NYC07FA083

The instrument-rated private pilot obtained a weather briefing and filed an instrument flight plan for a personal flight that would occur the following morning to attend a stock car race. On the morning of the accident, the pilot contacted air traffic control (ATC) to obtain an instrument clearance. The pilot was given a clearance and was told that it would be void in 10 minutes. Two minutes prior to the expiration of the ATC void time, the pilot departed.

Recorded non-volatile memory (NVM) data from the electronic cockpit flight display system were downloaded and analyzed by investigators, and these data greatly aided in the investigation. Analysis of the data indicated that, while climbing in instrument conditions, and most likely using the autopilot, the pilot attempted to increase the airplane’s climb rate by using the vertical speed bug on the electronic (glass) Primary Flight Display (PFD) in the cockpit. However, the NVM data revealed that the PFD had failed to “align”, and was not fully functional. As a result of the PFD not being aligned, the bug settings (altitude and vertical speed) were suppressed, and not transmitted to the autopilot computer. Therefore, when the pilot attempted to increase the airplane’s vertical speed using the PFDs vertical speed bug, the changes were not transmitted to the autopilot computer and the autopilot did not command an increase to the rate of climb. Subsequently, the rate of climb momentarily increased twice, most likely due to pilot control inputs via the control wheel. (A review of the PFD Pilot's Guide revealed that it does not provide information to the pilot regarding the suppression of the bug settings when the PFD is not aligned.).

Soon after the last vertical speed bug change, the airplane’s vertical speed began to fluctuate (and oscillate between climbing and descending) significantly. These fluctuations continued for the remainder of the flight (about 9 minutes). During this period, the airplane completed three left-hand orbits at varying altitudes. The pilot then reported an “autopilot problem” to ATC, and he requested a clearance to divert to a nearby airport. The airplane’s course deviated again to the left, after which the pilot again reported a problem with the autopilot. The airplane continued to fly for another 4 minutes as the vertical speed continued to oscillate. The airplane’s ground track remained erratic and was not consistent with a course toward the diversion airport. The airplane entered an uncontrolled descent, impacted terrain, and was destroyed.

Examination of the wreckage revealed that the pitch trim jackscrew was in the full nose-down position. No preimpact malfunctions of the autopilot or flight control systems were discovered. The nose-down trim condition likely resulted from the pilot pulling on the control wheel with the autopilot engaged which would have resulted in the trim running nose down. Exams of the airplane’s other systems, structure and engine did not reveal any evidence of preimpact malfunctions or failures.

Risk factors for spatial disorientation were present at the time of the accident, including IMC and maneuvering flight. Analysis of the airplane’s flight path during the accident flight, and the pilot’s communications with ATC, indicates that the pilot became spatially disoriented while attempting to maneuver with reference to a degraded PFD. The pilot was instrument current and qualified and he had accumulated about 291 hours operating the electronic flight display system. He had received formal training in two models of airplanes equipped with the same flight display system, and also experienced a failure of the display system on a previous airplane. Thus, the pilot was likely familiarized with the functionality of the PFD.

Analysis of the NVM data from the PFD revealed that the PFD had not aligned during its “power cycle” of the accident flight, nor had it aligned on the previous power cycle that occurred on the day of the accident. The recorded data do not explicitly indicate the reason for the failure to align on either cycle; however, the previous power cycle was 333 seconds in duration, and the data indicate that the airplane was moving (taxing) at that time. Movement during this initial, critical alignment phase hinders the system’s ability to align as it attempts to calculate the steady state bias of its rate sensors, and the PFD screen will display a message that states: “INITIAL AHRS ALIGNMENT / REMAIN STATIONARY / OK TO TAXI in xx SECONDS”.

After the 333-second power cycle, the system was shut down and restarted for the accident flight. No movement was recorded during the critical alignment period of this power cycle, although the airplane did taxi during the “OK TO TAXI” period of the alignment sequence. The investigation could not conclusively determine why the PFD would not align; however, the airplane departed 375 seconds after the PFD power cycle start-up, which may not have been sufficient time for the system to align. Typical time to align while remaining stationary is 166 seconds. The time to align when taxiing during the “OK TO TAXI” period varies considerably, based on the taxi motion and duration. During a previous taxi event that was similar to the one on the day of accident, the time to align was 416 seconds.

