DAQ System Has a Linear Bias When Measuring Voltage?

[TomDyson]

Hello,

I have a setup using a SCXI-1102 plugged into a PCI-6023E. When I measure a voltage in my test rig the value reported by the card appears to be off by constant factor of about 8%. I know the correct voltage because I've tested it with a multimeter. At the same time, I know that other signals on the card (mostly thermocouples) are correct. Even stranger, the voltage read is always greater than the actual voltage. I've spent hours on this problem and I can't figure out anything that could cause this behavior. I'd really like to be able to use my LabView setup to capture this data at the same time as everything else instead of having to write it in a notebook

I'd appreciate any input you might have on this.

Thanks,

Tom

[Doug_L._Bear]

What is the voltage signal that you are trying to read in (DC or AC? and what voltage level are we talking?).  Have you tried running this through other channels on the card, or is it specific to this channel?  If you wire a known voltage (such as a 9V battery) into the channel are you still reading an 8% offset?

[TomDyson]

This channel is monitoring a signal in the millivolts range (from around 50 to around 350 mV).

I had not tested a known voltage as you suggested. So I did that and the DAQ setup agreed with the precision voltmeter agreed within .0006 volts. Given my experience yesterday I was expecting it to disagree, so I setup my generator again to attempt to reproduce the problem I had yesterday. I haven't been able to reproduce the problem. I don't think that I've changed anything to improve the situation (In fact, the setup has even more alligator clips.), but there it is.

In any case, DAQ now seems to agree with my multimeter within about 1 mV. Problem solved?

I had suspected that a similar problem might be affecting another channel that is reading a signal in the microvolts range but now I don't know.

[TomDyson]

So I hooked everything back up for a test and the bias returned. Overreporting by 8% as before. However, this time I also had my meter set up on the other channel, which was off by about 3% in the other direction. Here is what is different that I can think of:

1) My test earlier only looked at one channel at a time. I saw some stuff about ghosting when scanning multiple channels. However, I just used my simple test program again and observed the 8% bias.

2) My voltage limits were different in my simple program than in the full rig.

2) Channel 4 (I am using channel 5) was not hooked up to anything before. I used this channel to see if there was a channel-to-channel variance earlier (there wasn't) but then it was unhooked again. I just now unplugged it and the bias went away! So some progress... I hooked channel 4 up while channel 5 was unhooked and found that the voltage is correct. I hooked 5 up to my known voltage source while channel 4 read the generator. This seemed to work?

So then I hooked up channel 5 to the other signal I want to measure and the error returned as before. What is going on?

[Doug_L._Bear]

Tom,

This sounds like crosstalk or ghosting to me because when reading one channel alone the reading is fine.  But when other channels are connected they effect the previously okay channels.  I would recommend you peruse through the following articles from ni.com.

How Do I Eliminate Ghosting From My Measurements?

Field Wiring and Noise Considerations

[TomDyson]

All right, I think that I've figured out how this is happening, even if I don't really understand why. It was not ghosting or crosstalk (I thoroughly eliminated them).

The two signals I was trying to measure are voltages across two different parts of the same circuit. I figured out that if I hook up just the (-) lead for one sensor while the other is reading it causes the signal to be wrong. I guess there is some sort of loop being created through the terminal block? I thought that each multiplexing circuit had its own bias and pullup resistors so there shouldn't be any way for a loop to form between two? In any case, having eliminated this as a potential issue regarding my other signals, I'll just continue to use my multimeter for this particular reading (though it is annoying). Unless there is an easy fix to this that I don't know about?