Interesting speculation that the 22-degree offset in the two AoA sensor readings in the Lion Air crash corresponds closely with a "bit flip" error in the digital representation of the value.
Also that the AOA sensor in question had just been replaced prior to Lion Air 34, the flight with MCAS upset and recovery prior to the fatal Lion Air 610. And further, there have been reports that MCAS alternates each flight which AOA sensor it uses for decision making; if that's true, a single faulty AOA sensor would not result in back to back flights having MCAS upset.
i.e. to speculate: the fault may not be in the the AOA sensor, it could be corruption in the communication between sensor and computer, or in the computer.
And more speculation. One contra argument I have with "bit flip" is that's usually transient. To consistently get a particular bit flip form of corruption? That's pretty weird, I don't have an explanation for that. The AOA sensor is a stream of data. And the preliminary report for Lion Air 610 shows a consistent disagreement between the two AOA sensors for the entire flight, almost like it's some kind of installation or calibration related error. But I'd like to believe each of these repair procedures have a verification procedure. These are really complicated systems, so I'd expect I'd want to do a test before repair, then conduct the repair, then do another test. Are the before and after test results consistent with repair expectations?
2 comments
[ 2.8 ms ] story [ 10.6 ms ] threadi.e. to speculate: the fault may not be in the the AOA sensor, it could be corruption in the communication between sensor and computer, or in the computer.
And more speculation. One contra argument I have with "bit flip" is that's usually transient. To consistently get a particular bit flip form of corruption? That's pretty weird, I don't have an explanation for that. The AOA sensor is a stream of data. And the preliminary report for Lion Air 610 shows a consistent disagreement between the two AOA sensors for the entire flight, almost like it's some kind of installation or calibration related error. But I'd like to believe each of these repair procedures have a verification procedure. These are really complicated systems, so I'd expect I'd want to do a test before repair, then conduct the repair, then do another test. Are the before and after test results consistent with repair expectations?