v10.2, 2026-10-08. The v10.1 guides with every decision made through 2026-10-08 written in. What is still open is in Pending Amendments v10.3, the only open list. Nobotic USV, Cape Dory 25D conversion. What this is. The boat's electronics: every box and every hole through its walls, the wiring, the sensors, power distribution (the bus, its fuses and breakers), the emergency bank, the lights, the comms hardware (cameras, AIS, Starlink, Iridium, GPS), the CAN network, the two team-built release and monitoring nodes, the battery watchman's box (the watchman is the small board that watches the battery and reports it; it controls nothing) and the parts to buy. Where each fact lives. This plan is the home for the boxes, wiring, hole lists, sensors, power distribution, the emergency bank, nav lights and comms hardware. The Compute Bay Housing and Wiring Plan is the home for what is inside the two compute boxes, the watchdogs, the central node's relay map and command paths, and electronics cooling, and it is the final word on the two-island layout, the lightning watch and the surge rules. Every setting, limit, threshold, timer and address: System Settings Sheet. Parts and prices: Bill of Materials. Cables to buy: Wire and Cable Buy List. The watchman's build and the charge-stop wiring: Battery Box Building Guide Part 8. The wing's own electrics since the slip ring (the ring, the wing compass, the tip wind meter and the wing box): Wing Power System sec 1A.
| Part | What it holds |
|---|---|
| Sections 2 to 7 | The design |
| Section 8 | The buy list |
| Section 9 | Open questions: in Pending Amendments v10.3 |
| Appendix A | Every box and every hole |
| Appendix B | Order status |
Companion documents: Glossary; Design Decisions and Rationale (why); Node Boxes Explained (the wiring pictures); the Box Wiring Sheets (Box_Wiring_Sheets_v10_2, one drawing per box); Compute Bay Housing and Wiring Plan; System Settings Sheet; Wire and Cable Buy List; Battery Box Building Guide; Lightning Ground System Guide; Steering System Handbook; the battery software notes (BMS Software Change Notes, Reference folder); Wing Power System; Serial and Network Summary (the port audit of 2026-09-08/09). Terms used here.
| Term | Plain meaning |
|---|---|
| Island | Gear fed from one battery, with no wire path to the other group |
| Node | A small team-built box with a microcontroller (a small computer chip) in it |
| CAN bus | The shared two-wire network the boat's devices talk over |
| Gland | A screw-in fitting that clamps and seals a cable entering a box |
| Drip loop | A sag in a cable, so water runs to the bottom and drips off |
| Ferrite | A clip-on magnetic collar that soaks up surge current riding on a cable |
| Staked | A dab of adhesive on a plug so vibration cannot walk it loose |
| De-energized-closed | A relay that passes current with no power on it, so a dead relay does not drop the load |
| Opto board | An input board that reads switch contacts across a gap of light, so no wire crosses |
| PoE | Power over Ethernet: the network cable carries the device's power as well as data |
| PoE class | What a PoE device asks the switch for (802.3af up to 15.4 W, 802.3at up to 30 W); the switch holds that much back, used or not |
| SFP module | Turns a switch's SFP slot into a port; the copper one gives an RJ45 jack, data only |
| Web API | The switch's own documented commands; the Pi uses them to switch a port on or off |
| SNMP | A standard way to read a network device's state; read-only on the TSW202 |
| H-bridge | A four-switch circuit that runs a DC motor either way; the steering drive is one |
| RS-232, RS-485 | Two wired serial standards: RS-232 is one device per cable, RS-485 a shared two-wire line |
| FTDI | Maker of the USB-to-serial chips this plan allows; each has a unique serial number, so the computer can tell them apart |
| Remote on/off cable | Victron's cable into a solar charger's VE.Direct port: voltage on its yellow wire lets the charger run, none stops it |
| MPPT | One of the five Victron solar charge controllers |
| TVS diode | A surge diode: it clamps a voltage spike before it reaches the board |
| INA226 | A small chip that reads volts and amps on one feed |
| IMU | A motion sensor chip; with GPS track it gives heading |
| UART | A plain serial port on a chip |
| N2K | NMEA 2000, the marine CAN standard the em-trak speaks |
| Isolated | Of a converter or adapter: its two sides share no wire, so a surge or a ground shift on one side cannot pass to the other |
| Slip ring | Sliding contacts that pass wires across a turning joint; here, the wing's turning joint |
| Wing box | The small box below the deck where the wing's wires enter the hull; the slip ring's wires land in it (A17) |
There is no section 1.
Both are the CXCESNS 11.8 x 8.3 x 3.9 in aluminum boxes already on hand. What goes inside them, how they mount and how they stay cool is the Compute Bay Housing and Wiring Plan's (secs 2 to 4). Every hole: Appendix A1 and A2 here. The watchdogs' parts, in short (the watchdogs themselves: Compute Bay Plan secs 4.1 to 4.3). WD-A sits in the Pi A box and WD-B in the Pi B box. Each holds its Pi's 24 V feed, ahead of the Pi's 5 V converter, through the normally closed contact of a 5 V 1-channel relay module (AEDIKO): coil off means Pi on, so a dead Nano leaves its Pi running. Each watchdog's own 5 V (the Nano, the relay coil and the input side of its optocoupler) comes from its own Pololu D36V6F5, ahead of the relay, so the relay never cuts it. The alive line and the VESC kill line cross on HiLetgo PC817 optocoupler boards. WD-B reads the lightning detector through a 4-channel two-way level shifter (the 3.3 V detector to the 5 V Nano) inside the Pi B box, and never cuts Pi B's power for lightning. None of this adds a box hole.
Compute Bay Plan sec 3.3.
Compute Bay Plan sec 3.4, and its USB rule in sec 4.1.
Pi B is through-bolted to the settee panel and Pi A is glued to the hull: Compute Bay Plan secs 2.2 and 2.3. Plastic boxes: sec 7.4 here.
Compute Bay Plan sec 3.5.
Compute Bay Plan secs 2.3, 2.4 and 3.6, with the pass/fail heat test. The four boxes glued to the hull for cooling, the network switch box among them, and the aluminum plate that cools the motor controller: Compute Bay Plan secs 2.3 and 4.4.
Inside: Compute Bay Plan sec 4.1. Every hole: Appendix A1. Its bond stud wires the box to the lightning copper.
Inside: Compute Bay Plan sec 4.2. Every hole: Appendix A2.
Power from the emergency bank only; every data link optical or transformer coupled; no metal shell or shield tying Pi B's box to Pi A's (plastic glands; plain UTP Ethernet, so the metal of its bulkhead jack touches no wire or shield, A2); no bond stud; everything else on the emergency bank on Pi B's side. The rules and why: Compute Bay Plan secs 5.2, 6 and 7. Same cabin, a few feet apart, is fine.
Compute Bay Plan sec 4.1: one 4 TB NVMe SSD and the Hailo-8L in the carrier's two M.2 slots, nothing on USB.
| Box | What is in it | Bus | Holes (all glands unless noted) |
|---|---|---|---|
| Central node (near the battery, busbars and compute boxes; switches loads, fires releases) | Sec 4.1, A3 | Control | about 14 in use + vent; spares 17, 20 and 25 not drilled (count at fit-out) |
| Release Node B (stern, near the transom boxes) | Sec 4.2, A15 | House bus end | 4 + vent |
| BMS box (as bought, bolted to the outside of the battery box's aft wall, never inside it; "the watchman box") | Sec 6.7, A4 | None (USB to the hub box) | 4 + vent |
| Aft junction (stern, passive, no Nucleo; plastic) | A6 | None | about 14 + vent; measure the box before the loom |
| Network switch box (aluminum, beside the aft junction box, glued to the hull for cooling, not wired to the lightning copper) | The TSW202 switch, its 54 V supply and the SFP module (A6a) | None (always-on 24 V feed) | about 10 + vent, 9 RJ45 couplers |
| Network surge board (not a box: a small board on a bulkhead near the deck, where the outside network cables first come inside) | The five Ethernet surge protectors and the small tinned bus bar for their grounds (A6a; Lightning Ground System Guide sec 7.4) | None | None: the cables plug into the protectors |
| USB hub box (small, beside Pi A) | The powered USB hub (fed from Pi A's 5 V), its 12 V converter, the dAISy, the science serial adapters and the wing's two readers (A1a) | None (two USB leads to Pi A) | About 10, listed in A1a; gland sizes at fit-out |
| Iridium A box, Iridium B box | A11 | None (serial to its Pi) | 3 each + vent |
| Detector mini-box | A12 | None (7-core lead to WD-B) | 1 + vent |
| VESC A box (VESC B box: BOAT 2 ONLY; Boat 1 sails with one VESC) | Bought sealed enclosure, with the extra bus capacitors (A7) | Control | 8 |
| Battery box | Battery Box Building Guide. No brain inside; no heater feeds, no gas-sensor leads (deleted, do not re-propose) (A9) | None | 3 glands + vent + relief stub |
| Transom terminal box (stern, passive, already on hand) | DIN strip only: the release trunks fan out to the drogue boxes and the cord pullers of the Jordan series drogue (JSD: a long line of small cloth cones streamed from the stern in a storm) (A16) | None | 6 + vent |
| Starlink enclosure | Deck; expendable, its Ethernet through a surge protector on the network surge board near the deck (A6a; Lightning Ground System Guide sec 7.4). Enclosure heater DELETED, do not re-propose | None | 2 |
| Wing box (small, below the deck where the wing's wires enter the hull; decided 2026-09-26, below deck Mike 2026-10-02) | The hull end of the slip ring: its eight wires land here, each through a fuse or surge part at its entry (A17) | None: the trim actuator keeps its own short CAN link, off both buses, to the CAN adapter in the USB hub box | At least 3 (A17); gland sizes open |
| Native devices, no drilling | em-trak (house, channel fails to ON); the Ewellix steering actuator, a plain DC device on the central node's drive, not a bus device (A14) | Tee + drop each |
Box count. Plastic: central node, Release Node B, transom terminal, BMS, aft junction, two Iridium, detector = 8 plastic boxes, plus three aluminum boxes (the two compute boxes and the network switch box), the small USB hub box, the PoE light driver's small dry box (sec 6.4) and the small wing box below the deck, where the wing's wires enter the hull (A17). The network surge board near the deck is a board, not a box (A6a). No FSESC box on either boat (A5). Every plastic box carries a membrane vent (Gore-type, M12 screw-in, 1/2 in hole, high on a box wall), so it breathes through the membrane instead of pumping moist air through its gland threads.
The circuit is the release channel steps (Compute Bay Plan sec 10.3) put onto screw-terminal boards. No soldering on the boat, no soldered field wires, and no push-on jumper pins anywhere in release logic. One-time bench soldering is allowed at the workbench, before burn-in and before anything is installed: a board may get its terminal blocks soldered on, or small solder links removed (Mike, 2026-09-27). After that every logic line between a Nucleo and a relay, MOSFET (a transistor switch) or opto board lands under a screw terminal or a latched connector. Each Nucleo is staked in its screw carrier after burn-in (a bench run that finds early failures).
The base:
Release channels (Compute Bay Plan sec 10.3 owns the channel steps). The release bus has its own feeder, 12/2 on a 30 A breaker at the bar (stud 4, sec 7.6), then the enable relay, then per channel a MOSFET switch module to the burn wire (channels 1 and 2) or the puller motor (channels 3 to 6). Each puller has a 1N5408 flyback diode (it soaks up the motor's kick at switch-off) under the terminal screws at its reversing relay's input side (MOSFET drain to release bus), not across the motor, so it works both ways; the burn wires need none.
