to oldschool1948
Steve has more bench experience with these so I wont even comment on his figures or claims. But I’ll provide a couple of data points that are useful to anyone trying to digest why jitter is worse in A than B.
First, SPDIF and its variants (including AES/EBU and toslink, which are merely physical layer manifestations) have the DAC as the SLAVE to the MASTER clock in the sending device (transport, CD player, adapter, streamer, whatever).
USB and Ethernet are asynchronous. the sender sends bits to a buffer until told to hold on... and the DAC clocks things out on its own, using whatever clock circuitry it has.
Beyond clock specs, we have to deal with detection of that clock, which might be on the lead (cab;e) on the internal bus, whatever. Lead lengths may vary, thresholds for detecting a clock pulse may vary, noise may interfere -- many functions can interfere with timing by small (to us) but large (to the music) amounts. Remember that the entire sample time slot for a 44k signal is 20 mSec (1/44000 sec) - yea simplified for illustration.
I suspect noise is a significant contributor since linear power supplies seem to improve sources in subjective testing and we KNOW (due to error correction) when there are errors and that the base clock was the same. I conclude that this ids also why the same signal over AES is reported to sound better than over Coax. We KNOW that optical toslink has a big jitter component, but the advantage of ground isolation (due to, well, no ground)
G
Steve has more bench experience with these so I wont even comment on his figures or claims. But I’ll provide a couple of data points that are useful to anyone trying to digest why jitter is worse in A than B.
First, SPDIF and its variants (including AES/EBU and toslink, which are merely physical layer manifestations) have the DAC as the SLAVE to the MASTER clock in the sending device (transport, CD player, adapter, streamer, whatever).
USB and Ethernet are asynchronous. the sender sends bits to a buffer until told to hold on... and the DAC clocks things out on its own, using whatever clock circuitry it has.
Beyond clock specs, we have to deal with detection of that clock, which might be on the lead (cab;e) on the internal bus, whatever. Lead lengths may vary, thresholds for detecting a clock pulse may vary, noise may interfere -- many functions can interfere with timing by small (to us) but large (to the music) amounts. Remember that the entire sample time slot for a 44k signal is 20 mSec (1/44000 sec) - yea simplified for illustration.
I suspect noise is a significant contributor since linear power supplies seem to improve sources in subjective testing and we KNOW (due to error correction) when there are errors and that the base clock was the same. I conclude that this ids also why the same signal over AES is reported to sound better than over Coax. We KNOW that optical toslink has a big jitter component, but the advantage of ground isolation (due to, well, no ground)
G