CJC accuracy (SCXI-1102 + TBX-1303 + K-type Thermocouples)

[steorn]

We have a PXI-1045 chassis connected to a Windows XP SP3 PC via MXI Express (using a PXI-8360 MXI Express card).

In the PXI-1045 chassis we have a PXI-6259 connected to a SCXI-1000DC Chassis via Connector 0.

In the SCXI-1000DC chassis we have a SCXI-1102 operating in Multiplexed Mode. Connected to the SCXI-1102 we have a TBX-1303 with 28 K-Type Thermocouples operating between 0 and 40 degrees centigrade.

We write all our own code using LabWindows/CVI and there is no possibility of us using VI.

In our code we create the thermocouple channels using 'DAQmxCreateAIThrmcplChan' and specify 'DAQmx_Val_BuiltIn' as our CJC source. We then read these Thermocouple channels continuously using 'DAQmxReadAnalogF64'.

The TBX-1303 is isolated in it's own faraday caged room and is free from any erratic air currents. That is to say the TBX-1303 is very well shielded from external temperature fluctuation influences.

We then proceeded to test the accuracy of the aforementioned system using the traditional Ice Bucket approach , the temperature of which we seperately monitor to ensure it is indeed maintained at zero degrees.

Our tests have confirmed our earlier suspicions that the whole system is 1.2 degrees adrift across our intendeed operating range (0 to 40 degrees centigrade). We have verified we are using the correct pull-up resistors on the TBX-1303.

I personaly have read every post about CJC accuracy and have browsed every reference on thermocouple measurment provided by NI. Consequently, we believe we have indeed configured an ideal system for our needs.

We understand that K-Type thermocouples are not 100% accurate but we are at a loss to explain such a large error and suspect that the CJC is not operating as expected. We have tried to recalibrate the CJC, providing two reference temperatures via the Ice Bucket and a heater but the CJC does not appear to retain our recalibration efforts.

Our primary question then is what could explain the 1.2 degrees error ?

Our secondary question is why doesn't the CJC retain our attempts to recalibrate it ?

We have contacted NI directly a number of times on the issue but have so far been unable to talk to anyone with real experience in the field of temperature measurement.

Many thanks in advance for any help a good technical engineer could provide.

Cheers, Shaun.

[Henrik_Volkers]

Basic uncertainty of type K is easily in the range of 1.2°C!   Do you know how good your batch of type K TCs is??

1.2 °C is about 48µV for type K

one more test: short cut ( about 2cm copper wire) each channel right at  the CJC. Should read 0V resp. the CJC temperature on all channels.

(yes, it's about 28*2*4 times you have to screw or unscrew 😞 )

To calibrate the CJC sensor you need to place a reference sensor (RTD with traced calibration)  in good thermal contact to the CJC-sensor. If you find an offset you can do a compensation calculation in LabWindows. (However, the CJC calibration should work... ) 

A good source for TC knowledge:
Manual on the use of thermocouples in temperature measurement,
ASTM PCN: 28-012093-40,
ISBN 0-8031-1466-4

255 pages dedicated to TC

(Page1): 'Regardless of how many facts are presented herein and regardless of the percentage retained,

                all will be for naught unless one simple important fact is kept firmly in mind.

                The thermocouple reports only what it "feels." This may or may not the temperature of interest'

Message Edited by Henrik Volkers on 05-04-2010 05:54 PM

[steorn]

Hi Henrik,

I should have mentioned our K-TCs are quite good, we're expecting an absolute max of 0.6 in our operating range.

I like the terminal shorting idea, so will give it a go.

We could code the compensation ourselves but it would be great it we could get the system to remember our CJC recalibration efforts, such as in MAX or, even better, on the actual device iteself. Perhapps we're missing something but I fail to see what.

I also stumbled opon your book reccomendation in one of your ealier posts and have already placed it on order 😉

On a secondary note, we've just set up a series of really stringent tests as we feel we have a component in the system which is not working as expected. We should complete these in a week and then i'll be able to...err....<cough>...shed more light on the problem 😉

Cheers, Shaun.