The Fault Tracks With the Weather: A Decision Tree
Why this matters
"It only acts up when the weather turns" is a real, useful clue that most techs waste, because "weather" is not one thing. It is temperature, humidity, dew point, wind, sun, and pressure, and each is a separate variable that fails a system a different way. Chase "bad weather" as a single cause and you flail. Decompose it into the one variable that actually matters and the fault gets specific fast.
Safety first: water plus power, and after a storm
Weather faults have a habit of putting water and electricity in the same place, which is a shock and fire hazard, not just a puzzle. If the trigger is rain, humidity, or flooding and the symptom is anywhere near energized equipment, de-energize the circuit at its source and verify dead with a meter you have just proven on a known live source before you probe. After a storm with lightning or a grid event, treat the equipment as possibly damaged: inspect protection and grounding before you re-energize anything. A wet, live fault is worth stopping the diagnosis for.
Start here: is it really weather, or something weather drags along?
Before decomposing weather, rule out coincidence. Weather correlates with a lot of other things: heating and cooling runtime climbs with temperature, occupancy changes with the season, humidity tracks rain which tracks storms which track power events. You need more than one instance, and you need non-failures. If it failed once, on one bad day, you have an anecdote, not a pattern. Collect several failures and confirm the system runs fine in the opposite weather before you trust the correlation.
Decompose the weather into one variable
| Variable | Mechanism | The tell |
|---|---|---|
| Temperature, hot | Expansion opens a connection, components drift, fluids thin, heat rejection fails | Fails on the hottest part of the day or after long runtime |
| Temperature, cold | Contraction opens a fit, fluids thicken, freeze blocks or cracks | Fails on cold mornings or in a freeze, clears when warm |
| Humidity / dew point | Condensation bridges contacts, insulation resistance drops | Fails on muggy days with no rain, worst near a cold surface |
| Solar radiant gain | One side or an attic heats far past the air temperature | Fails on sunny afternoons on the sun-facing side, fine at night at the same air temp |
| Wind | Pressure change, movement, wind-driven rain past a seal | Tracks the gusts, not the temperature |
| Precipitation | Intrusion, shorts, corrosion that conducts when wet | Fails during or after rain, hand off to intrusion tracing |
Solar gain is the one techs miss. Two days at the same air temperature behave differently if one is sunny, because radiant load is not air temperature. If a fault is worse in the afternoon sun and gone at night at the same reading, you are looking at radiant heat, not ambient.
Reproduce the variable indoors
Once you have named the variable, recreate it and watch the suspect. Warm a connection with a heat source for a hot-temperature fault, chill a component with cold spray for a cold one, raise local humidity for a moisture fault, or load the part while it is already hot. You do not need the sky's cooperation once you know which single variable to fake. Have your meter or your eyes on the suspect at the moment it lets go.
When you cannot summon the weather
Some weather will not arrive on the clear day you visit. Set up the next win: document the variable you suspect and the signature you found (corrosion at a damp point, a heat-discolored terminal, water tracks that run sideways), inspect for that mechanism, and give the customer a short list to log at the next failure (temperature, sun or shade, rain, wind, and time of day). If the diagnosis demands it, plan the return for the triggering weather.
Recap
- If water meets power or a storm hit, make it safe first: de-energize, verify dead, inspect protection.
- Confirm it is really weather, not a season-linked proxy, using several failures and some non-failures.
- Break "weather" into one measurable variable: hot, cold, humidity, solar gain, wind, or rain.
- Reproduce that one variable indoors and measure at the moment of failure.
References
- Trade-standard practice for environmental and intermittent-fault diagnosis
- NFPA 70E principles for de-energizing and verifying dead before work near moisture
- Manufacturer environmental ratings, sealing, and surge-protection specifications
- See related: Teasing Out an Environmental Trigger the Customer Dismissed; Finding the Hidden Trigger Behind an On-and-Off Fault