Cord puller reversing relays, for shore tests (the test and the lock rule: Compute Bay Plan sec 10.3). Each of the four pullers runs through its own 24 V DC DPDT reversing relay in this box (two switches that move together; MY2NJ type, 8-pin, with LED, rated 10 A; Amazon B0GY3ZM4YB, $9.49 each, four), plugged into a screw-terminal DIN socket, nothing soldered; it swaps that motor's two wires. All four coils run from relay channel 11, "reverse", with no diodes across the coils (channel 11 is a relay contact, not a chip). No extra wires aft: the trunk core count (A3 entries 3, 4) does not grow. Bench checks: read the puller's stall current at its bench test against the relay's 10 A rating; read the coil + and - as printed on the relay (a "with LED" coil is polarity marked) before wiring the sockets; check each DIN socket's pin numbers against the relay's, since sockets differ by maker. Cord puller fire guard (decided 2026-09-26). One slow-blow 5 x 20 mm glass fuse per puller, in a screw-terminal DIN fuse holder on this box's rail, between the puller's MOSFET module and its reversing relay. Without it each puller's thin wires sit behind the release bus's 30 A breaker alone, and a short or a hot stall could heat them. Each pull runs a set time, with no stall sensing (System Settings Sheet sec 5). Fuse size, set at the one-puller bench test: read the current at a 30 lb pull, then take the next size up in the kit; if a 64 s stall does not blow it, the motor's stall current is under the fuse, and the 64 s limit decides. Bench test, with the fuse fitted (also Build Guide sec 11.1): pull the 30 lb load, then hold the motor stalled for the full 64 s; watch for smoke and feel the can; if it smokes, fit the next size down and test again. Parts: MECCANIXITY UK5 DIN fuse holders, 6 pack, Amazon B0DSW8RTTP (four used, the pack has end stops); QBGTFAK slow-blow 5 x 20 mm fuse kit, 0.5 to 10 A, Amazon B09685NBKH. Prices not checked. In-service exercise:
| Cadence | Test | What it proves |
|---|---|---|
| Monthly | Exercise the enable relay with release-bus voltage readback | Releases nothing, because every MOSFET stays off |
Monitoring. Two DS18B20 chains: chain 1 the six battery probes, chain 2 compartment air. A BME280 barometer breakout on the I2C chain (the box breathes through its membrane vent, so it reads outside pressure). A humidity breakout. Nothing else. Not fitted: branch current meters, MOV or SPD status contacts (every clamp is metered at each refit and at the post-strike check), a VE.Direct reader, a high-water sensor anywhere, gas sensors. Do not re-propose them. The cell-temperature trip on chain 1 is the battery trip. Its levels, rate rule and release: System Settings Sheet sec 4; the rule: the battery software notes (Reference folder) sec 4. Relay modules. Two 24 V 8-channel relay modules. The channel map is the Compute Bay Plan's (sec 11); this plan, the Box Wiring Sheets and the lid labels follow it. The second module is probably on the shelf already (Bill of Materials sec 15.8). Channels 9 and 10 each leave the box by their own gland (A3 entries 27 and 28). Channel 13 is the horn (Mike, 2026-10-08), its coil driven from the Nucleo's native pin A2: the contact feeds the horn's 24 V through a 5 A inline fuse in this box, out by its own gland (A3 entry 30) on about 15 ft of 16 AWG pair to the horn low on the instrument post (Instrument Post Guide sec 3). No photocell relay; the TSW202 is on no channel (sec 6.1). Inputs. The two Water Witch states, the post wind sensor and rudder encoder A (PWM, sec 5.6) come in from the aft junction on one 12-core trunk (one is likely enough; count at fit-out). The two bilge manual-on feeds (channel 6 for pump 1, channel 12 for pump 2) ride either that trunk or the 16/2 switched-load runs (sec 7.2): settle it at loom time. Each joins its pump's + terminal, after the Water Witch, with no diode (Mike, 2026-09-28): the Water Witch runs the pump with the node dead, and shore can run either pump for a set time. Post wind sensor. The Calypso ULP CMI1018 (UART/I2C), wired straight to a Nucleo UART, no level chip; it draws 0.25 mA, so the cable run costs nothing. Its Pololu D36V6F5 regulator (5 V out, 50 V max in) sits in the aft junction box just after the wind sensor's fuse (A6); put its terminal block on at a workbench, off the boat, then pot it. Never 24 V into either Calypso meter: both take 3.3 to 18 V and their inputs carry clamping diodes, so 24 V destroys them. 5 V, not 12, because the meter's serial data swings to its supply rail. The CMI1017 RS-485 unit at the wing tip stays as the boat's backup wind meter; its wires come down through the slip ring to the wing box (A17).
A second, physically separate release chain: its own 12/2 feeder on a 30 A breaker at the bar (stud 2, sec 7.6), own enable relay, own MOSFET pairs, the same two-conditions interlock, commanded over the house bus. Each drogue box's two release channels split across the nodes, one fired from the central node and one from Release Node B, through separate glands on separate routes. Either channel alone cuts the strap. One dead box cannot remove every release on the boat. Release Node B's live channels are exactly the two drogue B-horns, on boxes 1 and 2. Five Nucleo pins. Inside, and every hole: A15.
Node addresses, message IDs, release channels, relay channels, probe serials, I2C addresses and opto inputs are printed on each lid, and recorded on the System Settings Sheet sec 9. Node Boxes Explained has the drawings. This table is the final word on the release-channel map. Node Boxes Explained, the Box Wiring Sheets, the Compute Bay Plan, lid labels and firmware message names all follow it. Release channels are not the relay module's channels (sec 4.1). Command words: the two-step is ENABLE then RELEASE; status wording is "channel N released".
| Channel | Action | Central node | Release Node B |
|---|---|---|---|
| 1 | Drogue box 1 lid strap | Horn A (3 coils) | Horn B (3 coils) |
| 2 | Drogue box 2 lid strap | Horn A (3 coils) | Horn B (3 coils) |
| 3-6 | The four JSD cord pullers on the T8 trigger cords: BRINGSMART 24 V worm-gear motors with pulleys, bought 2026-09-21; each through its own reversing relay and fuse in the central node (sec 4.1) | All four | - |
| 7-9 | RESERVED, BOAT 2 ONLY: the aux-rudder release (solenoid 1, solenoid 2, burn wire). Unwired on Boat 1 | (Boat 2) | - |
| 10-15 | RETIRED on Boat 1 (the six steering pull-pin paths; Boat 1 has no pins). BOAT 2 ONLY: Boat 2 keeps one pull-pin, on its primary actuator, so a dead actuator can let the rudder trail for the aux rudder; its release paths (two from the central node, one from Release Node B: Steering System Handbook sec 5) get channels when Boat 2 is wired | - | - |
| 16 | RETIRED (the trim-linkage decouple pulse; every wing-trim failure already ends in feather by passive means, so no channel and no gland). Number not reused | - | - |
NMEA 2000 Micro-C (M12 A-coded 5-pin), 120 ohm: a twisted shielded data pair plus a 22 AWG power pair in one jacket. Pre-molded cordsets only, cut once for drops; no field-made connectors.
One straight backbone per bus, stubs under about 3 ft. Never a star, never a ring. CONTROL BUS: power tee [END] -- [4-port: Pi A, Pi B] -- VESC A -- central node [END] HOUSE BUS: power tee [END] -- [4-port: Pi A, Pi B] -- em-trak tee -- Release Node B tee [END] (VESC B drop and the Electrak tee: Boat 2 only)
12 V, fed in once per bus: a 5 A breaker at the main bus board, 16/2 at 24 V to a sealed 24 to 12 V 5 A (60 W) converter at the compute multiport, and 12 V a couple of feet on to a power tee beside the 4-port tee (Mike, 2026-09-23: 12 V then never wanders across the boat). One converter per bus keeps the control and house split. 12 V is inside the NMEA 2000 window of 9 to 16 V, so any ordinary NMEA 2000 device can be plugged in later; 24 V would kill nearly all of them. Never feed bus power from inside a Pi box, so a Pi A fault cannot darken a bus; hungry loads have their own feeds. Sizing: about 12 W needed, buy 60 W. The pair feeds only the CAN transceivers and the isolated side of the two Pi CAN HATs, roughly 50 to 100 mA each, on four control drops and five house drops (sec 5.4): about 0.5 A at 12 V per bus. The nodes and the VESC make their own 5 V from their own 24 V feeds. The standard caps a power segment at 3 A, 36 W at 12 V. Buy the potted IP67 kind with the lower no-load draw: two converters at even 1 W idle cost about 48 Wh a day, more than the whole nav-light budget. About 2 A over ten feet of 16 AWG loses a tenth of a volt. Nothing else hangs off the pair (22 AWG, 3 A per segment): no nav lights or wet-lab instruments. Check at fit-out that the em-trak takes its own feed on relay channel 4 and uses the drop for data only (if it draws from the pair, the house bus peaks near 2 A, still inside 60 W), and that nothing on the pair wants 24 V (the em-trak takes 9.6 to 31.2 V). Micro-C parts are 60 V rated. Both buses run at the one bit rate (System Settings Sheet sec 9). The TSW202 is not fed from the pair.
| Device | Bus | Why |
|---|---|---|
| VESC A, central node (which drives the Ewellix steering actuator) | Control | Steering, release channels, motor. Nothing non-essential goes on it |
| em-trak, Release Node B (bus end) | House | So no single bus failure strands both release nodes (VESC B is Boat 2 only) |
| Viatrax Primelink A | House, a listen-only tap | A separate monitoring and black-box path (Build Guide sec 8.3) |
| Pi A, Pi B | Both | Only the Pis touch both |
| The wing's trim actuator (Thomson Electrak MD) | Neither: its own short CAN link through the slip ring (A17), which Pi A reads through a USB to CAN adapter in the USB hub box (A1a) | Kept off the boat's two buses (decided 2026-09-26); both channels of each Pi's CAN HAT carry those buses |
| The battery watchman | None (sec 6.7) | It controls nothing; its stream reaches Pi A over USB |
The rules. Each device has one CAN port and lives on exactly one bus. Only the Pis touch both, and cross-bus information moves through Pi software, so a babbling device, likeliest on the house side, cannot delay steering or throttle traffic. BOAT 2 ONLY: the house bus also carries the backup VESC with its phase-transfer contactor, and the Electrak steering actuator (J1939) rides the control bus, its own overload cut-out the backstop. Island rule. No emergency-bank device on either bus except through an isolated transceiver, which is Pi B's HAT.
NMEA 2000 (em-trak), VESC's own frames and ours from the two nodes share the wire fine; the Pi's software handles each. No J1939 (a truck and machinery CAN protocol) on the hull buses. The only J1939 aboard is the wing's trim actuator, the Thomson Electrak MD, on its own short CAN link through the slip ring (A17). BOAT 2 ONLY: the Thomson Electrak HD adds J1939 to the control bus, so the VESC takes a controller ID clear of any J1939 address (System Settings Sheet sec 9), and the two boats share one software build.
The primary rudder. The steering actuator is the Ewellix CAHB-22-F4E-3000529-BAA0PT-000, bought 2026-09-10: a plain 24 V DC actuator, 2,250 lbf (10,000 N), 11.8 in stroke, 0.4 to 0.5 in/s, a self-locking screw (it holds position unpowered, no brake), built-in position feedback, no CAN. Full data: Steering System Handbook; settings: System Settings Sheet sec 3. How Pi A and the node command it, the angle cross-check and the jam logic: Compute Bay Plan sec 10.2. The drive (decided and ordered 2026-09-27): one Pololu G2 High-Power Motor Driver 24v13 (Pololu item 2992), Amazon B07LH77MHR, , $79.99, in the central node. It takes 6.5 to 40 V (the bank peaks near 29 V) and runs 13 A with no heat sink, against the Ewellix's 8.5 A working and 10.9 A at clutch trip; moves are short and at most 20 percent of the time. It takes 3.3 V logic straight from the Nucleo and has its own current-sense output, so there is no separate current sensor. The bigger 24v21 (item 2995, same size and pins) also fits if it turns up. Not the Dual G2 18v22 shield: it tops out at 30 V, only 1 V over the charging peak, and it is two drivers on pin headers. Not the two 30 A relay modules as a reversing pair: the voyage plans about 190,000 rudder moves, and relay contacts switching a motor typically last around 100,000 (inferred from typical relay data). Fallback if the Pololu is unavailable: a Cytron MD20A plus an ACS712 30 A current board (the MD20A tops out at 30 V, only 1 V over the charging peak). Bench work, off the boat, before burn-in: solder on the two 5 mm power terminal blocks that come with it, and one 8-way 0.1 in screw terminal block on the logic row (Hxchen 8-pin, 20 pack, , on hand). After that every wire lands under a screw. The steering feed is 14/2 on its 20 A breaker (sec 7.6); if it will not fit the 5 mm blocks, land it on DIN terminals and run short 14 AWG jumpers. Wiring (A3 entries 12 and 13): 24 V from the steering breaker (entry 12) to a DIN terminal, through the power-cut relay's closed contact, to G2 VIN; G2 GND to the node's 24 V negative on its own 14 AWG return. G2 OUTA and OUTB to the Ewellix red and black (entry 13). Logic: PWM to D9, CS to A0, GND to Nucleo GND, DIR to D13, SLP tied to 3.3 V (always awake), FLT to an expander input. Ewellix feedback: brown to 3.3 V, white to GND, green (the wiper) through a 1k resistor (from the 1/4 W kit on hand) under DIN terminal screws to A1; no capacitor, the firmware averages the readings. Read the pin names off the board. Steering power-cut relay. One of the two 24 V 30 A relay modules on hand (bought for the relay-pair drive, now free; the other is left over), in series with the Ewellix feed ahead of the drive, wired de-energized-closed, its coil on an expander output. A dead node, a dead driver or a broken wire leaves the actuator fed. Cutting a jammed or runaway actuator takes a live node; a failed drive usually fails stuck on, and this relay is then the only stop. Rudder angle is read two ways: encoder A on the stock (an AS5048A magnetic encoder; the node reads its PWM output, a pulse-width signal, on one timer pin) and the actuator's own position feedback. The cross-check and the jam logic: Compute Bay Plan sec 10.2. Boat 1 steering has no backup, accepted on purpose. Bench test (Steering System Handbook sec 8 step 5): full stroke both ways logging current and wiper; chock the rod and see the stop within about 1 s; pull the Nucleo's power mid-move and see the rod stop; an hour of cycling at 20 percent duty, then feel the board. Rod-travel life: Ewellix publishes none (Reference/Ewellix_CAHB_Series_Catalogue.pdf), so rod travel sets no move budget (the voyage's rod travel, and why no travel limit is quoted: System Settings Sheet sec 3). The deadband (the small rudder error the boat ignores) is a software number (System Settings Sheet sec 3); no halved-move plan.
