Start of impact:
Not before Bank Angle exceeded zero at 20:40:32
At which angle depends on compressed strut state of Left MLG
When we assume it's in the 5°-12 region, that gives us a time of start between 20:40:33 and ...34. Only 1 second uncertainty!

End of impact:
Latest when the Bank angle started to reduce, because that will move up the wing tip (instead of the fuselage and right wing suddenly losing height). Maybe it takes a split second to unload the wing tension.
So end of impact around 20:40:35.0...35.5 mark

Duration in the 1 - 2.5s range. Can we narrow it down further?

I'm also looging at the Vertical Acceleration Graph, but can we read something from that?

From our position we cannot rule out any erroneous system influence regarding rudder during the takeoff roll. Because the direct link is augmented with the auto-coordination/yaw-damper function.
Not knowing Airbus at all, I wonder what the partial deviation rudder pedals to actual rudder surface FDR could mean. It's not simply a delay.

https://cimg5.ibsrv.net/gimg/pprune....7642139525.jpg

I thought about it. Since it happened in the seconds of the wing impact, the Left Outer Aileron was likely to pressed back. See contact mark on picture.

https://cimg4.ibsrv.net/gimg/pprune....d0e3b2b16c.jpg
When the wing flexed away the aileron took some of the force? So the wingtip contacted earlier, then for a shorter time the aileron? Or was it a momentary impact from the marker alone? But wouldn't this show a sudden pushback and a slow extension of that aileron?

Does FDR show actual hardware position of the outboard aileron at all?

Yes, that must have been a moment of wingtip scraping, and it conincides with the moment the right MLG lifted off the ground, but the left MLG still has weight on it.
This means that the force around the longitudinal axis was present already at Vr and banked the aircraft immediately, while the left MLG hasn't even come away from the runway.
I am not sure how it works with this bird> does it have a crossover speed, below which the ailerons are unable to compensate the rolling forces caused by the almost fully deflected rudder surface?
If yes, the question is not why the aircraft rolled, but goes back to the question, why the pedal deflected so much to the left a few seconds before rotation and unstick.

So, someone correct me if I am wrong, but the only thing I see is the rudder being deflected to the left, during the ground roll. As they get closer to Vr, this input increases, and later the PF comments that he was almost full left rudder to correct for their left deviation (…????…). Upon rotation with a lot of left rudder it rolled to the left, and (both) pilots put in right stick input, and the aircraft started to roll right, the pilot(s) added right rudder, and the aircraft rolled too far too the right. Eventually the oscilations stopped.
I see nothing the aircraft did that isn’t explained by the inputs the PF made. And his comments to the FA that with the bus you never know what it is doing don’t really help.

Where did you see THAT comment?

Crew statement:
Crew interview

Mind what you don't see and can't feel, hear, sense. I refrain from premature conclusions.

CVR from the docket. 20:46:32.0
INT1"that was a ah full left rudder to keep it on the, on the runway and the ah"

So do I. No conclusion was drawn. I asked if my observation was incorrect.

Thanks, I see. That comment refers to the left rudder that was necessary to keep it on the centerline during the takeoff roll due to the right crosswind..
It is NOT referring the unexpected left turn just about rotating and becoming airborne.
That part is referred to in the 2nd part of his remaining sentence
The "tits up" event is when the aircraft made a surprise turn to the left and approched the the runway edge during V1-Vr, as if the right crosswind had been replaced by a crosswing from the left, or, the amount of force necessary to deflect the rudder pedal to the left had decreased significantly, allowing it to be deflected much more than during the takeoff roll - with the same force from the captain's left leg.

Yes. I fly the A320. You would need a small amount of rudder input for the crosswind that was reported. It looks like they were left of center, and in a left turn on the ground. His statement suggests that he was making the input that caused that. If you rotate holding full left rudder, and again, it looks like he did, a left bank after rotation would be expected. I still don't see anything the aircraft did that wasn't commanded by the pilot.

A useful source is in Dsc 27-20-10-10 which shows how the modes change on rotation.When the aircraft is on the ground (in “on ground” mode), the sidestick commands the aileron and roll spoiler surface deflection. The amount of control surface deflection that results from a given amount of sidestick deflection depends upon aircraft speed. The pedals control rudder deflection through a direct mechanical linkage. The aircraft smoothly transitions to “in flight” mode shortly after liftoff.

