The Tells That a Fault Is Condition-Dependent

Why this matters

A condition-dependent fault is one that only appears when some outside variable is in a certain range: hot enough, humid enough, loaded enough, moving enough. If you diagnose it as a constant fault, you will meter a healthy system for an hour, find nothing, and either declare no fault found or throw a part at it. Recognizing early that a fault is tied to a condition changes the whole visit, because now you are hunting the condition, not the component. Reading the tells is what tells you which kind of fault you are standing in front of before you waste the trip.

What condition-dependent actually means

The fault is real, but it is gated. Below the trigger, everything measures fine. Cross the trigger, the fault appears; drop back under it, the fault clears. The component is not broken in the pass-fail sense, it is marginal, and the condition is what pushes it over its own edge. That is why a static bench test or a quick cold check clears it every time. You are testing on the safe side of the gate.

The field key of tells

Any one of these should make you stop treating the fault as constant.

• Patterned intermittency, not random. It comes and goes, but the customer can attach it to something: afternoons, rain, when the second appliance runs, after an hour. Random has no anchor; patterned points straight at a variable.
• It tracks a measurable variable. The symptom worsens as temperature, humidity, load, or runtime climbs, and eases as they fall. If severity moves with a number you can read, that number is your suspect.
• It has a threshold, not a slope. Fine, fine, fine, then fully faulted past a point, then fine again below it. A hard trip point is the fingerprint of a marginal part meeting a condition, though a truly random dropout with no threshold argues instead for a loose connection or a failing joint, so confirm the trip point repeats.
• It self-recovers. No one fixes it and it comes back on its own. Constant faults do not heal; conditional ones appear to, because the condition passed.
• It only shows in transition. The fault fires at startup, at shutdown, at a mode change, at the moment of a state swap, and never during steady running. Transition stresses (inrush, thermal shock, pressure spike) are conditions too.

The usual suspects to test against

Once you believe it is conditional, you only have a short list of variables to check, and most systems live and die by a handful:

• Temperature. Cold-start-only or hot-soak-only faults. Expansion, contraction, and semiconductor drift all track heat.
• Humidity and moisture. Tracking, corrosion at a joint, a sensor reading damp air. Worse on wet or muggy days.
• Load. Fine unloaded, faults under demand. The classic reason a static reading passes; hand it to the static-versus-loaded article.
• Time and duty cycle. Needs to run long enough to heat-soak, fill, or accumulate. Hand it to the can't-run-long-enough article.
• Movement and vibration. A fault that comes with a bump, a door slam, a compressor start. A physical joint or a chafed conductor.
• Supply variation. Voltage, pressure, or flow that only sags at certain hours or under the neighbor's demand.

Pin the condition by changing one thing

Recreate the trigger deliberately and change one variable at a time. Warm the suspect area, add the load, run it to soak, flex the harness, wait for or simulate the wet condition. Safety branch first: if reproducing the condition means energizing, pressurizing, or heating a system with a known hazard, verify guards and shutoffs before you provoke it, and never recreate a condition that risks gas, arc, or scald to chase a comfort complaint. When the fault appears exactly as you cross the condition and clears as you back off, you have both the cause and the proof.

When you cannot make it happen

Some conditions will not cooperate inside a service window, the weather is wrong, the soak is too long, the load is the customer's real life. That is not failure, it is a different method. Set up captured evidence, ask the customer to record time and conditions when it happens, or schedule the return for when the condition is present. Do not manufacture a constant-fault diagnosis just to leave with an answer.

References

• Trade-standard practice for intermittent and condition-dependent fault diagnosis
• Manufacturer operating-range and environmental specifications
• See related: The Fault You Can't See Because You Can't Run It Long Enough; The Pattern That Only Shows Across Multiple Visits