Written 2026-10-19 at the v10.3 promotion. v10.3 is the v10.2a guides with every decision made through 2026-10-19 written in: the USB hub on standoffs with its plugs held by zip ties; the wing CAN link at 250 kbit/s; the wing compass at about 23 Wh a day; the wing can's three PG9 glands and the slip ring brush bracket; the one-piece cable road and the sealed mast bore; the actuator shield left open at the wing end; the nose lead set by the Stage 10 balance; the drogue U-bolt collars and the sideways-pull wording; the gross tonnage; and more. The "Closed at v10.3" list near the end names each. This is the only open list. Part 1 is work with nothing left to decide. Part 2 is questions for Mike. Checked for accuracy and consistency 2026-10-19 (the same day): about 290 mismatches fixed in the guides and four order tracker cells; what is left is item 439 (Part 1) and item 440 (Part 2), and the "Closed by the v10.3 check" list near the end names what closed. Item numbers keep their v9.9 meaning. New items start at 438. The closed record is the "Closed at v10.3" list and the lists after it near the end of this file; older closed lists (v10.2 and before) are in Archive/Pending_Amendments_v10_3.md, Archive/Pending_Amendments_v10_2.md, Archive/Pending_Amendments_v9_9.md and Archive/Pending_Amendments_v9_8.md. Software items live in the software repo.
466. Motor coolant loop DROPPED; the jacket is filled three quarters with coolant and plugged (Mike, 2026-10-29: "yes, drop the loop and fill a quarter air"). Worked out in Reference/Motor_Dry_Running_2026-10-29.md (estimates, not maker figures; no ME1616 rating without coolant exists). Why: 250 W and 500 W run all day in every water on the route with no loop (winding about 80 to 165 F); about 1 kW holds all day except in the hottest sealed hull; 2 kW lasts about 1.5 to 3 hours before the 212 F derate, then the motor holds about 1 to 1.5 kW, which matters only for a 16 to 20 kt lee-shore claw-off (Build Guide 5.3). The loop as written would not have flowed: Build Guide 5.2 put the cooler plate below the jacket outlet, but hot coolant rises; and even working, an 8 x 12 in plate through the fiberglass only about doubled the cooling, so "2 kW without time limit at any sea temperature" was too strong. Gains: a simpler mount (no hoses, the ports can face any way), no coolant leak or air lock, no cooler plate to make and glue, about 14 lb lighter (guess), no liquid path for motor hum around the rubber mounts. The fill: ordinary car antifreeze, 50/50 premixed (universal extended-life), never concentrate on its own, pink RV antifreeze or plain water. The jacket holds about 12 fl oz (guess, 6 to 22). On the bench, motor turned as it will sit, pour into one port with the other open until it comes out, note how much went in, draw out a quarter, then plug both ports with two M16 x 1.5 plugs (below) on two M16 bonded seal washers from the kit on hand, Tef-Gel on the threads, snug plus a small nudge. Brass is fine, corrected the same day (Mike, 2026-10-29: "Brass fittings is how the motor came"): Motenergy itself puts brass in this aluminum case, and car engines carry brass plugs in aluminum heads, so Claude's worry about brass pitting the case was overstated. Where the motor sits (Mike, 2026-10-29): the inside of the boat is a generally dry, open space, and the bilge, where the pumps are, is forward of the motor and about 18 in below it; Claude had wrongly put the motor in "a damp, salty bilge", following Design Decisions R-3.5 ("a belt would run 17,000+ hours in a damp salt bilge"), which needs the same fix. Brass or stainless both do. Gap: Build Guide 3.5 and the 5.1 table say the motor came with no coolant plumbing; it came with brass fittings in its ports (Mike). Probably those are the brass M16 x 1.5 to 1/2 in barbs the guides list as bought, since no M16 pair is logged as bought (guess). They were hose barbs (Mike, 2026-10-29), so they come out and two plugs go in; and the "bought" M16 barbs in the guides are these, not a purchase (closes 440 (j)). The plugs, Claude's pick from Mike's Amazon search: uxcell 304 stainless M16 x 1.5 inner hex plug with seal ring, 2-pack, $9.48 (Amazon B0982SK9R9, seen 2026-10-29). Its socket head stands lower than an outside hex, which helps the tight mount, and its thread is about 3/8 in (10 mm) long, short enough not to bottom in the port (inferred). Fit each on an M16 bonded seal washer from the kit on hand rather than the rubber ring it comes with. A quarter air, not full and not a tenth: full has no room for the coolant to swell; a quarter gives about 15 psi at 200 F and 21 psi at 220 F, mostly water vapor, and it cannot boil below about 270 F at that pressure; a tenth gives about 34 psi at 220 F and sits too near the point where the swelling coolant fills the pocket. The jacket's pressure rating was not found. No firmware change: the 212 F / 248 F derate and the 194 F shore alert stand. Closes 335 (d) and 440 (j) (the barbs are not used) and takes the motor cooler plate out of 86b and 455. Written in: the order tracker (Parts, the loop row REMOVED and a new jacket plug row; Buy list, the plugs), 2026-10-29. To write in: Build Guide 3.5 (the coolant exits; "came with no plumbing" becomes brass fittings in its ports; build order step 5 becomes the jacket fill), 5.1 (Cooling and Continuous rating rows), 5.2 (rewrite the cooling text: sealed, three quarters full, no loop, the run times above), 5.3 (Operating modes: Max reserve 2 kW for about 1.5 to 3 hours, then about 1 to 1.5 kW; lee-shore limits); Design Decisions R-3.5 (the "No oil bath" paragraph; "damp salt bilge" in the belt paragraph becomes the dry, open space above the bilge) and R-8.1C (three hull-glued boxes, not four); Compute Bay Plan sec 2 and the Lightning Guide's hull-glued plates paragraph (no motor cooler plate); BOM 15.5 (the loop line becomes the plugs and the coolant); Glossary (Thermosiphon goes); Reference/Aluminum_Stock_List_Boat1.md row 7. *Documents: as named.*
465. Boom glass: cloth along and across, plus 1 extra layer on the side faces (Mike, 2026-10-28: "yes, along and across, plus 1 extra layer"). The boom keeps the guide's 3 layers of 17 oz biax over the front third and 2 over the rest, but the cloth is cut so its threads run along and across the boom, not at 45 degrees, and both side faces get 1 more layer for the first 12 in. Worked out: the boom's bending strength just behind the wing goes from about 8,500 in-lb (45-degree cloth) to about 24,000 in-lb, about 6 times the 60 lb breakaway case with a 1.5 margin (Reference/Boom_Root_Strength_Calc_2026-10-28.md). About 0.2 lb more glass plus about 0.3 lb more nose lead (a guess). *Documents: Wing Construction Guide Stage 12 (the boom); the boom-root pocket build sheet, page 3.*
464. Tail size and boom length stay as they are (Mike, 2026-10-27: "Tail size stays. Boom length stays."). The tail stays 27 x 11 in, about 2.0 sq ft, and the boom keeps the tail 58.8 in behind the pivot. Worked out 2026-10-27: 6 in more boom gives about 10 percent more authority for about 3.2 lb counting nose lead; a tail 3 in taller gives the same for about 1.2 lb; the bigger tail is a bolt-on swap later if sea trials ask for it, the boom is not. No document change; recorded so the question is not reopened. *Documents: none.*
454. The wing wiring tube moves off the trailing edge: Josh's channel plan, DECIDED (Mike, 2026-10-27: "pex is specced for the wing"; Josh's email, 2026-10-25; worked plan Claude, 2026-10-26). Open within it: how the cables get from the wing can to the root end of the PEX, since PEX barely bends (Mike, 2026-10-27); Claude's plan: item 461. The guide glues the 5/8 in bore wiring tube into the trailing-edge inch before Stage 11 packs it solid (Wing Construction Guide Stage 10). Josh reports it cannot fit, and the wing's own shape agrees: at the 27.7 in chord the NACA 0015 section is only about 0.43 in thick where the ribs stop and tapers to nothing, while a 5/8 in bore tube needs about 1-1/4 in of thickness, found only about 3-1/2 to 4 in forward of the trailing-edge point. Brandon is holding the first skin on this. Josh's fix, with Claude's worked details, recommended as a package: (a) Where: a straight channel cut spanwise into one face of the foam stack, centered 24 in back from the nose (3.7 in forward of the trailing-edge point, wing about 1-1/4 in thick there), from just above the root closeout to the tip closeout, tube bedded in thickened epoxy, filled flush, skinned over. The channel face is the face the boom and the actuator fairing mount on; the copper lightning strap then goes on the other face, which already matches the boom clamping plate and its jumper bolt (Stage 10; this also settles the open "strap: which face" bullet below). Since neither face is committed yet, Brandon cuts the channel in the face he skins second and that face becomes the boom face, so the first skin can go on now. (b) The tube: one 10 ft stick of 3/4 in PEX water pipe, 7/8 in outside, about 0.68 in bore (more than the 5/8 in minimum, so the 0.26 in actuator cable and the wing cable, guessed 0.3 in, C5 not bought yet, pass with the pull string). Plastic, so nothing new for lightning. Sand the outside so the epoxy grips it. About $5 to $9 at Home Depot (Apollo PEX-B 3/4 in x 10 ft, $5.23, ); one stick covers both lengths. This replaces the unpicked "sealed wiring conduit, $40" line in the wing materials list. (c) Past the boom root: do not run the tube through the solid zone (the trailing 4 in packs solid over a 6 in band at about 44 to 50 in up; a tube at 24 in from the nose would clip its forward edge). Two lengths instead, both ending inside the actuator fairing: root to fairing, about 44 in plus the sweep, and fairing to tip, about 31 in plus the sweep. Each fairing end sweeps gently up toward the boom face and stops just under where the skin will be, end taped shut and dammed so no epoxy enters, string inside; at Stage 10 the skin is cut open over each end inside the fairing footprint, just clear of the 6 in solid band (about 43 in and 51 in up), clear of the four boom bolts, and each mouth gets a heavy thickened-epoxy fillet to the fairing floor (this stands in for the guide's "one sealed fitting"; alternative, Josh's pick: let short stubs stand proud before skinning, like the socket tube at the root closeout, and skin around them). The fairing drawing, which must be checked against the Electrak MD's STEP file anyway, is drawn to cover both mouths. The actuator cable leaves at the fairing; the wing cable crosses the fairing between the two mouths and carries on to the tip. Prove each string pulls freely after gluing. (d) Drainage: the root length's low point is its root end, so gel-fill that end with silicone around the cables (the guide's second listed fix); the tip length's low mouth opens inside the fairing, which has its own drains, so it needs no weep hole. (e) While the stack is open: take the aft foam out of the boom-root zone and pack it solid now, before skinning, instead of digging it out between cured skins at Stage 10; the face doublers still go on over the skins at Stage 10, cut around the mouths. (f) Knock-ons at write-in: Wing Construction Guide Stages 7 to 11 (channel before skinning; Stage 11 packs the trailing edge with no tube in it) and the Stage 4 and 6 cable-road wording; Wing Power System 1A, 8.6 and 11; Trim Actuator Guide sec 4 (the fairing-floor gland line becomes the filleted mouths); the "trailing-edge conduit" name throughout. The mast wall hole item below stands; the root crossing just gets about 4 in shorter. The epoxy fill plus tube adds roughly half a pound aft of the pivot, inside what the Stage 10 balance measurement absorbs. *Documents: Wing Construction Guide Stages 7 to 11; Wing Power System 1A, 8.6 and 11; Wing Trim Actuator Installation Guide sec 4; Lightning Ground System Guide 5.1; Reference/Wing_Conduit_Relocation_2026-10-26.md.*
457. Cord pullers paired by corner (Mike, 2026-10-27: "by corner, otherwise split into 2-wire cables to run"). Each 18/4 carries the two pullers at one quarter: one cable for the port quarter (drogue 1's and drogue 2's port T8s), one for the starboard. Each cable runs straight to its corner and opens there; by drogue would have meant running loose pairs, in effect 2-wire runs, across the transom to the far corner. Accepted cost: a dead cable leaves one leg of both drogues held, so neither can be let go until it is fixed (Claude's note; a cable in the dry hull rarely dies). The transom terminal box is since dropped (item 456): each 18/4 now runs in one piece to its corner. Written in 2026-10-27: Electronics and Wiring Plan A3 entries 4 and 4a; Wire and Cable Buy List C3; Transom Drogue Box Guide sec 6; Box Wiring Sheets A3 and R1; Stern_Release_Wiring_2026-10-27.
