ELECTRICAL
Back to Explainers

A later normal reading never erases a reproduced failure

A circuit that passed a test after the fault fired is not a circuit that passed the test. Proof survives calm weather.

Conceptual definition

A reproduced failure is an intermittent fault you have caught red-handed: observed under a reproduced trigger, recurring, never a single unrepeated spike. That observation is proof the fault exists. It stands on the reproduction itself, not on whether you can reproduce it again on demand.

A normal reading is a measurement taken when the fault is not firing. The fault was not present at that moment, so the measurement proves nothing about the moments it does fire. It is a reading of a calm circuit, not a verdict on the faulty one.

This is the UNPROVED != EXCLUDED rule in action. A clean reading taken later has not proved anything. It has certainly not un-proved the failure you already caught.

Why it varies

Intermittent faults reproduce sometimes and not others, and every reason for that is about conditions, not about proof.

None of that invalidates the earlier proof. The reproduction was witnessed. Witnessed does not expire.

What it is not

It is not "the fault fixed itself." Faults do not fix themselves. Dormancy is not repair.

It is not "the first test was wrong." A witnessed reproduction under a real trigger was not a bad test. The calm reading that followed was not a better one.

It is not permission to clear the codes and ship the car. Codes cleared after a reproduced failure are a clean slate on a guilty circuit. The customer becomes the reproduction trigger, and the comeback is your fault twice over.

Failure modes

Shops lose money on this principle in predictable ways.

How SpeedNeeds uses it

The caller QTSs state the rule outright: a later normal reading does not erase a reproduced failure. QTS-POWR-E9 and QTS-FUEL-A2 carry it in prose because their whole verdict depends on it.

QTS-ADV-X3 is built on it. Its result-3 and result-4 are honesty routes: a clean walk run against an unfaithful reproduction proves nothing, and zero confirmed events after faithful attempts ends the walk with the fault unproved, not exonerated. The procedure refuses to let a calm walk overwrite a proven event.

Stop guessing. Prove the next move. A proof you witnessed beats a normal reading you took when the fault was sleeping.

Caution and edge cases

The principle cuts one way only. A reproduced failure stands as proof that the fault exists. It does not prove the fault is present right now. Dormancy is real: a fault you caught Tuesday can be genuinely invisible Thursday, and parts you replaced on the strength of Tuesday's proof still need verification that the signature is absent at the original observation point.

Do not let this principle become "we proved it once, stop testing." The proof survives calm weather, but repair verification is still mandatory: the caller's signature must be absent at the original observation point, and the repaired segment must test clean under load.

And the bar for entry never drops. This rule protects reproduced failures. A failure was never reproduced if it was seen once and never again. That is not protected by this principle; that is the single-spike rule rejecting it.

Closing clarity

The fault you caught red-handed does not get acquitted by the fault you could not catch. Normal readings measure calm moments. Reproduced failures define guilty circuits.

Write the failure down, keep the witness, and stop letting calm weather talk you out of what you already proved.