Getting temperatures right

Cal Swanson
Senior Principle Engineer
Single Iteration Div.
Watlow Electric Mfg. Co.
St. Louis, Mo.

There is more to getting accurate temperature measurements then just attaching a thermometer at the location of interest. Technicians must know not only the inherent accuracy that particular sensors will provide, but also how factors in the environment create further measurement uncertainty. In addition, it's important to know the sensor-calibration techniques available to reduce this uncertainty.

Thermocouples are the smallest, fastest, and most-durable temperature sensors. They can withstand very high temperatures, harsh mechanical punishment, and are simple to operate. The size of their sensing junction promotes rapid response to temperature along with close placement to the desired point of measurement. Their durability and simplicity makes them good candidates for embedding into other devices.

Thermocouples are most at risk from accuracy, noise, and precision error. However, proper compensation for these shortcomings can produce extreme accuracy and precision. Typically, short runs of insulated and shielded thermocouple wires are combined with balanced, lowpass-filtered differential amplifiers to avoid common-mode voltage offsets. But these techniques require relatively complex calibration procedures.

Thermocouples are classified by type codes that identify the materials making up the thermocouples. For example, a Type T thermocouple uses copper and a copper-nickel alloy. The National Institute of Standards and Technology (NIST) publishes a standard temperature profile for each thermo-couple type.

Variations in metal purity and poor alloy homogeneity can shift a thermocouple profile away from those standards. The problem becomes worse when there are long lead runs. High accuracy without extensive calibration requires thermocouples made with a minimum number of elements — such as a Type T, J, or G.

Thermistors perform best in applications that need high-accuracy over a relatively narrow temperature range. Typically, they work at less than 300°C, though some special versions are capable of handling over 1,000°C. Normally, thermistors cannot endure either the high temperatures or mechanical stresses that thermocouples can handle. This makes them difficult to use in applications where these factors are not well controlled. Encasing the sensor in a protective metal enclosure helps overcome some of the limitations; but the enclosure slows the sensor's thermal response.

Because thermistors tend to be larger than thermocouples, they respond more slowly to temperature changes. They may also be subject to additional location and heat transfer errors compared to thermocouples in the same spot. Local signal conditioning is still recommended for thermistors, though it is simpler than that required for thermocouples.

The thermistor response to temperature is decidedly nonlinear. Small changes in temperature create relatively high changes in resistance near the point of maximum sensitivity. But resolution drops quickly as the temperature moves away from the point. The addition of padding resistors in a voltage-divider-type circuit helps make the response more linear.

Thermistors can be made in relatively uniform batches, but batch-to-batch variations can degrade precision or accuracy. There are no NIST standards for thermistors, so there may be response variations between similar products from different manufacturers.

Resistive-temperature devices, or RTDs, give stable and precise measurements, and are candidates when accuracy over a prolonged time is important. Their accuracy and precision often exceeds that of both thermistors and thermocouples. RTDs typically follow the Deutsche Industrie Normen (DIN) or Joint Information Systems Committee (JISC) national standards. So off-the-shelf RTDs have consistent temperature response curves regardless of their batch number.

The larger physical size of an RTD means they respond more slowly than thermocouples. Though the melting temperature of an RTD is high, RTDs are delicate compared to thermocouples and thermistors. They do not handle aggressive physical applications well. This makes it difficult to embed RTDs in custom mechanical devices. Metal-sheathed assemblies help improve their robustness, but again at the cost of even slower response times.

Wire and termination resistance associated with long lead lengths and multiple connections for a typical 100- RTD becomes a significant source of error. Often three or four-wire RTDs are used to get the highest accuracy. Electronics can dynamically remove errors associated with lead resistance, but bring a trade-off in cost and number of wires needed to take the measurement.

Electrical noise generated by external sources creates additional measurement difficulties. The problem is made worse when low power levels used to minimize self-heating errors produce low-level signals. The use of differential, ungrounded, and shielded elements helps mitigate noise problems as they do with thermocouples.

One way to reduce self-heating errors while maintaining high signal levels is to cycle power to the RTD. Special electronic circuits apply full power and take readings only 10% of the time. The reading stays in electronic memory while power to the RTD is off. The 10% duty cycle limits RTD self-heating without reducing signal strength.

Any knowledge base on sensor types should consider inherent accuracy in terms of durability, range of operation, and susceptibility to external noise. It should also include sensor use in terms of temperature range, the required level of accuracy and repeatability, handling and installation ruggedness, whether calibrated or grounded, and the type of environment it's used in.

Awareness of sensor accuracy isn't the only knowledge required to assure correct temperature measurements. A broad knowledge of how sensor choice, placement, and environmental factors contribute to reading errors is also needed. In addition, a familiarity with calibration techniques used to reduce and compensate errors also bolsters temperature accuracy.

It is nearly impossible to sense temperature exactly where you need to sense it. At the very least, the finite size of the sensor displaces the sensing element from its attachment. The result is that the sensor ends up sitting some distance away from the desired measurement location. Thermistors and RTDs are at greater risk for location errors simply because they're bigger than an equivalent thermocouple.

Sensor location affects temperature measurement accuracy. Knowledge of the surrounding heat sources and sinks lets technicians allow for location errors. However, it's often tough to get this information so errors persist. The best advice is to just use a small sensor placed as close as possible to the temperature source.

