Temperature Sensor Types: RTD vs PRT vs Thermocouple vs Thermistor vs Infrared

The main temperature sensor types are resistance thermometers (RTDs and their high-accuracy platinum subset, PRTs), thermocouples, thermistors and non-contact infrared or radiation thermometers, and they differ mostly in accuracy, temperature range, response speed and how they drift: PRTs and RTDs give the best accuracy and stability over a moderate range, thermocouples cover the widest and hottest range but drift more, thermistors are extremely sensitive over a narrow band, and infrared reads surfaces without touching them but depends entirely on emissivity. Choosing the right one is only half the job. Whichever sensor a Singapore QC lab, semiconductor fab, cold-chain operator or test house relies on, its reading is only trustworthy while its calibration is current and traceable, because every one of these sensor types drifts in its own way. This guide explains how each type works, where it fits, how it fails, and how it is calibrated to a traceable, audit-ready standard.

The two families: contact and non-contact

Every temperature sensor falls into one of two families. Contact sensors (RTDs, PRTs, thermocouples and thermistors) must physically touch the thing being measured and reach thermal equilibrium with it. Non-contact sensors (infrared and radiation thermometers) read the thermal radiation a surface emits from a distance. The two families are calibrated in completely different ways, against different references, and Unitest holds separate SAC-SINGLAS accredited scopes for each: UNI-T001 for contact temperature sensors and UNI-T008 for non-contact infrared and radiation thermometers, both under accreditation number LA-2023-0845-C. Knowing which family your instrument belongs to is the first step in scoping its calibration.

RTD: the resistance thermometer workhorse

A resistance temperature detector, or RTD, works on a simple, stable physical principle: the electrical resistance of a metal changes in a predictable way as its temperature changes. Most industrial RTDs use platinum with a nominal resistance of 100 ohms at 0 degrees Celsius, which is why you often see them labelled PT100. Push the temperature up and the resistance rises along a well-characterised curve; measure the resistance accurately and you can infer the temperature.

RTDs are the workhorse of regulated industry because they are accurate, repeatable and stable over months of use. They suit the moderate ranges that dominate pharmaceutical, biologics and food processing: fridges, freezers, autoclaves, water baths and process lines. Their limitations are a slower response than a thin thermocouple and a ceiling on temperature, since platinum RTDs are not used at the extreme highs that thermocouples reach. RTD resistance still shifts gradually with repeated thermal cycling, mechanical shock and contamination, which is exactly why calibration is not optional. Unitest calibrates contact RTDs under its accredited scope, and the method is covered in depth in our guide on how to calibrate an RTD or PRT probe.

PRT: the platinum precision subset

A platinum resistance thermometer, or PRT, is the high-accuracy end of the RTD family. The word "platinum" is doing real work here: a PRT is built and characterised to a far tighter specification than a general industrial RTD, and standard platinum resistance thermometers (SPRTs) sit at the very top as reference-grade instruments used to realise the International Temperature Scale of 1990 (ITS-90). If an RTD is the reliable saloon car of temperature measurement, a PRT is the same idea engineered to laboratory-reference tolerances.

PRTs are what a calibration laboratory uses as its own working and reference standards, and what a metrology-conscious manufacturer buys when a general RTD is not accurate enough. Because they are precision instruments, they need careful handling and regular calibration: a dropped or mishandled PRT can shift its characteristic and quietly lose the very accuracy it was bought for. Unitest calibrates both RTDs and PRTs across a laboratory range of -80 to 660 degrees Celsius (and on site down to -95 degrees Celsius), with best measurement uncertainties as low as 0.01 degrees Celsius, all under SAC-SINGLAS accreditation UNI-T001. See the RTD probe calibration and PRT probe calibration pages for the accredited detail.

Thermocouple: widest range, most drift

A thermocouple works on a different principle again. Join two dissimilar metals and the junction produces a tiny voltage that varies with temperature. Different metal pairs (the familiar types K, J, T, N, R, S and others) suit different ranges, and collectively thermocouples cover the widest span of any sensor type, from cryogenic lows up to well over a thousand degrees Celsius where furnaces, kilns and heat-treatment live. They are also cheap, rugged, and fast to respond because the junction can be made very small.

The trade-off is accuracy and stability. Thermocouples drift more than RTDs, especially at high temperatures where the metals oxidise and their composition slowly changes, and their signal is small enough that lead wires, connectors and the reference (cold) junction all introduce error if not handled correctly. A thermocouple that read correctly last year can be several degrees out after a hard season in a furnace. For any thermocouple whose reading feeds a batch record, a process control loop or a safety interlock, periodic calibration against a traceable reference is what keeps it honest.

