Two numbers, one of them wrong
This is a smaller-sounding fault than most on this index and it is one of the more consequential, because every decision you make about the box rests on the number you believe.
Settle it first. Put a calibrated thermometer in the middle of the box, away from the coil discharge and away from the door, and give it twenty minutes to stabilise. Compare. If they agree within a degree or so, your controller is honest and whatever else is happening is a refrigeration problem. If they do not, you have a measurement problem and everything you thought you knew about this box is unreliable.
Why sensors read wrong
Drift
Thermistors and RTDs change their characteristics slowly over years. A sensor that was accurate at installation can be several degrees out after a decade. Nothing has failed; it has simply aged, and without a periodic check against a reference nobody would ever know.
Mounted in the wrong air
The most common cause by far. A sensor in the evaporator discharge stream reads several degrees colder than the box. A sensor near the door reads warm every time it opens. A sensor buried behind stacked product reads a pocket of air that has nothing to do with the rest of the box. In every case the sensor is working perfectly and reporting a place that is not representative.
Corroded or loose connection
Temperature sensors work by resistance. Add resistance at a corroded terminal and the controller reads it as a different temperature — usually warmer, which keeps the system running longer than it should. Walk-ins are damp, and terminals in damp places corrode. This cause also produces erratic, jumping readings rather than a steady offset.
Damaged cable
Sensor leads get pinched by panels, chewed by pests, and cut during other work. Partial damage changes resistance; full damage usually makes the controller display an error or go to a default.
The dangerous direction
A controller reading warmer than reality makes the system run too long. The box gets too cold, product freezes, and the compressor works more than it needs to. Wasteful, visible, and not a safety issue.
A controller reading colder than reality is the one that matters. The system shuts off believing the box is at setpoint, and the box is actually at 45°F. Your log, filled in from the display, records a compliant box all week. Nothing looks wrong to anyone. You find out when product spoils early, or when an inspector takes their own reading.
This is precisely why Massachusetts food safety practice expects temperature verification with a calibrated thermometer rather than acceptance of equipment displays, and why the independent reading is the one worth logging.
Checking it yourself
- Calibrated thermometer in representative box air for twenty minutes. Compare to the display.
- Find where the sensor actually is. Walk the box and look. If it is in the discharge air or by the door, you have probably found your answer.
- Watch the display for ten minutes. Steady but wrong suggests drift or placement. Jumping around suggests a connection.
- Check whether anything has been moved, stacked or rebuilt in the box recently.
What a proper check looks like
We verify sensor accuracy against a calibrated reference rather than against the display, inspect and remake the sensor terminations rather than just looking at them, confirm the sensor sits in representative air and move it if it does not, and write the measured offset on the report. If a sensor is more than a degree or two out it gets replaced, because the cost of a sensor is trivial next to the cost of running a box on a number that is not true.
Questions we get asked
Which reading should I write in the log?
The independent calibrated thermometer, always, and that is also what an inspector will expect. A temperature log built on an uncalibrated controller display is not evidence of anything.
How often should the sensor be checked against a reference?
At least annually, and on any box where the log and the display have ever disagreed. It is a two-minute check with an ice bath or a calibrated reference and it is one of the cheapest pieces of assurance available.
The reading jumps around erratically. Is the sensor failing?
Often it is the connection rather than the sensor. Corroded or loose terminals in a damp environment produce exactly that erratic behaviour. It is worth checking the terminals before condemning the probe.