Avionics Editing and Proofreading Services

The autopilot disconnects itself twice a month, always in cruise, and never while a technician is watching. Each time, the shop runs its standard tests, finds nothing, writes "no fault found, system tested satisfactory, returned to service," and reinstalls the same unit. Six months and four such reports later, nobody has asked whether the fault appears only above a certain altitude, at a certain temperature, or after a certain number of hours since the last power cycle. No report has recorded the conditions under which the fault failed to appear, only that it did not appear during the test.

No fault found and troubleshooting reports are the documents we see most from repair stations. Return-to-service statements and release documentation come with them, along with interference and intermittent fault write-ups. Also in the pile are software and configuration change records, bench test and ground test documentation, and the correspondence sent to an operator about a recurring, unresolved fault. Our editors work on the report written after a fault that refuses to reappear on command.

The conditions of the test are what turn a "no fault found" report into evidence rather than a closed file, and the failure is a report that records what was done without recording what it could not reproduce. "Tested satisfactory" says the unit worked on the bench. It says nothing about whether the bench resembled the conditions that produced the fault. The conditions of the original occurrence belong in the report precisely, taken from the pilot report: altitude, phase of flight, outside air temperature, and elapsed time since power-up.

Those conditions then have to be set against what the bench could actually achieve, because a ground run at 70 degrees cannot rule out a fault that appears at cruise altitude after two hours aloft. Every prior occurrence of the same fault on the same unit belongs in the report as well. Four "no fault found" reports on one component are a pattern of intermittent behavior, not four unrelated non-events. What was checked should be itemized rather than summarized as "system tested," because a reader six months later needs to know whether the wiring, the connectors, and the software configuration were checked or only the unit itself.

"Fault not present during test" and "fault does not exist" are different claims, and only the first is supported by a ground run. Where a fault cannot be reproduced on the ground, the report should say so plainly rather than let the limitation disappear into a satisfactory result. A recommendation for what would actually catch an intermittent fault belongs at the end, whether that is a flight test, a data logger, or a period of extended monitoring. Reports written this way eventually find the fault instead of returning the same unit to service four more times.

Everything you send us is treated in confidence, including fault histories, technical data, and correspondence with operators or manufacturers. We are editors, not licensed avionics technicians or engineers, and we offer no opinion on fault diagnosis or airworthiness. What we can do is make sure the report says exactly what was and was not established.