Swallow — DC cable routing, physical plan

v0.8 · 9 September 2026 · transom left, bow right, port at the top · cropped at the battery bay — nothing electrical lives forward of it
A working plan, not an as-built record. None of this is fitted yet. The drawing exists to settle where the heavy cable actually goes before any of it is cut, because on a twenty-seven foot boat the difference between a schematic and a route is several feet of expensive copper and a good deal of swearing. Positions and cable routes are geometric; component boxes are schematic.
🔴 DO NOT BUNDLE THE HEAVY RUNS. This is a design requirement, not neatness. ABYC's derate for bundled DC conductors is a flat ×0.7 for any bundle at all — not scaled by count, so two cables tied together is a bundle. That takes 2 AWG from 210 A down to 147 A, and a 200 A Class-T protecting a 147 A conductor is over-fused. The same applies to P6, P7 and P12. ⇒ separate clamps, an air gap between cables, and no zip-tying the heavy positives to each other. Running parallel but spaced is not a bundle; running in contact is. The whole 200 A Class-T decision depends on this, and it is the kind of requirement an installer undoes by tidying up.
🔴 The engine block ground is not confirmed. It is drawn dashed, at the aft end of the engine on the port side of the gearbox, which is where something that looked like a ground was seen. But the wire there is thin and black, and a cranking return should be about the gauge of the crank positive. So either that is not the block ground, or it is and it is badly undersized — which would be a live candidate for the intermittent starting fault, since a high-resistance return gives exactly that symptom. The test needs no cable-hunting: measure the volt drop from the engine block to the start battery's negative post while cranking. Above about 0.2 V and that is the fault.

Plan view

Plan view, transom to the forward edge of the battery locker. Aft berth ghosted — it is below the cockpit sole, so in plan it projects under the lazarette. Cable routes are geometric between regions but indicative where they pass behind fixed items such as the water heater.

Detail — the port locker

Detail — the port locker complex at ≈ 2.2× the main plan. Three athwartships zones. The panel's contents are an elevation, so they stay in panel-layout-dc.html; only its terminals appear here. ⚠️ The two 150 A breakers are drawn in the panel strip for clarity. In practice they mount on a small sub-panel on the BACK of the fold-down panel, carried on the GX Touch's mounting rods — so they face you the right way up when the panel drops, and the whole heavy circuit stays on the panel instead of crossing the hinge twice. Stud positions on the 6011 are indicative, drawn as BAT inboard, facing the shelf the feeds come from, and LOAD outboard. 🔑 CONVENTION, and the drawing is laid out to obey it: on every protective device the BATTERY side is at the BOTTOM and the LOAD side at the TOP. Each breaker is stacked directly above its own Class-T, so power reads upward through the picture and the two sides of a device can never be confused. P10 comes off the breaker's bottom edge for exactly that reason — it is on the battery side of the trip mechanism.

