How Do PT100 and RTD Temperature Sensors Work?

Detailed Explanation

Resistance Temperature Detectors measure temperature through the predictable relationship between temperature and the electrical resistance of a pure metal. Platinum is the standard material for precision RTDs because its resistance-temperature relationship is highly stable, well-characterised, and internationally standardised in IEC 60751.

The PT100 Standard

At 0°C, a PT100 has exactly 100 Ω. As temperature increases, resistance rises nearly linearly at a temperature coefficient of 0.00385 Ω/Ω/°C (the "alpha" value, derived from the IEC 60751 standard for platinum purity and annealing). In practical terms, resistance increases by 0.385 Ω for every 1°C rise in temperature.

Over the industrial range:

The response is not perfectly linear — there is a small quadratic correction at temperatures below 0°C, captured in the Callendar-Van Dusen equation used in IEC 60751. For the range 0°C to 850°C the linearity is good enough for most industrial applications without software correction; below 0°C, a lookup table or the Callendar-Van Dusen polynomial is required for accuracy better than ±0.5°C.

Wire Connection Methods

The single most important practical consideration in RTD measurement is how the sensor connects to the measurement electronics. The RTD's resistance is small, and connecting wire has its own resistance. A 1 Ω cable resistance error on a PT100 represents a 2.6°C measurement error — unacceptable for precision work.

2-wire connection: The excitation current path and the measurement voltage path share the same two wires. Cable resistance (R_lead × 2) adds directly to the measured resistance. Only acceptable for short cable runs (< 0.5 m) at low accuracy requirements. Simple and cheap — fine for oven control where ±2°C is good enough.

3-wire connection: One wire carries excitation current; the other two are used for voltage measurement, with the third wire introduced to compensate for cable resistance. The assumption is that all three leads have equal resistance — a reasonable assumption for cables with the same wire gauge and length. Compensation is built into most RTD transmitter ICs (MAX31865, ADS1247, etc.) and removes most of the cable resistance error. 3-wire is standard for industrial measurements where cable runs are long but perfect accuracy is not required.

4-wire (Kelvin) connection: Two wires carry the excitation current; two separate wires carry the voltage measurement. Because the voltage sense wires carry negligible current (high-impedance voltage input), their resistance causes negligible voltage drop and therefore zero cable resistance error. This is the correct choice for any measurement requiring better than ±0.1°C accuracy, or for long cable runs where lead resistance is large and asymmetric. Requires four-wire cable from sensor to measurement electronics, but the accuracy is fundamentally limited only by the quality of the RTD, the stability of the excitation current, and the resolution of the ADC.

Signal Conditioning Circuit

An RTD produces a small resistance change — a PT100 spanning 0–200°C produces a resistance change of 0 to 77 Ω. To resolve 0.1°C temperature steps, you need to resolve ~0.039 Ω differences. With 1 mA excitation current, that corresponds to a 39 µV voltage change — requiring a 12-bit or better ADC with a front-end amplifier.

Basic constant-current excitation:

• Drive a known current (e.g. 1 mA) through the RTD with a precision current source or a resistor from a stable supply voltage.
• Measure the voltage across the RTD (4-wire sense) with an instrumentation amplifier.
• Calculate temperature from V_measured / I_excitation = R_RTD, then look up temperature from the IEC 60751 table.

Ratiometric measurement: A more robust approach uses the same excitation current through both the RTD and a precision reference resistor. The ADC measures the ratio of the RTD voltage to the reference resistor voltage — this makes the result independent of excitation current variations. The MAX31865 RTD interface IC uses this approach: it includes the excitation source, the reference resistor, and a ratiometric 15-bit ADC in a single package, returning temperature directly over SPI. For a complete wiring, register, and firmware guide see using the MAX31865 with a PT100.

Wheatstone bridge: Balances the RTD resistance against a reference resistor, producing a differential output voltage that is zero at a reference temperature and proportional to resistance deviation. Suited for narrow-span measurements around a specific operating point (e.g. monitoring whether a process is within ±5°C of setpoint). An instrumentation amplifier amplifies the small bridge imbalance voltage for ADC measurement. See instrumentation amplifier basics for the amplifier stage.

For practical embedded implementations, the MAX31865 (or equivalent PT100/PT1000 interface ICs from Analog Devices) is the lowest-effort path: the IC handles all signal conditioning and returns calibrated resistance (and optionally temperature) over SPI. Custom discrete designs give you more control over update rate and power consumption but require careful attention to excitation current stability and reference resistor quality.

RTD vs Thermistor vs Thermocouple

RTDs suit high-accuracy, long-term stable measurements. For lower-cost, lower-accuracy applications below 150°C, an NTC thermistor (see thermistor basics) is often easier to interface. For temperatures above 600°C or where absolute accuracy matters less than thermal response speed, thermocouples are preferred.

For applications that need to interface multiple temperature sensors digitally without custom analog signal conditioning, see the digital temperature sensor interface page — dedicated 1-Wire and I2C temperature ICs (DS18B20, TMP117) are simpler to implement than RTDs for many use cases.

Design Considerations

• Excitation current and self-heating: Keep excitation current at or below 1 mA for PT100, 0.3 mA for PT1000. Verify self-heating against your accuracy requirement, especially in still-air environments where convective cooling is poor.
• Reference resistor quality: For discrete designs, the excitation current reference resistor must have a low temperature coefficient (≤ 25 ppm/°C) and high stability. A standard carbon-film resistor (±100–200 ppm/°C) will dominate the measurement error — use a metal-film resistor or a precision thin-film type.
• Guard against EMI on RTD leads: Long cable runs to an RTD act as antennas. Add an RC filter (10 Ω + 10 nF) at the measurement input to attenuate high-frequency interference before it reaches the ADC or instrumentation amplifier.

Common Mistakes

• Using 2-wire connection with long cable runs, then being surprised when readings are systematically offset from known reference temperatures — the offset is exactly 2 × R_lead × 0.385°C/Ω.
• Confusing PT100 (100 Ω/0°C) with NTC thermistors (typically 10 kΩ nominal) — the RTD's resistance is an order of magnitude lower, and the signal conditioning circuit is completely different.
• Over-driving the excitation current to get a larger signal, then introducing self-heating errors that overwhelm any gain in signal-to-noise ratio.
• Forgetting linearisation below 0°C — a simple linear approximation from the datasheet works above 0°C but introduces meaningful error in sub-zero applications.