PXI slot utilization and PC lockup/BSOD

[ChrisGruver]

Hello,

I have a test station with the following equipment and PXI slot:

PXI hardware:

 - 1085 chassis

 - 8398 controller

 - Keysight VNA, slot 3

 - Astronics OmnibusII, slot 4

 - 4081 DMM, slot 5

 - Pickering MUX card, slot 6

 - 2564 switch cards qty 4, slots 7,9,11,12

 - 8431/16 RS-422 serial, slot 16

 - 8430/16 RS-232 serial, slot 17

PC

 - Dell Precision 5820, 64GB RAM, Intel Xeon W-2225 running Win10 21H2

We recently switched to the Dell PC mentioned above from an Industrial Rackmount made by BSI Computers.  The old PC was not near as powerful as the new one but seemed to work better for my application.  I think I have gotten past the issues of the Dell PC but need some help making sense of it all.  The PC had lockup/BSOD issues at random intervals so I thought using the MXI compatibility SW should help.  After enabling MXI compatibility I no longer had PC lockup/BSOD but I was not able to collect all necessary data from the RS-422 comport card.  The comport is set to 230400 baud and the device will transmit 32 ~200 byte messages every second for 50 seconds, it is very important that I collect all data, if one of the 32 messages is thrown away the test will error.  My software would throw an error saying it couldn't collect the data in order which tells me my driver parsing the data threw away a message because of a checksum error or the message somehow didn't get to the serial port buffer (I didn't create a breakout cable to collect the comport data, but that may have been the next step).  I had seen the comport error on another PXI system, same model Dell PC, and disabling MXI compatibility seemed to correct that.  I would also have trouble sending a RS-232 message to a device, the message is simple, just sending a couple of bytes and waiting for an ACK.

Since enabling MXI compatibility was not an option due to the missing messages and inability to get the ACK from the 232 message I decided to move all PXI cards towards the controller card, eliminating blank slots.  It was a shot in the dark to move the PXI cards, I figured that since I had trouble with the serial communication trying them in a different slot may be beneficial.  I moved the 2564 cards to slots 7-9 and 11 (skipped slot 10) and moved the 8431 to slot 12 and 8430 to slot 13.  After moving the cards I still received PC lock up/blue screens so then I enabled MXI compatibility and now my system seems to work perfectly.  The only blank slots I have now are 2, 10 and 14-18.

Has anyone had similar issues or know why these issues could occur?  I need to prove that the PC and PXI systems we use are good for mass deployment.

[Robert__R]

Hi Chris,

I'd suggest disabling ASPM to see if that fixes the lockup/BSOD issues. Part of this KB explains how to do that:
https://knowledge.ni.com/KnowledgeArticleDetails?id=kA03q000000x0MKCAY&l=en-US

It looks like you're using several older (non-express) cards.  There are 2 PCI segments in the 1085 chassis, so the distribution of boards between the left half and right half can be important for bandwidth and latency. I don't know how active the cards are (typically switch cards and mux cards are "set and forget"), but a good strategy may be to split cards across segments -- 1 serial card on the left, 1 on the right; 2 switch cards on the left, 2 on the right; etc. Distribution within a segment is unlikely to be important.

There are other possibilities as well. The right half of the chassis has a faster/wider link to the host so it might help to have any high bandwidth devices in the right half and the serial cards in the left. There are also cases where that's worse, though.

- Robert