Works Cold, Fails When Hot: A Thermal Fault Tree

Why this matters

The hardest faults to catch are the ones that hide when you arrive. A unit that runs fine first thing in the morning and quits after an hour, then works again once it cools, is showing a thermal fault: a part whose electrical or mechanical behavior changes as it heats up. These calls get misdiagnosed constantly because the technician tests a cold unit, finds nothing, and leaves - and the customer calls back the next day. Knowing which components fail hot lets you target the test window and catch the fault in the act.

Start here: confirm it is actually thermal

Before chasing parts, prove the pattern. A true thermal fault has three traits:

• It works on a cold start.
• It fails after a predictable run time or once warmed.
• It recovers after it sits and cools, with no parts changed.

If resetting power makes it work again immediately even while still hot, you may have a latching safety or control reset, not a thermal component fault. If it fails on cold start too, it is not thermal - look elsewhere.

Safety note: a unit that fails hot is often failing because it is overheating. If you find scorching, smoke, or a hot enclosure, treat it as an overheating hazard and de-energize before probing. Do not run a smoking unit to "catch it failing."

Branch 1: it overheats and a safety shuts it down

The most common cause is not a tricky component at all - the unit is genuinely getting too hot and a thermal limit is protecting it.

• If a high-limit or overload trips after a run and resets when cool, the unit is overheating. Chase the heat, not the limit: blocked airflow, fouled cooling, failing fan, or overload. See the overheating decision tree.
• Replacing the limit switch here is a mistake - it is doing its job.

Branch 2: an electrical component changes with temperature

Several common parts shift resistance or lose capacity as they heat, and fail only when warm.

• Connections: metal expands when hot. A marginally loose connection can be tight and conductive cold, then expand open and lose contact hot. This is the classic "hot intermittent." Check terminals and lugs warm, with a meter or by a careful touch test after power-down.
• Capacitors: a weak start or run capacitor can hold just enough charge cold to start a motor, then fade as it heats, causing a hot no-start. Measure microfarads warm and compare to the nameplate.
• Semiconductors and control boards: transistors and relays can pass cold and fail conduction hot. Suspect these after connections and capacitors are cleared.
• Windings: an insulation weak spot can pass cold and short or open once thermal expansion bridges or breaks it. Resistance readings often look fine cold and drift hot.

Branch 3: a mechanical part binds when hot

• If a motor or pump runs cold and seizes or drags hot, suspect a bearing or clearance that closes up with thermal expansion, or lubricant that thins and stops protecting at temperature.
• If it gets harder to turn by hand once warm, the fault is mechanical expansion, not electrical.

How to actually catch it: test in the failure window

You cannot diagnose a hot fault on a cold unit. The method:

1. Let the unit run until it fails, or get as close to the failure point as is safe.
2. The instant it fails, take your readings - voltage at the load, current draw, and the temperature and continuity of suspect connections - before it cools.
3. Compare those hot readings to the cold baseline you took on arrival. The component that changed is your fault.

Carry an infrared thermometer to find the hot spot fast and a clamp meter to read current without breaking the circuit. The part that reads normal cold and abnormal hot is the one to replace.

Reading the pattern

• Trips a thermal safety after a run equals genuine overheating - fix the heat.
• One connection or capacitor reads fine cold, bad hot equals a thermal-intermittent component.
• Shaft turns free cold, binds hot equals mechanical expansion.

The discipline is patience: stay until it fails, then read it before it recovers.

References

• NFPA 70B, recommended practice for electrical equipment maintenance
• Manufacturer nameplate values for capacitor microfarads and rated current
• Trade-standard practice on testing intermittent faults at operating temperature
• See related: "Overheating Equipment Diagnosis Decision Tree" and "Loose Connection vs Failed Component"