05-28-2012 11:10 AM - edited 05-28-2012 11:13 AM
I am trying to measure the signal from a linear potentiometer (resistance = 5000 ohm) using the NI 9219 and cDAQ 9178 in the 2-wire resistance mode. I ran some initial tests using the pot and 9219 and found a drift in the signal. This is what I did :
The pot was displaced to a particular position and was arrested in that position and I was measuring the resistance reading from the pot over a period of 3 days. I was sampling at 1 Hz. I was carrying out the test in my lab and there could be temperature fluctuations (max 5° C fluctuation) due to the AC shutting off and on. This is what I obtained -
As you can see, the signal drifting quite a bit (max resistance change = 3.2 ohms). Is this kind of behavior expected with the 9219 in the 2 wire resistance mode ? if so is there any other module to get the signal from potentiometers or do you think I am doing something wrong ?
05-28-2012 12:08 PM
It certainly looks like typical temperature related behavior.
The specification sheet for the 9219 gives Gain Drift as +/-15 ppm of reading and Offset Drift as +/-3 ppm of reading. If this were the only error source, it would require a temperature change of ~ 200 degrees. The potentiometer probably has a larger temperature coefficient than the 9219, but even that will probably not account for all the variation.
On the 10 kohm range the excitation is 2.2 V. Interestingly, the description says it excites with a current source, which is not specified. If it has a maximum compliance voltage of 2.2 V at the maximum resistance of 10.5 kohm, then the current is about 200 uA. At 408 ohms the disspation in your device is about 5 mW at 2.2 V or 18 uW at 200 uA so self-heating should not be an issue.
Can you repeat the test with a short circuit on the input? Or try it with a metal film or wirewound resistor. Cover the DAQ chassis and the resistors with a cardboard (or styrofoam) box to minimize air currents. Also note whether the computer and the DAQ chassis have been powered on for some time or start cold.
I did not see any warm up time specification in the 9291 documents, but many devices specify an hour or more before fully complying with the specs. The first 12 hours of your data seem to have a different character from the later part. It could be that the entire system takes that long to warm up.
05-29-2012 12:37 PM
I agree with' Mr.' Lynn
Could be temperature related, if possible you may want log temperature too while testing.
It is interesting how A/C schedules affect precise measurements, I can sometimes tell when data was collected during the weekend versus weekdays due to a different A/C schedule being used.
Were the connections to the potentiometer made with solder or secure screw terminal connections?
05-29-2012 12:58 PM
Thanks for your reply. I will do a test with a wirewound resistor and let you know of the results when it is done. The DAQ assistant tells me that the current being provided is 500uA (and is not allowing me to change this value.. if I try changing it, it is giving me a message that only 500uA can be selected). Yesterday, I did a test similar to what I explained earlier but this time I covered the DAQ with a box and this is the result I got:
The variation of the signal kind of follows the change of temperature in the room. May be this time I will log the temperetaure as I do the test.
05-29-2012 01:02 PM
The 9219 has terminal spring connectors on it. So, I just hooked in the wires from the pot to the connectors. Do you suspect any loose connections ?
I will log the temperature when I run the test again.
05-29-2012 01:51 PM
I would not trust the spring connectors for stable long term precision measurements. No evidence. Just a gut feel that you may be trading convenience for performance.
I will be interested in seeing the correlation with temperature also.
05-29-2012 01:57 PM
Yes, I was concerned about loose connections- probably nat an issue in this case.
Yes, please log the temperature.
05-30-2012 11:37 AM
I was just looking into the potentiometer specs and data sheet more carefully and found that they specify a recommended wiper current of 1uA. They also gave this figure in a seperate manual
The 9219 is providing it with a 500uA current. I am guessing this might be the problem. Any suggesutions how to get the current down to 1uA ?
05-30-2012 05:29 PM
That may be most of your problem. It never is a good idea to exceed the manufacturer's recommendations, esepcially when trying to get the best possible performance.
I would suggest that you apply a fixed voltage to the ends of the potentiometer and measure the voltage at the wiper. To keep the current below 1 uA you must keep the voltage at the wiper below 1 V if you connect directly to the NI 9219. Otherwise buffer the voltage with a low input current op amp. The datasheet recommends this. Make sure the excitation voltage is stable.