Not understanding why the ouput of Concat is working differently for MatrixWriter and DAQmx AnalogOutput

[jeehan-phillip-lee]

This is the workflow in question (file: analogouptuttest.zip), where I was trying to make the NI device output a square wave for 3 seconds and then reset the voltage to 0. The sequence that I'm reading with MatrixWriter after Concat seems to be as intended - I get the square wave for 3 seconds and then 0.1 seconds of 0 values returned. However, the voltage output is behaving unexpectedly, where the square wave gets continued for 3.1 seconds and then remains high. In the image shown, all the buffer sizes for the nodes that require them are matched (to 100 samples = 0.1s) - but I also tried increasing ScalarBuffer size , this makes the square wave continue for that much longer.

I got the idea to do this from this old discussion - but I'm not really sure where I am making a mistake. Would greatly appreciate any help on this, thanks in advance.

[banchan86]

Hi @jeehan-phillip-lee if you are only playing the 3.1s waveform once, can you try setting the SampleMode property in AnalogOutput to FiniteSamples?

[jeehan-phillip-lee]

I ultimately would like the waveform generation to be arbitrarily long depending on some condition, which is why I'm sure if FiniteSamples would work (plus I am feeling as if I have some fundamental misunderstanding of one of the nodes in the original workflow due to the fact that I can't figure out why it's not working).

I did try your suggestion as below:

where buffersize is 3000 for FunctionGenerator and 3100 for AnalogOutput, and it seems to work as intended. For this, is it accurate to understand the count in Take as the number of buffers, rather than number of Samples? I have it set to 1 which works, whereas 3000 will give an error.
Thanks again.

[banchan86]

Hi @jeehan-phillip-lee!

So, in your original workflow, I believe the reason why setting it to ContinuousSamples does not work is because once the observable sequence is terminated (since you just uploaded that one sample buffer), the sample buffer that is uploaded gets replayed continuously (3.0s square wave + 0.1 low). This might be easier to see if you adjust it to something like 1.5s square wave + 1.5s low. It also explains why there is a mismatch between the written values in MatrixWriter and the actual voltage output.

If you, for instance, remove the Take and put a Timer with a period of say, 10 seconds, before the Function Generator, then the observable sequence will not terminate, and AnalogOutput in ContinuousSamples mode will keep "listening" for new buffers to play.

Ultimately what you use (ContinuousSamples vs FiniteSamples) will depend on what your task logic will be.

Re: yes Take refers to the number of buffers here.

[jeehan-phillip-lee]

once the observable sequence is terminated (since you just uploaded that one sample buffer), the sample buffer that is uploaded gets replayed continuously (3.0s square wave + 0.1 low)

I tried switching to 1.5s and 1.5s as you suggested, and the signal never gets brought down to what ScalarBuffer specifies. I also tried using a FunctionGenerator with amplitude 0 in place of ScalarBuffer, and this results in the square wave stopping - but only after much longer than the time set in Timer->TakeUntil. I guess the signal generation in FunctionGenerator shouldn't be treated as synchronous so my original approach of using TakeUntil is wrong - but still not sure why the ScalarBuffer value is never getting uploaded.

If you, for instance, remove the Take and put a Timer with a period of say, 10 seconds, before the Function Generator, then the observable sequence will not terminate, and AnalogOutput in ContinuousSamples mode will keep "listening" for new buffers to play.

I think I am totally confused now. Isn't the default behaviour of FunctionGenerator to not terminate already? How does Timer make it not terminate?
And if the sequence going into AnalogOutput isn't terminating, how can I address the original problem of making sure the AnalogOutput is being brought down to 0?

Thank you for your patience.

[banchan86]

Hi @jeehan-phillip-lee! Sorry, I know this is frustrating. If you want to drop me an email at [email protected], we could arrange a quick remote session, I might be missing something and its a bit hard to troubleshoot without access to the hardware.