FMGEC Resets, Resynchronisation, and Temporary Anomalies

The FM is software, and software crashes. The FMGES's philosophy is automatic restart plus automatic reconciliation — the crew enters only when the ladder runs out. Article 19 covered how to live after the collapse; this article covers how the system saves itself first, the escalation ladder, and the officially catalogued pits.

 software hiccup → RESET (2–3 s) → RESYNC (≈25 s) → recovered      [self-healed]
    ↓ repeated hiccups
 multiple dual resets (4 in 3 min; other standard 3 in 2 min) → memory wiped [data lost]
 multiple resets (5 in 4 min; other standard 4) → SINGLE LATCH                [locked single]
 multiple resync failures (several in 5 min; other standard 3 consecutive) → INDEPENDENT
 both sides latched → DUAL LATCH → raw data + NAV B/UP                        [article 19]

Per FCOM DSC-22_20-40:

When the FM software cannot work properly, or receives instructions to perform impossible operations, it automatically resets itself. A resynchronization with the other FM always follows. When the reset is a minor one, the system will recover by itself. One single reset lasts a maximum of 2 s to 3 s, followed by a resynchronization.

A resync takes about 25 s — and is not always reset-triggered: the two FMs also launch one on their own when cross-comparison finds a disagreement (article 19's crosstalk, as an immune reaction). And the caution that defines the golden discipline:

CAUTION If four dual FM resets occur in 3 min, pilot-entered data is lost. If a dual reset is identified, it is recommended that the flight crew does not perform again the last MCDU actions for 1 min (in order to avoid a potential second dual reset, leading to the loss of pilot-entered data).

Four dual resets in 3 minutes (the other FG standard: three in 2 minutes) wipe everything you typed — plan, weights, CI, cruise level, preselections, tuned navaids. The one-minute discipline: after recognising a dual reset, do not repeat your last MCDU action for a minute — your entry was probably the trigger, and pressing it again is wiping the library by hand.

1. Reset in progress — five symptoms and one prohibition

Per FCOM DSC-22_20-40:

While a RESET/RESYNCH occurs: ‐ The ND shows MAP NOT AVAIL. ‐ The MCDU reverts to the A/C STATUS page, with PLEASE WAIT displayed in the scratchpad with the FM FAULT light illuminated. ‐ Autotuning of NAVAIDs (VOR, DME, ADF ) are lost on the failed side. ‐ AP and managed modes may be transiently lost (reversion to HDG / V/S or TRK/FPA). ‐ If the flight crew presses a key while the scratchpad is showing PLEASE WAIT, there is no change at MCDU level. This behavior is normal. The flight crew should not respond by pulling the MCDU C/B.

MAP NOT AVAIL; the A/C STATUS page with PLEASE WAIT and the FM light; autotuning gone on the failed side; a transient fall to basic modes — and a dead keyboard that is normal. The prohibition in the last line matters: do not pull the MCDU circuit breaker — it only scrambles the resynchronisation (article 03's PLEASE WAIT discipline, sourced). One cold postscript for the recovery: WAIT 1 min after the 'PLEASE-WAIT' message has disappeared before engaging the AP/FD of the failed FMGC — a freshly rebooted computer is not immediately handed the aircraft.

2. SINGLE LATCH — locked out, two-stage response

Per FCOM DSC-22_20-40:

If five successive resets occur in 4 min, the failing FMGEC will latch, and single mode operation will start.

(The other standard: four successive resets.) Symptoms: the failed side's MAP NOT AVAIL with SELECT OFFSIDE RNG/MODE, its MCDU back at the MENU page, its AP/FD gone (ECAM AP OFF plus FM 1(2) FAULT). The response has two stages: ① engage the healthy side's AP and managed modes; ② manually reset the latched computer with the overhead FMGEC reset button — success means an automatic resync back into dual; failure means pulling its reset breaker and setting the FM source to BOTH ON the healthy side (formally entering article 19's SINGLE procedure). The manual reset's three trump cards:

If this occurs in flight, reset one FMGEC at a time. ... The reset of the FMGEC disconnects the onside autopilot. ... The manual FMGEC reset via the reset button does not increment the reset counter: there is no limitation to the number of resets commanded manually via the reset buttons.

One at a time in flight; the reset drops the onside AP as a side effect; and manual resets don't count toward the latch — only automatic ones do (an electrical transient can defeat a manual reset; on the ground, engines-off retry is available).

