NTSB Factual Report

On November 30, 2001, about 0928 central standard time, a Mitsubishi MU-2B-40, N9052Y, piloted by a private pilot, sustained substantial damage during a hard landing on runway 14 at St Paul Downtown Holman Field Airport (STP), near St Paul, Minnesota. The 14 CFR Part 91 business flight was operating on an IFR flight plan. Instrument meteorological conditions prevailed at the time of the accident. The pilot and three passengers were uninjured. The flight originated from Southwest Michigan Regional Airport, near Benton Harbor, Michigan, at time unknown, and was landing at STP at the time of the accident.

The pilot reported in a written statement:

I departed Benton Harbor (B.E.H.) on 11/30/01 at 0800 using

runway 14. It was an hour and half flight to S.T.P. It was an ILS

approach. The glide slope and DME [distance measuring equipment]

were in-op [inoperative]. I was using the autopilot to fly the ILS

localized timed approach. I was looking for the runway and watching

the time. The autopilot inadvertently disengaged, turning the airplane

to the left. I righted the airplane just before touchdown. It landed on

the mains with the nose slightly high. It came down, and the front

wheel and strut broke. The airplane skidded on the nose down the

runway. I kept it on the runway by using the brakes.

The Federal Aviation Administration (FAA) issued a ferry permit to allow the airplane to be flown to Green Bay, Wisconsin, for repairs.

The pilot reported in another statement:

After receiving a ferry permit from the FAA, I departed St. Paul, MN

the morning of December 17, 2001, to fly my MU-2, N-9052Y to

Green Bay, Wisconsin for repairs of damage to the aircraft in a hard

landing at St. Paul Minnesota, as stated above. This flight was to be

flown VFR, day, single pilot only. About 10 miles east of St. Paul

and level at 2,500 feet, I engaged the autopilot. A few minutes after

the autopilot was engaged, the nose of the aircraft pitched down and

turned sharply to the right. I immediately attempted to disengage

the autopilot, which was futile, and told the traffic controller of my

problem. He in turn reported that a small airport was to my left and

behind my current position. I saw the airport ) Lake Elmo (21D))

and in order to fly the airplane to the airport, I had to use extreme

backpressure, left rudder, plus differential power to maintain

enough control to reach the airport and land on a 2,497 foot runway

without using flaps or reverse prop and with no damage to the

airplane. ... After I got the aircraft back to Green Bay, Wisconsin, I

took it to [a repair station there] where [mechanics and the FAA]

each inspected the autopilot system. They were able to duplicate the

runway trim on the ground where it was determined that the problem

of the autopilot trim runaway was a faulty connection in the autopilot

computer.

FAA inspectors, safety investigators from the airplane manufacturer, and representatives from the repair station examined the airplane. The repair station stated:

The autopilot system was engaged and disengaged several times and

all flight director modes engaged. Mode Annunciator test function,

all lamps test and illuminated when mode selected and coupled to

autopilot.

Electric manual trim and automatic pith trim moved elevator trim

tab in the correct direction and speed.

The servos in pitch, roll, and yaw had adequate strength and speed

to move the respective control surfaces. No mechanical binding or

slack noted in rigging or capstans and servo mounts.

The attitude gyro Sperry VG14A (primary attitude reference for

M4D autopilot, FD112V flight director and weather radar) was

removed from its mount for field troubleshooting.

Voltage measurements were taken at the autopilot computer, of 200mv

per degree for pitch and roll output from vertical gyro referenced on

flight director. Readings corresponded with pilot's primary attitude

reference with respect to pitch and roll.

Autopilot system performed satisfactory until the autopilot computer

model 5536F part number 400495-8501 was moved in its shock

mount at which time the yoke was pulled violently right in roll axis

and forward in pitch axis. This finding is consistent with [the pilot's]

experience, as you reported to us, during landing on November 30,

2001 had the autopilot remained engaged. This malfunction was

repeated five times ... However during closer examination of the

autopilot computer connections pin 53 of jack 2 of autopilot

computer was found to be oversized and discolored in a manor

consistent with electrical arcing. Pin function described as servo

motor power in the M4D maintenance manual.

The MU-2 airplane flight manual's emergency procedures stated:

In case of emergency, the autopilot can be overpowered manually to

correct the attitude, but the autopilot must immediately be disengaged.

If the autopilot remains engaged, the autopilot will trim the airplane to

oppose the pilot's actions. This could result in a severely out of trim

condition.

The manual stated:

The autopilot can also be disengaged by any of the following methods:

1. Operate trim switch UP or DOWN

2. Position inverter switch to OFF momentarily and then return

to original position

3. Turn off the ELECTRIC MASTER SWITCH

The pilot reported he accumulated 11 hours of pilot in command time in the accident airplane make and model.

Parties to the investigation were the FAA and Mitsubishi Heavy Industries America, Inc.

NTSB Probable Cause

The autopilot's computer connections arcing and reported disconnection during the localizer approach and the pilot not performing a go around prior to the hard landing. A factor was the pilots lack of total experience in the accident airplane's make and model.