One rule throughout: a tree, not a loop — exactly one path from any node back to the common negative.
HOUSE batt (−) --[2 AWG]--> SmartShunt "BATT−" <- the ONLY thing on the shunt's battery side
START batt (−) -+-[2 AWG]--> SmartShunt "SYS−" --> 5026 NEGATIVE BUS -+-> bilge pump (−)
| +-> Cerbo GX + GX LTE (−)
| +-> fridge (−)
| +-[2 AWG]-> 8377 NEG BUS --> panel loads
| (the one link across the hinge)
+-[2 AWG]--> 2203 BULKHEAD STUD <-[jumper]- ENGINE BLOCK <-+- Starter (−)
(system negative) +- Alternator (−)
<-[2 AWG]-- Lazarette COMMON − lug (both chargers)
SmartShunt aux ....> START batt (+) voltage only - sense, no current
SmartShunt B+ ....> 100 A breaker, BATTERY side short run; keeps counting through a trip
| # | From → To | Cable | Protection | Notes |
|---|---|---|---|---|
| P1 | House batt (+) → house Class-T | 2 AWG | Class-T 200A | on the shelf beside the battery, inside ABYC’s 7 in. 2 AWG is rated 210 A, so 200 A protects the feeder with no overload window |
| P2 | House Class-T (load) → 100 A breaker → 6011 BAT-1 | 2 AWG | 100 A breaker | 🔑 100 A because it protects the 8377’s buses, which are rated 100 A — a 200 A Class-T cannot. Do not tidy this up to match the start side. Resettable, so an accidental panel overload never costs a Class-T fuse |
| P3 | Start batt (+) → start Class-T | 2 AWG | Class-T 200A | same arrangement. 200 A on both sides so the boat carries one spare denomination |
| P4 | Start Class-T (load) → 150 A breaker → 6011 BAT-2 | 2 AWG | 150 A breaker | the only breaker in the crank path. Cranking at ~100 A is 67 % of rating, and the breakaway peak lasts tens of milliseconds — far too fast for a bimetal |
| P5 | 6011 LOAD-2 → starter solenoid | 2 AWG | cranking conductor | 🔴 straight to the solenoid through a grommet — it does NOT use a bulkhead stud. Carrying the switched crank feed is this cable’s whole job. 16 ft loop = 0.25 V, 2 % |
| — | 6011 LOAD-1 → 8377 positive bus | 2 AWG | — | unbroken — the panel feed has no device between the switch and the bus |
| P6 | House charger lug → house Class-T (battery side) | 8 AWG pigtail → 2 AWG feeder | 40A | fuse sits at the charger end, on the pigtail — the smallest conductor, and shorter than ABYC’s 7 in allowance. ⚠ The only cable still running the full ~6 ft to the lazarette, the studs being single-circuit |
| P7 | Start Class-T (load) → 2204 bulkhead stud | 2 AWG | — | the unswitched start-positive node — 250 A, tin-plated brass, sealed and isolated from the bulkhead |
| P8 | 2204 → alternator B+ | 2 AWG | none — deliberately unfused | a fuse in an alternator output is a load-dump mechanism, and landing unswitched means turning the battery switch off while the engine runs cannot damage it — which is why that rule is not on the operations cards. Protection is mechanical: abrasion sleeve, secured every 18 in, chafe guard at every bulkhead pass |
| P9 | 2204 → start charger (+) lug | 2 AWG → captive lead | 30A | ⚠ confirm the IP67’s captive-cable gauge before finalising the fuse |
| P10 | 100 A breaker (load) → +ve bus (Blue Sea 5026) → 6011 BAT-1 | 2 AWG | the 100 A breaker | 🔑 The block is a TAP, not a link in the chain. Feed in and feed out are two rings stacked on the same #10-32 stud, so house current crosses ring-to-ring and never enters the block’s bus — which is why 12 positions on a 100 A block is not a pass-through problem. No intermediate fuse: 2 AWG is protected by the 200 A Class-T, the block sits behind the 100 A breaker, and its bus carries only what its own fuses distribute. The 40 A MAXI and the 5006100 holder are deleted. ⚠️ The rule that replaces them: unswitched positions are for small always-on loads, so the branch total stays well under 100 A. ⚠️ Two 2 AWG lugs on one #10-32 stud — check stud length, torque to 24 in-lb and re-check it, because with ring-to-ring conduction the joint IS the clamping force |
