cDAQ with 9205 module single chanel versus scanning

[Mike Webster]

I have an interesting issue where I can't place my finger on the cause.

I have a set of pressure transducers that return 4-20mA signal.  I have a 250-ohm resistor in the path to convert that signal to a nominal 1-5V and it's being read by a 9205 plugged into a cDAQ system.  The measurements are taken differentially at 10kHz, but the negative sides are all tied back to common anyway.

I have a calibration module that reads just the channel under calibration.  When reading one of these transducers upon entering calibration mode, the reading is roughly the same as when running in multiplexed mode reading all the transducers (there are 8 channels being read when multiplexing).  After about 10-15 seconds, that signal will increase about 15mV over the course of 2 or 3 seconds, enough to throw my calibration off by about 15 psi.

Can anyone else come up with a plausible explanation for this behavior?  I'm probably just going to rewrite the calibration module to multiplex like the rest of the application and just pluck out the signal of interest, but I would really like to know the root cause for this.

[Doug_L._Bear]

Just the fact that it happens when you are reading a single channel as opposed to multiplexed leads me towards crosstalk or ghosting on your lines, especially if you are using all the lines.  It does seem strange that it doesn't happen for 10-15 seconds after begginging a single sampling channel, that sounds like more than enough time for residual voltage to bleed off.  Regardless, I would give a look through these documents and see if anything strikes gold with your system.

http://digital.ni.com/public.nsf/allkb/B9BCDFD960C06B9186256A37007490CD

http://zone.ni.com/devzone/cda/tut/p/id/3344

http://digital.ni.com/public.nsf/allkb/73CB0FB296814E2286256FFD00028DDF

Another thing you may want to take a look at is the accuracy of your resistors, I don't think it could a fluctuation consistently like you are seeing, but it is definitely something to consider.  If you multiplex across two far apart channels (like AI0 and AI8) do you see the same thing? Same voltage rise and time and all that?

[KyleBryson]

Hi Mike,

Can you confirm that your grounding connects all of the negative sides of the transducers, the COM pin of the module, and your system ground together?  A slow drift of a signal could certainly be grounding related.  Also, is your chassis grounded in some way - through the ground lug on the end of the chassis?  Are you connected to a desktop or laptop system?  Have you tried measuring using RSE or NRSE mode (since your measurements are single-ended anyhow)?  If those make the problem go away, it would point to a grounding issue.

Let us know if you find a solution to your issue, or some more interesting clues!

Regards,
Kyle

[Mike Webster]

I guess I was just hoping to strike something obvious here.  Unfortunately, the system is remote to me now and I'm having to work through my client, so a lot of this troubleshooting can't be done until my next onsite visit.  For the time being, I've modified my calibration module to do multiplexed sampling like the rest of the program and to just pick out the signal of interest.  Hopefully, that will take care of it, but I'd still like to have known the reason for it.

All of the low sides are tied back to a common ground which is tied back to system ground which is tied to earth ground.  One thing I don't think we did was to tie those common grounds to the AI ground on the cDAQ module.

The PC is a desktop.  I don't think it's resistor heating/drift since the signal coming back from the transducer is constantly on while the system is on regardless of whether or not I'm sampling and the signal "resets" if I stop sampling then restart.

Anyway, I should find out if my stopgap solution worked later today sometime.

Thanks for the help,

Mike

[KyleBryson]

Hi Mike,

Because the NI 9205 is an isolated module, if you don't connect anything to your COM terminal, you might see some odd behavior from the analog front end of the device floating around.  You're effectively referencing it through your signal wiring, but the paths through the differential amplifier will be very high impedance, so it wouldn't take much noise to affect your measurements.

Regards,
Kyle