Gibralter have recently introduced RNP(AR) approaches (last year?). Boeing and Naverus got down to 0.11nm minima on the 737. Maybe Airbus tried to outdo them and went for a 0.1nm minima. To achieve this it seems they used some software intended for tracking an engine failure/engine out. Bank angle limits on say an ILS are single figures, but on an RNP(AR) approach it is 30 degrees - but that should be a constant bank to achieve an RF leg. Somehow it got lost in translation and Airbus software thinks you can now roll thru a 60 degree arc. A320 roll rates are 30 degrees/second clean, 25 deg/sec dirty and only 15 deg/sec in manual flight.
The RNP(AR) software “operates outside normal flight control laws” and it seems roll rates are unlimited. The software gets stuck in a “loop” rolling thru a 60 degree arc. It will do the same in the SIM if you put in a spot wind of 40|45 knots (crosswind) and mod/sev turbulence - on an RNP(AR) approach. It is too sensitive and too reactive.
It’s a bit like the Airbus Rudder Limiter. It seems to have got lost in translation and software also limits rudder pedal movement. As little as 1 inch of rudder pedal travel will achieve full rudder deflection for the given speed, as Air Canada A319 ACA 190 encountered 10 Jan 2008. Explained in in the report under “Rudder Control System”. (TSB GC CA a08w0007 2008)
Should a bank angle limit become a roll limit (60 degree arc), should a Rudder Limiter limit rudder pedal travel or just rudder deflection?
The crew in Gibralter most likely experienced some Spatial Disorientation and what was perceived/reported as “turbulence” was in fact as a result of g loading and spoiler deflection/buffet. Recommendation for spatial disorientation is to keep the autopilot engaged - crew did everything by the book. Info plus some accident history on site code7700.com under spatial disorientation.