PCI 6225 residual voltage in analog input channel

[test-lab-joe]

Hello, I am new to the forums and just starting to work hands on with NI hardware/software/etc.

I am using MAX (diffential setting) to monitor an analog input channel (ai71) through a PCI-6225 with an SCB-68. The voltage displayed in MAX exhibits sometimes steady 10.6 volts and sometimes intermittent 0 to 10.6 volt noise and/or square-ish wave. I have observed the noise/wave decay to zero. The voltage displayed in MAX is changes (seemingly randomly) when a voltmeter is used to measure voltage across pins 1 and 35 (with no signal wires connected)

When an externally generated square wave (2.7 volts DC) is applied to pins 1 and 35 in the SCB-68 the value displayed in MAX is dominated by the 0-10.6V "noise", while a voltmeter simultaneously across pins 1 and 35 shows the square wave.

Any suggestions? Thank you in advance.

[crossrulz]

What is generating the square wave?  Are the voltages referenced to the AI ground on the DAQ at all?

[test-lab-joe]

Square wave from motor hall sensor. When using the voltmeter there is no ground reference. Unclear if/how the AI is grounded. What is the preferred way to ground the AI to the DAQ? I came across a wiring schematic for direct pass through method that led me to simply connect signal + to pin 1 and signal - to pin 35.

[crossrulz · Accepted solution]

If you dig into the specs of the DAQ, the input voltages must be referenced to the AI ground or you could damage the board.  Have a really good read of this article: Field Wiring and Noise Considerations for Analog Signals.  Since you are likely dealing with a non-grounded differential signal, what you want to do is add a resistor to each side of the signal to ground.  That article recommends resistors from 10kOhm up to 100kOhm.

[test-lab-joe]

Thank you for pointing out that document. It was helpful. I confirmed zero with low noise by shorting three different ways (pins 1 and 35 to 9 on both sides of signal cable, pin 1 to 35 which seemed less reliable). I also tried the 10k bias resistors but that seemed to negate the signal. Reasonable success was achieved by grounding pin 35 to pin 9. I'm currently getting an authentication error that prevents me from providing kudos.