JPY | USD

Reap the Benefits of RTD Temperature Sensors Without the Interface Complexity

By Bill Schweber

Contributed By Digi-Key's North American Editors

From healthcare, instrumentation, HVAC, and automotive applications to the Internet of Things (IoT), temperature is the most widely sensed real-world parameter, and knowing the temperature with the right balance of accuracy, precision, and repeatability is critical for many applications. A widely used choice for a temperature sensor is the resistance temperature detector (RTD), a precision metal element usually made of pure or nearly pure platinum. A platinum-based sensor has a fully detailed, repeatable, and characterized resistance-versus-temperature transfer function, so RTDs are widely used in scientific and instrumentation applications.

However, to fully realize the performance potential of this seemingly simple two-terminal sensor, the designer must understand the various ways of driving it and measuring its resistance to determine temperature. Further, many applications require multiple RTDs, so the interface approach and associated circuitry must also match the application.

What designers need are RTD-specific components that address and overcome the RTD’s inherent idiosyncrasies. This article shows how ICs from Texas Instruments, Maxim Integrated, and Analog Devices, along with an evaluation board from Microchip Technology can be used to simplify their application.

How RTD sensors work

Somewhat similar to the thermistor, the operating principle of the RTD is deceptively simple. It is a platinum wire or thin film, sometimes with other precious metals added such as rhodium, with a known nominal resistance and a positive change in resistance as a function of temperature (i.e., positive temperature coefficient or PTC). RTDs can be fabricated with many different nominal resistance values, with the most common being the Pt100 and Pt1000 (sometimes written as PT100 and PT1000) with a nominal resistance of 100 ohms (Ω) and 1000 Ω, respectively, at 0⁰C.

Common ways of constructing the sensor include winding the platinum wire around a glass or ceramic support, or using platinum in a thin-film fabrication (Figure 1). Due to their widespread use and need for interchangeability, an international standard, DIN EN 60751 (2008), defines the detailed electrical characteristics of platinum temperature sensors. The standard contains tables of resistance versus temperature, tolerances, curves, and temperature ranges.

Image of RTDs use (left to right) thin-film, glass, and ceramic fabrication techniquesFigure 1: These RTDs use (left to right) thin-film, glass, and ceramic fabrication techniques. (Image source: WIKA Alexander Wiegand SE & Co. KG)

Standard platinum RTDs operate over the range of -200⁰C to +800⁰C. Their key attributes include high stability, repeatability, and accuracy, provided they are properly driven by a current or voltage source, and their resistance is measured as a voltage across their two terminals using a suitable analog front-end (AFE) circuit, with the voltage readings linearized for highest accuracy.

The RTD’s resistance changes fairly dramatically with temperature, which adds to their suitability for high-precision measurement. For a standard Pt100 device, the resistance changes from about 25 Ω at -200⁰C to about +375 Ω at +800⁰C. The average slope between 0°C and +100°C is called alpha (α), or temperature coefficient, and its value depends on the impurities and their concentrations in the platinum. The two most widely used values for alpha are 0.00385055 and 0.00392.

RTDs are offered in thousands of specific models from many sources. An example is the Vishay Beyschlag PTS060301B100RP100, a 100 Ω platinum RTD with ±0.3% basic accuracy and a ±3850 ppm/°C temperature coefficient in an 0603 SMT package. It is one member of the PTS Series of 100 Ω, 500 Ω, and 1000 Ω leadless SMT RTDs that come in 0603, 0805, and 1206 packages, respectively. These devices are fabricated using a homogenous film of platinum deposited on a high-grade ceramic substrate and are conditioned to achieve the correct temperature coefficient and stability. The sensor elements are covered by a protective coating designed for electrical, mechanical and climatic protection, and meet all relevant IEC and DIN standards for performance and compliance. Due to its small size, the 100 Ω device in the 0603 package features a fast response time in free air of under two seconds to within 90% of its final resistance value.

RTD linearization

RTDs are fairly linear but still have a curved, monotonic deviation. For applications needing accuracy to one degree or a few degrees, it may not be necessary to linearize the RTD transfer function since the deviation is fairly small (Figure 2). For example, between -20⁰C and +120⁰C, the difference is less than ±0.4⁰C.

