External Power

Two ground-power receptacles sit aft of the nosewheel bay — EXT PWR A and EXT PWR B — each able to take a 90 kVA ground power unit (GPU). On paper they look like a symmetric pair. They are not. EXT A is the primary receptacle and EXT B the secondary, and that asymmetry runs through everything in this article: the panel logic (ON versus AUTO), the back-up mode after a controller failure, the ground-service network, and the dispatch rules. Understanding external power is largely understanding why A and B are not interchangeable.

This article does three things. First it establishes the A/B asymmetry and the design reason behind it. Second it follows the GPU's electricity through the gates it must pass before it can touch the aircraft network — the interlock pins, two relays, and a protection table. Third it covers the GAPCU's third role: the same box that runs the APU generator (see APU Generator and GAPCU) also manages both external sources, and even with its microprocessor dead it keeps one path alive for EXT A. Throughout, the boundary with neighbouring articles is kept tight: the ground-service-network detail belongs to Ground Service and Maintenance Bus, the No-Break Power Transfer mechanism to No-Break Power Transfer, and the full priority walk-through to Network Priority and Normal Supply.

1. The twin receptacles — A is primary, B is secondary

The division of labour is stated in the first sentence of the AMM external-power description: a GPU may supply the whole network through EXT PWR A and/or EXT PWR B, but the ground-service network recognises only EXT PWR A. Four independent pieces of evidence make EXT A the unambiguous primary receptacle:

1. Panel logic. Pressing the EXT A pushbutton is an ON command — a direct order: the AVAIL legend goes out, the blue ON comes on, and EPC A closes at once. Pressing the EXT B pushbutton is an AUTO request — EPC B closes only if the APU generator is off the line, and the AUTO legend stays lit even after the APU has taken over (the request is parked, not cancelled).
2. Back-up mode. After a GAPCU microprocessor failure the analogue back-up circuit can control only EXT A (§6).
3. Maintenance bus. The forward-cabin MAINT BUS switch is latched on "EXT A parameters healthy" (developed in Ground Service and Maintenance Bus).
4. Ground-service network. It hangs directly on EXT A's EPC upstream, so a single A-cart can energise the service network without lighting up the whole aircraft:

"The AC ground service distribution network can be supplied: ‐ either normally from the aircraft network, ‐ or by the ground power unit ( external power source through EXT PWR A) upstream of the external power contactor, without energizing the whole aircraft network."

Per AMM 24-42-00. This is the engineering bottom of the line-station habit "plug into A only and you can power the service network without bringing up the entire aircraft".

Why is B "AUTO" and not "ON"? There is now a verbatim FCOM basis, not merely a synthesis. When ground crews can connect only one cart, the procedure tells them to use EXT PWR A, and it gives the reason:

"If only one electrical power is available, the ground personnel should connect EXT PWR A since: ‐ EXT PWR B does not permit to supply directly GND/FLT buses without energizing the total aircraft network. ‐ The simultaneous use of EXT PWR B and APU GEN is not possible."

Per FCOM PRO-NOR-SOP-04. The second bullet is the root of AUTO: EXT B and the APU generator physically cannot coexist — EXT B's "home" is the left half-network, the same side the APU generator holds with first priority. So B is built as a request that queues behind the APU: press it while the APU is on the line and EPC B stays open; switch the APU off and EXT B takes over automatically, with the AUTO legend lit throughout to show the request is pending.

[!warning]- "Connect to A or to B, it makes no difference — press it and you get power"

Incorrect. EXT A pressed is ON (a direct command — EPC A closes immediately). EXT B pressed is AUTO (a request — EPC B closes only with the APU generator off the line). With the APU on the line, pressing B merely queues the cart; the AUTO legend stays lit while the APU keeps the bus. The two receptacles are not interchangeable.

A configuration note (GPCU vs GAPCU). FCOM describes the external-power controller as a two-way option — "A Ground Power Control Unit (GPCU) … or A Ground and Auxiliary Power Control Unit (GAPCU)" (per FCOM DSC-24-10-20-10). On the GPCU build the unit manages only external power and generates the NBPT synchronisation reference; on the GAPCU build the same unit also manages the APU generator and its frequency/voltage. This article describes the GAPCU build — the "one box, three jobs" of §5 is a GAPCU characteristic. When FCOM elsewhere speaks of "the GPCU sending the synchronisation reference", that is the GPCU build's wording for the same function.

