Cold-junction error of 2°C

[m610]

I'm using a PCI-6229 to measure temperatures using K-type thermocouples and a thermistor channel for cold junction compensation.  The thermistor cricuit I made seems to be working fine.  It's reading room temperature nicely, which happens to be 20.7°C, and is essentially the same as what I'm getting on three different thermocouple panel meters, plus it agrees with my cacluations of the voltage-temperature relationships for this thermistor.  This thermistor itself is installed in an unused terminal on the CB-68LPR board, right in there where the thermocouple extension wires connect.  It fit perfectly in one of the little holes.

But when I use this thermistor channel for CJC my thermocouple measurements the readings are consistently about 2°C low, even when using the same thermocouples as I use with the panel meters.  The -2°C error is constant across all of my thermocouple channels. 

With the measurement junctions of the thermocouples being at the same temperature as the CB-68LPR, the thermocouple voltages at the points where the thermocouple extension wires are connected to this terminal block should be 0 volts.  I confirmed with a voltmeter. So the PCI-6229 board should be seeing 0 volts on the thermocouple channels, which per NIST tables, represents 0°C.  It seems that all it has to do now is add the thermistor temperature measurements.  Best I can figure,either this addition is not being done right, or 0 volts is not taken to be 0°C on this board, or something else is going on.

I'm using DAQmxCreateAIThrmstrChanIex to configure the thermistor channel and DAQmxCreateAIThrmcplChan to configure the thermocouple channels.  The CJC sourse is DAQmx_Val_Chan.

Any idea of what is going on, what I am doing wrong, etc.?

Thanks.

[CWolf · Accepted solution]

It sounds like you are doing everything right, but thst is fairly close to the accuracy of the system.  K type thermocouples only have an accuracy of 1.5 degrees.  Also, the voltage from a K-type thermocouple changes by about 40 microvolts/ degree, and the accuracy of the 6229 (when it is in its smallest range of +/-0.2V) is 112 microvolts.  It is usually recommended to perform some signal conditioning on thermocouple inputs before reading them with a DAQ card.  There SCXI and SCC modules that already have the signal conditioning and CJC included.  These may enable you to get better results, but you are still limited by the accuracy of the thermocouple.

-Christina

[m610]

Thanks.  That certainly explains it.  I had not looked at that spec.  I was thing more along the lines of 0.2v / 2^16 = 6 microvolts = resolution.  But, not accuracy.

I guess I could use some conditioners that provide the CJC plus amplification.   I could do with 8 of them.  Does NI have any for use with the M-Series boards?

[John_P1]

Hi m610,

We do have the SCXI 1112 available that provides 8 channels with 100x amplification and per-channel CJC intended for use with thermocouples.  Each channel also has a 2 Hz lowpass filter built-in which may or may not be desirable for your application to reduce noise.

To use the 1112 you would also need an SCXI chassis and a 1349 adapter to be able to control it with your M Series.

Having said this, you will still have some degree of inaccuracy inherent to the thermocouple regardless of what you use to measure the voltage output.  Depending on the temperature range and accuracy that you require there may be other options available that can give you more accurate readings.

Best Regards,

John

[m610]

Thanks for the suggestions.

A 2 kHz filter should be fine for thermocouples.   My temperatures won't be changing anywhere that fast.;)  It'll help remove the 2kHz noise from the fluorescent lighting.

I'll be measuring temperatures up to 900C.  That limits my choices.  For accuracy I'm looking for the 0.75% or ±2.2C that k-type thermocouples offer.

I'm probably going to go with three DIN-mountable modules.

[aNIta B]

Hi M610-

       Page 30 of the specs lists the accuracy as ±1°C for 1000°C temperature measurements.  So, you'll want to take that into consideration along with your thermocouple accuracy. If you're looking for DIN mountable, you may want to check out the SCXI-TBX terminal blocks that are DIN-rail mountable.  For example, there's the TBX-1303 for the SCXI-1100 and 1102.  The 1102s specifications are listed here.  Have a good one!