DAQ card voltage offset problem

[RobTuley]

Hi,

I'm not the most knowledgeable about electronics so hopefully this will be an easy problem to solve. I am trying to aquire a voltage signal from across a wheatstone bridge pressure sensor using a National Instruments PCI6024E DAQ card, and I'm having huge problems with a voltage offset. I've built my own bridge, and circuit is as below:

+5V from voltage calibrator ---------------------------
| |
| |
\ / /
R1 \ R2 / /
\ / /
Connection to ch 2 ADC---| |
A | | B
| |
Connection to ch 10 ADC---------------------|
\ / /
\ R3 / R4 /
\ / /
| |
| |
0V connection and AIGND ---------------------------
connection on DAQ card |
and earth on power supply |
_____
___
_


I thought the problem might be with my pressure sensor so I built my own bridge to test, but the problem is the same with this - I get a large voltage offset.

For example: With the circuit above with R1 = R2 = R3 = R4 = 562 ohms, I connect the circuit up without the DAQ channel connections, and then get the expected voltages of 2.5V and 2.5V when probing location A and location B respectively. However when I connect my DAQ channels and reprobe, I get the expected 2.5V at location A but a value of 2.37V at location B. thus when sampling within labview in differential mode I get an offset value of 0.13V. Given that the FSO of the pressure transducer is 50mV, this is a substantial problem.

I know for definite that I'm in differential mode with the DAQ card so the signal should be fed straight into a floating source instrumentation amp and I've tried virtually every combination of earthing possible.

This has been driving me nuts for a week so any help much appreciated!

Rob.

 

[IRstuff]

So EE101 math says that you seem to have a 5123 ohm resistance in parallel with R4. TTFN

 

[RobTuley]

Yup - I realise i could just stick another resistor across 'R2' to balance the offset out, but surely this isn't standard practice? I am a little worried that something strange is going on in the DAQ card which I don't really know about and which will screw up my results, and the manual hasn't really given me many clues.

Cheers for the math though. Rob.

 

[IRstuff]

Have you tried to reverse the connections and/or use other channels from the ADC board? TTFN

 

[Heydave]

Beyond the accuracy of the resistors used, you can also look for dissimilar metals around the bridge at different temperatures. This causes offset voltages. Using a portable volt meter instead of the measurement board will tell you if the voltage you see is real. Other possible causes are ground loop currents and unbalanced input currents on the board. AC noise from power circuits can effectively be rectified and appear as DC voltages in the measurement board. This rectification can come from the timing of the sampling or from signal conditioning circuits. There could also be excessive input currents on one leg of the board caused by static discharge damage.
A common trick to cancel measurement offsets in a balanced measurement circuit is to use an input mux to sample twice but in opposite polarity. The results are subtracted thus cancelling the offset error and doubling the result. If the offset is from your mux or before, say due to static damage, then this solution will not work. Some protective diodes are known to cause similar problems as static damage with light shining on them. I have designed automated fault location equipment with individual bridges spred out over 6 miles creating serious problems because of all these issues. It is not easy to balance a bridge to a few milivolts with over 100 volts of noise and a few microamps of stimulus current.
Good luck.

 

[melone]

Where is your other connection on your differential signal? Depending on how far away, and ground leg impeadance, you may easily get 130mV offset in your "measured" value.

 

[jbartos]

Suggestion: There may also be common-mode currents or voltages over the ground connection to the better ground.