Applications involving heating invariably require temperature measurement instrumentation. Temperature instruments typically consist of a thermowell, a sensor and a transmitter to convert the sensor data to an industry-standard signal such as 4 to 20 mA or fieldbus digital communications. Temperature sensors come in many shapes and sizes, but for industrial use, two technologies cover the majority of applications: thermocouples (TCs) and resistance temperature detectors (RTDs).

It is not difficult to find resources discussing the differences between the two technologies. But, for purposes of this article, which is part one of a two-part series, we will focus on some of the more nuts-and-bolts aspects of the selection process. In addition, this article will describe ways to improve temperature-measurement performance regardless of the selected sensor technology.

Thermocouples vs. RTD: Basic Considerations

To begin, let’s briefly review the basic differences between the two sensor technologies. Figure 1 shows the applicable temperature ranges for several types of thermocouples and a typical RTD.

A thermocouple is a closed thermoelectric circuit consisting of two wires of dissimilar metals joined at both ends. A voltage is created when the temperature at one end or junction differs from the temperature at the other end. This phenomenon is known as the Seebeck effect, and it is the basis for thermocouple temperature measurements. The voltage created is proportional to the temperature difference at the two junctions. This means a thermocouple cannot provide a temperature reading without an external temperature reference.

An RTD uses a characteristic found in many metals whereby the electrical resistance increases as temperature increases. This is a phenomenon known as thermal resistivity. A given metal will have a given resistance at a given temperature. If this is properly quantified, an RTD can provide an accurate temperature reading.

Thermocouples Remain Popular

Thermocouple technology is the oldest electronic temperature-measuring method, with the Seebeck effect recognized for almost 200 years. It remains the most common technology, but it is losing market share to RTDs. A thermocouple is much like a differential pressure reading: it can tell you the difference between two points but not the absolute value of either. One of the temperatures must be known to calculate the other, which is why the two junctions are called sensing and reference.

In a high-school physics class, the reference junction is usually ice water, which provides a specific temperature reference. In industrial applications, the transmitter will have its own temperature sensor, typically a thermistor, to provide a reference reading. The reference-junction sensor can be anywhere and at any temperature as long as it is measured accurately. Consequently, the reading accuracy from a thermocouple is no better than the reference-junction accuracy.

Thermocouples are made in types designated by a letter indicating the combination of wires. For example, a Type K thermocouple uses one wire made from chromel and the other from alumel, with each metal having different thermoelectric characteristics. Each thermocouple type has its own temperature range limitation (figure 1).

One of the biggest advantages of a thermocouple over an RTD is its ruggedness. Because the sensor consists of two wires, typically welded together, it is not delicate. So long as the junction is sound, the insulation is intact and the wire itself is not corroded, a thermocouple will work. The wire used is usually of a relatively heavy gauge, so a thermocouple can withstand a lot of punishment and vibration. Even so, naked-wire thermocouples are relatively rare. The sensing junction is normally inserted into a stainless steel sheath. This produces a variety of effects.

RTDs for Greater Accuracy

RTDs have a resistive element with leads attached. The elements and leads are packed in insulation and inserted into a protective sheath. The resistive material is usually platinum because of its high accuracy, excellent repeatability and exceptional linearity over a wide range. Platinum exhibits a large resistance change per degree of temperature change, making it a good choice for RTDs.

Whereas the signal from a thermocouple is a voltage, the signal from an RTD is resistance, measured in ohms. To overcome the effect of the lead-wire resistance unduly influencing the temperature reading, RTDs use wiring methods designed to compensate for the resistance of the lead wire.


FIGURE 1. Different thermocouple types have different and wide effective measurement ranges. RTDs (far right) have a much narrower range. Types R and S thermocouples are shown as a single column because they effectively cover the same range.

Given the cost of platinum, RTDs are designed to use as little of it as possible. The actual sensor element typically is fine wire wrapped around a ceramic mandrel or a thin-film deposition on a substrate. In either case, both are normally encased in insulation and inserted in a stainless steel sheath.

Accuracy, Response Time and Durability

The conventional wisdom says that for applications where a high degree of accuracy is needed, an RTD is the better choice for several reasons:

  • If the thermocouple reference-junction sensor changes, it will cause the reading to change.
  • Corrosion of the thermocouple wires can cause reading drift.
  • RTDs generally have a higher degree of reading stability.
  • RTDs have accuracy classes that include very specific tolerances.
  • Accuracy for a thermocouple depends upon the metallurgy of the wires. If the alloy is at one extreme of the specification, or if it changes due to prolonged exposure to heat, it can allow readings to drift. Over time, some degree of drift is inevitable.
  • RTDs can be matched to a specific transmitter. This allows for an exceptionally high degree of accuracy.