3. Frozen targets — the most insidious failure shape

Per FCOM DSC-22_20-40:

In some cases, one FMS may fail and latch (no auto-recovery). The FMS may continue to send valid guidance targets to the FG that are frozen. ... In that case, the FG does not detect the FMS failure. Both AP and FD may remain available. If the AP is engaged on the affected side, the failed FMGEC continues to guide the aircraft in managed mode (speed, vertical and lateral) with the frozen targets.

The FM dies, but the targets it emitted with its last breath still look valid — the FG suspects nothing, the FMA doesn't move, the AP and A/THR stay engaged, and the aircraft is quietly steered by a dead computer's frozen intentions. Recognition is circumstantial: the failed side's ND showing MAP NOT AVAIL and GPS PRIMARY LOST, its MCDU frozen, the opposite MCDU announcing INDEPENDENT OPERATION (possibly with FM 1(2) FAULT and CAB PR LDG ELEV FAULT on ECAM). The response, with its famous trap:

‐ If the failure affects FMGEC1, and if the FMGEC1 is still failed and latched before the approach: DO NOT USE AP1 for the approach ‐ Before disconnecting AP2: REVERT to selected speed DISCONNECT A/THR Note: It prevents the A/THR from switching to the frozen speed target of the failed FMGEC1 when the flight crew disconnects the AP2 for landing.

Ignore the failed side, don't fly its AP, attempt a short reset when workload permits. And the trap: FMGEC 1 latched, approach flown on AP2 — before disconnecting AP2 for landing, first go selected speed and disconnect the A/THR. Otherwise, the instant AP2 drops, the A/THR mastership defaults back to FMGEC 1 (article 16: no AP → FMGEC 1) — and the thrust starts chasing a speed target frozen half an hour ago, on short final. This is the chapter's only "the FMA is perfectly normal and perfectly wrong" case — the boundary of article 06's "the FMA reports intent": when the intent's source is dead, the FMA doesn't know. Monitoring falls back to trajectory-versus-expectation plus the circumstantial list.

4. After the wipe — the rebuild sequence

The residue after a data-losing dual reset (official list): the bias is lost so the FM position falls back to MIX IRS (article 20); VOR/DME autotuning recovers on IRS position but ILS and ADF autotuning cannot; managed modes won't engage on either axis; CAB PR LDG ELEV FAULT; REENTER ZFW/ZFWCG. The rebuild, in the book's order:

If the FMGC remains in Takeoff phase, CHANGE the vertical mode in order to recover the next flight phase. SELECT DIR TO the required downpath waypoint. SELECT LAT REV at the downpath waypoint and redefine DESTINATION. SELECT the FUEL PRED page and enter the ZFW and ZFWCG values. SELECT the PROG page and enter CRZ FL. SELECT the PERF page and enter CI. CHECK or REENGAGE (as appropriate) the relevant speed/Mach target and vertical mode.

Escape the wrong phase, DIR TO a downpath fix (article 21's master key), re-hang the destination, then weights, cruise level, cost index, and re-check the targets — finishing with a navigation-accuracy check (a MIX IRS more than 20 NM adrift is a legitimate scene for the otherwise extreme-caution manual position update — article 20). The order is deliberate: first make it flyable (DIR TO), then make it able to count (weights and CI).

5. FMS DEGRADED MODE — the half-survivor

Per FCOM DSC-22_20-40:

‐ The predictions are lost on the affected FMS(s). All predictions are replaced by dashes on MCDU pages. ‐ The vertical profile and managed speed computations are lost on the affected FM(s) (in Approach phase, VAPP is correctly computed and managed speed is available). ‐ ... ‐ Lateral path computation and guidance remain available. ‐ The MCDU displays the 'FM PREDICTIONS LOST' message.

After repeated resets the system may settle here: predictions all dashes, vertical managed gone, lateral kept (FM PREDICTIONS LOST) — save what can be saved. With the approach exception: VAPP still computes and managed approach speed survives (article 17's 700 ft memory family). The recovery key: re-entering the cost index forces a prediction recomputation — but the CI cannot be entered directly in the approach phase, so the official detour runs through article 22's phase machine: enter the current altitude as CRZ FL to force the cruise phase back, enter the CI, then manually re-activate the approach phase.

