Temperature Measurement for Everyone

Electronic circuits generate heat. A temperature change can cause resistors to drift or cause transistors operating in their linear range to saturate. Therefore, you may have to measure the temperature of individual circuit components during a design evaluation test or during a field failure test. You may also have to measure the temperature in an enclosure to verify that the product’s cooling vents and fans work properly. When you assemble a rack of instruments, you should measure the temperature inside it (with the doors closed) and, if necessary, provide an escape for the hot air.

When you measure the effectiveness of your cooling vents and fans, you often need to measure at least two temperatures. The difference between the vent’s input and output temperatures tells you how well the electronics are being cooled.

You also can use temperature measurements to get an indication of how much power a component consumes. For example, you might want to measure the temperature on both sides of a power transistor’s electrical insulator.

Probes and Meters

To measure temperature, you typically need a probe and a meter. For the probe, you can use a thermocouple, a resistance temperature detector (RTD), or a thermistor. You also can use an IC sensor as part of a temperature-measuring circuit. Finally, you can use an infrared thermometer to make a nonintrusive measurement. Table 1 highlights typical temperature ranges for each type of sensor.

Thermocouples, RTDs, and thermistors are available as just the sensor element with short lead wires or with the sensor mounted in a protective sheath. If you’re measuring the temperature of an IC, then you should use a sensor without a sheath. (A sheath may be too big to fit into a small area, and it may not make sufficient contact to get good thermal transfer.) Attach the sensor to your IC with tape or use an adhesive that conducts heat.

For measuring the temperature inside an enclosure or rack, use a probe with a sheath. Some sheaths come with mounting hardware attached. These probes might have plates with screw holes for mounting to flat surfaces or clamps for attaching to pipes.

Thermocouples are the most popular temperature probes. They’re the least expensive and easiest to use, and they have an accuracy of between 18C and 28C. You can mount the sensor several hundred feet from the electronics that measure its output. Thermocouples come in different types depending on their composition. The most popular thermocouple today is called a “type K.” If you have a DMM that can measure temperature, it probably works with type-K thermocouples. Your meter may also accept other thermocouple types such as type J or type T, and it may accept RTDs.

The simplest thermocouple assembly consists of two thermocouple wires crimped or welded together at one end. The opposite ends of the wires should be at the same temperature as each other so you have identical temperature differences along the lengths of both wires. Most thermocouple meters have a heat-conducting block to minimize any temperature difference between the wires at the meter.

While using just unsheathed wires is certainly easy and inexpensive, the thermocouple is prone to damage unless you permanently mount it. And because the wires may not be shielded, they are susceptible to EMI from computers, machinery, electrical lines, and so forth. Thermocouples generate a low-level voltage. At 1008C, a type-K thermocouple generates about 4 mV,¹ so EMI can interfere with your measurements.

When you use a thermocouple to measure temperature on metal surfaces, check for possible ground loops or shorts. If the metal surface is grounded, contact with the sensor may ground the sensor, too. If you use a thermocouple enclosed in a sheath, verify if the sheath or the thermocouple wire is grounded. You might short out the sensor or introduce interference.

RTDs offer better accuracy (less than 18C) and better long-term stability than thermocouples. Their temperature range isn’t as wide as most thermocouples, but it well exceeds the range you need for measurements on electronic circuits or even on vehicle engines. RTDs are less susceptible to EMI than are thermocouples.

Instead of generating a voltage like thermocouples do, RTDs change resistance with temperature. The most popular RTD, known as a Pt 100 sensor, consists of a platinum wire that has a nominal resistance of 100 V at 08C. The resistance-temperature curve for a Pt 100, which your meter will linearize, is based on a polynomial equation in which the most significant constant, a, is 0.00385 V/V/8C (also called the European curve).

You can get better than 18C accuracy with RTDs, but the resolution you get depends on your measurement instrument. RTDs typically change resistance by less than 0.4 V/8C. With 1 mA of excitation current, that translates into 0.4 mV/8C. So, your instrument needs to resolve better than 0.4 mV for 18C resolution.

