Master DAC for a Slave soundcard

[Irakli]

Hello,

Title says it all, I would like to design a DAC with its own clock and slave everything else to it.

I searched the forum about the subject. There is a lot of talk but I could not find anything completed and practical. Also, all the bits and pieces are here too. We just need to bring everything together.

Here is the simple rationale behind this project:

1. All music nowadays is on the computers and people who care have decent sound cards. All decent sound cards accept external clocking. Some accept Word Clock and some take clock on S/PDIF.
2. If we sync DAC and sound card, we would not need FIFO buffer at the DAC input.
3. precision master clock next to the DAC reduces jitter (so it is said).

I want this thread to be strictly about the practical design aspects of this particular DAC. Here is the plan: we agree on specifications and assess whether it is doable, than components, circuit topology, etc. The final product would be a PCB.

I would like to know if there is an interest among the forum members and if so I will publish the tentative specs for discussion.

picture shows what I have in mind.

[wakibaki]

Yes, it is true that many cards will sync to an incoming SPDIF, but I'm not sure that they will use this clock when in output mode.

If you can confirm that a particular card will use an external clock for output, then it seem an obvious and easy solution to the jitter issue.

w

[Irakli]

Originally Posted by wakibaki

Yes, it is true that many cards will sync to an incoming SPDIF, but I'm not sure that they will use this clock when in output mode.
w

very good point. I need to investigate that further.

[Irakli]

"Word Clock Out: Sends word clock (sample clock) to another digital recorder. Word clock is always output, whether it is generated by the internal clock or passed through from the word clock input."

This is a phrase from EMU-1212 card manual. So it seems that pass through is possible.

[wakibaki]

Originally Posted by Irakli

So it seems that pass through is possible.

Yes, but this is still not an explicit statement that the card will clock out data from the computer memory under the control of an external clock. DACs in some cases have facilities to use 2 clocks, out of sync, but in that case there is provision to repeat or drop samples when necessary to maintain 'quasi' synchronisation, so it is necessary to know exactly what is going on in the soundcard, and how the bitclock is being derived.

Forgive me for being so exacting in understanding what is going on in the card, but there is little point in proceeding further if the effort is not going to produce a real solution, but rather a compromise of the type already available.

w