Dynamic thermal errors are transient errors that are typically difficult to compensate. This is because every material within the thermal system has its own special thermal conductivity and capacity. Of the three most widely used sensor types, it is typically the thermocouple that best minimizes transient errors mostly because it is the smallest with the shortest time constant.

Sensors receive conductive, convective, and radiative inputs that contribute to measurement inaccuracy. These types of errors come from ambient conditions that heat up or cool down the sensor — often along specific pathways. For example, thermally conductive electrical wire leads on a sensor may pick up heat from a nearby heating element and conduct it to the sensing device. Type E and J thermocouples use alloy leads that are less thermally conductive. That makes them good candidates for reducing this kind of error.

The third form of measurement error applies to thermistors and RTDs. This error comes from heat dissipating inside the sensing element. The temperature inside the sensor rises so it becomes hotter than the environment. As mentioned earlier, strategies for minimizing this error include keeping sensor current low or pulsing sensor power with a low duty cycle to keep average power dissipation low.

Operating or cycling all three sensor types near their temperature limits speeds deterioration that produces a drift from the initial profile. Thermistors and RTDs are usually well sealed from the environment and thus are less susceptible to internal corrosion. However, these sensors are usually connected to copper wires that are subject to deterioration.

Three or four-wire RTDs can effectively correct errors created by lead-wire corrosion. They do this by measuring the resistance of the sensing element only at the element versus the combined resistance of the element and connection wire. This helps give RTDs the greatest overall stability of the three sensor types. Thermistors usually exhibit some initial drift, but are generally stable after initial aging. Thermocouples exhibit more complex behavior because the voltages produced are a direct result of the dissimilar metals used in their construction. The output voltage curve shifts as the metal ages and deteriorates and the alloy formulation changes.

Forced airflow on and around a sensor measuring the temperature of a surface contributes to false readings because of heat-transfer error. This is because convective currents add or remove heat from the sensor or surface. If the atmosphere is at a different temperature than the surface, or the measurement environment is moist, you must treat heat flow associated with convection as if it were another heat source or sink.

Applications subject to extreme mechanical motion, vibration, or high intensity acoustics should avoid sensors that are fragile or that have small wire gage leads. The most common wire failures are near connection points where there is the greatest amount of flexure. Mechanical motion or vibration can also stimulate internal resonance inside the sensor leading to early failure. Thermocouples are generally the most durable of the three sensor types because many of the alloys in the wires are more ductile. This lets thermocouples handle additional motion and vibration better than the others.

Besides fatigue, cables in motion can also generate low-voltage triboelectric effects. For microvolt sensors, such as thermocouples or RTDs, these effects contribute to measurement uncertainty. This is particularly true if the motion stimulating the effect is of the same order as the thermal response to be measured.

Thermocouples and RTDs generally have the lowest noise immunity of the three sensor types. Shielding and grounding these sensors improves their immunity from potential noise sources caused by capacitive, radio frequency (RF), and electrical current offsets. However, it is tough to make them immune from magnetic sources.

The environment in which sensors operate can often contain large motors, solenoids, and other high-current devices that create transient currents or magnetic surges. Sensor types such as thermistors and RTDs that need external electronics may give bad measurements if there are power brownouts. In addition, large inductive spikes can create circulating currents that alter ground potentials near the sensors. This effect then biases the voltage read from the sensor to create false readings.

Thermistors used to measure temperatures near their lower limits may approach resistances of 100 k or more. The high resistance coupled with long runs of thermistor wire form an antenna that picks up noise. While it is possible to filter out most of the noise, the potential for biasing the measurement becomes greater because a dc charge called an electret effect can collect on the wire.

Protect sensors from outside sources of electrical and magnetic interference. Keep the sensor and lead wires away from interference sources; shield the wires and sensors; and pay close attention to isolation and grounding of the electronics. It's best to minimize noise by keeping sensor lead wires short and converting the signals into digital form as close as possible to the point of measurement.

Inherent accuracy errors are commonly corrected by calibrating the sensor in a controlled isothermal liquid bath. The temperature readings from the bath then get compared against a standard reference. An alternative way to characterize accuracy is point calibration. It immerses the sensors in an ice bath or other standardized freeze point to provide a point of calibration. The other temperature readings are extrapolated from that calibration point. However, point calibration only works if the assumptions hold up that are used to calculate the other temperature values from the calibration point.

If only relative accuracy is important, an array of sensors can be calibrated to each other by immersing them in a common bath at a known temperature, such as a 0°C ice bath. The temperature in the bath can then be slowly raised, while all sensor responses are tracked. For best results, the calibration bath should span the same temperature range as the intended measurement. Additionally, the rate of temperature increase should be slow, relative to sensor responsiveness, to reduce time-transient errors.

The limiting factor for minimizing inherent sensor error is the un-certainty of the calibration process for both accuracy and precision. Generally, thermistors and RTDs have better inherent accuracy than thermocouples; but all three types of sensors require calibration to get accuracies down to 0.1°C. However, it is more challenging to calibrate thermocouples than thermistors and RTDs because calibration must consider both hot and cold-junction temperature errors.

MAKE CONTACT
Watlow Electric Mfg. Co., (800) 492-8569, watlow.com