“This article discusses the history and design challenges of resistance temperature detector (RTD)-based temperature measurement systems. This article also covers RTD selection and configuration trade-offs. Finally, the paper details RTD system optimization and evaluation.
By: Jellenie Rodriguez, Applications Engineer Mary McCarthy, Analog Devices
This article discusses the history and design challenges of resistance temperature detector (RTD)-based temperature measurement systems. This article also covers RTD selection and configuration trade-offs. Finally, the paper details RTD system optimization and evaluation.
Why is RTD temperature measurement important?
Temperature measurement plays an important role in many different end applications such as industrial automation, instrumentation, condition monitoring (CbM) and medical equipment. Regardless of the monitoring environmental conditions or the drift performance of the correction system, high accuracy and precision are very important. There are various types of temperature sensors available, such as thermocouples, resistance temperature detectors (RTDs), Electronic bandgap sensors, and thermistors. Which temperature sensor to choose and how to design it depends on the temperature range being measured and the desired accuracy. For temperatures between -200°C and +850°C, RTDs offer an excellent combination of high accuracy and good stability.
What are the main challenges of temperature measurement?
► Current and voltage selection. RTD sensors are passive devices and do not generate electrical output by themselves. The resistance of a sensor is measured using excitation current or voltage, that is, a small current is passed through the sensor to generate a voltage. How to choose current/voltage?
► Is the best choice for a specific design a 2-wire, 3-wire or 4-wire?
► How should RTD signals be conditioned?
► How can the above variables be adjusted to use converters or other building blocks within specification?
► Connecting multiple RTDs in the system – how to connect sensors? Can some modules be shared between different sensors? What is the impact on the overall performance of the system?
► What is the expected error of the design?
RTD Selection Guide
The resistance of an RTD sensor is a function of temperature in some well-defined way. The most widely used RTDs are platinum Pt100 and Pt1000, which are available in 2-wire, 3-wire and 4-wire configurations. Other RTD types are made of nickel and copper.
Table 1. Common RTD Types
The most common Pt100 RTDs come in two shapes: wirewound and thin film. Each type is constructed according to several standardized curves and tolerances. The most common normalization curve is the DIN curve. DIN stands for “Deutsches Institut für Normung”, which means “German Institute for Standardization”. The curves define the resistance versus temperature, normalized tolerance, and operating temperature range of a platinum 100Ω sensor. Its defined RTD accuracy starts with a base resistance of 100Ω at 0°C. DIN RTDs have different standard tolerance classifications. These tolerances are shown in Table 2, and they also apply to Pt1000 RTDs used in low-power applications.
Table 2. RTD Accuracy – Class A, Class B, 1/3 DIN
When choosing an RTD sensor, both the RTD itself and its accuracy are considered. The temperature range varies with element type, and the accuracy shown at calibration temperature (usually at 0°C) varies with temperature. Therefore, the temperature range to be measured must be defined, taking into account that any temperature below or above the calibration temperature will have wider tolerances and lower accuracy.
RTDs are classified by their nominal resistance at 0°C. The temperature coefficient of Pt100 sensor is about 0.385Ω/℃, and the temperature coefficient of Pt1000 is 10 times larger than that of Pt100. Many system designers use these coefficients to obtain approximate resistance-to-temperature conversions, but the Callendar-Van Dusen equation provides a more accurate conversion.
For temperature t ≤ 0°C, the formula is:
For temperature t ≥ 0°C, the formula is:
t is RTD temperature (°C)
R0 is the resistance of the RTD at 0°C (R0 = 100 Ω in this example)
A = 3.9083 × 10−3
B = −5.775 × 10−7
C = −4.183 × 10−12
RTD wiring configuration
Another sensor parameter to consider when choosing an RTD is its wiring configuration, which affects system accuracy. There are three different RTD wiring configurations on the market, each with its own advantages and disadvantages, and may require different techniques to reduce measurement errors.
