PhilNYC;153834 Wrote: > This is essentially a FIFO buffer (my Dodson DAC does the same thing, > albeit with a much bigger buffer than the Lavry), and in this > architecture, you need to deal with things like buffer > overflow/underflow, because the data is still "streaming"...and this > still requires some "cooperation" between the incoming clock and the > re-clocker. This will still contain some inherent jitter. >
OK, so we agree that loading it all into memory eliminates jitter entirely. The next question is whether the Lavry (which I use as an example because it's the only one I've read the white paper for) accomplishes this as well. What they do is use a CPU to check the state of the buffer. If it's getting full, they increase the speed of the clock (which controls the output); if it's emptying out they decrease it. This keeps the buffer around half full all the time. They make this (tiny) adjustment every ten seconds or so for a typical timing mismatch. Now, while it's possible (but implausible) that this adjustment process could have an audible effect on the sound, it's still got nothing to do with, and is totally independent of, input jitter. It depends only on the difference in _average_ clock rates, not when each individual rising edge arrives. In your language, cooperation between the two clocks is not necessary - you could just check the buffer at regular intervals, and if it's, say, greater than 3/4 full or less than 1/4 full make an adjustment. -- 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