2.3 Resistance Temperature Detectors (RTDs), Thermistors & Thermowell Installations

Key Takeaways

  • A Pt100 has a nominal resistance of 100 Ω at 0 °C. Select the matching resistance curve and wiring mode at the transmitter.

  • Two-wire readings include both lead resistances. A three-wire circuit can compensate matched leads. Four-wire sensing uses separate voltage and current paths.

  • An NTC thermistor’s resistance falls as temperature rises. Use the element’s own curve or coefficients and account for excitation and self-heating.

  • Check thermowell material, connection, immersion, probe fit and mechanical suitability for the process. Apply ASME PTC 19.3 TW within its scope.

Last updated: November 2026

Use the correct resistance curve

An industrial platinum resistance temperature detector (RTD) changes electrical resistance predictably with temperature. Platinum RTDs can use wire-wound or thin-film elements. Check the element’s tolerance class, applicable temperature range, packaging and vibration rating. The transmitter excites the device and interprets the resistance using a configured table or equation. IEC 60751:2022 specifies the industrial platinum relationship and requirements.

A Pt100 is nominally 100 Ω at 0 °C. A Pt1000 is nominally 1,000 Ω at 0 °C. The nominal IEC platinum curve has an average coefficient from 0 to 100 °C of about α=0.00385\alpha=0.00385 per degree Celsius, where

α=R100−R0100R0.\alpha=\frac{R_{100}-R_0}{100R_0}.

The Pt100 nominal resistance at 100 °C is about 138.51 Ω. Its average change over the interval is about 0.385 Ω/°C. For the same curve family a Pt1000 has ten times the resistance change per degree, so a given series lead resistance has a smaller effect on inferred temperature. For accurate conversion, use the nonlinear curve. With temperature TT in degrees Celsius, the Callendar–Van Dusen form is

R(T)=R0[1+AT+BT2](T≥0),R(T)=R_0[1+AT+BT^2]\quad (T\ge 0), R(T)=R0[1+AT+BT2+C(T−100)T3](T<0).R(T)=R_0[1+AT+BT^2+C(T-100)T^3]\quad (T<0).

For the IEC curve the numerical coefficients are A=3.9083×10−3A=3.9083\times10^{-3}, B=−5.775×10−7B=-5.775\times10^{-7} and C=−4.183×10−12C=-4.183\times10^{-12} when TT is entered in degrees Celsius. Texas Instruments’ RTD guide gives the coefficients and electrical circuits. Some transmitters support other curves or individually characterized sensors. If a 100 Ω nominal element actually gives 139.20 Ω at 100 °C, the IEC Pt100 input interprets it as about 101.8 °C. The measured resistance is higher than the configured curve expects at 100 °C. Match the sensor and configuration before adjusting calibration.

Excitation can warm the element by dissipating I2RI^2R. At 0.50 mA and 100 Ω, that is 25 µW. The resulting temperature error depends on the assembly and thermal conditions. Use the instrument’s specified excitation and the probe’s dissipation data when evaluating it.

Find which leads contribute to the reading

ConnectionMeasurement modelPractical check
Two-wireRread=RRTD+RL1+RL2R_{read}=R_{RTD}+R_{L1}+R_{L2}. The same leads carry excitation and convey the measured drop.Calculate or measure both leads’ total resistance and judge its contribution against the measurement tolerance.
Three-wireA suitable input uses paired lead paths or an extra measurement to compensate lead resistance.Follow the transmitter wire grouping and terminal drawing. Mismatched lead resistance and circuit errors can leave a residual.
Four-wire (Kelvin)Current leads excite the element, and separate high-impedance sense leads measure voltage at the element ends.Land the four leads in their assigned terminals. The excitation circuit must still operate within its limits.

In a two-wire check with 1.95 Ω measured total lead resistance, the indicated Pt100 temperature is approximately 5.1 °C high using the 0–100 °C average slope. A single 0.975 Ω conductor makes a pair of 1.95 Ω in that case. Use the actual curve slope near the operating point for a more accurate estimate. Two-wire wiring may be acceptable when the calculated error is within the specification.

