It's my understanding that:

• A thermocouple is formed any time two dissimilar metals are joined together, but only the joining of certain metals and of known purity, produces a predictable and repeatable voltage that is high enough to easily measure at a given temperature.
• Thermocouple extension wire must be of the same metals that the thermocouple is made of, or else additional thermocouples are generated in series with the measurement thermocouple.
• Terminal blocks used to join thermocouples to thermocouple extension wire must also be made of the these metals. Or else, same.

I have been looking for methods of bonding thermocouples to metallic objects to measure their temperature, and in doing so find myself challenging what I think I know.

According to some credible sources on the internet, it is acceptable to simply weld the thermocouple junction to the metallic device under test. It seems this is how grounded thermocouples are made; the thermocouple inside is welded to the stainless sheath:



But this introduces a 3rd metal to the mix. Won't it effect the voltage produced?

I have also read (not in whitepapers but in responses to questions asked online) that it is acceptable to solder or braze the thermocouple wires together, and to the device under test. This introduces at least one more (but most likely several more) metals to the mix. Still not a problem? How not?

Finally there is the issue of the terminals at the measurement device. I have dissected several different industrial temperature controllers and found the thermocouple input contacts appear to be made of nothing special. And the MAX31856 breakout boards that I'm using right now also appear to have "ordinary" terminals for the thermocouple input. Both the MAX31856 and industrial controllers support several types of thermocouples, so even if they did use special metals for the input terminals (and for the PCB traces I suppose), it would only be suitable for one type of thermocouple. So I suspect this is just plated copper as usual, and the effect of this dissimilar metals junction this accounted for in whatever Cold Junction Compensation algorithm the device uses. Is that correct? If so, then why is it not OK to use ordinary copper wire as thermocouple extension wire? Unless I'm missing something, all that does is move the dissimilar metals junction further away from the measurement device.

EDIT: please allow myself to answer myself. I remembered. The measurement device measures its own temperature and the cold junction is assumed to be at the same temperature, so the cold junction must be kept as close to the device as possible or else the assumption is wrong.

 

You are asking a number of different questions. I will try to answer them as simply as possible.
Introducing a third metal between the two thermocouple wires by brazing them together will not change the properties of the thermocouple because: The emf generated by the junction of one of the wires to the brazing material is offset by the opposite polarity emf generated by the junction of the second wire to the braze.
The cold junction of a thermocouple is the point at which both thermocouple wires are connected to a common metal, e.g. copper wires. The emf generated by the junction of one thermocouple wire to the copper is cancelled out by the junction of the other wire to the copper. That is why compensating cable with similar properties to the thermocouple materials is used to ensure that the cold junction is in the right place to make a measurement.
Grounded thermocouples are usually measured using amplifiers with differential inputs.
I hope this answers your questions in a way that you can understand.

 

I did design an engine production test stand many years ago, and there was quite a bit of concern about running the thermocouple wires to terminal strips. So I came up with an easy cheating trick. The secret is to remove about an inch of insulation from each of the conductors, and then twist the wires together tightly, and then put the twisted wires into the terminal strip. So the terminal strip had no effect, the folks demanding terminal strips got what they asked for, and it worked quite well. So that is how to avoid the dis-similar metals problem.