DAQ System Hanging Up on "Write to Text File"

[diarmaede]

Hi All,

I am collecting data using two SCXI chasses with 1120 amplifier modules and SCXI 1112 thermocouple modules.

I am monitoring 12 input voltages and 4 thermocouple inputs.

I have set the samp,e rate to 2500S/s, and I am letting the software figure out how many samples to take each measurement.

Most of the time, the system runs fine, at about a 16ms loop time.  However, every few seconds, the system will hang up for ~500ms or more between readings.  This only happens when I am logging data to a text file.  When I have the logging function turned off, the system never hangs up.

Any thoughts as to what could be causing this and how to fix it?

Many thanks!

[RavensFan]

Duplicate Post

[Mark_Yedinak]

Yes, do your logging in a parallel task so that it will not impact the data collection. When you do it within the same loop as your data collection you take the hit for the file I/O. There is a variety of ways you can pass the data to the logging task. You can use a queue or a notifier to pass the data. If you have your data accumulated during your data collection you could use a functional global to store the data. Both loops can access the data via the functional global. Since your data set is not that large I would use a queue to pass the data.

[diarmaede]

If I have broken a rule in posting a question about a related but significantly different issue, I apologize.  However, I have ruled out the primary failure mode I suspected in my first post, and have moved on to another suspected failure mode, thus the change in subject line.

[RavensFan]

It looked like you had the same text in both messages.  If you truly have a different question, then post a new thread.  But then I would suspect your message would look substantially different.

If you have a related question, then you should keep it in the same thread that way the continuity of the discussion continues.  Since you said, "hey I don't think it's this problem anymore, now I think it's this", you are basically continuing with solving your same problem and should stay in the same thread.

[diarmaede]

Hi Mark,

This sounds perfect, but I lack the technical knowledge to put your ideas into code.  Any chance I could get you to either post code (I am running LV 2010, V10.0, 32-bit) or post a picture of the associated block diagram?

I found the "Queue Basics" VI in queue.llb.  Am I on the right track?

Thanks again.

[Mark_Yedinak]

Yes, that is basically what you want to look at. You should also use the "Find Examples" to look at the shipping examples for queues.

Here is a very basic piece of code which illustrates what I was suggesting.

This simulates collecting data in the upper loop. Every 250 points it posts the collected data to the queue for the lower loop to process. In this example I do nothing more than update the indicator. In your case you would want to put the code there to write it to file.

[diarmaede]

That's what I thought you meant.  Thanks for posting it.

My main thread uses state machine architecture.  Looks like I would have to build a second loop, outside the state machine, to do the logging.  Please let me know if that is not correct.

[Mark_Yedinak]

Yes, that is what I have been suggesting. If you keep the logging in your state machine you will take the hit on the performance because the file i/o is slow. If you separate it out your data collection loop will be unaffected by the data logging.

[diarmaede]

Understood.  I am such a gumby at this, I just wanted to make sure I was comprehending the obvious.  I will set it up and run as you suggest.

Thanks for your help and patience.