2. Architecture — the external-power chain

Read the chain top-down from the cart to the network. The GPU delivers three-phase 400 Hz 115/200 V, up to 90 kVA, through a plug that carries the three phases plus neutral and a pair of small interlock pins (E/F):

GPU  (3-phase 400 Hz 115/200 V, <= 90 kVA)
  │  plug: phases A/B/C + neutral N + E/F interlock pins
  ▼
Receptacle 20XG (EXT A) / 19XG (EXT B)  —  aft of nosewheel bay
  │            E/F pins ─ DC interlock loop ─► GAPCU IMR
  │            criterion: V(E/F) < 42 VDC = plug seated, cart live
  ▼
Feeder adapter 30XG ─► current transformer 21XG (A) / 22XG (B)
  │            3-hole CT   (a generator channel uses a 6-hole CT)
  ▼
EPC A 4XG  (GAPCU + ECMU 2)     EPC B 3XG  (GAPCU + ECMU 1)
  ▼
Transfer corridor  BTC1 ── SIC ── BTC2
            EXT A feeds the RIGHT half;  EXT B feeds the LEFT half

┌──────────── GAPCU 40XG — external-power functions ───────────┐
│ IMR  A/B   interlock monitor  ◄─ E/F < 42 VDC + channel OK    │
│ EPRR A/B   ext-power ready    ◄─ GPU supplying + protections OK│
│ Protections: UV / OV / UF / OF / sequence / interlock-V /     │
│              overcurrent / open-cable / EPC-failure           │
│ Three independent supplies:                                   │
│   (1) GPU via EXT A, direct (internal TR makes its own 28 VDC)│
│   (2) aircraft DC back-up 28 VDC  (BITE only — NOT to the EPC)│
│   (3) APU PMG, direct                                         │
│ Microprocessor dead ─► analogue back-up still controls EXT A  │
└───────────────────────────────────────────────────────────────┘

The control side of EPC A — how a pushbutton press reaches the contactor — is a separate path from the power feed above. Per the wiring schematic (AMM/ASM 24-41-03):

Press EXT A pb 10XG ─► EPRR A closes ─► EXT PWR A flip/flop relay energised
        │
        ├─(a) ground signal direct ─────────────► ECMU 2  (runs EPC A logic eqn)
        └─(b) ground signal via relay 25XG/26XG ─► ECMU 1
        │
        ▼
EPC A 4XG closes ─► feeds the BTC/SIC corridor
        ON blue lit · AVAIL out · NOT-IN-USE out
Receptacle lights 7XG (NOT IN USE, white) + 8XG (AVAIL, amber)
        driven through diode module 1820VD; 9LP = receptacle lamp-test button

Three load-bearing facts the schematic pins down: (i) EPC A is under dual control — an EPC-A CONTROL module plus ECMU 2 (the contactor terminal connects to both); (ii) the flip/flop relay is driven by EPRR closing, then its state is relayed to the ECMUs via 25XG/26XG — this is the physical "how the pushbutton state reaches the ECMU"; (iii) the receptacle lights are driven through an isolating diode module (1820VD) with a lamp-test button (9LP) on the receptacle housing. The B side (AMM/ASM 24-41-06) mirrors this with ECMU 1.

[!warning]- The ECMU cross-assignment: EPC A ↔ ECMU 2, EPC B ↔ ECMU 1

The contactor-to-ECMU mapping is crossed, not aligned: External Power Contactor A is managed by ECMU 2, and External Power Contactor B by ECMU 1. This matches the half-network wiring — EXT A on the right, EXT B on the left — and the same crossover that puts the APU contactor and EXT B on side 1 and EXT A on side 2 (see Electrical System — Overview). Expecting A↔1 / B↔2 is the common error.

Key numbers and parts:

3. The E/F interlock — the plug's safety pin

Besides the three phases and neutral, the plug carries a pair of small pins, E and F, running a DC interlock loop. The criterion is purely the voltage on those pins:

"the Interlock Monitoring Relay (IMR) A (B), This relay, when energized, connects a holding supply to the ground power unit line contactor. It is energized when the DC input voltage (pin E and F receptacle) is lower than 42 VDC and either the external power A (B) channel fault signal is absent or the Analog/Digital (A/D) fault signal is present."