458. Rename "trim tab" to "tail flap" everywhere (Mike, 2026-10-27: "Tail flap is fine"). On an airplane a trim tab is the small tab that takes the load off the stick; the hinged back 2-3/4 in of our tail is the steering surface itself, so "trim tab" was the wrong name. Only the word changes: the tail stays bolted to the boom and the flap stays the one moving part. At the next doc set, rename it in Build Guide secs 6.1, 6.4 and 6.5, Wing Construction Guide (sec 1, Stages 10 and 12), the Wing Trim Actuator Installation Guide (its title becomes the Wing Tail Flap Actuator Installation Guide), Design Decisions R-6.2 and R-6.4, Wing Power System, Bill of Materials, Start Here and the Glossary (the "Trim tab" row becomes "Tail flap", and the "Wing tab spring link" becomes the "wing flap spring link"), and the two 3D model pages in Reference. *Documents: all of the above.*
459. Build Guide sec 6.4: the wing's angle follows the tail flap about one half to one, not "roughly one to one" (found 2026-10-27). Measured plain-flap data (NACA TN 796, Figure 1(b): 0.50 for a flap a quarter of the chord) put the wing at about half the flap angle, so the 30 degree flap stops give about 15 degrees of wing angle. That is still all the wing can use: the NACA 0015 section stalls at 13 to 15 degrees at 15 to 27 knots (Sandia SAND80-2114, Figure 7) and the whole short wing a few degrees later. Fix the sentence at the next doc set; nothing else changes. The numbers and sources: Reference/Wing_Airfoil_Measured_Data_2026-10-27.md. *Documents: Build Guide sec 6.4; Wing Trim Actuator Installation Guide sec 1.*
456. Transom terminal box DROPPED: the release cables run straight from the central node to the stern (Mike, 2026-10-27: "yes, drop the box"; Claude's proposal from Mike's "what other leftover incorrect things might exist?"). The box was planned when the stern took twenty-plus wires (drogue confirms, bridle switches, pull-pin stations, the aux rudder), all now deleted or Boat 2 only. What is left is 12 wires on three cables that would leave the box almost as they came in, so it only added two screw joints per wire, in a box on the transom, on the one circuit nobody can fix at sea (Box Wiring Sheets A16 said as much: at a dozen wires, direct runs are competitive and the joints go away). Do not re-propose a stern terminal box for Boat 1. The runs now: (a) The fire trunk: the CONEPY 14/4 tinned (0.40 in, fits a PG11, Mike 2026-10-27), from the central node to inside the hull near the transom, in one piece. There the jacket is opened, glue-lined heat shrink goes over the cut jacket end (Electronics plan sec 7, cut cable ends), and each fire pair runs on as its own two cores in a sleeve to its own potted transom gland and on to its lid coils: drogue box 1 on red (+) and black (-), drogue box 2 on white or the lightest core (+) and the fourth core (-). No joint from the central node to the coil tails. Each pair's + is the release bus; its - goes to its own MOSFET switch at the central node. (b) The puller cables: two 18/4 shielded tinned (0.25 in, fits a PG7, Mike 2026-10-27), each in one piece from the central node to its quarter, paired by corner (item 457): one for the port quarter's two pullers (drogue 1's and drogue 2's port T8s), one for the starboard. The jacket opens at the corner and each pair goes to its motor: red (+) and black (-) to drogue 1's puller, white or lightest (+) and the fourth core (-) to drogue 2's. Each pair's wires come from its puller's reversing relay at the central node. The shield drain is tied to the 24 V negative at the central node end only and trimmed and sleeved at the corner (Claude's pick, as on the wing cable, standing unless Mike says otherwise). (c) At the central node, entries 3 and 4 become three entries, one cable each (Mike's yes to a third gland, 2026-10-26): the 14/4 in a PG11 (3/4 in) and each 18/4 in a PG7 (1/2 in). Labels both ends, ferrules on every end, drip loops, chafe sleeve where the cables pass anything hard. The two routes still never share a gland, bundle or trunk: Release Node B's own runs stay separate. What goes: the transom terminal box and all its entries (A16), the UK2.5B blocks and DIN rail it would have used (back to the shelf), the 14/4 tail pieces and the 18/4 splices (less 18/4 used: the 50 ft roll also feeds the wing actuator run and maybe the channel 9 run, so measure what is left after the wing run). The box on hand (9.5 x 5.5 in, Mike 2026-10-26) is free for another job. The build sheet drawn for it, Transom_Terminal_Box_Build_Sheet_2026-10-26, is moved to Archive\. The plastic box count drops from nine to eight. Written in (2026-10-27, at Mike's "update all other boxes that are affected"): Electronics and Wiring Plan sec 3 box table and count, sec 7.2, sec 7.5 ferrite list, A3 entries 3, 4 and 4a, A16 (now a DELETED note), the totals; Box Wiring Sheets index, Sheet 0, A3 (drawing and notes), A16 (now a DELETED note) and R1 (drawing, table and notes), PDF rebuilt; Wire and Cable Buy List C3; Transom Drogue Box Guide secs 6 (puller wires; the plugged gland line); the new drawing Stern_Release_Wiring_2026-10-27 (.html master, .pdf to print). Node Boxes Explained names no terminal box and needed no change. The cables' gland sizes (14/4 PG11, 18/4 PG7) are also noted, at Mike's ask (2026-10-27), in Wire and Cable Buy List C3, C4 and C6, Transom Drogue Box Guide secs 5 and 6, Electronics plan A3 entry 27, and Box Wiring Sheets A16 and R1.
455. Heat-carrying epoxy grit: the #120 aluminum oxide on hand, for every hull-glued bed (Mike, 2026-10-26: "120 grit will be plenty"). Settles the open grit line in Compute Bay 2.3 and Build Guide 5.2 (#120 bought, #220 or #320 written) and item 289 (8). The same bed glues the Pi A box, the network switch box and the VESC A box floor (the motor cooler plate is gone: item 466). 12 lb on hand (Amazon, 2026-09-11); none to buy. The bed fills the hull curve, so it is far thicker than a #120 grain (about 0.005 in); a finer grit matters mainly in paper-thin glue lines (Claude, agreeing). The "prove the grit on hand on the bench" wording goes with it. Written in: the VESC A box guide and print sheet (2026-10-26), order tracker Parts row "Heat-path grit". To write in: Compute Bay 2.3, Build Guide 5.2, BOM (the grit line), and the Pi A box guide's open item on the hull bond grit.
453. Backup battery board: no on/off switch on the plus lead; the 30 A ANL is the off switch for work (Mike asked 2026-10-24, "does it need a on/off switch for pos?"; Claude's pick, standing unless Mike says otherwise). The backup battery must never be off at sea, and with no crew a switch that can be left off will one day be off, the same reason the VESC's power button became a shorted plug (Design Decisions, the motor controller). A switch also adds two more bolted joints in the one lead the whole backup island hangs on. For work on the board: undo one nut and lift one end of the ANL fuse, which kills everything on the board's + bar. Then cover the two 50 W panels (they feed the 75/10 in any daylight) and pull the Orion-Tr's green 2-pin remote plug, which turns it off. The board's breakers are auto-reset with no off lever (Mike, 2026-10-24: "breakers aren't switchable btw"), so breaker 13 cannot cut the Orion-Tr's input: its two input wires on the board's left strip stay live from the main battery until the main battery switch (Blue Sea 6006) is off. Turn that off only to touch those two wires. Whether the Dyness has its own power button: not found (Claude, 2026-10-24). *Documents: Electronics and Wiring Plan A10 (one line under the backup battery board); Backup_Battery_Board_Layout_2026-10-15 notes (written 2026-10-24).*
452. VESC A box: no surge diode on the feed (Mike, 2026-10-24: "delete it", after the pros and cons). The 1.5KE33CA that VESC_A_Box_Wiring_2026-10-14 drew across the posts is gone; the Electronics plan's A7 never named one (settles item 401 t). Why: the 32 kWh battery and the four 1,000 uF capacitors hold the feed far below the diode's 31.4 V turn-on, its 1,500 W for about 1 ms is tiny against motor regen if the battery is cut off while the motor turns, and when it fails it fails shorted, which here blows the 250 A motor fuse and ends propulsion until someone is aboard. Not fitted, do not re-propose: a diode on the VESC feed, plain or fused. The Chanzon 20-pack's count for the hull boxes drops by one if the VESC was counted. Written in: VESC_A_Box_Wiring_2026-10-14 (drawing and Power note), the build guide VESC_A_Box_Build_Prep_2026-10-24. To write in: Box Wiring Sheet A7 ("TVS on the feed"), BOM 15.6 and Lightning Guide 7 diode counts if they include the VESC.
451. VESC A box: the Pi A cable's landing at the Flipsky (Mike, 2026-10-24: "yes to pi a cable", to Claude's recommendation). The kill core goes to the servo plug's PPM pin and the ground core to the servo plug's GND beside it. A 10k resistor from COMM 3.3V to PPM, in the harness, holds the line high (motor allowed) whenever the PC817 is off, whether or not the firmware turns on the pin's own pull-up (not known). So a loose kill wire cannot stop the motor; the motor still ramps to zero if CAN and the serial line both go quiet (Build Guide 9.2). COMM's one GND serves the motor sensor alone, so no thin JST-PH wires are spliced; every VESC ground pin is one ground (inferred). TX and RX stay on the COMM plug. Settles item 419 (c) and the Pi A box guide's open question on where the PC817 output lands (V2 on the kill core, G on the ground core, at the Pi A end). The 10k comes from the resistor kit on hand; nothing to buy. Written in: VESC_A_Box_Wiring_2026-10-14 (drawing and pin lists, 2026-10-24), the build guide VESC_A_Box_Build_Prep_2026-10-24. To write in: Electronics plan A7 (Plugs paragraph: the COMM pigtail no longer splits its ground), System Settings Sheet sec 2 (kill switch row: the 10k pull-up), Build Guide 5.1 (layer 3), Compute Bay 4.1 (the kill channel's output landing), Box Wiring Sheet A7.
450. VESC A box: the CXCESNS aluminum box, its floor as the cooling plate, and the motor leads re-lugged inside (Mike, 2026-10-24: "yes to both, buying the box today"). (a) The box is a CXCESNS cast aluminum box, 11.8 x 8.3 x 3.9 in, the same as Pi A, Pi B and the network switch box; Mike is buying it 2026-10-24 (Amazon, $65.99, $60.71 with the coupon). Replaces BOM 15.5's "VESC A waterproof enclosure, ON HAND" ($40 estimate, model never named). (b) The box floor is the cooling plate: the floor glues straight to the hull in the heat-carrying epoxy (Compute Bay 2.3), the way the switch box does (decided 2026-10-03), and the Flipsky bolts to four 316 flat-head studs set through the floor from outside in epoxy before gluing, heads flush, on the Minus Pad 8 gap pad; the posts sit on the same kind of stud. No separate plate and no board-size hole in the floor. Closes "Cooling plate size" (Part 1B). Heat: roughly 60 W at 2 kW (a guess at 97 percent), a few watts at the usual 150 to 250 W, through about 98 sq in of floor. The box stays wired to nothing (Lightning Guide 6.2): the Flipsky's metal base touches only the gap pad and the studs, and an aluminum-board base is insulated from its copper (inferred). A desiccant pack goes in, since the floor runs near sea temperature. (c) The motor's lugged phase leads cannot pass a PG16 (its seal takes 0.39 to 0.55 in cable; a heavy 3/8 in lug is about 3/4 in wide, a guess): cut each lug off, feed the lead in through its PG16, crimp a new 3/8 in lug inside, and bolt it face to face to an 8 AWG 3/8 in lug on the Flipsky's own lead of the same letter (U to U, V to V, W to W) with a 3/8-16 x 1 in 316 bolt, washers and a nyloc, glue-lined heat-shrink over each joint. Settles item 419 (d). To buy: three 3/8-16 x 1 in 316 bolts (none that length on hand), and three lugs for the motor lead if it is not 2 AWG (its size is printed on the jacket; not on record). Layout and gland plan, redrawn 2026-10-26 from Mike's photo of the parts in the box: the Flipsky is about 2.4 x 4.3 in with its battery leads out one end and its motor leads out the other (switch and sensor plugs on one side, CAN and USB on the other, servo and COMM at the motor end); the posts sit under the battery glands along one short wall, the capacitor bank and WAGOs beside the Flipsky under the + post; the box has a ledge with holes along each short wall and a boss low on each long side. Glands, all 1 in below the rim, with the box drawn lying long side left to right, battery end on the left and corners numbered 1 top-left, 2 top-right, 3 bottom-right, 4 bottom-left (sheet redone 2026-10-26 in the style of Mike's Pi A box sheet, at his request): + and − PG16 on the battery end, 2-3/8 and 5-7/8 in from corner 1; the three phase PG16s on the motor end, 2-1/2, 4-1/8 and 5-3/4 in from corner 2; C PG7 on the top side, 7-3/8 in from corner 1; S and U PG7s on the bottom side, 6-5/8 and 7-5/8 in from corner 4. The Flipsky's power button and its lead come out (the shorted plug replaces them); its USB lead stays coiled inside for VESC Tool with the lid off (Claude's pick). Flipsky size and holes (2026-10-26): the case is 130 x 68 mm (5-1/8 x 2-11/16 in, Flipsky's spec), not the 2.4 x 4.3 in read off the photo (the photo's camera sat close, so the deeper Flipsky looked small). Its four mounting holes are blind and threaded in its finned bottom plate, 40 x 120 mm (1-9/16 x 4-23/32 in), centered, read off Flipsky's own photo of the bottom (inferred; thread M3 or M4 is a guess); it has no ears. Full-size pattern: Flipsky_Hole_Pattern_2026-10-26 (.html master and .pdf). Hold-down (Mike, 2026-10-26: "holes"): four 316 flat-head screws from outside, up through the floor into the Flipsky's own holes, countersunk flush (90 degrees if metric), set before the box is glued to the hull; the Flipsky then stays in for good. Replaces the floor studs and nylocs for the Flipsky; the posts keep their studs. Motor lead entry (Mike, 2026-10-26): he decides once the motor is mounted and the distance known; the choices are three PG21 glands (1-1/8 in holes, seal 0.51 to 0.71 in) passing the motor's own lugs uncut (Claude's pick; glue-lined heat-shrink built up on the lead if it is under 0.51 in), the PG16 re-lug plan above, or the joints outside the box (Claude advises against: salt on bolted joints). Any motor lead may go to any Flipsky lead, one to one; the order only sets the direction, set in VESC Tool, but bolt all three before the motor setup wizard. Build guide (Pi A guide format, with the floor plan, hole map, wire list and gather checklist): VESC_A_Box_Build_Prep_2026-10-24.html (top of the folder); the two-page print, the floor plan with every hole and the notes, is VESC_A_Box_Build_Sheet_2026-10-26 (.html master and .pdf, Mike, 2026-10-26: the full guide printed 21 pages). Written in: VESC_A_Box_Wiring_2026-10-14 (updated 2026-10-24; the leftover 1.0 in clamp drawing removed), order tracker (Parts row "VESC boxes", Buy list). To write in: Electronics plan A7 (box, cooling, motor leads), Build Guide 5.1 and 5.2 (the plate paragraph), Compute Bay 2.3 and 4.4 ("motor controller plate" becomes the VESC box floor), Lightning Guide 6.2, Design Decisions (the four hull-glued boxes), BOM 15.5 (the enclosure line and the three bolts), Box Wiring Sheet A7 notes, Wire and Cable Buy List A3 ("no joint in the 300 A path" becomes one bolted joint per phase).