| F1 | 5026 → AC panel backlighting | 16 AWG | 1A | 🔑 Run 16 AWG to everything on this block, and note the fridge is unswitched with its own switch at the icebox — fuse at the source, switch at the load. ⚠️ That takes the unswitched draw from ~10 Ah/day to 30–50, so a lost shore connection flattens the bank in two to three days rather than ten: a VRM shore-power-loss alarm is a commissioning requirement. Then any fuse denomination protects the wire, and you never size a fuse down to a thin factory pigtail. Fuses become pull-out-and-replace instead of cut-and-recrimp |
| F2 | 5026 → Cerbo GX + GX LTE | 16 AWG | 5A | |
| F3 | 5026 → bilge pump | 16 AWG | 10A | |
| F4 | 5026 → fridge (Isotherm GE80) | 16 AWG | 15A | |
| N1 | House batt (–) → SmartShunt “BATT–” | 2 AWG | — | 🔴 the ONLY thing on the shunt’s battery side. Bond the house post to anything else and house current finds a path around the shunt — every state-of-charge figure becomes wrong but plausible |
| N2 | Start batt (–) → SmartShunt “SYS–” | 2 AWG | — | the start battery sits on the system side, so cranking current never passes through the shunt |
| N3 | SmartShunt “SYS–” → 5026 negative bus | 2 AWG | — | the bay busbar is gone; the bay holds batteries only |
| N3b | 5026 negative bus → 8377 negative bus | 2 AWG | — | 🔑 the one negative link that crosses the hinge. Every unswitched load — bilge, Cerbo, GX LTE, fridge — returns to the 5026’s bus instead, so both legs terminate on the fixed back wall and nothing flexes when the panel opens. Confirmed: the negative bus has a single heavy stud, so both 2 AWG lugs stack on it and house return crosses ring-to-ring — the bus carries branch returns only, exactly as on the positive side. |
| N4 | Engine block → 2203 bulkhead stud | 2 AWG | — | the system-negative node. Starter and alternator negatives land on the block, which lands here on a short jumper |
| N5 | 2203 → start batt (–) post | 2 AWG | — | the one long run forward. 🔑 Each stud replaces two long cables with one long run and two short jumpers, so every long cable terminates where it can be seen and re-torqued |
| N6 | Lazarette COMMON – lug → 2203 | 2 AWG | — | both chargers share the one negative lug; it reaches the shunt’s system side through the stud and the start battery post |
| S1 | SmartShunt aux → start batt (+) | sense wire | fused at the start Class-T load side | start-battery voltage only. The aux does voltage or temperature, not both. 🔑 A low-voltage alarm here catches a wrong charger profile, a dead charger, a tripped breaker and a failing battery, all at once — which is why the controls-free IP67 was the right choice |
| S2 | SmartShunt B+ → 100 A breaker, BATTERY side | supplied, own inline fuse | inline | a short run on the shelf rather than the length of the boat — and 🔑 it keeps the shunt powered through a breaker trip. The shunt counts amp-hours rather than measuring state of charge and only re-synchronises on a full charge, so an unpowered shunt comes back not knowing what happened in between. Still downstream of the Class-T, so the two-number rule holds |
| C1 | SmartShunt → Cerbo VE.Direct 1 | VE.Direct | — | bay → panel; VE.Direct comes in lengths to 10 m for pennies, which is why the Cerbo sits with the screen |
| C2 | House charger IP43 → Cerbo VE.Direct 2 | VE.Direct | — | lazarette → bay → panel |
| C3 | Cerbo → GX Touch 50 | HDMI + USB | — | short — the Cerbo sits directly behind the GX Touch on the DC panel. This pair is the length-constrained one and neither extends gracefully |