| Camera | Where | What it watches | Status |
|---|---|---|---|
| Foredeck fisheye | Glued-on G10 pad on the foredeck | Up and forward at the wing, rig and foredeck | A 360 degree outdoor PoE fisheye (Mike, 2026-09-28): Hanwha XNF-8010RV, bought 2026-09-30 (open box, $132); wired PoE on an RJ-45 plug from the TSW202 (Bill of Materials sec 15.8) |
| Forward and aft post cameras | High on the post, on glued-on 1/4 in G10 pads tabbed with two layers of biax (fiberglass cloth); nothing through the tube wall | Forward: ships and ice. Aft: the post base and boot, transom, drogue bridle, rudder head | The two Reolink RLC-520A domes, on hand |
| Interior camera | Inside the cabin | The compute boxes, battery box and bilge | The first Reolink FE-P (an indoor unit), on hand (Bill of Materials sec 15.8). Indoor-rated: never on deck |
| Underwater camera | Angled epoxy-glass through-hull tube low on the hull | The prop hub and rudder | Axis P1245 Mk II modular sensor head, received |
One Cat6 each (outdoor Cat6 on deck); the two post cameras, the foredeck fisheye and the interior camera each pass an Ethernet surge protector on the network surge board near the deck (A6a; Lightning Ground System Guide sec 7.4). IR LEDs off (System Settings Sheet sec 5), no trigger wires. Cameras run on events plus a schedule, not continuously; per-port PoE control through the switch saves power, with no relay channel. Below about 50 deg S the forward post camera and the thermal camera run together into the AI ice path. Cape Horn version only: a forward thermal camera, an uncooled LWIR FLIR Lepton 3.5, for ice detection day and night, AI-fused on the Hailo-8L (Build Guide sec 3.9); not fitted on the Atlantic version. The underwater camera also trends fouling and watches the shaft collar anode. It looks through a JGS2 fused-quartz window kept clear by 275 nm UV-C light; the head swaps afloat through the inboard cap, and its dry main unit inside takes the PoE. It is ISOLATED, never wired to the lightning copper: plain UTP Cat6, shield broken at the camera end, off the anode circuit. Build: Underwater Camera Port Guide. Camera port UV-C lights. The two Seoul star boards do not fit beside the camera head in the 1-1/4 in bore and are not bought. On trial (three ordered 2026-09-28 on eBay, ships from China, due Oct 19 to 23): small 275 nm UV-C LEDs on round aluminum boards 0.39 in (10 mm) across, fitted above and below the lens on a flat mask in front of the camera face. Check each is real UV-C with the QuantaDose UV-C test card before building. Each board gets two wires, soldered once at the bench. They run from an LM317 current regulator (on hand), starting at 20 mA with a 62 ohm set resistor, and the regulator is fed through the normally closed contact of central node relay channel 9 (A3 entry 27; the on and off times: System Settings Sheet sec 5). The mask, the mount and the build: Underwater Camera Port Guide. Open: whether to have a factory-made board built for the same layout, carrying four name-brand UV-C LEDs soldered by the maker; Mike's go settles it. One network switch on the boat: the Teltonika TSW202 at the aft junction. Eight PoE ports, two SFP slots. Not fitted: the Netgear GS308EP and GS305EP, and any switch inside the Pi A box. The switch, its 54 V supply and the SFP module sit in their own aluminum box beside the plastic aft junction box (A6a), and every network cable ends there. The five Ethernet surge protectors are not in that box: they are on a small board on a bulkhead near the deck, where the outside cables first come inside (A6a). That box is one of the four glued to the hull for cooling, glued straight on like the Pi A box (Compute Bay Plan secs 2.3 and 4.4); inside it the switch and its supply clip to a DIN rail on a side wall (A6a).
| Port | Device | Power |
|---|---|---|
| 1 to 5 | The five cameras | PoE |
| 6 | The PoE light driver, which runs both interior disc lights (sec 6.4) | PoE |
| 7 | Pi A | data only |
| 8 | Pi B | data only |
| SFP 9 | Starlink, on a copper SFP module (10GTEK SFP-T, two bought) | data only |
| SFP 10 | Free |
The switch is always on, from the house branch, never on a relay channel (shedding it would kill the cameras, Starlink's data path and the Pi-to-Pi Ethernet at once). Always-on costs about 72 Wh a day. Cameras save power port by port. Pi A switches a port through Teltonika's documented Web API, then reads it back over SNMP, so command and check take different paths (the request, the MIB and the port settings: System Settings Sheet sec 5). Pin the firmware version: Teltonika renamed the port setting between firmware 1.01 and 1.05, so code for one silently fails on the other; test on the exact firmware that flies and record the working request. PoE budget. The Mean Well DDR-120B-48 supply stays (no upgrade): about 112 W available, cameras about 60 W set aside, plus the light driver at 25 W in 802.3at or 12 W in 802.3af, leaving 22 W or 36 W to spare. The 24 V feed current stays about 5.5 A. Bench checks before the loom (Build Guide sec 11.1): 1) turn a port off and on through the Web API, recording the exact request and firmware version; 2) read the same port back over SNMP; 3) read watts per port, recording the command and units; 4) find whether a per-port PoE power limit or priority exists; 5) power all seven PoE devices at once and read the total. Hardening rules, all pass/fail:
Post mounting: Instrument Post Guide (nothing through the tube wall, one bond spur, a short wire to the lightning copper, per metal body; cables in their own glands with drip loops). IR LEDs off in settings.
Two separate receive chains, each on its own antenna, so no single antenna fault deafens both. A third, the RTL-SDR listener and its whip, is DELETED, do not re-propose.
The transmit whip lives under the panel-free deck patch, stood as near upright as headroom allows. No AIS antenna above deck.
Ethernet to the TSW202 at the aft junction (SFP port 9, sec 6.1), power on relay channel 1. Expendable, but its Ethernet passes an Ethernet surge protector on the network surge board near the deck (A6a; Lightning Ground System Guide sec 7.4). No enclosure heater (deleted, do not re-propose: the dish has its own snow-melt heater). Deck hole: the instrument post's 1 in PVC deck stub. The GPS and Iridium A coaxes and Starlink's Cat6 and power pair come below through it with the post bundle (A13; Instrument Post Guide sec 8). Not fitted: four separate 1 in standpipes and their compression deck seals. Iridium B keeps its own deck hole (sec 6.5).
Nav lights: one red/green bi-color lantern on the centerline at the bow and a white stern light on the post, on relay channel 3, lit from computed sunset to sunrise (GPS position and the date), wired through the normally closed contact so they fail ON. The two all-round whites on the post come on only when it is dark (the same computed rule) and the motor is running (channel 7, through a 30 A relay in the aft junction). Fog mode (Mike, 2026-10-08): when fog is likely, or when shore sends the signed command, channel 3 lights the bow and stern lights in daytime too, and the whites come on whenever the motor runs. The relay logic: Compute Bay Plan sec 11; the fog thresholds: System Settings Sheet. Channel 3 carries the light current directly: these lamps are small and need no 30 A relay, unlike the motoring whites. How channel 3 fails toward light, and why a dead Pi never leaves the boat dark: Compute Bay Plan sec 11. Not fitted: a hardware photocell (it drew power around the clock). DELETED: current sense on the light circuit and a Pi backup channel (the cameras show the lights). No converter anywhere in the light circuit. Every fixture takes the bus voltage as it is: the TRUE MODS bow lantern (12 to 24 V), the TRUE MODS stern light (9 to 33 V) and the two TRUE MODS all-round whites (12 to 24 V), so the fail-ON wiring means what it says. The two attwood bi-colors are 12 V only and come off the boat. Draw figures: Instrument Post Guide. TRUE MODS sealed LED; no Hella (do not re-propose). All five TRUE MODS lights are on hand; the layout uses four (the bow lantern, the stern light and the two all-round whites), so one is an extra, and nothing more is bought. Their USCG certification, ranges and heights on the post: Instrument Post Guide. Check that each light's body carries "USCG" and its distance; the stern light must be marked for 2 miles. Budget about 30 Wh per night. Interior lights. The PoE Texas AT-LED-24V kit: one PoE-powered driver and two 4 in LED disc lights, on TSW202 port 6. The driver is a bare board, so it lives in a small dry box of its own; its Cat6 comes in on one bulkhead jack, taped like every other (A6a). It has no current limiter: do not add a third light. The discs are dry-location parts; mount them where no condensate collects and no drip can land in them. They light the interior camera for colour, which infrared cannot give.
Modem A near Pi A (main bank, always on); modem B near Pi B (emergency bank, always on). Box and holes: A11. Each coax runs straight to its quarter-wave stub arrestor (a short coax stub that shorts a lightning surge to ground but passes the signal). The stub is the only coax protection, because a stub cannot wear out. Antenna domes (Instrument Post Guide): one clear acrylic dome per antenna on a white base plate, rubber gasket, oversize holes and snug screws, a 5/64 in (2 mm) weep hole at the flange low point. A sits aft beside the post, its coax down through the post's deck stub (sec 6.3); B sits on the foredeck about 18 ft forward with its own small deck hole, so the two chains share no dome, deck hole, cable or fuse. Each dome sits where no solar panel shades it and it shades no panel.
Antennas: each receiver has its own, on separate spots: one on top of the pylon, one on a deck pad near the post, clear of the post's boot (Instrument Post Guide). The four AEDIKO waterproof active GPS antennas bought go with the GPS boards and suit the deck pad (Mike, 2026-09-26); the NEO-7M pack's own four antennas become shop stock (Claude's reading). No arrestor and no bias tee on either GPS coax (the rule and why: Compute Bay Plan sec 14.1). The protection is the two separate chains (different receivers, antennas and boxes), the ferrite at the wall, the low cost of the receiver, and Starlink's own position fix as a cross-check.
Two boards in the BMS box as bought (A4), bolted to the outside of the battery box's aft wall, neither of them inside the battery box: the Electrodacus SBMS0, the watchman, and the JKBMS JK-B2A8S balancer. They watch and control nothing: no load, no charger. Their build, the taps, the probes, the settings and the charge-stop wiring: Battery Box Building Guide Part 8; the settings: System Settings Sheet sec 1. Its USB stream to Pi A is required: the battery percentage comes from it, because the SBMS0's pack shunt is the only battery current sensor aboard. Both of its shunts sit on the positive side, on the main bus board (sec 7.6). The base board has no data port, so the stream needs the Electrodacus WiFi/USB add-on board (ordered). Its USB lead goes to the USB hub box (A1a) and reaches Pi A on the hub's own lead (A4 entry 4, A1 entry S). No CAN, no RS-485 from the BMS. The serial reader, the stream-gone-quiet alert and "no charge for a solar day" are required software (the battery software notes, Reference folder): a quiet stream loses the battery percentage. The DS18B20 chain comes into the watchman box in the tap bundle and leaves by its own PG7 gland to the central node, where the cell-temperature trip lives (sec 4.1, A4). Charging is on by default; only a shore command stops it, through relay channel 8 in the central node. Every failure leaves the boat charging. The relay's place in the channel map: Compute Bay Plan sec 11. The wiring and its bench test: Battery Box Building Guide Part 8. The deadman, the Pi-silence timer and the resistor: System Settings Sheet sec 1. The thin battery positive enters the central node by A3 entry 22. The five remote on/off cables meet on the main bus board, tied together, and one run goes to relay channel 8 (decided 2026-09-26). Cold charging. No automatic stop. Pi A alerts shore when a pack probe reads cold, and shore can then send the stop (System Settings Sheet secs 1 and 4; Battery Box Building Guide, "The cold, and no heaters"). Heaters: not fitted, do not re-propose. None on the battery box, and none on the wing, which has no packs since the slip ring.
The science instruments are in the Science Payload Strategy; the IMUs and the working barometer below also feed science. Where each sensor's wiring and settings live is named in its row.