447. USB hub box bottom wall: one Altelix 4-port split gland for the four round cables (Mike, 2026-10-22: "4 port"). Split gland B takes the BaroVUE 10 cable (port 2), the HygroVUE 10 cable (3), the SBE 45 FTDI line (4) and the DO sensor FTDI line (5), in place of four PG7 glands; one 1-3/16 in (30 mm) hole. From the 2026-10-14 five-pack on hand, nothing to buy. Each cable must be 0.15 to 0.30 in (4 to 7.6 mm) across to seal (Altelix's page); check before cutting. The 24 V feed (16/2, oval) keeps its own PG9 gland, sealant-filled: a slit seal closes only round a round cable. So the hub box's wall entries are: top, the 5-port split gland and the PG7 for the hub's 5 V lead; bottom, split gland B and the PG9. Drawn in USB_Hub_Box_Wiring_2026-10-20 the same day. *Documents: Electronics plan A1a (wall entries) and the gland totals; Box Wiring Sheet A1a.*
446. USB hub box: one 1.5KE33CA surge diode across its 24 V input (Mike, 2026-10-22, fitting it in the built box: "Where does the diode go"). Settles the mismatch in item 442: Box_Parts_With_Photos_2026-10-15 listed it, no guide did, and the first drawing said none. It sits on TB1-1 (+24 V) and TB1-2 (0 V), one leg under each, ahead of the YRDZXG, legs short and sleeved; the CA part is two-way, so it has no right way round. Same part and same place as the other boxes' power inputs (Compute Bay 14: TVS on each power input), from the 20-pack on hand. The no-fuse decision (item 426) stands. Drawn in USB_Hub_Box_Wiring_2026-10-20 the same day. *Documents: Electronics plan A1a; Compute Bay 14; Box Wiring Sheet A1a; BOM 15.4 (TVS count).*
446. The wing data cable is Mike's shielded Cat6 on hand, about 5 ft; its shield is grounded at the wing box only, as item 403 has it (Mike, 2026-10-21: "yes to unshielded to wing", then "turns out cat6 is shielded to wing box, does the shield need connected to anything?"). Mike's run is about 5 ft: the wing box and the hub box both sit in the middle of the boat. The 25 ft in the docs was never a measured run, only the length the Monoprice cable is sold in ("25 ft reaches anywhere inside the hull"). On the 5 ft run a shield is not needed (the twisted pairs do the noise work; the three Taidacent surge modules at the wing box and the isolated adapters in the hub box do the lightning work), so plain Cat6 would have done; but the Cat6 on hand turned out shielded, and a shield that is there gets landed the one way: twist the braid or drain into a short pigtail, sleeve it, land it on the wing box's 24 V negative terminal with the brown pair and the three Taidacent ground terminals; at the hub box trim it flush and sleeve it, touching nothing. Never both ends (a ground loop through the boat's wiring), never the hub end only (it would carry a strike surge into the box beside Pi A). Left floating at both ends it would do no harm but no good either. (Claude's reasoning, the same as item 403's, which Mike OK'd 2026-10-09; treated as decided since it only restores that rule.) Mike's Cat6 is 23 AWG, so most likely solid: strap it within a few inches of each box so it cannot flex at the terminals; its wires go into the terminals bare, no ferrule. Colors as before: blue CAN, orange tip meter RS-485, green compass RS-485, brown ground. Fit: shielded 23 AWG Cat6 is often about 0.28 to 0.32 in across (Claude's guess, not checked against this cable), against split gland port b's 0.15 to 0.30 in (item 445); if it will not seal there, it takes its own PG9 gland on the hub box's bottom wall, as before item 445. Nothing to buy: the Monoprice line (Amazon B077H5LL5N, $14.98) is dropped. Claude also compared the 22/12 shielded cable on hand (C1): it fits port b (0.24 in) and is stranded, but its wires are not twisted in pairs, so the Cat6 is the better pick. The other Cat6 runs stay unshielded as written, and the Pi B run must stay unshielded. Drawn in USB_Hub_Box_Wiring_2026-10-20 the same day. *Documents: Wire and Cable Buy List C8 (the 25 ft and the buy line) and the totals; Electronics plan A17; Compute Bay 10.2; BOM 15.6; order tracker (done 2026-10-21). The shield rule in Electronics plan 7.4, Compute Bay 7, Wing Power System 8.7 and Box Wiring Sheets A14 stands as written.*
449. The BaroVUE 10 and HygroVUE 10 cables: each sensor's own Campbell cable runs whole into its hub box gland, nothing to buy (Claude, 2026-10-21, from Mike's "what kind of cable for baro and hydro vue?"). From the BaroVUE 10 manual (rev 04/2022, copy: Reference/Campbell_BaroVUE10_Manual_2022-04.pdf): its cable plugs into a removable 5-pin screw plug on the sensor; maximum cable length 3 m (10 ft); supply 9 to 28 V; the pressure fitting is a 1/8 in barb. Campbell's replacement cable (BaroVUE10CBL) is 30 in (campbellsci.com). So the BaroVUE sits inside the hull within reach of its 30 in cable from the hub box's gland 2 (Claude's pick, standing unless Mike says otherwise); the hose from the Gill port does the long run, and a sensor beside the hub box is still above the hose's low point and drain (Barometer guide sec 4). This settles the "fitting to confirm" check in Barometer guide sec 1: a 1/8 in barb, sealed, so no sealed box; the 1/4 in ID hose needs a 1/4 to 1/8 in barb reducer at the sensor. The HygroVUE 10 ships with 5 m (16.4 ft) of cable (campbellsci.com product page); what length was ordered on Josh's account is not known. Its mounting spot is still open (Barometer guide sec 1); pick it within its cable's reach of the hub box, or join on screw terminals with the leftover 22/12 shielded (C1), which SDI-12 allows. *Documents: Barometer and Pressure Port secs 1 and 4; Electronics plan A1a; USB hub box drawing; BOM 15.11 (the barb reducer, pocket change); order tracker.*
448. The SDI-12 adapter's screw order, settled from the maker's manual (Mike, 2026-10-21: "can't read + or -", then "please research and give me board reference"). LiuDr SDI-12 USB Adapter User Manual 2021-02-26, p. 10 (copy: Reference/LiuDr_SDI-12_USB_Adapter_Manual_2021-02-26.pdf): "The markings are on the underside of the board to save space. When you are facing the connector's holes with the board right-side-up, the pins are ALWAYS "-", bus, "+" from left to right." Use the top-left SDI-12 block: its wire holes face the middle of the board (Claude's reading of Mike's photo), so with the board as photographed it reads left −, middle S, right +. Middle (S) to TB1-8 top, left (−) to TB1-4 top, right (+) empty; the jumper as shipped (USB 5V), Ex. PWR empty. Drawing: SDI12_Adapter_Wiring_2026-10-21 (.html master, .pdf to print). *Documents: the drawing; USB hub box drawing (name the block and the screw order).*
447. No more wire is ordered for now (Mike, 2026-10-21: "don't need more wire ordered at this time"). Held: the second 16/2 roll (Wire and Cable Buy List A9) and the second 14/2 roll for the backup battery board (A8); the wing data cable needs none (item 446, the Cat6 on hand). The 16/2 case (Claude's reading): the six switched-load runs from the central node to the aft junction box (C2) are written as about 20 ft each, 120 ft in all, "which is why the second A9 roll is needed"; but the central node sits near the main bus board and the board to the stern is 10 to 12 ft (Mike, Release Node B's run), so each run is nearer 12 ft, about 72 ft in all (inferred, not measured). The wire size stands: 22 AWG at 12 ft and about 3 A still loses about 5 percent of 24 V on the Starlink pair, over the 3 percent rule, so the runs stay 16/2. Leftover 14/2 and 12/2 may stand in for 16/2. Buy a roll only when the one on hand runs out at fit-out. Also found in the same check: the "good to" column of the branch breaker table (Electronics plan 7.6) is the longest run each wire size allows, not a run length, so the wing box's "34 ft" is not a length. The tracker showed Mike's x (ordered) on the second 16/2 roll and on the wing data cable with no order mail for either: if either comes, it goes back. *Documents: Wire and Cable Buy List A8, A9, C2 and the totals; order tracker (done 2026-10-21).*
445. The wing cable enters the USB hub box by the 5-port split gland's free port b, not its own PG9 gland (Mike, 2026-10-20: "why not run the wing cable through the PG25 gland open spot?"). Port b was freed when Iridium A moved to Pi A (item 444). The Altelix port seals on 0.15 to 0.30 in (4 to 7.6 mm) cable (Altelix's page), so the shielded Cat6 must measure inside that: most shielded Cat6 does (Claude's reading, not measured; the cable is not yet picked, item 403's Monoprice Cat6A S/FTP or any on hand). The cable lays into the slit from the side, so its bare end needs no threading; it runs down inside the box to TB2. One fewer hole (the bottom wall since item 447: split gland B and the 24 V PG9). Drawn in USB_Hub_Box_Wiring_2026-10-20 the same day. *Documents: Electronics plan A1a (wall entries) and A17 (the cable's hub box end); Box Wiring Sheets A1a and A14; Wire and Cable Buy List C8 (cable at most 0.30 in across).*
444. Iridium A's USB cable goes to Pi A's second USB port, not the hub (Mike, 2026-10-20: "yes", to Claude's offer once item 443 freed the port). The RockBLOCK cable's USB plug goes into its DSD TECH SH-G01A USB isolator inside the Pi A box, and the isolator into Pi A's second USB port, so a hung hub can no longer cut off the main satellite link. The cable enters the Pi A box by its existing 4-port split gland (entries S and 7 share it: the hub's lead and Iridium A's cable, two ports plugged); the isolator is tied down and staked by the USB rule. The hub box no longer holds Iridium A's isolator, and its split gland port b is plugged. Settles item 436 (c). Drawn the same day: USB_Hub_Box_Wiring_2026-10-20 and Iridium_to_Pi_Wiring_2026-10-15 (the "at the Pi" side). *Documents: Electronics plan A1 (entry 7 now Iridium A's USB cable; contents: the isolator), A1a (Iridium A lines go), 6.5; Compute Bay 4.1, 14 (Iridium USB links); System Settings Sheet sec 9 (Iridium A on a Pi A port); Box Wiring Sheets A1, A1a, A11; Glossary (USB hub port list).*
443. The wing's CAN adapter plugs into the USB hub, not Pi A's second USB port (Mike, 2026-10-20: "plug into hub", after Claude explained the Oct 7 reason and advised keeping it on Pi A). Reverses that part of the 2026-10-07 wing readers decision (item 386): the DSD TECH SH-C30G stays in the hub box and takes a hub port, so the trim actuator's control now depends on the hub like the wing's two readers already did. What follows: Pi A's second USB port is freed (Iridium A takes it, item 444); the hub box's 5-port split gland port b is freed (plugged since item 444); hub ports in use eight of ten (SDI-12, SBE 45, DO sensor, wing reader, CAN adapter, dAISy, SBMS0, the MOTU M2's lead), two spare. Drawn in USB_Hub_Box_Wiring_2026-10-20 the same day. *Documents: Electronics plan A1 (entries S and 7, the split gland), A1a (the CAN adapter line, "never a hub port" goes; split gland ports; hub port list); Compute Bay 4.1 (Pi A's two USB ports); Design Decisions R-8.1H; Wing Power System 8.5 to 8.7; Wing Trim Actuator Installation Guide sec 3; Box Wiring Sheets A1 and A1a; Glossary (USB hub).*
442. The MOTU M2 sits in its own box, not in the USB hub box (Mike, 2026-10-20: "we already decided MOTO in its own box"); and the hub box wiring drawing. The earlier decision was never written down: every guide still said "Open: where the MOTU M2 sits", so the first draft of the drawing put it inside the hub box. Now settled: the MOTU M2 has its own box; only its USB lead comes into the hub box, by port e of the Altelix 5-port split gland, to a hub port. Iridium A's USB cable no longer comes into the hub box at all (item 444). The MOTU's own box is not picked and not drawn. Drawing: USB_Hub_Box_Wiring_2026-10-20 (.html master, .pdf to print; the v10.2 drawing of 2026-10-14 went to Archive): page 1 where the parts sit (not to scale, Mike: "don't need to scale"), page 2 the wiring, page 3 every wire from and to. New in it, Claude's layout, standing unless Mike says otherwise: (a) two DIN terminal strips on one short rail with the Waveshare reader: TB1 (24 V in, the YRDZXG's leads, 12 V and SDI-12 out) and TB2 (the wing cable: tip meter on 1 and 2, compass on 3 and 4, grounds on 5 to 7 bridged, CAN on 8 and 9); (b) the BaroVUE 10 and HygroVUE 10 each in their own PG7 gland, joined on TB1 (one cable per gland); wire colors from Campbell's manuals (BaroVUE red 12 V, black ground, white SDI-12, blue and yellow RS-232 cut and taped; HygroVUE brown 12 V, black ground, white SDI-12, clear shield to 0 V); (c) the Waveshare's channel 1 for the tip meter, channel 2 for the compass. Still open: the hub box itself (not picked; about 10 x 8 x 4 in or bigger). The surge diode mismatch with Box_Parts_With_Photos_2026-10-15 is settled by item 446 (the diode goes in). *Documents: Electronics plan A1a (MOTU line and its "Open", the second split gland, the glands list); Box Wiring Sheet A1a (same); Science Payload Strategy and Compute Bay 4.4 (where the MOTU sits); Box Parts With Photos.*
400. Order tracker rows still to settle (2026-10-08; brought up to date 2026-10-19). The BOM's status words are up to the tracker at v10.3. Still open: the Waveshare 7-port marked ordered Oct 7 has no order mail (cancel it if it was placed); the fuse links row prices nine, against 18 tap links plus 4 at the shunts; Reference/Left_To_Buy_2026-09-26 still says about $4,580; the tracker's puller fuse holder row links B0DSW8RTTP, the cancelled UK5 6-pack's listing, while the holder bought is Amazon order 114-6285250-6767424 (listing not logged). *Documents: the order tracker; Left to Buy; BOM 15.4, 15.7.*
