Unexpected comb filtering on acquisition (NI6210)

[mathol]

Hi,

I am working on a project to record signal from accelerometers using an usb6210 card, and while validating in the lab with signals generated from a low frequency generator, I came upon an unexpected behavior that I cannot explain or get rid of.

When I compare the power spectral density of the acquisition of a sawtooth wave (50Hz, Fs= 10kHz) to another sawtooth wave with the same characteristics generated directly on the computer (python), we see that the recorded signal's PSD has a profile as if some comb filtering had occured. (see pictures attached).

We are recording in differential, but tried with only one channel (Ai0, asymetrical with reference) and with Ai1 and Ai8 connected to ground through some resistor and the behavior is still the same.

For information we use advanced TDMS asynchronous library for the writing of samples.

Would somebody have a clue on what could be the reason for this behavior?

Thanks,