Choosing the right temperature sensor

Temperature sensors are an effective way to measure temperature, but which should you use, and for which application? Sensors expert Moore Industries shares some tips.

A very efficient way to measure temperature is with either a thermocouple (T/C) or a resistance temperature detector (RTD). Both sensors perform contact temperature measurement; that is, they are attached to the object or in contact with a process medium, and measure its temperature at the point of contact. For industrial use, RTDs and T/Cs come in several versions: flexible, straight, fixed immersion probes and sanitary sensors, among others. Sensors can be inserted into thermowells or protection tubes, welded into place on boiler tubes or other objects, or clamped down for surface measurements.

Both sensors are widely preferred because of their ruggedness, durability, and the wide ranges of temperatures they can measure in industrial applications. RTDs and T/Cs are generally used with a signal conditioner or transmitter which converts the output signal into a voltage or current, such as 1-5V or 4-20mA. Each also has distinct advantages in certain applications, where T/Cs are typically used for extreme temperatures and faster response times, while are used mostly for their accuracy, repeatability and higher immunity to noise interference.

Thermocouple types

A thermocouple consists of a pair of dissimilar metals welded or fused together at one end to form the ‘hot’ or measuring junction. A small voltage or electromotive force (EMF) is then generated by this union of dissimilar metals, and temperature is then derived by measuring the voltage levels and comparing it to known element type lookup tables that are stored in temperature transmitters or thermocouple input cards.

A sometimes overlooked measuring junction that is vital to this measurement is the ‘cold’ junction or ‘reference’ junction. Since many terminals are brass or nickel-plated brass, in effect there is another thermocouple created when you hook the T/C lead wires to the input terminals of a temperature transmitter or T/C input card. Therefore, to cancel out the EMF created at the terminals the measuring device must monitor the temperature at the terminals so it can effectively cancel out any unwanted voltage levels created at the terminal block or input block.

Thermocouples come in a variety of types and cover a wide range of temperatures. Each T/C type has a specific temperature range and set of environmental characteristics that make it suited for certain applications. All thermocouples come in three types of configurations: grounded, ungrounded and exposed. Ungrounded is the most popular, while grounded and exposed are used for faster response times.

While isolated in the beginning, over time ungrounded T/Cs have a tendency to go to ground. Therefore, care should always be taken when installing T/Cs to ensure that you are using instrumentation with proper isolation levels. We have seen too many occurrences where the T/C packing or insulating material breaks down over time and causes a ground loop in a non-isolated measuring circuit.

Type T—This thermocouple is best used in extremely low temperatures, and is widely used in the food processing and medical industries, where subzero temperatures are most likely to be maintained. These thermocouples have special metal properties that are excellent in oxidising and reducing atmospheres. They are also the most stable in cryogenic temperature ranges.

Type J & K —These are the most common sensors, and are used in all industries because of their wide ranges. They are also the most linear of all thermocouples. The K T/C has a greater temperature range and tends to be used in foundries where the sensor is exposed to the elements. The K type T/C also has reduced corrosion properties when exposed to oxidising atmospheres.

Type E —When temperatures become critical and measurements are in small increments, it may be necessary to use the E type T/C. This is because they have the greatest change in mV over the smallest temperature change, allowing for very accurate readings in the chemical and medical industries. This thermocouple is not subject to corrosion where the moisture content is high at very low temperatures.

Type R, S, B —These thermocouples are used in areas where the temperatures are too high for J & K type T/Cs. For example, in steel mills and glass foundries, the temperatures are much greater—as high as 2800°F. These T/Cs have longer life expectancy at high temperatures when protected by a thermowell.

Type C — This can reach temperatures as high as 5000°F, and is widely used in space vehicles, nuclear reactors, industrial heating, or wherever extremely high temperatures may be present. The only downfall to these thermocouples is that they become extremely brittle at high temperatures and require extra protection.

RTD types

Metals will change in resistance when subjected to a temperature change. This relationship can be predicted by the use of a constant (alpha)—the temperature coefficient of resistance. A metal’s alpha may be changed by combining it with another material, or by mechanically stressing it, thus forming a resistance temperature detector (RTD); that is, a sensor whose resistance changes with temperature.

