What common applications use temperature measurement?
Temperature sensing is ubiquitous. Historically, temperature sensors have had well-known applications in environmental and process control as well as in test and measurement and communications. Contemporary applications include all those, plus high-volume consumer applications, a range of automotive applications, and medical products from MRI imagers to portable ultrasound scanners.
New communications applications range from basestations of all sizes to cellular handsets. Temperature sensors also can be found inside automotive engines and transmissions, where controllers adjust operating parameters based on temperature. And, they’re on circuit boards with big, fast processors and FPGAs.
What kinds of devices are available for temperature sensing, and how do they work?
There are traditional temperature sensors and silicon-based temperature sensors. Traditional sensors comprise thermistors, resistance temperature detectors (RTDs), and thermocouples. They are analog devices, so their outputs must be digitized before they can be used in a digital-control loop.
Thermistors are usually based on ceramics or polymers, and RTDs are based on metals. RTDs can be used across larger temperature ranges than thermistors. Because they’re purely resistive, thermistors and RTDs need external voltage sources.
In contrast, thermocouples use junctions of dissimilar metals to generate output voltages. These are proportional to temperature differences, rather than the absolute temperature of their surroundings.
Temperature sensors don’t have to be analog. There are silicon-based temperature- sensing ICs that output precise digital representations of the temperatures they are measuring. This simplifies design of the control system, compared to approaches that require external signal conditioning and an analog-todigital converter (ADC).
How do silicon temperature sensors work?
Some silicon temperature sensors take advantage of the principle that when two identical transistors are operated at a constant ratio of collector current densities, the difference in their base-emitter voltage is proportional only to absolute temperature. Other sensors are based on the behavior of the base-emitter voltage, VBE, of a diode-connected transistor, which varies inversely with its temperature. The rate at which this voltage varies is a very consistent –2 mV/°C, but the absolute value of VBE varies from transistor to transistor. Compensating for this variation involves comparing ?VBE at different values of IE.
Some silicon temperature sensors produce an analog voltage output (“VPTAT,” for “voltage proportional to absolute temperature”). Others convert VPTAT to a current output.
In digital-output temperature sensors, an amplified version of VBE from the sense transistor may be compared to a bandgap voltage reference and the result input to a S-? or successive-approximation register ADC to provide a digitized output. Precision may be 13 or 16 bits, with the least significant bit used as a sign bit.
An alternative digital output scheme uses pulse-width modulation (PWM), with temperature proportional to the ratio of ontime to off-time. Because on-time is fixed, these sensors can be used in an on-demand single-shot mode to minimize power consumption.
What kind of accuracy does a digital temperature sensor offer?
Accuracy varies with the temperature range of the part. For digital temperature sensors designed to read from 0°C to 70°C, it is possible to achieve ±0.5°C accuracy. With packaged parts designed for wider ranges, e.g., –55°C to 175°C, typical specs are ±1°C from 130°C to 150°C, or ±1.5°C from 150°C to 175°C for bare die.
The choice of packaging and temperature range depends on how the sensor is used. Underthe- hood automotive applications favor a wider temperature range and bare die. Medical, consumer, and instrumentation applications may only have to deal with temperatures from 0°C to 70°C, and ease-of-assembly considerations may make it more attractive to perform the extra heat-flow calculations required for converting measured junction temperature to surface-mount package temperature or metal-can temperature.
What flexibility does a digital temperature sensor offer?
Registers can be programmed for “high,” “low,” and “critical” temperatures (see the figure). The measured value is compared with these limits, and over events or under events trigger certain pins on the package. The measured temperature is communicated to the system controller via a serial bus that is also used for configuration and for loading those high, low, and critical registers.
Alternatively, digital temperature sensors work essentially like thermostats, toggling their output state (open-drain or push-pull) when a factory-preset trigger point is crossed. Trip points run from 35°C to 115°C in 10° increments. Hysteresis can be set via an external pin.