Dozy, contrary to your observation, the data shows that the initial L>R>L etc roll was very quickly contained and controlled.
Hi PJ2,
Help me read the traces, because what I see is a considerable amount of roll oscillation for about 30 seconds immediately after disconnect, which dovetails nicely with the near-full deflection of the inner ailerons (timecode approx. 2:10:05-2:10:35). Immediately after this the inputs reduce in magnitude and roll starts to stabilise (consistent with getting a feel for the aircraft), but shortly thereafter the aircraft enters the approach to stall regime and control seems to run away from the PF.
As a non-pilot I don't have the subjective knowledge to say how long it should take to contain, but 30 seconds seems like a fairly long time to me.
Please correct me if I'm barking up the wrong tree.
@HN39 - I don't think he's misquoting - at least not dramatically. What he's doing is interpreting the phrase as meaning things happened sequentially rather than concurrently. Whether due to translation issues or problems with wording, it's difficult to say one way or the other when taken out of context. Put in context with the rest of the report though, it's clear they mean that the aircraft climbed concurrently ("at the same time" - for non-native English speakers) with the increase in pitch attitude.