Huge scan backlogs on RTSI slave

[shantaram]

I am synchronizing AI of 2 PCI-6031E boards by sharing channel clock on RTSI. I am using a continuous buffered acquisition and an intermediate FIFO buffer to save scans into a file parallely. For all channel clock rates, I am getting huge scan backlogs and a subsequent -10846 error on the slave, where as the master seems to behave normally. How do I counter this?

[Spencer S.]

Greetings,

You have indicated that you are routing the channel clock between the master device and slave device. Are you also routing the scan clock between these devices? If you are using the slave device's internal scan clock, the behavior you are seeing may be a result of the rate of the slave device's scan clock being too low. Most likely, if you are sharing the channel clock, you will also want to share the scan clock. Additionally, if you need further assistance, posting your code may be useful.

Good luck with your application.

Spencer S.

[shantaram]

hello,

I disabled scan clock on the master, so that round robin scanning is performed(our sensors have 20 meg resistors, so they need more settling time than the max channel clock rate). However, I am not doing anything with the scan clock on the slave. I am not sure if this is the way to go. pls advise.

Attached is my code. The sub vi that takes care of the synchronization is AI synchronize.vi

[Spencer S.]

Thanks for attaching your application. I was able to successfully run it with two DAQ boards. The slave's scan backlog would jump around a bit, but it never continually grew. How many channels are you sampling? What sampling rate are you using? Below, I have attached a modified version of one of your subVIs. It performs all of the required DAQ operations. Can you run it with your DAQ boards to see if it behaves similarly to your application (scan backlog growing, etc.)?

Spencer S.

[shantaram]

Hi Mr. Spencer,
Thanks for the modification. It works with my DAQ boards in similar fashion as is. But, I guess, there are a lot of other things (for eg: saving data to a file, using queues as intermediate buffers) in my original program that may be slowing the retrieval from the AI buffer down. Are queues fast enough for my application or should I be using RT-FIFOs? I did not have any problem using queues with just 1 DAQ board.

[shantaram]

Oh, I forgot to mention that I didnot have any problems even at maximum sampling rate of the board. I was using all 32 channels on the PCI board at a scan rate of 3000. Never a problem with your "Justdaq.vi".
Ram.

[Spencer S.]

I believe that queues should be suitable for this application. To discover what is causing this behavior, I would suggest that you begin removing functionality from your application. File I/O may be the best place to start, since it can be very processing-intensive.

Spencer S.