A relatively simple solution for obtaining accurate temperature measurement embraces three factors: The inherent accuracy of a particular sensor type, the environmental factors that can affect the sensor’s accuracy, and sensor calibration techniques.

However, thermocouples are the type most vulnerable to noise, inaccurate readings, and precision error. When extreme accuracy and precision are required, many of these shortcomings can be compensated with short, insulated, shielded thermocouple lead wires; balanced, low-pass, filtered differential amplifiers; and complex calibration procedures.

Also, differences in metal purity and alloy homogeneity can make thermocouple temperature profiles deviate from the National Institute of Standards and Technology’s (NIST’s) standards. This can be a problem when long lead wires are needed between the sensing element and the signal conditioners. A thermocouple containing a minimum number of elements, such as a type T, J, or G should be used when extreme accuracy is required, but a means for calibration it is not available.

Thermistors
Thermistors are ideal for measurements that require high accuracy and sensitivity over a narrow range of temperatures (typically less than 300º C), but they cannot endure the high temperatures or mechanical stresses that thermocouples can. This makes them unsuitable for applications and assembly operations where these factors are not well controlled. To compensate, the sensors can be encased in a protective metal enclosure, but this slows their response to temperature changes. Because some thermistors can handle temperatures to 1000º C, their slower response is traded off for their higher temperature capability.

Although thermistors are less subject to different types of errors, local signal conditioning is still recommended. Because thermistors tend to be larger than thermocouples, they are slower to respond to temperature changes and can have larger heat transfer errors than thermocouples.

Thermistors produce relatively high but non-linear changes in resistance near their maximum sensitivity point with small changes in temperature. However, padding resistors can be added in a voltage divider circuit to obtain a more linear response. Thermistors can be made fairly uniform within a batch, but batch-to-batch variations can be a problem when high accuracy is needed. Also, NIST does not publish standards for thermistors, so variations may be seen among different manufacturer’s devices.

RTDs
RTDs are extremely stable, make precise measurements, and maintain high accuracy over long periods. Often, they are more accurate and precise than thermistors and thermocouples. RTDs conform to Deutsche Industrie Normen (DIN) and Joint Information Systems Committee (JISC) national standards. Consequently, off-the-shelf RTDs maintain consistent tolerance specifications regardless of batch number.

Compared to thermistors and thermocouples, RTDs are more fragile, and although the melting temperature of an RTD element is high enough to survive many high-temperature manufacturing operations, they do not survive aggressive mechanical compaction processes. For this reason, they are difficult to embed into custom mechanical devices. Metal-sheathed assemblies can help overcome the fragility factor, but this, coupled with their larger size, often makes their response time slower than  thermocouples.

For a typical 100 ohm RTD, wire and termination resistance from long lead lengths and multiple connections can become a significant source of error. Three or four wire RTDs are often used to obtain the highest accuracy. Sensor conditioning electronics can remove errors from lead resistance, but there is usually a tradeoff between cost and number of wires needed for the measurement.

RTDs suffer measurement errors from self-heating, but this can be reduced considerably by applying an extremely low bias or a 10% duty cycle instead of a constant bias. Although this approach helps alleviate one problem, it generates another: noise that can affect the RTD measurements. However, the noise from this and other external sources can often be minimized by using differential, ungrounded, and shielded elements.

General measurement errors
Temperature measurement errors arise from several sources, which include location, transients, and heat transfer. Often, it is difficult to sense the temperature at the exact point where it is needed because the sensor has a finite size or is surrounded by an encasement that displaces the sensing element. Then the sensor’s actual temperature will be different from that at the intended measurement point. However, such location errors can be compensated when surrounding heat sources and sinks are known. A simple solution, which avoids complex calibration techniques, uses the smallest sensor possible, placed as close to the temperature source as possible. Thermistors and RTDs have larger errors than thermocouples located in the same place, due to their larger size.

Sensors are also susceptible to heat transfer errors and measurement inaccuracies from conduction, convection, and radiation inputs. Higher and lower ambient temperatures can move heat into and out of the sensor through their thermally conductive lead wires. Type E and J thermocouples are alloys with less conductive leads, which make them most suitable for minimizing these kinds of errors. Self-heating errors apply to thermistors as well as RTDs. The same compensation strategies used for thermistors can be used for RTDs.

The lead-wire corrosion problem is reduced for RTDs by using three or four-wire units that measure the resistance of the sensing element instead of the connection wire. This gives RTDs the greatest overall stability of the three sensor types. Thermistors usually drift a little initially, but they generally stabilize after some aging. Thermocouples behave in a more complex way, because the voltages they produce come from their different metals and alloys, which change as the metals age and deteriorate.

When measuring temperature on a surface, forced airflow on and around a sensor can contribute to a false reading because of heat-transfer error. Convective currents add or remove heat from the sensor and measurement surface. When the atmosphere and surface temperatures differ, or the measurement environment is moist, convective heat flow must be considered as if it were another heat source or sink.

Thermocouples and RTDs have the highest noise sensitivity of the three sensor types. Shields and proper grounds can reduce noise susceptibility and offset currents due to capacitive coupling and radio frequency interference, but magnetic sources are a little more difficult to shield.

Sensors often operate in the same environment with motors, solenoids, or high-current producing devices that generate high transient currents and magnetic surges. Signal conditioners for thermistors and RTDs should contain power supplies that are immune to these droops and surges when operating from the same power mains. Otherwise, the signal conditioners’ power supply changes could adversely affect the sensors’ temperature readings. Also, large inductive spikes can generate circulating currents that alter ground potentials near the sensors. These spikes can bias the voltage output of the sensor and produce false readings.When thermistors measure temperatures near their lower extremes, their resistance may approach 100k or more. With such high resistance, long thermistor lead wires can act as antennas and add noise to the measurement system. Most of this can be filtered out, but a small dc charge, called the electret effect, can bias and affect the measurement. The best method to protect sensors from outside electrical and magnetic sources is to keep the sensor and lead wires as far away as possible from the interfering source, shield them, and use appropriate electronics isolation and grounding. Also, keep sensor lead wires short and convert the analog signals into a digital format as close to the measurement point as possible.

Sensor calibration techniques
A common method for correcting inherent accuracy errors is to calibrate the sensor in a controlled isothermal liquid bath and compare temperature readings to a standard reference. A point calibration method immerses the sensors in an ice bath or other standard freezing point medium such as a gallium-freeze bath at 29.7647° C.

The limiting factor for minimizing inherent sensor error is the uncertainty (including both the accuracy and precision) of the calibration process. Generally, thermistors and RTDs have better inherent accuracy than thermocouples, but all three types require calibration to obtain accuracies down to 0.1° C. Calibrating thermocouples is more difficult than thermistors and RTDs because the procedure must consider both hot and cold junction temperature errors. DW

For more information, see www.watlow.com