Per AMM 24-41-00 §6.A. In plain terms, the E/F pins are the physical handshake that says "the plug is fully seated and the cart is supplying normally". An abnormal pin voltage — DC above 42 V, or AC bleeding in above 13 V pk-pk — means the plug is loose, the cable is damaged, or the cart has a fault: the IMR drops out, the EPC's holding supply is removed, and external power leaves the network at once.

[!warning]- "To remove external power you must first go to the cockpit and press the EXT pushbutton before unplugging"

Not so on the protection side. The interlock makes unplugging self-safe: if the E/F voltage goes abnormal, the IMR drops out, the holding supply is removed, and external power disconnects automatically. (This is the hard protection layer — the procedural layer still asks the crew to request disconnection only with the pushbutton at AVAIL; see §8 and §10. The two layers are "follow the procedure, with hardware as the backstop", not "pull it whenever you like".)

The note on "28 VDC": the manual gives only the "< 42 VDC" criterion for the interlock pins. The figure 28 VDC belongs elsewhere — it is the GAPCU's internally generated supply (its own internal TR) and the aircraft-DC BITE back-up voltage (§2). It is not the interlock-pin voltage.

4. Two relays, nine protections — the GPU's security check

The GPU's electricity must pass two relays before it can reach the EPC (per AMM 24-41-00 §6.A):

• IMR — the physical interlock gate (§3): plug seated, cart live, E/F voltage normal, channel fault-free.
• EPRR (External Power Ready Relay) — the electrical-quality gate: the GPU is connected and running, and the undervoltage / overvoltage / underfrequency / overfrequency / phase-sequence / channel-fault checks are all clean.

Both gates passed, three things follow together: the EPC may close, the green AVAIL legend comes on, and the receptacle-area lights light. The protection table shares its lineage with the generator channels but carries external-power specifics (per AMM 24-41-00 §6.A):

On a trip the action is uniform: IMR and EPRR both de-energise and the GPU leaves the network; while a protection is active the ON legend is inhibited from lighting (per the §6.A NOTE). So "I pressed the EXT pushbutton but no ON/AUTO legend came on" most often means a protection caught it — not a broken button.

Two finer points on the table. First, overload (> 277 A) does not, by itself, trip the contactor — it sheds galleys. The TRIPS column of the overload row is blank: it issues a ground signal to the ECMU to shed galleys automatically (the same script as a generator overload, see Galley and Commercial Loads). What actually disconnects the GPU is the subsequent overcurrent EPOC — if current is still excessive after shedding, the inverse-time curve trips EPRR and IMR. So overload and overcurrent are two tiers: shed first, disconnect only if shedding is not enough. Second, the underfrequency 2nd-stage reset carries an IMR RESET + CYCLE EPS condition (a 2nd-stage UF reset requires cycling external power).

Who actually closes the EPC. The GAPCU's "ready / quality OK" is not the final word — the ECMU runs the closing logic equation:

"‐ the EXT PWR A (B) flip/flop relay is energized, if the EPRR A (B) is closed. (1) The EXT PWR A (B) flip/flop relay status is transmitted to ECMU(s) by the following interface: ‐ directly a ground signal to the ECMU 2 (1) which manages the logic equation of the EPC A (B), ‐ through the relay 25XG (26XG) a ground signal to the ECMU 1(2)."

Per AMM 24-41-00 §7.A. The EPC's final closure is decided by the ECMU running its logic equation — inputs being the flip/flop relay state, the AC main-contactor states, and the GCU/GAPCU data — the same "GCU + ECMU dual sign-off" structure as the GLC (see GCU and AC Generation Control). And the blue legend is explicit about what it does not mean:

"NOTE: The blue ON (AUTO) legend indicates that the pilot action has been taken into account but not that the GPU connected to EXT PWR A (B) is already supplying the aircraft electrical network (because of the NBPT time delay or the APU gen availability)."

Per AMM 24-41-00 §7.A.

