# Thermocouple socket wrong way round on board - problems?

Just come across a board where the K-type thermocouple socket has been tracked the wrong way around. A partial fix is to forced the plug in the wrong way around as well, but this will cause a mismatch in the metals at the connectors leading to an error in signal.

Question is, how much of an error? The plug/socket combination can be assumed to be at constant (room) temperature. My guess is "not much", given that we only need accuracy to a degC between 10degC and 40degC. Opinions?

[Yes, I know it's a service nightmare waiting to happen unless well documented but no choice until we do a re-layout. Fortunately it is the first time we have used this feature]

UPDATE - the problem just got simpler. The socket is tracked correctly but the component (socket) pin numbers on the circuit diagram are wrong. This means the wiring loom and plug are reverse wired, which is easy to correct - no re-layout!

• Can't you do a cut/strap on the board? Jul 27, 2017 at 10:29
• @Finbarr Maybe, but its messy. There is no perfect solution Jul 27, 2017 at 11:11

It's tricky reasoning about thermocouple junctions, so it's worth going back to basics. This is the conventional temperature practical measurement situation.

simulate this circuit – Schematic created using CircuitLab

The voltmeter reads Tmeas-Tref. The introduction of disimilar metals into the thermocouple loop does not matter as long as both Tref junctions are at the same temperature.

When we remove the copper wire and voltmeter, as the junctions are at the same temperature, and move the voltmeter somewhere else, then we get the classic thermocouple diagram from basic physics text books, where the voltmeter obviously reads Tmeas-Tref.

simulate this circuit

In plugging the thermocouple into the socket backwards, what you are proposing is the following

simulate this circuit

The voltmeter reads Tmeas + Tref - 2*Tsocket

Any difference between the socket temperature and the reference temperature will be added two-fold to the measured temperature. If that difference is expected to be <<0.5C, then you may get away with it.

If not, then it would be better to plug the thermocouple in the right way round, and modify the firmware to negate the measured difference from reference.