6. Resets on approach — the 700 ft double insurance pays out

Per FCOM DSC-22_20-40:

Above 700 ft AGL: ILS/GLS tuning may be lost. The loss of ILS/GLS tuning, due to a dual reset, will cause a loss of the LOC and G/S, and the disengagement of the APs and FDs. ... Below 700 ft AGL: A single or double reset does not affect an ILS/GLS approach below 700 ft AGL. ILS frequency/GLS channel is locked and AP/FDs remain engaged. ‐ CONTINUE the approach.

Above 700 ft, a dual reset can take the tuning, the beams and the autopilots — unstable by 1000 ft or no visual: go around. Below 700 ft, neither a single nor a double reset touches the ILS approach — the frequency is locked in the MMRs (article 11's ILS TUNE INHIBIT!) and the AP/FDs hold: continue. The freeze that read like a restriction in article 11 turns out to be a life-jacket: it blocks your slip of the hand and it blocks a dying FM from taking the frequency with it. Non-precision approaches have no such insurance — a master-FMGC death there means managed everything is gone; decide at the stabilisation height.

7. TAB — the official catalogue of known pits

The temporary-abnormal-behaviour chapter is the manufacturer's own diary of known software behaviours (the table below provides a list of all Temporary Abnormal Behaviors) — the AUTO FLT family runs to roughly 25 entries, each with triggers and recommendations. Four high-frequency/high-consequence cases:

• Crossing the 180th meridian — per FCOM: a single or a dual FMS reset with automatic recovery may occur when the aircraft crosses the 180th meridian. Defensive move: park a FIX INFO radial on the line and be mentally ready (the symptoms are section 1's five).
• DIR TO/ABEAM with three steps — a TO fix outside the plan + three or more STEP points + a stepped fix on the direct segment → a dual reset; the failed insertion leaves the TMPY hanging — retry it three more times (four in 3 min) and the library is wiped. Avoidance: in that geometry use plain DIR TO or DIR TO/INTCPT, not ABEAM. The one-minute discipline's most concrete incarnation: a failed action is not to be re-pressed.
• SEC contaminating the active plan (spurious forward fuel transfer) — an AOC uplink loads a SEC carrying a stale MIN DEST FOB; the value won't clear, EXTRA computes wrong, and activating the SEC transfers the contamination into the active plan — the TAB title names the endpoint: potentially a spurious forward fuel transfer (a wrong fuel conclusion driving the trim-tank logic). Recognition: EXTRA ≠ EFOB at destination − MIN DEST FOB. Cure: re-enter a MIN DEST FOB (to get the system value back, enter any value then clear it).
• UPLINK TO DATA prompt mis-seated — change the runway inside SEC then activate, and the INSERT UPLINK prompt can appear on the wrong runway's page. Discipline: before inserting, verify the T.O DATA against the T.O RWY; prompt on the wrong page → send a fresh TO DATA REQUEST (article 28's four gates, patched).

The rest of the catalogue (know where to look, not by heart): enhanced-LOC-capture misbehaviour (article 11), a QFE-takeoff early CLB engagement (QFE-fitted aircraft), an uncommanded climb near T/D, DIR TO RADIAL IN failing to sequence, MCDU freezes entering DIR TO/HOLD/OFFSET, VFE NEXT mis-drawn above twenty thousand feet (article 18's neighbour). The habit: a weird symptom goes to the TAB first, the reset button second.

8. The decision tree

Symptoms appear → place yourself on the ladder (one self-healing reset? latch? independent? degraded?) → self-healed: wait a minute after PLEASE WAIT clears before re-engaging that side's AP → latched: healthy-side AP → manual reset, one computer → failing that, reset breaker + BOTH ON → both latched: raw data + RMP tuning + resets one by one → nothing returns: backup navigation (article 19). Three abstinences throughout: never pull the MCDU breaker, never re-press the same action inside the minute, never reset both computers at once in flight. The FCTM adds: consider the QRH reset after a single failure; with both lost, rescue the FG first (AP/A-THR back) before mourning the FM. And two neighbours filed here for completeness: prediction weirdness is not a reset scenario (the no-reset three-step triage — article 19); a spurious engine-out takes E.O. CLR, not a reboot.

[!warning]- Five misconceptions this article corrects (1) A dead keyboard under PLEASE WAIT is normal — and pulling the MCDU breaker only scrambles the resync. (2) The one-minute discipline is not superstition — your last entry probably triggered the dual reset, and repeating it four times in three minutes wipes everything you typed. (3) A perfectly normal FMA can be perfectly wrong — the frozen-target failure guides the aircraft on a dead computer's last words; before disconnecting AP2 on that approach, go selected and kill the A/THR, or the thrust will chase a fossil speed at 50 ft. (4) The below-700 ft tuning freeze is protection, not restriction — it blocks both your hand and a dying FM. (5) Manual resets never latch the computer — the counter only counts automatic ones; press away (one side at a time).

