Hi everyone I was wondering what sort of output I would get if I flipped a thermocouples polarity at a connector that was far away from the main thermal junction.
The answer is that the output will be unpredictable, based on temperature differences that can't be compensated in the usual way.
Normal thermocouple practice is that there are exactly two points at which temperature affects the measurement: the connection of the two wires at the probe end (the "hot junction" by convention, although it could be any temperature), and the connection of each wire to the measuring circuit (the "cold junction" by convention). Each wire produces a voltage by the Seebeck effect according to its Absolute Thermoelectric Power and the temperature difference across its length. The difference of these voltages in the two wires is the measured quantity.
To know the temperature of the probe end, you must know the cold junction temperature because they can only be measured together. That is the reason "cold junction compensation" is required. If you introduce a third thermojunction midspan by "flipping the polarity" of the wires, you must now know its temperature as well; this could be provided by a thermistor or RTD at that point, but it needlessly complicates the apparatus, so this design is avoided. Instead, the same alloy wires are used all the way back to the measurement circuit (thermocouple extension wire).