Graph of Maxim Pt100 RTD resistance vs. temperatureFigure 2: Pt100 RTD resistance vs. temperature, shown with the straight-line approximation for 0°C to +100°C. (Image source: Maxim Integrated)

However, the RTD is often used in precision applications needing accuracy to a tenth or better of a degree and so linearization is needed. Linearization can be implemented by computation in software or by a look-up table. For highly accurate linearization, the Callendar-Van Dusen equation is used:

Equation 1

where T = temperature (°C); R(T) = resistance at T; R0 = resistance at T = 0°C; and A, B, and C are RTD specific constants.

For α = 0.00385055, the DIN RTD standard defines the Callendar-Van Dusen coefficient values A, B, and C as:

A = 3.90830 x 10-3,

B = -5.77500 x 10-7, and

C = -4.18301 x 10-12 from -200°C to 0°C, and C = 0 from 0°C to +850°C (this has the benefit of reducing the polynomial to a simpler, second-order equation.)

RTD connections

As a passive, two-terminal resistor, the RTD interface drive and sensing circuits are simple in principle, and the drive can be a voltage or current source. In the most basic form with a voltage source, the RTD leads are connected to the source, as is a stable known resistor (RREF) placed in series that usually has the same nominal value as the RTD (Figure 3). This forms a standard voltage divider circuit. The voltage across both the RTD and series resistor are measured, and simple voltage divider calculations are then used to calculate the RTD resistance. Accuracy can be improved by measuring the voltage across the known resistor, along with the voltage across the RTD.

Diagram of Maxim simplified RTD signal conditioning circuitFigure 3: This simplified RTD signal conditioning circuit uses the RTD in series with a known reference resistor (RREF) and a current source; the voltage across the RTD is measured along with the voltage across the reference resistor to calculate the RTD resistance. (Image source: Maxim Integrated)

While simple, this arrangement has many sources of potential inaccuracy, including changes in the source voltage, reference-resistor temperature coefficient, connection-lead current-resistance (IR) drop, and even the temperature coefficient of the copper connection leads, which is about +0.4%/˚C. To partially overcome these error sources, the RTD is often instead used in a ratiometric Wheatstone bridge configuration.

However, the bridge and voltage drive approach still has weaknesses. A ratiometric arrangement such as the bridge has a well-known nonlinear relationship of its own, independent of the nonlinearity of any bridge element. Therefore, this relationship must be factored into the calculations that correct for the nonlinearity of the RTD element, which complicates the algorithm and adds to the processing load.

For these and other reasons, the RTD is almost always used with a current source. This allows full control over the drive situation and provides opportunities to more directly compensate for voltage drop and temperature related changes in the connection leads. Depending on the application and distance between the RTD and its AFE, designers can use two, three, four, or four-wire with loop connections (Figure 4).

Diagram of interconnection between the RTD and the AFE can use two, three, or four wiresFigure 4: The interconnection between the RTD and the AFE can use two, three, or four wires; the latter can be a paired four-wire connection or have a separate loop for two wires. (Image source: Texas Instruments)

The two-wire connection is the simplest, least bulky, and least costly. However, it is suitable for accurate results only when the wires connecting the Pt100 RTD to the AFE circuit have very low resistance of under a few milliohms (mΩ), where the wire resistance does not become significant compared to the RTD resistance. Typically, this restricts the distance to about 25 centimeters (cm) but is also a function the gauge of those wires, which tend to be thin due to the physical installation configuration and constraints. It is possible, of course, to correct for the voltage drop using calculations. However, this adds to complexity, especially if the lead wire resistance is affected by temperature.

For longer distances up to about 30 meters (m), the three-wire approach is used. Here, the circuit monitors one side of the current loop with a Kelvin connection, measuring voltage drop in the resistance of the loop and then compensating for that drop. This method assumes the voltage drop in the non-Kelvin lead is the same as in the Kelvin lead side.

The four-wire approach uses full Kelvin sensing to monitor both sides of the current loop of the RTD. This approach offers precision in eliminating the effect of lead resistance, regardless of differences between the two current source wires. It can be used for distances of hundreds of meters but has the highest material and wire bulk impact.