According to the flight display manufacturer, the failed alignment in both power cycles would have resulted in red X’s being displayed on the PFD. Based on the information recovered from the accident unit, it is likely that red X’s were displayed at the time of the takeoff. Regardless, the airplane was equipped with traditional standby flight instruments (specifically an attitude indicator, an altimeter, an airspeed indicator, a magnetic compass), and dual global positioning systems, which should have provided sufficient information for the pilot to have maintained situational awareness to allow for continued safe flight after failure of the PFD.

The investigation revealed a history of numerous failures of multiple types that affect the flight display system’s reliability. Additionally, examination of the accident airplane’s magnetometer also revealed evidence of fretting corrosion in flight critical components which could result in further malfunctions and failures. However, the investigation did not confirm whether any malfunction or failure of the electronic flight display system was present at the time of the accident takeoff, other than the PFD’s failure to align prior to departure.

History of flight

On March 24, 2007, about 0920 eastern daylight time, a Piper PA-32R-301, N324ST, was destroyed when it impacted terrain following a loss of control while maneuvering near Jacksonville, Maryland. The certificated private pilot and two passengers were fatally injured. Instrument meteorological conditions (IMC) prevailed for the flight that departed Harford County Airport (0W3), Churchville, Maryland, about 0906, destined for Virginia Highlands Airport (VJI), Abingdon, Virginia. An instrument flight rules (IFR) flight plan was filed for the personal flight conducted under 14 Code of Federal Regulations (CFR) Part 91.

According to transcripts provided by the Federal Aviation Administration (FAA), the accident pilot contacted the Williamsport Automated Flight Service Station via telephone on the day prior to the accident to obtain a weather briefing and to file an IFR flight plan. During the briefing, the pilot indicated that he was going to fly to VJI so that he could attend a NASCAR race in Bristol, Tennessee.

The next morning, the pilot contacted Potomac Terminal Radar Approach Control (TRACON) to obtain his IFR clearance. At 0858, the pilot was given a clearance and was told that the clearance would be void at 0908 (10 minutes). The pilot acknowledged. Eight minutes later the pilot departed and contacted Potomac TRACON shortly after departure.

The airplane’s Avidyne Entegra flight deck instrumentation was capable of recording some flight data and other parameters. Safety Board investigators downloaded these data, and compared them to air traffic control (ATC) recorded radar and voice data provided by the FAA. The flight deck instrumentation data revealed that the Primary Flight Display's (PFD) Attitude Heading Reference System (AHRS) had failed to align twice on the day of the accident; once on the electrical power cycle preceding the accident flight, and again when the PFD was started up for the accident flight.

According to Avidyne, if the PFD is not aligned, then heading data, Horizontal Situation Indicator (HSI) navigation data, and attitude data are removed from the display and replaced with Red “Xs". The PFD also cannot be used to control the autopilot computer, limiting the available functionality of the autopilot. The PFD would still however, display altitude, airspeed, and vertical speed.

According to the recorded data, prior to takeoff, the vertical speed “bug” on the PFD was set at 750 feet per minute (fpm). The airplane then took off, maneuvered to a westerly heading on track to the Westminster VOR, and at 09:06:37 established an indicated climb rate of about 450-500 fpm. At 09:07:11, the airplane was at an altitude of about 1,440 feet msl and ATC instructed the pilot to climb and maintain 8,000 feet. The PFD altitude bug changed from 5,000 to 8,000 feet. At 09:08:24, the airplane was at an altitude of about 2,000 feet msl and ATC asked “verify you are climbing”. The pilot responded with “say again please.” ATC replied, “climb and maintain eight thousand,” which the pilot acknowledged.

