I am using an SMU (PXIe-4138) to force a 1A current through a conductive path with a total resistance around 2 Ohms. Part of that conductive path contains a slip-ring. The slip-ring rotates and experiences momentary, somewhat small (20m-100m Ohms), resistance changes. For a "bad" slip-ring the resistance changes can be large (1 Ohm)
I am digitizing the voltage created across the path (by the 1A current and ~2 Ohms) using a PXIe6363 analog input in a differential fashion across the path. I get better noise immunity using differential measurement as the environment is inherently noisy.
It appears that most SMUs have isolated (non-ground referenced) outputs. So it seems to me that I should have bias resistors (10k-100k range) present between AI+ and AIGND, and between AI- and AIGND. This is described here:
Do I need external bias resistors?
I seem to get normal looking data without them present. With them present (100k) I seem to ensure that the voltage will always be in the AI input's acceptable common mode range.
Using a diff measurement approach was a backup for us. It is a little bit of a pain to add the bias resistors, so I don't want to do that unless it is truly necessary.
The 4138 reads back voltage. Is there a reason you are trying to read the voltage across the wire on a separate card instead of using the 4138 to report the voltage across the wire?
Thank you for your reply. We originally had some other things in our measurement sequence and interconnect that led us down the DAQ road. However, your point is a good one. We were using the differential input to the DAQ for noise considerations and because our code was already set up. I haven't done it using just the SMU, so maybe it performs just as well from a noise perspective. In any event, if we use the DAQ analog input, I think we need the resistors in there. I'll record some data w the SMU and let you know what we see vs the DAQ. Thanks again.