By contrast, thermocouples tend to win when response time and durability are paramount. Thermocouples also are better for high process temperatures. Other thermocouple advantages include:

  • Thermocouples do not need as much insulation within the sheath, so there is less material to slow heat flow.
  • Thermocouples can be welded directly to the inside of the sheath (provided the application allows for a grounded sensor), speeding up heat transfer.
  • For extreme applications, the thermocouple junction can be welded directly to the surface to be measured to provide the ultimate in fast heat transfer and response.
  • Thermocouples are more robust. RTDs are more delicate physically, so they can break more easily due to shock or vibration.
  • Specific thermocouple types can be used in applications from -328 to 2640°F (-200 to 1450°C).

Manufacturing Technologies Evolve

Given the simplicity of thermocouples, their construction and manufacturing techniques are mature, and there is not much in the way of new developments. RTDs, by contrast, are moving through a transition.

As mentioned previously, the traditional construction of RTDs has been the wire-wound version where very fine platinum wire is wound into a helix or around a ceramic form. These are built by hand, and their quality depends upon the skill of the assembler to be extremely precise with respect to assembly and lead mounting.

Also as mentioned earlier, there are thin-film RTDs where the platinum sensor element is deposited on a substrate like a tiny printed-circuit board. This process is automated using the same techniques as computer chips and integrated devices. While the quality of a hand-made RTD depends on the skill of the assembly person, thin-film sensors made by an automated process can be extremely precise and consistent. These types of RTDs also tend to be more shock and vibration resistant due to their smaller mass and support of the platinum element by the substrate.

Many users still regard wire-wound sensors as the gold standard and specify them. Given the advances in thin-film production technology, however, the only legitimate technical reason to call for a wire-wound sensor is for cryogenic applications.

temperature sensing

FIGURE 2.Thermocouple and RTD technologies are different in concept, so they bring different considerations when analyzing performance tradeoffs.

Highest Accuracy from an RTD

RTDs offer two methods for approaching applications where the highest accuracy is necessary. First, there are accuracy classes A and B (figure 2), which are verified using techniques outlined in IEC 60751. Naturally, these command premium prices justified by the tight manufacturing tolerances and precise calibration required.

Another approach is to fine-tune an RTD sensor to a specific transmitter (figure 3). In a laboratory environment, it is possible to put the sensor into a dry-block calibrator set at temperatures close to the expected operating range of the final installation. The transmitter can be adjusted to match the exact output of the sensor. So, if a reading at 329°F (165°C) is 33.79 Ω, the transmitter can be set manually to match those two values.

Obviously, this takes some time. It is normally necessary to set a series of temperature values above and below the most critical value. The result is a specific temperature curve for that sensor/transmitter combination, making it possible to provide the highest degree of accuracy within the critical operating range.

Some users wanting to make this type of adjustment make the process more expensive than necessary by combining this technique with a Class A sensor. If this hand-calibration process is being done, it can be applied to a conventional sensor because RTDs are generally stable. Once the curve is corrected, it will stay that way, so it is not necessary to purchase a high accuracy sensor if hand calibration and matching to a specific transmitter is planned.

RTD Sensor

FIGURE 3. By using a dry-block calibrator and Hart communicator, a transmitter and RTD sensor can be individually matched to create a highly precise measurement setup.

Improving Accuracy Overall

Accuracy of a given application depends on more than just the sensor. There is a chain of components between the sensor and the final input point, which can be on anything from a local transmitter display to a plant-wide control or monitoring system. Every termination, and even the cable itself, can cause reading drift by interfering with the voltage or resistance measurement. The transmitter or input card also will have its own error factor, so it is important to understand how these all add up. Suffice it to say, an expensive Class A RTD supported by lower-grade components will be no better than the worst link in the chain.

In the real world, applications requiring the highest degree of accuracy are relatively rare. More often than not, repeatability is as important — if not more so. For many processes, maintaining the same temperature value is more critical than knowing exactly what it is. In other words, stability may trump accuracy. So long as operators see the reading at the outlet of a reactor remaining stable, the process can operate safely even if the reading is consistently some number of degrees off. If other critical variables are what they should be, there is probably nothing to worry about.