NTSB Factual Report

On August 4, 2002, at 1415 central daylight time, a Cessna 172S, single-engine airplane, N431ME, was substantially damaged when it impacted terrain after a loss of control during takeoff/initial climb near Marietta, Oklahoma. The airplane was owned and operated by the pilot. The private pilot and his two passengers received serious injuries. Visual meteorological conditions prevailed for the 14 CFR Part 91 personal flight, and no flight plan was filed. The flight was originating at the time of the accident, and was destined for Broken Arrow, Oklahoma.

In a telephone interview with the National Transportation Safety Board (NTSB) investigator-in-charge (IIC) on August 12, the pilot stated that prior to takeoff on runway 35, the flaps were set to 10 degrees and the autopilot was not engaged for the soft field takeoff procedure. He further stated that the airplane rotated about one-third of the way down the runway, and shortly after liftoff it pitched up about 10 degrees and entered an uncommanded right turn with about a 20-degree bank angle. The pilot further stated that he didn't immediately respond to the situation "as I was stunned by what was happening." The pilot reported that he did not see the autopilot light illuminate and did not think to disengage the autopilot. He then said the aircraft entered a stall, the left wing dropped, and the airplane hit some trees.

According to the Pilot/Operator Aircraft Accident Report (NTSB form 6120.1/2), received by the NTSB IIC on August 21, the pilot reported that after a normal preflight he taxied down to the extreme south end of the runway, noting the wind sock was indicating no wind. He further reported, "A north departure was chosen since this field slopes down to the north." The pilot also stated that the takeoff roll was normal, with rotation delayed until "near the end of the runway." The pilot further reported that shortly after takeoff "the autopilot unexpectedly engaged and caused the airplane to enter a climbing turn to the right." The pilot also reported that he manually overpowered the autopilot but had entered a stall to the left, which caused the aircraft to impact tree tops.

In three written statements, two provided to the NTSB IIC, the other submitted to an FAA inspector, the passenger who occupied the right front seat, a commercially rated pilot who was currently working on his flight instructor certificate, reported setting the flaps to 10 degrees, which was confirmed by the pilot. He further reported that the takeoff roll began at the south end of the runway and that the pilot maintained back pressure to keep the nose wheel off the ground. "The aircraft accelerated normally and became airborne just after the main wheels crossed a paved road that runs perpendicular to the runway." He also reported hearing a "thump, thump" from the main gear as the aircraft crossed the road that bisects the grass strip, and at the same time the stall warning sounded intermittently for 3 to 4 seconds, and then stopped for 2 to 3 seconds. The passenger also stated that after the brief silence, the stall warning began to sound again, "only this time it was a continuous warning rather than intermittent. When this continuous warning sounded, I turned to my left to observe the airspeed indicator and noticed that the aircraft was in a definitive pitch up attitude." The passenger also reported that he continued to watch the airspeed decrease as the stall warning continued its aural warning for about 8 to 10 seconds, after which the right wing "dipped" [dropped], "and it was then I observed the pilot pull back on the controls and knew our goose was cooked. The aircraft 'fluttered' [drifted] towards the west for a few seconds when the left wing began clipping trees. At some point the aircraft struck a branch or trunk large enough to whip it [the aircraft] ninety degrees to the left." The passenger also reported that he had wished he had pushed forward on the controls, and that he recognized the aircraft appeared to react exactly as he had experienced in flight training during "power on departure stalls." The passenger also stated that he was sure the autopilot wasn't on for takeoff, but wasn't sure if the pilot might have inadvertently engaged it after takeoff.

In a written statement provided by the rear left seat passenger to the NTSB IIC, the passenger reported from the time she got in the airplane the pilot seemed to be in a hurry. She also stated that it seemed strange to her that he didn't do a "run up" before takeoff. She further reported that just as they were getting ready to takeoff, the right front seat passenger asked the pilot, "10 degrees of flaps?" The pilot replied, "Oh yea, 10 degrees of flaps." The passenger also stated that she felt a hard bump on the runway. "I thought we had run over something. I looked back over my left shoulder and saw the road."