384. Lewmar hatch: one check left. The Lewmar Medium Profile size 60 is written in at v10.2 (Build Guide 3.2, 4.4, 11.2; Design Decisions R-3.2; BOM 15.2). Its cut-out is 20 x 20 in, where the guides sized the opening at 24 x 30 in: check nothing that must come out at refit is bigger than 20 in square (Open line, Build Guide 3.2). Its weight is not known, so the Design Decisions R-1.3 weight table still carries the cover's 18 lb. Of the 30 3/8-16 x 2 in hatch bolts on hand, twelve are rig base edge bolts (Mike, 2026-10-12); the other eighteen are the rig base's outer-angle edge bolts too, with six more to buy (settled 2026-10-19, item 439 (l)). *Documents: Build Guide 3.2; Design Decisions R-1.3.*
8a. Drawings for shapes already decided. The wing can's drawings are on the welder's sheet (item 343); the 2026-09-24 concept drawing went to Archive 2026-10-08 (it still showed a base ring and a can-wall bolt). Its old blanks: the pipe cut length (the Spec Sheet formula should take off 1/4 in, not the whole 3/4 in lip, because the lip slips over the pipe; inferred from the drawing, check before cutting) and the labyrinth edge. Still stale: Rev H wing sheets 2 and 6 (the Wing Trim guide no longer points to them; the pipe cut-length note is in the Spec Sheet since v10.0). *Documents: the wing drawing set, Wing Can Spec Sheet.*
276. Cord puller mount: Mike is making a new mount (2026-09-26), neither of the two drawn. Log its shape, how it is fixed and where the hull hole goes once it is made (Open line in Transom Drogue Box Guide 6 and Design Decisions R-3.6C). The motor screws are written in at v10.2 (M4 pan-head 316 with flat washers, 11/64 in holes, no countersink, blue thread locker); the tracker shows only black 304 M4 x 10 mm screws and an alloy kit, so check the shelf for 316 pan-heads before the motors go on. *Documents: Transom Drogue Box Guide 6, Build Guide 7.3, Reference/Cord_Puller_Mount_2026-09-24.md.*
335. Motor mount: what is left (written into Build Guide 3.5 at v10.1). v10.1 carries the measured shaft (8-3/8 in, so the L-110 and the Lasdrop fit), the face plate with two feet, the plywood trial fit (Reference/Motor_Mount_Trial_Fit_Guide_2026-09-30), the L-110 figures and the M16 x 1.5 coolant ports. Still open: (a) Mike's trial fit measurements (A to G), then Claude draws the cut file; (b) the plate metal, aluminum or stainless, and the V-block's material (3/4 in okoume with a cast epoxy V is where Mike's notes leaned); (c) the height check before drawing metal: the notes' numbers say the beds rise above the shaft toward the front, so rails set from the back would run into the bed at the front; the straight bar clamped on the shaft settles it, and the stack must count the rubber under the foot; (d) the coolant barbs: moot, the loop is dropped and the ports plugged (item 466, 2026-10-29); (e) the idea in Mike's handoff notes of a pillow block bridge (ledges glassed to the valley walls, a bar across, the block on a rubber pad) is in no guide; Mike to say if it stays. Also at the shaft: the keyway (a machine shop cuts one if none), straight or tapered end, and the cracked bed near its aft end (open, probe, fill before the rail goes on). *Documents: Build Guide 3.5, BOM 15.5, Design Decisions R-3.5G.*
346. Longer prop shaft: ordered (Mike, 2026-10-01), what is left. The 31 in shaft is written into Build Guide 3.5 and 5.1, BOM 15.5 and Design Decisions R-3.5, and is on the order tracker as ordered; seller, price and due date not logged. At the first fit, check the prop hub clears the rudder's front edge with the rudder hard over each way. Claude's opinion, Mike's call: put the pillow block just behind the Lovejoy, not right behind the seal, so the shaft hangs as little as possible past its last bearing. *Documents: Build Guide 3.5; BOM 15.5.*
352 and 353. What is left of the two mismatch lists of 2026-10-02 (after v10.2). v10.2 settles most of them; these are still open, each needing a decision or a look: (a) the release feed's FHAC0002ZXJ fuse holder (Compute Bay 10.3, Electronics plan A15) may be redundant now the feeder is on a 30 A breaker; (b) settled and checked at v10.3 (Pi B on the backup battery board's 5 A breaker, no fuse in its box; no hub box fuse); (c) Pi B's wall: entry 9 is a metal SMA bulkhead, against "nothing conductive touches the wall" (Compute Bay 7.3); (d) the battery trip "isolates the bank", but Electronics plan 7.6 has no contactor or disconnect besides the manual switch: what does the isolating is not written; (e) the release bus readback: Compute Bay 10.2 and 11 give an expander pin to it, the Build Guide says no release-bus readback, and Electronics plan 4.1 has the opto board read only the Water Witch states; (f) WD-B "puts the full data on the bus" (Compute Bay 13.1), but WD-B has no bus link (likely fix: WD-B hands the strike log to Pi B over its serial line); (g) the hub box and wing box bar studs: the v10.0super closed list says the branch bar's end stud, the Electronics plan 7.6 table and the 2026-10-06 drawing say studs 8 and 3 (Open lines in A1a and A17; Mike's word on the table closes both); (h) how Starlink's Cat6 and power pair plug into the Mini, and its power wire, 14/2 in Wire List A8 against one of the 16/2 runs in C2; (i) where the HygroVUE mounts and the RockBLOCK's serial logic level. *Documents: Compute Bay 4.2, 7.3, 10.2, 10.3, 11, 13.1; Electronics plan 4.1, 7.6, A1a, A10, A15, A17; Wire and Cable Buy List A8, C2; BOM 15.4.*
436. Left over from the v10.2a write-in (Claude, 2026-10-18). Each is an Open line in the guides or a mismatch the write-in found. (a) The 5 V surge diode on the lightning detector lead: not picked (Compute Bay, Electronics plan A12, BOM). (b) The software must accept signed commands over Starlink as well as Iridium; the Starlink rules count on it (Build Guide sec 8, System Settings Sheet). (c) Iridium A's USB cable goes into the USB hub, so a dead hub also cuts Iridium A, and no guide says the hub stays on in safe mode: needs a line in Compute Bay and the Electronics plan once settled (Claude's view: Iridium A into one of Pi A's own two ports would avoid it, but both carry the hub and the CAN adapter). (d) The two split gland choices for the Iridium cables (Electronics plan A2 and A1a). (e) and (f) settled and written in at v10.3. (g) The dated wet lab box drawing (Wet_Lab_Box_Layout_2026-10-15) still shows a vent and puts the surge diode where Reference/Wet_Lab_Box_2026-10-14 does not. (h) Mismatches found, not settled: Build Guide puts Primelink B on the emergency bank, sec 4.3 and the Electronics plan do not; Electronics plan A3 entry 22 may break one cable per gland; the UV-C boards are $14.80 in the BOM and $37.10 on the tracker. Detail: Archive/_v102a_work/reports/A.md to F.md. *Documents: as named.*
438. Left over from the v10.3 write-in (Claude, 2026-10-19). (a) The slip ring brush bracket cannot be short: the post's top is about 10 to 11 in below the mast plate and the brush rides the 3 in of drive tube just under the drive bar, about 7 to 10 in above the post's top, so a bar from the post's side must climb past the ring's tabs and the 6 in drip cup and come back in to the tube: about a 10 in stick-out arm of 1/4 x 1 in bar, which may flex under the brush's spring (Claude's figures from the heights in Wing Power System 1A; the brush also sits 3 to 6 in above the hand-holes' top edge). Its length, bend and thickness go on the can's build drawing (Open line in the Wing Can Spec Sheet). (b) Settled at the v10.3 check: the wing box's 5 A fuse is on the box's short DIN rail in every guide. (c) Settled (Mike, 2026-10-19, "yes"): the wing counts in the daily total; see the Closed at v10.3 list. (d) Settled at the v10.3 check. Detail: _v103_work/reports/A.md to D.md. *Documents: as named.*
439. Left over from the v10.3 accuracy check (Claude, 2026-10-19). The check fixed about 290 mismatches in the guides; these need a number, a look or a measurement first. (a) The always-on network switch draws about 72 Wh a day (Electronics plan 6.1), but the Build Guide 4.6 hotel table has no line for it, and nothing says the 160 Wh "Pi CM5 + sensors" line holds it; if it is outside, the hotel load is about 915 Wh a day, not 845, and secs 4.6 and 4.7 move. Finding where the 160 came from settles it. (b) Build Guide 4.4's panel stations are shorter than the panels: the stern station is 60 in for a 62.2 in panel, and the middle station 67.2 in for three 24.1 in panels crosswise; from the pylon flanks to the post the panels fill the run butted end to end, with no gap for the edge fillets. Re-lay the stations from the panel sizes with the four deck checks already open for 4.4. Reference/Deck_Solar_Panels_and_Mounting_2026-09-26 also says "16 panels" against its own list of 14. (c) Build Guide 5.3 gives 850 to 900 rpm and 4.16 kt at 2 kW, but Design Decisions R-3.5D and the prop pitch study say the motor runs out of voltage near 900 rpm on the 25.6 V bus, where the 12 x 10 prop takes about 1,720 W; the lake power sweep (Build Guide 11.2) settles the 2 kW rows and the lee-shore figures. (d) Build Guide 1.4 and Design Decisions R-1.4 say the conversion costs about 0.12 to 0.14 kt, but the per-100 lb rate they give puts the added weight at about 0.18 to 0.35 kt: say which weight the figure is for and re-figure. (e) Design Decisions R-1.3's drogue row says 95 lb but its parts add to 99; recount it from the parts as built (the R-1.3 totals and Build Guide 1.3 move with it). (f) Compute Bay 5.3's backup island rows do not multiply out: Pi B at 3 to 3.5 W for 8 percent of a day is about 6 to 7 Wh, not 10 to 15, and the rows add to about 23 to 31, not 25 to 32; the working behind them settles it. (g) Build Guide 3.1 counts the three pylon bulkheads among the watertight ones; supported from the keel area with okoume, they may not close off the hull past the battery box; the bulkhead drawing of 2026-09-15 settles it. (h) A third shaft collar anode came with the Piranha cutter (order tracker; same parcel, no delivery mail): when the parcel is found, the Build Guide 3.5 and Design Decisions R-3.5E anode counts follow the BOM. (i) The post's copper run is given as about 7 ft, two stock sticks (Lightning Guide 5.3 and 9.1, Instrument Post Guide 12.2, BOM 15.2), but by the guides' own stations it is about 11 to 14 ft (Claude's figure); add it to the bar-run measuring in the moved-out lightning questions, and do not cut to 7 ft. (j) Iridium chain A's arrestor lead: the deck stub and the post's deck terminal are about 2 ft or more apart, not a few inches (Open lines in Lightning Guide 7.1 and Instrument Post Guide 12.3); the post guide's own rule puts each arrestor at its own hull entry, so measure the lead at fit-out. Instrument Post Guide 11.1 also puts the deck terminal "aft of the boot", against "port side" in 12.2. (k) The horn on the post is not on the Lightning Guide's bonding list (sec 6.2); whether its body takes a spur depends on whether the maker ties the body to a 24 V lead. (l) Settled 2026-10-19 without a measurement: the stack through the outer angle's 1/4 in stainless leg is 1-1/2 in, so the 3/8-16 x 2 in bolts fit; all 36 edge bolts are that bolt, 30 on hand and 6 more to buy (Bolt Depot 16131, tracker Buy list); Rig Base sec 5, BOM 15.2. (m) Where the Starlink and GPS antenna mounts spur to the copper (Open line in Lightning Guide 6.2 item 9): none is on the post; pick the mounts first, and never spur GPS B's mount to the copper through its coax shield. (n) Wing trim's step on the load-shed ladder is not named (Open line in the Settings Sheet sec 4). (o) Boat 1 Route sec 2's Leg 1A motor levels and the Routing Guide's "a fast passage waits north up to about three weeks" still follow the 1,675 W array: fold them into the deck solar change rework (item 313). (p) Barometer and Pressure Port 3.4 figures the arm load at 32 in; the arm is at 44 in, which roughly makes its share about 13 percent of the post load (Claude's figure; the order of failure holds); re-run it at the real height (Open line there). (q) The post's 5 x 5 in head plate overhangs the 4 x 4 in tube by 1/2 in a side, and the air terminal stack on the port face is about 3/8 in thick, so a centered plate's edge meets the bar: draw how they clear. (r) The Trim Actuator Guide buys four rod-end clevises (two on the link, two for the pushrod ends); the order tracker's Buy list row has two; settle the count when the link and pushrod ends are drawn. (s) The Kamoer dosing pumps are still "arriving October 2 to 14" in the Science Payload Strategy and Wet Lab guides; the tracker has them on the way with no delivery mail: check the shelf. (t) Item 2's displacement range (5,570 to 5,775 lb) is a pound or two behind the Build Guide's 5,780 (secs 1.3, 6.9 and 11.2 now agree). *Documents: as named.*
460. Build Guide sec 6.5's upwind limit (60 to 70 deg true) looks optimistic; check at sea (Claude, 2026-10-27, from Mike's "how tight to the wind"). A worked estimate that adds the hull's windage and the keel's sideways slip puts the wing-only limit near a beam reach: heading about 80 deg off the true wind, real path about 90 deg (80 to 100), the same from 5 to 30 knots, flat water. Not measured; the hull windage and keel grip are estimates. At sea trials, log heading, track and true wind with the motor off at the closest heading the wing holds; then fix the sentence at the next doc set and tell the route planner the real figure. Detail: Reference/Wing_Upwind_Estimate_2026-10-27.md. *Documents: Build Guide sec 6.5; Routing Guide.*
419. Gaps found drawing the VESC A box wiring (Claude, 2026-10-14). (a) No doc names the motor temperature pins: drawn on the Flipsky sensor plug's TEMP and GND, the usual VESC place (inferred). (b) The 2 AWG feed: Electronics plan A7 and the 2026-09-26 capacitor drawing say 5 to 6 ft; Reference/Main_Bus_Board_Drawing_2026-10-06 says about 10 ft. The routed length settles it; the Wire Buy List's 20 ft of each covers either. (c) One COMM ground pin serves both the UART backup and the motor sensor; the docs say the pigtail splits but not how (the JST-PH wires are about 26 to 28 AWG, below a WAGO 221's 24 AWG floor). (d) How the motor's phase lugs meet the Flipsky's phase outputs is not written. Drawing: VESC_A_Box_Wiring_2026-10-14.
