NI Board Analog Input Delay

[Adrian Orozco]

Hello,

I am using a pulseblaster as a pseudoclock to trigger the data acquisition from a PCIe-6363 analog input. The sequence starts with time elapsed before the data acquisition is triggered. For instance, the time elapsed is due to waiting for a MOT to load and the signal is a fluorescence signal. I found that there is an unwanted delay between the signal that is to be acquired and the acquisition window as a function of the time elapsed before acquisition. This delay is linear with the elapsed time. I checked to see if this delay also occurs when using LabVIEW and found no delay. Could you shed some light on a possible issue that could be causing this delay?

In addition, I found no delays when triggering the PCIe-6363 digital and analog outputs.

Best,

Adrian

[Philip Starkey]

Hi Adrian,

Thanks for reaching out. I don’t know of anything that would explain why this happens with labscript and not LabVIEW. It does very much sound like this though: https://github.com/labscript-suite/labscript-devices/pull/102 

I would have expected to see the same issue in LabVIEW though if that was the case.

Might be worth a look anyway. You’ll need to update to the latest development version of labscript devices (it’s not on PyPI/conda yet), provide a constant frequency reference clock to the NI card (it really should be derived from whatever external clock reference is used for your pseudoclock or else you may just trade one drift for another), and add some extra keyword arguments to your connection table/experiment scripts (updated docs for this feature are here: https://docs.labscriptsuite.org/projects/labscript-devices/en/latest/devices/ni_daqs/#ai-timing-skew )

If that doesn’t solve the issue, could you attach an example HDF5 file that contains the acquired data, with an explanation of what you see (with times) vs. what you expect to see? It doesn’t have to be real if you need to avoid sharing sensitive research. Any minimal example that includes the acquired traces will do as long as you can detail how the acquired data differs from what you expected.

Cheers,

Phil