[!warning]- "I pressed EXT A and the blue ON lit, so the GPU is now supplying the network"

Not necessarily. The blue legend means only "your action has been taken into account" — it does not confirm the GPU is feeding the bus, because of the NBPT time delay or because the APU is still on the line. To confirm what is actually supplying the network, read the arrows on the ELEC AC SD page, not the pushbutton legend.

A useful teaching image for the whole chain: bringing external power onto the network is like a visitor entering a secure building. The E/F pins are the access card (a physical interlock — if the card is not properly read, the door does not open); the EPRR is the security scanner (voltage / frequency / phase sequence — any alarm and the visitor is turned away); EXT A is the resident contractor with a private back door (the analogue back-up when the GAPCU is dead) and access to the back-of-house (the ground-service network and MAINT BUS); EXT B is the temporary visitor who can only queue at the front desk (AUTO) and does not get in while the resident (APU GEN) is inside.

5. The GAPCU's NBPT veto

The GAPCU is the frequency-reference source for No-Break Power Transfer, feeding a synchronisation reference to the GCUs. It carries one uncompromising rule:

"If the GAPCU finds that its acquisitions can cause a NBPT with non-synchronous alternators, the computer inhibits frequency references sent to GCUs, which prevents any NBPT. In any case, the NBPT is inhibited definitively until the GAPCU power supply is completely cut off."

Per AMM 24-41-00 §6.A. The meaning for operations: NBPT would rather not happen than happen badly (a non-synchronous parallel connection is an instantaneous circulating-current shock). And the inhibit is latched — once a suspect condition has occurred, every source transfer on that aircraft will break-transfer until power is completely removed (every AC and DC source, batteries included) and the controller cold-starts.

[!warning]- "The NBPT flickered once — just restart the source transfer and it's fine"

Conditional. The inhibit is latched: "the NBPT is inhibited definitively until the GAPCU power supply is completely cut off". After one suspect event the aircraft will break-transfer on every transfer for the rest of the day, until it is fully de-powered (batteries disconnected too) and cold-started. On the ground, "why does every transfer flicker now?" can trace back to one event recorded earlier that morning.

Where does the reference come from? Inside the GAPCU is the FRU (Frequency Reference Unit), generating two independent references:

"The FRU is located in the GAPCU. The GAPCU generates two separate frequency references. The left reference (FRU L) for the left half of the aircraft (channels 1 and APU). The right reference (FRU R) for the right half of the aircraft (channel 2). … In case of FRU failure or FRU reference signal being out of limits, each GCU regulates the frequency to its own internal 400 Hz reference and inhibits the NBPT for its channel."

Per AMM 24-29-00 §6.M. For an external source to take part in an NBPT there is a hard synchronisation window — "the EXT PWR frequency is 400 HZ plus or minus 10 HZ, the three phase average is 115 VAC plus or minus 5 VAC" plus the PR equation true (per AMM 24-29-00 §6.M): all three conditions together. Maintenance can read the external-power BITE codes — two separate families: the GCU family FC 889–892 (IDG synchronisation) and the GAPCU family FC 889–893 (APU synchronisation plus SIC-state 893). The NBPT mechanism itself is developed in No-Break Power Transfer.

6. Microprocessor failure — the back door reserved for EXT A

A GAPCU microprocessor failure does not kill external power outright:

"In case of microprocessor failure, the GAPCU is still able to control a ground cart connected to the External Power A receptacle."

Per AMM 24-41-00 §6.A. The back-up is a purely analogue circuit running at reduced protection: phase sequence (non-ABC inhibits), OV 127.5 ±2.5 V (135 ms), UV 102 ±2.5 V (4 s), OF 430 ±5 Hz (3.5 s), UF 365 ±5 Hz (135 ms), OC 288.5 ±11.5 A (9 s); galley-shedding is still commanded; all the indicator lights work, but there is no external-power display on ECAM and the BITE does not function; reset is via the EXT A control switch itself. The design intent is blunt: a GPU stuck at an outstation with a dead GAPCU can still feed the aircraft through the A receptacle overnight or for troubleshooting — one more reason EXT A is the primary receptacle. EXT B has no analogue back-up.

