reference written by a chip designer which describes in great detail (it's about ten pages long) how that works. In a nutshell it is a PLL, but in the digital domain and after upsampling by a huge factor, which allows it to have a bandwidth of around 3 Hz. Apparently that is much better than a more standard approach where one uses a PLL to recover the clock from the S/PDIF directly. However it is also clear that it doesn't not entirely elminate jitter (since it's still using the S/PDIF edges to construct the clock). If done according to his description it's not going to add jitter, except possibly to a signal with very very low jitter to begin with (and then it might increase it slightly). However, there is another, entirely different approach possible, which - as far as I can see - completely and totally eliminates the effects of input jitter. This does not use a PLL at all, because it does not reconstruct the clock from the incoming data stream. As I said above, simply imagine having a huge buffer in your DAC. Now run the audio stream for, say, one hour (the length of a CD). Record the entire thing in the buffer. You now have it stored as a digital sequence which (barring bit errors) is identical to the sequence on the CD, and has nothing at all to do with any jitter in the S/PDIF signal that carried it. Now, after waiting one hour, you get to listen to your jitter-free music as the DAC plays out the data, using its own internal crystal clock (which can be extremely clean). Not very convenient, because you had to wait so long, but this totally eliminates the effects of transport jitter (if not, I'm waiting for someone to tell me why). Now since this is rather inconvenient, you can be more clever and reduce that initial pause to a nearly imperceptible one, and that's what the Lavry does (according to their white paper). This does not use a PLL because it does not reconstruct the clock from the incoming S/PDIF - it uses its own clock - and therefore I fail to see how it can be affected by jitter. -- opaqueice ------------------------------------------------------------------------ opaqueice's Profile: http://forums.slimdevices.com/member.php?userid=4234 View this thread: http://forums.slimdevices.com/showthread.php?t=29450 _______________________________________________ audiophiles mailing list audiophiles@lists.slimdevices.com http://lists.slimdevices.com/lists/listinfo/audiophiles