The 2-wire configuration is the simplest but the least accurate because errors in lead resistance and its variation with temperature cause significant measurement errors. Therefore, this configuration should only be used in applications with very short leads or where high resistance sensors such as Pt1000 are used, which minimizes the effect of lead resistance on accuracy.
The 3-wire configuration has the advantage of using three pins, making it the most used configuration and useful in designs where connector size is minimized (only 3 connection terminals are required, compared to 4-wire terminals for 4-wire RTDs). The 3-wire configuration also offers a significant improvement in accuracy relative to the 2-wire configuration. Lead resistance errors in 3-wire configurations can be compensated for by different calibration techniques, which are described later in this article.
4-wire is the most expensive but most accurate configuration. This configuration eliminates errors caused by lead resistance and the effects of temperature variations. Therefore, a 4-wire configuration achieves the best performance.
RTD configuration circuit
High-accuracy RTD sensor measurements require precise signal conditioning, analog-to-digital conversion, linearization, and calibration. A typical design of an RTD measurement system consists of different circuit stages, as shown in Figure 2. While the signal chain may seem simple, there are several complexities involved, and designers must consider complex component selection, connection diagrams, error analysis, and analog signal conditioning challenges. The above factors affect the overall system board size and bill of materials (BOM) cost due to the high number of modules involved. But the good news is that Analog Devices offers a plethora of integrated solutions. This complete system solution helps designers simplify designs, reduce board size, shorten time to market, and reduce the cost of the entire RTD measurement system.
Figure 1. RTD Wiring Configuration
Figure 2. Typical RTD Measurement Signal Chain Module
The three RTD wiring configurations require different wiring techniques to connect the RTD to the ADC, as well as other external components and ADC requirements, such as excitation current and flexible multiplexers. This section will discuss each RTD configuration circuit design and considerations in more depth.
When designing RTD systems, Sigma-Delta (Σ-Δ) ADCs offer several advantages. First, sigma-delta ADCs are capable of oversampling the analog input, minimizing external filtering and requiring only a simple RC filter. In addition, they support flexible selection of filter type and output data rate. In mains-powered designs, built-in digital filtering can be used to suppress AC power interference. 24-bit high-resolution ADCs such as the AD7124-4/AD7124-8 have a peak resolution of 21.7 bits (max). Other advantages include:
► Analog input with wide common mode range
► Reference input with wide common mode range
► Capable of supporting ratiometric configuration
► Buffered reference and analog inputs
Some sigma-delta ADCs integrate many functions, including:
► Programmable Gain Amplifier (PGA)
► Excitation current
► Reference/Analog Input Buffer
► Calibration function
These ADCs significantly simplify RTD design and reduce BOM, system cost, board space, and time to market.
For this article, the AD7124-4/AD7124-8 are used as ADCs. These two devices are low noise, low current precision ADCs with integrated PGA, excitation current, analog input and reference buffer.
The ratiometric configuration is a suitable and cost-effective solution for systems using resistive sensors such as RTDs or thermistors. With the ratiometric method, the reference voltage and sensor voltage are derived from the same excitation source. Therefore, the excitation source does not need to be very precise. Figure 3 shows an example of a ratiometric configuration in a 4-wire RTD application. A constant excitation current powers the RTD and precision resistor RREF, and the voltage developed across RREF is the reference voltage for RTD measurements. Any change in excitation current will not affect the accuracy of the measurement. Therefore, the ratiometric approach allows the use of a noisier and less stable excitation current. Excitation current has better noise immunity than voltage excitation. The main factors to consider when choosing an excitation source value are discussed later in this article.
Figure 3. 4-Wire RTD Ratio Measurements
IOUT/AIN shared pin
Many RTD system designers use sigma-delta ADCs that integrate multiplexers and excitation currents to support multichannel measurements and flexibly connect excitation currents to individual sensors. ADCs such as the AD7124 allow a single pin to be used as both an excitation current and an analog input pin (see Figure 4). Since IOUT and AIN share pins, only two pins are required per 3-wire RTD sensor, which is beneficial for increasing the channel count. However, in this configuration, the large value resistor R in anti-aliasing or electromagnetic interference (EMI) filtering in series with the RTD will introduce errors in the RTD resistance value, so the value of R is limited. Because of this, it is generally recommended to have dedicated pins for each excitation current source to avoid introducing errors into RTD measurements.