[!warning]- "If the GAPCU (microprocessor) fails, external power is completely dead"

Half true. After a microprocessor failure EXT B is indeed lost, but EXT A survives — "the GAPCU is still able to control a ground cart connected to the External Power A receptacle". The analogue back-up takes over with reduced protection and normal lights — the only missing pieces are the ECAM display and the BITE.

7. The dual-source power split and the receptacle lights

On the ground the APU generator and EXT A can run together — split, not paralleled:

"They can supply technical and commercial loads up to a maximum power of 115 KVA for the APU generator and 90 KVA for the EXT PWR A (the technical loads is approximately 50 KVA)."

Per AMM 24-41-00 §3.B NOTE. The "technical loads ≈ 50 kVA" is a handy mental baseline: with the galleys and other commercial loads off, a single 90 kVA GPU is comfortable; on a hot day with two packs cooling, full galleys, and cargo work, one source gets tight — which is exactly why the twin receptacles and the "APU + EXT A together" capability exist (more so on freighter builds).

The receptacle-area lights serve the ground crew, not the flight deck (the pilot reads the pushbutton legends):

The AVAIL light is amber, confirmed by FCOM: "This amber light comes on to indicate that external power is available and the voltage is correct." (per FCOM DSC-24-20). One AMM §7.A sentence reads "red EXT PWR A (B) AVAIL", but the same paragraph's NOTE and the FCOM panel description both say amber; the "red" is taken as a typographical slip. NOT IN USE lit means "this cart is connected for nothing — it can be removed"; NOT IN USE out means "it is supplying — do not unplug".

8. Controls, ECAM and the priority picture

Panel skeleton (per FCOM DSC-24-20). EXT A pb: AVAIL green (parameters healthy) → momentary press → EPC A closes, AVAIL out, ON blue on; press again → EPC A opens, back to AVAIL. EXT B pb: AVAIL green → momentary press → EPC B closes only if the APU generator is off the line, AUTO on; press again → opens, back to AVAIL. The AUTO legend stays lit after the APU takes over (the request is pending).

The same FCOM page carries the network priority, and the narrative version states the whole logic:

"Each AC BUS is supplied in priority order by : ‐ the corresponding engine generator. ‐ the APU generator or the external power A (if both are connected, the APU generator has priority for the left side bars, and the external power has priority for the right side bars). ‐ the external power B (if both external powers are connected, B has priority for the left side bars and A has priority for the right side bars). ‐ the other side engine generator. The APU generator or an external power may supply all the network. … The generators cannot be connected in parallel (except on ground during No Break Power Transfers)."

Per FCOM DSC-24-10-30-10. This is the direct source for "EXT B → left half, EXT A → right half". The AMM gives the hardware landing of which receptacle carries which bus:

"make sure on the SD that: ‐ EXT A supplies busbar AC2 ‐ EXT B supplies busbar AC1."

Per AMM TASK 24-41-00. So EXT A feeds the right half (AC BUS 2) and EXT B feeds the left half (AC BUS 1) — the physical landing of "EXT A has priority on the right, EXT B on the left". The full priority walk-through, including the five-source ordering, is in Network Priority and Normal Supply.

On the ECAM ELEC AC SD page in the GROUND configuration, both source symbols appear with APU + EXT A on the line, and green arrows show which source feeds which half (APU left, EXT A right). The MEMO ELEC EXT PWR is shown in green while external power is available and turns amber once more than one engine is turning (per FCOM DSC-24-20 MEMO) — the prompt that it is time to disconnect external power.

9. Pilot scenarios

1. One cart at an outstation. Connect A. Four reasons: A is the ON direct command; the MAINT BUS recognises A; the GAPCU back-up mode keeps only A; and the ground-service network hangs directly on A's EPC upstream (one A-cart powers the service network without lighting up the whole aircraft, §1). A cart on B is merely "queued" while the APU is on the line — and EXT B and APU GEN cannot be used at the same time (§1).
2. Press EXT A, nothing happens (no ON). First, did AVAIL light beforehand? If it did, a protection caught the press (cart voltage / frequency / phase sequence) — have the ground crew check the cart. If AVAIL never lit, the interlock gate did not pass (plug not seated / E/F loop abnormal) or the cart is not delivering.
3. Reading the dual source on ECAM. With APU + EXT A both on the line, the GROUND-configuration ELEC AC page shows both source symbols, with green arrows for who feeds which half (APU left, EXT A right). The ELEC EXT PWR MEMO is green while external power is available — and turns amber after one or more engines start (a reminder to disconnect).
4. External power still connected at engine start. No alarm needed — the engine generators are top priority and a GEN on the line automatically displaces external power (seamless on-ground NBPT). But the amber MEMO and the ground handover still apply: a plug left physically attached is an FOD/burn hazard.
5. Outstation GAPCU failure (no external power on ECAM). Recall the §6 back door — EXT A's analogue back-up can still supply (lights normal, only the ECAM display absent), with the degraded analogue protection set. For dispatch, consult the operator MEL (§10).

