PJ2, re # 1805, 1819, DJ77, #1793, et al.
As I understand, there is a step change in the value of stall warning at the transition between normal and alternate laws (upper vs lower tables in # 1818) and it is also possible to infringe the lower, alternate boundary, in turbulence or gentle maneuvering at high altitude.
Both values of stall warning are a function of Mach as aerodynamic stall AOA decreases with increasing mach No, and alternate law has no stall protection; thus at higher speeds the warning is at a lower AOA. The details are explained in
What is a stall.
Also at higher Mach No, trim (THS) is biased to counteract the Mach induced nose up pitch moment (Mach trim in conventional aircraft); I assume that this moves the THS slightly nose down.
At the moment of changing law (airspeed mismatch / failure), the control response is identical except for the stall warning. In addition, the aircraft aerodynamics and the control surfaces, THS/elevators, are matched - the aircraft is 'in trim'.
“Hypothesis”. If we consider a slow blockage of the temperature sensor (cf Vh-EBA ~ 30secs), at some time before the pitot blockage, and that this resulted in an increase in computed Mach (cf VH-EBA); then the Mach Trim function could be incorrectly biased nose down. In auto flight, this would probably be countered by an up elevator demand (all small amounts), but the aircraft would be technically mistrimmed as the wing aerodynamics and the control surfaces, would not be matched, i.e. the wing was not creating a nose up moment – no real Mach increase, but the trim was set to a real nose down position due to a false Mach increase.
When changing to Alt Law, - the point of pitot blockage and autopilot disconnect, the aircraft would be mis-trimmed for level flight. If the trim aspects (THS, control stick – elevator) were not zeroed out electronically (and I can’t see how this might done at this time), then the resultant effect might offset the 1g level flight demand such that nose up control input was always required, i.e. the zero control position did not result in 1 g flight – in this instance level flight. An obvious question is what exactly does the FDR record as ‘control demand’?
Following on, the PF would require a nose up movement just to fly level, but at some point, this should command auto trim (THS) – nose up, but what the null trim datum would be, either the stick centre (no demand), or offset THS datum due to previous false Mach, is unknown.
Again, the values might be small, but sufficient such that the PF would have to control a mistrimmed aircraft, perhaps with increased sensitivity and tendency to over control which could trigger the earlier stall warning.
Add to this any control activity with regaining altitude; cf VH-EBA, -300ft (more nose up demand and thus THS nose up), and any latent perception of increasing Mach (false value), again nose up attitude demand to reduce speed.
This is a very complex technical scenario dependant on where the trim datum (null control) might be positioned when changing to alternate law with erroneous Mach. The elements which trigger the mismatch are highly likely to be present; the details in the control law are unknown.
However, the hypothesis might provide a reason for the propensity for nose up control and trim movement (FDR), the early stall warning, and in the extreme (and perhaps most important), the mis trimmed interaction with normal control in alternate law might provide a reason why the trim ran fully nose up and ‘apparently’ did not move in the opposite direction, as the THS never achieved its (false) null position.