What a record must achieve
A good failure report is not a formality: it is the technical memory that will avoid repeating the analysis next time. It must allow someone who was not present to reconstruct the sequence, assess whether the proposed cause is supported and confirm that the action was effective.
Minimum content
- Equipment identification: type, serial number, location, version or configuration.
- Description of the symptom: what was observed, when, and under what operating and load conditions.
- Evidence: photographs before handling, measurements with their method and instrument, logs and retained parts.
- Analysis: hypotheses considered, which were ruled out and by what test.
- Direct cause and root cause, stating the degree of certainty.
- Actions: immediate corrective action and long-term preventive action, with owner and date.
- Verification: how and when it was confirmed that the failure did not return.
Separating facts, interpretations and conclusions
The most damaging confusion is mixing what was measured with what was assumed. “The capacitor was swollen” is a fact; “it swelled because of overvoltage” is an interpretation that needs support. Writing them separately makes it possible to review the reasoning later.
Follow-up
A failure is not considered closed when the repair is done, but when it has been verified that the preventive action was carried out and worked. A record without follow-up becomes a dead file. It is also advisable to classify with a common vocabulary (equipment, failure mode, cause) so that similar failures can be searched and counted.
A brief example
A board comes back three times with the same damaged regulator. The record of the first failure showed a high temperature in that area that nobody addressed; the third review catches it thanks to that note, and the heat dissipation is corrected. Without the history, each repair would have looked like a new case.