WPR17IA086

A mistrim condition while the autopilot was engaged for reasons that could not be determined based on available evidence, and the pilot’s subsequent failure to confirm the command bar status before disengaging the autopilot. Contributing to the accident were the pilot’s misinterpretation of the event based on a previous unrelated experience, a jammed trim switch, and autopilot disconnect switch which broke off in flight, and the stress induced by the physical exertion required to maintain control of the airplane.

After performing a series of uneventful landings in the single-engine airplane at a local airport, the pilot flew a return leg with the autopilot engaged. As he approached the airport, he disengaged the autopilot using the autopilot disconnect button on the control wheel. The airplane then immediately pitched up aggressively with an immediate loss of airspeed such that the pilot feared the airplane was about to stall. He applied full forward pressure on the control wheel, and the airplane pitched almost directly nose down and began to quickly build airspeed. He then reduced the control wheel pressure, and the airplane pitched back up again. He was able to maintain an approximate level attitude with full forward control wheel pressure, although the control forces were so great that he needed to use both hands and his knee to keep the control wheel forward. During the event, the pilot reportedly attempted to adjust pitch with the trim wheel, but it was jammed. He also attempted to disengage the autopilot by pulling its circuit breaker; however, the breaker was both beyond his reach on the right side of the airplane and blocked by the right arm of the control column. Other methods of responding to this emergency were available to the pilot; however, due to the jammed trim wheel (there are no emergency procedures that cover this contingency) and a previous experience involving a throttle cable failure, the pilot was convinced he had a flight control failure rather than an autopilot anomaly and did not attempt any alternative actions. Ultimately, the pilot was able to regulate pitch with engine power and landed the airplane gear-up. Examination of the airframe did not reveal any mechanical failures that would explain why the landing gear did not extend, and following the incident, the landing gear selector switch was found in the down position, and the landing gear circuit breaker was in the pulled (deactivated) position. It is likely that the circuit breaker was inadvertently pulled by the pilot as he tried to disengage the autopilot, and as a result the landing gear did not extend as he had thought. This misdiagnoses further highlights the stress the pilot was experiencing. No discrepancies were observed in the airplane’s flight control system, and an important service bulletin related to pitch trim runaway had been applied. Examination of the autopilot system found that the left side rocker of the manual electric trim switch would intermittently fail to return to the neutral position after use. In this state, the trim servo solenoid remained engaged, which locked the trim wheel. However, pulling the pitch trim circuit breaker would have released the solenoid and resolved the anomaly. The damaged rocker switch was therefore likely the reason the pilot was unable to manually adjust the airplane trim during the emergency. Disassembly of the pitch servo revealed that a bearing in its drive motor had seized and was causing the servo to intermittently stop operating. The design of the servo was such that in the event that it locks, the motor assembly can activate the pitch trim servo to compensate, which could result in a mistrim configuration when the pilot disengages the autopilot. Other possibilities exist that could cause the airplane to be in a mistrim condition with the autopilot enabled, including the pilot manipulating the controls or adjusting the engine speed with the autopilot engaged; however, the pilot stated that he did not do this during the flight. To prevent deactivation of the autopilot under a mistrim condition, the flight manual specifically calls for pilots to check the command bar on the flight command indicator prior to disengagement. There is no evidence the pilot did this and, as a result, was likely caught off guard by the mistrim condition. Additionally, the pilot did not perform the autopilot pre-flight check prior to flight, which may have alerted him to the anomalies with the trim switch and would have alerted him to another unrelated problem with the pitch trim servo, which was operating slower than required. Even if the pilot disconnected the autopilot without checking the command bars, the required control forces, while high, would still have been controllable. However, the trim position as found after the incident would have required an extreme application of force (almost 300 lbs) by the pilot at the control wheel to maintain a neutral elevator position, an amount well beyond what the autopilot system could command. Therefore, it is likely that this trim finding was a result of the autotrim system reacting against the pilot’s control input while he attempted to maintain pitch control. It could not be determined why the autotrim system did not completely disengage during this phase of the flight. The pilot’s previous experience with a throttle failure event likely affected his ability to accurately diagnose the cause and identify an appropriate remedy for the multiple conflicting system cues and failures in a timely manner; his performance was further degraded by the physical workload required to maintain control of the airplane.

