DCA24LA328

NTSB incident investigation · Cirrus Design SF50 · Austin, TX · September 14, 2024

Incident · No injuriesClass 3 · LimitedFinal report
Date
September 14, 2024, 12:48 local (17:48 UTC)
Place
Austin, TX · ADSB position at same time of EA activation in on board data
Injuries
1 uninjured
Damage
Minor damage
Phase of flight
Approach
Aircraft
N442CVCirrus Design SF50
Defining event
Approach-IFR initial approach: Midair collision
Weather
Visual (VMC) · Day
Status
Completed · report October 22, 2026
Report
Approved by Office Director
Investigation class
Class 3
NTSB office
Major Investigations Division
Parties
Cirrus Aircraft, FAA/AVP110
Case key
195334

From the NTSB’s public API. By law, NTSB reports can’t be used as evidence in lawsuits over the incident.

On this page

Probable cause

The unintended activation of the Emergency Autoland button on the overhead panel due to the airplane’s collision with an unknown object. Contributing to the incident was the pilot’s inability to initially deactivate the Emergency Autoland system due to a software malfunction.

Findings · 2

NTSB’s coded findings. Those marked are part of the probable cause; the rest are factors or other findings. Under each, the cases with the same finding.

Object/animal/substance: aircraftEnvironmental issues › Physical environment · Unknown/Not determinedCirrus Design SF50: 1 caseAll aircraft: 116 casesProbable cause
Auto flight system: autopilot systemAircraft › Aircraft systems · MalfunctionCirrus Design SF50: 2 casesAll aircraft: 28 casesProbable cause

Sequence of events

The events NTSB coded, in order, by phase of flight. The defining event is the one that best describes the accident. Under each, the cases with the same event anywhere in their sequence.

Analysis

During approach to land with the autopilot engaged, in visual meteorological conditions, the airplane struck an object while at a height of about 1,540 ft above the ground, and at a speed of about 170 knots, as the airplane began a left turn toward the final approach course. Just prior to the strike, the pilot briefly saw a white object through the right (copilot side) windscreen, which he initially surmised was a bird. He described the impact as an “extremely loud” booming noise and was like hitting a “Mack truck”. He checked the windscreen for damage and confirmed that the airframe parachute system had not deployed. The airplane’s Emergency Autoland (EA) system had activated, likely due to the collision, and he overheard repeating automated emergency radio transmissions in his headset announcing the airplane’s callsign and declaring an emergency (a function of the EA). These transmissions interfered with his communications with the air traffic control tower, and his initial attempts to deactivate the EA system by pressing the autopilot disconnect button (AP DISC) several times, were unsuccessful. He noticed damage to copilot side sun visor, the copilot side oxygen mask had fallen from its stowed position, a fire extinguishing switch was damaged, the cabin pressure dump switch had activated, and the transponder had changed to the emergency code of 7700 (another feature of the EA system). On-board recorded data indicated that the EA system was ultimately deactivated, though the pilot was unsure if or when it had. He reported that he had no difficulty controlling the airplane, which responded to his flight control inputs as well as flap and landing gear selections, when and as expected. He was able to land the airplane without further incident. Postlanding photographs showed white, red, and blue paint transfer marks leading from the lower left edge to the upper center of the right windscreen, which was inconsistent with a bird strike. The object struck was not identified, as explained later below. The airplane’s on-board recorded data indicated that the EA system had activated twice, in succession. The data indicated neither of the autopilot modes that are a prerequisite for an automatic activation of EA were engaged either time, suggesting that the activations were initiated by the manual button in the overhead panel. The first activation occurred nearly simultaneously (within 1 second or less) when the cabin pressure dump system activated, and likely at the same time as the equipment disturbances in and near the overhead panel. This, along with 1) post accident testing of an exemplar EA button revealed that it would activate when subjected to shock and 2) the pilot’s testimony that he never pressed the EA button, suggests that the first EA activation was most likely due to a shock load to the button during the object strike. However, the on-board recording system did not detect a signal from the button at that time. This was likely due to therecorder’s 1 second sample rate for the EA button status. The EA button (as well as the AP DISC button) is a spring-loaded momentary contact type, which does not remain pressed or latched when pushed. A brief actuation of either of these buttons, shorter than 1 second, may not be detected by the recording system. About 10 seconds after the first EA activation, the recording system detected a signal from the AP DISC button for a single recoding sample. However, the EA did not deactivate. Post incident research and testing by the airframe manufacturer using exemplar avionics revealed two repeatable software defects related to the AP DISC button function. First, the AP DISC button had no effect if pressed during a certain unquantified period of time between about 8-10 seconds after the start of an EA activation. This may explain why the EA system did not deactivate in response to the (single) recorded AP DISC button press. The AP DISC button status was also recorded at a 1 second sample rate. Had the pilot pressed the AP DISC button more than once during the ‘defect’ period, it would likely not have deactivated EA. Further, multiple brief presses of the AP DISC button could have occurred and not been recorded due to the 1 second sample rate. Second, the aural autopilot disconnect warning tone, which is normally sounded through the airplane’s audio system whenever the autopilot is disconnected, would not sound if the AP DISC button was pressed at any time while EA was active. The on-board recorded data indicated that first EA activation was terminated about 20 seconds after it began. This may have been in response to a press of the AP DISC button by the pilot, but if so, the press must have been too brief and between recording samples, as none were recorded. The second EA activation began about 3 seconds after the first one was terminated. At that time, the recoded data captured a signal from the EA button, indicating it had activated. Examination of the EA button assembly did not reveal any obvious damage, though the button was found to be “slightly loose” compared to an exemplar unit. As noted above, an exemplar button was found to be susceptible to activation when shocked, however no apparent source of shock was identified at the time of the 2nd activation. The pilot was adamant in his interview that he did not press the button at any time. The investigation was unable to determine what led to the 2nd button activation. This EA activation was terminated about 20 seconds after it began, and the recording system did not detect a press of the AP DISC button. As noted above, it is possible for brief press of the AP DISC button to not be recorded, however this could not be confirmed. The investigation was unable to identify the object that collided with the incident airplane. The potential for collision with an unmanned aircraft system (e.g. drone, model aircraft) or an experimental balloon were considered. The strike occurred about 30 seconds afterthe airplane passed by a hobbyist remote control model aircraft field, which was an FAA approved location for unmanned aircraft system (UAS) operations. At the time of the strike, the airplane was 1.8 nm northwest of the model aircraft field, at an altitude of 1,540 ft above the ground. This was well outside the field’s lateral dimensions, and about 1,140 ft above its 400 ft altitude restriction. Given the lateral distance from the model aircraft field, it is unlikely that a remote pilot located at the model aircraft field was operating a UAS flying at the location of the strike, with direct visual observation of the UAS (by either a remote pilot or a visual observer), as required by the FAA and the model aircraft club operating rules. The safety officer for the model aircraft field reported that a polling of club members revealed no known conflicts with the incident airplane. Whether or not a UAS was being operated from the model aircraft field or from any other location, either without direct visual observation, or because of a lost radio communication, could not be determined from the available evidence. While the FAA requires that some UAS be operated with a remote identification (RID) system, a representative from the FAA reported that: there are no archived flights of drones or RID broadcasts remote pilots do not file flight plans Drones are not equipped with radar transponders neither drones nor RID message elements appear on air traffic control radar displays The Civil Air Patrol, which had previously launched some experimental balloons from various locations in and around Austin, reported that they had not conducted any launches on the day of the incident.