The resistance element of an RTD is the temperature-sensing component at the tip of an RTD probe. Typically, the sensing element cannot be used in its bare form, so it is built into a probe or an assembly that can withstand the various conditions of a specific application.

As with T/Cs, the output of an RTD is converted to a temperature by a transmitter or other device that can deal with the sometimes nonlinear output of a RTD. A ‘smart’ transmitter can be further ‘trimmed’ to increase the accuracy of an RTD over a measurement span of interest.

The most commonly-used materials for RTDs are platinum, nickel, and copper. These metals each have a specific temperature curve that is best suited for use in certain temperature ranges.

Platinum RTD—The most popular RTD. Platinum elements have the ability to withstand oxidising conditions, where others would corrode. This material has an operating range of -32 to 1562°F (-200 to 850°C). A special consideration when selecting a platinum RTD is its resistance. The choices are, typically, 100, 200, 500, and 1000 ohms @ 0°C. Each RTD type represents distinct degrees of sensitivity in the relationship to resistance. The greater the change in ohms per degree value, the greater the resolution and accuracy in narrow ranges.

Copper RTD—The most linear RTD. The chief advantage of copper RTDs over platinum and nickel is that it is the most linear of metals available. This linearity simplifies the process of accurately scaling the RTD’s reading. The most popular uses for this type of sensor are in motor windings, generators, and turbines. Copper RTDs have a much smaller resistance change ‘span’ than platinum and nickel which limits its range to -58 to 482°F (-50 to 250°C). Copper RTDs also have a tendency to oxidise at high temperatures.

Nickel RTD—A good compromise, or the best of both worlds for a price. Nickel is the ‘middle of the road’ RTD, costing a bit more. It has a base ohm resistance value of 120 ohms at 0°C. Nickel RTDs tend to become non-linear at temperatures above 300°C. Nickel RTDs have a temperature range of -112 to 608°F (-80 to 320°C), but perform well when protected by a thermowell.

RTD elements

An RTD may consist of a coil of wire, foil, or a thin film of material deposited onto a suitable insulating core. The manner of supporting the metal is crucial in sensor design because of the effect of thermal expansion and the strain on the resistance. An ideal mounting would impose no strain to the metal being used.

Wire wound—a sensing element made with a precision piece of wire wound into a coil, and encapsulated in ceramic, mica or a glass insulator. This element is then held in place with an adhesive and packed with aluminum oxide or magnesium oxide. The result is that each turn is free to move and the strains from thermal expansion are negligible. Partially-supported elements are well suited for industrial applications. Designs are available that are quite shock resistant (up to 30g causes no change in calibration) and they can be used from -260 to 800°C. However, the fabrication of this type of element is generally very expensive.

Thin film—a sensing element that is fully supported. Platinum is deposited on the surface of an insulating support piece (usually ceramic). Manufacturers have designed elements that permit simple adjustments of the resistance to increase accuracy. For example, the film can be made in the form of a grid, and filaments in the grid can be severed with a laser beam to produce the needed result. Film-type sensors are rugged and inexpensive because of automated fabrication methods. The only downside is that thin-film elements are more susceptible to strain effects, but are still quite satisfactory and very accurate for most applications.

Rigid vs flexible sensors

A standard rigid temperature sensor, made by virtually every sensor manufacturer in the world, consists of a T/C or RTD sensor element protected inside a rigid stainless steel shaft.

Typically, the T/C or RTD element is installed inside the bottom two inches of a stainless steel tube, which is then filled with mineral insulated powder, such as magnesium oxide (MGO), and sealed to prevent moisture penetration. A rigid sensor assembly is commonly used with a thermowell. While this solution has worked for 30 or more years, rigid sensors have always posed installation and maintenance problems because of the difficulties involved in working with a rigid sensor, keeping the correct spares, and replacing them in ‘sagging’ or dirty thermowells.

When used with thermowells, for example, a rigid sensor has to be the correct length to fit. That means a plant must keep several different lengths of spares in stock to fit every thermowell. If a thermowell sags from extreme heat or fills with debris, a replacement sensor often will not fit, and the thermowell needs to be replaced. Replacing a rigid sensor can be difficult.