An onsite examination of the aircraft was conducted by an FAA inspector and Parties to the NTSB investigation. The angles of the tree limbs that were severed and the angle of impact marks to the left wing suggested the aircraft impacted trees in a 15-degree nose down and 50-degree left bank attitude. There were numerous sections of 3 - 4 inch diameter limbs ranging from 12 inches in length to several feet in length, which exhibited propeller cuts to each end of the severed sections of tree limbs. Flight control cable continuity was confirmed to all primary flight control surfaces from the cockpit to the control surfaces. Flaps were extended 10 degrees. There was no post-impact fire. The seats, seat rails and seat restraints were not damaged. The cabin environment was intact with no buckling signatures.

The airplane's flight control system consists of conventional aileron, rudder, and elevator control surfaces. The control surfaces are manually operated through cables and mechanical linkage. The aircraft was equipped with a Bendix/King KAP 140 two axis autopilot with altitude preselect, a Bendix/King transponder, dual Bendix/King KX 155A Nav/Comm radios, an a Bendix/King KLN 89 GPS navigation computer.

The KAP 140 autopilot system is comprised of a computer/controller, interconnected to the GPS and navigation system displays, pitch and roll actuator servos, control yoke switches, and an electrical system pull-type circuit breaker. It provides an autotrim feature during autopilot operation, and manual/electric trim when the autopilot is not engaged. The electric trim system is designed to be fail safe for any single inflight trim malfunction. Trim faults are visually and aurally annunciated. A lockout device prevents autopilot or electric trim engagement until the system has successfully passed a preflight self test. An automatic preflight self test begins with initial power application to the avionics bus.

The autopilot system may be activated by pushing the autopilot engage/disengage button, heading select button, or the altitude hold button.

The autopilot system will disengage if an electrical power failure occurs, an internal autopilot system failure, pitch accelerations in excess of plus 1.4g or less than plus .6g (if produced by a servo runaway, not by maneuvering), a turn coordinator failure, activation of the roll or pitch axis annunciator, activation of the A/P Disconnect/Trim switch on the control yoke, pulling out the autopilot system electrical circuit breaker, or turning off the avionics master or airplane electrical master switch.

The autopilot system components were functionally tested at the facilities of their manufacturer, Honeywell Aerospace Electronic Systems of Olathe, Kansas, under FAA supervision on October 22, 2002. The components tested comprised the KC 140 flight computer, KCM 100 configuration module, KS 270C pitch servo, KS 271C roll servo, and KS 272C pitch trim servo. According to Honeywell, the components were tested both as a system (connected together as a system test harness), and individually.

The first tests was to recover stored error codes from the previous power cycles, and to thoroughly test the autopilot as a system to observe and document any unusual characteristics, particular attention being paid to the possibility of autopilot self-engagement or uncommanded autopilot or pitch trim operation. No unusual operation was noted, the system performed as designed, with no self-engagements or unusual characteristics of operation of any of the servos. When the system was powered for the first time on the test bench, the current power cycle was 213. The most recent power cycle where any error codes were logged was power cycle 192. The power cycle of the accident flight would have been 212, and as previously mentioned, had no error codes logged. None of the error codes logged on previous power cycles (flights) would indicate a self-engagement of the autopilot.

The autopilot components were then tested individually, as they would be in production test or following repair or overhaul. The testing of each unit was observed by an FAA Flight Standards District Office inspector. For test results refer to the Honeywell KAP 140 Autopilot tests, attachments to this report.

The Honeywell test report concluded that upon completion of the previously described testing, both as a system and as individual components, the autopilot system demonstrated compliance with all of the design requirements and all of the components tested met all of the required specifications for a new unit. No anomalies were noted during any phase of the testing that would indicate the system had previously experienced a failure that might cause autonomous engagement or uncommanded servo drive.

NTSB Probable Cause

The pilot's failure to maintain adequate airspeed which resulted in a stall.