AES16e-50 sync problem

edit: I tested both the stagebox and the aes16e-50 cards with both desks.

I have been having similar problems with the lynx AES16e-50 hooked up to a Midas DL32 stage box running RML labs Software Audio Console ( SAC ) as the software. I don't think it is a clocking issue. For me It seems to be a dropped buffer issue between the DAW software and the Lynx Mixer? I could not get clean audio out at 1 buffer 64 samples, 1 buffer 128 samples or even 1 buffer 264 samples, but when I tried 3 buffers at 32 samples it worked. I then tried 2 buffers at 32 samples and it remained stable with no audible glitches. SAC has a window in the software Where it reports slipped buffers and while I was trying all this it never reported any slipped buffers yet something was definitely going on.

Phase 2:

AS I started to test further with more audio source input and screen changes, plugin window opening etc..

I've found that the SAC program ( which has several "settings" to choose from such as Realtime Priority, High Priority, Single CPU etc.. ) needed to be set to High Priority NOT Realtime Priority in order to have mostly clean audio. I say mostly, because with this setting I got rid of the constant "dropped Buffer" sounds but SAC Itself will now drop the occasional buffer and register the event in its window.

This is what has lead me to the possibility that there could be some conflict in the transfer between the DAW and The Lynx mixer? i.e. both cannot have "Real time priority" just a theory as the lynx mixer does not report slipped buffers as far as I can tell.

What DAW type software was used in testing with the Lynx AES16e-50?

Is it even possible for the card/lynx mixer to run fine at 2x32 samples but NOT at 1x higher buffer sizes? this would seem strange to me.

Thoughts?

Hal

Mixer version is 2.00 build 23, firmware version is 14.0 but I have already tried version 13.2 with no success.

I installed my Motu 896mk3 to the same computer and programs and plugins run nice with same samplerates and buffer lengths so after all it really seems to be a hardware issue.