Thermocouples operate on the principal that when two dissimilar metals touch, a small voltage is generated as a function of the hot junction temperature. ‘Hot junction’ is thermocouple terminology for the location where the two dissimilar metals touch. By connecting the thermocouple to a process controller or data logger, we create two additional junctions. The two junctions, referred to as cold junctions, also generate voltages. The cold junction voltages cause a large systematic error in the thermocouple output, significantly larger than those stated in thermocouple datasheets.
Cold junction compensation is the process of accounting for the cold junction voltage to prevent it from reducing the thermocouple’s accuracy. There are several strategies for cold junction compensation, including placing the cold junctions in an ice bath; this is common in calibration labs due to its high accuracy but impractical in industry. The standard compensation strategy is to measure the cold junction temperature and to calculate and output an equal and opposite voltage which cancels out the cold junction voltages. This sounds complicated but is normally performed by a simple Wheatstone bridge circuit with a thermistor.
- The temperature at the cold junction, is called the cold junction temperature
- The temperature at the hot junction is the process temperature
- The (uncompensated) output voltage is the sum of the voltage generated by the hot and cold junctions
Fortunately for process engineers, they need not worry about implementing cold junction compensation. Almost all process controllers and data loggers with thermocouple inputs have built in cold junction compensation. Furthermore, it is possible to purchase standalone thermocouple amplifiers with built in cold junction compensation. The standalone amplifiers convert the thermocouple signal from a mV voltage to any of the common output signal forms (e.g. 0-10 Vdc). Cold junction compensation isn’t perfect, so lookout for the compensation error in the datasheet. Compensation error is normally stated as measurement error per degree of temperature change (relative to a reference temperature of 0 °C), e.g. 0.05 °C/°C .
Calculating the error caused by uncompensated cold junctions
The measurement error due to a lack of cold junction compensation depends on the ambient temperature, process temperature and the type of thermocouple used. The error can be calculated using the following two calculators provided by Fluke (click on the images to access the calculators). Calculator one provides the correct thermocouple voltage for a chosen temperature. Calculator two uses this voltage and the cold junction temperature (which they refer to as the reference junction temperature) to calculate the uncompensated temperature. The error due to a lack of cold junction compensation is the difference in temperature between calculators one and two.