NI USB-6218 sample rate inconsistent?

[TechnoSam]

So instead of getting one sample per loop, go for 5 or 10? Or more? I can see how that might help...

[Dennis_Knutson]

Something like 10% might be more reasonable. For example, 1000 samples/sec and 100 samples.

[TechnoSam]

I'm not sure if I understand... Can you give me pseudocode? That would be awesome.

[Dennis_Knutson]

Have you looked at the getdata function I mentioned and the samples parameter?

[TechnoSam]

Yes, I did. Basically, it doesn't really fit my purpose. Because I need to graph the data in real time. So someone can watch it. Maybe it won't make a difference if I grab just a few, but yeah...

[Dennis_Knutson]

You don't think getting data every 100ms is fast enough? What sort of refresh rate is on your computer monitor or on your tv?

[Dennis_Knutson]

Sorry, meant to finish it saying you can adjust rate to whatever you want but you need understand that displaying fewer samples will mean displaying less of s complete waveform.

[TechnoSam]

It's possible that I could. I'll have to talk to my supervisor.

I'm trying it out right now though.

I use getdata(ai, 50) and give it a 100 ms pause every loop.

After a few iterations, it says

"Error using daq.analoginput_nidaqmx_1_22/getdata

     OBJ is not running and no data is available. Use START before calling GETDATA

Error in analoginput/getdata

     varargout(1) = getdata(uddobj, samples,dataformat); "

I started ai outside of the loop, and the first few iterations it doesn't have any problems. So I have no idea what's going on now...

[TechnoSam]

Just checked, I can't use the getdata(). I have to get single values and send them off. Not totally sure why.

So is there anything that could cause the slow down? My code is just "get a value, wait, repeat." I have no idea why that wouldn't work.

[Dennis_Knutson]

First, I don't see a wait in the for loop of your original code. Second, as I mentioned, you are using software timed input. On a non-real-time os, the timing resolution can vary widely. That was the whole point of the recommendation to change to hardware timed input where the sample rate is meaningful.