Figure 4. 3-wire RTD, IOUT/AIN pin sharing
4-Wire RTD Connection Diagram
A 4-wire RTD configuration performs best. Compared to the other two configurations, the only issues facing the system designer are the cost of the sensor itself and the size of the 4-pin connector. In this configuration, lead-induced errors are eliminated through the return line. The 4-wire configuration uses Kelvin sensing, with two wires carrying the excitation current to and from the RTD, and the remaining two wires sensing the current in the RTD element itself. Errors caused by pin resistance are eliminated by the system itself. The 4-wire configuration requires only one excitation current, IOUT, as shown in Figure 5. Three analog pins from the ADC are used to implement a single 4-wire RTD configuration: one pin is used to excite the current IOUT, and two pins are used as fully differential input channels (AINP and AINM) to sense the voltage on the RTD.
When designing with multiple 4-wire RTDs, a single excitation current source can be used and directed to different RTDs in the system. By placing the reference resistor on the low side of the RTD, a single reference resistor can support all RTD measurements. That is, this reference resistor is shared by all RTDs. Note that if the reference input to the ADC has a wide common-mode range, the reference resistor can be placed on the high or low side. Therefore, for a single 4-wire RTD, a reference resistor on the high or low side can be used. However, when multiple 4-wire RTDs are used in the system, it is advantageous to place the reference resistor on the low side because one reference resistor can be shared by all RTDs. Note that some ADCs have built-in reference buffers. These buffers may require some headroom, so if the buffers are enabled, headroom resistors are required. Enabling the buffer means that stronger filtering can be connected to the reference pin without introducing errors such as gain errors within the ADC.
2-Wire RTD Connection Diagram
The 2-wire RTD configuration is the simplest configuration, as shown in Figure 6. The 2-wire configuration requires only one excitation current source. Three analog pins from the ADC are used to implement a single 2-wire RTD configuration: one pin is used to excite the current IOUT, and two pins are used as fully differential input channels (AINP and AINM) to sense the voltage on the RTD. When designing with multiple 2-wire RTDs, a single excitation current source can be used and the excitation current directed to different RTDs in the system. By placing the reference resistor on the low side of the RTD in a 4-wire configuration, a single reference resistor can support all RTD measurements. That is, this reference resistor is shared by all RTDs.
The 2-wire configuration is the least accurate of the three wiring configurations because the actual resistance value measured includes both the resistance value of the sensor and the resistance values of leads RL1 and RL2, thereby increasing the voltage measurement on the ADC. If the sensor is remote and the system uses very long wires, the error will be large. For example, the equivalent resistance of 25 feet of 24 AWG copper wire is: 0.026Ω/foot (0.08Ω/meter) × 2 × 25 feet = 1.3Ω. Therefore, the error due to 1.3Ω wire resistance is: (1.3/0.385) = 3.38°C (approximately). Wire resistance also changes with temperature, which again adds to the error.
Figure 5. Single and Multiple 4-Wire RTD Analog Input Configuration Measurements
3-Wire RTD Connection Diagram
Using a 3-wire RTD configuration can greatly improve the large errors caused by the lead resistance of a 2-wire RTD configuration. This article uses a second excitation current (shown in Figure 7) to cancel out the lead resistance errors created by RL1 and RL2. Therefore, four analog pins from the ADC are used to implement a single 3-wire RTD configuration: two pins for excitation current (IOUT0 and IOUT1) and two pins as fully differential input channels (AINP and AINM) for detection Voltage on RTD.