Factual narrative

On September 14, 2024, at 1248 central daylight time, a Cirrus SF50, N442CV, experienced an unintended activation of its Emergency Autoland (EA) system at or near the time the airplane was struck by an unknown object during approach into Austin Bergstrom International Airport (AUS), Austin, Texas. The airplane sustained minor damage due to the impact, and the pilot was not injured. The airplane was operated as a Title 14 Code of Federal Regulations Part 91 personal flight from Cincinnati Municipal Airport/Lunken Field (LUK), Cincinnati, Ohio to AUS. Object Strike According to the pilot, during a left turn toward the final approach course, with the autopilot engaged, he briefly saw a white object through the right windscreen and then heard an “extremely loud” booming noise. He further described it as sounding like he “hit a Mack truck”. He later noticed that several items in or near the overhead panels were disturbed. The right-side sun visor had broken, and the copilot oxygen mask fell from its stowage area. The cabin pressure dump switch had activated, and one of the “PUSH to DISCHARGE” fire bottle activation buttons had separated from the overhead panel and fell to the floor. The manual EA activation button was located overhead in a panel just aft of the disturbed items. The EA system was designed to autonomously land the airplane, in the event of pilot incapacitation. It can be manually activated via this button, or in some situations can be automatically activated as described later below. Once activated, it manages airplane control, navigation, emergency radio and transponder communications to land at the nearest appropriate airport. Figure 1. Overhead Panel. (Source: Cirrus Aircraft) The pilot initially thought that the airframe parachute had deployed. After he noticed that the copilot side sun visor was damaged, he checked the windscreen for damage (it was scratched but intact), and the parachute status, to determine if it had deployed (it had not). He then heard an automated voice in his headset that was repeating the airplane’s callsign (N442CV) and declaring an emergency, which was his first indication that EA had activated (this automated radio transmission is part of the programmed sequence of events that occur when the after EA is activated). As the automated radio transmissions occurred, they prevented him from communicating with the AUS air traffic control tower. He also noticed that the transponder had been set to squawk 7700 (another item in the EA activation sequence) and the control tower confirmed receiving 7700. He recalled that he had repeatedly pressed the autopilot (AP) disconnect button, however the automated radio transmissions repeated 3-4 times before they eventually stopped. He was then able to inform the tower that he was able to make a normal landing and receive landing clearance. The remainder of the flight was uneventful. After landing, the pilot wiped the airplane down and noted there was an “oily residue” onthe windscreen. He did not retain any samples of the residue. The pilot surmised that the airplane had struck a bird. However, a photograph provided by the airplane manufacturer showed scrape marks and blue, red, and white color transfer marks that extended diagonally from the bottom of the right windscreen near the center post toward the top and outboard side the windscreen, see figure 2. Figure 2. Scrape and color transfer on windscreen. (Source: Cirrus Aircraft) A review of Federal Aviation Administration (FAA) automatic dependent surveillance broadcast (ADS-B) airplane tracking data, revealed that the airplane had flown past the Austin Radio Control Association (ACRA) model aircraft field, located about 8 nautical miles northeast of AUS. As the airplane flew to the west on a track of about 268°, it was abeam the model aircraft field at 1247:08 which was about 0.65 nautical miles to the south. At that time the airplane’s altitude was about 2,300 ft above the ground, according to on-board recorded flight data from the airplane’s radio altimeter. Figure 3. Airplane track – (direction of flight was west followed by left turn to the south). According to the ADS-B and on-board recorded flight data, the EA activation and cabin pressure dump system activation occurred about 30 seconds after the airplane passed by the model aircraft field, and at that time was about 1.8 nautical miles to the northwest. The EA function activated as the airplane was at a radio altitude of 1,540 ft above the ground at speed of 167 knots. Figure 4 shows the location of the activation relative to the model aircraft field location and size. Figure 4. Strike location. (Image source: FAA, annotation NTSB) The safety officer from ACRA reported that they had polled their membership (about 120 total members, 20-30 of which fly on a regular basis), and none reported knowledge of, observing, or involvement with N442CV. They also advised that all ACRA operations of model aircraft, also known as unmanned aerial systems (UAS) were, by the Academy of Model Aeronautics rules and federal aviation regulations, below 400 ft above ground level. Most of the members aircraft weigh under 15 lbs. and none exceed 35 lbs. In the event of a loss of radio communication, most of the aircraft flown at the field are configured to fly in a 360° circle and hold altitude, until a connection is re-established. First person view operations, in which the remote pilot uses goggles or a display to view video from an on-board camera (and not directly viewing the UAS from their position), are permitted at the field but require a second observer pilot to maintain direct view of the UAS at all times. The FAA requires that certain unmanned aircraft systems (UAS), including some hobby drones and model aircraft, be equipped with a remote ID (RID) device that broadcasts identification and aircraft position. RID is not required by the FAA for UAS operations within approved