I use long (10-m) copper-constantan thermocouples for work to measure temperature. The temperature gradient is only present in the 15 cm nearest the hot junction; the remaining 9.85 m of these thermocouples contain no temperature gradient. Currently, we use 10-m copper wires and 10-m constantan wires that go all the way from the hot junction back to the datalogger.

Since the temperature gradient exists only in the 15 cm nearest the hot junction, is it necessary to use constantan and copper wires to go all the way from the hot junction to the datalogger? It'd be a lot cheaper for us to use constantan only for the 15 cm nearest the hot junction.

Based on my knowledge of the thermoelectric effect, the Seebeck effect, and Kirchhoff's law, I believe that we only need to use the different types of metals across the temperature gradient nearest the hot junction. For the remainder of the thermocouple where there is no temperature gradient, I believe we can use copper wires for both the constantan and the copper side of the circuit, and a quick online search confirms this suspicion. I am not an electrical engineer, though, so I'd love to hear from someone who knows more than me.

Here's a sketch:

Thermocouple Sketch

Since temperature gradients only exist across the two wires connected to the hot junction, it seems like currents will only be induced in these two wires. The two segments of copper wire farther away from the hot junction will not have any current induced in them; the only current that will flow through them is that which gets induced in the two wires that are part of the hot junction.

  • 2
    \$\begingroup\$ Anywhere you have dissimilar metals, you have a thermocouple. The cold junction compensation is for the transition from thermocouple wires back to copper. \$\endgroup\$
    – Kartman
    Nov 10, 2021 at 11:42

1 Answer 1


If you change wire material, then you establish a junction that you have to know the temperature of. That's why thermocouple extension wires and sockets are made of the thermocouple materials, so you don't need to know or control their temperature.

By all means, put a junction from thermocouple wires to copper at the sense end, and copper back to thermocouple for connection to your datalogger (I assume your datalogger has a thermocouple metal input rather than a copper input), but then you will have two junctions whose temperatures you need to know in order to offset the reading on the data logger. If they're equal, then no correction needed.

If your logger has a copper input (yikes, unknown connection block temperature!), then you could go thermocouple to copper once near your sense point, and copper all the way to your logger. You would still need to know the temperature of that connection, but you wouldn't lose any accuracy at the logger connection block.

  • \$\begingroup\$ Thanks for the reply. If you change wire material but one of the wires does not change in temperature across its length (like in the photo I just added), why will this new junction be a thermocouple? \$\endgroup\$ Nov 10, 2021 at 15:14
  • \$\begingroup\$ @DavidMoore You need to look at the physics of it for that. Look for diagrams of the Seebeck effect. There's stuff flowing back and forth along the length of a junction metal that has a temperature gradient across it, even if its just one leg. \$\endgroup\$
    – DKNguyen
    Nov 10, 2021 at 15:17
  • \$\begingroup\$ I think I understand. We aren't measuring temperature at these two new cold junctions, so although we will probably still see an increasing signal as temperature increases, we won't be able to measure temperature exactly. Is this correct? \$\endgroup\$ Nov 10, 2021 at 17:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.