Figure 6. Single and Multiple 2-Wire RTD Analog Input Configuration Measurements
Figure 7. Single and Multiple 3-Wire RTD Analog Input Configuration Measurements
There are two ways to configure a 3-wire RTD circuit. Method 1 puts the reference resistor on the top side, so that the first excitation current IOUT0 flows to RREF, RL1, and then flows to RTD; the second current flows through the RL2 lead resistance, and the resulting voltage cancels the voltage drop on the RL1 lead resistance. Therefore, a well-matched excitation current can completely eliminate errors caused by lead resistance. If the excitation currents are not so well matched, using this configuration minimizes the effects of the mismatch. The same current flows to RTD and RREF; therefore, any mismatch between the two IOUTs will only affect the lead resistance calculation. This configuration is useful when measuring a single RTD.
When measuring multiple 3-wire RTDs, it is recommended to place the reference resistor on the bottom edge (Method 2), so that only a single reference resistor can be used, thus minimizing the overall cost. However, in this configuration, one current flows through the RTD, but two currents flow through the reference resistor. Therefore, any mismatch in IOUT will affect the value of the reference voltage and the cancellation of the lead resistance. When there is an excitation current mismatch, the error of this configuration will be larger than that of method 1. There are two possible ways to calibrate the mismatch and mismatch drift between IOUT, thereby improving the accuracy of the second configuration. The first method is to chop (swap) the excitation current, perform a measurement at each stage, and then average the two measurements to achieve calibration. Another way is to measure the actual excitation current itself and then use the calculated mismatch at the microcontroller to compensate for that mismatch. More details on these calibrations are discussed in CN-0383.
RTD system optimization
Examining the system designer’s problems reveals different challenges in designing and optimizing solutions for RTD applications. Challenge one is the sensor selection and connection diagrams discussed above. The second challenge is the configuration of the measurement, including ADC configuration, setting excitation current, setting gain, and selecting external components, while ensuring that the system is optimized and operating within ADC specifications. Finally, the most critical question is how to achieve the target performance and determine which error sources contribute to the overall system error.
Fortunately, there is a new tool
RTD_CONFIGURATOR_AND_ERROR_BUDGET_CALCULATOR, which provides a hands-on solution for designing and optimizing RTD measurement systems from concept to prototyping.
► Helps understand proper configuration, wiring and circuit diagrams
► Helps understand different error sources and supports design optimization
Designed around the AD7124-4/AD7124-8, the tool allows customers to adjust settings such as excitation current, gain, external components, and more. It will indicate out-of-bounds conditions to ensure the final solution is within the ADC’s specifications.
Figure 8. RTD Configurator
Selection of Excitation Current, Gain and External Components
Ideally, we tend to choose a higher excitation current to produce a higher output voltage and maximize the ADC input range. However, since the sensor is resistive, the designer must also ensure that the power dissipation or self-heating effects of large excitation currents do not affect the measurement results. The system designer may choose a high excitation current. However, to minimize self-heating, the excitation current needs to be turned off between measurements. Designers need to consider the impact of timing on the system. Another approach is to choose a lower excitation current to minimize self-heating. Timing is now minimized, but designers need to determine if system performance is affected. All scenarios can be tested by RTD_Configurator_and_Error_Budget_Calculator. The tool allows the user to balance the selection of excitation current, gain and external components to ensure that the analog input voltage is optimized, while adjusting the ADC gain and speed to provide better resolution and system performance, i.e. lower noise and offset errors.
To understand the resulting filter curve, or to gain a deeper understanding of the transition timing, the VirtualEval online tool provides details.
Both the ADC input and the reference input of the sigma-delta ADC are sampled continuously by the switched capacitor front end. For the RTD system in question, the reference input is also driven by an external reference resistor. It is recommended to use an external RC filter on the analog input of the sigma-delta ADC for antialiasing. For EMC purposes, the system designer can use larger values of R and C on the analog and reference inputs. Large RC values can cause gain errors in the measurement because the front-end circuitry does not have sufficient time to settle in the time between two sampling instants. Buffered analog and reference inputs prevent such gain errors, allowing unlimited R and C values to be used.
