Temperature Sensing

Temperature Sensing

Thermopile Sensors and Thermopile Arrays can be used for a large variety of non-contact temperature measurement applications like in-ear or forehead thermometers, industrial process control or human presence detection in public buildings. To understand the potentials and limitations of thermopiles as temperature sensors, it is important to understand their working principles including how they convert infrared radiation to a voltage and then to a temperature value.

Infrared Basics

Thermopiles sense the electromagnetic radiation which is emitted from the surface of any object or body with a temperature above absolute zero (-273.15°C). This radiation has a broadband spectral distribution that depends on the surface temperature of the emitting body and can be described by Plancks radiation law. This is shown in the image below:
As you can see from the image, higher surface temperatures have two effects. First, the total emitted energy increases. This is described by the Stefan-Boltzman law:
The emitted energy increases in proportion to 4th power of absolute (K) temperature. Therefore a slightly higher object temperature will generate a more-than-slightly higher level of radiation. Small change in absolute temperature with larger change in emitted energy due to 4th power dependence, results in easier to detect temperature differences. While you may not see this effect clearly in the double logarithmic scale in the first graph, it is made even more obvious in the next graph, where the y-axis is not scaled in a logarithmic format:
Second, the peak wavelength of the radiation spectrum shifts to shorter wavelengths at higher temperatures. This is described by Wiens displacement law:
For ambient temperatures of 27°C the peak wavelength is at 9.66 µm. For 1000°C the peak wavelength shifts to 2.28 µm and for the surface temperature of the sun of about 6000°C the peak wavelength is about 0.46 µm which is in the visible spectrum of the human eye.

For most temperature measurements in the range of 0°C to 1000°C, the peak wavelength and therefore most of the emitted radiation is in the mid and far IR range which is - depending on the definition - between 3 µm and 15 µm:

Thermal Infrared Sensors

After learning the most important basics about thermal infrared radiation, we can now have a look at the working principle of a thermal infrared sensor. As depicted in the next figure, there are four elements which have to be considered. First, the radiation source. This is the object, body or surface which emits infrared radiation due to its surface temperature. Second, the infrared radiation propagates from the radiation source through the atmosphere towards the sensor. Before the radiation is detected by the sensor element, it can be manipulated by IR optics. The optics is the third element and can consist of a filter and/or lens. The last element is the actual infrared sensor which converts the incoming radiation to a signal, usually a voltage value suitable for display or further interpretive signal processing and action.

Influence of the radiation source

An ideal radiation source is called a black body. It emits the maximum possible thermal radiation for each wavelength and its characteristic is exactly described by Plancks radiation law. A black body will not reflect any light nor has it any transmission of light. This means that 100% of the incoming light is absorbed by the black body. This absorbed radiation energy raises the temperature of the absorbing black body and therefore the radiation emitted by that black body is also slightly increased according to the Stefan-Boltzman law, which says that higher temperature objects emit more radiation. Simply stated: The absorbed radiation is re-emitted; or in other words the absorption and emission coefficients are the same and equal 1.

In the real world, no perfect black body exists and the emissivity of the object will be lower than 1. Consequently, the emitted radiation will be lower than described by Plancks radiation law. If this radiation is detected by our sensor, it will also show a lower temperature because it receives less radiation than expected. To compensate for this effect we have to know the surface emissivity of the object to be measured. Emissivity is a factor accounting for the deviation of the real object from a perfect black body.

Conveniently, the human skin is an almost perfect black body with an emissivity of about 0.98, while most metals and other typical industrial targets have lower emissivity factors which may also change with temperature and oxidation of the metal in complex ways. Therefore, the measurement of the surface temperature of metals is much more difficult.

Influence of the Atmosphere

While the influence of the non-ideal black body behavior of real objects can be accounted for if we know the emissivity of the surface, compensating for the influence of the atmosphere is not always as easy. To understand this, it is important to understand how  infrared light transmission is affected by the atmosphere.

The atmosphere contains gases like H2O, CO2, O3, N2O, CO, CH4, O2 and N2 to name just the most common ones. When the infrared radiation from our object of interest interacts with one of these gas molecules, the light may get scattered or absorbed. This happens at characteristic wavelengths for each gas corresponding to their molecular structure. Suddenly the smooth and continuous spectrum of infrared black body radiation is seen to be intersected by deep dips where the absorption of one gas or another has attenuated the energy selectively by wavelengths characteristic of the gas that was present. While certain spectral ranges show high absorption in normal atmospheric conditions, the radiation in other ranges can pass almost unhindered. The high transmission spectral ranges are referred to as atmospheric windows. There are two atmospheric windows that are important for our thermal sensors:
  • 3 to 5 µm (mid infrared)
  • 8 to 14 µm (far infrared)
If we limit our measurements of temperature or of trace gas concentration to these spectral ranges, the influence of the atmosphere on  measurement accuracy is improved. Nevertheless, even in these atmospheric windows the transmission is not 100%.  Beer’s Law tells us that if the radiation passes longer distances through the atmosphere, this small absorption in the window region will become increasingly relevant for the measurement results as distance increases. If the sensor is placed close to the measurement object, the minor absorption in the atmospheric windows can be neglected in most cases.

Influence of the Optics

The next element in the system is the optics. To limit the received radiation to a defined spectral range, optical filters are necessary. While an optical filter may limit the spectral range to e.g. 8-14 µm, it will itself have a transmission factor which is less than 100%. This results in a further reduction of the infrared radiation which is received by the sensor. All these effects have to be considered in order to determine the surface temperature of the object of interest. At least you should know about these effects to get an idea of how accurate your measurement will be.

Working Principle of Thermopile Sensors

Infrared radiation, diminished by its emissivity, by some atmospheric absorption, and by the transmission of optical elements finally reaches our thermal sensor element. We can now have a look at how a thermopile sensor works. The conversion of incoming radiation (heat) to an output signal (voltage) is based on the thermoelectric effect, also called the Seebeck effect. The thermoelectric effect describes the separation of charge carriers inside an electric conductor due to temperature differences. If one side of the conductor has a higher temperature than the other, the charges will be in disequilibrium. The degree of charge separation depends on the material and is a constant called the thermoelectric coefficient. It can be either positive or negative. If we connect two conductors with differing thermoelectric coefficients on one end and subject it to a high temperature T 1 , we get a thermocouple. Then we can measure a charge difference (= a voltage) between the two other ends which are located at a distance and at a lower temperature T 2 .

As the thermoelectric coefficients (α1 and α2) are material constants, the relationship between temperature difference ΔT and output voltage is strongly linear and can be described as follows:
However, the output voltage of a single thermocouple is very small - in the range of µV. Therefore it is normal that many thermocouples are connected in series to get a higher output voltage value. This collection of thermocouples is commonly referred to as thermopile. Nevertheless, even with a thermopile producing higher signal voltages, low-noise signal processing is needed to generate a useful output signal voltage with good signal to noise ratio..
Share by: