0
$\begingroup$

Is there any in-flight built-in-test in the airplanes separate from the in-flight monitoring?

$\endgroup$
  • $\begingroup$ Of what? Flight controls? Avionics? $\endgroup$ – Carlo Felicione Oct 9 '19 at 2:29
  • $\begingroup$ What are you trying to test? $\endgroup$ – KorvinStarmast Oct 9 '19 at 18:53
3
$\begingroup$

The thing that separates a BIT/BITE (E=Equipment) from continuous fault monitoring is that the continuous fault monitoring simply responds to exceedances of various paramters or values outside of specified limits in real time as they happen (say where an analog voltage signal of a device drops below a minimum of X or exceeds a maximum of Y), whereas a BIT/BITE test is a specific routine where simulated fault signals are injected into a circuit to see if a fault monitor is doing its job.

BIT or BITE tests of electronic controllers can be categorized into three types:

  • Start Up BIT: Most common. A test routine automatically run on power up of the controller.
  • Initiated BIT: Controllers may have a manually initiated BIT/BITE that can be run any time, but this is usually triggered by a button on the controller itself in the avionics compartment, or through a maintenance access portal in the aircraft maintenance diagnostic system, and usually doesn't have a flight crew initiated facility (but may in some cases). These are normally intended for troubleshooting.
  • Continuous BIT: A controller may run an internal self test routine on a continuous cycle while powered up, where a controller has fault detection circuits that can't be allowed to have dormant failures, even for the duration of the flight.
$\endgroup$
  • $\begingroup$ How faults can be injected while the controller is working? Doesnt it degrade its performance? $\endgroup$ – shooshool Oct 8 '19 at 13:24
  • $\begingroup$ Not necessarily. The faults are injected by software internally and the microprocessor doing the active fault monitoring can be told it's part of the BITE and to ignore the result. $\endgroup$ – John K Oct 8 '19 at 13:44
  • $\begingroup$ I do not design this equipment for airplanes, but this is largely the same in industrial equipment. Real time monitoring is an error/warning system with complex responses. Built in test prevents a device becoming operational (or limits operational mode) when it fails. However, I have not had to design anything as elaborate as simulated faults during BIT/POST , generally we rely on OK signals from components and some functional verification. $\endgroup$ – crasic Oct 8 '19 at 16:05
1
$\begingroup$

Yes, in-operation BITs exist.

For example some older analog fuel flow meters have two BITs, one for zero flow and one for nominal flow. The zero flow BIT will move the needle up to a non-zero flow value, while the nominal flow one will move them down to a different low one. This way the instrument can be tested in any condition without risk that the results of the BIT could be misinterpreted.

$\endgroup$
1
$\begingroup$

On modern complex aircraft systems, each LRU (Line Replaceable Unit) will be conducting a number of built in tests on itself continuously as well as checks that only occur when the unit is powered on

For example, the power supply voltages will be monitored for too high/too low voltages, memory devices will be checked for read/write faults, interfaces with other systems will be checked for a frequent stream of data, data buses may perform a loopback test of themselves to check that it is working etc etc.

Any faults may be reported to a centralised maintainance/fault system which will determine whether the pilots need to be informed of the fault.

$\endgroup$
  • $\begingroup$ Is this different from continuous fault monitoring? For example, we know that sensor faults are detected by monitoring redundant sensors and using a majority voting algorithm. If one sensor is different from the others, it will be considered faulty. The same is for redundant outputs of flight control computers. What I mean is that is this monitoring the in-flight Built-In test or it is something separate that is done in addition to this monitoring? $\endgroup$ – shooshool Oct 8 '19 at 12:40
  • $\begingroup$ @shooshool Two separate items. BIT will typically be checking the hardware/interfaces of the unit. Voting schemes are to ensure availability of the system even if one sensor/device fails but the the BIT flags reported by a unit may be used by a voting mechanism to ignore the data from one source. For example if one of the airspeed sensors reports an internal fault from a built in test, the computing systems using that sensor will ignore it's data. $\endgroup$ – scotty3785 Oct 8 '19 at 12:47
  • $\begingroup$ A/c controllers can be like that but have to go to greater lengths to mitigate the risks of dormant failure modes. But most of the time a BIT on power up is sufficient. You will also see,especially with controllers related to flight controls that the entire computer channel is split into two completely separate sub-channels, with one actually in control and one "going through the motions" in parallel, like a mime, & another chip watching both to make sure they do the same thing. Ea sub channel will use a different brand of microprocessor to prevent architecture related defects affecting both. $\endgroup$ – John K Oct 8 '19 at 16:19
-4
$\begingroup$

Absolutely. Every flight is a built-in in-flight test. For example, if you perform a 2-g turn and the wings don't fall off, you've demonstrated that the aircraft can sustain 2 g's without structural failure.

$\endgroup$
  • 1
    $\begingroup$ How is flying a turn a built-in test? $\endgroup$ – Bianfable Oct 8 '19 at 13:59
  • $\begingroup$ Maybe answer only applies if the plane eventually enters the turn on its own when the pilot releases the controls. In this case the turning tendency (lack of inherent tendency to stay wings-level) would be built in to the design. This would be a good description of most modern aircraft. $\endgroup$ – quiet flyer Oct 8 '19 at 14:04
  • $\begingroup$ I don't think you are using the usual definition of a built-in-test. $\endgroup$ – AEhere supports Monica Oct 8 '19 at 14:10

Not the answer you're looking for? Browse other questions tagged or ask your own question.