How does a thermocouple work, and do I really need an ice bath (part 2 of 2)?

Modern thermocouple instruments automate the cold-junction-compensation process.

Part 1 of this two-part series introduced thermocouple cold-junction compensation (CJC). This part will elaborate on the concept. Table 1, presented in part 1 and repeated here, lists a subset of the NIST voltage and temperature data for the type T thermocouple.Figure 1 shows a typical measurement setup, employing a type T thermocouple and a 40°C cold-junction temperature.

Figure 1. This setup shows a type T thermocouple with a 40°C cold junction.

Right. I noted that based on Table 1, the 12-mV voltmeter reading corresponds to 250°C. Why can we not add the 40°C reference temperature to 250°C to determine that the temperature of interest is 290°C?
First, note that the relationship between a thermocouple’s voltage V and temperature T is defined in terms of the Seebeck coefficient S, where S=DV/DT. From Table 1 you can estimate that S for a type T thermocouple equals about 14.862 mV/300K, or 49.54 µV/K. However, S itself is a function of temperature, so the type T thermocouple’s voltage-temperature curve (blue line in Figure 2) is not linear, and we can’t simply add temperatures that way.

Figure 2. A thermocouple’s voltage-vs.-temperature curve (blue) is nonlinear, as the linear red line emphasizes.

How do we proceed?
Figure 3 illustrates the approach. At the top, we have a type T thermocouple with a cold-junction temperature of 40°C, and the voltmeter reads 12.013 mV. We surmise, however, that because the cold junction is 40°C instead of 0°C, the voltage should be higher. We need to find this “missing voltage,” as it’s sometimes called. To do that, we can use another type T thermocouple to measure the first thermocouple’s cold-junction temperature, and then drag out our ice bath to use as the second thermocouple’s 0°C cold junction. As shown on the voltmeter on the bottom right of Figure 3, the “missing voltage” is 1.612 mV.

Figure 3. In this setup, a second type T thermocouple and voltmeter find the “missing voltage.”

Of course, we didn’t really need the second thermocouple and ice bath to do this. We can get the 1.612-mV value directly from Table 1. Now, we simply add the missing voltage to the original voltage to get 13.625 mV, which from Table 1 corresponds to 279°C.

How does CJC work in real life?
A modern thermocouple and voltage-measurement instrument makes the process simple by employing an internal cold junction (Figure 4). The instrument uses a thermistor to continually monitor the cold-junction temperature of each input channel and compensates accordingly. Unlike in Figure 4, the instrument won’t display intermediate results. In fact, it might be a faceless LXI instrument that displays nothing at all.

Figure 4. A modern thermocouple instrument employs a thermistor (T) to measure the cold-junction temperature.

What else should I know?
Keep in mind that not any thermocouple will operate over the entire IEC range for that thermocouple type. The maximum and minimum temperatures for any particular thermocouple will depend on factors such as the wire gauge, the type of insulation, and your accuracy requirements. Consult the manufacturer’s data sheet before buying.

I should also note that “dragging out the ice box” now is not nearly as inconvenient as it was in the 19^th century. For example, Isotech makes an all solid-state 0°C reference, and Omega Engineering makes a calibration chamber that uses real ice. Whether you choose a standalone 0°C temperature cold-junction box or an instrument with internal CJC depends on many factors. The standalone cold-junction approach can double your wiring requirements—you have to connect your thermocouples to the cold-junction box and then connect the cold junctions to your voltage-measuring instrument. If you rarely change your test setup, and if your sample rates are low enough that you can use a scanning DMM to sequentially read the voltages across all your channels, then a standalone cold-junction box may be your optimal solution. However, if you need high sample rates, require many channels, and change your test setup frequently, then an instrument with multiple high-speed channels and internal CJC may be the way to go.

Where can I read more?
Omega engineering has an introduction to CJC. Tegam has an article on the CJC used in its thermocouple thermometers. Beamex has a blog post on the topic. DATAQ Instruments has information on using thermocouples at less than 0°C.

Finally, this paper from the University of Saskatchewan provides a clear, concise, yet rigorous explanation of thermocouple basics. It explains, in detail, why voltages develop across metal wires when the ends are at different temperatures. Contrary to popular belief, there is no voltage developed at the junction of a thermocouple’s two wires. The two wires generate different voltages across them, thus the measured voltage at the cold-junction compensation point stems from those different voltages. That’s why the cold junction is made from a metal block, to keep its connections at the same temperature. A difference in temperature between the two cold junction connections makes things far more complicated.