413. Gross tonnage: over 4, where the Coast Guard says it starts to matter (Mike, 2026-10-14). The USCG asked Mike the boat's gross tonnage; he did not know, and they said it only matters if it is above 4. No guide gives it. Worked by the USCG simplified rule for boats under 79 ft (46 CFR 69.203 and 69.209): gross tons = 0.5 x length x beam x depth / 100 for a sailing boat (the wing makes it one; a powerboat uses 0.67, about a third more). Length 25 ft, beam 8 ft (Build Guide 1.1). Depth is from the top edge of the hull side amidships down to the bottom of the keel; on a faired full keel the keel counts and only 75% of that depth is used. Masts, the wing, cabins and deckhouses do not count, and loading does not change it. So for this hull, gross tons = 0.75 x depth in feet, and depth = the 3.5 ft draft + the hull side height above the water amidships. Under 4 gross tons needs depth under 5 ft 4 in, so a side height under about 1 ft 10 in. Mike, 2026-10-14: the side height is more than 1 ft 10 in, so the boat is over 4 gross tons and the USCG's extra paperwork applies (about 4.1 at a 2 ft side, 4.5 at 2 ft 6 in, 4.9 at 3 ft). Do not try to get it under 4: only cutting the hull sides down would, at a cost in height above the water and self-righting. Net tons are 90% of gross (sailing, motor in the hull), about 4, under the 5 net tons needed for federal papers (46 USC 12102), so the boat can only be state registered under US rules, which bears on the flag choice in the rules and permits item (1B list below). Same for both boats: Boat 2's bow armor does not change the outside size. Still open: what the extra paperwork above 4 gross tons is (not found in the federal rules), and the real side height amidships (deck edge to waterline at design load, or deck edge to keel bottom directly) for Form CG-5397; the numbers given must be the real hull's. Written in at v10.3: the rule and the result in Build Guide 1.1, pointers in 11.5 and Boat 1 Route sec 4, the side height and the paperwork as Open lines. *Documents: Build Guide 1.1 (add the side height and the tonnage figure once known) and 11.5; Boat 1 Route sec 4.*
401. Numbers and small gaps found at v10.2, nothing decided (2026-10-08). (a) The A1 and A2 IMU disagreement limit has no number (System Settings Sheet sec 7, TO BE SET). (b) Done at v10.3: the wing CAN link at 250 kbit/s (Settings Sheet sec 9). (c) The SBMS0's serial speed is set nowhere (Open line in the Battery Box guide and the Settings Sheet; from item 363). (d) The JK balancer's 1.2 W ceiling (about 29 Wh a day at most) is not in the Build Guide 4.6 hotel table. (e) Done at v10.3: the wing compass at about 23 Wh a day (Wing Power System 5, Build Guide 4.6). (f) Electronics plan 7.6's 100 A MEGA sum still books about 1.5 A to the hub box; the hub now draws from Pi A's line (the total barely moves). (g) The hub box's SMA bulkhead likely needs a short SMA male-to-male lead inside to reach the dAISy's socket (inferred; no doc names one). (h) Overtaken: each tap cable has its own gland (Mike, 2026-10-13; Battery Box guide), so there is no bundle gland. (i) Which way the switch box's right-angle patch cords must point: read it off the switch before ordering (Electronics plan A6a). (j) Design Decisions R-8.3A and Instrument Post Guide sec 10.4 still say motion correction comes off "IMU A"; Pi A now has A1 and A2. (k) Electronics plan sec 8 says two of three 8-channel opto boards are spare; the tracker has five Fafeicy boards on hand. (l) The Lightning Guide's bar run list totals about 40 ft; item 364 estimated about 45 ft against 48 ft on hand: the run measurement (item 309) settles it. (m) The Transom Drogue Box Guide's order list still names 3/8 in 316 thimbles ordered 2026-09-26; those went back Oct 1, and a US Stainless 316 5/16 in part came Oct 7 (probably thimbles, Claude's reading): check it is thimbles and fits the bridle rope, then fix the guide. (n) IMU placement: "one at each end of the plate" (IMU A1 and A2) and "the converter at the far end of the plate from the IMUs" cannot both hold; the Box Wiring Sheets draw A1 and A2 at the two corners of one end, the converter at the other; Compute Bay 4.1 and Electronics plan A1 should say which. (o) Battery Box Part 8 gives the farthest tap as about 35 in (Mike); the tap drawing draws W4 at about 42 in. (p) Inline fuse holders: nine on hand and seven used at the bus board, while v10.2 adds at least five more inline fuses (Pi A's 3 A hub lead, the hub box's 24 V, the wing box's 5 A, the tip's 1 A, the horn's 5 A) plus one per post feed (Claude's count): count the shelf before the loom. (q) The one 1/8 in G10 sheet on record (13.2 x 11.8 in) makes one 11.2 x 7.6 in compute box plate, not two (inferred). (r) Lightning Guide 7.7 asks for a TVS on Pi B's lightning detector lead; no part is named, and the 1.5KE33CA (31 V) does nothing on a 3.3 V line. (s) Electronics plan A2 entry 5 puts an optocoupler in Pi B's box; Compute Bay 4.1 runs the alive line through Pi A's PC817 board: whether a second board goes in Pi B is not written. (t) Box Wiring Sheets A7 draws a TVS on the VESC A feed; the Electronics plan's A7 text names none. (u) The MOTU M2 is on a hub port (written 2026-10-09), but where it sits and how its cable reaches the hub box are not written. (v) The bus board drawings put the battery switch ahead of the battery shunt (from Battery Box Part 8's wording); no guide says which comes first. The charge-stop 1k resistor and the remote on/off terminal strip belong on the bus board, but no guide says which bar or stud they land on, so they are not drawn. *Documents: as named.*
404. Safe mode: points for Ken (software repo, 2026-10-08). (a) The guides say safe mode runs "its IMUs"; Ken's decision said IMU A, which is now A1 and A2: confirm. (b) The command signatures' sequence counter is kept on the SSD (Build Guide 8.3), which safe mode does not use: how the replay check works in safe mode is not written. (c) Still to prove on a real CM5: that its startup reads the header pin to pick the safe copy (Compute Bay 4.3, Open line). *Documents: Compute Bay 4.1, 4.3; Build Guide 8.3; software repo.*
320. Wet lab and camera port: what is left after v10.1. v10.1 writes in: no drip-tray leak sensor; the 360 degree foredeck fisheye (Hanwha XNF-8010RV; bought 2026-09-30 open box on eBay, $132, written into the BOM, Electronics and Wiring Plan, Design Decisions and System Settings the same day); the Iwaki bought; no paint on the de-bubbler; the simple reducing-tee de-bubbler with the Rain Bird drip emitter bleed; the flow meter's 3 to 24 V; the SBE 45 drainback sentence; and the camera port's UV-C as layout H (small 275 nm LEDs on a flat mask above and below the lens; the Seoul stars do not fit and are not bought). Still open: (a) where the SBE 45 goes, above or below the waterline (Mike's call; the claim that Sea-Bird wants it above is not in the SBE 45 manual; below, it never drains; the limit is room: 13.3 in tall with 6 in clear below and 12 in above); (b) the biocide reservoir: settled, the filter housing was ordered 2026-09-30 and is written in; (c) the camera port: three LEDs were bought (one eBay order, 20-15209-84182, 3 on 10 mm boards, $14.80, here 2026-10-09) against four in layout H; the mask and how three sit are drawn when they arrive; test each with the QuantaDose card first; the factory ring board (four Marktech MTSM280UV1-C3510-24) stays the fallback if they fall short. *Documents: Flow-Through Wet Lab Design 6 and 8; Underwater Camera Port Guide; BOM 15.8 and 15.11.*
319. The wet lab's 24 V feed: one check left. The steady 24 V converter (XWST, bought) is written in at v10.1. Open: where the flow meter's pulse wire lands, since its pulse now swings to 24 V. *Documents: Flow-Through Wet Lab Design 4 and 12; Electronics and Wiring Plan 7.6.*
2. Displacement margin. Estimates run 5,570 to 5,775 lb against the 5,500 lb basis (re-summed at v10.0). The Build Guide 11.2 weighing is the gate. Design Decisions R-1.3's weight table was re-summed at v10.0 with the can at 110 lb (Boat 1 about 5,775 lb; the range 5,570 to 5,775). Two of the three estimates sit past the 5,600 lb re-check line, so expect the weighing to matter. For Boat 1 alone, which leaves out the Cape Horn package, the strip and bottom-up estimates are about 5,435 and 5,505 (Build Guide 1.3), so only the itemized 5,775 is past it; the quoted range 5,570 to 5,775 mixes the two boats. *Documents: Build Guide 1.3 and 11.2, Design Decisions R-1.3.*
22. Heading wander. Log one hour of compass heading under self-steering once the boat floats. It checks the steering deadband (how far heading may wander before the rudder moves), which item 92q left as it is. *Documents: Steering System Handbook 10, System Settings Sheet 3.*
4. Camera head against the tube. Measure the Axis P1245 head against the 1-1/4 in bore of the G-10 tube before the tube is glued in. Aim check at 38 degrees. *Documents: Underwater Camera Port Guide 5 and 6.*
61. Bare shaft for the seal. Now part of item 173's haul-out list. The Lasdrop DrySeal D-100 is on hand; polish the shaft where its lip rides. *Documents: Build Guide 3.5, BOM 15.5.*
17. Release trunk core count. Count at loom time. Fire pairs are 12 to 14 AWG. The four cord pullers now add power cores. *Documents: Electronics plan A3 entries 3 and 4, Wire and Cable Buy List C3.*
87. Aft junction box size. Planned at 12 x 8 x 5 in with 10 RJ45 couplers and about 21 entries. It now also holds the post wind meter's 5 V regulator (item 194). Measure before the loom; it may need a bigger box. (2026-10-30: the couplers moved to the switch box, so about 14 entries; the build sheet of item 467 needs a plate about 10 x 6 in and a base at least 3-1/2 in deep for its lower gland row.) *Documents: Electronics plan A6, Compute Bay 4.4.*
110. Antenna bundle through the post stub. Pass the real bundle (GPS, Starlink and Iridium A coax, plus the post cables) through a sample stub before the real stub is glassed in. Cable ends go through bare; connectors are made up after. If it will not pass, say so before the deck is closed. *Documents: Instrument Post Guide 8.*
126. Count the shelf before buying. The 2 AWG roll length (68 ft needed). The TVS bag: 1.5KE33CA surge diodes or 1N5408 flybacks. IP68 glands (about 120 needed, 50 on record). M12 vent plugs (one: the central node only, item 434). ADuM1201 isolators (three ordered, two delivered). Tylaska T8 shackles (four for Boat 1; the order looks like eight). Lead: 14.4 lb on hand; how much the wing nose needs now comes from the Stage 10 measurement, since the packs, pod and panels have left the wing. Then strike the Wire list buy lines that Mike's 2026-09-21 notes put on the shelf (NMEA 2000 cords, Cat6, 6 AWG, serial cables). *Documents: Wire and Cable Buy List, BOM, Electronics plan App B.*
29. Duckbill cracking pressure. At the Battery Box Part 10 check. Grease the lips first, then read light breath pressure on a clear vinyl tube taped to a yardstick. *Documents: Battery Box Building Guide Part 10.*
94. Compliance link preload. 4 lbf to start, tuned at sea from the link-open alert. Write the fit-out numbers in the wing records. *Documents: Wing Construction Guide records, Wing Trim Actuator Guide.*
128. Prove both carrier slots. Fit the 4 TB NVMe drive and the Hailo-8L together on the carrier (the second carrier is on hand) and run both, before the wiring is committed. *Documents: Build Guide 8.1, Compute Bay 16.2.*
88. TSW202 bench test, five checks. Turn a PoE port off and on through the Web API; record the request and the firmware version. Read the same port over SNMP. Read per-port watts. Find out whether per-port power priority exists. Power all six devices (the five cameras and the light driver) at once and read the total. Pin the firmware. *Documents: Build Guide 11.1, Electronics plan.*
33. Nichrome bake-off. 28 AWG against 30 AWG, wet, at 24 V. The winner goes in Transom Drogue Box Guide sec 9. *Documents: Transom Drogue Box Guide 9, Build Guide 11.1.*
127. Iridium modem bench test. One of the two jobs that gate the wiring. The other is the Flipsky bench check (item 254). *Documents: Electronics plan, Compute Bay, Build Guide 11.1.*
172. Motor wiring: what is left. The sensor question is settled in Build Guide 5.1 (the 75200 Pro reads sin and cos on ADC1 and ADC2; a divider unless the bench check in item 254 says not). Left: size extra bus capacitance at the controller with the cold plate work, and measure the three phase leads before siting the controller. The controller is here (2026-10-07). *Documents: Build Guide 5.1, Electronics plan A7, BOM 15.5.*
173. The shaft run, at haul-out. Mike measured 8-3/8 in of bare shaft forward of the stern tube (2026-09-29), so the run fits with the L-110 and the Lasdrop; nothing shorter is needed on Boat 1 (written in at v10.1). Left: confirm with calipers that the coupling end is straight, not tapered; check for a keyway; and check Design Decisions' PYI PSS figure (7 in of bare shaft) against the study's 4.7 in. *Documents: Build Guide 3.5, Design Decisions R-3.5C.*
86b. Measure for the three hull-glued boxes. The hull laminate thickness at each spot, and the size of each plate: Pi A, the motor controller plate, the network switch (the motor cooler plate is gone: item 466, 2026-10-29). Taken at the boat. *Documents: Compute Bay 2, Build Guide 5.2.*
