Thermocouples

The principle

Join two dissimilar metals and a small voltage appears that depends on the temperature difference between that junction and the other end of the circuit. That is the thermoelectric effect, and a thermocouple exploits it.

The critical consequence: a thermocouple measures a temperature difference, not an absolute temperature. To get an absolute reading, the temperature at the point where the thermocouple wires meet ordinary copper — the cold junction — must be measured separately and added in. This is cold junction compensation, and every thermocouple input card or transmitter does it automatically using a temperature sensor at its own terminals.

It follows that if the terminal temperature at the input card is not what the compensation believes it to be, every thermocouple on that card reads wrong by the same amount. A cabinet with a failed cooler or an open door produces exactly this.

Common types

Type	Materials	Notes
K	Nickel-chromium / nickel-alumel	The general-purpose type. Wide range, cheap, widely available. Drifts in some conditions.
J	Iron / constantan	Good sensitivity, limited upper range, the iron leg corrodes.
T	Copper / constantan	Good at low temperatures, including cryogenic service.
E	Nickel-chromium / constantan	Highest output of the common base metal types.
N	Nicrosil / nisil	Better stability than type K at high temperature.
R, S, B	Platinum alloys	Very high temperature, high accuracy, expensive; used in furnaces and fired heaters.

Practical points

• Extension and compensating cable must be of the correct type and correct polarity all the way to the cold junction. Any length of copper wire in between creates an unintended junction and an error. Getting the polarity reversed produces a reading that moves the wrong way as temperature changes — a classic and very recognisable fault.
• Thermocouples are low output and high impedance relative to noise, so screened cable, single-point screen earthing and separation from power cables matter more than usual.
• Grounded, ungrounded and exposed junctions: grounded junctions respond faster but can create earth loops; ungrounded junctions are electrically isolated and are the usual choice; exposed junctions respond fastest but are unprotected.
• Thermocouples drift with use, particularly at high temperature and in contaminating atmospheres. They are cheap to replace and should be, rather than trusted indefinitely.
• They cannot be checked for continuity meaningfully with an ordinary ohmmeter in the way a resistance thermometer can — a low resistance says the circuit is intact, not that the sensor is accurate.