So if the Angle of Attack sensor is providing dodgy data, what can cause that? How does it obtain its data?
From what I have seen on Mentour, The AoA system consists of two independent vanes that point into the airflow around the nose of the plane. One on either side of the aircraft. The fault can be in the physical vane itself, the way the vane's physical position is translated into data, or in the transmission of that data to the flight control computer.
Crucial to the problem is that MCAS uses the pilots AoA vane only. The Lion Air plane that crashed had a fault in the AoA system that was noticed the day before, resulting in some maintenance work. Its significance to MCAS was not understood and, IIFC, the flight crew on the fatal flight were not notified of the fault.
On the Ethiopian plane the faulty data caused the nearly continuous activation the captain's stick shaker. While the FOs column was not affected. This caused significant pilot confusion. According to the Aviation Herald, Ethiopian Airlines had not yet update their flight manual with the Emergency Airworthiness Directive Boeing issued in November, after the Lion Air crash. Information that if absorbed by the crew might have prevented the crash.
So like most accidents, both were a combination of a many things related to a new design. I interpret the preliminary information to be that the MCAS was not robust enough for its implementation at a crucial time in the flight without significantly more training than was suggested by Boeing or offered by the airlines to the pilots and maintenance crews. MCAS may not have been robust enough period. The vane appears to be the weakest point. But further investigations will undoubtedly reveal more.
From the data recovered it looks like ET302 may have had a vane detachment (they have an internal counterweight that would swing it to maximum deflection, and there was a brief period where it 'almost' returned to normal readings, this was at a period of extreme negative G forces, suggesting that the reading was from the counterweight 'floating'- of note was that at that same instant the sensor reading failed to full deflection,the heater sensor failed as well.....
Another thing that may have led to the crews confusion is that although the captains shaker only was going off, apparently the yokes are connected, so the FO yoke may have had some stickshaker bleeding through
A third point of confusion was the airlines requirements for 'no additional training' required for the MAX, but one major difference was the trim yoke switches, in the earlier 737 variants, you could disable all AP functions with one, the other disabled AP control over the stabilizers- BUT allowed electrical trim to continue working. The MAX changed this to disable all AP with both switches- this may explain why the MCAS was turned back on again a couple of times late in the flight- at the speed they were going by that stage, it is literally impossible to trim out the MCAS input to the stabilizers by hand with the trim wheels, and they tried to activate the electrical trim motors as per previous models, only for this to not work as expected (a change which was only noted in the MAX manuals with a 'throwaway line' that the function of the yoke switches had been renamed- nothing said about the fact they had been rewired in a different configuration- no training need then!!!)
Worse when they finally reactivated what they thought was the elec trim circuits (or possibly they deliberately reactivated the entire system- they didn't much to lose by that stage) , it also reactivated the faulty MCAS- but this wouldn't have apparent at first, as it was still held off until they stopped trying to retrim, it waits 5 seconds and then starts winding even MORE trim on...
Apparently the 'official' way to get out of the overloaded trim is to decrease thrust and nose down, while winding off (anything up to a hundred turns!!!!)- not really an option when you are only 1000ft off the ground... (and some media reports have been misinterpreting their altitude to boot which doesn't help) yes they did get to nearly 9000ft before it all went pear shaped, giving the impression they had climbed a bit, what many don't seem to know is that Bole is 7625ft above sea level and where they crashed is higher again- so they were never far off the ground at all...
Of course none of this is guaranteed, but there are some hints from the released data so far that points to this possibly being the issue (I am shamelessly passing on some pretty smart peoples ideas here, I only fly private GA, but a mate flies 737NG's and there is a lot of discussion of the MAX issue obviously)
Be interesting to see what comes out officially in the end...