
IN SERVICE FEEDBACK
| Event date : 17 January 2023 | Time : 04:25 UTC |
| Place : East Midland Airport | Aircraft : ATR 72-202 |
1. Serious incident description
On 17 January 2023, the airline operated an ATR 72-202 on a scheduled commercial flight from Belfast airport (EGAA) to East Midland airport (EGNX).
As they approached East Midland airport, they were told to expect to hold due to delays caused by the Low Visibility Procedures. The aircraft entered the hold as directed and after approximately 10 minutes ATC gave radar vectors to intercept the ILS CAT II for Runway 27.
The weather report for East Midland airport at 03:50 UTC was: wind 270°/8 kt, visibility 400 m freezing fog, cloud broken at 100 ft and the temperature and dewpoint both -2°C.
At approximately 500 ft radio altitude the aircraft entered cloud. Shortly after that the ELEC caution on the Centralized Crew Alerting System (CCAS) illuminated. This indication directed the crew to check the electrical indications on the overhead panel where they noted that the DC GEN 2 fault amber light was flashing.
The First Officer’s EADI and EHSI were also flashing in time with the ELEC and DC GEN2 captions. The autopilot (AP) disconnect horn sounded and the flight director (FD) modes on the EADI disappeared, although the FD guidance bars remained visible and appeared to give sensible indications. The crew decided to go around. As power was applied for the go-around there were numerous audio and visual warnings including EGPWS warnings for “terrain ahead” and “too low gear”.
During the climb, both pilots recalled hearing sounds that they believed were electrical relays cycling. The ADU was also flashing, and the crew were unable to reselect the AP.
Once level, the crew made a PAN call to ATC but heard no response. They then tried to contact ATC on the distress frequency 121.5 MHz but still received no response. They reselected the East Midland Tower frequency, selected 7600 on the transponder, then broadcast a MAYDAY. The pilots heard a faint message from ATC to contact East Midland Radar. They changed to the appropriate frequency and from that point communications were restored. The crew requested radar vectors for a diversion to Birmingham International Airport (EGBB).
The weather report for Birmingham airport at 03:50 UTC was: wind 230°/1 kt, CAVOK and the temperature -5°C and dewpoint -6°C.
As the DC GEN 2 fault amber light was flashing, the crew carried out QRH actions for a DC GEN 2 fault, which involves selecting that generator to OFF. When the DC GEN 2 was switched off the audio and visual warnings stopped, the First Officer’s screens stopped flashing and went blank, and several caution lights remained illuminated in the overhead panel. The commander recalled that the “Bus Tie Contactor (BTC) did not close” as would have been expected so the DC Bus 2 was unpowered.
The left navigation receiver could be tuned to the ILS for Runway 33 at Birmingham International Airport, but all automatic flight functions remained unavailable. The commander flew a manual raw data ILS approach to Runway 33 and the aircraft landed without further incident and taxied to stand.

RADAR GROUND TRACK
2. Maintenance History
Routine scheduled maintenance (same as all ATR version) had been completed on the aircraft two days before the incident. During the check, the No 2 engine DC starter-generator (DC GEN 2) brushes were found worn to their service limit, so it was replaced with an overhauled unit. The replacement starter-generator had been overhauled and bench tested in accordance with the manufacturer’s Component Maintenance Manual (CMM) six months previously and was supplied with a valid EASA Form 1 Authorized Release Certificate.
3. Electrical system
If one of the DC generators fails, its dedicated GCU isolates the affected DC network and the BPCU closes the BTC so that the other DC generator can provide power to both DC main buses. The BTC can also be isolated (ISOL) via a pushbutton switch on the flight deck overhead panel, to prevent the main DC 1 and DC 2 buses being linked, but the BTR is also isolated, preventing AC 1 and AC 2 from being linked. The 1PA contactor further allows DC Bus 1 or DC Bus 2 to power the ESS, STBY and EMER Buses, if either or both of the 3PA and 58PA contactors are unable to be controlled by the MFCs.

DC/AC NETWORK SCHEMATIC
4. DC starter-generator
The starter-generator operates in two modes. Firstly, as a starter during the engine start sequence (up to 45% NH). At this point, the unit stops providing starting torque. When NH then reaches 61.5%, the unit begins to operate as a DC generator, providing power to its respective DC Bus network.
NH is measured by a magnetic speed sensor within the starter-generator. The speed sensor assembly consists of a conical metal speed disc with equally spaced holes around its face, that is attached to the unit’s rotating shaft, and a magnetic sensing unit positioned above the disc. As the disc rotates, the holes pass the sensor. The change in air gap detected as the holes pass creates an output voltage from the magnetic sensor which is translated into the NH measurement.

