High current causing noise on thermocouple input spoiling reading.

[NPrice]

I have a J-type thermocouple I'm measuring using a PCI-6036E and an SCB-68. The thermocouple is attached directly to a charging/discharging capacitor - once the rms current through this cap reaches above about 80A noise on the thermocouple input causes the temperature reading I'm getting through LabVIEW to sink to below -10000 (which is obviously incorrect). Below about 80A rms it works fine and when I connect the thermocouple to a meter instead of the SCB-68 it works fine. Is there any kind of input buffer I should be using inside the SCB-68 or some kind of software compensation I can perform on the reading?

Thanks!
Neil

[chutla]

You can limit the maximum current with a series resistor. Is there any reason you need to be using that kind of current? As long as it behaves linearly, I think you can get by with a lot less than that. Capacitors themselves can be a huge source of noise under these conditions. It's best to solve the problem at the source, even if you CAN compensate somewhere else.

Eric

[NPrice]

Eric,

Thanks for the response. Unfortunately we do need to run that kind of current through them (up to about 150A rms max but I'm not sure offhand).

As I mentioned in the previous post using a meter we can still read the temperature at the higher current values and I'm just interested to know if it's possible with my setup to read more accurately through LabVIEW.

Thanks
Neil

[chutla]

Hello again!

Well, it's best to clearly define what you're seeing as "noise". Have you looked at the signal with an oscilloscope to see if it really is random noise, or something else? If it's random noise, you can probably get rid of it with a simple low pass filter. If it's something like an oscillation that occurs sporadically, you may have to approach it differently. In any case, I'd look very carefully at the waveform and see what you're really dealing with before you draw any conclusions.

Hope this helps some.

eric

[LesHammer]

Or if it is periodic you can minimize it by averaging over the period.
Since you said that the meter readings were much better than what your LabVIEW program was getting, I'm guessing that the meter averaged over a period of time but your LV program did not. (You would need to average the raw thermocouple output, not the temperatures.)

Les Hammer

[NPrice]

Les,

Do you know if it's possible to interact with the raw thermocouple output using LabVIEW? Using the DAQmx inputs the reading I get *is* the temperature rather than any raw data.

Thanks,
Neil

[Brian_Beal]

Are you using the other analog inputs on the card? If you are, disconnect all except your t/c and try to read your temperature. If the analog inputs aren't isolated, then there could be a high common mode voltage between inputs. This would cause the problems you are seeing.

[NPrice]

Brian,

Thanks for the suggestion. We've already tried this (even though we believe our inputs are isolated) but it has no effect on the thermocouple input.

Thanks,
Neil

[Brian_Beal]

Neil-

I'm pretty sure that your inputs aren't isolated. I'm convinced that that is the problem. I've seen this problem with Fieldpoint hardware, and was unable to use it for my project. We are reading thermocouples near 230 amp heaters. I had to use a Yokogawa Darwin unit that has 500V common-mode voltage isolation, which is different than regular isolation. Most analog input devices have a common-mode voltage limit of about 5 volts. Even though your thermocouple may be putting out millivolts, that differential is riding on top of some other voltage that is being induced by your equipment. An averaging filter may make your data look better, but it won't be correct.

If you are only looking at one t/c, you may want to get a signal cond
itioner that converts the temperature reading to a higher level analog.

Brian

[Bharatinder Sandhu]

Hi Neil:

To troubleshoot further, can you please describe your set up; how the thermocouple is connected to the capacitor and the DAQ Card?

Thanks,
Bharat Sandhu
Applications Engineer
National Instruments