At 09:08:37, the data indicated that the PFD’s vertical speed bug setting changed from 750 fpm to 950 fpm. About 09:09:24, the setting increased to 1,050 fpm. At the same time, the pitch attitude reversed its downward trend to an upward trend. About 16 seconds later, a vertical speed was established at about 1,100-1,200 fpm, which lasted for approximately 43 seconds. During this 43-second period, the pitch attitude continually increased at a relatively constant rate, and airspeed decreased at relatively constant rate from about 115 knots to about 94 knots.

Also during this 43-second period, the vertical speed bug setting reduced twice; first to 850 fpm and once again to 750 fpm. However, the vertical speed remained at 1,100-1,200 fpm. After this period, the vertical speed began to fluctuate, at times reaching peaks of 5,000 fpm down, and 7,500 fpm up. The fluctuation continued until the end of the recording. Altitude, pitch attitude, airspeed, and vertical acceleration also continued to fluctuate until the end of the recording, with trends consistent with the vertical speed trend.

Twenty seconds after the vertical speed fluctuations began, at about 09:10:53, the airplane began the first of three 360-degree turns to the left. As the turn began, the trend of roll attitude was initially to the left followed by a reversal and brief trend to the right before reversing again and remaining left wing down (though oscillating in value) for the duration of the 3 “orbit” turns to the left. During these turns, the vertical speed, airspeed, and pitch and roll attitudes continued to fluctuate. About halfway through the set of turns, at 09:12:15, the pilot transmitted to ATC: “we seem to be having some type of difficulty here could uh you give us assistance please”.

At 09:13:01, after the airplane completed the three turns, the pilot transmitted: “we’re with ya we were off getting back on course here we had a little difficulty with the auto pilot.” At this time, the airplane’s track was approaching its original course line to the Westminster VOR. At 09:13:35 the pilot requested a heading to the Frederick airport, so they could “do a stop”, and he received a clearance to fly a heading of 280 degrees. Also over the next 30 seconds, the airplane’s track stabilized at about 274 degrees magnetic. Altitude and airspeed continued to fluctuate during this time; however the pitch attitude was relatively stable when compared to the previous and subsequent pitch oscillations.

At 09:14:08, the airplane again began a turn toward the left. At 09:14:46 the “next waypoint” parameter changed from EMI (the Westminster VOR) to KFDK (Frederick Municipal Airport). At 09:15:23, the pilot again reported to ATC that he was “having some problems with the autopilot”. After the left turn, the airplane did not return to a course toward Frederick, and did not maintain any course/heading for longer than about 12 seconds. Altitude and airspeed continued to fluctuate significantly, with the exception of one period of about 20 seconds where the altitude was relatively stable around 2,600 feet (pressure altitude) and the airspeed remained around 110 knots. Over the last 2 minutes of the flight, the fluctuations in altitude and airspeed appear to be somewhat periodic in nature.

The last recorded data points indicated a pressure altitude of 707 feet, indicated airspeed 148 knots, and a descent rate of 6,824 fpm.

Aircraft information

According to FAA and airplane maintenance records, the accident airplane was manufactured in 2006. The airplane’s most recent annual inspection was completed on March 1, 2007. At the time of the inspection, the airplane had accrued 74.9 total hours of operation.

The airplane was equipped with an Avidyne FlightMax Entegra Integrated Flight Deck, which included two 10.4-inch displays, one of which was the PFD and the other of which was the Multi Function Display (MFD).

A set of standby flight instruments were also installed which consisted of a magnetic “whiskey” compass (attached to the center post of the windscreen), a conventional attitude indicator, conventional airspeed indicator, and a conventional altimeter (all of which were located to the left of the PFD in a vertical line). All could be used by the pilot to fly the airplane in instrument meteorological conditions in the event that the PFD malfunctioned.

Additionally, the airplane was equipped with dual Garmin GNS 430 global positioning systems (GPS), and an S-TEC System 55X autopilot, with full-function heading, navigation, and altitude hold, with localizer/glideslope coupling, and altitude pre-select functions that were integrated into the PFD. Together with the audio control panel, these systems were located to the right of the PFD in a vertical line.

During the most recent annual a