Read the pattern, not the number
A single temperature reading tells you almost nothing about a swinging box. The shape of the curve over several hours tells you nearly everything. If you have a data logger or a controller with a history, pull it. If not, write the temperature on a sheet every fifteen minutes through one service period. It is tedious and it is worth it.
Three patterns and what each one means
A regular sawtooth, same amplitude every cycle
The system is cycling on and off normally but across too wide a band. This is a control setting: the differential — the gap between the temperature that starts the system and the one that stops it — is set too wide. On older mechanical thermostats this drifts as the bellows ages. On electronic controllers it is a setting someone changed, often to stop short cycling, and overcorrected.
Sharp spikes at even intervals, four to six times a day
Defrost. Every walk-in freezer and many coolers defrost on a schedule. A normal defrost causes a modest rise that recovers quickly. A spike that is large, or that takes an hour to come back down, means the cycle is running to its time-out limit instead of terminating when the coil is clear — usually a failed termination sensor, sometimes a defrost heater that has partially failed so the coil takes far longer to clear.
Irregular jumps with no clock pattern
Something physical and human. Door openings during deliveries, product loaded warm, a door left on the latch during a long prep. Match the timestamps against what the kitchen was doing. This is not a refrigeration fault and no technician can fix it, but the fix is usually free.
The sensor-in-the-wrong-place problem
This is worth its own section because it is so common and so invisible. A control sensor has to read representative box air. We find them taped to a wall next to the door, sitting directly in the evaporator discharge stream, buried behind stacked product, or dangling where someone moved it during a clean and never put it back.
A sensor in the discharge air sees cold before the box does, so the system shuts off early and the box drifts. A sensor near the door sees every opening and short cycles the system all day. A sensor buried in product reads the product, not the air, and responds hours late. None of these are broken parts, and all of them produce a swinging box.
Why swing matters more than operators think
A box averaging 38°F sounds fine on a daily report. A box that swings between 33°F and 45°F averages 39°F and spends part of every cycle out of compliance, with the product surface riding those swings. Health inspectors in Massachusetts are entitled to take the reading they find, not your average. Shelf life suffers too — produce and dairy degrade on the peaks, not the mean.
What we do on this call
We log the box rather than guess at it, check where the control sensor actually sits and what air it is reading, verify the defrost interval and termination method against what that coil needs, and set the differential to something the equipment can actually hold. Often there is no part to replace at all, which makes this one of the cheaper faults to resolve — and one of the more satisfying, because a box that has swung for two years can usually be made to sit flat in a single visit.
Questions we get asked
How much swing is normal?
Two to three degrees either side of setpoint on a cooler is normal and healthy. Six degrees or more is a fault. A brief rise during defrost is expected, but it should recover within thirty minutes and should not take product above 41F.
My logger shows a spike at the same times every day. What is that?
Almost certainly defrost. Count the spikes: four a day means a six-hour defrost interval. If the spike is large or slow to recover, the cycle is running longer than it needs to or the termination sensor has failed and it is defrosting on time-out instead.
Could it just be the door?
It can, and the pattern gives it away. Door-driven swings are irregular and line up with service times, deliveries and shift changes rather than with a clock.