Stations

Cable schedule

IDRunSizeNotes
P1House battery + → house Class-T, battery side2 AWGStraight up from the battery onto the shelf
P2House Class-T load → house 100 A breaker2 AWGInches — straight up, the breaker stacked directly above its own Class-T. 🔑 100 A, and the number is not arbitrary: it protects the 8377's BUSES, which are rated 100 A. Nothing else does — a 200 A Class-T sits upstream, so without this breaker a bus-level fault has no device sized to it. That is the whole reason the breaker exists, and it is why this one must not be quietly upsized to match the start side
P3Start battery + → start Class-T, battery side2 AWG
P4Start Class-T load → start 150 A breaker2 AWGSame arrangement. 🔑 150 A here, because this is the only breaker in the crank path: cranking at ~100 A sits at 67 % of rating, and the breakaway peak is over in tens of milliseconds — far too fast for a bimetal to see
P56011 LOAD-2 → starter solenoid2 AWGCrank positive, out of the switch and straight aft to the solenoid — not via a stud, because that is the switch's whole job. This is what makes cranking switchable. It still crosses the bulkhead, but through a grommet
P6Lug plate HOUSE + → house Class-T load2 AWG≈ 6 ft measured. 75 mV at 80 A. The only cable still running the full distance
P7Start Class-T load → 2204 stud2 AWG🔑 The shared unswitched run. Alternator and start charger both hang off the far end of it, so this one cable replaces what used to be two long ones
P82204 stud → alternator B+2 AWGShort jumper into the blind zone, made on the bench. 🔴 DELIBERATELY UNFUSED — a fuse in an alternator lead is itself a load-dump mechanism
P92204 stud → lug plate START +2 AWGShort jumper. The start charger's own fuse sits at the charger end, on its 8 AWG pigtail
P10House breaker INPUT side → fused positive busbar (unswitched)8 AWG, 5006100 + 40 A MAXI at sourceFeeds bilge · Cerbo · GX LTE · shunt sense. 🔑 Shares the breaker's AFT terminal with P2 — upstream of the trip mechanism, so it survives the master switch and a breaker trip, which is the whole point of it, while leaving the Class-T load stud at two conductors. ⚠️ The simpler route was considered and rejected on purpose: taking it straight off 6011 BAT-1 would be tidier, and it would go dark the moment the house breaker tripped. Do not "simplify" it back. Sized for a ceiling of ~40 A if a second bilge pump ever lands on that bus; realistic load is ~11 A. 10 AWG would do, 8 AWG has margin and is on hand. 🔴 It needs its own fuse at the source — see the assumptions
P11House 100 A breaker → 6011 BAT-12 AWGShelf up to the panel
P12Start 150 A breaker → 6011 BAT-22 AWGShelf up to the panel
—6011 LOAD-1 → the 83772 AWGPanel-internal, and unbroken: the protection sits upstream on the two 150 A breakers, so there is no breaker in this leg
N1House battery − → SmartShunt, BATTERY side2 AWG🔴 The only cable on that post. About 1.5 ft as drawn
N2Start battery − → SmartShunt SYSTEM side2 AWGStart current never crosses the shunt
N3SmartShunt SYSTEM → panel negative busbar2 AWGThe single negative arriving at the panel
N4Engine block ground → 2203 stud2 AWGShort jumper, made on the bench, fitted once. 🔴 Block landing point unconfirmed
N52203 stud → start battery − post2 AWG🔑 The mirror of P7 — one shared run doing two jobs, carrying the crank return and the chargers' common return together. Lands on the battery post, not the panel and not the shunt, which keeps the crank loop at 16 ft and the start post at two conductors
N6Lug plate common − → 2203 stud2 AWGShort jumper. ❌ This used to be drawn running the full 6 ft to the shunt — wrong. It terminates at the stud like everything else on that side, and reaches the shunt through N5
B1Panel → bilge pump + float, port aft14 AWG10 A fuse. Unswitched

Assumptions

Everything the drawing rests on that has not been measured twice. Each of these gets confirmed aboard before anything is cut.