SPEED SENSOR ASSEMBLY
GCUs monitor the supply from the DC generators. If a fault is detected, the GCU will open the generator contactor between the generator and its main DC Bus to isolate it. A DC GEN fault pushbutton caption will subsequently illuminate on the overhead panel and an ELEC caution will appear on the CCAS.
The GCUs also monitor the generator speed sensor output signal for fault detection at the 45% and 61.5% NH points. If the speed sensor signal is not detected correctly, the GCU will trigger the sequence of events to isolate the respective DC Bus and send a signal to the BPCU to close the BTC. The BTR will then close 10 seconds later.
If the BTC is cycled three or more times within two seconds, an inbuilt system protection locks it open so it will not tie.
5. Tests and research from the report
The DC Gen 2 starter-generator removed from G-NPTF was examined by the unit’s manufacturer.

DC GEN 2 STARTER-GENERATOR
The speed sensor output connector was removed from its housing for examination of wiring condition. The speed sensor output connector is connected to the sensor unit by two wires, each containing seven strands of 0.15 mm diameter. Removing the heat-shrink sleeve from the white wire showed the wire to be completely broken but, whilst held in place by the heat-shrink sleeve, close enough to make intermittent contact. Removing the heat-shrink sleeve on the black wire showed damage to the wire core.

SPEED SENSOR CONNECTOR & WIRES HEAT-SHRINK SLEEVE REMOVED FROM WIRES
The manufacturer’s overhaul and maintenance procedure require removal of the speed sensor connector from the wires to facilitate cleaning of the main generator unit. Excess wire length is provided at manufacture, to allow this to be performed multiple times during the unit’s service life. To reconnect the wires, a small portion of the wire’s insulating outer sleeve is removed to expose the wire core before soldering to the connector pins.
The starter-generator was overhauled, in accordance with the manufacturer’s CMM, at a third-party provider (Ref: Chapter 2. Maintenance history). The type of tool required to strip wire insulation is not specified within the CMM and the provider used a standard mechanical stripping tool.

MECHANICAL WIRE STRIPPING TOOL USED AT OVERHAUL
In contrast to the third-party provider, the manufacturer of the starter-generator confirmed that it uses thermal insulation stripping tools in its facility as it had previously observed the potential for wire damage by using mechanical tools.

THERMAL WIRE INSULATION STRIPPING TOOL
6. Contributing Factors from the report
A wiring defect on the DC GEN 2 speed sensor resulted in rapidly changing erroneous signals being sent to the GCU. This resulted in the rapid opening and closing of contactor 23PA in response to these inputs and, due to the rapidly fluctuating conditions, the BTC entered a self-protection mode and remained open for the remainder of the flight.
In consequence of the above, the crew lost a significant number of instruments and systems during the flight.
The wiring defect is associated with incorrect use of wire stripping tools at the third-party organization that had overhauled the starter-generator. The starter-generator manufacturer and the overhaul organization have identified several safety actions they intend to take to prevent a reoccurrence.
The starter-generator manufacturer took the following safety actions:
- To modify the CMM procedure for this unit to highly recommend the use of thermal wire strippers.
- Extend the use of thermal wire strippers to all other applications.
Next revision: CMM 24-32-65 rev 2 (P/N: 8260-124) and CMM 24-32-61 rev 13 (P/N: 8260-121/123)
The overhaul facility has stated it will take the following safety actions:
- An analysis of standard practice manuals to check the method of wire stripping specified, followed by an update of the relevant Technical Instruction to bring it in line with the standard practices.
- Clarified that tool choice is performed in the following order for each task: CMM, Standard Practice, Technical Instruction.
- The Method Department technicians have been informed of the issue and, where a method is not specified, they will assist the technician in assessing the best way to strip the wire.
- Wire stripping has been declared as an industrial process and training is to be performed.
- Technicians have been informed that the preferred method of wire stripping is to use thermal wire strippers.
7. ATR recommendations
ATR recommend the retrofit of the starter-generator applicable to P/N “8260-121 pre-amendment R” and “8260-123 pre-amendment G” via the Thales Service Bulletin 8260-12x-24-017. The reason is to improve the speed sensor reliability.
The final report in English is available for download at Final investigation report