95. Desiccant in Pi A. Whether Pi A also needs a desiccant pack once its back sits on the cold hull. Settle it with the 86b measurements. *Documents: Compute Bay 2.*
167. Measure the tiller parts. Poke the wood hard with an awl first. Then measure the tongue's height above the casting and whether it has room for a second hole, the hinge pin diameter, the tiller's width and depth at the 13 in mark (the actuator station) (for the plates), the pin's height above the cockpit floor (sets the base bracket height), and the actuator's eye and pin (Build Guide 3.8 says a reamed 7/16 in hole). *Documents: Steering System Handbook 2.4, 3 and 4; Build Guide 3.8.*
168. Hydrophone fin station. Mark the station on the hull now (starboard, middle third of the waterline length, 12 in above the bottom of the keel) so no stand or sling strap lands on it; fix it after the keel measurements. Before cutting, check from inside where the gland's hole comes out: at 12 in above the keel bottom it may open into the lead or its cap; raise the fin if so (Science Payload Strategy 5). The gland (Jacob WADI K257-1012-00) is bought and written in at v10.2; on arrival read its stainless grade (316 or not) and its thread length. *Documents: Bottom Painting Guide Step 0, Science Payload Strategy 5.*
214. Drogue cord pullers: bench one, then write how they work. No document says how a puller is switched (one direction or reversing), how it stops (a set run time or stall current), or what it really pulls on its 5/16 in pulley against the 30+ lb the Build Guide asks. The documents also disagree on the cord's path: Build Guide 7.3 and Routing Guide 12.4 run it through a sealed conduit glued through the transom; BOM 15.7 says no conduits and no transom glands. Bench one motor, time a full wind, measure the pull, then write it in. The stop is settled by item 301: a 64 s run and a fuse per puller. A mount plan is written (2026-09-24, Reference/Cord_Puller_Mount_2026-09-24.md): the motor on a 3 in piece of 2 x 2 in aluminum angle screwed to the free end of the G10 pad, pulley toward the U-bolts, cord straight down under the pulley edge. Release line, decided by Mike 2026-09-24: the motor pulls the blue 2 mm Dyneema line already on each T8's trigger; pulling it releases the shackle. No fid. The T8s were bought with no swivel. Hull passage, decided by Mike 2026-09-24: no rod. The line runs in a greased PTFE tube sized for it, 2 mm inside, 4 mm (5/32 in) outside, one 20 ft piece for all four, from near the T8, clipped along the outside of the hull, through the hull bedded in 3M 5200 (Mike: 5200, not a cable gland; 7/32 in hole, tube sanded, 5200 fillets both sides, clipped within 1 in of the hole both sides so it cannot work), up to the pulley; at least 12 in of tube outside the hull. Check first that the line pulls freely through 2 ft of the tube, bent 45 degrees (if it binds, 3 mm inside), and that each blue line is about 4 ft long (if short, a 2 mm Dyneema line replaces it). Test first: a scrap tube bedded in 5200 through scrap G10, cured a week, 10 psi of air with soapy water on the far side; 10 psi on a greased tube with the line in it; and the motor pulling the blue line through a full greased tube, at slack and with a realistic load on the shackle (Build Guide 11.1). An EPDM vacuum cap for 5/32 in line goes over the outside end of each tube, the line through a pinhole in it, whipped on behind a heat flare. Links: Left to Buy 2026-09-26 items 97 to 100. Written into the v10.0 guides (Build Guide 7.3 and 11.1, Transom Drogue Box Guide 6, Routing Guide 12.4, BOM 15.7). The mount choice is question 276; the at-sea exercise is question 277. Mike's alternative (2026-09-24), which I think is better: the same angle, 4 in long, through-bolted to the outside face of one of the glassed plywood triangles beside the pad, motor can up, 5200-bedded tube through bare hull beside the triangle (drawing Reference/Cord_Puller_Rib_Option_2026-09-24.png). Mike to pick one before anything is drilled. *Documents: Transom Drogue Box Guide 6, Build Guide 7.3 and 11.1, Electronics plan 4.3 and A16, Compute Bay 12, BOM 15.7.*
215. Victron remote cable length. The five remote on/off cables run from the central node to the five solar controllers, and no document records how far that is. Measure each run against the cable's own length; extend with 18/2 tinned wire if short (item 302). *Documents: Electronics plan 6.7 and A3 entry 22, Wire and Cable Buy List, Box Wiring Sheets.*
218. How the switch and the motor controller meet the hull. Item 86b glues the network switch and the motor controller plate to the hull for cooling. But the TSW202 sits inside the plastic aft junction box, and VESC A in its own sealed box on the cockpit web. Write how each reaches the hull: a metal plate through the box floor, or the box itself glued down (question 285 carries Claude's recommendation). The plates stay off the Lightning Guide's list of metal wired to the lightning copper (Mike, 2026-09-30; written into Lightning Guide 6.2). *Documents: Compute Bay 4.4, Electronics plan, Build Guide 5.2, Lightning Guide 6.2.*
219. Wet lab numbers to take on the bench. Loop volume: the Wet Lab uses 2.5 L for the dose and flush but also says about 1.2 L with two stations filled; measure it, then re-figure the 500 mg dose and the flush time. Time the fill and de-bubble at 1.5 L/min (still written as about 60 s). Set the restrictor coil length, (the uxcell PTFE tube went back 2026-10-02, so the coil tube is to pick). The two Taidacent parts bought 2026-09-07 are RS-12V/2S surge modules, not converters; the flow meter's supply is question 282. *Documents: Flow-Through Wet Lab Design 3, 8 and 10; Build Guide 12; BOM 15.11.*
254. Checks from the 2026-09-24 research, when the part is in hand. Each is written in its guide.
288. Arithmetic to re-check in Build Guide secs 4 to 6 (found in the v10.0 trim, 2026-09-26; each needs the working it came from, so nothing was changed). (1) Sec 4.6: the solar factor K is 0.72, but its parts (0.88 x 0.95 x 0.92 x 0.96) multiply to about 0.74; the harvest table uses 0.72. (2) Sec 5.3: the overall efficiency from electrical power to thrust is 0.37, but its listed parts (0.55 x 1.15 x 0.8 x 0.97 x 0.85) multiply to about 0.42; a factor may be missing (the controller?). The Routing Guide and Design Decisions R-5.3 use 0.37 too. (3) Sec 4.7 says conserve mode after the midpoint is 200 W; the sec 5.3 modes table lists "doldrums economy 250 W (conserve, post-midpoint)". (4) Sec 4.7's second day figure in each row (11.9 d and so on) is 90 percent of the bank used to empty, not the 90 to 10 percent the heading says: label it. (5) Sec 4.5: the 100/20 controller's clip is given on its input side (about 575 W) and the 75/15's on its output side (about 414 W): put both on one basis. (6) Sec 4.3 gives the emergency island 25 to 32 Wh a day (Compute Bay 5.3); the sec 4.6 hotel table carries 24 for the Pi B island: one figure, or say what each covers. (7) Sec 4.6 skips both the cutter sweep and a cutting campaign in HIBERNATE, STORM and below 30 percent charge; sec 5.3 says that only for the sweep: settle whether a campaign may run below 30 percent (a question for Mike once the working is checked). *Documents: Build Guide 4.3, 4.5, 4.6, 4.7, 5.3; Routing Guide 2.1 and 13; Design Decisions R-5.3.*
289. Bill of Materials figures to re-check (found in the v10.0 trim, 2026-09-26; nothing was changed, since each needs a receipt, an order or a count). (1) Cells: 36 at $100 is $3,600, but the line carries $3,439 paid. (2) Slewing rings: all three ($1,208) sit in Boat 1's wing cost, while sec 15.13 leaves Boat 2's ring out; about $403 of it is Boat 2's (question 287). (3) T8 shackles: $1,701 carried whole; if the box holds eight, about $851 is Boat 2's. (4) "VESC waterproof enclosures (primary + backup)", $80, but Boat 1 has one VESC. (5) The AS5048A angle sensor may be counted twice: a $60 wing pivot sensor in the sensor list, and one on hand in the wing pod line. (6) The nav lights line settled at the v10.3 check ($124: the attwoods went back); the surge parts line still does not add up from the prices written in it. (7) Settled: the Orion-Tr was $56.10 (order tracker). (8) The aluminum oxide grit mixed into the heat-carrying glue: the fine grit bought is 120; the spec is 220 or 320 (Build Guide 5.2). Settled 2026-10-26: the #120 (item 455). (9) The 2 AWG motor leads: BOM sec 15.4 says $41 to buy, the wire list A3 says none. (10) The G10 sheet stock ($719) is not marked bought, though the drogue pads are cut from the 3/8 in sheet. (11) The four SN65HVD230 CAN boards are on hand (bought 2026-08-24), but the Release Node B line still prices a pair (about $16) and the CAN line carries a $40 estimate. (12) The two BME280 boards came in the 2026-09-07 order; the working barometer line still carries a $15 estimate. (13) BOM 15.12's $12 of Belleville washers for the busbar and head-bus terminals looks like 15.2's second M6 pack ($7) counted again. *Documents: Bill of Materials; Wire and Cable Buy List A3.*
281. Drogue boxes, on arrival (not yet received, 2026-09-26). Count them (Boat 1 needs two). Magnet-test the body (steel: do not use it). Measure the wall thickness. Pack one drogue into a box before any metal is stripped, so an unstripped box can still go back; a packed drogue is probably 1,000 to 1,500 cu in against about 990 in the box (both guesses). If it does not fit: the next VEVOR size up (the range goes to 20 x 16 in), not a laid-up box. Size up in fiberglass only: Mike asked 2026-09-28 about the VEVOR 16 x 16 x 8 in STEEL enclosure on Amazon ($73.90); not for this, since painted steel rusts at every hinge and strap hole and cut edge, and epoxy holds poorly on paint. The fiberglass 16 x 16 x 8 in is at Home Depot (VEVOR DQXFSBLG40X40X201V0, $138.82 on 2026-09-25), about 1,400 cu in usable against about 990; Amazon has VEVOR fiberglass 19.7 x 15.7 x 7.9 in (B08ZYJ2QC4, price not checked; check two fit the transom). Transom width up high, where the boxes go, is not in any guide and not published (only the 8 ft beam); measure it before buying a box wider than 16 in. The v9.8 guide's two 22 in boxes suggest about 44 in was once thought to fit, but it was never measured. Boat 1's rudder hangs on the keel, so nothing on the transom is in the way except the quarter U-bolts and the cord puller tubes. The box on order is probably VEVOR's fiberglass 15.7 x 11.8 x 7.9 in (Amazon B091FG96KN). Mike, 2026-09-28: bolt the steel box to the transom and replace it at refits? Claude: it can work, with conditions, but fiberglass is still the better buy. Bolting fixes the glue problem. The risk is rust inside the box between refits: rust swells to several times the steel it eats, and the lid lives in a 1/8 in gap in that steel frame, so rust scale can lock the lid, the one failure that loses the drogue; rusty hole edges also saw the Dyneema hinge and strap. If steel: drill every hole first, then coat the whole box inside and out with epoxy barrier coat and bottom paint the mouth; make the lid fiberglass, not the steel door; bed the bolts and put plastic washers between steel and stainless. Costs against the $65 a box saved: that coating work, some of the hull's barrier coat (item 226 added one kit for the hull alone), and about 20 lb per steel box by Claude's figure (1.2 mm wall assumed, not checked), with the displacement margin already tight (item 2). Mike to decide. Start the Dacron tape salt-water soak now; it takes weeks. *Documents: Transom Drogue Box Guide 1, 3 and 8; Build Guide 3.6; BOM 15.7.* Ordered 2026-09-28 (mail): a VEVOR fiberglass enclosure 19.7 x 15.7 x 7.9 in ($74.01, due Oct 2), the size-up above, and a QILIPSU UL94-V0 16.4 in outdoor box ($82.15, due Sep 30), probably the second. The steel VEVOR of 2026-09-24 is not used.
259. Wing fold hinges: four White Water hinges, once one passes a pull test. Mike, 2026-09-26: the four he has (two pairs) should do for folding in calm air. Research, 2026-09-26, with Mike's photos (Reference/Fold_Hinge_Strength_2026-09-26.md): no maker rating exists; worked out from the 304 sheet, which Mike measured at 5/64 in (the listing says 1/16 in). Mounted with the feet on plate 1 aft of plate 3 and the flat leaves on plate 3's top, the rig's weight pushes straight down the upright legs: about 135 lb each against about 700 lb. The weak way is the tackle pushing the legs toward the bow while it holds the rig back: about 45 lb each with a 6 ft gin pole (about 65 lb with a 4 ft pole) against about 160 to 240 lb. So four will do with a gin pole, eased down smoothly. Test first: one hinge on scrap plates, mounted as it will sit, about 300 lb straight down on the barrel, then about 100 lb toward the bow side. If it bends, the welder makes the two heavy hinges with plate 3. *Documents: Build Guide 3.3.1 (fold steps: the hinges, the gin pole length), Rig Base Two Plate Design, BOM 15.6.*
309. Open questions moved out of the guides at v10.0super (2026-09-27). The guides no longer keep their own open lists; each question below came from one and names where. Where a guide still depends on it, it says "Open:" at that spot. Electronics, the computer boxes, settings and wire:
Rig base, wing can, battery box, drogues, steering and routing:
The wing and the slip ring:
Lightning, the post, barometer, science, wet lab and Boat 1's route:
310. Numbers that disagree, found at v10.0super (2026-09-27). Each needs a check, a receipt or Mike's word; none was changed.