| Sensor | Part | Where, and how it is read |
|---|---|---|
| IMU x3 | Bosch BNO085 (three bought 2026-09-08; all three fly) | IMU A1 and IMU A2 in the Pi A box, on the same I2C pair (a short chip-to-chip data link), A2 at the chip's second address, one at each end of the plate, as far from the 5 V converter as space allows; IMU B in the Pi B box (Compute Bay Plan secs 4.1 and 4.2). Each has a forward arrow on its plate, a heading offset and roll and pitch level offsets (System Settings Sheet). If A1 and A2 disagree past a set limit (System Settings Sheet), Pi A falls back to the GPS track and gyro, flags it and sends an alarm home. Heading, fused with the GPS track; tilt correction for the post wind sensor; yaw rate for the steering loop and the rudder-fault detector; heel, knockdown and inversion detection, the growler impact signature (Build Guide sec 3.9A) and STORM-entry triggers; wave spectra |
| GPS x2 | u-blox NEO-7M, one per island (hiBCTR modules, bought; the other two of the 4-pack are Boat 2 stock) | Separate antennas (sec 6.6); position only, accurate to a few yards. Heading comes from the IMUs and the track. The Starlink dish's own GNSS (over its local gRPC API) is logged as a cross-check. Iridium gives only a rough Doppler position |
| Wind, post (PRIMARY) | Calypso ULP CMI1018 ultrasonic (UART/I2C), bought 2026-09-10 | On the post's aft arm, level (Instrument Post Guide), wired straight to the central node's UART, 5 V from its own regulator (sec 4.1). Tilt-corrected off Pi A's IMUs. Not fitted: the Airmar 200WX ($1,700); do not re-propose |
| Wind, wing tip (cross-check) | Calypso ULP CMI1017 ultrasonic (RS-485, NMEA 0183), bought 2026-09-10 | On a short arm ahead of the wing-tip nose, upwind of the tip rod (Wing Construction Guide). Its RS-485 pair comes down through the slip ring to the wing box (A17) and on to the Waveshare 2-channel RS-485 reader in the USB hub box (A1a); its 5 V is made at the tip (Wing Power System sec 1A); 24 V destroys it. Turned into the boat frame through the wing angle from the wing compass. Sacrificial in a strike. It stays as the backup wind meter if the post is lost (Mike, 2026-09-26) |
| Wing compass (wing angle and turn rate) | WitMotion HWT905 compass and gyro, RS-485 version, IP67, bought 2026-10-02, on a G10 pad on the wing's tip cap (Wing Power System sec 1A) | Its gyro's turn rate feeds the active rate damper (Build Guide sec 6.4); its heading minus the hull's heading gives the wing angle, which turns the wing-tip wind into the boat frame. Its own RS-485 pair comes down through the slip ring to the wing box (A17) and on to the same Waveshare reader as the tip meter, on its own channel (A1a). It runs on 24 V (9 to 36 V, under 40 mA; it does not run on 5 V), from the tip branch's 24 V on its 1 A fuse, ahead of the tip's 5 V converter |
| Barometer | BME280 breakout on the central node's I2C chain (zero Nucleo pins) | The working barometer, read over CAN from the node. The node box breathes through its membrane vent, so it reads outside pressure. No weather pod and no AK09918 compass on the post. The science barometer is separate (Barometer and Pressure Port) |
| Cameras x5 | Sec 6.1 | Sec 6.1 |
| Forward thermal camera | Uncooled LWIR (FLIR Lepton 3.5) | Cape Horn version only (sec 6.1) |
| Battery temperature x6 | DS18B20, 1-wire chain 1 | Six pack probes (none at the motor jacket, bearing or shaft). The cell-temperature trip runs on this chain (sec 4.1; its levels: System Settings Sheet sec 4) |
| Combustible gas | DELETED, do not re-propose | Both sensors, their leads, the node input and the gas-alarm switching freeze are struck. The cell-temperature trip is the battery trip |
| Rudder stock encoder | Magnetic absolute (AS5048A) | True blade angle. Encoder A rides the sensor trunk to the central node and is cross-checked against the Ewellix's own feedback (sec 5.6). A permanent twist in the stock sits below the encoder and shows only as a lasting helm bias; the rudder camera watches for that (Build Guide sec 3.8) |
| Steering telemetry | At the central node | The Ewellix position feedback, the drive's own current sense (sec 5.6), the power-cut relay state and encoder A agreement; no J1939. For jam detection and screw-wear trending. Boat 2: the Electrak's own position and overload reports over J1939 |
| Shaft rpm | DELETED | A slipped coupling shows as motor rpm with no boat speed and no motor current, both already reported by the VESC |
| Shaft seal leak | DELETED | The bilge pump sensors cover it |
| Battery watchman | Electrodacus SBMS0 (USB text stream, required) | Sec 6.7. It reports 8 group voltages, min/max and spread, pack current, total solar current, SoC (state of charge, the battery percentage the load-shed schedule uses, from the pack shunt), pack temperature and a status word, sent to shore a few times an hour. A quiet stream is an alert: shedding falls back to a fixed cautious schedule |
| Bilge | Water Witch 230-24 x2, one per pump | Their 24 V pump outputs read as "pump running" through the central node's opto board (sec 4.1). No separate high-water sensor (do not re-propose), no capacitive level sensor, shaft-seal probe or pump current meters. Staged switching is hardwired; pump state, duty cycle and rate-of-rise go to shore (Build Guide sec 3.6) |
Every box entry is a nylon IP68 gland (IP68: dust-tight, good under water), as bought. No brass, no nickel-brass.
| Gland | For | Drill |
|---|---|---|
| PG7 | One cable of 1/8 to 1/4 in | 1/2 in |
| PG9 | Small bundles | 5/8 in |
| PG11 | Actuator bundles | 3/4 in |
| PG13.5 | Paired harness entries; the switched-load duplex runs (sec 7.2); the battery tap bundle unless it outgrows it (A4 entry 1) | 13/16 in |
| PG16 | 2 AWG (the VESC feed and the battery mains) | 7/8 in |
Nothing needs to unplug at a box, so no M12 receptacles anywhere, and a gland seals better. The only M12 joints are the tees on the CAN backbones and the molded couplings mid-cable on point-to-point runs.
Two small same-route cables may share one gland a size up, gaps between the jackets filled with sealant. Fast cure is fine; a filled gland cures before the salt-spray test. Grouping is allowed only for cables on the same route to the same box, serviced together. Never for release circuits, mains, or cables arriving from different directions. The transom terminal box's two puller glands (A16 entries 5 and 6) each take one 4-core cable carrying two pullers' power pairs: one cable per gland, so they are not grouped entries. Current grouped entries: the central node's 1-wire chains (A3 entry 18), the BMS box's tap bundle (A4 entry 1), the remote on/off bundle (A3 entry 22) and the switched-load runs (A3 entry 16, A6 entry 10).
The M12 disconnect is on the cable, outside the box: at a tee for CAN, at the molded male-female coupling of a cut cable for point-to-point runs. Female on the box side of the joint, male on the cable, boat wide.
A ferrite is a clip-on magnetic collar that soaks up high-frequency surge current riding on a cable. Genuine mix 31 only: Fair-Rite 0431 series or Palomar 31-xxx; no-name kits are the wrong material for lightning, which lives below 30 MHz. Only on cables that leave the cabin (Mike, 2026-10-14). A long wire up the post, across the deck or down from the wing picks up the surge from a nearby strike; a short cable between two boxes inside the cabin picks up almost nothing, and a direct strike goes down the lightning copper, where no clip-on core matters. A cable with its own surge protector (the cameras, Starlink's Ethernet, the solar strings) needs none. One ferrite on each cable below, outside the box it enters, just below the entry, sized to the cable. About 11 boat wide:
| Cable | Box | Core ID |
|---|---|---|
| GPS A coax | Pi A (A1 entry 11) | 0.30 in |
| GPS B coax | Pi B (A2 entry 9) | 0.30 in |
| Iridium A and B antenna coax, one each | each Iridium box (A11) | 0.30 in |
| Post wind meter cable, post lights cable | aft junction (A6 entries 3 and 4) | 0.30 in |
| Horn pair up the post | central node (A3 entry 30) | 0.30 in |
| Bow lantern feed | aft junction (A6 entry 19) | 0.30 in |
| Each hull cable from the slip ring (how many: the cable choice, Wire and Cable Buy List C4) | wing box (A17) | 0.30 or 0.5 in |
None on any other cable: not on the motor phase leads (too short for the clamp, Mike, 2026-10-15), not at the BMS box, the battery box, the USB hub box, the release nodes, the transom box or on the runs between the cabin boxes. On hand: the order log shows 15, 4 and 1 of three Amazon ferrite listings (August), at least 20 cores; none to buy. The 1.0 in phase clamp bought 2026-09-13 is not used. A ferrite must never hang on a cable: one swinging below a gland wears the cable at the gland lip. Glue the core to the bulkhead or settee structure just below the box, with 3M 4200 on the sanded shell, cable through it with a little slack each side (this and the instrument post cusp caps are the only two places 4200 is used on the boat; sec 8, Sealant); or lay a small cable's two-pass loop against the structure; or, where a box has room, tie a small core down inside on the plate. Keep ferrites off the hull skin and away from the lightning straps. A cable clamp on a glued-on pad a few inches below each box supports the drip loop.
This section is the home for the bus, its fuses and breakers, and every branch. The surge rules: Compute Bay Plan sec 14. Peak currents. The motor at 2,000 W and 25.6 V: about 78 A sustained, and about 180 A for a bounded CUT-BURST (Build Guide sec 5.3). Combined solar: about 50 A new (1,400 W at 27.6 V), about 35 A at the mid-mission planning output. The 300 A Class T at the bank is the hard ceiling. The main fuse is the 300 A Class T from the JIAKUPA kit ($22, bought), bolted to the outside of the battery box's aft wall at mid-height on the + side, not on the main bus board; the + main lands on it from its gland, and a sleeved 2 AWG run goes on to the Blue Sea 6006 battery switch on the main bus board, so the whole run is protected. The fuse stays in series through every override; a CUT-BURST at about 180 A does not trouble it. Not fitted: a Blue Sea 5119 (do not re-propose). Mounting steps, and the care if its bolts go through the wall: Battery Box Building Guide Part 8. Not yet measured: room on the wall for the 4.8 x 2.0 in fuse block, its cover and the cable bend. Three rules for the mains at the fuse:
Clamp both mains a few inches from the fuse block so they do not tug on it; torque the fuse stud nuts and paint a witness stripe across each. The motor branch takes a 250 A MEGA fuse under the main fuse, at the busbar. The main bus is two generic 400 A busbars with four 3/8 in studs each (bought); each main branch leaves through a MEGA fuse in its own holder (four holders bought). The main bus board sits on the hull side beside the battery box (Mike, 2026-09-27). No Victron Lynx anywhere (deleted, do not re-propose). No main-BMS bypass contactor: the bank runs through no BMS gate, so there is nothing to bypass; the Class T stays the short-circuit protection. The only mechanical switch aboard is the manual battery disconnect (the Blue Sea 6006) for haul-out servicing. The watchman's two shunts sit on the positive side. The SBMS0 reads a shunt only with 4 to 76 V on it, so on the negative side it would read nothing (its manual, Reference/SBMS0_v03d_Manual.pdf). The 300 A battery shunt goes between the Class T fuse's output and the + 400 A bar; the 100 A solar shunt between the + solar bar and the + 400 A bar. The - side is plain cable. Each sense wire sits at battery voltage, so each gets a 30 AWG fuse link at the shunt end, like the battery taps, and each shunt keeps a cover. The sense leads: A4 entry 3; their wiring: Battery Box Building Guide Part 8. Fuses and breakers, not a breaker panel (a panel is dead weight with contacts to corrode). Every small branch that can be an auto-reset breaker is one: 24 V stud-type auto-reset breakers (they trip on heat and reset as they cool; bought 2026-09-07), landed on 150 A bar pairs, so a short glitch cannot leave a load dead for the mission: the CAN power tees (5 A), both bilge pumps (5 A), the box feeds (Pi A 10 A, the central node 15 A, the aft junction 20 A), and since 2026-09-28 every other line off the branch bar too (the stud table below). Blade-fuse blocks and most inline holders are out. Bench trip-test every breaker at about twice its rating, and hold rated current for minutes without tripping. Mount high, terminals down under the covers, drip loops. Breakers, not fuses, on the releases, Iridium A and steering too (Mike, 2026-09-28). A fuse that blows on a passing fault leaves that function dead for the mission; a breaker comes back. Still fused: the five solar outputs and the two bus surge modules (reasons below), the watchdog rails, the cord pullers (one slow-blow fuse each, Transom Drogue Box Guide sec 6), the post feeds (one fuse per feed at the aft junction, A6; Instrument Post Guide), the wing box's own 5 A and the tip branch's 1 A (A17), and the emergency bank's DIN strip (A10). Not fused, on purpose: the charging stop's thin positive wire from the main bus board's positive bar to the central node, limited by the 1k resistor at the bar (sec 6.7). Fuse sizing rule (no crew to replace a blown fuse). Size every fuse below its wire's current rating and at the next standard size above the true worst-case continuous load plus margin, not the minimum, with time-delay types where there is inrush. Never oversize past the wire rating: a fuse that will not protect the wire is worse than none. Critical always-on functions (Iridium, the watchdogs, the steering actuator feed) get parallel-fused feeds from different buses, so one fuse clearing does not kill the function. Which lines share a bar stud. Fifteen lines leave the + 150 A branch bar, each through its own breaker (Mike, 2026-09-28). The JUXILPP bar has eight small studs (counted on the part) plus a second big stud at the end opposite the feed. Breakers are numbered 1 to 15 top to bottom as they sit on the board (drawing Reference/Main_Bus_Board_Drawing_2026-10-06). The four heavy or life-critical lines get a stud each; the others pair up, and three share the big end stud. The landings are chosen so no two jumpers cross (Mike, 2026-09-29). Never put a backup pair on the same stud. Wire sizes are for a 3 percent drop at 24 V; "good to" is the longest one-way run at that size.
