Confirming a Part Is Actually Bad
Why this matters
The fastest way to lose money and trust is to swap a good part. You charge the customer, the symptom comes back, and now you are eating a return trip and explaining why your first fix failed. "Looks bad" is not a diagnosis. A part is confirmed bad only when you have a measurement, a comparison, or a substitution that proves it. This card is the discipline that separates a tech who fixes things from one who throws parts at them.
The three ways to prove a part is bad
There are only three honest ways to condemn a part. Everything else is a guess dressed up as a diagnosis.
- Measure it against spec. Read the part's actual value and compare it to what the nameplate, the rating, or the service data says it should be. A capacitor reads its microfarad value; a heating element reads its resistance in ohms; a sensor reads a known value at a known condition. Out of tolerance is proof.
- Compare it to a known-good identical part. When there is no published spec handy, put the suspect next to its twin. Two identical pumps, two identical relays, two legs of the same circuit. The one that reads or behaves differently is the outlier.
- Substitute and observe. Drop in a known-good part and see if the symptom clears. This is the strongest proof when you can do it safely and cheaply, because the equipment itself renders the verdict.
If you cannot do at least one of these, you have not confirmed anything.
Measuring against spec
This is the cleanest method when a spec exists. The rule is simple: know the expected value before you read, then read.
- Find the rating. It is on the nameplate, the part body, the wiring diagram, or the service literature.
- Put the part in a state where the spec applies. Many specs are "cold" or "at rest" values; some are "under load." Reading a hot element against a cold spec gives a false fail.
- Take the reading on a meter you trust, on the right range, with good leads.
- Compare. Most parts carry a tolerance, often a percentage. A reading inside tolerance is good; outside is bad. A reading of zero or infinite (open or shorted) is almost always bad regardless of tolerance.
The trap here is a partially-failed part that still reads "close." A capacitor at the bottom edge of tolerance may still start a motor on a mild day and fail on a hot one. When a reading is borderline and the symptom is intermittent, treat borderline as bad.
Comparing to a known-good twin
When you have no spec but you do have an identical second unit, comparison is fast and decisive. The key word is identical: same part number class, same rating, same age range if you can manage it. Compare the same measurement under the same condition on both. A meaningful difference points at the odd one out.
Be careful comparing to a part of a different size or rating; "different" then proves nothing because they were never supposed to match.
Substitution: the strongest proof, with one rule
Swapping in a known-good part and watching the symptom clear is the most convincing test there is, because the system answers for itself. But there is one rule you break at your peril: never substitute into a circuit you have not first checked for the condition that may have killed the original part. If a part failed because of a short, an overvoltage, a blockage, or a stuck mechanical load, the new part dies the same way the instant you power up, and now you have burned two parts and learned nothing. Find and clear the cause first, then substitute.
When you cannot fully confirm
Sometimes the part is sealed, the spec is unpublished, or substitution is not practical on the spot. When you genuinely cannot reach a clean confirmation:
- Use the heat-or-cool test or a tap-test to force an intermittent fault into the open so you can at least see it change.
- Eliminate everything around the part. If the inputs are good and the outputs are wrong, the part in between is condemned by exclusion, which is a legitimate diagnosis when documented.
- Tell the customer plainly what you confirmed and what you inferred. Honesty about a probable-cause diagnosis beats a confident wrong one.
The habit to build
Before you reach for a replacement, ask yourself one question: "What measurement or test proved this?" If the answer is "it looked old" or "it usually is the problem," stop and prove it. Parts that usually fail still need confirming, because the day you assume is the day it was something else.
References
- Trade-standard component verification practice (measure-to-spec, compare-to-known-good, substitution)
- Manufacturer service documentation for published component tolerances and test conditions
- See related: The Heat-or-Cool Test to Force a Fault; The Scene Tells the Story Before You Touch