For the AD7124-4/AD7124-8, the analog input buffer is automatically enabled when using an internal gain greater than 1. Since the PGA is placed in front of the input buffer and the PGA is rail-to-rail, the analog input is also rail-to-rail rail. However, for reference buffers, or using the ADC at a gain of 1 and the analog input buffers enabled, it is necessary to ensure that the necessary headroom is provided for proper operation.
The signal level output by the Pt100 is very low, on the order of a few hundred mV. For best performance, a wide dynamic range ADC can be used. Or use a gain stage to amplify the signal before applying it to the ADC. The AD7124-4/AD7124-8 support gains of 1 to 128, allowing the design to be optimized for various excitation currents. Multiple options for PGA gain allow designers to trade-off between excitation current value and gain, external components, and performance. The RTD configuration tool will indicate whether the new excitation current value can be used with the selected RTD sensor. It also gives appropriate recommendations for precision reference resistors and reference headroom resistors. Note that this tool ensures that the ADC is used within specification – it will show possible gains that support the relevant configuration. The excitation current of the AD7124 is output compliant; that is, the voltage on the pin supplying the excitation current requires some headroom with respect to AVDD. The tool will also ensure compliance with this compliance specification.
With RTD tools, system designers can guarantee that the system will operate within the operating limits of the ADC and RTD sensors. The accuracy of external components such as reference resistors and their contribution to systematic errors will be discussed later.
Filtering options (analog and digital 50 Hz/60 Hz rejection)
As mentioned earlier, an antialiasing filter is recommended for use with a sigma-delta converter. Embedded filters are digital, so the frequency response folds around the sampling frequency. To adequately attenuate interference at the modulator frequency and its multiples, antialiasing filtering must be used. The sigma-delta converter oversamples the analog input, so the design of the antialiasing filter is greatly simplified, requiring only a simple single-pole RC filter.
When the final system is put into field use, it can be very challenging to deal with noise or interference from the environment in which the system is located, especially in applications such as industrial automation, instrumentation, process control or power control that require noise immunity while not being able to Generates too much noise and affects adjacent components. Noise, transients, or other sources of interference can affect system accuracy and resolution. Interference can also occur when the system is powered by AC power. The AC power frequency is 50 Hz and its multiples in Europe and 60 Hz and its multiples in the United States. Therefore, when designing RTD systems, filter circuits with 50 Hz/60 Hz rejection must be considered. Many system designers want to design a general-purpose system that can reject both 50 Hz and 60 Hz.
Most lower bandwidth ADCs, including the AD7124-4/AD7124-8, offer several options for digital filtering, and the notch frequency can be programmed to 50 Hz/60 Hz. The selected filter option affects the output data rate, settling time, and 50 Hz/60 Hz rejection. When multiple channels are enabled, each channel switching requires a settling time for the conversion result to occur. Therefore, choosing a filter type with a longer settling time (ie sinc4 or sinc3) reduces the overall throughput rate. In this case, a post filter or FIR filter can be used to provide reasonable simultaneous 50 Hz/60 Hz rejection with shorter settling times, thereby increasing the throughput rate.
The current consumption or power budget allocation of the system is highly dependent on the end application. The AD7124-4/AD7124-8 feature three power modes that allow trade-offs between performance, speed, and power consumption. Portable or remote applications require low-power devices and configurations. For some industrial automation applications, the entire system is powered by a 4 mA to 20 mA loop, so the maximum allowed current budget is only 4 mA. For such applications, the device can be set to medium or low power modes. The speed is much lower, but the ADC still provides high performance. If the application is an AC powered process control, the current consumption can be much higher, so the device can be set to full power mode, and the system can achieve much higher output data rates and higher performance.
Error sources and calibration options
Once the desired system configuration is known, the next step is to estimate the ADC-related and systematic errors. These errors help the system designer understand whether the front-end and ADC configuration is meeting the overall target accuracy and performance. RTD_Configurator_and_Error_Budget_Calculator allows user to modify system configuration for best performance. For example, Figure 9 shows a summary of all errors. The systematic error pie chart shows that the initial accuracy of the external reference resistor and its temperature coefficient are the main contributors to the total systematic error. Therefore, an external reference resistor with higher accuracy and better temperature coefficient must be considered.