| No. | Line | Stud | Breaker | Wire | Good to |
|---|---|---|---|---|---|
| 1 | CAN control bus (its converter) | 1 | 5 A | 16/2 | 90 ft |
| 2 | Release Node B (stern) | 2, alone | 30 A | 12/2 | 12 ft (4 percent at 25 A, fine for burn wires) |
| 3 | Central node | 3 | 15 A | 16/2 | 18 ft |
| 4 | Release bus A (central node) | 4, alone | 30 A | 12/2 | 9 ft (12/2 to match Release Node B, Mike 2026-10-01) |
| 5 | Steering (central node) | 5, alone | 20 A | 14/2 | 10 ft |
| 6 | Aft junction | 6 | 20 A | 16/2 | 16 ft |
| 7 | Iridium A | 7, alone | 5 A | 16/2 | any |
| 8 | CAN house bus (its converter) | 8 | 5 A | 16/2 | 90 ft |
| 9 | Wet lab, its one 24 V converter | End | 15 A | 16/2 | 22 ft |
| 10 | Pi A box | End | 10 A | 16/2 | 36 ft |
| 11 | Bilge pump 1 (through its Water Witch) | End | 5 A | 16/2 | 22 ft |
| 12 | USB hub box (A1a) | 8 | 5 A | 16/2 | 45 ft |
| 13 | Orion feed (the Orion-Tr is at the Dyness, A10) | 6 | 10 A | 16/2 | 17 ft |
| 14 | Wing box, for the slip ring (A17) | 3 | 5 A | 16/2 | 34 ft |
| 15 | Bilge pump 2 (through its Water Witch) | 1 | 5 A | 16/2 | 22 ft |
How each size is set (2026-09-30, the smallest that passes both checks): the wire must carry more than its breaker (tinned 105 C marine wire in open air: 16 AWG about 25 A, 14 AWG about 35 A, 12 AWG about 45 A), and the run must lose under 3 percent of 24 V at the real load. "Good to" is the longest one-way run that stays under 3 percent; if a measured run is longer, go up one size. A bigger wire is always allowed, so leftover 14/2 or 12/2 can stand in for 16/2. The duplex (/2) is the run from the breaker out to the load: + and - in one jacket, the - back to the - branch bar. The jumper from a stud to its breaker is one wire, the + only, the same size as its line and as short as the layout allows. It sits ahead of the breaker, so only the 100 A MEGA guards it: keep it under 7 in, or sleeve it in split loom where it is longer. Where each line goes. Four go into the central node: its own feed, release bus A, steering (through the node's drive) and the wet lab (through relay channel 2, which switches it). Seven go straight to their own box or device: Release Node B, Iridium A, the aft junction, Pi A, the USB hub box, the wing box and the Orion-Tr (the node switches the Orion-Tr by a thin wire on its remote terminal; its power does not pass through the node). Four do something else: each CAN line feeds a 24 to 12 V converter that powers a bus, and each bilge line runs through its Water Witch to its pump. Breakers on the board as built (Mike's photo of 2026-09-28, Reference/Main_Bus_Board_Mockup_Photo_2026-09-28.jpg; drawing Reference/Main_Bus_Board_Drawing_2026-10-06): thirteen, read from their colors (clear 5 A x6, red 10 A x2, blue 15 A, yellow 20 A, khaki not yet read, green 30 A x2). The two green are the release breakers; the khaki one is steering if it reads 20 A. The other two are on hand (Mike, 2026-10-06): a 15 A for the wet lab and a 5 A for Iridium A. Nothing left to buy. Read every rating before wiring. Release feeders: 30 A breakers (Mike, 2026-09-28); a 30 A breaker holds a 25 A burn for 60 seconds without tripping. Both feeders are 12/2 (Mike, 2026-10-01). Release bus A runs 12/2 to the central node, near the board; Release Node B runs 12/2 to the stern, 10 to 12 ft (Mike). Both paths end at the transom boxes: A's goes on from the central node through the 14/4 fire pairs, about 10 to 12 ft, while B's fire wires are short. With A's feeder on 12/2, the two paths lose about the same at the coils, about 1.1 to 1.2 V (under 5 percent) at 25 A on the feeder and 12.5 A per horn (Claude's working), fine for burn wires. On 14/2, A lost about 1.4 V, the weaker side. Releases fire one at a time. The burn-wire bake-off confirms the current (Transom Drogue Box Guide sec 9). The 100 A MEGA feeding the small branches stays 100 A. The fifteen breakers add to 185 A, but the worst that draws at once is about 70 A: about 31 A running (node and its relay loads 5, Pi A box 4, aft junction 5.5, both CAN feeds 2, Orion-Tr charging 5, both bilge pumps 3.5, the wet lab 4, the USB hub box 1.5, the wing box 0.6, Iridium A under 0.5), the steering actuator at 10 to 14 A for a few seconds, the trim actuator at 2.6 A, and one release bus at 25 A for 60 seconds. Releases fire one at a time, never both at once (Compute Bay Plan sec 10.3); both would make about 95 A, still under a slow MEGA fuse's opening point of roughly 110 percent of its rating. Not fitted: 4 AWG and a 150 A fuse. Do not re-propose: the link from the 400 A bar to the 150 A bar pair is 6 AWG, good for roughly 100 to 120 A in tinned marine cable, so 150 A would leave it unprotected; 100 A also opens on a partial fault such as a 110 A chafe. Steering feed: a 20 A breaker. Twice the Ewellix's 10.0 A rated current covers brushed-motor inrush and still protects the actuator's own 14 AWG red and black motor pigtail (its current figures and its built-in heat switch: System Settings Sheet sec 3). The two CAN bus feeds: 5 A breakers at the bar, then 16/2 to the two 24 to 12 V converters at the compute multiport (sec 5.3). The wet lab feed: one steady 24 V converter (decided and bought 2026-09-28, $49.99). The wet lab's line (no. 9 in the stud table: its own 15 A breaker, 16/2, switched by central node relay channel 2; 15 A so it never trips on the converter's start or a full 10 A output, Mike 2026-10-06) feeds one potted buck-boost converter (it steps the voltage up or down as needed, so its output holds steady however the bank's voltage moves): 9 to 36 V in, a steady 24 V out, 10 A (XWST, Amazon B097CYG8HG, the 24V-10A option). It feeds everything in the wet lab: the Iwaki pump, the Kamoer dosing pump, the Gulper air pump, both valves (valve 2 is sold as 9 to 24 V, below the bus's charging voltage), the DO sensor (10 to 30 V), the SBE 45 (8 to 30 V) and the DIGITEN flow meter (3 to 24 V). The pressure sensor (5 V) and the temperature probe (3.3 V) take their small power from the board that reads them, over their own cable. There is no wet lab leak sensor: the bilge pump sensors cover a wet lab leak. The wet lab itself: Flow-Through Wet Lab Design. Open: where the flow meter's pulse wire lands, since its pulse now swings to 24 V; drawing the wet lab's wiring settles it. Solar controllers on the main bus board, five: 2x Victron SmartSolar 100/20 then 3x SmartSolar 75/15, left to right (Mike, 2026-10-01: only two 100/20s; the third 100/20 ordered 2026-09-29 was cancelled before it shipped). The one SmartSolar 75/10 is the emergency bank's (A10). The panels and where they sit: Build Guide sec 4.4. Voc is the open-circuit voltage. The cold figure is at -4 F (-20 C) panel temperature, worked from Renogy's -0.25 percent per degree C for its rigid ShadowFlux (the flexible's figure is not published):
| Controller | Panels | Nameplate | Voc at 77 F | Voc at -4 F | Output at 27.6 V |
|---|---|---|---|---|---|
| 100/20 no. 1 | two 100 W pairs | 400 W | 67.2 V | ~75 V | 14.5 A |
| 100/20 no. 2 | two 100 W pairs | 400 W | 67.2 V | ~75 V | 14.5 A |
| 75/15 no. 1 | one 100 W pair (cold rule below) | 200 W | 67.2 V | ~75 V | 7.2 A |
| 75/15 no. 2 | one 200 W, stern, port | 200 W | 44.1 V | ~49 V | 7.2 A |
| 75/15 no. 3 | one 200 W, stern, starboard | 200 W | 44.1 V | ~49 V | 7.2 A |
| 75/10, emergency bank | the two 50 W in series | 100 W | 45.2 V | ~50 V | 3.6 A |
A pair is two 100 W panels in series; pair panels that sit side by side where the layout allows, so both see the same shade. The most on one controller is 400 W, 14.5 A at 27.6 V, inside a 100/20's 20 A. Two pairs on one controller make 7.3 A of short-circuit current, inside every input, and need no fuse of their own; never put three pairs on one controller. Cold rule for 75/15 no. 1. A 100 W pair reaches about 75 V at -4 F panel temperature, which is a 75/15's limit; past it the controller can be damaged. A 100/20 takes 100 V, and a lone 200 W panel stays near 49 V, so only this one controller is at risk. At sea neither boat's route comes near that cold. Ashore, on any night forecast below 20 F, take the PV leads off 75/15 no. 1 (PV first); the others can stay connected. The five solar branches are fuses, in sealed holders within a few inches of the busbar: 25 A on each 100/20 and 20 A on each 75/15, about 1.25x each controller's capped output. Not breakers: an auto-reset one that keeps closing into a hard short fed by a 32 kWh pack can weld its contacts, and nobody is aboard to reset a manual one. A blown fuse loses one controller; the worst single loss is 400 W, 29 percent of the 1,400 W (100/20 no. 1 or no. 2), and a 200 W panel or one pair (any 75/15) is 14 percent. The blade fuse inside each SmartSolar (20 A in the 75/15) guards only against reverse polarity: a chafe fault on the cable to the busbar is fed from the battery and never passes through it. Keep the two 100/20s within 10 ft of the busbars. Wiring a 75/15: terminals left to right BATT +, BATT -, PV -, PV +, LOAD -, LOAD +; leave LOAD empty (the loads live on the busbars). BATT runs on 14/2 duplex tinned (a 25 A fuse protects 14 AWG, and 2 ft loses under 1 percent), PV on 16/2 or bigger (16/2 is good to about 25 ft; the 12/2 left on the roll does as well), BATT to the solar busbar pair through the 20 A fuse on the positive. Connect the battery first and PV last; to take one out, PV first, battery last. Never leave PV connected with the battery disconnected: the controller is not built to survive it. Heat sink ground screws: leave them empty (Mike, 2026-10-06). Each SmartSolar has a ground screw on its black heat sink. The case is isolated from the battery and PV terminals, so a ground wire there is optional, and on this boat it would add a second path for strike current next to the battery bus's one bond wire (Lightning Ground System Guide sec 7). Nothing goes on it on any of the five. Surge modules at the bus (decided 2026-09-26 and 27; the rules: Compute Bay Plan sec 14.1). The battery bus takes two surge modules of one class. Each sits on its own 25 A blade fuse in a sealed inline holder at the + solar bar, its leads from the 12/2 tinned duplex on hand, run together, on a short DIN rail on the main bus board between the + and - solar bars, which reach the 400 A bars by short 6 AWG links (Mike, 2026-09-29), the + link passing through the solar shunt: the leads stay a few inches long and cross no other wiring. No breaker: a breaker can trip on the surge itself, and an auto-reset one cycles into a shorted module. At 24 to 29 V a blade fuse's 1,000 A break rating is a small risk, and the 300 A Class T is the backstop. Each module is the 2-pole type with a third, ground, terminal at the bottom: jumper it to the module's DC- and land one minus lead on the - solar bar, so the bus keeps its one bond wire and gets no second lightning connection (Lightning Ground System Guide sec 6.2). Each charge controller gets its own PV surge module, on a short DIN rail just below it on this board, across its PV leads (the part and the wiring: Compute Bay Plan sec 14.1). Surge parts elsewhere (Compute Bay Plan sec 14.1): a TVS diode at every DC-DC input, the Taidacent RS-12V/2S modules on the two node CAN drops (two on hand; not SM712 chips), the stubs on the Iridium coax, nothing on the GPS coax or on the VESC UART and kill pair (by choice), and a fuse or surge part on every wire into the wing box (A17). TVS diodes: 1.5KE33CA, a Chanzon 20-pack on hand (check the bag holds the 1.5KE33CA, not the 1N5408 flybacks). About a dozen are needed at the hull box inputs, plus whatever the wing box takes, against the 20 on hand: count the bag first; a second 20-pack ($5.89 last time) is Left to Buy item 64, bought only if the count is short. Iridium stubs: two on hand, one for each chain (Mike, 2026-10-07); the Fairview order WFM215964 is cancelled. The MOV-class and PV surge modules are plain heat-disconnect types, no status contacts, no signalling.
The voltage a strike induces in a circuit grows with the area the circuit encloses and how close it runs to the down conductors, so routing is the first lever against lightning (Compute Bay Plan sec 14.2):
Prices and status: the Bill of Materials (BOM), at the section named on each line. Purchase records: the order tracker, BOM_v10_0_Order_Status.xlsx; returns are on its "Send back for refund" tab. Below is only what the BOM does not say.
Full list with links: the Wire and Cable Buy List (A1 to A4).
Open questions for this guide are in Pending Amendments v10.3.
Every entry is a nylon IP68 gland unless it says RJ45 coupler or SMA; gland holes as sec 7.1, SMA 1/4 in, RJ45 coupler about 7/8 in (per part), vent 1/2 in, bond stud 1/4 in. A CAN drop is the male half of a cut Micro-C cable: it plugs into the tee outside, bare end in through a PG7 gland. A point-to-point run is one Micro-C cable cut in the middle, glanded at both ends. Ethernet crosses a wall on an IP67 RJ45 coupler, coax on an SMA F-F bulkhead. A mix-31 ferrite only on the cables that leave the cabin (the list: sec 7.5), outside the entry, just below it, glued down, never hanging on the cable; deep drip loops on every cable. No M12 receptacles on any box. On the compute boxes, Pi B's entries all go on its bottom face and Pi A's on any face but the glued back face (Compute Bay Plan sec 2.3). A compute box bottom face is 11.8 in: up to 14 holes on 1.2 in (30 mm) staggered centres fit, with room for vent and stud.