FAA-Recognized Identification Areas (ofwhich ACRA was one) but does require that RID equipped UAS’s to have the RID operating while flying in such areas. A representative from the FAA reported that there was no archived data of any RID broadcasts available. Representatives from the Civil Air Patrol, which had previously launched some experimental balloons from various locations in Austin, reported that they had not conducted any launches on the day of the incident. Flight data were recovered from the N442CV’s central maintenance computer (CMC) and recoverable data module (RDM). The data showed that the EA system activated and deactivated twice in succession, however details about the initiation and termination of EA were not consistently captured in the recorded data. Emergency Autoland (EA) Activation and Deactivation According to the airplane’s pilot information manual, EA (also referred to as “Safe Return Autoland”) can be activated by: 1) Manually pressing the overhead EA pushbutton. 2) Automatically, if the autopilot’s “Automatic Level Mode” has been engaged for one minute. 3) Automatically, if the autopilot’s Emergency Descent Mode is engaged, and the airplane descends through 15,000 ft pressure altitude. EA can be deactivated by: 1) Pressing either the left (pilot) or right (copilot) AP disconnect button located on the control stick for one second. 2) Pressing the AP button on the autopilot control unit in the instrument panel. According to the on-board recoded data, at 1247:38, while at an altitude of about 1,540 ft above ground level and at a speed of 167 knots, the automatic flight control system (AFCS) mode changed from “selected altitude” mode to “level” mode, and within one second, the recorded CMC parameter EMER AUTOLAND ACTIVATING parameter changed from FALSE to TRUE, indicating that EA waiting/warning period mode began, and a warning message was displayed on instrument panel advising of the activation. In addition to this messaging, by design, the waiting/warning period lasts for 10 seconds, during which the autopilot will enter the “level” mode (as noted above), an automated audio cue announcing “Emergency Autoland Activating” will sound on the airplane’s audio system (through the headsets and cockpit speaker), and a visual annunciator next to the manual activation button will illuminate. The EA system will then wait until the end of the waiting/warning period, before activating other EA functions, including the automated radio calls and transponder tuning. However, at the time of EA initiation, there were no other indications in the CMC or RDM data that either of the two automatic means of initiation had occurred. Further, the RDM parameter that captured the status of the manual activation button did not indicate that the button had been pressed. The manual activation button is a spring-loaded momentary contact style, which after pressed, returns to its normal resting position. Additionally, the RDM parameter that records the button’s status (pressed or not pressed)is sampled at 1 Hz. As a result, if the button were pressed (or, activated by some other means, such as due to the collision) briefly, and between the 1 second samples, it may not have been captured in the recorded data. The data indicated that the waiting/warning period lasted for about 10 seconds, as designed. The CMC parameter that monitored the status of the autopilot disconnect (AP DISC) buttons, which were also momentary contact style and recorded at a rate of 1 Hz, indicated that AP DISC button was pressed during a single recorded sample at 1247:48, just prior to the end of the EA waiting/warning period. However, the EA system did not deactivate. After the waiting/warning period, the EA continued to the next phase of full activation and continued for another 10 seconds, before it was deactivated. The AP DISC parameter did not capture a button press at the time EA deactivated. About 3 seconds later, the EA waiting/warning period activated again, and this time the recorded data showed that the overhead manual activation button was pressed. The warning period was active for 10 seconds, followed by the full EA activation, which interrupted about 10 seconds later. When EA deactivated, The AP DISC parameter did not capture a button press. The pilot reported that he did not press the EA button at any time. The EA overhead panel and button were removed from the airplane and examined at the airframe manufacturer’s facility under the oversight of the FAA. The examination revealed that when pressed, the button would activate electrically before it travelled its full mechanical stroke (the electrical contacts would close when the button was only partially pressed). The button was found to be “slightly loose” when compared to an exemplar unit, though no obvious damage was noted. The exemplar button would consistently activate briefly when subjected to shock loading directed parallel to the axis of the button’s motion. Research and testing performed by the airplane manufacturer revealed two software programming defects in an exemplar integrated avionics system. The first was that during the EA activation sequence, there is a short window of time near the end of the 10 second warning period, during which the system would not respond to an AP DISC button press. If the button was pressed during the window, EA would not deactivate. If the button was pressed before or after the window, EA would deactivate. The precise duration of the window was not determined; however, it occurred between 8 and 10 seconds after initial activation of EA. The second defect was that the autopilot disconnect warning tone, which is designed to sound through the airplane’s audio system whenever the autopilot is disconnected, would not function if the AP DISC button was pressed at any time while EA is active. The airplane manufacturer advised that these 2 issues are scheduled to be addressed in a future software update.

Aircraft

Registration
N442CV
Reported as
Cirrus Design SF50 (NO SERIES)
Category
Airplane
Serial number
0442
Built
2023
Engine
Williams FJ33-GA/397147, 2,000 lbs thrust, turbo fan
Damage
Minor damage
Flight
Personal · Part 91
Registered owner
Pp100 LLC
Airworthiness certificate
Standard, Normal
Departed
LUK · Lufkin, TX
Destination
Same as departure
Flight plan
IFR
Airport
AUS · Austin-Bergstrom Intl, runway 18L, 9,000 × 150 ft, concrete
Equipment
ADS-B, Airframe parachute, Angle of attack indicator, Auto-pilot, Electronic multifunction display, Electronic primary flight display, Onboard weather, Stall warning system

Matched to the registry by N-number and the event date; the registry’s history has gaps, so older matches can be wrong. Compare the reported make and model.

Flight crew

As NTSB recorded them. NTSB leaves out names.

Role
Pilot
Age
69
Injuries
None
Certificate
Private
Airplane ratings
Multi-engine land, Single-engine land
Instrument ratings
Airplane
Medical
Class 3, With waivers/limitations
Last flight review
May 24, 2424
Seat
Left
Restraint
3-point
Flight hours
20,000 total · 225 in this make and model · 35 in the last 90 days · 15 in the last 30 days · 3 in the last 24 hours

Weather

Conditions
Visual (VMC) · Day
Wind
200° at 10 knots
Visibility
10 statute miles
Lowest ceiling
25,000 ft broken
Temperature
91 °F, dew point 70 °F
Altimeter
29.91 inHg
Observed at
KAUS, 9 nm away
METAR
METAR KAUS 141753Z 20010KT 10SM FEW150 BKN250 33/21 A2991 RMK AO2 SLP115 T03330211 10339 20233 58004=

Docket · 12

The files NTSB released with the investigation on October 22, 2026, on NTSB’s docket system: statements, records, photos and specialists’ reports. 22.6 MB in all.