Calibration Drift in a Protective Device, How to Catch It

Why this matters

A protective device does not fail all at once, usually. It drifts. A relief valve that should open at a set pressure starts opening a little early, a thermal overload that should trip at a set temperature starts tripping a little sooner, every month, every season, every thousand cycles. The device still "works," it just no longer works at the number stamped on it. A tech who only asks "did it trip, yes or no" misses drift entirely, and either replaces a good device for nuisance trips it did not cause, or leaves a drifted device in place past the point where it protects anything at all.

What drift actually is

Calibration drift is the gap between a device's stamped or designed trip point and its actual trip point, growing slowly over the device's service life. It is different from a device being wired wrong, sized wrong, or outright failed. A drifted device is still functional, still moving, still responding to the parameter it monitors, it has simply lost the precision of exactly where it responds.

Drift has two directions, and they are opposite problems:

• Drifting more sensitive (trips early): produces nuisance trips on legitimate load. Annoying, costly in callbacks, but safe.
• Drifting less sensitive (trips late or not at all): the device stops protecting at its rated threshold. This is the dangerous direction, because the system runs past the point it should have shut down, and nobody notices until something downstream fails.

Common causes of drift, by generic mechanism

• Mechanical fatigue. Springs, bimetal strips, and diaphragms lose precision with every cycle. A device rated for a service life measured in cycles, not years, can drift well before a calendar-based replacement schedule would catch it.
• Corrosion and contamination. Corrosion on contacts increases resistance and changes the effective trip current on an electrical device; scale or debris on a mechanical seat changes the pressure or flow needed to actuate it.
• Thermal cycling. Repeated heating and cooling stresses the same mechanical elements that drift from fatigue, and accelerates it in devices that see frequent on/off cycling rather than steady running.
• Age and dwell time. A device that sits unused for long periods, then is expected to actuate precisely, can drift from lubricant breakdown, spring set, or seal hardening, even with zero operating cycles.

How to actually test for drift

Drift is only caught by comparing actual behavior against the rated threshold, not by watching whether the device trips at all.

1. Get the rated setpoint from the nameplate or manufacturer documentation. Do not estimate it.
2. Reproduce the condition under controlled, instrumented load (see the load-testing decision tree for the method) and record the exact reading, current, pressure, or temperature, at the moment of actuation.
3. Compare the actual actuation point against the rated setpoint. A device consistently actuating within the manufacturer's stated tolerance band is functioning correctly, even if it feels early or late to the customer. A device consistently actuating outside that tolerance band has drifted.
4. Repeat the test. A single reading could be a measurement error or a momentary condition. Drift is a repeatable, consistent offset from the rated point, not a one-time anomaly.
5. Check the direction. Note whether it is tripping early (sensitive drift) or late (desensitized drift), because the two call for very different urgency, as covered above.

What to do with a confirmed drift

• Drifted more sensitive, within a modest margin: document the reading, discuss with the customer, and monitor or replace on a normal schedule. The system is still protected, just prone to nuisance trips.
• Drifted less sensitive, or drifted past the manufacturer's tolerance band in either direction: treat as a safety-relevant finding. A device that no longer protects at its rated point should be replaced, not adjusted in the field, unless the manufacturer specifically documents an approved field recalibration procedure for that device. Most protective devices are not designed to be field-recalibrated; they are designed to be replaced when they drift.
• Never adjust a protective device's setpoint in the field to compensate for a customer's nuisance-trip complaint unless the manufacturer's documentation explicitly supports a field adjustment for that model. An undocumented field tweak is how a device stops protecting at the point it was engineered to protect.

The mental model to keep

A protective device that trips is not automatically working correctly, and a protective device that has not tripped yet is not automatically fine. The only way to know is to measure the actual actuation point against the rated one, under real load, and compare. Trip or no trip tells you the device moved. The measured setpoint tells you whether it moved at the right time.

References

• Manufacturer documentation for rated setpoints, tolerance bands, and approved field-calibration procedures
• Trade-standard practice for periodic testing of relief valves, thermal overloads, and circuit protective devices
• See related: The Device Trips Under Load But Tests Fine at Rest, Decision Tree; Why a Protective Device Doing Its Job Correctly Can Still Feel Like the Problem