It unfortunately seems that the Dell PCIe Link Speed change was not a valid solution. It worked for a few days, but eventually the problem returned.
I have been able to isolate the problem to two of our pieces of software. One of our programs, Source Control, writes various signals to the AO, while another program, BigAccDAQ, pulls data from a PXI 6229 in the same PXI crate.
I can confirm that the AO runs into this problem only when BigAccDAQ pulls data from the 6229 at a 'fast' rate of more than a few hertz. I find it incredibly hard to believe that our newer and faster windows machines are struggling to keep up with what is now an antiquated set of hardware in our PXI crate (which hasn't changed at all in the process).
The intermittency is still quite baffling to me, but perhaps most bizarre is that setting the clock speed to an intermediate number between the 'working' 200kHz and our standard 250kHz does not register an error, but it does cause our 6229 to stop reading in data. That is, it gives us many of the problems associated with the error, but it doesn't generate an error message.
I can also confirm that AO regeneration and Data Transfer Request Conditions had no effect on the problem.
If anyone has any advice, I'm all ears. At this point I've more or less given up and am trying to convince myself that lowering the AO clock speed won't affect our system in any meaningful way, which I've almost succeeded in doing.
