You Fixed One Fault but the Symptom Remains: Decision Tree

Why this matters

You found a real fault, you fixed it, you verified the repair, and the customer says it is still doing the thing. This is the moment a job becomes a callback if you handle it wrong. The instinct is to doubt your own repair and start swapping parts. The discipline is to work it as a branch: confirm the first fix truly took, then decide whether you misattributed the cause or whether a second, independent fault is the one actually driving the symptom. This tree keeps you from unwinding good work while you find the rest of the story.

Start here: is the lingering symptom a safety symptom

Before you diagnose further, check what the remaining symptom is. If it is a gas smell, a carbon-monoxide reading, a shock, heat, or arcing that is still present after your repair, stop and make it safe first - de-energize, shut the fuel, take it out of service - and only then continue. A persistent safety symptom after a repair is a red flag that a live hazard is still in the system, and it is never something to keep energized while you think.

Step 1: prove the first fix actually worked in isolation

Do not assume your repair took just because you did it. Test the thing you fixed on its own terms.

• If the repaired component now passes its own test (the new capacitor reads to spec, the reseated connection is tight and cool, the cleared drain flows), your fix is good. Move to Step 2.
• If it does not pass, your fix did not take, or you fixed the wrong instance of the right kind of fault. Redo or re-verify before you go looking for a second cause. Chasing a phantom second fault when the first one is not actually resolved is how a one-hour job turns into an afternoon.

Step 2: read how much the symptom changed

The size of the change points to the shape of what is left.

If the symptom is unchanged - your fault was real but it was not the cause of this symptom. You fixed something that needed fixing and misattributed it. Set it aside as a legitimate find and diagnose the symptom fresh, as if you had just arrived, without anchoring on the system you already touched.

If the symptom is better but not gone - this is the additive case, and it is the most common. A second fault is contributing the remainder. Measure the residual against spec, then hunt the second contributor in the same signal path (a filter cleared but a belt still slips, a valve opened but a line still restricts). Do not stop at "improved"; the customer called about the whole symptom.

If the symptom went away and came back - suspect an intermittent second fault your repair briefly masked, or a first fix that is already degrading. Try to catch it in the act; an intermittent needs conditions reproduced, not a static test.

Step 3: separate the two faults deliberately

Once you suspect a second fault, do not shotgun. Isolate.

• Change or disable one variable at a time and watch the symptom move.
• If fixing the first fault moved the number partway and fixing a suspected second closes the gap, you have confirmed two independent contributors, not one.
• If nothing you touch moves the residual, widen the search; the real driver is somewhere you have not looked, not in the parts you keep swapping.

Step 4: verify against the original complaint, not your repair

Close the loop on what the customer actually reported. Reproduce the original conditions - the same load, the same cycle, the same time of day if it mattered - and confirm the symptom is gone, by measurement, not by "seems better." A repair verified only against your own work, and not against the complaint, is how the same call comes back next week.

Recap

1. If the remaining symptom is a hazard, make it safe before diagnosing further.
2. Prove the first fix passes its own test before blaming a second fault.
3. Unchanged means you misattributed; better-but-not-gone means an additive second fault; gone-then-back means intermittent.
4. Isolate one variable at a time to confirm two independent causes.
5. Verify against the original complaint conditions, by the numbers.

References

• Trade-standard practice on post-repair verification and complete diagnosis
• Manufacturer troubleshooting documentation listing multiple contributors per symptom
• See related: When Two Separate Faults Are Hiding in One Symptom; Telling One Fault From Two That Look the Same