PXI

cancel
Showing results for 
Search instead for 
Did you mean: 

Streaming from multiple devices fails if in certain slots

Solved!
Go to solution
Another thing I noticed - The data is being acquired and streamed to disk, but the GUI is not updating.  While the cursor was jumpy and nothing was happening on my GUI, I had Windows Explorer up, and I kept refreshing it.  I could see the streaming file increasing in size with each refresh.  So apparently, changing the slot assignments is somehow affecting the threading of the application, which is something I've suspected for some time now.
0 Kudos
Message 21 of 28
(2,454 Views)
Thanks for the update....in these latest tests what devices did you have in the chassis?  Is there anything in slot 2?  If there is, can you pull it out and run your tests again?
 
The reason that I ask is every fourth slot shares an interrupt line, so slot 2 and slot 6 share an interrupt line back to the CPU.  Also, the behavior you're describing sounds like there are too many interrupts for the system to be able to handle at the same time as updating the GUI.
 
-Jack
0 Kudos
Message 22 of 28
(2,450 Views)

After running my program some more times, it appears that the streaming is NOT happening during the beginning of the requested collection time.  When I refresh Explorer after a few seconds, the file size is still small.  It then grows more rapidly than it would in real time with each refresh.  I tried setting the collection time to 60 seconds to see what's going on.  It takes about 8 seconds for the task VIs to indicate that they've read a block of data.  Each task then indicates high-speed bursts of Events Received updates.  The bursts get shorter and begin to slow down as the acquisition progresses.  I was able to collect 60 seconds of data, but the behavior of the program was completely different than with the 'good slot' test, which starts immediately and updates the Events Received counters at regular intervals.  It seems like setting up the tasks is taking much longer in some slots, and the system is trying to catch up.

Has anyone asked the developers if the slot number has any effect on how DaqMX initializes, performs threadings, etc.?  I've been asking this question for weeks now, but haven't gotten an answer.  As I've repeatedly stated, the ONLY DIFFERENCE in the test program is the change in the Physical Channel assignment.

0 Kudos
Message 23 of 28
(2,451 Views)
Just saw your last post, Jack.  Yes, the ZTEC GPS card is in slot 2.  I have previously run numerous tests without a card in this slot and had the same problem.  I'll try pulling the card out and repeating the test.
0 Kudos
Message 24 of 28
(2,450 Views)

OK - Same results as I've previously reported.

 

Slots 2-4 are empty, and a 6120 is in slot 5-8.

These work: 5&7, 5&8, 6&8, 6&7. Events Received updates with no delay.

These don't work: 5&6, 7&8.  Events Received is significantly delayed, then plays catchup.

0 Kudos
Message 25 of 28
(2,447 Views)

Jack,

Tried your program as well.  Same slots cause similar problems, although your program actually behaves more badly than the one I have that also streams to disk. Your program 'errors out' with -200010. It never finishes both acquisitions.  Go figure...

0 Kudos
Message 26 of 28
(2,446 Views)

Jack,

I'm sending you my test program so that you can have exactly what I'm running.  I am able to do streaming at 800KHz using slots 5&7, 5&8, and 6&7.  Slots 5&6 and 7&8 exhibit the delayed start and catchup behavior.  Maybe you can run this exactly as is on your system before Robert comes and fetches the chassis (provided if I get authorization to release it). 

0 Kudos
Message 27 of 28
(2,433 Views)
Solution
Accepted by topic author wired

For all interested parties,

I neglected to post the resolution to this issue.  The problem is in my PXI-8186 controller - I have an older model, which apparently has this bug.  It was fixed for the newer models of controller (anything above Rev C, if I recall correctly).

0 Kudos
Message 28 of 28
(2,172 Views)