This website uses cookies primarily for visitor analytics. Certain pages will ask you to fill in contact details to receive additional information. On these pages you have the option of having the site log your details for future visits. Indicating you want the site to remember your details will place a cookie on your device. To view our full cookie policy, please click here. You can also view it at any time by going to our Contact Us page.

Tutorial: Understanding the tricky thermocouple

26 January 2009

Following the article ‘Challenges of Temperature Sensing’ we received comments that suggested thermocouples were not characterised well. Peter Welander goes into more detail on these often misunderstood devices.

Illustration 1
Illustration 1

Here is the main statement made in the article that readers took issue with, and clarification: ‘Thermocouples indicate temperature by providing a very small voltage signal generated by a junction of dissimilar metals.’ This describes the apparent action, but technically it is not true. (Mea culpa.) Am I splitting hairs? To an observer on earth, you could say that day and night are caused by the sun orbiting the earth. It describes the effect, but is wrong. In the same way, you deserve to know more of the truth about thermocouples.

The voltage that provides the measurement is not generated only at the junction. Although some sources suggest this is the case, there is nothing magic about the meeting point of the dissimilar metals.

What does matter is that the junctions are the ends of the wires, and that is important. The voltage is actually generated over the entire length of a wire, anywhere there is a temperature gradient.

Maybe this verbal illustration will help. Think of a wire as a hose filled with water and with both ends capped. If you lift one end of the hose higher than the other, gravity will cause pressure in the hose. Anywhere there is a gradient, pressure will develop. (This is an imperfect model, but work with me.) If you cut the caps off the hose, keep it completely full of water, and connect the ends to a differential pressure sensor, the sensor will always read zero (assuming a normal sensor) because the two ends of the hose are at the same level. Regardless of how the hose may go up and down, all gradients even each other out.

Thermocouples work because heat creates a thermoelectric voltage in a wire. This is the Seebeck effect. Anywhere there is a temperature gradient, there will be a voltage because electrons want to flow from hot to cold. The voltage value per degree temperature difference is the Seebeck coefficient and depends on the characteristics of the specific wire alloy. But if the circuit is not complete and the voltage has nowhere to go, you can’t use it.

Bringing both ends of the wire to a voltmeter (Illustration 1) would be like the hose analogy because the ends will be the same temperature and the voltage evens out.

Illustration 2
Illustration 2

But, if you use two different kinds of wire that have a different Seebeck coefficient, you can measure the difference in the voltage produced by the two wires over the same temperature gradient. (Illustration 2) That’s why thermocouples use dissimilar metals. They’re chosen because they produce different voltages for a given temperature gradient—hence, the common belief about the junction creating the signal. What you’re measuring is the voltage difference created by the overall temperature gradient of the two types of wire. The junction is simply the ends of the wires, and that is the relevant measuring point.

Thermocouple purists get hot about this (pun intended) because they believe if you don’t understand these basic facts, you won’t be able to apply thermocouples well, you won’t understand how and why they drift, and you will tend to dismiss them as inaccurate. For example, people who think the junction is the working bit of the thermocouple tend to forget about the rest of the wire, and this is dangerous. Anywhere there is a physical or chemical change in the wire, it can affect the reading. If it’s corroded somewhere in the middle away from the junctions, it can change the voltage and resulting reading.

Similarly, mismatched extension wire, poor connections, and lax maintenance can all degrade thermocouple performance. A technician looking only at the sensing junction may say it’s OK and not understand why his or her thermocouple has drifted due to problems with the wire in another area. If you don’t understand the real causes, you come to the wrong conclusions.

If you have high quality wire with carefully measured Seebeck coefficient characteristics, have it in an application where it is protected from physical and chemical changes, and have reasonably sophisticated signal processing, thermocouples can indeed be very accurate and very stable.

Three illustrations:

1: If you have the same kind of wire coming to both sides of the voltmeter, your reading will be zero because the temperature gradient on both sides will be the same and they will balance each other.

Illustration 3
Illustration 3

2: If you have two kinds of wire with sufficiently different Seebeck coefficients, you will see a voltage difference because one wire will produce more voltage than the other for the same temperature gradient. If you know the thermoelectric characteristics of the wire and know the temperature at the voltmeter, you will be able to calculate the temperature of the flame from the voltage measurement.

3. This shows the more traditional thermocouple configuration.

The sensing wire extends from the flame to the ice bath, which provides a known reference temperature.

(Real-world process installations do not normally use an actual ice bath, but a second temperature measuring device.) The voltage is created by the temperature gradient in the wires, anywhere in their length where a gradient exists.

—Peter Welander, process industries editor, peter.welander@reedbusiness.com,

Control Engineering Process Instrumentation & Sensors Monthly


Contact Details and Archive...

Related Articles...

Most Viewed Articles...

Print this page | E-mail this page