TC CJC Accuracy

[ASTDan]

Hello,

I have what I hope is an easy question.

I noticed the NI USB Thermocouple modules have a CJC accuracy of 1 degree. Can this be improved using an external CJC souce i.e. ice bath?

http://www.ni.com/pdf/manuals/371566e.pdf

Also 1 degree F of error seems high. Is this normal? What techniques can improve this accuracy? Can the accuracy be improved?

Thanks

Dan

[TobyD]

QUOTE (ASTDan @ May 7 2008, 06:50 AM)

Hello,

I have what I hope is an easy question.

I noticed the NI USB Thermocouple modules have a CJC accuracy of 1 degree. Can this be improved using an external CJC souce i.e. ice bath?

http://www.ni.com/pdf/manuals/371566e.pdf

Also 1 degree F of error seems high. Is this normal? What techniques can improve this accuracy? Can the accuracy be improved?

Thanks

Dan

I can't address your question specifically because I have not dealt with this hardware, but I can speak in general terms...

When dealing with any temperature measurement device you have to worry about accuracy and repeatability. A published spec only says what the unit is required to do, but most of the time the actual performance is better than the spec. You can calibrate out errors in accuracy by measuring a series of known source temperatures (or by comparing measurements of multiple sources with measurements from a high accuracy device) and generating a lookup table that shows the offset from the actual temperature to the measured temperature.

We generally do this by taking 5-10 measurements about 10 seconds apart and averaging them. If we are measuring a 200°C source and our average measurement is 200.7°C then at 200°C we have an offset of -0.7°C. Repeat this process different temperatures until you can generate a relatively good offset curve. Then any time you get a measurement from your device, add the offset at that temperature to give you a more accurate result.

Repeatability is a bit trickier. Really all you can do is characterize your device and hope that it is better than the published spec. We generally find that on high quality hardware the repeatability is much better than the published spec. Taking an average of measurements over time can also help to minimize the damage done by repeatability error.

Occasionally you should recheck your device against a known source to ensure your offset table is still accurate. Calibrations can drift over time. There are also labs that will do the characterization for you and generate the offset table (we send our temperature standards to NIST for calibration and then use those to calibrate our units that don't require such high accuracy) but it can be expensive.

-Toby

[xtal]

QUOTE (ASTDan @ May 7 2008, 09:50 AM)

Also 1 degree F of error seems high. Is this normal? What techniques can improve this accuracy? Can the accuracy be improved?

The accuracy of the measuring device can only be as accurate as the transducer. Here's a table of TC accuracies I found: http://www.microlink.co.uk/tctable.html. Omega has simlar charts based on the number of coefficients you use to calculate the polynomial.

In our specs, we usually use an accuracy of +- 3 deg C for TCs in general regardless of how we measure them. Also, TCs can drift over time depending upon what the operating temp is and oxidation levels. We've seen a 5 deg C drift for type K thermocouples used in high temps over the course 2-3 years.

For situtations where we need better accuracy, we use precision RTDs.