10. Dispatch and reset

Dispatch (per the operator MEL, chapter 24-41). External-power faults split into three classes, not one:

The routing fact worth holding: there is no "EXT PWR FAULT" caution entry — an external-power fault does not raise a dispatch-level ECAM caution, because external power is a non-essential ground source. The only external-power caution that routes into chapter 24 is ELEC EXT PWR A(B) OVERLOAD, which routes to the galley-supply item 24-26-01 (the same routing as a GEN / APU GEN overload — all three are handled by "confirm automatic galley shedding works", echoing the §4 overload-sheds-galleys behaviour; see Galley and Commercial Loads).

Engine start on external power — start both engines before disconnecting:

"In order to ensure that the entire network remains electrically supplied, it is recommended to start both engines before the disconnection of the external electrical power. … When the external power supply is disconnected: ‐ If only one engine is running: A power transient occurs when the external electrical power is disconnected, as the No Break Power Transfer (NBPT) function is not available. ‐ If both engines are running: No power transient occurs, as the NBPT is available."

Per FCOM PRO-NOR-SUP-ENG. This closes the loop with the §5 NBPT mechanism: with one engine running NBPT is unavailable and disconnecting external power produces a transient; with both running NBPT is available and the transfer is seamless. Note the two layers of removing external power — the procedural layer (the crew requests disconnection only with the EXT pushbutton at AVAIL, because "Disconnection of the external power with the EXT PWR pb-sw ON may cause injury to the ground engineer", per FCOM PRO-NOR-SUP-ENG), and the hard-protection layer (§3 — if a live plug is forced out, the E/F interlock disconnects the system automatically).

Reset (per QRH). The authoritative reset procedure:

"GPCU/GAPCU In flight: Not applicable. On ground: GPCU/GAPCU reset may be attempted, if the AVAIL and ON lights do not light up. 1. Press and release the affected EXT A(B) pb to reset the functions of GPCU/GAPCU 2. Check the GPU is connected to the aircraft, then started 3. Check AVAIL light is ON and press the EXT A(B) pb to connect the GPU. If unsuccessful, check GPU voltage and frequency (bite is available on batteries only)."

Per QRH ABN-02 System Reset Table. Three points: the external-power reset is a ground-only action (not applicable in flight); the trigger "AVAIL and ON do not light" is exactly scenario 2 above; and reading the external-power BITE codes requires the aircraft to be on batteries only — the precise limit behind the §6 "BITE does not function". The QRH has no standalone EXT PWR abnormal procedure; external power appears in the QRH only on this reset table.

Self-test

[!note]- Q1. What is the panel-logic difference between EXT A and EXT B, and the design reason?

EXT A pressed = ON (EPC A closes directly). EXT B pressed = AUTO (EPC B closes only when the APU generator is off the line; once the APU takes over, the AUTO legend stays lit to show the request is parked). The reason has a verbatim FCOM basis: "The simultaneous use of EXT PWR B and APU GEN is not possible." (FCOM PRO-NOR-SOP-04). EXT B's home is the left half-network, which overlaps the APU generator's first-priority side, so B is built as a queued request — avoiding a pointless source swap and letting B take over automatically once the APU is switched off.

[!note]- Q2. What are the E/F pins, and what does 42 VDC govern?

The plug's interlock pins, running a DC interlock loop into the GAPCU's IMR. The manual gives only the "< 42 VDC" criterion (28 VDC is not the pin voltage — it is the GAPCU's own internal supply / DC-BITE back-up). With E/F below 42 VDC and the channel fault-free, the IMR energises and the EPC gets its holding supply; with DC above 42 V or AC above 13 V pk-pk bleeding in, the loop reads a loose plug / damaged cable and the IMR drops out, disconnecting external power at once. It is the "plug must be seated before it powers" safety pin.

