However, it is somewhat difficult to follow what inputs the YD is getting, and what effect NCD data may be having on the Ny Accelerometer inputs to the YD calculation.
The answer to that I think is that there isn't any Ny accelerometer input to the YD calculation when Alt2 is operative. So as as I know, the sole short term input to the YD in this state is rate of yaw. Ny does affect the YD operation when in normal law in that the lateral turbulence damping commands are fed through the YD, and if you want to reduce motion at the back end of the cabin then an accelerometer on the rear bulkhead would be a good starting point but as I understand it this is a pilot selectable option
The DFDR Lateral Acceleration (sideslip) trace is reflected in the YD output and its amplitude appears to have increased marginally during the descent and has a natural period of around 7 - 8 seconds which only gets interrupted by prolonged left bank commands. As pointed out earlier, the clockwise heading change is greatest when the Ny driven oscillation is dampened by hard over and held left SS inputs. So my feeling is that the PF managed to dampen the Ny oscillations that could have easily resulted in a total LOC.
The DFDR trace is Ny, which is not simply the lateral acceleration due to sideslip - there is a substantial contribution from the gravity component when banked. And as I said, at high AoA bank generates sideslip directly although there will be additional slip associated with accelerations along the y axis coming from sideforce. Up to 02:13:00 there is very little rudder application, then it is active for about 45 secs after which it goes quiescent again, and while the rudder application was active there is little sense of any reduction in the rolling motion, so I'm not at all sure that your feeling is justified.
Having written the above, then noted your latest comments, I am no further ahead in resolving what I still consider is the YD driving the Rudder and "leading" the Roll/Bank. I've had a look at the Pitch Attitude trace and find that maximum yaw coupled with side-slip tended to occur when pitched up with a RH roll bias.
I'm sticking with my feeling that it is lagging.
You say that, "In Alt2 the rudder deflection is controlled by pedal angle and yaw damper commands", and yet in the period we have been looking at, only YD commands have been presented to the rudder.
That is simply because the PF made no rudder pedal movement surely?
How or why does the Lateral Acceleration trace then synchronize with the YD / Rudder traces? To answer my own question, I have previously posited that the aircraft was effectively 'fish tailing'
They both synchronize with the aircraft dutch roll/roll oscillation motion - the measured lateral acceleration with bank angle plus some additional 'g' from yaw movements (assuming that Ny comes from that rear bulkhead accelerometer); the YD traces synchronize with yaw rate.
Its not important to the outcome, though helpful in understanding what is actually happening in what appears on the surface to be an aircraft in a benign stall and retaining some lateral stability, i.e. avoiding the 'spiral dive'
Yes I agree it is not a be all/end all point. One additional remark may be relevant though - when you get into or towards that high AoA/roll oscillation condition it tends to be a situation where roll due to sideslip is high and weathercock stability low, which is the condition for more positive spiral stability, which makes the reason for the continued roll right even more mysterious! [I've seen this before - in my youth we tested a free flight model which had this roll oscillation and even when we cut the complete fin off it simply wallowed in a straight line with no spiral divergence]