When the aircraft is in the “in flight” mode, normal law combines control of the ailerons, spoilers (except N° 1 spoilers), and rudder (for turn coordination) in the sidestick. The pilot does not need to use the rudder for turn coordination. While the system thereby gives the pilot control of the roll and heading, it also limits the roll rate and bank angle, coordinates the turns, and damps the dutch roll.

The full pedal delection is about the famous 10 inch.

A recognizable xwind of 15 kts requires about 2 inches, later reducing to about 1. During the final stages before rotation it is pulsating a bit (towards 0) to prevent overcompensation and fighting your own PIO.

​​​​At roatation I get the most consistent and smoothest results by centering everything. Immediately post liftoff the personal de-crab techniques differ.

Last tested 3 hours ago.

The CVR reads to me very clearly. Also the recorded parameters look consistent to tracked inputs on the controls.

To reiterate, I find constant 1 inch (10% of travel) input during high speed borderline excessive.

The plot shows the initial deviation compensated by full into the wind sidestick (which has no effects whatsoever) and a cross-control full rudder eventually, which only worsens the situation and is recored as such.

That's fine, but then it begs another question that maybe you can explain.
We also know that the pilot placed an unusual amount of input force to the left pedal during the early stages of the takeoff roll. (He stated this during the interviews as a "feeling", and he also was surprised about simulation playback later on.)

So in that case, why didn't the aircraft run off the runway earlier?

I find it rather suspicious, that someone is able to keep a fairly steady heading during the takeoff roll, but suddenly mixes up left and right leg just about becoming airborne. .
This sort of left/right mixup theory was already tried (and failed) during the attempts to explain the Boeing rudder reversal accidents in the 1990's. FDR's at that time didn't show the actual rudder deflection, so it was easier to make such a claim, assuming that the rudder pedal and control surface must have moved together. But Boeing's expert was proven wrong., pilots don't mix up left and right legs after a few thousand hours. So "the captain suddenly forgot everything he knew, and mixed up his left and right leg" theory seems very odd again.
By the way, the aircraft never flew again, it was scrapped, which I find a little unusual. Perhaps a PR move from the airline.

I wonder Is there any data about the spoilers in the FDR? I could not find any.

edit: I am curious about that because from the outline of the event, the captain had to input more than usual left rudder to keep the aircraft accelerating on the runway. Why? Maybe:
- much crosswind stronger than expected?
- nose wheel steering slight misalignent?
- asymmetric braking of the main wheel?
- excessive right spoiler deployment?
- ?

I could attribute the event to the pilot's complacency and lack of attention, if the wrong inputs had been provided only during the rotation time period.
With the excessive left rudder input already present during the early part of takeoff rol, I can't help thinking of a right yawing tentency he tried to keep under control, and which disappeared in the moments of becoming airborne.

Modern FDR should record it, but it's neither in the plot not in the tabular "raw" data inside the docket.

I think nobody challenges this anymore. It's quite clear

So either it's the PF trying to cover up mishandling or ...

... any influence yet unknown to us.
​​​​​​​

Time,Time Hrs,Time Min,Time Sec,Altitude Press,Accel Vert,Aileron-L,Aileron-R,
Airspeed Comp,Altitude Radio 1,Elevator-L,Elevator-R,Eng1 N1 Act,Eng2 N1 Act,
Gear WOW-L,Gear WOW-N,Gear WOW-R,Heading,Pitch Attitude,Roll Attitude,
Rudder,Rudder Ped,Sidestick Lat-L,Sidestick Lat-R,Sidestick Long-L,Sidestick Long-R

That's in the tabular data. Anything not plotted in the Graph?

Brushed up my FDR-to-Excel Skills (it's tough using a non-US-version with its other decimal and time format support.)
Here's higher resolution of the roll and rudder values (right axis, minus=left is up). Only added the WoW as other data so far

https://cimg2.ibsrv.net/gimg/pprune....bbd347d102.jpg

What right yawing tendency? There is a left yawing tendency in the heading starting some time before the main upset. And after this left yaw started left rudder pedal was continously increased to more than twice the deflection (and an enourmous deflection at that) compared to when the left yaw started. There is nothing suspicious in the traces regarding what the aircraft did compared to what the control input was. Only problem is the inputs don't seem to make any sense. I guess we will never really know what the reason for those inputs were.