| C4 | Cerbo VE.Direct 3 → future solar MPPT | spare | — | solar-ready, no rework. Leave a pull-string lazarette → coachroof while cables are being routed anyway |
| C5 | VM-3P75CT → Cerbo | VE.Bus | — | shore V / A / kWh on the Cerbo and VRM — live total draw for AC load management |
| C6 | Cerbo → GX LTE 4G-A → VRM | USB | — | ⚠ data only — the LTE takes its DC supply from the 5026, sharing a branch with the Cerbo so their grounds match as the manual requires. Fixed 1.0 m USB-A lead, so it must sit within a metre of the Cerbo. 4G-A is the correct band variant for North America. Needs a Mini-SIM and ~$10–20/month. Keep the antenna clear of metal; fibreglass is RF-transparent |
| — | Start charger IP67 | — | — | Bluetooth only — it cannot connect to the Cerbo. Its battery is watched by the shunt’s aux input instead (S1) |
Gauges are sound minimums for these currents and the cranking path; on long runs, step up one size for voltage drop. Every battery (+) branch is fused within ~7″ of the post (ABYC).
Cable summary — three gauges (tinned marine; red for +, yellow for –): 2 AWG = every heavy run — both mains, the crank feed, the alternator, and every main negative · 8 AWG = the charger pigtails · 16 AWG = everything on the 5026. No 1/0 — at 16 ft and ~100 A the cable is not the problem, the joints are. 🔴 Do not bundle the heavy runs: 2 AWG is 210 A open, but bundled DC derates to ~147 A, which is under the 200 A Class-T. Run them separately or the fuse stops protecting the wire.
| # | From → To | Cable | Protection | Notes |
|---|---|---|---|---|
| A1 | Dock pedestal → shore inlet (SmartPlug) | 50 ft 10/3 cordset | dock 20 A | the dock is a 20 A service (5-20R receptacles; there are no 30 A slips at DCYC) ⇒ 16 A continuous is a permanent design constraint. The pedestal carries receptacles only — no breaker, no disconnect — so switch the boat’s AC main off before unplugging. Inlet is a SmartPlug: large flat contacts and a positive latch, which removes the twist-lock’s char-and-fail mode |
| A2 | ELCI MAIN — the shore disconnect | Blue Sea 8100 (breaker 3106100) | 30 mA · 30 A 2-pole | A-Series ELCI main, 30 A double-pole, 30 mA leakage, 120 V AC, own enclosure (5.25 × 3.75 × 3.50 in deep), red reverse-polarity LED, green ON LED. UL-based and ABYC E-11 compliant by design. 🔑 It is now the ONLY main — the 8097 downstream is a branch panel with no main of its own, so the earlier two-mains-in-series arrangement is gone. Position: at the panel, a deliberate departure. E-11.11.1 would place it at the additional overcurrent protection within 10 ft of the inlet (E-11.10.2.8.3), but nothing in that run is accessible or testable, and protection nobody tests is not protection. ⚠ Accepted residual: a ground fault inside the ~20 ft inlet→panel feed returns via the green conductor and never passes through the toroid. The dock GFCI covers that at DCYC; nothing does elsewhere. The feed itself is 8 AWG behind a 20 A dock breaker. TEST monthly — no bench test needed |
| A2b | Inlet L + N + G → 8100 → 8097 buses | 8 AWG 3-core | — | all three conductors pass through the 8100. Its outputs land on the 8097’s hot, neutral and grounding buses, which are part of that panel |
| A2c | Surge protection | — | — | Deliberately not fitted. Nothing aboard now diverts line or neutral energy into the earth path by design, so the isolator is not asked to carry surge current. If ever added: Type 2, MCOV ~150 V not 275 V (a European-rated arrester will not clamp low enough for 120 V equipment), status window, on the line side of the ELCI (downstream of an RCD it looks exactly like a residual current and trips it every storm), on the shortest practical lead — what limits an arrester is the inductance of its own connections |