Thermistor: high sensitivity, narrow band

A thermistor is a semiconductor whose resistance changes very sharply with temperature. The most common kind, an NTC (negative temperature coefficient) thermistor, drops in resistance as it warms; a PTC (positive temperature coefficient) type does the opposite. That steep response makes thermistors extremely sensitive and able to resolve tiny changes over a narrow band, which is why they turn up in medical devices, laboratory instruments and precise cold-chain monitors where small deviations matter.

The catch is that a thermistor's response is strongly non-linear and its useful range is narrow compared with an RTD or thermocouple, and its characteristic can shift with temperature history and material ageing. Here the honesty point matters, and Singapore auditors care about it: at Unitest, thermistor calibration is a traceable calibration by comparison against calibrated reference standards, and it is not on our SAC-SINGLAS accredited schedule. It is genuinely traceable to national standards through the references used, but it must never be described as an accredited calibration, because it is not. We explain that distinction, and why traceability still holds up, in our article on thermistor calibration and why traceability matters and on the thermistor calibration service page.

Infrared and radiation thermometers: reading heat from a distance

Non-contact thermometers do not touch anything. They collect the infrared radiation a surface emits and convert it to a temperature. That is invaluable when the target is moving, electrically live, too hot to touch, or simply out of reach: a spinning bearing, an energised switchboard, a moving production web, a furnace interior. Handheld infrared "guns" dominate facilities, food and HVAC spot checks, while fixed radiation thermometers and pyrometers handle industrial process control at high temperatures.

The convenience hides a real trap. A non-contact reading depends on the target's emissivity (how efficiently that particular surface radiates), on the distance-to-spot ratio (whether the instrument is actually seeing only the target and nothing around it), and on the spectral band the detector uses. Get the emissivity setting wrong and the reading can be off by several degrees while looking perfectly plausible. Unitest calibrates infrared and radiation thermometers under accredited scope UNI-T008, across 35 to 500 degrees Celsius, at emissivity 0.90 to 1.00 and a spectral band of 8 to 14 micrometres, with the calibration referenced to traceable blackbody sources. The infrared thermometer calibration and radiation thermometer calibration pages set out that scope, and our article on emissivity, blackbody and accuracy unpacks why non-contact is not the same as no-calibration.

How the types compare at a glance

• Accuracy and stability: PRT best, then RTD, then thermistor (within its narrow band), then thermocouple. Infrared depends heavily on emissivity control.
• Temperature range: thermocouples widest and hottest, RTDs and PRTs moderate, thermistors narrow, infrared set by the instrument and target.
• Response speed: thin thermocouples and small thermistors fastest, sheathed RTDs and PRTs slower, infrared effectively instant on the surface it sees.
• Main drift mechanism: RTD and PRT from thermal cycling and shock, thermocouple from oxidation at heat, thermistor from ageing, infrared from optics, contamination and emissivity error.
• Calibration at Unitest: RTD, PRT, infrared and radiation are SAC-SINGLAS accredited; thermistor is a traceable calibration by comparison, not accredited.

Why the sensor type changes how you calibrate

The reason all this matters to a quality manager is that each sensor type drifts differently, so each needs a calibration approach and an interval suited to its failure mode. A furnace thermocouple oxidising at 900 degrees Celsius needs watching far more closely than a stable PRT in a 4 degrees Celsius fridge. An infrared gun that has been knocked around a plant needs its optics and emissivity response checked, not just a single-point comparison. A thermistor in a critical medical monitor needs a traceable characterisation across its narrow working band. Treating every sensor as if it drifts the same way is how a facility ends up with a certificate that looks fine and a measurement that is quietly wrong. We cover how to turn these differences into sensible calibration intervals in how often to calibrate temperature sensors.

The traceability chain is what an auditor actually checks

Whichever sensor you use, the value of its calibration is the unbroken chain that links it back to national and international standards. For contact sensors, that chain runs through fixed points defined by ITS-90, such as the triple point of water, realised in reference cells and transferred through calibrated PRTs. For non-contact instruments, it runs through traceable blackbody radiation sources. When a pharmaceutical, semiconductor or food-safety auditor picks up a sensor's certificate, they are checking exactly this: is the calibration traceable, is it accredited where it claims to be, does it state an uncertainty, and does the calibrated range actually cover how the sensor is used? A sensor without that chain is, for compliance purposes, an unknown reading with a number printed on it.

Match the sensor, then keep it honest

Pick the sensor type that fits your range, accuracy and response needs, then commit to keeping it traceable, because the best sensor in the world is worthless once it has silently drifted out of tolerance. If you are not sure which of your temperature instruments are in scope for accredited calibration and which are traceable by comparison, we will tell you plainly. Send us your sensor list and required ranges and we will confirm what is covered under our SAC-SINGLAS scope and return a clear quote, with no obligation. You can reach our Singapore laboratory through the contact page.