During takeoff from runway 17 at Melbourne the Boeing 737-300 aircraft experienced a single, rapid, right rudder deflection. This occurred when the aircraft was accelerating between V1 (135kt) and Vr (141kt). The co-pilot was handling the aircraft and was able to apply left rudder and arrest the developing yaw. The co-pilot advised that he applied between 2 and 3 inches of left rudder movement. The takeoff and climb were completed without further problems and the flight proceeded normally to Adelaide.
An inspection of the aircraft failed to find any airframe or engine anomalies that could have contributed to the reported event.
After consultation with the manufacturer it was determined that there was two possible causes for the right rudder deflection, namely a gust or a short duration rudder pedal pulse. Accordingly, the rudder power control unit (PCU), the standby rudder power control unit (SPCU), and the digital flight data recorder (DFDR) were removed for examination.
The DFDR recorded a maximum right pedal deflection of 0.17 inch. Because the DFDR only samples rate of pedal deflection twice per second it is possible that the maximum right pedal deflection recorded was not necessarily the peak of that particular deflection. The manufacturer advised that simulation of the event showed that a 3.3 inch right pedal pulse with a duration of 0.5 seconds would be needed to generate the lateral acceleration and heading changes recorded by the DFDR. This equates with the crew's report of between 2 and 3 inches of corrective rudder movement being applied.
The PCU was bench tested in accordance with the manufacturers requirements and found to be serviceable.
The manufacture advised that an identified cause of involuntary rudder movement was the binding of the input lever on the SPCU resulting in pedal feedback if the input lever binding force is sufficient to overcome the feel and centering unit restoring forces. Previous analysis of the SPCU input lever binding conditions indicates that approximately 20 lb resistive force at the input lever is required before any measurable uncommanded rudder pedal movement would occur. A mechanical binding force of this magnitude would be repeatable and would worsen over time. A limit of 1 lb binding force on the input shaft has been mandated. The removed SPCU was tested and the input lever binding force was found to be below the mandated limit. All other test requirements were found to be satisfactory.
A comparison of the airspeed and ground speed recording on the DFDR showed a fluctuation of airspeed at the time of the event accompanied by a fluctuation in the angle of attack vane. This suggests that a wind gust could have affected the airspeed and contributed to the yawing motion experienced by the aircraft.
The takeoff had been commenced from intersection C on runway 16 with a reported surface wind of 10 to 15 knots from 130 degrees. Four minutes prior to the B737 takeoff a B747 aircraft had departed using the full length of the same runway. The parameters for the B737 takeoff would normally preclude wake turbulence from the B747 being considered as a factor.
The investigation was not able to determine if the gust that affected the aircraft was due to a local atmospheric disturbance or was as a result of lingering wake turbulence from the departing B747.