Are thermocouples not affect by EMI?

[Topher925]

Are thermocouples not effected by EMI?

I've been reading a lot of papers lately about temperature measurement. Most them involve temperature measurement in electrochemical devices such as batteries and fuel cells. Using micro-thermocouples seems to be common practice but this makes absolutely no sense to me. How is it that you can use a thermocouple or thermistor to measure temperature in an environment where EM fields are prominent? Granted most batteries and fuel cells don't have current densities beyond 1.5 A/cm² but can the induced current from the EMI just be integrated out?

 

[berkeman]

I've been reading a lot of papers lately about temperature measurement. Most them involve temperature measurement in electrochemical devices such as batteries and fuel cells. Using micro-thermocouples seems to be common practice but this makes absolutely no sense to me. How is it that you can use a thermocouple or thermistor to measure temperature in an environment where EM fields are prominent? Granted most batteries and fuel cells don't have current densities beyond 1.5 A/cm² but can the induced current from the EMI just be integrated out?

The thermovoltage is differential between the two wires, and EMI will generally be coupled to the twisted pair in common-mode. So your measuring circuit does need to have a good CMRR, and maybe some CM filtering as well. Worst case, you could use twinax cable (shielded twisted pair) to connect to the thermocouple, and ground the shield well to suck up the EMI fields.

 

[Topher925]

The thermovoltage is differential between the two wires, and EMI will generally be coupled to the twisted pair in common-mode. So your measuring circuit does need to have a good CMRR, and maybe some CM filtering as well. Worst case, you could use twinax cable (shielded twisted pair) to connect to the thermocouple, and ground the shield well to suck up the EMI fields.

I can see how a good CMRR and a good CM filter will help/eliminate EMI under steady conditions but what about under heavy transient operation (20%->100% power within 100ms)? Keep in mind that in batteries and fuel cells the group current travels in one direction but not necessarily the local current. Could electronic filtering alone yield an accurate (<.05C) degree reading under these type of conditions? Most of these applications are using sensors around the size of 100 microns where there is no room for shielding.

 

[berkeman]

I can see how a good CMRR and a good CM filter will help/eliminate EMI under steady conditions but what about under heavy transient operation (20%->100% power within 100ms)? Keep in mind that in batteries and fuel cells the group current travels in one direction but not necessarily the local current. Could electronic filtering alone yield an accurate (<.05C) degree reading under these type of conditions? Most of these applications are using sensors around the size of 100 microns where there is no room for shielding.

Not sure I'm following the description of the physical situation. A thermocouple is formed by the physical contact of two different metals, and a thermovoltage is generated based on the junction temperature. Since the junction is metallic, it cannot be in any conductive contact with anything conductive (like battery electrolyte). To operate correctly, it should be well-coupled thermally to the object to be measured, but certainly electrically isolated.

If you are referring to capacitive displacement current interfering with the measurement, that is why twisted pair is used to transfer the differential thermovoltage to the measuring instrument. Twisting the wires keeps capacitively-coupled noise from appearing as a differential noise voltage term.

Sorry if I'm missing what you are asking about. Do you have any setup drawings or pictures that would help us to understand your concerns?

 

[berkeman]

Actually, my description of the thermocouple junction and thermovoltage is not really correct... A better explanation from wikipedia.org:

Thermocouples measure the temperature difference between two points, not absolute temperature. In traditional applications, one of the junctions—the cold junction—was maintained at a known (reference) temperature, while the other end was attached to a probe.

Having available a known temperature cold junction, while useful for laboratory calibrations, is simply not convenient for most directly connected indicating and control instruments. They incorporate into their circuits an artificial cold junction using some other thermally sensitive device, such as a thermistor or diode, to measure the temperature of the input connections at the instrument, with special care being taken to minimize any temperature gradient between terminals. Hence, the voltage from a known cold junction can be simulated, and the appropriate correction applied. This is known as cold junction compensation.

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

 

[Averagesupernova]

Considering the low output impedance of most thermocouples (between the 2 wires) I find it hard to believe EMI can be a problem. I believe you are most likely to have temp errors due to the device measuring the temp at the open end of the pair. A thermocouple develops a voltage based on the difference in temperature from the shorted end of the pair to the open end. So to get a true measurement of the shorted end of the pair you have to know the temp of the open end of the pair. Usually done with another type of sensor.

 

[Averagesupernova]

Berkeman, it doesn't necessarily matter if the shorted end of the thermocouple is in contact with something else conductive. It depends more on the measuring device. I've taken a digital exhaust gas temp gauge for automotive applications which uses a K-type thermocouple and simply stuck the ends of the wires in a solder pot to monitor temp. The molten solder made the connection. I no longer work for the particular company but last I knew it was still in use. It stays in the solder constantly.

 

[Topher925]

Considering the low output impedance of most thermocouples (between the 2 wires) I find it hard to believe EMI can be a problem.

Are you certain of this?

The offending sources deliver leakage currents through capacitance and imperfect insulation and spray magnetic and electric fields at your signal wiring. The resulting injected voltages or currents can overload and paralyze signal amplifiers.

Thermocouple and RTD signals, being at low dc millivolt levels are the most susceptible, so I will use these in examples that follow.

http://www.eurotherm.com/training/tutorial/instrumentation/holland/emi.htm

I always thought of thermocouples makeshift antennas that measure temperature. I know that interference from the 60Hz power lines in the walls can make a thermocouple useless if an integrator isn't used to filter out the noise. Its this same reason why thermocouples can not be used inside transformers, motors, RF transmitters, etc.

Berkeman,
I should have specified. These applications require thermocouples and their leads to be extremely flat and non-intrusive as possible, similar to MEM thermistors (which I also wonder about). This requires that they not be twisted and they not be shielded. I believe most use an enamel insulation similar to that of magnet wire. I would like to post a paper of an example, but I don't think that would be ethical.

 

[FredGarvin]

I really don't kno the answer, however, I did a test a while back that had our test subject subjected to an extremely intense EMI field. We still used TC's for the monitoring of temperatures. The big restriction we had was that we could not have any wiring coming out of the test object so we had to include a converter to go to fiber optics.

 

[Proton Soup]

i think berkeman is mostly right, except i think the twisted pair keeps out magnetically-coupled noise, and the shield on it will keep out the capacitively-coupled. if you keep your measurement circuit electrically-isolated and pass the data over an optic link, that will also improve things by eliminating ground loops and common-mode noise. most temperature measurement is also probably very low in frequency, so filtering out say >30Hz shouldn't mess you up. the main problem is that the signal is just so small and you've got to account for any other thermocouple junctions you inject with dissimilar metals, like say from solder joints.

 

[Topher925]

So in a nutshell, thermocouples are not immune to EMI and precautions should be taken. Some of the data I have been looking at contains very large amounts of noise which isn't accounted for by the authors other than stating that its just the nature of using very small instrumentation. Most of this data was collected under steady state operation, I would presume that it would be worthless if collected from a transient operation.

 

[dlgoff]

This might be of interest:
http://www.omega.com/Temperature/pdf/Twisted_Shielded_Case.pdf."