Most industrial RTDs use three wires. One wire carries excitation current, typically 1 mA. Another wire carries the sense signals from the element. The third is a common wire for both sensing and excitation-current return. In applications that require the best accuracy, you can use a four-wire RTD, which separates the sense return and the current return, thereby eliminating IR losses on one of the wires. Typically, you won’t need four-wire RTDs for everyday measurements; calibration labs typically use four-wire RTDs to calibrate working probes.

Thermistors are also resistance devices, but they’re made of semiconductor material rather than platinum wire. Unlike RTDs, some thermistors have a resistance that decreases with temperature—these are called negative temperature coefficient (NTC) thermistors.

Thermistor resistance-to-temperature curves are also highly nonlinear, but today’s instruments can easily compensate for the nonlinearities. Because of their high sensitivity, thermistors usually require only two-wire connections. To minimize IR losses, use short wires—no longer than just a few feet—between the sensor and your measurement instrument.

A thermistor probe is the best choice when you need the highest resolution, that is, when you want to measure small changes in temperature. If you’re tying to measure the temperature coefficient of a component such as a crystal over a narrow temperature range, then use a thermistor. Thermistors rival RTDs in accuracy and have better resolution than other sensors. A typical 10-kV (at 258C) thermistor has a resolution of 438 V/8C.² That’s more than a 1000 times increase in resolution over an RTD.

Thermistors have a narrow temperature range compared to thermocouples and RTDs. Thermistors work best if the temperature range is between 08C and 1008C, although some can work as high as 3508C while others work down to –808C.

IC temperature sensors contain not only a temperature sensor but also signal-conditioning circuits. These sensors will produce a voltage or current output that is linearly proportional to temperature. Some even digitize the temperature and produce serial outputs. Most of these devices are packaged in standard IC packages for mounting on circuit boards and they all require a DC voltage source, typically 2.7 V to 5 V.

IC temperature sensors may require you to design your own electronics to digitize the analog output or to interpret the digital serial output. Sensor manufacturers provide application notes with sample schematics to help you with your design. IC temperature sensors provide accuracies between 28C and 38C.³

Infrared thermometers measure the surface temperature of objects within their field of view. The size of the field is important if you’re trying to measure the temperature of a single object. The field of view is inversely proportional to the distance to the object. Check the manufacturer’s data sheet for this specification. If the distance is too great (field of view too wide), the thermometer will display the average temperature of more than one object, which will give you a false reading. You also have to be concerned about emissivity—how temperature relates to radiation.

You can use a variety of instruments to measure temperature with probes. Instruments range from handheld meters that accept one probe to PC-plug-in cards to panel meters to benchtop, portable, or rack-mounted data-acquisition systems capable of scanning hundreds of channels. Simple meters may not store any measurements while others store from several readings to more than 100,000 readings. PC-based systems can store millions of readings, limited only by the PC’s memory. Stand-alone systems can also store millions of readings.

Regardless of which type of probe and instrument you use, you must verify that they will work together. Many instruments can accept more than one thermocouple, RTD, or thermistor type. Check that you’ve set the instrument for the type of sensor you’re using. T&MW

FOOTNOTES

1. Temperature Handbook, Technical reference section, Omega Engineering, Stamford, CT. www.omega.com/temperature/Z/zsection.asp.  

2. Zurbuchen, John, “Precision Thermistor Thermometry,” Measurement Science Conference Tutorial, January 20, 1993. Measurement Science Conference, Newport Beach, CA.

3. “Low Voltage Temperature Sensors, TMP35/TMP36/TMP37,” Data sheet, Analog Devices, Norwood, MA, 1997, www.analog.com. 

FOR FURTHER READING

Garvey, Doris, “So, What Is an RTD?,” Sensors, August 1999, p.39. www.sensorsmag.com/articles/0899/39/index.htm.

Haussmann, Werner, “Thermistor and DMM Measure Temperature,” Test & Measurement World, October 1999, p. 19. 

 You can contact Martin Rowe at m.rowe@tmworld.com.