Start with the process requirements, not the sensor name

Neither a Pt100 nor a thermocouple is the best choice for every application. A Pt100 uses temperature-dependent resistance, while a thermocouple uses thermoelectric voltage. A suitable Pt100 is a strong candidate where stability and low uncertainty matter at low or moderate temperatures; a suitable thermocouple may be preferable for very high temperatures or a very small sensing tip.

A catalogue accuracy figure alone is insufficient. Consider process temperature, atmosphere, response time, cable length, input module, installation and maintenance together.

Resistance measurement and voltage measurement are different

A Pt100 is a platinum RTD with a nominal resistance of 100 Ω at 0 °C. The instrument applies a small measuring current, calculates resistance from voltage and converts it to temperature. A Pt1000 has a nominal 1000 Ω at the same temperature and requires suitable input configuration and excitation.

A thermocouple's two different conductors produce a small EMF within a temperature gradient through the Seebeck effect. Its measuring principle needs no external excitation current, although the connected PLC or transmitter still needs power. RTD and thermocouple inputs are therefore not directly interchangeable.

The illustration shows the measuring principles; the Pt100 side is a simplified two-wire connection. Thermocouple conductor colours distinguish A from B and do not represent a standard cable colour code.

Temperature range: distinguish the standard curve from product ratings

“A Pt100 only works up to 300 °C” is not a valid general limit. IEC 60751 defines platinum resistance–temperature behaviour; the common standard characteristic is considered over approximately −200 °C to +850 °C. This does not mean that every Pt100 probe covers that entire range or maintains the same tolerance class throughout it.

The thin-film or wire-wound element, sheath, internal connections, filling and cable determine actual product limits. An element may withstand a higher temperature than its connecting cable. Check the temperature range over which the specified tolerance class remains valid.

Thermocouple limits depend on the selected thermoelement type and probe construction. Type K is common in general industry; precious-metal types such as R, S and B are used at high temperatures, while other suitable types serve different atmospheres. The upper temperature in a reference table is not the probe's continuous operating limit.

Accuracy, linearity and stability

In suitable low- and moderate-temperature applications, a quality Pt100 can provide tighter sensor tolerances and good stability. Lead resistance, measuring current and input electronics still need attention. A Pt100 is not perfectly linear; precision conversion uses the Callendar–Van Dusen relationship or appropriate tables.

Thermocouple characteristics are also nonlinear. IEC 60584-1 specifies reference functions and tolerances for the different types. A calibrated thermocouple's performance cannot be judged by its generic type name alone. For both technologies, calibration, repeatability, tolerance and long-term stability are distinct concepts.

At high temperatures, thermocouple oxidation, contamination and loss of alloy homogeneity can cause drift. A Pt100 can also change through mechanical strain, thermal cycling and construction-related effects. An electrically intact sensor is not proof of an accurate reading.

Response time is not determined by the sensor name

A very fine exposed thermocouple can respond quickly; so can a small thin-film Pt100. Probe diameter, sheath thickness, element mass, thermal contact, flow velocity and thermowell determine the overall response. Compare catalogue figures only when the test medium and response criterion are comparable.

With similar construction, a grounded thermocouple normally responds faster than an insulated junction. However, choosing a grounded, ungrounded or exposed junction also changes electrical isolation. “Grounded” identifies connection to the sheath and does not itself establish a PE connection. Insulation between a Pt100 element and its sheath is a separate design requirement; “RTDs cannot be grounded” is not a valid general statement.

For an approximately first-order response, one time constant, τ, is the time to reach about 63.2% of a temperature step. Approximately 95.0% is reached at 3τ and 99.3% at 5τ. Five time constants are not exactly 100%. A real probe may not follow a single time constant under every condition.

Vibration and mechanical robustness

Mineral-insulated thermocouples can be made as small, robust probes; IEC 61515 addresses requirements for these thermocouple cables and assemblies. Industrial Pt100 probes can also be robust with suitable sheaths and element support. Compare actual product mechanical ratings for vibration, shock, bending and mounting conditions.

