Temperature sensor drift is a gradual change in a measurement channel’s output under comparable conditions. A reading that differs from an earlier baseline does not by itself prove that the sensing element has drifted: the reference, measurement location, wiring, input instrument, installation, or process may also have changed.
This guide separates sensor drift from initial accuracy, resolution, repeatability, and sudden failure, then outlines a practical approach for trending and calibration. It applies to RTDs, thermocouples, and NTC thermistors, with technology-specific effects depending on construction and environment.
Drift, accuracy, and repeatability are different
Accuracy describes closeness to a reference under stated conditions; repeatability describes how consistently a measurement can be repeated under the same conditions. Resolution is the smallest displayed or detected change. Drift concerns a change over time, which must be established by comparing measurements under sufficiently comparable conditions with a suitable reference and measurement method.
For example, an apparent offset can arise after moving a probe, changing a controller, replacing a cable, altering an RTD input configuration, or comparing different physical locations. Record the sensor, instrument, wiring, test point, installation, and reference conditions before attributing a change to the element.
What can contribute to long-term change?
- Thermal exposure: time at temperature and thermal cycling can affect an assembly; the mechanism and magnitude depend on element, materials, construction, and service conditions.
- Mechanical stress: vibration, bending, pressure, strain, or repeated installation can change contact, wiring, or element conditions.
- Moisture and contamination: damaged seals, cable transitions, or connectors can create leakage, corrosion, or intermittent behavior.
- Electrical and instrument changes: lead resistance, connection resistance, excitation, input configuration, reference-junction compensation, scaling, or electronics can change the indicated value.
- Measurement setup: different immersion, thermal contact, airflow, stabilization, or reference placement can produce a difference that looks like drift.
Technology-specific diagnostics matter. An RTD reading depends on its resistance curve, lead arrangement, and input circuit. A thermocouple system depends on the thermocouple type, wiring and polarity, reference-junction compensation, and instrument configuration. An NTC thermistor depends on its R/T curve and the controller’s conversion model. Use the documentation for the actual assembly rather than generic expected values.
How to confirm a suspected drift
- Record the observed change, when it occurs, sensor identity, operating history, and the maintenance or wiring changes made since the baseline.
- Recheck the physical measurement point, mounting, immersion or thermal contact, and process state.
- Inspect wiring, connector, terminal, insulation, grounding, and instrument configuration; verify sensor type, curve, and lead arrangement.
- Compare with a suitable reference at defined points using a repeatable setup. Allow stabilization and record location, conditions, equipment, and measurement uncertainty.
- Where needed, evaluate the sensor separately from the installed measurement channel to determine whether the change follows the probe, wiring, instrument, or installation.
Do not infer a failure from one comparison if the test conditions, reference uncertainty, or acceptance limits are unknown. If results remain outside the application’s defined limits, follow the applicable calibration, repair, or replacement process.
Build a useful calibration and trend plan
Set calibration intervals and acceptance limits from the process risk, required measurement uncertainty, usage, environment, prior results, and quality-system rules. There is no single interval that fits every sensor and application. Keep a baseline and record as-found and as-left results, reference method, points, conditions, instrument configuration, and any adjustment or replacement.
When trending a population, compare like-for-like sensor types, installations, instruments, and test points. Review repeated movement over multiple checks instead of treating one reading as a long-term trend. NIST describes relevant methods in its guidance on industrial thermometer calibration.
When drift is accompanied by failure symptoms
An abrupt open circuit, short circuit, intermittent signal, or damaged assembly is a different diagnostic problem from gradual change. For immediate reading problems, use our five-point temperature measurement troubleshooting checklist. For physical assembly failure modes, see the OEM failure-mode guide.
OEM selection and maintenance checklist
- define the measurement point, temperature profile, exposure, and required system uncertainty;
- specify the sensing technology, R/T curve or thermocouple type, wiring, materials, and instrument input;
- agree on calibration points, method, reference, acceptance limits, and records;
- document installation, baseline results, service history, and change control;
- review the complete channel when observed drift cannot be isolated to the probe.
For technology comparisons, see our RTD versus thermocouple selection guide and NTC calibration and curve verification guide. To discuss an OEM assembly or test plan, send the application details and drawing.
