Thermocouples are widely used for measuring temperature in industrial applications. They are cheap, rugged, reasonably accurate and are able to measure temperatures from -200°C to well over 1600°C. At their simplest a thermocouple is just a welded junction of two wires made from dissimilar metals, as shown in Figure 1, although they are available in a wide range of physical forms.
A simple K-type thermocouple. Note the Chrome-Alumel junction at the bottom of the picture. In this case the two dissimilar metals are welded together, and the junction left exposed. You can also get thermocouples in a wide range of packages (electrically isolated or not) and formats. They are relatively cheap and pretty rugged given their very simple construction.
The fact that any junction of two different metals will produce a temperature dependent voltage was discovered by Thomas Seebeck in 1821 (and is named the Seebeck efect in his honour). The voltage developed in a single junction is small – in the microvolt range – and depends on the materials used as well as the temperature. Measuring this voltage and extracting a meaningful temperature value is a non-trivial exercise.
Thermocouples are available off-the-shelf in a variety of standard types (defined by the materials used) shown in Table 1 below. The most common for general purpose use seems to be type K which are made from alloys of Nickel (Nickel-Chromium and Nickel-Aluminium) often referred to as “Chromel-Alumel” thermocouples. Selecting the right type and package for your application is beyond the scope of this article.
The challenge for the electronic designer is to pick out the relevant signal and convert that into some quantity that accurately represents the temperature of the junction. Unfortunately, this is not as simple as it might appear. Consider the schematic in Figure 2. As well as the “hot” junction A, that we are concerned about, there will also be other dissimilar-metal junctions in our circuit. For example, at some point B, there will be a transition from the thermocouple materials to traditional conductors such as copper. There may also be other transitions to other materials in the measurement circuit itself.
This diagram shows that in any thermocouple circuit there may be several dissimilar-metal junctions, each of which will contribute a Seebeck voltage. The effect of the two junctions at C cancel out since they are made of similar materials and are at the same temperature. This is not the case for the junctions at point B where the junctions are different. Together these junctions act like a single Chromel Alumel just like that at point A. This means the voltage we measure will effectively be proportional to the difference in temperature between the hot and cold junctions.
Hopefully it’s clear that we don’t need to worry about the metal to metal junctions at point C. As long as the transitions are between the same metals (copper to other, and other to copper in this case), and the junctions are at the same temperature, the Seebeck voltages will cancel each other out. At point B however, this is not the case since the two junctions are different (Chromel to copper and copper to Alumel).
In fact, we will develop a net voltage across the cold junctions that is the same as that between the two thermocouple materials (Chromel and Alumel in our case). In effect, the sense circuit will see a voltage that is proportional to the difference between the temperatures of point A (the hot junction) and point B (the cold junction). Assuming we want a measurement referenced to 0°C, we need to keep the cold junction at that temperature.
— ADVERTISMENT—
—Advertise Here—
One way to do this is to immerse the cold junction in an ice-water slurry. This is pretty difficult to do outside a laboratory, so we typically resort to cold junction compensation where we separately measure the cold junction temperature and apply the appropriate offset to the thermocouple voltage either in the analogue or digital domain.
Figure 3 shows how this works. For a given hot and cold junction temperature, we will measure a voltage Vuncomp that is proportional to the difference in temperature between the hot and cold junctions. To reference this to 0°C, it is necessary to add the voltage Vcomp to get a voltage proportional to the temperature refenced to 0°C.
The uncompensated voltage Vuncomp, is that measured in Figure 2. If the cold junction was held at 0°C, this would be proportional to the absolute temperature (in °C) of the hot junction. Cold junction compensation works by providing a voltage Vcomp, that compensates for the difference in temperature between the cold junction temperature and 0°C.
If you are going to roll your own thermocouple circuit, you will need to amplify the measured voltage by a factor of a few hundred. The op amp you chose should have low offset and, more importantly, low offset drift. You will most likely have to use a “zero drift” op amp.
One possible choice is the LT1049 from Analog Devices. Figure 4 shows an extract from the data sheet for this chip where it is being used in precisely this application. The gain is set to 246 to get approximately 10mV per °C from a K-type thermocouple. An LT1025A cold junction compensator is being used to add a compensating voltage to produce a 0°C referenced output. This latter device can support E, J, K, T, R & S type thermocouples.
This figure, extracted from the LTC1049 data sheet, shows how an analog thermocouple amplifier might be configured. Note the use of the LT1025 Cold Junction Compensation IC. This adds a small voltage proportional to the temperature in °C. Since we are dealing with microvolt-level signals, the op amp should be a zero drift or chopper type.
If you need a digital output there are plenty of devices such as the MAX31855 which provide signal conditioning, cold junction compensation and analog to digital conversion all in one package. These usually also provide fault detection to identify open or shorted thermocouples which may or may not be useful depending on your application. You can also perform the cold junction compensation in software if you digitise both the uncompensated thermocouple voltage and the temperature of the cold junction.
References
“Common Thermocouple Types.” Accessed April 24, 2023. https://www.tcaus.com.au/thermocouple/thermocouple-types.html.
“Thermocouple.” In Wikipedia, February 5, 2023. https://en.wikipedia.org/w/index.php?title=Thermocouple&oldid=1137656823.
“LTC1049 Datasheet and Product Info | Analog Devices.” Accessed April 24, 2023. https://www.analog.com/en/products/ltc1049.html.
“LT1025 Datasheet and Product Info | Analog Devices.” Accessed April 24, 2023. https://www.analog.com/en/products/lt1025.html.
— ADVERTISMENT—
—Advertise Here—
“MAX31855 Cold-Junction Compensated Thermocouple-to-Digital Converter | Analog Devices.” Accessed April 24, 2023. https://www.analog.com/en/products/max31855.html.
Sponsor this Article
Andrew Levido (andrew.levido@gmail.com) earned a bachelor’s degree in Electrical Engineering in Sydney, Australia, in 1986. He worked for several years in R&D for power electronics and telecommunication companies before moving into management roles. Andrew has maintained a hands-on interest in electronics, particularly embedded systems, power electronics, and control theory in his free time. Over the years he has written a number of articles for various electronics publications and occasionally provides consulting services as time allows.