Finally, the four-wire with loop approach gives the designer choices in how to measure the loss in the loop. The resistance of the loop connection wires can be measured as a simple resistance independently of the actual RTD loop, assuming the two extra leads are identical to the RTD leads. This approach may seem to be more of a headache than the direct Kelvin arrangement in terms of installation and calculations, but there are practical cases where it is physically difficult to provide regular Kelvin connections at the RTD. Nonetheless, this arrangement is not often used in modern installations because the four-wire and even three-wire approach can provide comparable results with appropriate set-up and calibration.

Note that the choice of using a two, three, or four-wire interface is independent of the RTD, and any RTD can be used with any of the choices provided there is space and access to make the needed physical connections. However, in physically small set-ups, the mass of the wire bundle may introduce thermal shifts and additional thermal time constants. In general, it’s good practice to keep the thermal mass of the sensing arrangement as small as possible relative to the mass being sensed.

Issues related to the connection leads and signal integrity go beyond just basic DC resistance. Noise is often a concern, and even though temperature is a relatively slow changing phenomenon compared to most noise signals, noise can still corrupt the signal at the AFE if it occurs just as the voltage across the RTD is being sampled or converted. In extreme cases, noise can saturate the front-end and “blind it” for a few milliseconds (ms) until it comes out of saturation.

For this and other reasons, the sense leads from the RTD should be balanced (sometimes called longitudinal balance) with equal impedance to ground if their length is greater than about one meter. The reason is that these parallel leads will likely have a common-mode voltage (CMV) and noise, but the differential front-end of the AFE can reject these. However, if the leads are unbalanced, the circuit will convert some of the common-mode signal to an unbalanced signal, which will not be rejected by the differential input of the AFE.

Pt100 vs. Pt1000 RTD choice

Since the most common RTDs are available with either 100 Ω or 1000 Ω resistance at 0⁰C, the obvious question is how to choose between them. As always, there are trade-offs and no single “right” answer, as it depends on the application specifics. Note that the linearity of the characteristic curve, the operating temperature range, and the response time are the same, or nearly so, for both Pt100 and Pt1000 RTDs, and their temperature coefficient of resistance is also the same.

The Pt100 RTD has lower nominal resistance, and therefore as noted earlier, it can only be used for short distances in a two-wire configuration as lead resistance will be significant compared to the RTD. In contrast, the lead resistance is a much smaller fraction relative to the Pt1000 resistance, making the Pt1000 better suited to longer two-wire runs.

Since the Pt1000 RTD has higher resistance, per Ohm’s Law (V = IR), it requires less drive current to develop a given voltage across it. A modest 1 milliampere (mA) current will yield a 1 volt drop at 0⁰C, and the voltage increases from that value as temperature increases.

However, there is a potential undesired consequence of higher voltages, as the RTD voltage may overrange the AFE front-end at higher temperatures. Also, the current source needs to have sufficient compliance to drive the fixed current value through the resistance. For example, 1 mA through 1000 Ω requires a current source compliance of a little above 1 volt, but as the RTD heats up and its resistance increases, the compliance needed increases proportionally. Thus, a high resistance RTD current source may require higher voltage rails to ensure adequate compliance voltage.

The lower current needed by the Pt1000 for a given voltage drop brings two benefits. First, less power is needed, which increases battery life. Second, self-heating of the RTD is reduced, which can have a major effect on the accuracy of the reading. Proper engineering practice is to use a current drive level that minimizes sensor self-heating, consistent with developing sufficient voltage drop and thus, resolution across the RTD.

This does not mean that there is little place for the Pt100 RTD. In fact, it is widely used in industry due to legacy reasons, and where lead length, low-power operation, and self-heating are not major factors. As low-impedance loops, Pt100 RTD installations are also much less sensitive to noise pickup compared to those with the Pt1000 RTD, which inherently has a loop impedance that is ten times higher.

There are also mechanical considerations in addition to the electric ones. Pt100 sensors are available as both wire-wound and thin-film constructions with different physical attributes, while Pt1000 RTDs are generally offered only as thin-film devices.

Note that for higher accuracy applications, other steps may be needed to minimize RTD self-heating error. One way to do this is to pulse the current through the RTD and then measure the voltage during the pulse period. The shorter the duty cycle of the pulse, the lower the self-heating error. However, this approach also requires a somewhat more sophisticated interface to properly manage the pulse timing and duty cycle, and synchronize the voltage reading with the pulses.

