optical out ?

[sfraser]

Maybe I don't understand everything that is involved here, but I often wondered why jitter is an issue at all between components such as a SB2/3 and a DAC or in my case my Bryston Suround Processor preamp. If the physical interconnect's and coax/optical cables create jitter or timing issues, why don't more DAC's have buffer space? With the prices companies are charging for these devices, surely a relatively cheap fix would be to provide sufficient buffer space on your digital input of your "DAC Device", and then re-clock? Is this to simple of a fix?

Scott

[LavaJoe]

Maybe I don't understand everything that is involved here, but I often wondered why jitter is an issue at all between components such as a SB2/3 and a DAC or in my case my Bryston Suround Processor preamp. If the physical interconnect's and coax/optical cables create jitter or timing issues, why don't more DAC's have buffer space? With the prices companies are charging for these devices, surely a relatively cheap fix would be to provide sufficient buffer space on your digital input of your "DAC Device", and then re-clock? Is this to simple of a fix?

Scott

Good question, and I don't know the state-of-the-art in DACs, but the older ones definitely do try to clean up the clock using a PLL (phase-locked-loop). They have to let the source clock (the SB3, e.g.) determine the clock rate itself (in other words, you cannot have an independent clock on the DAC), since small rate differences will eventually over or underflow the buffer. A PLL uses the source signal as the clock rate, but it cleans up the jitter by evenly spacing the samples (in essense, filtering out the jitter).

Audiophiles and stereo salespeople say that it's still a problem, since the more "correction" the DAC has to do to a jittery signal, the more new jitter (or residual old jitter?) is introduced indirectly. If the source is very clean, as the theory goes, the milder amount of cleanup causes fewer side effects electrically and the whole system will produce a cleaner output. The same theory is proposed to explain why even corrected errors from CDs in the transport cause degradation of sound quality - the act of correcting stresses out the system and creates jitter, electrical noise, or whatever.

Not sure I really believe these theories, but that's where the whole thing goes into the realm of "hard to prove."

[sfraser]

Good question, and I don't know the state-of-the-art in DACs, but the older ones definitely do try to clean up the clock using a PLL (phase-locked-loop). They have to let the source clock (the SB3, e.g.) determine the clock rate itself (in other words, you cannot have an independent clock on the DAC), since small rate differences will eventually over or underflow the buffer. A PLL uses the source signal as the clock rate, but it cleans up the jitter by evenly spacing the samples (in essense, filtering out the jitter).

Audiophiles and stereo salespeople say that it's still a problem, since the more "correction" the DAC has to do to a jittery signal, the more new jitter (or residual old jitter?) is introduced indirectly. If the source is very clean, as the theory goes, the milder amount of cleanup causes fewer side effects electrically and the whole system will produce a cleaner output. The same theory is proposed to explain why even corrected errors from CDs in the transport cause degradation of sound quality - the act of correcting stresses out the system and creates jitter, electrical noise, or whatever.

Not sure I really believe these theories, but that's where the whole thing goes into the realm of "hard to prove."

Not sure I believe them either . IMO, if you design a external DAC with X amount of buffer space, you should be able to eliminate all external jitter (jitter being time variances between data samples arrival at the DAC input). Now once that sample is buffered you should be able to stream it internally to the DAC, if it is designed correctly, a very controlled environment therefore not introducing any new jitter (or very little). In my mind a simple buffer eliminates all external jitter and any problems it may cause.

Regarding over/under flow of buffer space. I'm not sure that i follow. Assuming the SB2 is the source it's providing the clocking for the data arrival to the DAC device buffer . Even if it was clocking at bit fast, a 64 M buffer will more than handle it in all residential applications. 64 M of RAM purchased in bulk commercially can't run you more than a few bucks. Less than the gold leaf writing on the front of the DAC device (but thats marketing a real black art) .However as I stated earlier, maybe I don't comepletly understand the problem.

[LavaJoe]

Not sure I believe them either . IMO, if you design a external DAC with X amount of buffer space, you should be able to eliminate all external jitter (jitter being time variances between data samples arrival at the DAC input). Now once that sample is buffered you should be able to stream it internally to the DAC, if it is designed correctly, a very controlled environment therefore not introducing any new jitter (or very little). In my mind a simple buffer eliminates all external jitter and any problems it may cause.

Regarding over/under flow of buffer space. I'm not sure that i follow. Assuming the SB2 is the source it's providing the clocking for the data arrival to the DAC device buffer . Even if it was clocking at bit fast, a 64 M buffer will more than handle it in all residential applications. 64 M of RAM purchased in bulk commercially can't run you more than a few bucks. Less than the gold leaf writing on the front of the DAC device (but thats marketing a real black art) .However as I stated earlier, maybe I don't comepletly understand the problem.

There were a couple of recent threads about the idea of drift over time when clocks don't match:

http://forums.slimdevices.com/showth...=forever+drift
http://forums.slimdevices.com/showth...ighlight=drift

You may be right in that it may take a very long time to run out of buffer with small clock differences (and this may be far smaller than album length, giving a chance to resync periodically). But this sounds kludgy, and from a purist standpoint, if the DAC drifts from the source, the display of time, etc. on the source unit will be further and further away in time from what you are hearing. Another issue is that you'd want to full the buffer to the 1/2 way point when you start the music, so this would cause a delay proportional to the buffer size.

I think that if there is a way to keep from accumulating drift and still get good results, it's preferable not to use the large buffer/two clock method.