High amount of noise in thermocouple reading

[Peaches0204]

Hi all,

Currently I am currently trying to take some measurements in an aqueous environment using a J-Type thermocouple. I run into some problems with the measurement as the thermocouple works fine but as soon as the heat source (radio frequency probe) is introduced, the values either revert to 0 or a very large negative number. I expected these high frequency values to be blocked by the integrated low pass filter of the NI-9213 receiver that I am using but apparently I  was wrong. The radio frequency probe functions at roughly 400-500kHz. Anyone have any kind of input as to what should be done in order to eliminate this noise? I have thought of implementing an extra low pass filter but that is impossible to do once the data is input into LabVIEW given that the temperatures are just output as an array and cannot be filtered. Another thought was building a separate low pass circuit for all of the inputs before it passes into the receiver.

Any ideas and help would be greatly appreciated.

[ToeCutter]

I think your ideas are sound.

Something as simple as the right size of inductor in series or capacitor across your probe will probably do the trick. Low pass filtering in LabVIEW is no problem however the effectiveness may be limited as what is presumably high frequency noise has already been aliased onto your low frequency thermocouple measurement, so you may inadvertently end up smoothing out wanted signal. Depends how sharp your temperature changes are. There's a 'build waveform' function in LabVIEW that you can use if the problem is that you need a waveform type for your chosen filter block. You just wire the array and sample period to it.

[ToeCutter]

On second reading, filtering in LV probably won't help. Your signal is totally saturated by the RF pickup and you're not sampling fast enough (hardware limitation) to differentiate the noise spectrally.

[Peaches0204]

Its funny you say that it is a hardware limitation as I had talked with a NI representative recently before deciding on the hardware for the measurements and he stated that the filter within the device should function well enough to take out any RF signal.

[ToeCutter]

When I say hardware limitation- I only mean the limited sample rate.

I would expect the system to have filtering built in and I'm sure it does. Could just be that the probe is either putting out crazy high signals or the RF is being picked up on the card's internal circuitry post filtering.

[Peaches0204]

Most likely the first option of the probe putting out the high frequencies. Like I said in the range of 450kHz. Problem is that I need a live feed back of the temperature readout and I cannot achieve that with this little problem of mine. Best scenario to try the extra circuitry and build a low pass filter to pass the thermocouple through.

[0utlaw]

Hello Peaches,

In my experience, a railed thermocouple reading is usually due to a ground loop or similar.  Your description of the scenario (this happens only when the probe is introduced) as well as the readings you're getting (consistently railed to one extreme or the other) make me think this is more likely the culprit than unintended RF pickup- where is this RF probe powered from, and how is it grounded relative to the DAQ?  How long is the thermocouple probe? Where is the DAQ carrier/chassis relative to the RF probe?  Does the polarity of the signal switch from 0 to -<something> or vice-versa when the thermocouple probe's leads are swapped? IIRC the 9213 also has active open thermocouple detection, which may mean that a small current will flow through the thermocouple and it needs to be shielded if it is in contact with conductive elements.

It seems unlikely that there would be enough of a signal picked up on the thermocouple to cause this sort of behavior- you'd need to be "receiving" a differential signal across the two alloys on the approximate order of 10s of mV to rail the channel. 

This article may be helpful:

Field Wiring and Noise Considerations for Analog Signals

http://www.ni.com/white-paper/3344/en/

Regards,

[johnsold]

I agree with 0utlaw that this sounds like the equivalent of a ground loop.

Try fully shielded thermocouples with the shields grounded. Also wrap the thermocouple cables (inluding shields) through a ferrite core which has high permeability at the probe frequencies.

Lynn

[Henrik_Volkers]

Measuring µV with a loop (your TC) in a induction heater .. 

I assist johnsold about the ferrite core.

IF you currently have an unshielded TC make shure that it is (very well) twisted. 

How fast will your temperature increase?

[JÞB]

A J type thermocouple and a Induction heater?.

OK the currie points are

Iron @ 1043 K

Constantin @ 35 K

So The Iron will have a nice stable magnetic field no matter what is applied

the Constantin's magnetic filed will align only in the presence of an external EMF

Application of an inductive heating element changes the measured output of your probe.  I suspect those facts are somehow related.  You need those wires very well shielded to prevent inducing some rather strange currents.