| A3 | Inlet G → galvanic isolator → 8100 | 8 AWG green | VDI-32 ≥30 A | the isolator sits in the green earth only; L and N bypass it. Blocks sub-1.4 V galvanic DC, passes AC fault current. ⚠ ABYC A-28.9.2.1 requires fail-safe isolators to be labelled for a lightning strike — it fails in the harmless direction (safety ground survives, galvanic isolation may not), and does so silently, showing up as a disappearing anode months later. ⇒ test it after any close strike (two anti-parallel diodes; multimeter in diode mode). On the maintenance card |
| A4 | Permanent earth connections → 8100 grounding bus | green | — | 🔴 The engine-block bond and the bonding-system tie land on the 8100, not the 8097 — the 8100 is a fixed device that survives a panel change, and a branch panel is the thing most likely to be swapped again. The ground-to-DC-negative bond is made at the engine block and there is exactly one aboard. Two create a loop, and normal load current then divides between the negative conductor and the grounding conductor permanently — a stray-current corrosion path. Neutral is never bonded to ground aboard. ⚠ OPEN: the old panel’s ground bus carries seven conductors against the neutral bus’s five. Five neutrals is one in plus four branches exactly, which rules out a neutral-to-ground link; of the seven grounds, one is shore and four are branches, so two are unaccounted for — most likely the engine bond and a bonding tie already in place. Confirm before creating another: unplug from shore and ohm the ground bus to bare engine block, then ring each conductor out as the panel comes apart |
| A5 | CHGR branch → switched 2-gang outlet → both chargers | 10 AWG 3-core | 15 A | 🔑 The chargers plug in, they are not hard-wired. An IP66 weatherproof two-gang outlet on the lazarette board, each charger independently switchable at arm’s length — that outlet is the chargers’ isolator, which is why the DC side needs none. ~6 A AC for the pair. ⚠ Independent devices on independent gangs, but one shared 15 A breaker |
| A6 | W.HTR branch → water heater | 10 AWG 3-core | 15 A | Seaward S600, galley under-sink, ~12.5 A element — and Seaward specify a 15 A breaker. Engine-coolant heated as well, so the element is an occasional top-up rather than a routine load. Currently deferred and kept switched OFF |
| A7 | OUTLET branch → cabin receptacles | 10 AWG 3-core | 15 A | — |
| A8 | A/C branch → air conditioner | 10 AWG 3-core | 15 A | Mabru SC09VI, 9,000 BTU variable-speed inverter, 1.3–5.2 A, replacing the King D16000 whose compressor is mechanically dead. 🔑 An inverter compressor ramps from zero, so there is no start inrush at all — which is what makes 15 A comfortable where the old unit drew 18.4 A sustained. The existing raw-water circ pump (1.9 A) is retained and fed through the A/C unit, the SC09VI shipping without one. ⚠ The return-air rebuild is not optional: 9k sits at the bottom of the computed range, so ≥80 in² free return area is load-bearing for the sizing |
| — | Two spare positions | — | — | the fridge takes one, leaving one |
AC load management — 20 A dock, 16 A continuous. Mabru at full plus both chargers is ~11 A; throttle the A/C and cap the chargers and even the water-heater element fits. 🔑 The SC09VI’s 1.3–5.2 A turndown is what makes that possible — a fixed-speed unit would not co-exist with the element at all. The VM-3P75CT shows live draw. Boat-side hardware is all 30 A fed from a 20 A source, so the dock breaker is the only real protection for the shore cord — and it lives in a distribution panel elsewhere on the dock, not on the pedestal. Full AC detail: knowledge/electrical-system.md · by-location AC sheet: swallow-ac-schematic-zoned.html.