Where: inboard aft of the settee, glued to the hull below the waterline, back face flat to the hull and no entry in that face; at least 2 ft from the main battery and motor cables where space allows, for the IMUs (a guess at a safe distance). Bus: both. Inside: Compute Bay Housing and Wiring Plan sec 4.1, with desiccant and a humidity breakout. No load current passes through the box except its own feed and the hub's 5 V lead (entry 12). Its power path: the 10 A breaker (entry 1), then inside: TVS, INA226, the Mean Well DDR-60L-5 converter (5 V 12 A, 18 to 75 V in, isolated, on a short DIN rail in the power section at the far end of the plate from the IMUs; output set to 5.1 V, output negative tied to input negative), then the carrier. WD-A's relay (normally closed, sec 2) sits in the 24 V feed ahead of the converter. The USB hub's 5 V lead takes off the converter's output on a 3 A inline fuse (entry 12). Two IMUs, A1 and A2, one at each end of the plate (sec 6.8). The safe-mode wire runs inside the box from a WD-A output pin to Pi A header pin 37 (GPIO26), through a 15k and 22k resistor pair so the Pi sees 3.3 V, not 5 V; no box hole (Compute Bay Plan sec 4.1). Pi A's motor kill pin, when it is assigned, must not use pin 37.
| # | Entry | Through | Hole |
|---|---|---|---|
| 1 | Main 24 V feed, 16/2 on its own 10 A auto-reset breaker at the 150 A bar (end stud, sec 7.6; 6 A breakers come in 12 V only) | PG9 | 5/8 in |
| 2 | CAN control drop, from the multiport (shield drain to bond stud) | PG7 | 1/2 in |
| 3 | CAN house drop, from the multiport | PG7 | 1/2 in |
| 4 | VESC A UART throttle + GPIO kill: one Micro-C cut in the middle, glanded both ends; never shares a CAN cable. The kill line leaves WD-A through a PC817 optocoupler board inside | PG7 | 1/2 in |
| 5 | Iridium A serial (cut Micro-C; isolated inside) | PG7 | 1/2 in |
| 6 | Link to Pi B: the alive discrete only (cut Micro-C; a PC817 optocoupler board inside) | PG7 | 1/2 in |
| 7 | USB lead from the DSD TECH SH-C30G USB to CAN adapter in the hub box (the trim actuator's CAN link, A1a and A17), to Pi A's second USB port, not a hub port, so the wing's control never depends on the hub | PG7 | 1/2 in |
| 8 | Ethernet to the TSW202 in the switch box at the aft junction (A6a; switch port 7); outside face taped (A6a) | RJ45 coupler | per part |
| 9 | (not drilled: the Starlink run goes to the TSW202) | - | - |
| 10 | (not drilled: Pi B runs to the TSW202) | - | - |
| 11 | GPS A coax (no arrestor, no bias tee; check SMA, not RP-SMA) | SMA F-F | 1/4 in |
| 12 | The USB hub's 5 V lead out: a short 2-core lead from the Mean Well's 5 V output, on its 3 A inline fuse inside this box, to the hub's power input in the hub box (A1a). The dAISy and its coax are in the hub box (sec 6.2) | PG7 | 1/2 in |
| 13 | (not drilled: the E32 wing radio's 915 MHz whip went with the slip ring) | - | - |
| 14 | (not drilled: no RTL-SDR whip) | - | - |
| S | The hub's USB lead from the USB hub box (A1a), to Pi A's other USB port; no USB isolator (A1a). 11 entries + vent + bond stud; three more fit under the 14-hole cap | PG7 | 1/2 in |
| V | Vent | 1/2 in | |
| B | Bond stud 1/4-20 | 1/4 in |
Bus: none: two USB leads to Pi A, the hub's own lead (A1 entry S) and the CAN adapter's lead (A1 entry 7). Powered from: two feeds. The hub runs off Pi A's 5 V: a short 2-core lead from the Mean Well's output, on its 3 A fuse in the Pi A box (A1 entry 12), so the hub restarts whenever WD-A cycles Pi A (Mike, 2026-10-07). The box's own 24 V feed, 16/2 on its own 5 A breaker on the branch bar (line 12, sec 7.6), feeds only the 12 V converter, through one fused 24 V line inside (fuse size set at fit-out). Open: which stud, the end stud or stud 8 as in the sec 7.6 table; Mike's call settles it.
Not fitted: the DROK 24 to 5 V converter bought for the hub (the hub runs off Pi A's 5 V); the USB isolator (DSD TECH SH-G01A), which with one shared 5 V supply would isolate nothing; a relay to cycle the hub alone (weighed 2026-10-07; do not re-propose). If the hub drops off, Pi A tries a USB reset first, and if the hub stays gone it stops its heartbeat on purpose, so WD-A cycles Pi A and the hub together (software repo). Rules. FTDI silicon only for serial, never CH340 or CH341; a udev rule keyed to each adapter's serial number keeps each sensor's name after a reboot and sets each FTDI chip's USB latency timer (the serial addresses and the timer: System Settings Sheet sec 9). Order any instrument that offers RS-485 as RS-485. No RS-232 sensor ever lands on a Nucleo pin. No serial add-on boards (no quad-UART HAT; do not re-propose one). No other RS-232 cables until an instrument needs one. Every USB plug clamped within 2 in and staked (Compute Bay Plan sec 4.1). What crosses its wall (about 10 entries):
Open: the box itself (it is to buy: its material, and so whether it takes a membrane vent) and the gland for each entry; buying the box and having the cables in hand settles them.
Where: a few feet from Pi A on the settee panel. Bus: both, through the isolated HAT. Floats: nothing here is wired to the lightning copper. Inside: Compute Bay Housing and Wiring Plan sec 4.2, with desiccant and a humidity breakout; WD-B holds Pi B's 24 V feed through its relay's normally closed contact, and has its own D36V6F5 (sec 2). No bond stud. Plastic glands only. The Ethernet bulkhead jack is the Alrhso, whose body is zinc and plastic: on plain UTP cable its metal touches no wire or shield, so Pi B stays isolated (Claude, inferred; fine to use).
| # | Entry | Through | Hole |
|---|---|---|---|
| 1 | Emergency bank feed | PG7 | 1/2 in |
| 2 | CAN control drop (cut Micro-C drop, isolated HAT) | PG7 | 1/2 in |
| 3 | CAN house drop (cut Micro-C drop, isolated HAT) | PG7 | 1/2 in |
| 4 | Iridium B serial (cut Micro-C) | PG7 | 1/2 in |
| 5 | Link from Pi A: the alive discrete only (cut Micro-C; optocoupler inside) | PG7 | 1/2 in |
| 6 | AS3935 detector lead, 7-core, under 3 ft, to WD-B through a 4-channel two-way level shifter inside (the 3.3 V detector to the 5 V Nano). The detector logs strikes; WD-B never cuts Pi B's power for lightning | PG7 | 1/2 in |
| 7 | (blank; spare) | ||
| 8 | Ethernet to the TSW202 in the switch box at the aft junction (A6a; switch port 8), UTP, long run to the stern; outside face taped (A6a) | RJ45 coupler (Alrhso) | per part |
| 9 | GPS B coax (no arrestor, no bias tee) | SMA F-F | 1/4 in |
| V | Vent | 1/2 in |
Bus: control. Powered from: main bank branch, its own 15 A auto-reset breaker at the 150 A bar (stud 3, sec 7.6), 16/2; the release bus has its own 12/2 feeder on a 30 A breaker (stud 4), steering its own 14/2 on a 20 A breaker (stud 5), and the wet lab its own 16/2 on a 15 A breaker (end stud), switched by relay channel 2. Contents: everything in sec 4.1 (the staked Nucleo and its CAN and surge boards, the expander, the potted Klnuoxj 5 V converter, the DIN rail, the two relay modules, the steering power-cut relay and the Pololu G2 drive (sec 5.6), the MOSFET modules, the four MY2NJ reversing relays in DIN sockets with their 1N5408 diodes, the four puller fuses in DIN holders, the opto board, the two DS18B20 chains, the BME280 and humidity breakouts), and the charging path beside relay 8 (sec 6.7). No short relays, no branch current meters, no VE.Direct reader, no gas breakout.
| # | Entry | Through | Hole |
|---|---|---|---|
| 1 | CAN control drop | PG7 | 1/2 in |
| 2 | 24 V feed, from the 15 A breaker | PG9 | 5/8 in |
| 2a | Release bus feeder: 12/2 on its own 30 A breaker at the bar (stud 4), never the node's feed. Measure the 12/2's jacket, then size the gland | PG11 or PG13.5 | 3/4 or 13/16 in |
| 3, 4 | Release trunks to the transom terminal box: the two drogue node-A fire pairs (12-14 AWG) and the four JSD cord puller power pairs, on two 18 AWG 4-core cables, one per pair of pullers (Mike, 2026-10-02), 22 AWG for the rest. Open: which cores ride which trunk. Confirm, bridle, aux-rudder and pull-pin cores DELETED. Count cores at loom time; reverse adds none | PG11 x2 | 3/4 in |
| 12, 13 | Steering: the 14/2 feed in from the 20 A steering breaker (stud 5), and the cable out to the Ewellix: the drive's two 14 AWG motor leads (red and black) plus the actuator's three 22 AWG position-feedback wires, green (the wiper), white (the full-retract end) and brown (the full-extend end) (sec 5.6) | PG11 x2 | 3/4 in |
| 14, 15 | Aft trunk from the aft junction (12-core): two Water Witch states, encoder A, wind (Calypso CMI1018 UART), and the two bilge manual-ons (channels 6 and 12) unless they ride 16/2 runs (settle at loom time, sec 4.1). Drill entry 15 only if the fit-out count needs it | PG11 (x2 if needed) | 3/4 in |
| 16 | Switched loads out to the aft junction: six 16/2 duplex runs, three per gland, sealant-filled; or three and two, with the nav lights' run placed at fit-out (the two plans differed: settle it at loom time). One of the six is the wet lab's own line after relay channel 2 (entry 29) | PG13.5 x2 | 13/16 in |
| 17 | Spare, not drilled (it was the INA226 I2C chain to the branch shunts, DELETED). Drill it only when a cable needs it (Mike, 2026-09-26) | (PG7) | (1/2 in) |
| 18 | 1-wire chains x2 (two cables, one gland, sealant-filled) | PG9 | 5/8 in |
| 20 | Spare, not drilled (it was the battery bus and PV surge module status contacts, DELETED: not watched en route). Drill it only when a cable needs it | (PG11) | (3/4 in) |
| 21 | (not drilled: no gas sensor lead) | - | - |
| 22 | The thin battery positive in from the 1k at the main bus board, and the one run from the five VE.Direct remote on/off cables (ASS030550320), which meet on the main bus board, tied together (decided 2026-09-26; sec 6.7). One bundle, sealant-filled. Measure each cable's run at fit-out; extend a short one with 18/2 tinned marine wire joined at a screw terminal strip, not twisted or soldered: the line carries only a few milliamps, so any size from 22 to 16 AWG works, and 18/2 is easy to handle | PG9 | 5/8 in |
| 23 to 24 | (DELETED: the VE.Direct data cables; each MPPT's VE.Direct port takes its remote on/off cable, entry 22) | - | - |
| 25 | Spare, not drilled. Drill it only when a cable needs it | (PG9) | (5/8 in) |
| 26 | (not drilled: no cabin gas sensor lead) | - | - |
| 27 | Camera port UV-C feed, relay channel 9, out to the LM317 regulator and the UV-C lights at the underwater camera port | PG7 | 1/2 in |
| 28 | Orion-Tr remote on/off pair, relay channel 10, out to the Orion-Tr's remote on/off terminal (A10) | PG7 | 1/2 in |
| 29 | Wet lab line in: its own 16/2 from its 15 A breaker (line 9, sec 7.6), to relay channel 2; it leaves on one of the entry 16 runs and goes on through the aft junction (A6 entries 11 to 13) | PG9 | 5/8 in |
| 30 | Horn pair out: 16 AWG pair from relay channel 13 and its 5 A inline fuse, about 15 ft to the horn on the instrument post, up through the post's existing deck stub | PG9 | 5/8 in |
| V | Vent | 1/2 in |
Where: bolted through the battery box's aft wall on the minus side, opposite the main fuse, at mid-height or above, where the inside of the wall faces only the 1.75 in air gap and the aft G10 plate, never the cells; never on a long side wall, where the nuts would sit against the padding and the cells (Mike, 2026-10-03). Four 1/4 in stainless bolts, each hole bored oversize, filled with thickened epoxy and redrilled, each bolt bedded in sealant, fender washers inside, nylon lock nuts or blue Loctite; clear of the lid screws, the main cables and the fuse (the steps: Battery Box Building Guide Parts 4 and 8). All its glands face down, each cable with a drip loop. It is IP65 (spray, not standing water); a drowned BMS box only loses readings, since it controls nothing. Bus: none. Powered from: the pack, through the SBMS0's own supply lead; the JK balancer takes its own power pair, V- on the W0 tap and V+ on the W8 tap. Contents: Electrodacus SBMS0 with its WiFi/USB add-on board, JKBMS JK-B2A8S, a Winford BRK2X6-L-DIN 12-pin ribbon-to-screw-terminal board on a DIN rail (the SBMS0's tap wires land on it; a new 12-pin ribbon feeds it from the SBMS0, whose own ribbon is not cut), the butt splices from the JK's harness to its tap cable, and the sense leads from the SBMS0's two shunts. It watches and controls nothing: no charger wiring, no resistor (sec 6.7); if the SBMS0 dies, charging carries on and only the readings stop. No humidity breakout, no gas sensors. Lace the harnesses to the plate at each plug; dielectric grease on the plug seals only (sec 7.4). The terminal map, wire colors and splices: Battery Box Building Guide Part 8; the workbench sheet: Reference/BMS_and_Balancer_Wiring_Sheet_2026-10-03.pdf. Four holes plus the vent.
| # | Entry | Through | Hole |
|---|---|---|---|
| 1 | Signal bundle from the battery box: eighteen tap wires, nine for the SBMS0 and nine for the JK, each on its own ring and its own fuse link at the battery (A9 entry 3). The SBMS0's nine plus the XT1 thermistor pair on 11 cores of the 22 AWG 12-core shielded cable (its shield grounded at this end only, at W0); the JK's nine on the ENERJOUR 18 AWG 10-core cable (one core spare); and the DS18B20 chain. One bundle, laced and sleeved in adhesive-lined heat shrink so the gland seals on one round sleeve. No gas-sensor leads. Open: the gland at both ends (PG13.5 takes about 1/2 in; likely PG16, to about 9/16 in, or PG21, to about 11/16 in, from the glands on hand); measuring the sleeved bundle settles it | PG13.5, PG16 or PG21 | per gland |
| 2 | (not drilled: the five remote on/off cables meet on the main bus board, and one run goes to the central node, A3 entry 22) | - | - |
| 3 | Shunt sense leads, one twisted pair per shunt (nothing else goes on the shunts). Both shunts sit on the positive side of the main bus board: the 300 A between the Class T fuse's output and the + 400 A bar, the 100 A between the + solar bar and the + 400 A bar (sec 7.6). Each sense wire sits at battery voltage, so each has a 30 AWG fuse link at the shunt end | PG9 | 5/8 in |
| 4 | USB lead to the USB hub box beside Pi A (A1a): micro USB on the WiFi/USB add-on board, tied down right outside the plug; readings only, required | PG7 | 1/2 in |
| 5 | (DELETED: the charge-continue bypass) | - | - |
| 6 | DS18B20 chain out, to the central node (A3 entry 18) | PG7 | 1/2 in |
| V | Vent | 1/2 in |
Neither boat has an FSESC box. Boat 1 has one steering actuator, and Boat 2's steering backup is an aux rudder, not a second actuator. The number is kept so later box numbers stay put. Do not re-propose.
