Check Recent Change vs Component First on a New-Onset Fault Decision Tree

Why this matters

When a system that worked fine suddenly fails, the fastest path to the cause is almost never the component itself; it is whatever changed just before the fault appeared. New-onset faults have a trigger, and the trigger is usually a recent event: a part was replaced, a setting was bumped, the power blinked, the season turned, someone "fixed" something. A tech who skips the change history and goes straight to component testing can chase a perfectly good part while the real cause (a wire knocked loose during unrelated work, a filter installed backward, a setpoint changed) sits in plain sight. This article gives a rule for when to investigate the recent change first versus when to test the component directly, applicable across every trade.

Symptom presentation

The defining feature is sudden onset on a previously working system. The customer can often name a "before" and "after": it worked Tuesday, it failed Wednesday. That timeline is the single most valuable piece of intake data on the call, because it brackets the cause to whatever happened in between.

Contrast this with a gradual degradation (a slowly clogging filter, a wearing bearing, creeping corrosion), which has no sharp trigger and points at the component's own wear. New-onset faults point outward at events; wear faults point inward at parts. Telling them apart at intake decides the whole approach.

Quick checks at intake

Pin down the onset. Ask "when did it last work normally" and "what is different now." A sharp before/after answer means recent-change investigation; a "it's been getting worse for months" answer means component wear.

Ask what happened around the onset. Recent service (yours or another contractor's), a storm or outage, a renovation, a new appliance added, a thermostat or breaker someone touched, a part the customer replaced themselves. Owners often forget to mention a change they do not connect to the fault.

Read your own service history. A fault that began right after a prior visit is a disturbed-during-service fault until proven otherwise, and that is a fast, cheap thing to check first.

Isolation tree

Branch A: a known change immediately preceded the fault. Investigate the change first, always. If a part was just replaced and the symptom started, suspect the replacement (wrong part, wrong setting, miswire, a connection left loose) before suspecting anything else. The base rate that "the thing someone just touched is the thing that broke" is extremely high.

Branch B: the fault began after unrelated work in the same area. Check whatever the prior work could have disturbed: a bumped wire, a kinked line, a closed valve, a knocked sensor, debris dropped into a mechanism. Disturbed-adjacent faults are common and cheap to find, and they precede component testing.

Branch C: the fault began after a power event (outage, surge, brownout, storm). Suspect control electronics, breakers, and anything with a memory or a reset state first. A board that lost its program, a GFCI that tripped, a contactor welded by an inrush, a setpoint reset to default all present as new-onset component failures but trace to the power event.

Branch D: the fault began with a season or environmental change. A unit that fails the first cold week, the first humid week, or the first run after months idle is responding to a condition it had not seen recently, not to a sudden component death. Check the condition-sensitive parts (charge, seals, condensation, expanded/contracted fits) before condemning the core component.

Branch E: a setting or mode changed. Someone adjusted a thermostat, a timer, a valve, a switch, a breaker. Verify all user-accessible settings before any component test. A "broken" system is often a changed setting, and that check costs a minute.

Branch F: no change can be identified despite careful questioning. Now component-first testing is justified, because you have ruled out the high-probability trigger classes. Proceed to the component isolation tree for the symptom, but keep the door open: an unremembered change is still more likely than a spontaneous failure on a young, healthy unit.

Branch G: the unit is old and at end of life. Component-first is reasonable even with a possible trigger, because aged components fail on their own and a coincidental change may be a red herring. Weigh the unit's age against the strength of the change-correlation.

Confirming diagnosis

Confirm by reversing or correcting the suspected change and seeing the symptom clear. If restoring a setting, reseating a connection, or correcting a reversed install fixes it, the change was the cause and you have a confirmed, low-cost diagnosis.

If correcting the change does not resolve the fault, the change and the fault may be coincidental, or the change caused a downstream component failure (a surge that actually killed a board). Move to component testing, but note in your diagnosis that the trigger and the failed part are linked.

Record the trigger on the work order. "Failed after panel work last week, found neutral loose at the device" tells the next tech and protects against the same disturbance recurring.

Next steps

When the recent change is the cause, correct it and verify the system holds. When the change caused a component failure, repair the component and, where possible, address what allowed the change to damage it (surge protection after a surge, strain relief after a disturbed wire).

When no change is found and component testing confirms a wear failure, treat it as a normal end-of-life or wear repair and set the customer's expectation accordingly. Do not invent a trigger that the evidence does not support.

References

• ISO 13379-1 (Condition monitoring and diagnostics of machines; failure cause analysis).
• NFPA 70B (Standard for Electrical Equipment Maintenance; failure-after-service and condition-based diagnostics).
• ASHRAE Handbook (HVAC Applications; seasonal and post-service fault diagnosis).
• OSHA 29 CFR 1910 (technician safety during inspection of recently disturbed systems).