PXI4472B; problem with pseudodifferentiel

[benbol]

I do data acquisition of sensors which measure velocity of ground motion. I use the acquisition card PXI4472B which work in a pseudodifferential way only.This acquisition card is in the acquisition device NI PXI1042Q. I would like to subtract positive and negative signals of my sensor to reduce the noise. So I put negative signals to the shield because I believe (but I am not sure) that the DAQ will do the subtraction between my positive signal and my negative signal. I believe (but I am not sure) I can do that because the negative signals do not vary more than +-2.5V. But I saw in a documentation that we can't make a non-ground connection to the shield. Some people (where I work) said that it was possible because my negative signal doesn't vary more than +-2.5V. I would like to know if that I did was correct.
The other possibility was to use 2 channels, one for the positive signal and the other one for the negative signal. But don't I loose the dynamic range because I do the subtraction of my 2 signals after digitalisation (with a software)?
An other possibility was to do the subtraction before digitalisation with an ampli-op. But we measure little signals (few mV), so we don't want to do that by our own because we are afraid to make to much noise. Do you make that?

The final possibility was to use an other acquisition card. We have bought also the PXI3060E which is also in the acquisition device NI PXI1042Q. This one work in a differential way, so we don't have all this problems. But this acquisition card don't do vibration treatment, like antialiasing filters.

I have illustrated with 4 outlooks my four possibilities. I am open to other possibilities and all ideas will be welcome, but I prefer having answers for the first free possibilities.

Please help me. Thank you,

ben

[Benjamin_C]

Hello ben,

The NI-4472 has unbalanced differential inputs. With this type of inputs, the best is to measure a voltage from floating sources.
Indeed, if the sources is grounded, the source ground may be different from the NI-4472 ground. This difference is usually between 1 and 100 mV, but it can be higher if power distribution circuits are not properly connected. In these cases, the measruement may be inaccurate because of a ground loop. Nevertheless, there is a 50 Ohm resitor on the signal ground to reduce this current to a negligeable level. We estimate that if the shield is less than 2.5 V from the NI-4472 ground, the perturbations are not significants.
This is your case, so you should directly connect your signal as it is indicated in the NI-4472 User Manual (with the shield on the ground).

Regards,

[benbol]

Thanks a lot to have answered me. I would like to know also if I have to divide the amplitude of my signal by two because by subtracting negative signals from positive signals, the amplitude is multiplied by two. Is it in the conversion from volts to m/s (calibration of the sensor) that the division by two must be done (for example, 1um/s=10 V single ended, so 1 um/s=20V for the differential to divide the signal by two) ? In most of sensors which measure velocity, the technical paper gives us the calibration, but we don't know if they suppose that the sensor is connected on a differential way or not.

ben