Thermocouple extension wire compensation

[racegene]

I am trying to set up a system level calibration procedure for my data system which validates an end tolerance for accuracy.  This becomes problematic when I start getting into the thermocouple channels.  When you connect the mV calibrator to the thermocouple wire, it creates two thermal junctions which are not compensated for.  Thermocouple simulators can asses some of this, but they do not meet the accuracy spec for the channel due to the delays in thermal compensation.  I can hold +/-1.5uV on a parallel circuit with copper or +/-1.0uV directly wired to the TBX-1303, but the thermocouple circuit floats with temperature (+/-200uV) and has offsets due to the voltage generated at the junctions were the calibrator is connected (~250uV).  Any suggestions on how to address system calibration for this kind of a setup?

Thanks

[Henrik_Volkers]

 HJ--<TC>-----J---<ExW>--------CJ(TBX)

HJ - Hot junction (measurement point of intrest)

TC - Thermocouple

J -Junction to

ExW - Extention wire

CJ(TBX) - cold junction @ TBX

No more junctions?   Type of junctions? (keep contacts clean, equal temperature, ....)

Where do you want to place the simulator?

What is the accuracy spec, temperature range??

Type of simulator?

Do you have access to the HJ ? Can you use a mobile calibrator (small oven) ? For thermal calibration you will always have delays due to thermal stabilisation processes.

When it comes to higher accuracy, I'm not a fan of extention wires.

I have seen 4-5K drifts due to the bad extention wire and it's connection. 

Sorry, just questions, no answers...

Manual on the use of thermocouples in temperature measurement,
ASTM PCN: 28-012093-40,
ISBN 0-8031-1466-4

'page 1':  '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'

[racegene]

Thanks for the response Hendrik.

The wiring is as follows:

<CJ - TBX>---J-type extension wire---<J-type Connector>---J-type extension wire---<J-type Connector>---Copper Wire---<mV Calibrator>

There is no thermocouple connected as I am just trying to validate that the data acquisition channel is within measurement spec all the way to the point of connection of the TC. The only real issue seems to be the copper junctions to the J-type connector where the mV source calibrator connects. These junctions appear to create two thermal junctions which float with temperature. 

The accuracy spec for the channel with all hardware is +/-30uV.  As I mentioned previously, I have demonstrated that the system works with copper wire (+/- 1.5 uV measured with 3uV offset); so I know the channel and method is feasible. 

The constant thermal environment is not practical as well due to the environment.  The wiring goes between two different rooms with separate controls.  The room where the signal is applied tends not be very stable.  This is partly why I was trying to do the validation without involving temperature.

The calibrator I have been using is an Omega CA100.  This is actually made by Yokogawa, but I am not sure of their model number.  It functions in many ranges, but the TC simulator function updates the junction simulation quite slowly (10s).

Please let me know if this stirs up any ideas.  Thanks.

[Che_T.]

By the way, which signal conditioner are you using?  With your terminal block, it seems like the following are valid: SCXI-1100, SCXI-1102, and SCXI-1102B/C

[racegene]

I am using the SCXI-1102 in a SCXI-1001 chassis communicating to a PCI-6289 over a 5m SHC68-68-EMP.  All of the measurements have been taken with an express VI in LabView 7.1 and verified with MAX.  I have run the acquisition at 10kH and run an FFT Power Spectrum to verify that there is no noise contribution (signal at -40dB and noise at -160dB over 2Hz up to 5kHz).  I was somewhat suprised to see that there was no spike at 60Hz.  I guess shielded twisted pair really does work as well as claimed.

[Che_T.]

BTW, the thought just occurred to me.  Is there a reason why you'd want to use an external calibrator?  Does software calibration meet your needs?

[racegene]

The purpose of the external calibrator is traceability.  In order to validate a system level measurement, you must connect a calibrated standard at the connection point of the sensor to validate that you measure the sensor correctly.  This level of calibration ensures that you account for all factors of the measurement being taken.  The problem lies in the fact that most thermocouple calibrators are intended for calibrating instruments in isothermal conditions within an arm-reach.  This is not possible when you have measurement systems spanning isolated environments.

[Henrik_Volkers]

---

@racegene wrote:

Thanks for the response Hendrik.

The wiring is as follows:

<CJ - TBX>---J-type extension wire---<J-type Connector>---J-type extension wire---<J-type Connector>---Copper Wire---<mV Calibrator>

There is no thermocouple connected as I am just trying to validate that the data acquisition channel is within measurement spec all the way to the point of connection of the TC. The only real issue seems to be the copper junctions to the J-type connector where the mV source calibrator connects. These junctions appear to create two thermal junctions which float with temperature. 

[...]

---

That red marked connection is your cold junction (with unknown temperature)

Can you replace the copper wire with J-Type extension cable ?  Otherwise the temperature difference between that junction and the junction at your calibrator will give you an additional EMF.

[racegene]

Hendrik,  you are correct about having a second cold junction.  This is the part I have been trying to figure out how to compensate for.  Calibrators are intended to compensate at one temperature for both junctions.  When the connection are in seperate rooms, it amplifies errors due to the difference in temperature between the rooms (two cold junctions).  I tried running an extension wire so that the cold junctions are in the same room and the effect is stabilized, but still holds an offset error of about 3.5 deg C.

[Henrik_Volkers]

---

@racegene wrote:

Hendrik,  you are correct about having a second cold junction.  This is the part I have been trying to figure out how to compensate for.  Calibrators are intended to compensate at one temperature for both junctions.  When the connection are in seperate rooms, it amplifies errors due to the difference in temperature between the rooms (two cold junctions).  I tried running an extension wire so that the cold junctions are in the same room and the effect is stabilized, but still holds an offset error of about 3.5 deg C.

---

3.5 °deg for a Type J is about 150µV ... so a coarse error budged :

1K to th TC or extension wire

1K to the RCJ and DAQ

1K to the next extension and junctions

.5k to the second RCJ and the calibrator

My experience: if you want to measure with TCs with a u below 1K: Don't use extension wires and don't mix TC charges and have a individual calibration of the TC. 

BTW What is your temperature range ?  And what is the uncertaincy goal?

The calibrator is a portable device ... do you have ex conditions?