  1. Span 17.1 ft mast to transom, companionway at 8.5 ft — scaled off the factory drawing, checked against the published J. The drawing is cropped at 12.0 ft.
  2. Engine ≈ 2.2 ft long including the gearbox, occupying 6.3 → 8.5 ft.
  3. Alternator at the forward (belt) end; starter and solenoid on the port side against the water-heater bulkhead — confirm.
  4. 🔑 Shelf laid out to separate negatives from positives, aft to forward: SmartShunt · SHURflo · the start pair · the house pair. The shunt sits alone at the aft end with the pump between it and the positive group — maximum distance between a dropped tool and a bridged pair. 🔑 Each breaker is stacked directly above its own Class-T, so the hop between them is inches, the two Class-T load studs never meet, and every device reads battery-at-the- bottom. ⚠️ Sizes are drawn to fit the convention, not measured — check the real 5007 and 187-series footprints against the shelf before committing to this arrangement.
  5. ✅ Every terminal in the chain carries at most TWO conductors. That is what moving the breakers onto the shelf bought, together with taking the unswitched tap off the breaker's battery side rather than the Class-T. Both Class-T load studs carry two (charger feed in, breaker feed out); the house breaker's battery side carries two (P2 in, P10 out); every breaker output and every switch stud carries one. ⚠️ The only three-way nodes left are the two bulkhead studs, which are distribution points by design.
  6. 🔴 THE CHARGING FEEDS ARE ON THE BATTERY SIDE OF THE BREAKERS, AND THAT IS THE POINT OF PUTTING THEM THERE: NO BREAKER BETWEEN THE ALTERNATOR AND ITS BATTERY. P7 carries the alternator, so on the load side a trip — or a manual open while the engine is running — would be a load dump. That is the exact failure the unswitched, unfused alternator landing exists to prevent, and the reason "never switch off while the engine is running" is not on the crew cards. Supporting, in descending weight: the breakers are protective devices deliberately sited so as not to be used as switches; charging should survive a trip for the same reason P10 does; and the chargers already have a purpose-built isolator — they are plugged into independently switched sockets in the lazarette, so DC-side isolation was never the right tool for it anyway. Service isolation is the Class-T's job, or the battery post's.
  7. ⚠️ Cross-starting puts the full crank current through the 100 A house breaker — deliberately, in the "trip the start breaker, then COMBINE" case. At ~100 A for ten seconds that is 100 % of rating, and a thermal breaker carries 100 % indefinitely, so it holds even allowing for hot-locker derating. Noted only so that if a cross-start ever does trip it, the cause is understood rather than chased as a fault.
  8. ✅ P10 IS FUSED AT ITS SOURCE — Blue Sea 5006100 holder with a 40 A MAXI, at the breaker end. It was needed because ABYC allows a conductor to run unprotected only 7 in from its source (40 in if sheathed), and the busbar is at the panel while the house breaker is on the shelf — so behind only a 200 A Class-T the 8 AWG had nothing sized to it. (The earlier "keep it short and heavy, inside 7 in" note was written when the busbar lived in the bay beside the Class-T; moving it to the panel broke that condition, and the sentence guaranteeing it went with the bay.) ✅ Block confirmed: Blue Sea 5045, 100 A per block, 30 A per circuit. The block's own blade fuses do all the load protection, so this fuse only ever operates for a fault on P10 itself or on the block's bus. 🔑 40 A CHOSEN DELIBERATELY OVER THE 80 A THE CABLE WOULD ALLOW — 8 AWG at 105 °C outside engine spaces is 80 A (ABYC E-11 Table 6A), so 40 A is half the wire and always safe; under-fusing never creates a hazard, only over-fusing does. The reason is SPARES STANDARDISATION: the same 40 A MAXI already protects the 8 AWG charger runs, so the boat carries one size rather than two. That is the same argument that settled the 200 A Class-Ts — reaching for the wrong fuse in the dark, blowing it again, and not knowing why, is a real failure mode. ⏰ The one thing that would force a revisit: a high-flow SECOND bilge pump. That plus the Jabsco plus the electronics is ~31 A with both running — 78 % of the fuse, so it still holds, but nothing else on this bus comes close. Realistic load today is ~11 A. ⚠️ Figures assume 105 °C insulation (90 °C is 70 A, 75 °C is 65 A — read the jacket) and an unbundled run, per the warning at the top. ⚠️ Both 5007 covers fitted, which the build already specifies.
  9. 🔑 THE STUDS ARE NODES, NOT PASS-THROUGHS, AND THAT IS THE WHOLE POINT. Each carries three conductors and each replaces two long cables with one long run plus two short jumpers. 2204 is the unswitched start-positive node — start Class-T load in, alternator and start-charger out. 2203 is its mirror on the negative side — block ground and charger common in, one run forward to the start battery post. ⚠️ Each is a SINGLE stud, so only one circuit of each polarity can use them; the house charger positive (P6) therefore still runs the full distance on its own.
  10. ⚠️ Terminal counts. Start battery negative carries two ✅ (shunt, N5). The shunt's SYSTEM side carries two ✅ (start −, panel negative main) — the charger common no longer lands there. Both studs carry three, and the house Class-T load side carries three (charger feed in, breaker feed out, unswitched tap). All legal; ABYC allows four.
  11. 🔑 Cranking draw taken as ~100 A sustained, not 250 A. Backed out two ways that agree: the factory 4 AWG over a ≈ 20 ft loop reaches a 0.5 V drop — the practical floor for a diesel — at 100.6 A; and Bosch quote 65–80 A normal cranking for the WAI 31282N starter fitted. On 2 AWG over the new 16 ft loop that is 0.25 V, or 2 %. ⚠️ Still an inference. Clamp the crank cable during a start (DC/Hall clamp, max-hold) to confirm it. 🔑 And note the copper is no longer the dominant term — the Class-T, the breaker, the switch and eight terminations add about as much again, which is why the corroded joints matter more than the gauge.
  12. Locker zones drawn as parallel fore-and-aft strips — is that a plan split, or are they stacked vertically? Changes the cable entries, not the schedule.
  13. Bilge pump port, below the lazarette, under the aft berth, ≈ 4.2 ft.
Colours: red = DC positive, amber = DC negative, purple = bilge branch, dashed grey = undecided or unconfirmed. AC not shown. Geometry is driven by the ST block in the source.