Two boxes side by side. The aft junction is this plastic box and, beside it, the aluminum network switch box (A6a), which takes every network cable. Bus: none. Nothing in it talks on CAN, so the house bus ends at Release Node B's tee (sec 5.2) and no CAN cable comes in. Contents of the plastic box: DIN rail with screw terminal blocks (sec 4.1), one fuse per post feed; the post wind meter's Pololu D36V6F5 regulator just after the wind sensor's fuse (sec 4.1); motoring-white relay; encoder A landing. No photocell relay. No network parts: the switch and its supply are in A6a, and the Ethernet surge protectors are on the network surge board near the deck (A6a). About 14 entries (15 if the second trunk is drilled), plus the vent; measure the box before the loom. The entry numbers are kept, so the network entries that moved to A6a stay listed here as moved.
| # | Entry | Through | Hole |
|---|---|---|---|
| 1, 2 | (moved: the post camera Cat6 runs go to the switch box, A6a entries 1 and 2) | - | - |
| 3, 4 | Post cables: wind sensor (5 V up, Calypso CMI1018 UART down and on to the trunk), lights | PG7 x2 | 1/2 in |
| 5 | (plugged: no weather pod) | PG7, plugged | 1/2 in |
| 6, 7 | (moved: foredeck fisheye and interior camera, A6a entries 3 and 4) | - | - |
| 8 | (moved: Ethernet to Pi A, A6a entry 7) | - | - |
| 9 | 24 V feed, always on (line 6 from the main bus board, sec 7.6); the post feeds hang on it. Open: whether the switch's 24 V feed is jumpered across from here or runs to the switch box on its own (A6a entry 10); Mike to say | PG9 | 5/8 in |
| 10 | Switched loads in from the central node: the six 16/2 duplex runs of A3 entry 16, sealant-filled; one is the wet lab's own line from its 15 A breaker, switched by relay channel 2 in the central node (sec 7.6) | PG13.5 x2 | 13/16 in |
| 11 to 13 | Load bundles out: lights group, Starlink power, em-trak + the wet lab's own line | PG11 x3 | 3/4 in |
| 14, 15 | Trunk to the central node, 12-core (drill entry 15 only if the fit-out count needs it) | PG11 (x2 if needed) | 3/4 in |
| 16 | Bilge leads: the two Water Witch 230-24 pump-running outputs (24 V present = pump running) + the two manual-on feeds, pump 1 on relay channel 6 and pump 2 on channel 12. No high-water sensor | PG9 | 5/8 in |
| 17 | (plugged: no shaft sensors) | PG9, plugged | 5/8 in |
| 18 | Rudder encoder A lead (PWM output) | PG7 | 1/2 in |
| 19 | Side-light feed (the one centerline bow lantern) | PG7 | 1/2 in |
| 20 to 23 | (moved: underwater camera, Pi B, Starlink and the light driver, A6a entries 5, 8, 9 and 6) | - | - |
| 24 | (moved: SFP port 10 is in A6a) | - | - |
| V | Vent | 1/2 in |
The box: a CXCESNS aluminum box, 11.8 x 8.3 x 3.9 in, the same as the two compute boxes (Compute Bay Plan sec 3.1). Its back face is glued flat to the hull for cooling, the way the Pi A box is (Compute Bay Plan sec 2.3); entries go on the other faces, with drip loops. The box is wired to nothing: it stays off the lightning copper list, like the cooling plates (Lightning Ground System Guide sec 6.2). Contents: the TSW202 switch and its Mean Well DDR-120B-48 supply, set and sealed (sec 6.1), always on, never on a relay; the 10GTEK SFP-T copper module in SFP port 9. Nothing else: nothing lightning-related is inside this box. About 10 entries plus the vent. Keep the box well above bilge water. The switch and its supply clip to a DIN rail on a long side wall, not glued to the floor (Mike, 2026-10-03). The floor (the face glued to the hull) is too shallow: the TSW202 is 4.8 in (122 mm) deep and the box only 3.9 in. On the side wall the switch reaches about 4.8 in across the 8.3 in floor, plus about 1.2 in of plug, which leaves about 2.3 in to turn the patch cords. Heat goes from the rail to the box wall, the glued face and the hull: about 5 to 15 W in all (the supply loses about 9 percent, the switch makes 3 to 8 W). Right-angle patch cords at the switch ports, so no cord bends hard at its plug; a short right-angle adapter at a port is fine where a straight plug cannot turn in the 2.3 in. Tie each cord down within about 2 in of its plug, so the plug carries none of the cord's weight when the boat pounds. Open: which way the right-angle plugs must point; reading it off the switch, with the distance from the rail to the port face, settles it before the cords are ordered. Network jacks, boat wide (Mike, 2026-10-05). Every network cable that crosses a box wall comes in on a bulkhead jack (an IP67 RJ45 panel coupler, plug in on both sides), one per cable: the Alrhso couplers, zinc and plastic. The outside face sits in the hull's salt air, and salt is what corrodes the thin gold plating; the jack's cap seals only an empty jack. On each outside face: a thin smear of dielectric grease on the plug, plug in, then self-fusing silicone tape stretched to about twice its length, half-overlapped, from the jack's threads over the plug and 1 to 2 in down the cable, with no gaps, and a drip loop below. Jacks go on side or bottom walls, facing sideways or down, never on the top. Inside faces, within sealed boxes, need no tape. The jacks: this box 9, Pi A 1, Pi B 1, the Starlink enclosure 1, the light driver's dry box 1 (13 in all). The network surge board (near the deck, not in this box). The five Ubiquiti ETH-SP-G2 Ethernet surge protectors (outdoor-rated) and the small tinned bus bar for their grounds sit on a small board on a bulkhead near the deck, where the outside network cables first come inside: the two post cameras, Starlink, the foredeck fisheye and the interior camera. The bar's ground lead runs short and straight to the lightning copper, a foot or two at most (Lightning Ground System Guide secs 6.2 and 7.4). From each protector a short premade Cat6 runs down to its wall jack on this box, low in the hull. Keep the cables before the protectors (from outside) apart from the cables after them (down to the switch). Open: where the lightning copper runs near that bulkhead, and so which bulkhead takes the board; check at the boat. Network cables. Every network cable is premade, plugs and all; no plug is crimped on the boat. Coil the extra near the surge board in small tight coils about 8 in across, strapped, never one big loose loop, and keep the coils on the camera side apart from those on the switch side. No coils inside this box: a short premade patch cord, 6 in to 1 ft, runs from each wall jack to its switch port. No sharp bends at any plug. Check that each premade plug passes the deck holes and glands on its route; where one does not, use a split gland (a gland that opens to close around a cable that already has its plug).
| # | Entry | Through | Hole |
|---|---|---|---|
| 1, 2 | Post cameras: forward camera Cat6 (port 1), aft camera Cat6 (port 2); each from its surge protector on the surge board | RJ45 coupler x2 | per part |
| 3, 4 | Camera Cat6: foredeck fisheye (port 3), interior camera (port 4); each from its surge protector on the surge board | RJ45 coupler x2 | per part |
| 5 | Underwater camera Cat6 from the Axis P1245 main unit inboard (port 5); no surge protector | RJ45 coupler | per part |
| 6 | Cat6 to the PoE light driver in its dry box (port 6); no surge protector | RJ45 coupler | per part |
| 7 | Ethernet to Pi A (port 7) | RJ45 coupler | per part |
| 8 | Ethernet to Pi B (port 8), UTP, so the coupler's metal touches no wire or shield (A2); long run from the settee | RJ45 coupler (Alrhso) | per part |
| 9 | Ethernet to Starlink (SFP port 9 through the copper SFP module), from its surge protector on the surge board | RJ45 coupler | per part |
| 10 | 24 V feed for the switch, always on, from the aft junction's line (the switch and its 54 V supply hang on it). Open: its own run from the main bus board, or a short jumper from the plastic box's terminals (A6 entry 9); Mike to say | PG9 | 5/8 in |
| 11 | (not drilled: no ground lead; the surge protectors' ground is at the surge board) | - | - |
| 12 | (SFP port 10, free: a coupler only if wanted; not drilled) | - | - |
| V | Vent | 1/2 in |
A7, VESC A (control bus, 8 holes: CAN, UART + kill, battery + and - at PG16, phases x3 at PG16 with the 1.0 in mix-31 clamp outside, sensor lead). The controller is the Flipsky 75200 Pro V2.0 (Bill of Materials sec 15.5; as bought: Build Guide sec 5.1; settings: System Settings Sheet sec 2). Its power button is a plain on/off switch in the logic supply, so a crimped shorted plug replaces it: the switch plug's pins are GND, IGNITE and BAT+; join IGNITE to BAT+ and the board runs whenever the battery is connected. The board has no spark protection, and no precharge is fitted (Mike, 2026-09-29): the battery switch makes the first connection, and it is built for it. The box reaches the hull for cooling through an aluminum plate through its floor, set in heat-carrying epoxy and wired to nothing (decided 2026-09-26; Compute Bay Plan sec 4.4): size the plate before the box is drilled. Plugs. COMM: JST-PH 8-pin (5V, 3.3V, GND, ADC1, ADC2, ADC15, TX, RX); the pairs on hand fit it; its pigtail splits to the UART backup (TX, RX, GND) and the motor sensor (5V, GND, ADC1, ADC2). CAN: JST-GH 5-pin (GND, CANL, CANH, 5V, PWR.ON) on plain pre-crimped leads; wire CANH, CANL and GND only, PWR.ON left unconnected. Find pin 1 on each plug with a meter: GND rings to battery negative; powered, 5V and 3.3V read as marked. Bench test: power up with the shorted plug. Motor leads. The ME1616's three orange phase leads, about 2 to 3 ft, boots and lugs fitted, phases labelled: nothing to make up at the motor, no joint in the 300 A path. Measure their real length before siting the controller. In a CUT-BURST the phase current reaches about 300 A for seconds while the battery-side cap stays about 180 A; the short run soaks that up as heat. Bus capacitance (decided 2026-09-26). The long 2 AWG battery leads ring on every switching edge (owners report blown capacitors, then transistors), so extra capacitors sit right at the controller: four Nichicon 1,000 uF 63 V (Claude's sizing) on a small G10 board bolted to two insulated studs in this box, where the 2 AWG feed (5 to 6 ft, twisted) and the controller's own 8 AWG leads also land. Studs: MGI SpeedWare 3/8 in power junction posts, bought 2026-09-26, $18.95, here 2026-09-28 (order tracker; rated 160 A there, against about 80 A at 2 kW), in place of the Blue Sea 2017 Dual PowerPost picked the same day. Drawing: Reference/Controller_Capacitor_Bank_2026-09-26.png (no precharge, Mike, 2026-09-29). Motor position sensor. The ME1616's SinCos sensor harness is built to Build Guide sec 5.1 (six wires, two connectors, the divider crimped into the cable, potted, ratios on the label; crimp, no solder). The motor's sensor plug is an Aptiv Metri-Pack 150, 8-way, sealed; its mating half: Bill of Materials sec 15.5. Sin goes to ADC1 and cos to ADC2 on the COMM plug, both 3.3 V inputs; the sensor (an RLS RMB29 board) runs on 5 V and its tops reach about 3.6 V, so fit a divider on each signal, 10k from the sensor and 18k to ground, unless the bench check shows the tops stay under 3.2 V (not a digital level-shifter board; those are for on/off signals). Bench checks, before the harness is potted: 5 V on the sensor, turn the shaft by hand, read the lowest and highest sin and cos; the KTY84/130 temperature sensor pair reads about 600 ohm at 77 F; the board must have the 10k resistor on the motor temperature pin that the firmware assumes, and the motor must read room temperature within about 5 F at first power-up (System Settings Sheet sec 2). The fallback is sensorless: fine at cruise, weakest at full torque from a standstill. A8, VESC B: BOAT 2 ONLY. Boat 1 sails with one VESC: losing propulsion costs time, not the trip, because the wing sails the boat (Routing Guide 11.5). Boat 2's box has 8 holes (CAN, battery + and - at PG13.5, phases to the contactor x3 at PG16, contactor coil set/reset + sense at PG7), with the phase-transfer contactor and its coil driver; none of it is in the Boat 1 totals.