313. What the deck solar change leaves open (2026-09-28). The deck array drops from 1,675 W on paper to about 1,400 W.
338. The lightning copper through the deck (the junction and the pickup below the deck are written in, 2026-10-01). Also open since the junction became a bolted crossing: where the post bar, the spark-gap bar, the battery negative wire and the far spurs land (for example taps drilled in the cross bar beside the crossing, as sec 6.1 does); the cross bar layout at the bulkhead settles it (Lightning Ground System Guide 5.3, Open line). The copper above the deck: decided 2026-10-01 (Mike: "drop it"), written in the same day (see the Closed 2026-10-01 list): no copper above the deck; the main bolts carry the strike and the trunk picks up a side bolt below the deck. Drawing: Reference/Item338_Copper_Pickup_2026-09-30.png. The junction, decided by Mike 2026-09-30: with the keel block gone (item 344) it has two branches. One bar runs across the bulkhead to both hull sides (two sticks lapped 4 in end to end, two bolts, as the post run), nearly level, 5 to 15 degrees downhill toward each side, then curving onto the hull face (8 in radius or more) to its exit bolt. The trunk comes straight down and crosses it face to face, and the crossing is held by ONE 1/4 in silicon bronze bolt through its middle with a wide washer each side, both faces sanded bright with No-Ox-Id; meter it with the battery-and-bulb check (Lightning Guide 8.1), near 1/1000 V at 2 A (inferred). No plate is needed (Mike has no 4 x 4 in copper). The near-square turn is allowed: NFPA 780's 90 degree and 8 in limits apply to bends in a bar, not to a bolted joint where current divides. Washers, not bought: stainless 1/4 x 1 in fender washers on Amazon (, price not checked; fine here, since the washer only clamps), or silicon bronze 1/4 x 1-1/4 in from Fair Wind ($0.79, 3/64 in thick, ) or Green Boat Stuff (about $1.10, 1/16 in thick, ). Also left (found 2026-10-01 while writing it in): (a) how the two 1/4 in pickup bolts are held (through the tripled middle bulkhead and both angles, or into holes tapped in the 1/4 in leg), their length, and which side bolt is used; (c) the bond wire count: Lightning Guide 3, 6.2 and 10.1 say eleven, one wire each, but the can and the rudder stock now get none, and the Wire and Cable Buy List A5 says ten; (d) the bar total: BOM 15.2 says about 45 ft, the Lightning Guide about 40 ft; the trunk now starts below the deck, so measure it; (e) Lightning Guide 8.2 meters the other bolted joints but not the new pickup joint; add it when the pickup is drawn; (f) done 2026-10-08: the concept drawing Wing_Can_Fabrication_2026-09-24 and its script are in Archive, and the bond stud note is gone from the plate 3 drawing. *Documents: Lightning Ground System Guide 5.2, 5.3, 6.2 item 1, 9.2; Rig Base Two Plate Design secs 3 and 5; Wing Can Spec Sheet; the shop sheet (item 343).*
341. Figures to re-work for the XXS mast (found at v10.1). v10.1 scaled the mast-break figures by about 1.24 and marked them "scaled"; these still need working: (1) Design Decisions R-3.3: the slewing ring bolt joint's margin over the mast's plastic hinge falls from 1.47x to about 1.19x (inferred, 1.47 x 22,400 / 27,700); (2) Design Decisions R-6.9 and Build Guide 6.9 keep the Schedule 160 shedding band (22,000 to 33,000 ft-lb) beside the XXS mast-break figure; (3) the plain 3/4 in can lip runs just over 304's yield at the band top, carried by the 16 lip gussets, not re-checked; (4) the stability figures (R-1.3) were worked with about 10 lb less aloft; the inclining test settles it. *Documents: Design Decisions R-1.3, R-3.3, R-6.9; Build Guide 6.9; Wing Can Spec Sheet.*
342. Steering checks left after v10.1. The port side, the 13 in station, the 16 in cut, the stop blocks just past the actuator's ends and the rear mount as built are written in. Left: whether the blade clears the hull and prop at 28-1/2 degrees (hold it there before the blocks are bolted down); whether the cockpit floor has room out to about 30 in to port; and redraw Reference/Steering_Actuator_Top_View_2026-09-28.pdf Section B, which still shows the old angle bracket, once the rear pin's height above the floor and its distance forward of and to port of the stock are measured. *Documents: Steering System Handbook 3 and 4.*
25. Dacian at Electrodacus: one optional question. What the SBMS0 reports with the XT1 probe unplugged. The Battery Box Part 10 bench check measures it anyway, so send a mail only if that reading is unclear. Item 166 made the other two questions moot. *Documents: Battery Box Building Guide Part 8, BMS Notes.*
8b. Two shops. The mast weldment has no fabricator. The wing can needs a shop quote once the Spec Sheet blanks are filled (item 8a). The metal for both is on hand and both mill certs are in (item 332, written into the guides 2026-09-29). Shop leads in the Tulsa area and every metal job by package: Reference/Fab_Cut_Weld_Work_List_2026-09-29.docx (2026-09-29). *Documents: Wing Can Spec Sheet, BOM 15.6.*
89. Deliveries.
Documents: the order tracker, BOM.
96. Returns, with their windows. Now kept in one place: the "Send back for refund" tab of BOM_v10_0_Order_Status.xlsx (built 2026-09-26 from the mail). First deadline: the four KSD301 discs, 2026-10-04; then the bare MOOKEERF roll (about 10-08), the two Talentcell packs (about 10-21), the two 5 A MPPT boards (about 10-23), the Ecoflex pint (about 10-25), the three DG7512 boards and the heater pads on arrival, and the 5 in acrylic dome (drop off by 10-12). Found in the mail 2026-09-26: the second FE-P was cancelled 2026-09-21 before it shipped and the carbon brush pair was cancelled by the seller, so neither needs a return; the Lensun panels were cancelled and $334 refunded; the attwood lights went back, $71.20 refunded. Found 2026-09-30 (mail audit): the SmartSolar 100/20 story is corrected (Mike, 2026-09-30): of the two 100/20s ordered 2026-09-07, one came and the other, delayed, was cancelled 2026-09-29; its replacement (Amazon 112-8872850-2808228, $74.25 in points) shipped 2026-09-30 and arrives Oct 1. It is the second 100/20, not a third: keep it. The two-100/20s-then-three-75/15s layout stands. The VEVOR 16 in box was too small (Mike, 2026-09-29): VEVOR refunds on photos of it destroyed, nothing to ship. The SUPREMETECH 5 in dome is refunded ($16.93, 2026-09-30). *Documents: the order tracker, BOM (15.6 returns line, 15.8 FE-P line).*
38. Return the weather pod parts bought for the post: any pod housing or QwiicBus parts. *Documents: the order tracker.*
138. Fix Supply brass: check the certificate covers these fittings. The tracker logs Fix Supply's 2026-09-09 answer for order 000166489: certificate COCGSCIBRZTHDFTG-02, SCI bronze threaded fittings to ASTM B584 / B62, UNS C83600, which is a red brass. If it names the two 1 in tees and the two 1-1/2 x 1 in reducer elbows, write C83600 into the BOM and install them; if not, buy bronze. They stay out until then. *Documents: BOM 15.11, Flow-Through Wet Lab Design 4.*
261. Spain: permission for the boat to come into and leave the Canaries. Researched 2026-09-24: Reference/Spain_Autonomous_Vessel_Permission_2026-09-24.md, whose sec 7 is the step list; the steps are written into Boat 1 Route sec 5 and the Checks file, and the permission itself is still to get. The short of it: written permission from the Capitanía Marítima de Las Palmas (Spain's harbor master office there), once to enter Spain's waters and once to leave port, asked in Spanish, at least 48 hours ahead by law but in practice months ahead: have it before leaving Brunswick. The boat cannot clear itself in or out, so a Las Palmas ship agent or PLOCAN (Gran Canaria's ocean research center, which handled Saildrone in 2019) gives the Spanish address and does the paperwork. Expect a liability insurance question (Spain's floor for leisure boats is EUR 336,567). Steps: October, email PLOCAN and write to the Capitanía (+34 928 468 262); by mid-October, settle the flag and papers (Build Guide 11.5); November, insurance; by December 1, file the entry application; December, book the tow in. Biggest unknown: whether the boat may sail itself inside Spain's 12-mile zone or must be towed from outside it. Not researched: Portugal's rules, since the route passes inside Portugal's 200-mile zone near the Savage Islands. *Documents: Boat 1 Route; Build Guide 11.5.*
121. Left to buy. Rebuilt 2026-09-26 with a link for every item: Reference/Left_To_Buy_2026-09-26.md (and .docx); the total is at the top of that file (about $5,320 on 2026-09-28, after the deck solar change and a pass against the order tracker, whose Buy list tab now agrees; the deck panels and their tape are $3,300 of it), plus shop work by quote and the items that wait on a check. It replaces the 2026-09-24 list, now in Archive. Still waiting on something: the mast weldment and wing can (8b), the Plate 3 hinge stop welding, the waterjet order (haul-out), the deck solar and its tape (the four deck checks of Build Guide 4.4), the grit (86b), the release trunk cable (17), the bridle hardware (81). Before buying the Iwaki pump (bought 2026-09-28, $188.29), read the NRD manual for a minimum-flow note: it will run at 30 percent of its free flow. *Documents: Left to Buy, BOM.*
109. What is left of the 2026-09-15 buy list. To buy: a second 50-pack of M6 (1/4 in class) 304 conical washers ($7). To log: the Thomson MD part and price (forward Thomson's acknowledgement) and what the S1eX cost. Done: the foredeck fisheye (Hanwha XNF-8010RV, bought 2026-09-30, $132), Lasdrop, S1eX, cord pullers, tinned 18/4, wet lab PVC, the MOTU M2 (ordered), and the CAN boards (four SN65HVD230 on hand). *Documents: BOM, the order tracker.*
129. Money with a clock.
Documents: the order tracker, BOM.
139. Services and fees are on no list. Insurance (hull, liability, wreck removal); the flag or registry entry and the USCG notice; Iridium airtime for both modems (about $60 a modem to unlock, plus a plan); transport to the launch port, haul-out and launch; a crane or travel-lift for the 30, 60 and 90 degree heel test; a chase boat and a shore station; the lake power sweep and the inclining test; the PMI or hardness test; the 2205 fabrication with its dye-penetrant report; welding the twelve bolt heads to the rail. The BOM carries a 12 percent lump with no lines under it. Price these or mark them out of scope; Build Guide 11.2, 11.3 and 11.5 name most. *Documents: BOM.*
140. Test gear the launch gates need. On hand: the insulation tester (bought 2026-09-21), the 0.1 g scale and a come-along; the duckbill manometer is a vinyl tube on a yardstick. No pull tests, so no load scale (Mike, 2026-09-28). The 3M 2091 P100 filters are on hand; the respirator and HEPA vacuum are off the buy list (Mike, 2026-09-26). No thermal imager: Mike dropped it 2026-09-26. *Documents: BOM 15.1.*
142. Charges nobody has explained. DHL $270.84 (waybill 1296045575, invoice D35180194, a shipment of June 5); UPS $49.73; a $784 PayPal payment to Hunan YLWL on 2026-08-19. If any is boat parts, it belongs in the cost reconciliation. *Documents: BOM, the order tracker.*
224. The second relay module is probably on the shelf. Item 197's second 8-channel relay module (about $15) is on the BOM as to buy, but the tracker logs two HiLetgo 24 V 8-channel modules bought 2026-08-16 and 2026-08-18, and Electronics plan sec 8 lists two relay modules the node no longer uses. Check the shelf before buying; likely nothing to buy. Also likely on hand, per the tracker: the two 3/8-16 x 5 in 316 mast-socket bolts (Bolt Depot 2624658, check the length) and the sealed fuse holders (Mike's 2026-09-21 notes). *Documents: BOM 15.8, Electronics plan 8.*
227. Parts bought with no line in the design. Each needs a job written into the BOM, or it goes back. Mike says what each is for:
Documents: BOM, the order tracker.
