Temperature sensing is one of the most commonly specified attributes for monitoring and control of critical processes. So many physical and chemical processes rely on close control of temperature that even very simple processes can be ruined by operating in the wrong temperature regime. Think about refrigeration, food processing, manufacturing metal alloys, production of medicines, control of battery grid systems and so on.

Given the criticality of maintaining the proper temperature for a whole lot of process it is important to understand what the options are for monitoring and control of temperature. This article covers the most common sensing technologies, how they work and the pluses and minuses of each approach.

Source: Surasak/Adobe Stock

Thermocouples

By far the most common physical process that is sensed for industrial and commercial applications is temperature across the range from around -50° C to 150° C. Since the early 1800s, thermocouples did the bulk of temperature sensing. It was 1821 when it was discovered that when two dissimilar metals were intertwined and heated at the junction, that a small volage was generated across the junction. This voltage was roughly proportional to the temperature of the junction. This was dubbed the Seebeck effect, named for German physicist Thomas Johann Seebeck. Of course, different metal pairs give different results, so the thermocouple method has evolved into a family of different wire pairs depending on the temperature range being measured. One advantage of thermocouples is that with the various metal combinations, a wide range of temperature measurements can be achieved. Refer to the partial table shown below:

There are at least 11 “named” thermocouples that are used in a range of applications from petrochemical refining through aerospace applications. Operating temperatures up to 2,320° C are possible using refractory metals, however these materials are easily oxidized so they must be used in a vacuum or in an inert gas atmosphere. These are suitable for use in the vacuum of space.

Uses for thermocouples

As illustrated by the partial table shown above, certain thermocouple material pairs have become associated with specific processes or industries. Since thermocouples have been around for the longest of most common temperature measuring devices a number of dual alloy combinations have been tested and codified. Generally speaking, the dozen or so common combinations have been well characterized to U.S. National Institutes of Standards and Technology (NIST) standards and are very predictable. However, care must be used in designing thermocouple systems. They can be easily damaged if not mounted securely and they are usually placed as near to the heat source as possible. The electronics, in order to protect them, would need to be mounted far away from the heat source. This creates another attachment point with a dissimilar metal junction requiring the creation of a compensating junction (referred to as a “cold junction”). Clearly, some installations can get tricky. One unique use is based on the fact that thermocouples are voltage-generating devices. This makes them suitable for creating a voltage in space through radioactive heating of an on-board thermocouple array.

About RTDs and their uses

Another common temperature sensor is the resistance temperature detector (RTD). These work pretty simply by using the change in resistance of a metal alloy, usually of platinum, with temperature changes. Use of exotic metals makes these an expensive option. Versions are available that use nickel-platinum alloys; however, they are not as sensitive and are prone to drifting. These devices work best with a high surface area and relatively thin cross section. They are often fabricated by deposition of a thin platinum film on a ceramic structure. Like most temperature sensing devices, the point of temperature measurement is usually remote from the electronics. This brings up the issue of proper mounting, for these types of sensors as well. The mounting needs to be very secure to avoid damage or fatigue at the mounting site. These devices are fairly linear in the range of -200° to 850° C. Due to the good linearity, especially at high temperatures, they are commonly found in chemical processes, exhaust gas measurement, industrial boilers and other high temperature applications, including exhaust gas temperatures as part of the exhaust gas recirculation (EGR) system on passenger cars.

Thermistors

Another common temperature sensor device is the thermistor, which, like an RTD, is a resistance device. Thermistors are composed of a proprietary combination of metallic oxides, binders and stabilizers. The formulation and final size and packaging determines the shape of the temperature resistance curve. One common type is a bolt-on mounting, since the thermistor can be produced to be round with a hole through the middle, it can bolt directly to the surface. Versions made for high temperatures are more prone to corrosion just due to the fact that high temperatures tend to speed up oxidation processes. One typical protection is to encase the sensor in glass. Also there is no standardization for thermistors, meaning they are generally not interchangeable, which can negatively affect maintenance and consistency of performance from manufacturer to manufacturer.

Solid state temperature sensing (SSTS)

With the ubiquity of growth of internet-connected devices, there is a growing desire to make use of local area networks for connectivity. This shows up at the consumer level with “smart homes” that can monitor and control thermostats. They often have cameras built into their doorbells and interconnected fire and carbon monoxide alarms. Broadly speaking all components in the home, commercial settings and factories should be migrating toward internet connectivity. The issue with thermocouples, RTDs and thermistors is that they were not originally conceived as being used in this way. This situation favors the growth of the solid-state temperature sensor.

Since SSTS are built using semiconductor techniques, they can be built directly into the feedback circuitry and monitoring programs that are demanded by a network. For the automated factory environment, the bulk of temperature measurement requirements fit into the range of -50° C to 150˚ C. Using a semiconductor footprint holds a lot of advantages for a solid state solution.

Being able to use existing fabrication techniques for semiconductors has a tremendous advantage for solid state temperature sensors. In this context, existing component spacing, packaging options and mounting configurations can be substantially the same as existing manufacturing dimensions.

Additionally, components can be tested and calibrated using existing techniques (i.e. laser trimming for higher performance). Automated test fixtures, manufacturing data and calibrations can be automatically produced as part of the manufacturing process to provide manufacturing process control and lot control.

Since temperature control implies feedback, the control circuitry for providing this feedback can be incorporated directly into the control circuitry on the same board or at the chip level as desired. This control information can be shared throughout the network, since it would be easy to include in a networked system, allowing an operator to monitor and react to any alarm information in real time.

With all of the positive attributes that come with a solid state sensor, the uses are varied and flexible. They can be found in engines, battery monitoring systems for EVs, most factory environments, and they are now starting to migrate into the medical field potentially for use in portable implants for health monitoring.

How does solid state sensing work?

To start with, the basic sensor is a diode that uses one of the typical characteristics of a diode: namely V_t which is the voltage across the P-N junction as a function of the temperature. This is an inherently predictable non-linear relationship. The compensating components to linearize the output can be manufactured on the same portion of the PC board as the diode circuitry, which effectively reduces the non-linearity to an acceptable value at a reasonable cost. They can also incorporate a laser trimming process into the manufacture to fine tune the output further.

This device was introduced decades ago as a simple current output device with a scaled output of 1 µA/K, laser trimmed to produce 298.2 μA at 298.2 K (25° C). This has the advantage of operating much like a 40-20mA current loop, reducing its noise susceptibility. This is also easily converted to a voltage output, which can be digitized via an A/D converter to produce a voltage output easily incorporated in an IoT environment.

Conclusions

Although there are a range of temperature sensors available to designers, each type of sensor has its own unique mounting requirements and performance characteristics. In terms of development, solid state temperature sensors are the youngsters (thermocouples have been around for 200 years).

Recently, the growth of semiconductor-based manufacturing and the move to solid state designs has been a prime mover for all manner of electronics devices. A huge amount of infrastructure has been built around smart phones, flat screen TVs, computers, EVs and so on. The move toward solid state solutions, driven by software has been the source for a period of unprecedented innovation and connectivity.

It is this growth trajectory that has driven the move to solid state temperature sensors. The ease of incorporating a new type of temperature sensing device has created the solid state solution. It is the most flexible, it covers the “sweet spot” for most temperature sensing needs and it is readily incorporated in designs due to the wide infrastructure support. All of this adds up to a cost effective and expansive solution to a 200-year-old problem.