Parts & Tools

Known-Good Is a Discipline: Reference Parts, One Variable, and Honest Labels

By Alex Rodriguez, repair technician · September 24, 2026 · 5 min read · Leer en español

A test part you cannot trust produces results you cannot trust. How to keep a reference kit, change one variable at a time, and label pulled parts so today's test does not become next month's mystery.

Every bench diagnosis that involves swapping a part rests on one quiet assumption: that the part you swapped in is good. When that assumption is wrong, the test lies to you with complete confidence. Known-good testing is therefore not a part, it is a discipline. It starts with keeping a small reference kit on purpose: reference flexes for the models you see most, a reference screen, and a reference battery, each one verified working, set aside for testing only, and never sold, loaned into a repair, or left in a customer's phone overnight.

Verified is the word that does the work. A reference part earns its place by being tested in a phone known to be healthy, or by arriving as a pull from a device whose fault was proven to be elsewhere. Label the reference kit as what it is, store it apart from stock, and re-verify it whenever it is dropped, borrowed, or suspected. The kit is small because it has to be trustworthy, not comprehensive — a few parts you would stake a diagnosis on beat a drawer of maybes.

With a trustworthy reference in hand, the rule is one variable at a time. Substitute the single suspect part, test under the conditions that reproduced the fault, record the result, and restore or proceed before touching anything else. The moment two parts change together, the test stops being a test: improvement cannot be credited, and no improvement cannot be assigned. A/B testing is the same discipline stated as a comparison — original part, observe; reference part, observe; original part again if the result is surprising. It feels slower than swapping everything suspect at once. It is faster than doing the job twice and explaining a comeback, because each step actually removes a suspect from the list instead of rearranging it.

Labeling is where benches quietly win or lose this discipline. Every part pulled from a phone gets a label at the moment it comes out: which phone, which ticket, the date, and what it is suspected of — or, just as important, that it was pulled working and is not a suspect at all. An unlabeled flex in a tray becomes, within a week, a mystery part that someone will either throw away in frustration or, worse, reach for as a test part. That is how unknown parts enter the reference chain and poison it.

Only after this discipline has run its course — reference parts substituted one at a time, results recorded, originals labeled and cleared — has a part earned the right to blame the board. Board-level is a serious conclusion with serious costs attached, and it should be the survivor of a process, not the residue of a frustrating afternoon. Keep the reference kit honest, change one variable, label everything you pull, and your tests will mean what you think they mean.

Filed under: known-good parts · A/B testing · reference kit · bench discipline

Check your own log

Have a panic-full .ips file from a phone doing the same thing? Run it through the free analyzer — it reads the file locally in your browser and shows you the evidence it finds, the inspection targets it ranks, and the next checks it recommends.

Open the free analyzer

More from the Journal

Browse all articles