ICs simplify the RTD interface

As with their other resistor-based temperature sensing components, the RTD looks simple and its use should be, too. After all, it is a two-terminal resistor with no parasitics of consequence in the relatively slow-moving world of temperature sensing. Nonetheless, as with thermistors and many other basic sensors, we have seen that users of this transducer have a host of issues to consider, including drive, linearization, calibration, lead compensation, and more; the situation’s complexity increases when more than one RTD is used, as is often the case.

To address the issues associated with RTD interfacing, IC vendors have developed application-specific ICs that ease connection on both the analog RTD-facing side of the front-end as well as the conditioned output, even going so far as to include a complete, processor-compatible digital interface. For example, for basic interfacing to the RTD, the Texas Instruments OPA317IDBVT operational amplifier uses a proprietary autocalibration technique to simultaneously provide low offset voltage (20 microvolts (μV) typical, 90 μV maximum) and near-zero drift over time and temperature, and near-zero bias current. As a result, the op amp does not “load” or affect the RTD but is both “invisible” and consistent. The op amp operates from single-ended or bipolar supplies ranging from 1.8 volts (±0.9 volts) up to 5.5 volts (±2.75 volts), and its 35 μA (maximum) quiescent current make it a good fit for battery-powered applications.

One of the characteristics of this op amp is that it can be configured to work on signals that are very close to ground, as is the case for a “cold” RTD operating at a low current level and thus, with a low voltage across it. In contrast, many single-supply op amps have problems when the input and output signals approach 0 volts, near the lower output swing limit of a single-supply op amp. While a good single-supply op amp may swing close to single-supply ground, it may not actually reach ground. The output of the OPA317IDBVT can be made to swing to ground, or slightly below, on a single-supply power source by adding another resistor and an additional, more-negative power supply than the op amp’s negative supply (Figure 5). Adding a pull-down resistor between the output and the additional negative supply allows it to take the output down below the value that the output would otherwise achieve.

Diagram of Texas Instruments OPA317IDBVT can handle signals that are close to ground potentialFigure 5: By adding a pull-down resistor (RP) and an additional negative supply, the OPA317IDBVT can handle signals that are close to ground potential. (Image source: Texas Instruments)

Going beyond just the analog-interface op amp alone, the Maxim Integrated MAX31865 is an easy-to-use resistance-to-digital converter optimized for Pt100 and Pt1000 RTDs (Figure 6). The IC is available in tiny 20-lead TQFN and SOIC packages and can be configured for two, three, and four-wire RTD interfaces while providing an SPI-compatible interface on the processor side.

Diagram of Maxim Integrated MAX31865 RTD-to-digital converterFigure 6: The Maxim Integrated MAX31865 RTD-to-digital converter includes the analog interface, digitizer, and SPI output for two, three, and four-wire RTDs. (Image source: Maxim Integrated)

A single external resistor sets the sensitivity for the RTD being used, and a precision 15-bit delta-sigma ADC converts the ratio of the RTD resistance and reference resistance into digital form, for a nominal temperature resolution of 0.03125⁰C and an accuracy of 0.5⁰C under all operating conditions and extremes.

Many temperature measurement applications require the use of multiple RTDs, along with other temperature sensors, to fully instrument a test set-up. For these applications, the Analog Devices LTC2983 sensor-to-digital, high-accuracy digital temperature measurement system IC supports a multiplicity of sensors and options. It handles up to 20 sensor channels which can be a mix of two, three, and four-wire RTDs, thermocouples, thermistors, and even diodes (Figure 7). The IC can be programmed with the specific type of sensor and desired excitation, and then provide built-In standard coefficients for these sensors; it also supports custom, user-specified coefficients.

Diagram of Analog Devices LTC2983 universal inputsFigure 7: The twenty universal inputs of the Analog Devices LTC2983 can be mixed as needed for sharing among thermocouples, two, three, or four-wire RTDs, thermistors, and diodes used as temperature sensors. (Image source: Analog Devices)

It provides the digital results via an SPI interface in °C or °F, with 0.1°C accuracy and 0.001°C resolution. It operates from a single 2.85 volt to 5.25 volt supply and includes excitation current sources and fault detection circuitry appropriate for each type of temperature sensor, as well as cold junction compensation (CJC) for any thermocouples.

