Conclusions are out immediately?
What happens in a few days when the caps' sound settles in to something distinctly different?
Did more tests and let the caps burn in more. But my conclusion stays the same: Added electrolytics, even polymer caps seem to smear the impulse response in some way. It sounds better stock.
The audible differences between the original file and the recording (using a discrete DCS converter) are much smaller now than they are with all the SD converters I've got.
As for the clocking, the difference may only be there (or at least a lot more pronounced) with the digital output of my RME card set to "professional", which means the voltage is set for AES/EBU connections, exceeding the receivers specifications. I cannot reliably detect a difference otherwise.
For my needs the DAM1021 is perfect, with only the clock drift still bothering me.
How do you feed all you boards? Just curious
//
I bought precision linear PSUs for the DAM1021 boards. The audio output boards will be fed by a double linear PSU scheme with a single +/-20V PSU feeding several +/- 18V PSUs for 4 or 8 channels each (on the respective sub-boards), all based on standard linear regulators.
When I bought the DACs I assumed they would sync to the incoming clock, which is identical for all channels. But if all the DACs run their own clocks and just correct for the mismatch between the external and internal clock this might be a problem. Not a big one in most situations though, since minimal discrepancies only matter for material with overlapping signal, like drum multitracks with bleed.
Seems to me that an issue was raised regarding exactly how well all the boards would stay in sync. There were arguments made that the error would be small enough not to matter, others begged to differ. Despite all the interest and commentary about this dac, I don't recall a single build using them in a multichannel set up ever gracing these pages. I originally bought 4 boards intending to build a multichannel setup for an active 4 way system. The designer said initially that "it should work". There was never any actual proof given that it will work in fact, and due to the differences in opinion about the outcome, I decided to go a different route. I hope it works for you.
I will only know how it works when it is done. I remember, too, that Soeren said it should work. Sample/phase accuracy is mandatory for a speaker, but not necessarily for multichannel mixing.
I haven't been able to find documentation for J9 - could there be a way to sync boards? In the case of my build all boards are fed by the same clock anyway.
OK! - How many will you need?
I just got a RME Digiface USB from Thomann. It seems to drive my 1121 (Toslink) much better than my previous solution. A high class brand as I understand it. I like the little box!
I plan to get 2 more 1121s and then use the Camilla DSP platform (thread here!) to build my 2 way stereo system. Maybe I'll add 2 more channels and a center (stil all stereo!)
//
Just two. I'm currently running three different RME cards in the same computer, no problem. They work together like one big soundcard.
RME always had the best drivers, they're perfect for the digital side. Never liked the converters though. For most of the time they specced jitter to be better than 3 ns (clocking externally) or better than 1 ns (internal). Ns, not ps!
BTW: Are you planning on doing a 4K version of your linear 44,1 khz filter? I don't feel anything lacking now sonically with your filter, my output stage and the stock DAM, but you never know... ;-)
Last edited: