The main component of the ACARS system is the CMU. The CMU provides the receive-and-transmit interface through the VHF-3 transceiver and SATCOM for uplinked and downlinked messages. The CMU controls the VHF-3 communication system in data mode and the SATCOM system data link. The primary transmission medium is VHF. The CMU will automatically change to SATCOM when VHF becomes unavailable owing to ground station saturation or insufficient VHF coverage, and will automatically switch back to VHF when such service again becomes available. All messages sent from the aircraft are referred to as downlinks and all messages from ARINC, SITA or INMARSAT to the aircraft are called uplinks. Each downlink may be received by many different VHF RGSs, depending on the area of coverage. Typically, the RGS recording the strongest signal will provide an acknowledgment as an uplink. If there is no communication between the aircraft and RGS for 10 minutes, the CMU will automatically generate a downlink label "Q0" message referred to as a tracker message for flight-following, which is transparent to the crew and was not forwarded to Swissair (as it is not a chargeable message).
The flight crew access ACARS through one of the three MCDUs installed in the pedestal, which provide a variety of menu options from ACARS initialization to obtaining weather reports. Typically MCDU 3 is used for ACARS operations. All three communications radio panels provide the means to select either ACARS or VHF voice capability on the VHF-3 radio.
Uplink messages are printed on the multiport printer located on the aft pedestal or displayed on the selected MCDU. ACARS alert messages are displayed on the EAD and on the selected MCDU scratch pad line. Any alerts generated by the ACARS system are categorized as Level 0 events, displayed as cyan messages on the EAD or SD . Level 0 alerts are usually operational or aircraft system status information. There are three possible ACARS alerts: ACARS messagewhen data is received, ACARS NO COMMwhen ACARS fails to communicate, and VHF 3 VOICEwhen VHF-3 is in voice mode. As long as communications are possible over SATCOM, the Level0 alerts will not be displayed.
The ACARS VHF radio system operates on the following base frequencies, depending on where in the world the aircraft is operating: 131.55 MHz in the USA, 131.475 MHz in Canada through the Air Canada Dataplus network, or 131.725 MHz in Canada and Europe on the SITA network, and 131.450 MHz in Japan. The service providers may also "auto tune" or change the base frequency of the CMU in areas of heavy traffic.
The SATCOM system uses a geostationary satellite network, a global network of GES and AES to supply long-range, two-way data and voice communications. SATCOM, through the SDU, provides voice and data communication links between the aircraft and ground stations, via satellite, in conjunction with the audio management system, passenger telephone system for voice, and ACARS for data. The SDU also uses inputs from the inertial reference system to obtain position and attitude changes for the antenna steering. If navigation and attitude data from Inertial Reference System -1 or -3 are available when the SDU powers up, the SATCOM system will automatically log on to an SDU-selected GES. All satellite communications for this flight were handled through INMARSAT's Atlantic Ocean West satellite and the LRW GES located in Canada.
ARINC and INMARSAT audit logs for 2 to 3 September for HB-IWF were requested and provided by SITA via Swissair. According to ARINC and INMARSAT, their monitoring systems were functioning normally during the referenced time period, with no problems reported by other users. Furthermore, INMARSAT records show that the satellite telephone service was not used.
The INMARSAT data started with the log-off of the system at 1837:00 on 2 September 1998 when the aircraft was shut down in New York. The next message logged by INMARSAT from HB-IWF was on 2 September 1998 at 2330:18 as the SATCOM system was powered and it requested to be logged on. The log-on request was a Class 3, for voice and data communications, indicating that the CMU was operational.
After approximately 1 hour and 15 minutes, the aircraft requested the GES to be allocated capacity to send an air-to-ground message. The GES allocated the capacity at 0046:55, and the aircraft sent the message, which was acknowledged by the GES at 0047:06. At 0047:07 the GES sent an uplink acknowledgment for the downlinked media advisory message indicating a loss of VHF communications. Approximately seven minutes later, at 0053:56, INMARSAT logged a second (M1X) message, indicating that VHF communications were confirmed lost for greater than seven minutes. This seven-minute delay is a function of a "T7" timer in the CMU. During this time the CMU will go through the operator's requirement table of frequencies in an attempt to make VHF contact with another service provider. This process is repeated until the CMU finds an active frequency or is shut down. The next message was at 0125:52, at which time the aircraft SATCOM system requested to be logged on as a Class 2 (voice only), indicating that the CMU was not available. This log-on was acknowledged at 0126:01. This was the last message received from the aircraft by INMARSAT.
The ARINC data began with the OFF event (a confirmation that the aircraft had departed from JFK) at 0018:00. This message was sent on the third ARINC frequency, JFKA3, which is typically used as a terminal frequency. This message was logged at 0021:00, as it requires approximately three minutes for the CMU to calculate the estimated time of arrival in Geneva (0656:01). This information was sent to Zurich with the OFF event. This message was acknowledged by ARINC at 0021:07 at which time ARINC requested that the CMU change back to the base frequency of 131.55 MHz, which the CMU acknowledged at 0021:11. At 0021:15, ARINC acknowledged the change to the base frequency. This was the last communication with ARINC for 42 minutes, 59 seconds, until 0104:14, at which time the CMU sent a message indicating a change in operating frequency that is normally associated with a change in service provider, in this case from INMARSAT to ARINC. At this time VHF communications were again established with ARINC.
The flight crew used the ACARS service twice to obtain weather information. The first time was at 0113:13 when the crew requested the actual weather (M28 message) for LLSG (Geneva) plus the nine-hour forecast. The weather forecast was received in five data blocks and was completed at 0114:36. At 0114:37, a second weather request (M27 message) was sent requesting the weather forecasts for LLSG, KBOS, KBGR, and CVQM. This latter identifier is not a recognized airport code and most likely should have been CYQM, for Moncton. The weather forecasts for Geneva, Bangor, and Boston were received, but nothing for the CVQM request. This last request for weather was completed by 0115:18. There were no further recorded crew-initiated communications using ACARS.
At 0125:05, a normal Q0 downlink tracker message was sent by the aircraft for flight-following and was acknowledged by an ARINC RGS at 0125:08. If a communication is not acknowledged, it will be re-transmitted 10 seconds later. As the downlink was not re-transmitted, it may be assumed the acknowledgment was received by the aircraft. This was the last ACARS VHF communication recorded.
Other audit information provided by SITA records a total of nine messages prior to the OFF event being logged. Besides the initialization message by the crew, there was a special load notification to the captain informing him of a dangerous goods shipment consisting of 34 kg of dry ice, cargo information, and the final load sheets. The OUT event was also recorded at 23:50:15. Recording of the OUT event occurs when the aircraft is pushed back from the gate.
The ACARS audit logs show what appear to be two anomalies. The first was the lack of communication between the aircraft and ARINC from 0021:15 until 0104:14, and the second was that VHF coverage was lost at 0047:06 for 17 minutes, 7 seconds, until 0104:14 in an area of good ARINC ground station coverage.
Switching the VHF-3 radio from data to voice mode would force ACARS to SATCOM without delay. This would result in a media advisory downlink being sent by SATCOM, indicating a loss of VHF. No media advisory message was downlinked between 0021:15 and 0047:06. This indicates that the VHF radio remained in Data mode during this time, as the media message was not sent until 0047:06. The DFDR only reports the radio key events associated with the pilots' microphone keys, and not those associated with the use of VHF-3 in ACARS mode. During this period (from 0021:15 to 0047:06), 21 VHF-1 keying events and 2 VHF-2 keying events were recorded. These keying events occurred between 0022:33 and 0047:06. As there is no indication of a problem with SATCOM communications, the VHF-3 radio was likely not switched to voice mode during that time.
The ACARS CMU monitors the VHF-3 radio transmitter output during data downlink activities. If there is no feedback detected from the VHF-3 radio, the CMU issues a media advisory via SATCOM. No media advisory was downlinked between 0021:15 and 0047:06, when two tracker messages were expected to be downlinked. This indicates that the VHF-3 radio did not fail or lose power.
The ARINC audit data showed that the internal message counter in the CMU was at M63A when communications coverage was lost. When the coverage resumed with ARINC, the CMU message counter was at M66A, which indicates that two messages were generated but not received by ARINC. If these messages had been generated by the crew they would have remained in a message queue until they could be delivered. Also, when ARINC coverage resumed, the systems message counter was updated from S65A to S67A. If the CMU had a power interruption during this time it would, on initialization, reset the message sequence to zero. The message counter was not reset to zero but was incremented, which indicates that the CMU was functioning during this period to generate these messages.
After 10 minutes of no communication, the CMU sends a Q0 labelled tracker message used by the system for flight following. This is an automatic message and is sent without the knowledge of the crew. After the last Q0 labelled downlink test at 0021:15, and assuming no further crew or ARINC-initiated communication, a Q0 tracker message for flight following should have been broadcast 10 minutes later, at approximately 0031:11. For comparison, this would have been similar to the Q0 tracker message recorded at 0125:05 after 10 minutes of no communication. If the CMU does not get an acknowledgment to the Q0 downlink (e.g., owing to VHF congestion) after approximately 70 seconds (40 to 100 seconds) the CMU will switch to the base frequency and try again for approximately another 70 seconds. At the same time, the retry counter is started and, after four tries with no response, the CMU goes to the next frequency in the Operators Table of Frequencies. At this point a service advisory message label is sent, and the T7 timer is started. After seven minutes of no VHF communications, an M1X message is sent via SATCOM indicating a loss of VHF communications. During this seven-minute period, the CMU will go through the other frequencies in the operator's requirement table, attempting to make VHF contact. This process is repeated until the CMU finds an active frequency or is shut down.
A possible scenario for the ACARS communications gap, based on the fact that the system and message counters were updated during this time frame, is postulated as follows:
After the uplink at 0021:15 the tracker timer is started, and just before the tracker timer expires, a new application downlink message (missing M64A) is transmitted. This message is not acknowledged and, after four tries (which takes approximately 1.5 minutes), the CMU switches to the base frequency with the time now at approximately 0032:30. From the data received, the CMU base frequency on the inbound flight could not be determined. If the inbound base frequency was 131.475 MHz (the Air Canada Dataplus network) or 131.725 MHz (SITA Canada) then ARINC may have managed the communications by auto-tune uplinks similar to the auto-tune at 0021:07. If the CMU goes back to a base frequency of 131.475 MHz or 131.725 MHz (both of which are on the fringes of coverage at this time) and establishes contact with downlink message M64A, this might take an additional 2 minutes. Now the time is approximately 0034:30. At this time the tracker timer is started, and a Q0 tracker message is sent, at approximately 0044:30, but it is not acknowledged and the CMU performs "re-tries" over the next 1.5 minutes, to 0046:00. The service advisory logic runs in 1-minute loops, and the time is now 0047:00. The media advisory (label SA) is sent via SATCOM (which would be the missing S66A system message), and at the same time the T7 timer is started. The T7 timer expires and the label 1X message (missing M65A) is sent at 0053:51 for loss of VHF communications for greater than 7 minutes.
The audit data from Dataplus Network and SITA Canada were not available and the above cannot be proved unequivocally. However, the timing does give the scenario credibility. What can be said is that the CMU did not lose power during this time frame. Also, there is no evidence that the crew would have known about the apparent anomaly. When the crew used the ACARS for weather, it functioned normally, again indicating the system was operational at that time.
Once communications were re-established through SATCOM and then back to VHF communication at 0104:14, there were no other interruptions until 0125:52, when the AES made a log-on initiation request through SATCOM, which was acknowledged at 0126:01 as a Class 2 (voice only) mode. This was the last ACARS communication with HB-IWF.
The INMARSAT logs indicate that at 0125:52 the AES requested to be logged-on and the log-on was acknowledged at 0126:01 as a Class 2 (voice only). The log-on message shows it was an initialization and not a renewal (as occurs if handing off from one satellite to another). The installed software would normally prevent a change in log-on class if either voice or data capability were lost. Therefore, the log-on request was the result of a power interruption. The Class 2 log-on indicates that the CMU was no longer available at this time.
It is not known at what time power to the SATCOM system was interrupted. It could have occurred anywhere between 0053:51 and 0125:52. Typically, the SDU requires one minute from power-on to initiating a log-on if it has been under 24 hours since the last calibration, but it may take two minutes if a more complete calibration is required. There is no way of telling, from the data, the actual time from power-on to the log-on request. This indicates that the SDU lost power some time after the last recorded message at 0053:56, but powered-up between 0123:52 and 0124:52.
For the SATCOM system to be operational in voice-only mode the following components of the system also have to be operational: the SDU, the beam steering unit; the high-power amplifier; radio frequency unit 1; and the IRU-1, IRU-aux, or both. The SDU and radio frequency unit 1 are powered from the 115 V AC Generator Bus 1 (phase A) CB "SATCOM 1A," located in the AU panel at B-28. The beam steering unit and high-power amplifier are powered from CB "SATCOM 1B," located in the AU panel at B-29. IRU-1 is powered from the left emergency 115 V AC bus in the overhead panel at G-22, and IRU-Aux is powered from the 115 V AC Generator Bus 1 (phase A) AU panel at A-22. The IRUs have battery back-up in the event of a power loss.
However, the inertial reference system was active, allowing voice capability at the AES.
According to the SR Technics component inventory of "Time Controlled Components," the latest CMU was installed in HB-IWF on 22 December 1997 during an "L check". The CMU was manufactured by Allied Signal PN 965-0758-001, SN 0206. The reason for the installation at this time was an upgrade from Allied Signal, PN 3614291-4523E, to the Mark 11, PN 965-0758-001, as a prerequisite for the Future Air Navigation System. The change was covered by SR Engineering Order 513082 and Douglas SB MD11-23-059. This change also required a re-routing of wires to change the pin programming from 0110 to 1011, the MD-11 standard. According to the engineering order, all of the wire changes were accomplished in the avionics compartment. The following information pertains to the ACARS software installed:
The CMU is located in the MAR Shelf 4. The CMU is powered from 115 V AC Generator Bus 3 (phase A) through CB B1-33 mounted in the avionics CB panel on row D, location 27. From the CB D-27 terminal, wire B101-21-24 is tie-wrapped onto the other wires that are attached to the CBs on row D. These wires follow the grey-sleeved bus feed B117-4-8 and are routed to the lower aft corner of the avionics CB panel where they join all the other wires from the CB panel and become wire run AMA. The majority of these wires are then routed outboard and up toward one of the nine plugs in the avionics CB disconnect panel mounted at Station 396 on the aft cockpit wall. B101-21-24 connects to pin X of P1-323/R5-323 connector on the avionics CB disconnect panel, then via B203-110-24 (wire routing AAC down right side into avionics compartment) to pin Y of P1-134/R5-134 connector located in MAR Shelf 4. Then connects via B3044-20RD to pin 35JX of terminal strip S3-24 on MAR Shelf 4 and via 455-20RD to receptacle R5-1020C on the CMU.
In 1995, there were 23 complaints logged on the ACARS system. The majority of complaints were of OUT and IN events not being recorded, along with messages and weather reports not being received. There was one reference to a CB reset on 20 August 1995, but this did not resolve the problem of no OUT event times. The CMU was replaced four times in 1995. In 1996 there was only one reference to ACARS, and that was for a briefing card. In 1997, there were three records, one of which was for the ACARS printer missing some digits and an initialization failure. These last three complaints were cleared on the ground, with no fault found. In 1998, there was only one complaint, on 27 January 1998, for "mission status data lost during cruise," and the check action was "checked OK." The review showed no CB trips being recorded. From the time the new part number CMU was installed in December 1997, there was only one complaint noted.
The CMU was not found in the wreckage. Most of the avionics CB panel was re-built, along with those sections of identified bus bar from that panel. The remains of the row D bus bar, from D-22 to D-34, was recovered. The insulation material that covers the bus bar was mechanically damaged and had a "heat-shrivelled" appearance. None of the CBs were recovered from the panel, but a few of the white collar trip indicators were found trapped in the panel. Also recovered was a section of the AU panel containing the holes for CBs B-28 to B-30, C-27 to C-31, D-25 to D-30 and E- 24 to E-31. The white collars remained in D-26 and D-29. The white collar on D-26 was clean and did not exhibit any discolouration from soot, which suggests that it had not tripped prior to impact. The collar of CB D-29 was slightly discoloured, with a soot-like appearance, but this discolouration was not as uniform as examples of white collars, from other locations, that were covered in soot. This latter fact may indicate that the CB D-29 tripped near the end of the flight.
Approximately 3 inches of the wire from CB D-27, B101-21-24 was still attached to connector P1-323. No other wires were identified from this system. The aromatic polyimide wire insulation on the wire was intact and exhibited minor sooting but no discolouration of the modified aromatic polyimide resin topcoat. The connector P1-323 was generally intact and the red grommet material, although uniformly blackened from soot, was still resilient.
Recovered from the wreckage were four 24 AWG aromatic polyimide-insulated wires that exhibited melted copper damage as a result of an arcing event. Only one of these wires was positively identified as to its function: Fire 2 Engine Detection. It was not possible to determine whether the remaining arced wires were part of the ACARS CMU circuitry.
Analysis of the available logs shows that the VHF-3 radio had not been switched to the voice mode nor had a power failure occurred between 0021:15 and 0047:06, as both of these scenarios would have resulted in a media advisory being immediately downlinked via SATCOM. The media advisory sent at 0047:06, indicating a loss of VHF communications at this time, could have been the result of the crew switching VHF-3 from data to voice mode. This latter shows that the SATCOM system was functional at this time. The ACARS CMU was shown to have been functional during this time; its internal sequence generator was updated and not reset to zero, as would have been the case had it had a power interruption.
The ACARS CMU may have been operating temporarily on a VHF frequency other than ARINC (either Dataplus or SITA); it was not until 0047:06 that this coverage was lost, and it took until 0104:14 until coverage resumed with ARINC. The crew used the ACARS VHF communications twice after this, without any problems, to obtain weather reports.
The ACARS information was analyzed to determine whether the apparent anomalies noted could have played any role in the initiating of the on-board fire. As the crew never mentioned any problems with the system, and both the ACARS VHF and SATCOM modes functioned normally after the reported smell and subsequent confirmation of visible smoke, it was determined that the ACARS and SATCOM systems did not contribute to the initiation of the fire.
Table: ACARS Data Log
 AES stands for "airborne earth station" within the ACARS discussion.