For example, the CVR records only about 2 hours of data.

Overwriting important data must be avoided, and thus the recorders should know if a catastrophe has happened. Then they can stop recording (to protect the already recorded data from overwriting), and start signaling the rescue team.

What is the logic used to stop recording?


3 Answers 3


The CVR and FDR have impact switches to stop recording when they experience high acceleration. In incidents and accidents that don't involve high enough acceleration, the plane remains intact enough for the boxes to be immediately found and switched off manually. In more serious crashes, the boxes stop recording because of the physical destruction of everything but the memory units (only the memory units are built to survive crashes).

  • 2
    $\begingroup$ I had guessed: when the power cables to the unit have been ripped off. Is this a valid case? $\endgroup$ Apr 1, 2015 at 10:02
  • $\begingroup$ @aitchnyu: Yes; in many crashes, the flight recorders' chasses (which are armoured, cushioned, insulated, the whole nine cubic yards, though not quite as much as the memory modules themselves) survive completely intact, instead of just the memory modules, but the power cables (which are completely crash-unprotected) do not. $\endgroup$
    – Vikki
    Nov 21, 2018 at 23:01

In the United States, CVRs and FDRs have a 10 minute (+/- 1 minute) battery requirement to allow any available post-crash voice and data to be recorded, with power removed once a timer limit is reached.

While I'm not familiar with the internals of the FDR, simple circuitry could trigger the timer to start based on current flow from the battery, if not for an acceleration-based impact sensor suggested.

  • $\begingroup$ The wording of the linked rule seems to suggest that the independent-power-source requirement is more intended to allow the flight recorders to continue operating if the aircraft suffers an electrical failure in flight, rather than to ensure post-crash operation. $\endgroup$
    – Vikki
    Nov 21, 2018 at 23:10
  • $\begingroup$ ...if not for the rule in (d)(2) to remove that power after 10 minutes as well. $\endgroup$
    – Erich
    Nov 28, 2018 at 14:06
  • $\begingroup$ If it were meant to ensure post-crash operation, there wouldn't be any point in including a requirement to also power the cockpit area microphone for that length of time, given that the CAM won't be functioning after a crash anyways. $\endgroup$
    – Vikki
    Nov 28, 2018 at 21:59

In crash the last two hours before the crash is most important, so the recorder always recording and stopping when the plane crashes either because of crash sensor or simply due to loss of power and input as the wires leading to it are torn by the impact forces is the right thing.

However in some incidents¹ the data would also be useful. When the crew realizes something occurred that needs to be reported as incident, they are supposed to pull the circuit breakers after landing and tell the appropriate safety board to come for the data (which must be done before next flight). Alternatively if the aircraft is equipped with quick access recorder² they can instead preserve those data, which can be done by any maintenance worker, so the plane can be dispatched again.

However sometimes the crew fails to realize they should preserve the data (or fail to report the incident altogether and it may be reported by someone else like another crew, traffic controller, maintenance etc.) and get scolded for it.

¹ Incident is any occurrence in which safety margins were reduced, but nobody was hurt and there was no or only small damage (with damage to engines or landing gear always considered small). The kind of incident where investigators would most want to have CVR and FDR data and don't always get them is TCAS activations (near misses); they generally have ATC voice recordings and radar tracks, which ATC records in much longer loop, but the on-board recorders often get overwritten either before the planes land or because the crews fail to preserve them.

² The quick access recorder is not crash-worthy and not mandatory, but most airliners have it these days. It usually has longer recording loop and records more parameters, but the main advantage is that any mechanic with portable computer can download the data in couple of minutes and the aircraft can be dispatched again instead of waiting for the safety board investigator to come around.

  • 3
    $\begingroup$ "the last two hours before the crash is most important" - Almost always, but probably not in the case of MH370. $\endgroup$ Apr 1, 2015 at 9:24
  • 2
    $\begingroup$ @RedGrittyBrick: I wonder why the new types don't store longer recording. Quick calculation tells me the raw 4-channel sound would be around 1¼Gb/h which does not sound like that much (yes, it should probably not use the most integrated chips as resilience is more important, and it is probably stored in multiple copies but I would still expect a bit more to fit in the box). $\endgroup$
    – Jan Hudec
    Apr 1, 2015 at 9:44
  • 2
    $\begingroup$ Agreed. Maybe the businesses involved just don't voluntarily make improvements over what regulations require (set in the days of magnetic tape recording). I don't understand what significant obstacles there could be other than institutional inertia and maybe certification costs. $\endgroup$ Apr 1, 2015 at 9:56
  • 1
    $\begingroup$ @JanHudec oh, thanks! That makes sense. And the FDR will probably still have the data, which is something at least. $\endgroup$ Apr 1, 2015 at 12:12
  • 1
    $\begingroup$ @JanHudec: I know of at least one case (a near-MAC in the 60s or 70s; I'll add a link when I get home and can look it up) where the flightcrew of one of the planes involved pulled the CVR's breaker a few minutes after the fact, which significantly aided the NTSB's investigation; OTOH, this was back when they only recorded 30 minutes (usually plus one or two more to be on the safe side), rather than 120, so preserving the recording would have been significantly more time-critical than it would nowadays. $\endgroup$
    – Vikki
    Nov 21, 2018 at 23:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .