Sensor selection 101: Optimal temperature sensor selection: first step to accurate measurement

By Control Engineering Staff November 17, 2005

Achieving accurate temperature measurement is not simple. It requires knowing the inherent accuracy of particular sensor types, environmental factors that can create measurement uncertainty, and sensor calibration techniques available to reduce that uncertainty.

Thermocouples are the smallest, fastest, most durable temperature measurement solution. They withstand high temperatures and mechanical punishment and are simple to operate. They offer rapid temperature response times. The sensing junction can be placed close to the desired measurement point. Durability and simplicity make them suitable for embedding into other devices.

However, thermocouples—Type E and T, in particular—are at risk for accuracy, noise, and precision errors. When extreme accuracy and precision are needed, many of these shortcomings can be compensated for by using short runs of insulated and shielded thermocouple wires with balanced, low-pass filtered differential amplifiers to avoid common-mode voltage offsets—and by using some complex calibration procedures.

Lack of alloy homogeneity presents additional challenges. Deviations in metal purity and alloy homogeneity cause thermocouple profiles that deviate from National Institute of Standards and Technology (NIST) standards; this is a particular problem when long runs are required. When high accuracy is required without calibration, a thermocouple type that consists of a minimum number of elements, such as a type T, J, or G should be used.

Thermistors are suited for measurements that require high accuracy over a relatively narrow temperature range, typically less than 300 °C. However, they cannot endure high temperatures or mechanical stresses the way thermocouples do. Therefore, it is difficult to use them where these factors are not well controlled. Encasing the sensor in a protective metal enclosure can minimize this limitation, but this modification will reduce thermal responsiveness.

Although thermistors are less subject to error, local signal conditioning is still recommended (though it is much simpler than that required for thermocouples). Because they are physically larger than thermocouples, thermistors tend to have slower response times and may be subject to location and heat transfer errors more than similarly placed thermocouples.

When used near their maximum sensitivity point, small changes in temperature produce relatively high changes in resistance. Away from that point, they are less able to resolve temperature changes. A more linear response can be obtained by padding resistors through a voltage divider circuit.

RTDs and thermistors.

Thermistors are made uniformly in batches. However, batch-to-batch variations can create problems when high-precision accuracy is required. (No NIST standards exist for thermistors.) Under these conditions, special thermistors capable of working to temperatures of 1,000 °C should be used. Strategies for minimizing this effect include forming arrays from a single manufacturing batch.

RTDs are used when extremely stable and precise measurements are required or when accuracy over time is the most important factor. (Accuracy and precision of an RTD often exceeds a thermistor or a thermocouple.) RTDs follow DIN or Joint Information Systems Committee (JISC) standards and have good tolerance specifications. Off-the-shelf units are consistent, regardless of batch number.

RTDs are delicate. Melting temperature of the element is high enough to survive many high-temperature manufacturing operations, but not aggressive mechanical operations such as compaction. Therefore, they are difficult to embed into custom mechanical devices. Metal-sheathed assemblies can remove the fragility, but at the same time reduces response time. Their larger size also typically results in slow response times than comparable thermocouples.

Wire and termination resistance associated with long lead lengths and multiple connections can become a significant source of error in a typical 100 ohm RTD. Three- or four-wire RTDs are often used to achieve higher accuracy. The electronics can be constructed to dynamically remove error associated with lead resistance, but at a cost and the number of wires required to perform the measurement.

External noise can cause additional measurement problems, which can be mitigated much in the same way as with thermocouples using differential, ungrounded, and shielded elements. These effects can also be limited through optional electronics that perform 10% duty cycle measurements to limit self-heating power without reducing signal strength. However, using low-level signals (power) to drive an RTD may require even more measures to minimize effect of external noise.

Sensor considerations
When building a sensor knowledge base, a user should consider inherent accuracy in terms of durability, range of operation, and susceptibility to external noise influences. How the sensor will be used should be examined in terms of temperature range, level of accuracy and repeatability, handling and installation endurance, calibration or grounding, and environment.

—Cal Swanson, senior principal engineer,
Single Iteration , a division of Watlow Electric Manufacturing Co .
Contact Cal at