For RTD data acquisition designs where the team wants to create a tailored complete circuit but not “reinvent the wheel”, Microchip Technology offers the TMPSNS-RTD1 Pt100 RTD evaluation board. The board supports two RTDs and allows for user configuration of key operating parameters, including RTD current (Figure 8).

Image of Microchip Technology’s TMPSNS-RTD1 Pt100 RTD evaluation boardFigure 8: Microchip Technology’s TMPSNS-RTD1 Pt100 RTD evaluation board supports two RTDs and provides user configurability of key operating parameters. (Image source: Microchip Technology)

The evaluation board block diagram shows how it builds up the complete RTD interface channel function-by-function, so users can understand the circuit and then adapt it as needed (Figure 9). The board has an internal RTD and an external two, three, or four-wire Pt100 RTD can also be connected, along with a low-current current source to minimize self-heating. The voltage across the RTD is amplified using the MCP6S26 programmable gain amplifier (PGA). The PGA boosts the RTD voltage and also allows the user to digitally program the amplifier gain and increase the sensor output range. Additionally, a differential amplifier drives a 12-bit differential analog-to-digital converter (ADC). Finally, the converter output data is read out by the microcontroller using an SPI interface and sent to the host PC via the USB interface.

Block diagram of the Texas Instruments TMPSNS-RTD1 Pt100 RTD evaluation boardFigure 9: The block diagram of the TMPSNS-RTD1 Pt100 RTD evaluation board shows the AFE and associated signal path from RTD excitation/sensing through SPI interface. (Image source: Microchip Technology)

The associated user’s guide includes full installation and setup information as well as step-by-step instructions for the intuitive PC-based graphical user interface (GUI). This GUI allows users to set parameters such as number of samples, sample rate, PGA gain, internal RTD current, and external current (Figure 10).

Image of Texas Instruments TMPSNS-RTD1 Pt100 RTD evaluation board PC-based GUIFigure 10: By applying the supplied PC-based GUI, users of the TMPSNS-RTD1 Pt100 RTD evaluation board can adjust key operating points and evaluate resultant performance. (Image source: Microchip Technology)

To complete the documentation, the user’s guide includes a fully detailed bill of materials (BOM), schematic diagram, top and bottom pc board layouts, and silk screens.

Conclusion

Temperature measurement is a basic function and the RTD is a popular, widely used sensor for this application, even though its proper use can be deceptively complex. However, when driven and sensed with the appropriate circuitry, it is capable of providing high-precision and repeatability over a wide temperature range. As with any high-performance sensor, its characteristics must be understood to achieve optimum performance. As shown, ICs with differing levels of functional integration allow users to build RTD-based systems with minimal surprises and superior performance.

Disclaimer: The opinions, beliefs, and viewpoints expressed by the various authors and/or forum participants on this website do not necessarily reflect the opinions, beliefs, and viewpoints of Digi-Key Electronics or official policies of Digi-Key Electronics.

About this author

Bill Schweber

Bill Schweber is an electronics engineer who has written three textbooks on electronic communications systems, as well as hundreds of technical articles, opinion columns, and product features. In past roles, he worked as a technical web-site manager for multiple topic-specific sites for EE Times, as well as both the Executive Editor and Analog Editor at EDN.

At Analog Devices, Inc. (a leading vendor of analog and mixed-signal ICs), Bill was in marketing communications (public relations); as a result, he has been on both sides of the technical PR function, presenting company products, stories, and messages to the media and also as the recipient of these.

Prior to the MarCom role at Analog, Bill was associate editor of their respected technical journal, and also worked in their product marketing and applications engineering groups. Before those roles, Bill was at Instron Corp., doing hands-on analog- and power-circuit design and systems integration for materials-testing machine controls.

He has an MSEE (Univ. of Mass) and BSEE (Columbia Univ.), is a Registered Professional Engineer, and holds an Advanced Class amateur radio license. Bill has also planned, written, and presented on-line courses on a variety of engineering topics, including MOSFET basics, ADC selection, and driving LEDs.

About this publisher

Digi-Key's North American Editors