The Part Tests Good but the System Still Fails (Decision Tree)

Why this matters

You suspect a part, you test it, it passes, and the system still fails. This is the exact moment techs start firing the parts cannon, swapping the "good" part anyway, then the next one, then the one after that, spending the customer's money to buy back the confidence a passing test should have given them. A good test result is not a dead end, it is a door you are supposed to close. The skill is knowing what a pass actually proved, deciding whether to trust it, and then redirecting the whole search instead of re-testing or replacing a part you already cleared.

Start here: what a passing test actually certifies

A test result is a conditional claim, not a character reference. It says one specific parameter was within range, at the specific condition you tested, at the moment you tested it. It does not say the part is good for the system's real duty, and it does not say the part is the right part for the job. Read your pass narrowly and honestly. "The winding measured in spec at room temperature, unloaded" is what you proved. Whether that part performs under the system's real heat and load is a separate question your bench check did not answer.

Was the pass real: the one thing that reopens the part

Before you clear the part for good, close the single legitimate loophole: a pass at the wrong condition proves nothing about a fault that only happens at the right one.

• Did you test under a condition representative of where the fault occurs? A part that passes at no load is not cleared of a load fault. A part that passes cold is not cleared of a heat fault.
• Did you test the parameter that would actually be failing? Continuity good does not clear a part whose failure mode is under voltage stress.

If the test condition did not match the fault condition, the part is not actually cleared. Re-test it correctly, under the representative condition, before moving on. This is the exception to everything below, and it is the only honest reason to keep suspecting a part that passed.

The rule: a verified good part comes off the suspect list

Once a part has passed a test that genuinely represents its service condition, it is done being a suspect, and swapping it is not diagnosis, it is guessing with the customer's wallet. Replacing components that already passed under representative conditions is the costliest dead end in the trade and rarely fixes the fault. Hold the line: a cleared part stays cleared unless new evidence, not new frustration, reopens it.

Redirect the search, do not re-test the part

A system that fails with all-good parts has a fault that lives between or around them, not inside them. Point the investigation outward.

• The inputs feeding the part: supply voltage, control signal, feed pressure, command sequence. A good part fed a bad input fails through no fault of its own.
• The load or output side: a binding, blocked, or seized load makes a healthy part look bad.
• The interfaces: connectors, terminals, fittings, linkages, and wiring between good parts, the classic system-fault location precisely because you cannot pull them and bench-test them.
• The interaction: timing, sequencing, and signaling between parts that are each fine alone.
• The spec fit: a genuine possibility that the part passes its own test yet is the wrong rating or tolerance for this duty (see the related substitute-part article).

For the full mechanics of chasing a system-level fault, see the related article on a component that tests good while the system fails. The discipline this article is about is the decision to leave the cleared part alone and go look there.

Record why it passed and where the fault lived

Close the loop so the next person does not repeat your bench test.

• Note the exact test, the condition, and the result, so the pass is documented, not just remembered.
• Note where the fault actually lived once you find it, and why the part read good despite it.
• That record explains the apparent contradiction to the customer and keeps the next tech from re-testing the same good part on the next visit.

The judgment to bank: a passing test is a limited, conditional truth. Verify it was taken at the fault's condition, then trust it and redirect. Do not swap a part you already proved good because the system is still broken and you are out of ideas.

Recap

1. A pass certifies one parameter, at one condition, at one moment, nothing broader.
2. The only honest reopener is a pass taken at a condition the fault does not occur in; re-test correctly, then trust it.
3. A part verified good under representative conditions comes off the suspect list; do not fire the parts cannon.
4. Redirect to inputs, load, interfaces, interaction, and spec fit, and record why the part passed so no one re-tests it.

References

• ISO 13379-1 on component-level versus system-level diagnosis
• Trade-standard practice on verified diagnosis versus guess-and-swap replacement
• See related: Component Tests Good but System Fails; The Substitute Part Didn't Fix It