Implementing the right thermal sensing option in your embedded design - Embedded.com

Implementing the right thermal sensing option in your embedded design

Designers face many challenges when implementing thermal sensing solutions in embedded systems including thermistors, resistive thermal detectors (RTDs) and thermocouples. In this article, key design criteria such as temperature accuracy, power consumption, system cost and design complexity are addressed for each of these solutions.

We compare these solutions to silicon temperature sensor ICs and discuss how IC sensors provide flexible and cost-effective thermal-management solutions. We will also show to use integrated features such as integrated EEPROM, and under-/over- and critical-temperature monitoring to improve system performance.

Thermistors are the most common method used to sense temperature. Thermistors are made using semiconductor materials and can have either a positive or negative temperature coefficient (PTC or NTC, respectively).

The thermistor's resistance changes with respect to changes in temperature. PTC thermistor resistance increases when temperature increases, whereas NTC thermistor resistance decreases when temperature increases.

There are a few advantages to thermistor-based solutions. Thermistors are highly sensitive to changes in temperature, they have quick thermal response, and they are inexpensive. The biggest disadvantage is that thermistors are highly nonlinear over wide temperature ranges.

Figure 1: Shown is a thermistor circuit with a low-pass filter and a unity-gain buffer amplifier.

Figure 1 above shows a thermistor circuit with a low-pass filter and a unity-gain buffer amplifier. The low-pass filter (R2 and C1) network filters system noise from the sensor output, and the unity-gain buffer is used to drive resistive or capacitive loads.

The voltage across the thermistor (VTH) is proportional to the change in temperature. The graph indicates a linear response from 0 to 70 degrees Celsius . However, there is significant nonlinearity at temperature extremes. The change in resistance with respect to temperature is much less, when compared to the linear region. This requires some signal amplification, to improve measurement resolution at hot and cold temperature extremes.

Thermistors are inexpensive and provide accurate temperature monitoring, over a limited temperature range. To achieve high accuracy over a wider range requires a more complex design, including multiple gain stages at various temperature ranges using a programmable gain amplifier (PGA).

This provides a robust measurement solution but increases overall system cost. Additionally, thermistors require biasing current, which is set using R1 in Figure 1. Higher current increases temperature-measurement resolution.

However, it also increases temperature-measurement error. This is due to self-heat generated from the power that is dissipated across the thermistor. In most applications, other solutions such as IC sensors are better suited for extended-temperature applications.

Robust solution
Resistive temperature detectors (RTDs) are a robust temperature monitoring solution. These sensors provide excellent repeatability and stability characteristics. A designer can achieve high accuracy over several 100s of degrees Celsius range using RTDs.

This requires careful scaling, calibration and resistance to temperature conversion using the various RTD standards and specifications that are in place worldwide. A basic RTD circuit requires a constant current source for biasing and analog instrumentation circuit (such as an instrumentation amplifier) to instrument the voltage drop across the RTD.

This solution can be expensive due to the cost of the instrumentation amplifier and the manual operation required, including setting the measurement range and calibrating the gain and offset voltages. The amplifier output is typically connected to an ADC for digitization. Other circuits convert the change in resistance to frequency.

For example, the circuit in Figure 2 below shows a relaxation oscillator circuit, which uses RC network and a comparator to generate a frequency proportional to the change in temperature.

Figure 2: Shown is a relaxation oscillator circuit that uses RC network and a comparator to generate a frequency proportional to the change in temperature.

The frequency can be directly connected to an MCU for digitization. When designing an RTD circuit, the effect of self-heat due to power dissipation must be carefully considered.

RTDs have excellent repeatability and can provide a precise temperature-monitoring solution over a wide temperature range. Trade-offs include cost, design complexity and increased system power consumption due to the presence of multiple active components.

Thermocouples
Thermocouples have a wide operating temperature range of -270 to 1,372 degrees Celsius . The American Society for Testing and Materials defines a number of commercially available thermocouple classi- fications in terms of performance.

Types E, J, K, N and T are basemetal thermocouples, and can be used to measure temperatures from about -270 to 1,372 degrees Celsius . Types S, R and B are noble-metal thermocouples, and can be used to measure temperatures from about -50 to 1,820 degrees Celsius .

Thermocouples use two metal alloys such as Alumel and Chromel to measure temperature. The two metals are welded at one end, and open at the other end. The electrical characteristics of the wires at the welded point are temperature dependent.

A voltage is generated at the welded tip, which can be measured at the open end using a volt meter. The voltage magnitude increases and decreases in proportion to temperature changes. Thermocouples are highly nonlinear and require linearization algorithms.

The welded tip is referred to as the hot junction, and the open end is the cold junction. Temperature is measured by calculating the difference between the hot and cold junction temperatures. The temperature at the cold junction is used as a reference for the hot junction. The temperature at the cold junction is measured using thermistors, RTDs or silicon IC temperature sensors.

Figure 3: A typical thermocouple circuit.

The full-scale voltage range of a thermocouple is less than 100mV. Therefore, high-performance analog signal conditioning is required. The circuit in Figure 3 above shows a typical thermocouple circuit.

For industrial applications, the thermocouple is connected to the instrumentation system with EMI filters. The thermocouple terminals are connected to the positive and negative supplies through large resistors, so that the circuit can detect an open-circuit condition.

Auto-zero and chopper amplifiers can be used for signal conditioning, due to the low offset voltage and common mode rejection specifications. The cold-junction compensation circuit is implemented with a silicon IC temperature sensor, located on the PCB.