The error caused by the ADC is not the most important contributor to the total error of the system. However, using the AD7124-4/AD7124-8’s internal calibration mode can further reduce the ADC’s error contribution. Internal calibration is recommended at power-up or software initialization to remove ADC gain and offset errors. Note that these calibrations do not remove errors caused by external circuitry. However, the ADC also supports system calibration so that system offset and gain errors can be minimized, but this may add additional cost and may not be required for most applications.
For harsh environments or applications where safety is important, diagnostics are becoming part of the industry requirements. Embedded diagnostics in the AD7124-4/AD7124-8 reduce the need for external components to implement the diagnostics, resulting in a smaller solution size, shorter time, and lower cost. Diagnosis includes:
► Check the voltage levels on the analog pins to ensure they are within the rated operating range
► Cyclic Redundancy Check (CRC) for Serial Peripheral Interface (SPI) bus
► Memory mapped CRC
► Signal chain check
These diagnostics make the solution more robust. According to IEC 61508, Failure Modes, Effects and Diagnostic Analysis (FMEDA) of a typical 3-wire RTD application shows a Safe Failure Rate (SFF) greater than 90%.
RTD System Evaluation
Figure 10 shows some measurement data from Circuit Note CN-0383. This measurement was obtained using the AD7124-4/AD7124-8 evaluation board, which includes a demo mode for a 2-/3-/4-wire RTD, and calculated the corresponding temperature in degrees Celsius. The results show that the error of the 2-wire RTD implementation is closer to the lower limit of the error boundary, while the overall error of the 3- or 4-wire RTD implementation is well within the allowable limits. The higher error in the 2-wire measurement results from the lead resistance error described earlier.
Figure 9. RTD Error Source Calculation Program
Figure 10. 2-/3-/4-Wire RTD Temperature Accuracy Measurement Post Filter, Low Power Mode, 25 SPS
These examples illustrate that following the RTD guidelines above will enable a high-accuracy, high-performance design when used with ADI’s lower bandwidth sigma-delta ADCs such as the AD7124-4/AD7124-8. The circuit note (CN-0383) is also available as a reference design to help system designers quickly prototype. Evaluation boards allow the user to evaluate system performance, and each example configuration demonstration mode is available. Further, using the sample code generated by ADI (available from the AD7124-4/AD7124-8 product page), it is easy to develop firmware for different RTD configurations.
ADCs with sigma-delta architectures such as the AD7124-4/AD7124-8 are suitable for RTD measurement applications because they address issues such as 50 Hz/60 Hz rejection and the analog input has a wide common-mode range (reference input may also have). In addition, these devices are highly integrated and contain all the functions required for RTD system design. They also offer enhanced features such as calibration capabilities and embedded diagnostics. This level of integration, coupled with a complete system profile or ecosystem, will simplify overall system design, reduce costs, and shorten the design cycle from concept to prototype.
To make the design journey easier for system designers, the RTD_Configurator_and_Error_Budget_Calculator tool and online tool VirtualEval, evaluation board hardware and software, and CN-0383 can be used to address different challenges such as connectivity issues and overall error budget, improving the user’s design experience to a higher level.
This article has demonstrated that designing an RTD temperature measurement system is a challenging multi-step process. It requires choosing different sensor configurations, ADCs, and optimizations, and considering how these decisions affect overall system performance. Analog Devices’ RTD_Configurator_and_Error_Budget_Calculator tool and online tool VirtualEval, evaluation board hardware and software, and CN-0383 simplify the process by addressing connectivity and overall error budgeting issues.
About the Author
Jellenie Rodriguez is an applications engineer in the Precision Converter Technology Group at Analog Devices. Her main focus is on precision sigma-delta ADCs for DC measurements. She joined Analog Devices in 2012 and graduated from San Sebastian College-Recoletos de Cavite in 2011 with a bachelor’s degree in electrical engineering. Contact information:[email protected]