6220 periodically reading incorrect

[dbtestcon]

I am having a problem where the measurement reading periodically goes "unstable" and drifts off for 20 seconds or so and then comes back.  The actual voltage reading on the 68 pin terminal block does not change when measured with a voltmeter.  Attached is a picture (image2.jpg) of what the voltage does. This shows a drift from -0.115V up to -0.05V. Sometimes the voltage will oscillate wildly between +0.1V and -0.1V.

The attached picture (image1.jpg) is a screen shot of the VI doing the measurement on a load cell, doing a PID and then writing a voltgae out to a control valve. (please note that the data taken in image2.jpg is graphing the voltage data on the node between Daqmx Read and LC_zero.vi).  This VI is inside a timed loop which typically runs at 2kHz.  Running at 2kHz, this drift happens every 3min58secs.  If you double the speed of the loop, the occurs twice as fast. Likewise, slowing down the loop makes it occur every 8 minutes. 

Here is the setup:

PCI-6220 reading load cells. Setup for differential input (does it in RSE mode as well). 100k bias resistors are installed on each leg to ground.

PCI-6722 sends a calculated control voltage to hydraulic valves.

We are only using 2 channels out of 8 right now. I've tried it on different AI channels and get the same response. Only 1 channel however displays the problem (green/white) and follows a particular cable/load cell.

This is running under RT on a desktop PC.  The setup prior to the timed loop for the AI/AO tasks (dynamic) is show in image3.jpg,

Any thoughts as to why the reading from DaqMXRead would drift off from the actual voltage every 4 minutes?

The setup worked perfectly when we were measuring the voltage through an SCXI-1102 module before switching to the PCI-6220.

Message Edited by dbtestcon on 06-21-2006 10:58 AM

[Jesse_O]

Hello dbtestcon,

The behavior you are describing is very strange.  The fact that your application gets these unstable readings for 20 seconds every 3min58seconds (and that this time is based on your loop rate) makes me believe it is a software problem.  Have you been able to replicate this problem without using your existing code?  If you use shipping example or Test Panels in Measurement and Automation Explorer (MAX), are you able to see the same 20 second drift?

Are you able to replicate this in a smaller/simpler application, or is it limited to your current application.

Also, I did not completely understand what you meant by "I've tried it on different AI channels and get the same response. Only 1 channel however displays the problem (green/white) and follows a particular cable/load cell."  Does this mean that only one load cell has a problem, and that this load cell only has the problem when connected to a specific channel?  If you connect another load cells to the problem channel do they work fine? Or by same response did you mean that all channels give you this same unstable reading for this one load cell?

Regards,

Jesse O.
Applications Engineering
National Instruments

[dbtestcon]

Sorry that wasn't clear.... When I move the loadcell and cable to other AI channels (I tried 1,2,3,4, etc)  it was still a problem.  I've tried other loadcells and cables and the problem seems to float around.

We switched back to the SCXI setup and it worked perfectly with no drift or oscillating problems.  Why would my code work fine on an SCXI chassis and not on a PCI card? This makes me think hardware.  I'm going to try another PCI card and see if the problem still exists.

[Jesse_O]

dbtestcon,

There are many possible reasons that something could work on a SCXI module and not with a PCI-6220.  To start the 1102 has a 2Hz filter.  If there was a voltage spike (not detectable with a multimeter) that saturated the amplifier on your PCI-6220, this could be filtered out with the 1102.

To test your equipment you could connect one of your analog output channels from your PCI-6722 to your PCI-6220.  You could then measure this signal and see if everything works correctly.

If this works, you could try just reading in your load cell with shipping examples or MAX.

Regards,

Jesse O.
Applications Engineering
National Instruments