How Sensors Fail

Why this matters

A sensor turns something physical (temperature, pressure, flow, position, level) into an electrical signal a control can read. When a sensor lies, the control acts on the lie. A drifting temperature sensor makes a system overheat or short-cycle. A stuck pressure sensor lets a board run blind. The maddening part is that a bad sensor often produces a plausible wrong number, not an obvious fault, so the machine misbehaves while every reading looks reasonable. Learning how sensors fail keeps you from replacing good parts chasing a sensor's ghost.

The signal chain, not just the sensor

A sensor reading travels through a chain: the sensing element, the wiring and connectors, and the control input that reads it. A fault anywhere in that chain looks like a sensor problem. Before you condemn a sensor, remember most "sensor" failures are actually connection failures. Corroded pins, a chafed wire, and a loose terminal mimic a dead or erratic sensor exactly.

The failure modes

• Drift. The element slowly reads off true. It still tracks change, but the absolute number is wrong by a growing margin. This is the sneakiest failure because the system keeps running on a number that is quietly incorrect. A temperature sensor reading several degrees off, or a pressure sensor reading consistently high or low, points to drift.
• Open circuit. The sensor or its wiring goes electrically open. The control sees this as the extreme end of the scale (maximum or minimum, depending on the type) and often throws a fault.
• Short circuit. The opposite. A short pins the reading to the other extreme. Many controls flag open and short conditions specifically because they are out of physical range.
• Stuck or frozen. The reading holds at one value no matter what the real condition does. The element fouled, corroded, or failed internally. The tell is a number that never moves when it should.
• Erratic or noisy. The reading jumps around. Loose connection, water intrusion, electrical interference, or a cracked element. This is the one that causes random faults and intermittent shutdowns.
• Slow response. The sensor lags reality because it is coated, scaled, or mounted poorly. The control overshoots because it is acting on old information.
• Fouling and coating. A film of scale, soot, grease, or mineral buildup insulates the element from what it is supposed to measure. Common on anything in water, exhaust, or greasy air.

How to test one

The principle: compare the sensor's signal against a known truth.

1. Compare to a reference. Put a trusted instrument on the same condition. Read the actual temperature, pressure, or position with a calibrated tool and compare it to what the sensor and control report. A gap is drift.
2. Read the raw signal. Many sensors output a known relationship: a resistance that maps to temperature, a voltage or current that maps to pressure. Measure the raw output and look it up against the expected curve for that reading. Out of range means a bad element or bad wiring.
3. Wiggle test. With the system running, gently flex the harness and connector at the sensor while watching the reading. A jump means a connection fault, not a bad sensor.
4. Substitute a known condition. Where safe, force a known input (warm the probe in your hand, apply a test pressure) and confirm the reading moves correctly and proportionally.

Warning signs in the field

• A system that short-cycles, overshoots, or trips for no physical reason.
• A reading that looks plausible but does not match a reference instrument.
• A fault code naming a sensor that clears when you reseat its connector (a wiring fault, not the sensor).
• A reading pinned at the top or bottom of the scale (open or short).
• A number that never changes when the condition clearly does (stuck).

What to do about it

Confirm with a reference before replacing. Check and clean the connector first; it is free and it is often the real fault. If the element is fouled, sometimes a cleaning restores it, but a drifted element gets replaced, not adjusted, unless the device is designed for field calibration. Match the replacement by type and range; a sensor with the wrong range or curve will read wrong out of the box. After replacing, verify the new sensor against a reference so you know the whole chain reads true.

References

• Manufacturer sensor specifications (output curve, range, accuracy)
• Trade-standard practice for comparison against calibrated reference instruments
• NIST traceability principles for measurement and calibration
• See related: reading temperature differential (generic); the baseline reading you should always take