Circumstances
On Sunday, 23 September 2012, at 18:25 local time, a Cessna 206, registration F-GBBY, operated by a private company, was on its 12th skydiving flight of the day. The pilot took off from Runway 07 of Saint-Dié-des-Vosges aerodrome with four parachutists. At about 300 feet above ground, the pilot retracted the flaps and set climb power. He perceived that the engine was not delivering sufficient power to continue the climb and that the controls felt "mushy." He leveled the aircraft, switched fuel tanks, selected maximum power with the propeller in full fine pitch, but observed no improvement.
The pilot broadcast a MAYDAY call and attempted a left turn to land on the opposite runway direction. He reported that the aircraft appeared to sink whenever he tried to turn. Due to rising terrain ahead, he executed a forced landing in a field. During the landing roll, the nose wheel sank into soft ground, causing the aircraft to flip onto its back. Two passengers were injured; the aircraft was destroyed.
During the 8th and 9th flights that day, when reducing power before parachute drops, the pilot had noted fluctuations in manifold pressure. After those flights, he checked for oil leaks and performed ground engine tests, which revealed no anomalies. The 10th and 11th flights proceeded without issue. The pilot had first flown F-GBBY two days prior and considered it slightly less powerful than other Cessna 206s he had flown.
Technical Examination
Examination of the Continental IO-520-F engine revealed that the air-fuel mixer filter was clogged with a heterogeneous deposit. The deposit was not analyzed. Upstream and downstream fuel system filters were clean. Three of the six fuel injectors were blocked. After cleaning the filter, the engine ran nominally on a test bench. The aircraft and engine maintenance manuals specify that the filter should be inspected and cleaned every 100 hours.
Autopilot
The aircraft was equipped with a Navomatic 300A autopilot that controlled only the roll axis. When engaged without a higher mode, it maintains the bank angle selected by the control knob. This autopilot uses an electronic clutch, which does not provide the typical control force feedback when engaged. The company had informed the pilot via email that the autopilot was inoperative; therefore, he had not tested or used it during his three days flying the aircraft.
The autopilot switch was found in the ON position, the bank knob centered, and no higher mode selected. Video evidence showed that, during flap retraction, the pilot inadvertently activated the autopilot switch. With no higher mode active and the knob centered, the autopilot attempted to maintain zero bank. When the pilot tried to turn, the autopilot opposed the input, creating an impression of aileron ineffectiveness.
Maintenance History
The aircraft was purchased by the company in June 2011. Prior to the purchase, it had been stored outdoors for several months without flying. Before being returned to service, the aircraft and engine underwent a major overhaul. By the accident date, the aircraft had flown 298 hours since that overhaul.
The mechanic responsible for maintenance stated that the autopilot was overhauled and tested during the major overhaul. The last 200-hour inspection was performed on 19 July 2012, which included the tasks for the 50-hour and 100-hour inspections. The maintenance shop reported that the filter cleaning was carried out during that visit. The aircraft had flown 96 hours since that inspection.
Conclusion
The power reduction was caused by obstruction of the air-fuel mixer filter. The quantity of deposits indicated a long-term clogging process. The absence of deposits on upstream and downstream filters suggested that the filter check was likely omitted during the 100-hour inspections.
The pilot's sensation of roll control ineffectiveness was due to unintentional activation of the autopilot during flap retraction. The erroneous information provided to the pilot about the autopilot's operational status and the lack of unusual control forces prevented the pilot from identifying the autopilot engagement.