correcting temperature data when CJC source set to "constant" by mistake

[yafoteer]

I have a set of thermocouples taped to a UUT in an oven.  In my app, the CJC was set to "constant" (25C) by mistake during data acq.  (NI-9214 has a CJC on it and it has been selected now).  This was never a problem before because the ambient in our lab was always about 25C.  Yesterday the HVAC failed during the test and the ambient went up by several degrees, which means our data is skewed by several degrees.  

We have the delta between the oven setpoint and a thermocouple dedicated to measurement of oven temperature, for all data points. If you trust the oven setpoint (which we do) then I think we have a record of how the ambient temp skewed the data.  Right?  Delta is not constant, it increases with oven temperature, but ambient increased over time as well.  Max error is about 5C lower than the oven at 100C. 

Is it possible to apply a correction?  Can it possibly be as simple as applying the delta at each temperature? I think I understand what happened, but am I wrong about that?  

Thanks

[Henrik_Volkers]

The CJC compensation is not linear since the V-> T function of a TC is not linear.

I wrote a quick vi to show you how big the error compared to a linear correction would be.

You will find that the error isn't always Zero if the actual and fixed CJC match. That's due to the errors of the conversion polynomes used in the standard(s) (assuming NI implemented them right 😉 )

[yafoteer]

Thank you Henrik - I think I understand, but let me explain again and make sure I said it right.  

I have 3 records:

1: oven temperature as read back from the built-in PID controller.  We consider this a good reference to the actual oven temp.  

2: oven temperature as recorded by a TC hanging in space to measure the oven temp (CJC set to a constant 25C)

3: readings off of the UUT

The differences between 1 and 2 are not linear with respect to oven temperature.  I suspect they exist because the CJC was set to a constant and not compensating for the increased lab temp. 

The question is:  do the differences between 1 and 2 at each temperature give me a way to correct the readings from the UUT? 

For instance, I have readings from the UUT with a known oven temp of 60C.  I can see the measured oven temp (with the CJC set at 25C) is 58C.  Can I add 2C to the UUT reading to correct it?  If I do this at every measurement will it get me closer to the right answer or further away?  It won't be perfect and I don't expect great accuracy but if "better" is all I can get I will take it.    

[Henrik_Volkers]

Here is the 'correct' correction, as could be seen the error compared to a linear 'shift' correction seems to be very small.

And I wouldn't trust a TC (not recalibrated) to more than a u of 1-2°C (or 1-2%)

In my eyes the biggest error is correlation between your oven TC and PIP temp, but you can make a good estinate by looking at old tests.