Silicon temperature sensors
Many semiconductor manufacturers offer silicon-based temperature sensors. The operating temperature range of these sensors is typically -55 to 150 degrees Celsius . These devices can be categorized into three classes—logic-output, voltage-output and serial-output temperature sensors. IC sensors integrate many useful features that allow system designers to implement the design that best meets the application's requirements.

Temperature-sensor ICs require minimal design effort, and the integrated features can decrease overall system cost and minimize design effort, including:

1. Logic-output temperature Sensors . These sensors typically function as a thermostat, notifying the system that a minimum or maximum temperature limit has been reached. Sometimes referred to as “temperature switches,” these devices can be used to turn either a fan or warning light on when high- or low-temperature conditions are detected. Typically, the outputs are not latched.

Therefore, the switch will turn off when the temperature falls below or rises above the temperature set point. Most logic-output sensors have an integrated hysteresis of a few degrees Celsius, to prevent output chatter.

Figure 4: Shown are typical applications using logic-output temperature-sensor ICs.

Logic-output temperature sensors are available in either a “hot” option, where the output toggles as temperature increases; or a “cold” option, where the output toggles as temperature decreases. The hot and cold options ensure that the hysteresis is in the appropriate position, either below or above the temperature set point. Figure 4 above illustrates several circuits using logic-output temperature sensors.

2. Voltage-output temperature Sensors. .These sensors develop an output voltage proportional to temperature, with a typical temperature coef- ficient of 6.25mV, 10mV, or 19.5mV per degree Celsius .

Temperature-to-voltage converters can sense a -55 to +150 degrees Celsius temperature range and feature temperature offset, which allows reading negative temperatures without requiring a negative supply voltage. Typical operating currents are in the tens of microamps, minimizing self heating due to power dissipation and maximizing battery life.

The device output is typically connected to a standalone ADC or to a microcontroller with an integrated ADC, as shown in Figure 5 below .

Figure 5: The device output is typically connected to a standalone ADC or to a microcontroller with an integrated ADC.

3. Serial-output temperature Sensors. Typically, serial-output temperature sensors use a two- or three-wire interface to the host microcontroller. These devices have an integrated ADC that converts the analog output of the internal sensing element to a digital output. They can achieve temperature accuracies as high as 0.5 degrees Celsius , with a measurement resolution of less than 0.1 degrees Celsius .

Many serial-output temperature sensors provide user-programmable functions, such as over- and under-temperature alerts and integrated EEPROM for general-purpose data storage. These features can be used to simplify a design, increase design flexibility, improve temperature sensing accuracy, and lower overall system cost. The over- and under-temperature alert feature works in the same way as it does for logic-output temperature sensors.

Using the serial interface, the host MCU loads temperature trigger limits in degrees Celsius, into an internal register located in the silicon temperature sensor. When the desired temperature limit is exceeded, the sensor flags the host controller that an over or under-temperature condition occurred. This feature can be used to turn on a light or control a fan via a serial interface, without the need for the microcontroller to monitor temperature continuously.

This increases flexibility by freeing up the host microcontroller from having to continuously monitor the system. It also simplifies software and hardware development.

Many applications today require temperature accuracy of less than 0.5 degrees Celsius , over a fairly wide temperature range. Higher accuracy can be achieved with many silicon-based temperature sensors via a calibration lookup table that calibrates the sensor at various known temperatures.

The number of calibration points depends on the temperature range, the accuracy required and the non-ideal characteristics of the sensor. The accuracy vs. temperature graph in Figure 6 below demonstrates how a typical silicon temperature sensor's accuracy varies over temperature, before and after non-ideality error compensation.

Figure 6: Shown is silicon IC accuracy vs. temperature characteristics with and without compensation.

Figure 6 shows sensor accuracy before and after error compensation. The temperature sensor's non-ideality characteristics are illustrated in a predictable curve shape and can be described using a second-order polynomial equation.

The polynomial equation coefficients are generated by taking multiple temperature error points, from maximum to minimum temperature limits. The equation can be used to compensate the sensor temperature error by computing the equation at the measured temperature using a microcontroller.

The equation can also be used to generate a look up table that can be stored in EEPROM. Some temperature-sensor ICs have integrated 256bytes of EEPROM, which can be used to store the sensor non-ideality characteristics in a look-up table. The on-chip EEPROM can also be used for general-purpose data storage.

Temperature sensors bring with them a variety of advantages and disadvantages. No one type of sensor is appropriate for all temperature-sensing applications. To identify the appropriate sensor, the specific requirements of each application must be outlined.

Thermistors provide a useful, low-cost temperature-sensing solution for applications that operate over a limited temperature range due to its nonlinear characteristics. RTDs can be highly accurate over a several 100s of degrees Celsius range. RTDs require high-performance instrumentation systems that require manual scaling and calibration, which makes them more costly.

Thermocouples are most useful in applications that must operate in temperature extremes of less than -200 or more than 1,000 degrees Celsius . These sensors require high-performance instrumentation systems, which can be very costly. On the other hand, silicon IC-based temperature sensors simplify designs, while offering relatively high accuracy over a temperature range of -55 to 150 degrees Celsius . They also provide many integrated features that enhance system flexibility and performance.

( John Austin is Principal Product Marketing Engineer and Ezana Haile is Senior Applications Engineer in the Analog & Interface Products Division at Microchip Technology Inc. )

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.