[!note]- Q3. Walk the GPU's "security check" into the network.

Interlock gate (IMR: E/F pins normal) + quality gate (EPRR: GPU running and the nine protections all clean — OV 130 / UV 101.5 / OF 435 / UF 361·323 / sequence ABCN / interlock voltage / overcurrent / open cable / EPC failure) → the EPC may close and AVAIL lights. On any protection trip both IMR and EPRR de-energise and the ON legend is inhibited from lighting. Overload (> 277 A) runs the "shed galleys" script; overcurrent EPOC tells the ECMU to open the transfer contactor and isolate external power. And the final EPC closure is decided by the ECMU's logic equation, not the GAPCU alone.

[!note]- Q4. After a GAPCU microprocessor failure, what is the state of external power?

EXT A is still usable — a purely analogue back-up takes over with reduced protection (PSEQ / OV 127.5 / UV 102 / OF 430 / UF 365 / OC 288.5 A), galley shedding still available, all lights normal, but no ECAM external-power display and the BITE inoperative; reset is via the EXT A control switch. EXT B has no analogue back-up.

[!note]- Q5. On the ground with APU GEN and EXT A running together, how is the power split?

Split, not paralleled — the APU carries the left half (<= 115 kVA) and EXT A the right half (<= 90 kVA), together covering technical + commercial loads. Technical load is approximately 50 kVA, so a single source is enough for a "bare" aircraft; the dual source exists for full commercial load (galleys, cargo work, ground air conditioning), most of all on freighter builds.

[!note]- Q6. The blue ON came on after pressing EXT A — is the GPU now supplying the network? And does the GAPCU close the EPC by itself?

No to both. The blue legend means only "the pilot action has been taken into account", not that the GPU is feeding the bus (because of the NBPT time delay or the APU still on the line, per the AMM §7.A NOTE) — confirm supply by the ELEC AC SD page arrows. And the EPC's final closure is not the GAPCU's alone: EPRR closes → the flip/flop relay energises → its state reaches the ECMU directly and via relay 25XG/26XG → the ECMU runs the EPC logic equation (checking the AC main-contactor states and priority) before EPC closes — the same "dual sign-off" as the GLC.

Key takeaways

References

Per AMM 24-40-00 / 24-41-00 D/O (twin-receptacle layout, ground-service network recognising only EXT A, GPU 90 kVA, EPC dual control and the ECMU cross-assignment, IMR/EPRR two-relay chain and the external-power protection table, E/F interlock < 42 VDC, three independent GAPCU supplies, NBPT definitive inhibit, microprocessor-failure analogue back-up, dual-source power split with technical load ≈ 50 kVA, receptacle-light logic, flip/flop relay and the blue-legend NOTE) and AMM 24-42-00 (ground-service network supplied through EXT A upstream of the EPC) and AMM 24-29-00 (FRU L/FRU R in the GAPCU, the external-power synchronisation window); FCOM DSC-24-10-20-10 (GPCU vs GAPCU builds, twin receptacles), DSC-24-10-30-10 (priority order and no-parallel), DSC-24-20 (EXT A/B pushbutton logic, the amber receptacle lights, the priority note, the ELEC EXT PWR MEMO), PRO-NOR-SOP-04 ("connect EXT PWR A" rationale), PRO-NOR-SUP-ENG (engine start on external power, "start both engines before disconnecting", the disconnection warning); the operator MEL chapter 24-41 / 24-01-01 / 24-26 (the three dispatch classes, no EXT PWR FAULT entry, the OVERLOAD routing); QRH ABN-02 System Reset Table (the GPCU/GAPCU reset, BITE on batteries only). The receptacle-area-light "red vs amber" reconciliation and the visitor-building analogy are the author's integrative synthesis of the above; the AMM "red AVAIL" wording is read as a typographical slip in favour of the FCOM/AMM-NOTE "amber".

Independent study material, not an Airbus publication. Refer to current operator FCOM, FCTM, and QRH for operational use.