Self-test

[!note]- Q1. How long is a reset, how long a resync — and is a resync always reset-triggered?

Reset 2–3 s maximum; resync about 25 s. No — the FMs also initiate a resync on their own when cross-comparison finds a disagreement.

[!note]- Q2. The data-wipe count (both standards) — and what does the one-minute discipline prevent?

Four dual resets in 3 min (other standard: three in 2 min) wipe pilot-entered data. The discipline — don't repeat the last MCDU action for a minute — prevents you from re-triggering the reset and completing the wipe yourself.

[!note]- Q3. Keys dead under PLEASE WAIT — the right and the wrong reaction?

Right: wait; it is normal (and wait a further minute after PLEASE WAIT clears before engaging that side's AP/FD). Wrong: pulling the MCDU circuit breaker.

[!note]- Q4. The single-latch count (both standards) and the two-stage response?

Five successive resets in 4 min (other standard: four). Stage one: healthy-side AP and managed modes. Stage two: manual overhead reset of the latched FMGEC — success resyncs to dual; failure → reset breaker pulled + FM source BOTH ON the healthy side.

[!note]- Q5. The manual reset's three trump cards?

One FMGEC at a time in flight; it disconnects the onside AP; and it does not increment the reset counter — unlimited manual resets.

[!note]- Q6. Four circumstantial signs of the frozen-target failure — and the two actions before disconnecting AP2, with the mechanism?

Failed side: MAP NOT AVAIL + GPS PRIMARY LOST on the ND, frozen MCDU; opposite MCDU: INDEPENDENT OPERATION; possibly FM FAULT/CAB PR LDG ELEV FAULT on ECAM. Before AP2 off: revert to selected speed and disconnect the A/THR — else A/THR mastership defaults to the dead FMGEC 1 at AP2 disconnection and the thrust chases its frozen speed target.

[!note]- Q7. After a data wipe, which autotuning returns and which cannot — and the rebuild's first two steps?

VOR/DME autotuning recovers on IRS position; ILS and ADF autotuning cannot. First: change the vertical mode to escape the stuck phase; second: DIR TO the required downpath waypoint — flyable first, countable second.

[!note]- Q8. What does DEGRADED MODE preserve — and how is the CI re-entered in the approach phase?

Lateral path and guidance (predictions dashed, vertical managed lost; VAPP and managed speed survive in the approach phase). CI re-entry detour: current altitude into CRZ FL to force the cruise phase, enter the CI, manually re-activate the approach phase.

[!note]- Q9. A dual reset on an ILS approach — the fates above and below 700 ft, and the mechanism?

Above: tuning may be lost → LOC/G-S lost → APs and FDs disengage (go around if not stable/visual). Below: no effect — the frequency is locked in the MMRs (ILS TUNE INHIBIT) and the AP/FDs remain; continue. Non-precision approaches carry no such insurance.

[!note]- Q10. One-line trigger and avoidance for the three worst TAB pits?

180th meridian: crossing may cause a self-recovering reset — anticipate with a FIX INFO marker. DIR TO/ABEAM + 3 steps: dual reset — use DIR TO or DIR TO/INTCPT instead, and never re-press the failed insertion. Stale MIN DEST FOB via SEC uplink: wrong EXTRA and possible spurious forward fuel transfer after activation — re-enter a MIN DEST FOB.

Key takeaways

References

Reset/resync mechanics and counters (both FG standards), in-progress symptoms and the breaker prohibition, single-latch handling, the frozen-target scenario and its AP2/A-THR trap, the rebuild sequence, degraded mode and the CI detour, the approach 700 ft split, and the manual-reset properties per FCOM DSC-22_20-40 (reset section); the temporary-abnormal-behaviour catalogue and the four detailed cases per the FCOM TAB chapter and its ATA-22 detail pages. Single/dual failure techniques per FCTM PR-AEP-AUTOFLT. The ladder diagram and decision tree are integrative syntheses. GLS wording is quoted as printed; the function is not installed on the baseline airframe of this series.

Independent study material, not an Airbus publication and not endorsed by the manufacturer. Always defer to the current operator FCOM, FCTM, and QRH for operational use.