Bus: none. Five holes, plus the main fuse's bolts if they go through the wall and the BMS box's four bolts (row M). Build: Battery Box Building Guide.
| # | Entry | Through | Hole |
|---|---|---|---|
| 1, 2 | 2 AWG mains, + and -, in line with the terminal posts at measured post height, cables straight over the aft G10 plate (aft wall) | PG16 x2 | 7/8 in |
| 3 | Combined signal bundle: eighteen tap wires, nine for the SBMS0 and nine for the JK, the two wires of each joint on two different posts of that joint, one ring per stud, each fused at the terminal by its own 30 AWG tap-wire link (18 rings and 18 links), sleeved, laced; plus the XT1 thermistor pair and the DS18B20 chain (aft wall). The cables and the gland size: A4 entry 1 | PG13.5, PG16 or PG21 (A4 entry 1) | per gland |
| V | Gore vent, the breathing inlet (aft wall) | M12 | 1/2 in |
| F | The 300 A main fuse, bolted to the outside of the aft wall at mid-height on the + side (sec 7.6). Any bolt through the wall is bedded in sealant | Bolts | per part |
| M | The BMS box, bolted through the aft wall on the minus side, opposite the fuse, at mid-height or above (A4): four 1/4 in stainless bolts, each hole bored oversize, filled with thickened epoxy and redrilled, each bolt bedded in sealant. Drill before the pack goes in | Bolts x4 | per part |
| R | Relief stub: 1-1/2 in Sch 40 PVC glassed through the FORWARD wall, high, pointing up at 45 degrees into open cabin space, discharge forward away from the electronics; inside it a Dometic/Sealand 1-1/2 in toilet duckbill (OEM 385310076), lips out, flange clamped by a PVC coupling. No screen (the duckbill is shut except when the box vents; the cell-temperature trip is the protection). Not fitted: a bug cap or a fine flame mesh (do not re-propose) | 2 in hole saw | 2 in |
Deleted, do not re-propose: heaters, gas sensors (in the box or at the cabin deckhead), the high-water sensor, the overboard relief duct with its transom hole and burst membrane. The six DS18B20 pack probes and the cell-temperature trip are the protection.
A Dyness 24V 100Ah LiFePO4: 25.6 V, 100 Ah, 2,560 Wh, 20.9 x 8.11 x 8.54 in, 42.6 lb, plastic case, built-in 100 A BMS with low-temperature cutoff, ~$360. Not the Bluetooth version. Not the team-built 8x EVE LF50 + JBD 60 A bank. The unit is its own enclosure: no box is built, no hole schedule, no gland list. It carries Iridium B and the backup compute island with its watchdog, no steering actuator, and stays apart from the main bank until needed. Its loads are tiny: about 25 to 32 Wh a day (Compute Bay Plan sec 5.3). Its standby level and monthly exercise: System Settings Sheet sec 1.
Two recharge paths, independent of each other:
Where the Orion-Tr goes (2026-09-23): on the Dyness's shelf or bulkhead, within about a foot of it, so its output (which must float) reaches the fused DIN strip in a few inches, through its own 10 A fuse. Dry (it is only IP43), upright with a few inches of air round it, on wood or G10 with no bond wire; meter its case to each negative on the bench first. Its input is the 10 A branch from the main bus board, 16/2 across the boat. The 75/10 mounts beside it. Circuit connections: feed to Pi B, feed to Iridium B, the SmartSolar 75/10 in, and the Orion-Tr float charge in from the main bus. All land at the posts and a small fused DIN strip beside the battery, not through box glands. Open: the strip's fuse sizes for Pi B, Iridium B and the 75/10; the fuse sizing rule (sec 7.6) applied to each load and its wire settles them.
Contents: RockBLOCK 9603 carrier, an isolated 24 to 5 V converter, TVS. Isolated (Mike, 2026-10-02): each antenna's stub arrestor ties the coax shield to the lightning copper, and with an isolated converter that tie stops at the Iridium side, so the main negative (A) or the emergency island's floating negative (B) is not tied to the copper a second time. The converter (Claude's pick, 2026-10-09, standing unless Mike says otherwise): a Mean Well DDR-15G-5 in each box, 9 to 36 V in (both banks sit inside that), 5 V 3 A out against the RockBLOCK's 0.5 A at most, 4,000 V between input and output, screw terminals, on a short DIN rail on the box's mounting plate; 0.69 x 3.54 x 2.15 in (17.5 x 90 x 54.5 mm), so it stands about 2-1/2 in off the plate on its rail (Amazon B07B6553JY; $20.50 at Jameco). Its idle draw is not published. The box (Claude's pick, 2026-10-09, standing unless Mike says otherwise; Mike, 2026-10-09: the Pi boxes are full, so a bigger Iridium box): a TICONN IP67 ABS box, 5.9 x 5.9 x 3.5 in, with its mounting plate (Amazon B0B87THLGC, pick that size; price not checked), room on the plate for the converter, the RockBLOCK beside it and a TVS, and about 3 in inside from plate to lid. Not picked: a board-mount isolated module small enough for a tiny box (not sold on Amazon, and it must be soldered onto a board); the converter in the Pi box (the Pi boxes are full); the potted Klnuoxj (not isolated). Entries: serial to its Pi (PG7), coax straight to the stub arrestor (PG7), always-on power (PG7; for A, 16/2 on its own 5 A breaker, stud 7, sec 7.6), vent. A on the main bank near Pi A; B on the emergency bank near Pi B. Antennas and domes: sec 6.5.
Contents: SparkFun AS3935 breakout only. Entries: 7-core lead to WD-B (PG7), vent.
Entries: Ethernet to the TSW202 in the switch box (A6a; RJ45 coupler, its outside face taped as A6a says; SFP port 9), power (PG9). No heater entry. Deck: the post bundle, the GPS and Iridium A coaxes and Starlink's Cat6 and power pair share the instrument post's 1 in PVC deck stub and its 1 in liquid-tight conduit, sealed round the wires at the stub. The route, and the sample-stub check before the real stub is glassed in: Instrument Post Guide secs 7 and 8.
em-trak transponder (house, tee + N2K drop; channel fails to ON). The Ewellix CAHB-22 steering actuator has no bus and no box of ours: its cable runs to the central node (A3 entries 12, 13). Electrak HD tees on the control backbone are Boat 2 only. The wing has no electronics of its own since the slip ring (decided 2026-09-26): no battery, panel, charge controller, radio, computer or charging link, and no pod. Its trim actuator, tip wind meter and wing compass reach the hull through the slip ring and land in the wing box (A17). What stays on the wing, and the ring itself: Wing Power System sec 1A.
Bus: house bus end: its tee is the last on the house backbone and carries the terminator plug (sec 5.2). Powered from: its own 12/2 feeder on a 30 A breaker at the bar (stud 2, sec 7.6). Contents (sec 4.2): NUCLEO-G431KB on a Nano screw carrier (staked) + CAN transceiver breakout with a Taidacent RS-12V/2S surge module on the CAN drop, potted Klnuoxj 24 V to 5 V converter, fuse, enable relay (24 V 30 A 1-channel module), dual-MOSFET modules for the two live channels, mini DIN strip. No shorting relays, no opto board, no humidity breakout. All logic under screws.
| # | Entry | Through | Hole |
|---|---|---|---|
| 1 | CAN house drop from the end tee | PG7 | 1/2 in |
| 2 | 24 V feed: the 12/2 feeder on its own 30 A breaker at the bar (stud 2). Measure the 12/2's jacket, then size the gland | PG11 or PG13.5 | 3/4 or 13/16 in |
| 3 | Drogue box 1, node-B channel | PG7 | 1/2 in |
| 4 | Drogue box 2, node-B channel | PG7 | 1/2 in |
| 5 | (DELETED: pull-pin station; Boat 1 has single-actuator steering, no pins) | - | - |
| 6 | (DELETED: pull-pin station) | - | - |
| 7, 8 | (DELETED: the opto sense leads, with the opto board; a 4-hole box plus vent) | - | - |
| V | Vent | 1/2 in |
The dying-gasp reflex is DELETED, do not re-propose. Nothing is wired: no valve close lines to the backup island, no fuses for them, no pin on Pi B or WD-B. Valve 2, the duty valve, springs closed on power loss by itself (auto-return), and that alone is the boat-level protection. Valve 1 (stays-put 5-wire, open 24/7) needs no drive: a power-dead boat with valve 1 frozen open is acceptable, because the hull is still owned by the seacock and by valve 2. The valve types are in the Flow-Through Wet Lab Design.
Bus: none. Contents: DIN strip only, one landing per conductor; no electronics, nothing to fail. Fuses stay at the central node end.
| # | Entry | Through | Hole |
|---|---|---|---|
| 1, 2 | Release trunks in from the central node | PG11 x2 | 3/4 in |
| 3, 4 | Drogue box 1 and 2, node-A channels out | PG9 x2 | 5/8 in |
| 5, 6 | JSD cord puller cables out (the T8 trigger cords): one 18 AWG 4-core cable per gland, each carrying two pullers' power pairs (Mike, 2026-10-02) | PG11 x2 | 3/4 in |
| 7 | (Aux-rudder release out: BOAT 2 ONLY, unwired on Boat 1) | - | - |
| 8, 9 | (DELETED: pull-pin runs, with the single-actuator decision) | - | - |
| 10 | (DELETED: bridle tension switches; the speed drop and the aft camera confirm the drogue is out) | PG9, plugged | 5/8 in |
| V | Vent | 1/2 in |
The hull end of the slip ring (decided 2026-09-26; the ring, the wing's wiring and what stays on the wing: Wing Power System sec 1A). The ring's eight wires land here: the trim actuator's 24 V pair and CAN pair, the wing-tip wind meter's RS-485 pair and the wing compass's RS-485 pair (Mike, 2026-10-01). At the tip the 24 V pair feeds a branch on its own 1 A inline fuse: the compass runs on that 24 V, and a small converter makes 5 V from it for the tip meter alone (Wing Power System sec 1A). Every wire from the ring gets a fuse or a surge part at this box's entries, so a strike on the mast stops here and does not reach the other boxes. The 24 V pair gets the box's own 5 A fuse and a 1.5KE33CA surge diode; the CAN pair and both RS-485 pairs get surge modules where they enter the hull (Compute Bay Plan sec 14.1): three Taidacent RS-12V/2S, two ordered 2026-09-26 and the compass's bought 2026-10-07 (Wing Power System sec 8.7). The trim actuator stays on its own short CAN link, kept off the boat's two buses. No angle sensor cable: the wing angle comes from the compass on its RS-485 pair. No reader sits in this box: both wing readers are in the USB hub box (A1a). The wing box to hub box cable (Mike, 2026-10-07). One shielded 4-pair cable, one long run, one gland at each end: the actuator's CAN pair, the tip meter's and the compass's RS-485 pairs, each after its surge module, and a ground pair tied to the wing's 24 V negative in this box. The ground pair is needed because an isolated reader's signal side floats: it must share a reference with the sensors, or the signals drift out of range. RS-485 terminators off. The cable: any shielded Cat6, all copper, with a shield over all four pairs (Mike, 2026-10-09: "Any shielded cat6 should work"; what to look for, its pairs and colors: Wire and Cable Buy List C8). Its shield is grounded at this end only (Mike's OK, 2026-10-09), its pigtail on this box's ground terminal: the 24 V negative terminal, where the ground pair and the three surge modules' ground terminals also land (each module's ground terminal goes to that negative, as the battery bus surge modules' do, so the boat keeps its one bond wire). At the hub box the shield is trimmed back and sleeved, touching nothing. The strike comes in at the wing box, so whatever the shield picks up drains there, at the box built to stop it; grounded at the hub end instead, the shield would carry that surge into the box beside Pi A, which is what the isolated adapters are there to stop, and grounded at both ends it would close a loop through the boat's wiring (Claude's reasoning; Mike's OK, 2026-10-09; the ring's own shielded leads are grounded at the hull end the same way, Wing Power System sec 1A). What crosses its wall:
Open: the box's material, whether it takes a membrane vent, and the gland for each entry; picking the box and the cables settles them. Totals (from the entries above; Boat 1 boxes only): no M12 receptacles on any box; glands PG7 37 (A1 8, A2 6, A3 3, A4 2, A6 5 (1 plugged), A7 3, A11 6, A12 1, A15 3), PG9 14 (A1 1, A3 5, A4 1, A6 3 (1 plugged), A13 1, A16 3 (1 plugged)), PG11 13 (A3 5, A6 4, A16 4; two more if the second aft trunk is drilled at fit-out), PG13.5 4 (A3 2, A6 2), PG16 7 (A7 5, A9 2), plus the two release feeder glands (A3 entry 2a, A15 entry 2), PG11 or PG13.5 once the 12/2 jacket is measured, and the battery tap bundle's gland at each end (A4 entry 1, A9 entry 3), PG13.5, PG16 or PG21 once the sleeved bundle is measured; RJ45 bulkhead jacks 13 (A1 1, A2 1, A6a 9, A13 1, the light driver's dry box 1; 14 if the free SFP port gets one); SMA 3 (A1 1, A1a 1, A2 1); membrane vents 11 (A1, A2, A3, A4, A6, A6a, A11 x2, A12, A15, A16) plus the battery box Gore vent; ferrites about 11, only on cables that leave the cabin (sec 7.5), at least 20 on hand. The central node's spare entries 17, 20 and 25 are not drilled. Not yet counted: the glands of the USB hub box (about 10 entries, A1a) and the wing box (at least 3 entries, A17), and the light driver's dry box's other entries. The only M12 disconnects are the tees on the two backbones.
Each part's price and status: the Bill of Materials, pointed to from sec 8 (the wing's parts: its sec 15.6). Purchase records: the order tracker, BOM_v10_0_Order_Status.xlsx. Out, said only here (deleted or returned; do not re-propose): the ZED-F9P moving-base pair and two extra ADuM1201 isolators; the Thomson Electrak HD24B026 (cancelled 2026-09-11; the card was never charged, so no refund is due); Delrin pull-pin stock; two-part polyurethane topcoat and its primer (one-part Rust-Oleum Marine Topside everywhere above the waterline, Build Guide sec 3.1).