The TI one-current-source three-wire circuit measures V1=I(RRTD+RL1)V_1=I(R_{RTD}+R_{L1}) and V2=IRL3V_2=IR_{L3} at different sense points. It computes (V1−V2)/I=RRTD+RL1−RL3(V_1-V_2)/I=R_{RTD}+R_{L1}-R_{L3}. With equal lead resistances the lead error cancels in the ideal model. If RL1=1.10R_{L1}=1.10 Ω and RL3=0.90R_{L3}=0.90 Ω, the residual is +0.20 Ω. Current-source mismatch can contribute in two-source designs. Check the appropriate circuit when troubleshooting.

                four-wire Kelvin connection

           I+ lead ──┐           ┌── I− lead
                     ●── [RTD] ──●
           V+ lead ──┘           └── V− lead

In a four-wire setup, the voltage input draws very little current, so sense-lead voltage drops are small. For the ideal connection, RRTD=Vsense/IR_{RTD}=V_{sense}/I. At 0.50 mA and 69.25 mV, the computed resistance is 138.50 Ω, close to a nominal Pt100 at 100 °C. Excitation accuracy, sensor tolerance, drift, bad connections and self-heating still matter.

Select a thermistor for its curve and operating conditions

An NTC thermistor has a negative temperature coefficient and can have a large change in resistance per degree over a selected interval. A PTC thermistor’s resistance rises over its working region. Some PTC types switch rapidly near a characteristic temperature, making them useful for equipment protection. Consult TDK’s PTC product information. For precise NTC temperature measurement, use the specified resistance table or a fit such as Steinhart–Hart,

1TK=a+bln⁡R+c(ln⁡R)3.\frac{1}{T_K}=a+b\ln R+c(\ln R)^3.

Here TKT_K is absolute temperature in kelvin and resistance is entered in the units used to fit aa, bb and cc. See TI’s temperature sensor applications guide. The fit, accuracy range and allowed power are device-specific. In sensor selection, compare the temperature and environment rating, tolerance, installed response time and transmitter input. Thermocouples give a small voltage, while RTDs and thermistors require a resistance measurement.

Evaluate the thermowell and probe as an assembly

A thermowell is a closed-end pocket installed into the process. A suitable well forms part of the pressure boundary and separates an insert from the fluid. When the approved design and procedure allow, the insert can be accessed with the pressure boundary intact. Treat changes to the thermowell or process connection as pressure-boundary work. Check compatibility, ratings, connection, integrity, probe fit and allowable operating conditions. A short or poorly fitted sensor may respond slowly or read toward the connection temperature. A spring-loaded probe made for a particular well can maintain tip contact. Match its bore, length and compression using the drawings. Rosemount 114C thermowell data and 214C probe data give a matching example.

Straight, tapered and stepped stems differ in stiffness, mass, response and drag. Greater immersion can improve thermal measurement while increasing unsupported length. Re-evaluate both requirements before changing the installation. The well and its fit influence lag.

Flow past the stem can produce alternating vortices and cyclic forces. For a simple illustrative estimate fs=StV/df_s=St V/d, where StSt is the assumed Strouhal number, VV is fluid velocity and dd is the representative diameter. With St=0.20St=0.20, V=4.0V=4.0 m/s and d=0.016d=0.016 m, the estimate is 50 Hz. In-line excitation can occur at twice the shedding frequency. Compare the operating conditions and geometry using the approved engineering calculation. A simple estimate does not evaluate the static and dynamic stress, pressure and fatigue criteria.

ASME PTC 19.3 TW-2016 (R2025) covers specified bar-stock thermowells. Its published scope excludes wells fabricated from pipe and special surface forms such as spiral ridges. The design evaluation includes frequency, dynamic stress, steady-state stress and pressure limits. Consider the operating envelope and the applicable checks for in-line and transverse excitation. If a candidate fails, engineering reassesses geometry, support, process conditions or a different design and verifies thermal suitability again. A change in shank form or unsupported length needs a new evaluation.

Test Your Knowledge

In the stated one-current-source, three-wire RTD circuit, the subtracted lead measurement represents 0.90 Ω, and the lead in the RTD measurement contributes 1.10 Ω. What is the residual resistance error?

A

−2.00 Ω.

B

0 Ω regardless of lead mismatch.

C

+0.20 Ω.

D

+2.00 Ω.

Test Your Knowledge

Which base resistance and average coefficient apply to a nominal Pt100 using the stated IEC 60751 curve?

A

100 Ω at 0 °C and about 0.00385 per °C.

B

100 Ω at 0 °C and about 0.00392 per °C.

C

1,000 Ω at 0 °C and about 0.00385 per °C.

D

50 Ω at 0 °C and about 0.00427 per °C.

Test Your Knowledge

A proposed bar-stock thermowell’s simplified shedding estimate is 50 Hz. What must be established before the approved design is accepted?

A

Any frequency lower than the transmitter sample rate is acceptable.

B

The estimate alone proves the well is safe if it is below the natural frequency.

C

The well only needs the same probe insertion length as the previous installation.

D

The applicable frequency, stress and pressure checks pass for the actual geometry and operating conditions.

Sections you finish are checked off in the contents.