Did you update the value of r1 on the sidecar?

DPLL bandwidth?

Mid-low should work (based on my experience with Musiland). After 1 hour of warm up, try lower DPLL

Perhaps you should use a shield over the dac chip. Seems to have improved on lock issues. Some production dacs with ess sabre32 have shields. More info here: http://hifiduino.wordpress.com/category/test/ that might help with your search for gremlins.

Edit: you may also want to check the usb power. Sometimes computers have a pms period and 5v USB suddenly has a drop in voltage wich can cause all sorts of things.

Edited by user Tuesday, March 27, 2012 1:26:53 AM(UTC)  | Reason: Not specified

have come further now:
when I hook up s/pdif 1 and exa,they work fine
when I hook up only s/pdif 2(unconnected s/pdif 1 to input) it work fine as well, exa too
BUT when both are connected I loose lock on s/pdif 1 unless I connect s/pdif 2 minus to gnd!then s/pdif 2 is muted!
sidecar doesn´t change anything
seems like s/pdif 2 is the source for concern
puls transformer?
went over entire pcb several times and cleaned with isopropylalcohol (what a waste!)
any ideas?
best
Leif

What are the sources?
Optical, coax, AES?

Seems like something is strange indeed. But could be easier if you say what is connected to where. And how you have set up the inputs (AES, TTL or consumer spdif). Say f.ins that the spdif2 is your computers motherboard coax digital output. I allways get surprises with my computer/audio setup. I prefer using optical connections for this reason. Properly isolated usb when listening to music or high level situations, but optical for everything else (f.ins. video playback at low volume levels with normalization of the audio stream).

I somethimes get issues like that when connecting the same computer with both usb and coax. It seems that ground loops are hard to get around with computers proborably due to the large number of components that affect each other all the way down to poor ATX PSU design.

additional info:
as soon as minus wire from (isolated from chassis)rca chassis connector s/pdif 2 is connected to terminal block, s/pdif 1 looses sync and relocks once s/pdif 2 minus is joined to s/pdif 2 gnd

weird, eh?

best
Leif

Edited by user Wednesday, March 28, 2012 1:03:31 AM(UTC)  | Reason: Not specified

I just carried the (little ) sd card player down to my workshop and connected it alternately to s/pdif 1 and 2, not both at the same time!
I can bring down my denon dvd 1920 tonight and hook up both, but I didn´t bother with this weird problem
the seemingly identical consumer lvl s/pdif inputs sure don´t behave identical
1 works fine alone
2 works fine alone
when 2 minus wire is hooked up to rca connector even w/o source connected to 2, s/pdif one looses lock on connected source
, but relocks when jumper is installed between 2 minus and gnd or minus cable is removed from s/pdif 2 terminal block!
with jumper on s/pdif 2 between minus and gnd on terminal block the input signal is muted
this is all with s/pdif pcb connected directly to BIII with ribbon cabel
insertion of sidecar doesn´t matter
even tried another ribbon cable without change
sort of conclusion is that something is wrong with s/pdif 2 either on the s/pdif pcb or dac pcb
must try change dac board tonight
if no difference: faulty s/pdif input 2
or am I missing something?
best
Leif

Edited by user Wednesday, March 28, 2012 3:23:26 AM(UTC)  | Reason: Not specified

I assume you are doing this in conjunction with the selector switch?

Okay. It's 4am here so I am having trouble making out all of the iterations above, but here are a couple rules of thumb for the spdif module...

The + and - inputs for each channel are essentially differential inputs, connected to the primary of the input transformers. If you have unused inputs, short the +/- pins the gnd. If you are sharing a - connection between single-ended inputs (by joining them at the RCA), it's probably a good idea to short both - to gnd by jumpering - pin to gnd pin for both inputs, so only + changes.

Let me know how that works. It would probably be better to not join -, but let's see what happens.