Originally Posted by
xyze
Something I have never understood wrt this - is it possible for the a330 to be in the air with an airspeed of 60 knots or even 70 knots and NOT be stalled? If not then why inhibit the warning? Worrying about sensor accuracy seems to be missing the bigger picture wrt the purpose of the warning!
It is the design of the warning rather than the purpose. Invalidating the sensor readings (speed and AOA) at the ADIRU is not a bad design (and might have saved ET302). Failure to consider the possibility of known-invalid data further downstream is bad design.
The AF447 stall warning problem is even more subtle though - invalid data
was designed for, the problem is (it appears) that it was designed without state so
known invalid data resulted in "no warn". This was arguably incorrect, but known invalid data shouldn't result in "warn" either. With a stateful design a transition from stalled-data to invalid-data would not take you out of the "warn" state, not complicated (he says, having forgotten most of his NFA/DFA and statemachine stuff), just wasn't done that way.
Ironically if AF had taken the optional (standard on later buses) A330 feature that they said their pilots didn't need, the stall warning design issue would have been fixed (BUSS, optional on 330, fixes it). Whether it would have saved them, I don't know for sure, very possibly irrecoverable by the time they got to that point.