The airplane was examined by representatives from the NTSB and FAA, along with a technical specialist from Honeywell Aerospace, the autopilot manufacturer. Landing Gear Immediately after the incident the landing gear selector switch was found in the down position, and the landing gear and auxiliary fuel pump circuit breakers were both in the tripped (pulled) position. During recovery of the airplane the landing gear doors were all found completely closed, with no damage or gaps that would have indicated a partial extension. The landing gear and fuel pump circuit breakers were recycled, and the landing gear and fuel pump were tested with the airplane configured on jack stands. No anomalies were noted, and both the landing gear and fuel pump functioned correctly. Of note, the instrument panel lighting rheostat was found in the “dim” position, and as a result, determining the landing gear light status in the daylight condition was difficult. The findings were discussed with the pilot, who stated that it was possible that while he attempted to disable the autopilot, he may have accidentally pulled the landing gear and fuel pump circuit breakers, and the sound of what he thought was the landing gear extending, may have been the flaps operating. He also stated that due to the cabin lighting and urgency of the situation, he may have misinterpreted the landing gear indicators. Autopilot Operation The airplane was equipped with a Bendix/King KFC 200 flight control system (FCS). The FCS incorporated a two-axis (pitch and roll with altitude hold) autopilot and a flight director system. Pitch control was accomplished through a servo which drove the elevator via a bridle cable attached directly to the elevator control cables. A pitch trim servo drove the elevator trim tabs via a bridle cable attached directly to the elevator tab control cables. System capabilities included manual and automatic electric pitch trim, altitude hold, vertical trim, multiple nav capture and track modes, and control wheel steering (CWS). Pilot interface with the FCS was via both the panel-mounted KC 290 mode controller and several switches on the left horn of the pilot's control wheel. Self-test, mode select/engage, and vertical trim were controlled via the KC 290. The A/P DISC. TRIM INTERRUPT, CWS, and dual rocker electric pitch trim switches were on the control wheel. The automatic trim allowed the KFC-200 system to trim off elevator control surface pressures while the autopilot is controlling the elevator through the pitch servo. If the autopilot is not engaged, the pilot can use the manual electric trim switch or the trim wheel (located in the center console) to trim off elevator control forces. FAA certification regulations require that the autopilot must be capable of being manually overridden by the pilot, by using the control wheel. Autopilot Examination on the Airframe The red cap from the autopilot A/P DISC. TRIM INTERRUPT button on the left horn of the pilot’s control wheel had broken away from the switch and was found in the footwell immediately after the incident. The pilot stated that it likely came off as he attempted to gain control of the airplane. With the red cap removed, the switch plunger was free to sway within the switch case and required an asserted effort to activate. The elevator trim tab gauge indicated "18U" (18° nose-up), and the elevator tabs were found in the matching tab down (nose-up) position. The maximum trim tab positions were 27 ° nose-up, and 10 ° nose-down respectively. According to representatives from Textron Aviation, with an elevator trim setting of 18U, and an airspeed of 150 knots, the pilot would have needed to apply 296 lbs of force to the control wheel to hold the elevator in the neutral position. The entire elevator flight control and trim system was examined. All control cables were intact, and there were no indications of binding, failure, or foreign object interference. With the A/P DISC. TRIM INTERRUPT switch cap re-installed, and a metered vacuum source connected to the airplanes vacuum system, a pre-flight check was performed in accordance with the KFC-200 flight manual supplement and all autopilot annunciators illuminated, but the “autotrim trim failure” light flashed once, rather than four times as required. The A/P DISC. TRIM INTERRUPT switch was then reseated by movement, and the check was performed again, and no anomalies were noted, except section IV (7e.3) (manual electric trim run through full travel), which took 64 seconds to compete, rather than the nominal time of 34 to 50 seconds. The elevator control trim wheel could be moved easily by hand, and it was not locked as the pilot had indicated. However, during the examination, it was found that the left side rocker of the manual electric trim switch would intermittently fail to return to the neutral position after the trim-down position was selected. In this state, the trim servo solenoid remained engaged, and the trim wheel could not be moved. Pulling the trim circuit breaker disengaged the solenoid and allowed for manual movement of the trim wheel. Review of the elevator trim system electrical schematic revealed that with the left manual electric trim switch jammed in the fore or aft position, the trim solenoid would remain engaged, but the trim motor would not run. Autopilot Component Examination The autopilot drive and control components were removed and examined at the facilities of Honeywell Aerospace under the oversight of the NTSB. Examination revealed two discrepancies: the intermittent sticking of the left manual electric trim switch, and seizure of the KS270A Pitch Servo drive motor, which did not operate during its first test. Disassembly of the K270A unit revealed binding of its motor shaft, which could be overcome by rotating the tachometer gear end of the motor by hand. Once freed, the anomaly could not be repeated, and the K270A operated within specification. Using a Honeywell testbench mockup of the 200-series autopilot system, a simulation was performed to see how the autopilot would react to a jammed pitch servo motor. The system was tested in FD, ALT, AP modes, and the tachometer drive end of the pitch servo motor was held immobile. It was found that under this condition, if the capstan wheel assembly was rotated to engage the pitch trim microswitches, after a 3 second delay, the pitch trim servo would start to move. No errors or enunciations were trigged at any point during the test. The left-side of the trim switch was then activated in an attempt to simulate the findings on the airplane, and the pitch trim solenoid engaged. With the solenoid engaged the A/P DISC. TRIM INTERRUPT switch was pressed, and the solenoid immediately disengaged. It was also found that shutting off the avionics master switch disengaged the solenoid. The system was then set to FD mode, with the pitch servo motor still locked. Pressure was applied to the pressure port, and the pitch trim servo moved in concert with pressure changes. The pitch servo drive motor was subsequently removed and examined at the facilities of its original manufacturer, Allied Motion Technologies (previously Globe Motors, Inc.). Examination revealed that although the motor exhibited minimal wear signatures, the shaft bearing on the commutator side (tachometer end) had seized. It was found that the motor could be forced to rotate by hand, but only because the drive shaft was slipping within the bearing’s locked inner race ring. The motor was manufactured in 1992, and according to representatives from Allied, it had not been returned for service since manufacture. There is no life limit or recommended overhaul schedule for the motor, and Allied records did not indicate a similar failure had occurred to this type of motor before. According to the Honeywell representative, the autopilot does not specifically warn of a