@RWA:
The next phase started with the stall warning sounding. The PF appears to have responded with the correct drill at the time - 'TO/GA power and seek to maintain altitude
Nope. The drill for UAS was TOGA and 5deg nose up. At that moment (as per BEA) the plane was already 6deg nose up. So making persistent (for over minute as per BEA) nose up inputs to get attitide from (initial) 6 deg up down to 5 deg up is not certainly proper drill. The craft was still climbing (at 700fpm from 37500 to 38000) so maintain altidude does not seem yet to be an issue (or it'd rather dictate nose down).
The reason for nose up inputs (inputs, not surfaces position, but pilot inputs at the stick) remains unxeplained as for now.
This was the crucial 'second phase.' Whether this was the result of TO/GA power pushing the nose up, the PF's inputs, the THS going to 'full up,' or the thin air at 38,000 feet, or whatever, we don't know; my own feeling is that it was probably the result of a combination of all those factors. In any event, the aeroplane appears to have 'sat on its tail' and entered a deep stall.
Coffin corner has been calculated for that plane by one of the experienced posters here. It was determined to be 46000ft. Even if the air was somewhat warmer (as crew discussions as disclosed by BEA indicate) it was rather no worse than 43000ft -- still 5000ft above 38000ft achieved by the plane. And the fact is that the plane was slowed down to Mach 0.68 before stall warning came -- so there was a margin from 0.68 to somewhere around 0.85 -- hardly a corner.
iceman50:
Read the BEA report and try to understand it the THS went to the nose up position because the PF demanded it!
Exactly. It's stated explicitly. All the discussion about the reasons of trim going fully nose up is inventing (non)facts to fit someone's pet theories.
It then stayed there because autotrim was then disabled and all they had to do was use the trim wheel to reset the THS and it would have been annunciated on the PFD.
It seems more probable that there was no mode change to Abnormal attitude. Most probably because IAS below 60kts caused AoA readings to be treated as unreliable and thus disregarded by mode change logic. At the same time to get that (auto)trim down by the use of stick it'd require consistent nose down inputs lasting for time long enough to equalize previous nose up action. "Nose-neutral" inputs won't change (auto)trim, as someone few tens of pages back noticed.
BTW. Wrt AoA relaibility at slow speeds. There was recently thread about rejected takeoff above V1 incidedent and investigation body's take on in. One interesting tidbit pertaining to our discussion is AoA plot from FDR from that plane as presented in the report. As speed of the plane decreased below ~45-50kts measured AoA started to deviate and at ~40kts in went completely wild (like 60 or even 90deg). That plane was on the ground so it's real AoA remained allmost constant.