why does temperature measurement go crazy when heating is on

[rsusilo]

I am using a heater capable of heating to 1200C controlled by labview. I use a digital output to send signal to the solid-state relay in order to turn on/off the heater using K-type thermocouple as the signal input for the controller. I used cFP1808 with TC-120 module and 5 thermocouple channels connected. Everything is working fine but at high temperature above 800C, the temperature reading from some of the thermocouples go crazy, fluctuate up and down rapidly. Hence, my control system does not work as the temperature measurement needed as the input for the controller is messed up. The temperature readings go back to normal everytime the heater is off. I do not know why this is happenning. Any suggestion is welcome.

[Chris_12345]

Possibly this is due to the thermocouple wires picking up signal from the heater wires, although if the effect is a *sudden* transition at 800 C, this is indeed strange. Are the thermocouple wires twisted pair?

[rsusilo]

I think the thermocouple is picking up noice either from the heater wires or digital output module. I am not sure about thermocouple wires twisted in pair. Can you explain? I also monitored the voltage output from the thermocouple using a multimeter connected in paralel. The mV values did not fluctuate much unlike the the temperature reading (200-1200C) shown in Labview. The noise is not there all the time, it is on and off but enough to fail my temperature control.

[JÞB]

Do you have a graph of this phenomina?  the loss of control at 800C is very close the the currie temperature of iron are you sure that is not a J type TC?

[Chris_12345]

Twisting the wires around each other greatly reduces noise pickup from electromagnetic fields.

http://en.wikipedia.org/wiki/Twisted_pair

[rsusilo]

I am positive the thermocouple is K-type. The thermocouple wires were just regular K-type wire, not twisted pair.

[rsusilo]

I do not have the graph, but it looks just like noise with high frequency (less than 100C) or higher amplitudes (several hundred degrees) with lower frequency.

[.aCe.]

Please use an oscilloscope than a multimeter to monitor noise, you wont be able to see noise with a multimeter!

[JÞB]

Well there are a few other factors to consider then.

• Lead length:  TC actually generates the voltage across the temperature gradient in the leads (not directly at the junction) at 800C are the leads long enough that the thermal gradient does not react with the isothermal block?  BE CAUTIOUS! 800°C is dangerous- (You know that but, other readers might overlook this safety issue)  so don't just go touching the leads to feel for heat.
• Setup spacing:  Look at the set-up are the TC leads straight and unentagled?  Look for loops with other wires passing through the loop (Transformers) that would inject noise.  are they running parrellel to any high power wires? Other Noise sources?
• TC problems:  Are the TC and Leads rated to 1200°C? have you tried swapping TCs to see if one is marginal?

[-lucho-]

Hello, hope you are doing great!

Some simple things to check:

Be sure that the voltage  range you are using for the thermocouple is the correct for the temperature rating.
You can use bias resistor to ground, or a simple cable to ground for each of the thermocouples
If you are  using any connector block with gain setting be sure this do not make the measurement off scale.

Try different thremocuples. 

The twisted pair seems like a good idea, however is there a chance you can measure the voltage the thermocouple is generating at 800C, it is a bit dangerous.

We can actually  create a voltage task instead of a temperature task and check if we have the same behavior reading voltgae directly, If this is not the case we can create a custom scale for the linearity of the thermocouple and obtain temperature.

Luis A.

Applications Engineer.