Secret project (Svart AD/DA converters)

The ADC has an onboard master clock. The DAC slaves to the incoming spdif.

Yes. The ADC should be the master for all devices after it. An Apollo, or other device, should be set to clock off of the spdif signal in it's settings. The DAC should automatically sync as slave to the spdif that is fed to it.

If a unit is set for internal clock, it will get out of sync with the spdif timing as you'll get strange behavior, with pops, clicks, an odd sound being the symptoms.

How about the SP-1 (secret project 1)? Then depending on the configuration:

SP-1AD
SP-1DA
SP-1DAD

And the separate ADAT box, connecting up to 4 SP-1's, would be the SP-1ADAT.

Yes, that's generally the best for any products.  The ADC needs the best clock source simply for the acquisition quality.  Once the analog signal is acquired and converted that's the best the audio will ever be.  With SPDIF/AES the clock for the data is the data itself.  The SPDIF/AES receiver buffers the data and recovers the clock from the data itself.  This keeps from having to use a separate word clock and it also allows some devices to buffer and work on the data in non-real-time because the clock can be recovered at any point.

Your interface should then be set to clock from the recovered clock from the SPDIF/AES.  If you don't, since oscillators are close but not exactly the same, the data will slip out of sync with the internal clock of the interface.  it's best that the incoming data clock simply be the master here.

The DAC also recovers the timing from the data stream.

So the data and clock flow is this:

ADC/Master clock -> Interface/Slave -> DAC/Slave

There is no option for WC in or out.  Sorry, that was beyond the scope of this design.  Maybe in a future version or design I can make that happen if things work out. 

If you are concerned about clocking the ADC, I can guarantee that the oscillator being used is about 50-100x less jitter than average WC..

I'm not sure what you mean by "I suppose one could slave your DA to the system clock through spdif and clock the AD that way."

There is no reason to clock the ADC to anything(and there are no provisions for it), it's the master.  You set your interface as a slave and the DA is already set as a slave in hardware and it will follow whatever SPDIF info you send it.

You simply plug the ADC into your SPDIF input, set your interface's software as "SPDIF IN" for the clock source and then plug the DAC into the SPDIF output.

And enjoy using it.

If you are using multiple interfaces, they generally handle their own clocking.  Take for instance my use of an Alphalink system AND a soundcard in the same computer.  They don't share clocks.  Right now the soundcard software is set for SPDIF IN as the clock source.  The Alphalink uses it's own internal clock.  I simply let them be separate.  Once the audio passes into the analog domain, there is no reason to WC them together as the domains are not synchronized, nor need to be.

Typically there is no need to lock separate converters together in the time domain when using a DAW as the "tape deck".  This only comes into play with things like real tape machines and other devices that have an absolute need to be synchronized in time for start/stop sequences and such..  Or older devices that have very poor/cheap internal clocks.

I'm not even sure what you meant, so I hope this helped, or maybe you can clarify what you meant.

In this case, you would set your clock source as "SPDIF IN" in the Apollo settings(or whatever nomenclature it uses for that).  I've not used an Apollo but most every interface as such usually has this choice in the software settings somewhere.  This would allow the Apollo to recover the clock from the ADC SPDIF output and thus make the ADC the clock master.

Wait, I found this is the Apollo manual:

"When Apollo is set to use an external clock as the master clock source and a valid clock signal is detected at

the specified port, the External Clock LED is solid green. Apollo can be configured to use an external clock from

the Word Clock, S/PDIF, or ADAT inputs"

So you would do exactly as I mentioned above.

And if it's designed correctly, making the SPDIF the master, the WC output of the first slave should be derived from the clock recovered from the SPDIF source.

You might be able to daisy chain the SPDIF output from the first device out to the input of the second and so forth and just set them all to slave to the SPDIF signal too.  You'd put the DAC on the end of the chain.

There's no conversion in the satellite, so there wouldnt be any issue there.

OK, so the DAC board layout is pretty much finished. DRC has been run and no errors found. I've checked footprints on all the parts against the real parts and they all fit. I might tweak one copper polygon under the 15V regulator for better heatsinking but I think it's pretty much ready to send out to be made.

I'm on the fence about going straight to a production number of boards or doing expensive prototypes once again..

the benefit of a prototype is an absolutely confirmed build v.s. the possibility of some built not to designed spec ?

if you feel that a prototype is basically good insurance, could we help defray those costs ?  why should just you incur them ?

Cool, go for it svart!