What is the recommended way to record timestamps associated with measurement acquisitions from NI DAQmx, DCPower, DMM, etc.?

[sjdemartini]

General question/motivation

Our team records varying-duration acquisitions across several different NI cards, most notably DAQmx, DCPower, and DMM cards. We're using gRPC to configure the cards and collect measurements. It is very useful to be able to know the date/time that a measurement occurred (at least to within some level of precision) so that it can be correlated/aligned with other instruments (NI and otherwise).

What is the recommended way to infer or record timestamps (e.g. seconds since epoch, "2023-02-24T00:47:30.160005+00:00", etc) so that we can know when a given measurement was taken? In our investigation of NI APIs/documentation, it seems this is not supported in a native/straightforward way.

What we've been doing

In our gRPC client-side code now, we've added custom logic to attempt to "infer" timestamps of the recorded measurement data as we fetch it:

• For DAQmx: we record a "start time" t0 right before calling start_task, then calculate timestamps based on the "sample clock rate" attribute
  • Effectively: timestamp = t0 + (1 / sample_rate_hz) * current_sample_number
  • This seems similar to what is suggested in this old NI knowledge article, or "Option # 2" in this one.
• For DCPower: we record a "start time" t0 right before calling initiate, then calculate timestamps based on the "measure record delta time" attribute
  • Effectively: timestamp = t0 + measure_record_delta * current_sample_number
• For DMM: since it doesn't seem to be possible to configure a sample rate, we currently (1) record a start time t0 right before calling initiate, (2) record an "end time" t_end right before calling read_status, (3) call fetch_multi_point with array_size=read_status.acquisition_backlog, then (4) calculate timestamps for each point by "interpolating" the number of measurements between t0 and t_end.

With the above examples for DAQmx and DCPower, I do worry that there may be "drift" as continuous acquisitions get to be long-duration (many hours or days), which could make interpreting data error-prone. This seems particularly relevant if multiple DAQs, DCPower units, DMMs, and/or non-NI devices are all being used together and measurements need to be aligned. Also, not all NI cards seem to support all attributes in the same way (e.g. "measure record delta" is not always available).

With other brands of equipment (e.g. Keithley, Zurich Instruments, etc.), timestamps are natively returned/reported along with the measurements, which is quite helpful. It feels like there must be a way to do something more equivalent with NI.

Is there a better way? Are there other APIs/attributes/cards that would be of use to us? Any improvements we can make to our above logic? I've seen mentions of NI-SYNC and timing cards in various NI forums, though I don't know how/whether these do what I'm describing or if they can be used in association with the aforementioned cards. Thank you very much!