A recent development is the flexible sensor, which typically consists of a one inch, stainless steel sheath with an element and lead wires that are protected either with Teflon or fibreglass insulation. Flexible sensor wires can be trimmed to the correct length, simplifying the need for spare parts because one sensor size fits all. A flexible sensor often will fit into a sagging or dirty thermowell, and its replacement is much simpler, because only the enclosure cap has to be removed to insert a new sensor.

Lead resistance affects accuracy

Extension wires leading from the T/C or RTD sensing element to a termination point, such as a signal conditioner or transmitter, can affect performance.

For a T/C, extension wires can be critical. If excessive electrical or radio noise is nearby, T/C extension wires—from the sensor’s connection point to the transmitter—can pick up electrical spikes from up to 100 feet away. The T/C should be close coupled to the transmitter to filter out these noises.

An RTD can be 2-wire, 3-wire or 4-wire. A 2-wire extension lead is used where the lead wire is not very long and the accuracy is not critical. For higher accuracy, 3-wire RTDs have been used to compensate for the lead wire resistance error. A 4-wire RTD provides the highest degree of accuracy and long term repeatability, as a 4-wire monitoring circuit does not compare circuit resistances on each side of the element but instead uses a constant current source through the outer legs and takes a voltage measurement on the two internal legs. These internal lead wires are tied to two high impedance terminals, which virtually eliminates current flow into these legs, allowing the measuring device to isolate the RTDs resistance (a voltage measurement created by the current flowing across the RTD element) only and allows for a more accurate measurement. The transmitter can then cancel out all errors due to length and resistance imbalances between the leads.

Resistance variance may develop between RTD lead wires because of corrosion. Extension wires corrode at connection points such as splices, terminal blocks, and between the element and lead wires. Monitoring devices that are hooked up to 2-wire and 3-wire RTDs cannot provide lead wire imbalance correction. So unless you can guarantee that the corrosion or resistance imbalance will be exactly the same in all legs, temperature accuracy and repeatability will eventually degrade with these elements over time. To avoid these costly errors in critical temperature measurements, install 4-wire RTD sensors.

Recent developments in extension wire and protective cables have simplified certain applications. For example, measuring temperatures in boilers and furnaces previously required the use of long, rigid thermocouples, which were welded to tubes and surfaces. When these T/Cs had to be replaced, it required shutting down the boiler, going inside, removing the old welded T/Cs, welding in new T/Cs, and replacing the T/C extension wire all the way to the termination panels outside the boiler.

Today, T/C sensor wire is available for use up to 1,600°F. This makes it possible to use flexible sensors inside protection tubes, so the sensor can be replaced from outside the boiler or furnace without shutting down the process. The protection tube is welded to the boiler tubes and routed to the outside. A flexible sensor slides down the protection tube all the way to the sensing point, and is held in place with a spring.

Thermowells, protection tubes, and pipe wells

After you have selected the proper type of sensor for your application, you should next consider the best type of protection for it. This would be a thermowell, protection tube or a ‘pipe well’.

Thermowells are generally used when process or operating conditions are extreme. Tanks and pipes almost always use thermowells, as it is very inconvenient and often costly to empty a tank or stop process flow in a pipe that needs a replacement sensor. Protection tubes can be used when the conditions are not as severe.

A ‘pipe thermowell’ is a concept where users can build their own well from Schedule 40 pipe. The cost is about 1/10th of building and shipping a long 10ft, 40ft or even a 100ft foot stainless steel thermowell.

The key to the pipe thermowell is the WORM NOSE, which attaches to any standard length of steel pipe with 1/2-inch NPT threads. The pipe is threaded into a flange or plate with a 1/2-inch NPT connection. Our WORM sensor slides down into the pipe, resting inside the WORM NOSE. Because the WORM sensor can be built to any length, a ‘pipe well’ can be as long as needed.

Sensing temperature in an industrial application does not have to be complex or expensive. As we’ve seen, RTDs and T/Cs are able to provide the necessary accuracy, and transmitters convert the sensor’s signal to a more robust 4-20mA output signal used by most control systems. Flexible sensors allow you to fit a sensor in places you may not have thought possible. Before committing yourself to a traditional solution, check with your temperature sensor system provider and see what the alternatives are.

Mick Carolan is an employee of Moore Industries-Australia.