The Diagnosis That Doesn't Add Up: A Decision Tree

Why this matters

Sometimes you've done everything right and the answer still doesn't make sense. The readings contradict each other. The part you condemned tests fine on the bench. The symptom doesn't match any failure you know. That uneasy feeling is information. The worst move is to force a conclusion just to close the call. A diagnosis that doesn't add up is usually telling you that one of your assumptions is wrong, and this tree helps you find which one.

Start here: stop and name the contradiction

When the picture won't resolve, don't push harder in the same direction. Stop and state, in one plain sentence, what doesn't fit. "The motor draws normal current but won't turn." "There's voltage at the device but it won't energize." "The leak is here but the pressure drop is somewhere else." Naming the contradiction out loud forces it into the open and points at the assumption to test next.

Branch 1: Is the test itself wrong?

Most "impossible" readings come from a measurement error, not a haunted machine. Check this first.

• Wrong point. Are you measuring exactly where you think, or one connection over? A reading taken on the wrong side of a switch or splice tells a true story about the wrong thing.
• Wrong tool or mode. Right meter, right setting, right range? Confirm it on a known-good reference before you trust a surprising number.
• Bad lead or connection. A loose probe, a corroded test point, or a poor ground gives readings that look like component faults.
• System not in the failing state. Are you measuring under the condition that fails, or at idle? A reading taken when the fault isn't active is just noise.

If the test is sound and the contradiction survives, move on.

Branch 2: Is a hidden component in the path?

The schematic in your head may be missing something the real system has. Contradictions love hidden elements.

• A safety device, interlock, or limit switch you didn't account for is open and breaking the circuit or flow.
• A second control, relay, or valve in series is doing something you didn't expect.
• A previous repair changed the system. Someone jumpered a switch, swapped a part with the wrong rating, or added a workaround.
• The "obvious" supply isn't reaching where you assumed, because of a break, blockage, or shutoff upstream.

Trace the full path end to end, in the real equipment, not from memory. Hidden components hide in exactly the spots you skim past.

Branch 3: Is your model of the failure wrong?

If the test is clean and the path is fully traced, the problem may be your mental model.

• You're looking at a victim, not a cause. The part you condemned failed because something upstream stressed it. It tests fine on the bench because the stressor isn't there.
• Two faults at once. Rare, but real. Two small problems can produce a symptom that matches no single failure. Split them: fix the one you're sure of, then re-evaluate.
• The symptom is normal. Sometimes the "fault" is the system working as designed and the customer's expectation is off. Verify against spec before you chase it.

Decision summary

The discipline: change one thing, re-test

When a diagnosis won't resolve, resist the urge to swap several parts and hope. Change one variable, re-measure, and watch what moves. A contradiction that survives a clean test, a full path trace, and a corrected mental model is rare, and when you do hit one, it's usually two faults stacked. Slow down, isolate, and let one confirmed fact lead to the next. Forcing a tidy answer onto a messy reading is how callbacks are born.

References

• Trade-standard troubleshooting methodology (observe, isolate, test, verify).
• Manufacturer service documentation and diagnostic flowcharts (generic).
• OSHA safe-work practices for energized and pressurized equipment.
• See related: The Too-Simple and Too-Complex Trap; Trust the Data or the Story Decision Tree.