What changes with two-, three- and four-wire Pt100 connections?

In two-wire measurement, both lead resistances add to the sensor resistance. Near 0 °C, Pt100 sensitivity is approximately 0.39 Ω/°C. A total additional resistance of 1 Ω across both leads can introduce roughly 2.6 °C of error in this region. This example concerns total lead resistance, not the resistance of just one lead.

A three-wire connection reduces lead effects using sufficiently matched relevant lead resistances and a suitable input circuit; mismatches can leave residual error. A four-wire connection separates current and voltage sensing paths, greatly reducing lead-resistance effects. Neither arrangement removes the sensor's own tolerance or every electronic error.

Thermocouples need suitable cable and reference compensation

For a thermocouple, thermoelectric material and polarity matter as well as cable resistance. Use suitable extension or compensating cable. These cable categories are not synonyms; preserve the specified type and temperature limits.

Transitions to copper form the reference connection region. An additional error can arise when this region is at a different temperature from the instrument's represented terminal temperature. The instrument accounts for reference temperature through cold junction compensation. A Pt100 measures resistance and does not require thermocouple-style CJC.

Pt100 self-heating: more current is not always better

RTD measuring current dissipates P = I²R in the element. Through 100 Ω, a 1 mA current produces about 0.1 mW; 2 mA produces 0.4 mW. Doubling current quadruples the heating power.

The resulting temperature rise depends on heat removal by the medium and on element construction; power alone does not specify an error in °C. Self-heating in air may differ from that in flowing liquid. Thermocouple measurement requires no excitation current, but input leakage and diagnostic currents can still introduce separate errors.

PLC inputs, long distances and electrical interference

Use a suitable RTD input for a Pt100 and a suitable thermocouple input on the PLC. A universal module also needs correct wiring and sensor configuration. A raw sensor is not a direct replacement for a standard 0–10 V signal. A suitable temperature transmitter can convert either technology to 4–20 mA or 0–10 V, which the PLC then scales through a standard analog input.

A VFD, motor cable or switched power circuit can affect both measurement systems. Consider cable routing, shielding, input isolation and ground references together. A transmitter near the sensor can help over long distances; a 4–20 mA output does not guarantee galvanic isolation.

Tolerance, SLE and total uncertainty

SLE, Special Limits of Error, is used in ASTM-based thermocouple terminology for tighter initial tolerances. It does not automatically map directly to an IEC class. Read the standard, type, temperature range and specified limits together.

A sensor's ± error limit is not directly its contribution to the system's standard uncertainty. Arithmetic addition may be used for a worst-case limit; uncertainty evaluation considers distributions, sensitivity coefficients and correlations. For example, two uncorrelated standard uncertainty contributions already expressed in the same output unit can be combined as u_c = √(u₁² + u₂²). Putting all catalogue ± figures into this formula without appropriate conversion is also incorrect.

Wiring, CJC and drift can be prominent in thermocouple systems; lead imbalance and self-heating can be prominent with a Pt100. Both require consideration of mounting, electronics and calibration. Thermocouple system measurement uncertainty explains this distinction in more detail.

Choose according to the application

For stable process measurement with low uncertainty, a Pt100 is a good starting candidate if its actual temperature rating is suitable. At very high temperatures, evaluate an appropriate thermocouple type. For a small tip and rapid changes, compare actual probe specifications for both technologies; in high vibration, do not choose by sensor name alone.

A 0–300 °C range does not automatically mean Pt100, and a particular percentage error target does not automatically mean Type J. Compare total costs including wiring, input module, transmitter, calibration and maintenance. A simple base-metal thermocouple can be economical; the same generalization does not apply to platinum-rhodium types.

PTC and NTC sensors may be considered separately for narrow-range, threshold or protection applications. Non-contact temperature measurement may suit moving or inaccessible surfaces, but emissivity, field of view and reflected radiation errors require separate assessment. None of these technologies replaces all the others.

Related topics

How Pt100 sensors work
How thermocouples work
Cold junction compensation
Thermocouple junction styles