468. Whole-boat box review: the aft junction checks run on every other box (Claude, 2026-10-30, from Mike's "please review the entire build and other boxes to verify all these things"). Read-only review of every box against the guides (Electronics plan Appendix A, sec 7, Box Wiring Sheets, the dated drawings), looking for unfused feeds that leave the cabin, fuses bigger than their wire, gland rule breaks, glands that do not fit, missing surge diodes, signal level mismatches and docs that disagree. Each line says how sure: known from the text, inferred, or a guess. Most serious: (a) The central release enable relay is channel 5 of an 8-channel module (Compute Bay 10.3 step 2), but it carries the burn current, 12.5 A per horn on a feeder sized for 25 A, for up to 60 s; HiLetgo-class 8-channel modules use relays rated about 10 A (inferred from the usual SRD-24VDC parts; no doc states this module's rating). Release Node B uses a 30 A single-channel module for the same job. Claude's view: use one of the spare HiLetgo 30 A modules for the central enable too. (b) Switched outputs at the central node with no fuse of their own, so a short trips the node's own 15 A breaker and the node (steering, the cell-temperature trip, releases) drops: ch 4 em-trak and ch 9 camera port UV-C fail ON, so a short there comes back at every breaker reset and locks the node out (known); ch 6 and ch 12 bilge manual-ons put the pump in the bilge behind the node's 15 A instead of its own 5 A (inferred); ch 2 passes up to about 10 A of wet lab converter current through a module contact (inferred). Ch 1 and ch 3 are fused at the aft junction (item 467). Claude's view: a small slow-blow fuse on every switched output on the node's rail (holders and the fuse kit on hand). Compute Bay 11 also says the fail-ON loads stay on with the node's feed lost; they cannot, since their contacts are fed from that feed (known). (c) Nav light voltage: the bow lantern and both post whites are TRUE MODS 12 to 24 V (Instrument Post Guide 3), on a bus at 27.6 V charging and up to 29.2 V (known); Electronics plan 6.4 says every fixture takes the bus voltage as it is. Only the stern light (9 to 33 V) is rated for it. I don't know whether the 12 to 24 V lights survive 29 V. (d) The cord puller fuses sit on the minus side (between the MOSFET and the reversing relay); a chafed + core onto its shield (drain at 0 V) bypasses the fuse, leaving only the 30 A breaker on 18 AWG (inferred). Moving each fuse to the + side fixes it. (e) The post wind meter's 5 V data goes straight into Nucleo D0 and D1 with no level chip; those pins take 5 V only while the Nucleo is powered, and the meter runs from the aft junction's always-on feed, so with the node off it drives 5 V into an unpowered pin (guess from ST's pin ratings). Surge diodes: (f) the central node has one TVS (at its 5 V converter); Lightning Guide 7.7 says every power input (entries 2a, 12, 29 have none) (inferred); (g) Release Node B has none in A15 or Sheet A15, on a 10 to 12 ft feed (inferred); (h) WD-B's D36V6F5 has none drawn though Lightning Guide 7.7 lists it (inferred); (i) the switch box has none and A6a says nothing lightning related is inside, against Compute Bay 14.1's every-converter rule; write the exemption or fit one (known); (j) the wet lab box diode is across the XWST output in A18 and Sheet A18 but at the input in Wet_Lab_Box_Layout_2026-10-15 (known); (k) science data lines: Lightning Guide 7.5 keeps their surge protectors, A18 runs the DO and SBE 45 lines with none (known); (l) Lightning Guide 7.6 still says the BMS and VESC feeds are surge-clamped downstream; neither is now (known). Glands: (m) Iridium A's 16/2 feed in a PG7 (A11, Sheet A11-A); every other 16/2 gets a sealant-filled PG9 (known). (n) Central node entry 16 and aft junction entry 10 put three 16/2 runs in one PG13.5; 7.2 allows two, and three oval jackets are about 0.6 in across against about 0.47 in a PG13.5 seals (guess on the sizes). (o) The BMS box entry 3 puts two loose Cat6 pairs in one PG9 (keep them in one Cat6 jacket, or a gland each) (known). (p) The three slip ring hull cables, of different sizes, share one deck gland, which has no size or buy line (known). (q) Duplex entries not marked sealant-filled: central node 2, 2a, 12, 29, 30, Release Node B 2, Starlink enclosure power (known). (r) The split gland is "PG25" in 7.1, A1, A1a and A2, but "M30, 1-3/16 in hole" in the hub box drawing and item 447 (known conflict). (s) The VESC A box's two Micro-C cords in PG7 may be over 1/4 in (guess). (t) Wing box: five PG9s barely fit the 5.9 in wall, and the shielded Cat6 may need a PG11 (guess). (u) The wet lab drawing's staggered rows sit about 0.9 in apart against about 1.0 in needed (inferred). Fit: (v) The switch box's Mean Well DDR-120B-48 is about 4.9 in tall (seller spec) in a 3.9 in deep box: the rail must run up the wall, which no doc says; the jacks then land on the two end walls, which no drawing shows (supply height known, crowding a guess). (w) The BMS box size is never given, so its four entries cannot be checked (known). Wire against fuse: (x) the 300 A Class T and the 250 A MEGA each guard 2 AWG (about 210 A at 105 C); the boat wiring rule (ABYC E-11) allows up to 150 percent on these, so it passes there, but 7.6 says never past the wire rating (Claude's note: reword 7.6). (y) The wing tip's 1 A fuse is at the tip, so the 22 AWG pair up the mast is guarded only by the wing box's 5 A (inferred). (z) Six stud-to-breaker jumpers on the bus board drawing look over 7 in with no split loom shown (breakers 15, 14, 13, 2, 4, 1) (inferred from the drawing grid). (aa) The hub's 5 V lead has no wire size, so its 3 A fuse cannot be checked (known). Docs that disagree or are stale (known): (bb) shield rule 7.4 ("everywhere else it floats") against A3 entries 4 and 4a (18/4 drains to 0 V), the detector lead's drain at Pi B, and the HygroVUE shield landed in the hub box; (cc) vent: 7.4 says the central node only, A9, the Battery Box guide and the totals keep the battery box Gore vent; (dd) the bus board drawing still shows line 13 as 16/2 (14/2 since 2026-10-15) and two MEGA holders against 7.6's four; (ee) the relay modules' trigger jumper setting is written nowhere, though A18 says to set it as the central node's; (ff) PC817 for the island-alive line drawn in both Pi A and Pi B; (gg) the CAN drops' 12 V pair has no stated landing at either Pi HAT, and at Pi B a wrong landing ties the floating island to the main negative (inferred); (hh) no home for the UV-C LM317 and its resistor, and the camera cap's one gland already holds the camera cable (inferred); (ii) the wing box is A17 in the plan but lives on Sheet A14; (jj) Wet_Lab_Box_Layout_2026-10-15 and Reference/Wet_Lab_Box_2026-10-14 still show a vent; (kk) Reference/Wing_Electronics_Wiring_Sheet_2026-09-24 describes the deleted wing tub but the Instrument Post Guide still cites it; (ll) Iridium_to_Pi_Wiring_2026-10-15 still says one USB isolator to buy (both bought 2026-10-16); (mm) Reference/Battery_Tap_Wiring_2026-10-03 still says "tap bundle"; (nn) the horn's 5 A fuse is missing from 7.6's still-fused list. Claude's recommendations on (b), (c) and (e), asked 2026-10-30 (Mike: "what do you recommend for fuses? need to be sensible", "do we need voltage reg to 12v for the lights?", "#5 please research"): (b) four more 5 x 20 mm slow-blow fuses in DIN holders on the central node's rail, each right after its relay contact: ch 4 em-trak 2 A (the B921 averages 135 mA and peaks 2 A at 12 V, published; about half that at 24 V, inferred), ch 9 UV-C 0.5 A (the smallest in the kit; the LEDs draw about 20 mA), ch 6 and ch 12 bilge pump manual-ons 5 A each (the same as each pump's own breaker). None on ch 2 (its own 15 A breaker upstream), ch 7, 10 and 11 (inside the cabin, a few milliamps). The four puller fuses move to the + side, same sizes. Fuses from the QBGTFAK kit on hand; holders: count the MECCANIXITY pack (central node 8, aft junction 5, wet lab box 5). (c) No 12 V converter: a converter is one more part that can fail dark, and the guides keep the light circuit converter-free so fail-on means what it says. TRUE MODS publishes no upper limit for its 12 to 24 V lights (I don't know whether they take 29 V); its stern light is sold as 9 to 33 V, so the brand states a wide range when it has one (inferred: the 12 to 24 V ones may really stop near 24 V). Cheap guard with parts on hand: six 1N5408 diodes in series ahead of the bow lantern (after F4) and six ahead of the whites (after F3), on DIN blocks in the aft junction; about 0.7 V each at these small currents, so about 25 V at the light at full charge and about 20 V with the bank low, both inside 12 to 24 V. A diode usually fails shorted, which leaves the light lit. The stern light needs none. (e) Researched: the Calypso ULP manual (UART/I2C version) gives power 3.3 to 18 V and does not state the data level (marine-products-direct.com copy of Calypso's ULP STD UART/I2C manual, read 2026-10-30); a user report says 5 V TTL (secondhand). ST's own staff give the rule for an unpowered STM32 five-volt-tolerant pin as at most min(VDD + 3.6 V, 5.5 V), which is 3.6 V with the chip off (ST community answer for another STM32; the G431 datasheet's table could not be fetched, so the exact figure is inferred). So 5 V into D0 with the node off is over the limit, if the meter's data is 5 V. Fix that works whatever the meter's level: power the meter from the central node's own 5 V, down two spare trunk cores through a 0.5 A fuse at the node, so the meter and the Nucleo are always on or off together; the aft junction then loses its 5 V regulator and fuse F1 (a change to Sheet A6 and the build sheet, so Mike's yes first). The meter draws 0.25 mA, so the 30 ft or so of 22 AWG costs nothing. A divider on the data line would only work if the meter's level were known. Checked and fine: the other branch breakers against their wire, the stud pairings, the backup board's fuses and ANL, Pi A and Pi B wall fits, the battery box and BMS cable sizes and core counts, the stern release cables across all five documents, the central node's wall fit. *Documents: as named in each line.*
467. Aft junction box build sheet: what drawing it found (Claude, 2026-10-30, from Mike's "make a layout for the aft junction box like you did for the others, with step by step"). Drawing: Aft_Junction_Box_Build_Sheet_2026-10-30 (.html master, .pdf to print, four pages: the plate lid off, the holes with the bottom wall seen from outside, every wire, notes and 21 build steps; built by Reference/_make_Aft_Junction_Box_Build_Sheet_2026-10-30.py). It keeps Sheet A6's walls and part places (post cables on the top wall, everything else on the bottom wall; feed terminals left, nav lights middle, relay top right, sensor terminals right). (a) DECIDED (Mike, 2026-10-30: "yes to fuses"): the two deck feeds get a fuse in the aft junction box, the bow lantern 1 A (F4) and Starlink power 5 A (F5), slow-blow, in DIN holders on rail 1; drawn into the build sheet the same day; the Electronics plan A6 and 7.6 "still fused" list and the order tracker's fuse holder count are still to write. The question was: fuse the two deck feeds that pass through unfused? The bow lantern (ch 3) and Starlink power (ch 1) leave this box with no fuse; a short on deck trips the central node's own 15 A breaker, which feeds the node and every relay channel, and because ch 3 fails on, a shorted bow lantern would trip it again each time it reset. Questions: (b) Entry 11 ("lights group") has no wire: the post lights cable (entry 4) carries the stern light and the whites, and the bow lantern has entry 19 (settles the "Lighting exits at the aft junction" fit-out flag if Mike agrees). (c) Entry 13 holds two cables to two places (em-trak, wet lab line), against the grouping rule (Electronics plan 7.2); Claude's view: the wet lab line takes entry 11. Claude's calls, standing unless Mike says otherwise: (d) the post feed fuses are DIN rail holders with 5 x 20 mm slow-blow fuses, 1 A each, three of them (wind, stern light, whites), since the post lights cable carries two feeds from two sources; the BOM 15.8 line said blade fuses and no holder was picked; (e) the motoring relay module runs off ch 7 itself (DC+ and IN on ch 7's red, trigger jumper high), so it draws nothing while off; (f) encoder A runs on the central node's 3.3 V sent up the trunk, so its pulse stays at 3.3 V for pin D6; trunk cores 1 and 2 wind data, 3 signal ground, 4 and 5 encoder, 6 and 7 the pump running states, 8 and 9 pump 2 on if it rides there, 10 to 12 spare; (g) no shield lands in the box; the trunk's drain goes to 0 V at the central node only, like the stern release cables; (h) the bottom wall's eleven glands (twelve with 15) go in two staggered rows, 1 in and 2-1/4 in below the rim, since one row needs about 12-1/2 in; that needs a base at least 3-1/2 in deep and a plate about 10 x 6 in (item 87). Still open as before: the post lights 4-core (item 440 (e)), the wind data line's surge part (item 440 (f)), pump 2's manual-on path, and the bow lantern, bilge and encoder cables. *Documents: Electronics plan A6 and 7.6 (fuse kind and count, entries 11 and 13, trunk cores, encoder power); Box Wiring Sheet A6; BOM 15.8 aft junction line; order tracker (fuse holders) once answered.*
463. Boom-root pocket: the actuator, link, boom joint and both PEX ends in one shaped pocket inside the wing, on its centerline, under one lid (Mike and Claude, 2026-10-27; an idea, not decided). Mike, 2026-10-27: everything inside the 28 in chord, on the centerline, a bolt-on cover, two separate PEX tubes into it (one from the root, one to the tip); "I'm not emotionally tied to anything. I just want to do this a good way." The first version (an open box from the spar to the back edge with 1/4 in G10 top and bottom walls and a clamp bar) came to about 3 lb over the original plan; Mike: "That's A LOT!", "Why so much open space? Could add glass against foam to strengthen?" Second version, drawn: Wing_Boom_Bay_2026-10-27 (.html master, .pdf to print, two pages; built by Reference/_make_Wing_Boom_Bay_2026-10-27.py). Proposed: (a) one shaped pocket cut from the boom face that hugs the parts: an actuator box from the spar's back face (10.3 in back of the nose) to 19.2 in, 42.6 to 48.5 in up; a link slot 19.2 to 22.4 in, 46.2 to 47.8 in up; a blade box 22.4 in to the back edge, 42.6 to 47.8 in up; about 85 sq in, against 148 for the first version. (b) The foam stays everywhere else; the pocket's walls and floor get 2 layers of biax, lapped onto the far skin and turned out 3/4 in as the lid ledge. No G10 walls. (c) Both PEX tubes stay 12 in back of the nose and end in the pocket's bottom and top walls; the wing cable crosses on the lid side of the actuator. (d) The actuator hangs on the wing's centerline, rod line 47 in up, rear pin 11 in back on a G10 lug glued and bolted to the spar's back face, with the cradle and strap; spring link 18.2 to 26.7 in. It sits this far forward because on the centerline it needs about 2-1/2 in of wing thickness, which ends about 19.4 in back (rib table), and because the spar is the strongest place to pin it; it also takes about 2-1/2 lb of lead off the nose against the boom-root spot (inferred). (e) The boom's front end is a G10 blade, 1/2 in thick, 8 x 3 in, 43 to 46 in up: front 5 in in the pocket, back 3 in glued inside the boom's root, which is deepened to 42.6 in and back to its normal 2-1/2 in by 35 in back. Worked out: about 20,000 psi under the 60 lb breakaway load about 42 in out, about 3,700 psi under a normal gust (G10 about 60,000 psi, a typical figure); same strength as the first version's 3/8 x 5 in blade, about 1/3 lighter (0.8 lb), because a thicker, shorter blade is stiffer for its weight sideways. (f) Four 5/16 in 316 bolts, 2 in apart along the wing and 1.8 in up and down, heads potted in a far-side seat (filled epoxy and G10, glued and taped to the far skin), so nothing goes through the far skin; fender washers and nuts on the lid side, no clamp bar. (g) Pushrod in a 3/8 PEX liner inside the boom. (h) The lid is the pocket's own cut-out skin plus one layer, gasketed, about 15 #10 screws into rivet nuts in the ledge, raised about 3/8 in over the back 3 in; drain at the low back corner, breather at the top. (i) Build: carve the pocket, glue in the lug and seat, glass it, refill with the carved foam taped as a plug, skin the face whole, cut the lid out. Weight, rough guesses: about 2.2 lb of parts against about 2 to 3 lb for the original plan's plate, solid patch, extra layers, outside fairing and actuator pads, so about even. If Mike says yes, this replaces the hardpoint, firm base, corner pocket and branch in item 461, the outside fairing and boom plate in item 462, and the boom's four-bolt joint in Wing Construction Guide Stage 10. Found 2026-10-28 (Mike: look for opportunities for improvement): the drawing shows the actuator pulled in; its 2 in (50 mm) stroke moves everything on the rod line 2 in aft, so the link's rear joint goes from 26.7 to 28.7 in back, 1 in past the back edge, into the boom's front face. Fix proposed: a glass-lined 1 in hole, 2-1/2 in deep, in the boom root's front face on the rod line (the root is 1.5 in wide there; clears the blade top by 1/2 in), and a raised strip about 1-1/4 in wide in the lid over the link's travel, where the wing is thinner than the link's 7/8 in caps. Also proposed: flat-head screws countersunk into the blade into nuts cast in the seat (nothing sticks up at the bolts); a 1/8 in G10 pad under the ledge at each lid screw, and the lid's front edge hooked under a lip at the spar so it needs screws on three sides only (about 11); a slack loop in the wing cable tied to the pocket's top wall so the actuator lifts out past it. Not yet drawn; Mike's yes asked 2026-10-28. Third version, drawn 2026-10-28 (Mike: "bolts thru other skin help", "maybe g10 strip under screws rather than a bit under each screw", "draw all these things and suggestions with details, 3 view"): Wing_Boom_Bay_2026-10-27 now three pages (side view, three views, details). Changes: (j) the four blade bolts go through the far skin: 5/16-18 316 flat heads countersunk flush in the far face, through the seat and the blade, fender washer and thin nylon-insert nut on the lid side; bolts moved to 24 and 26 in back, 1-3/8 in long; the nut tops sit 0.63 in out, under the lid's 0.66 in underside, so no raised part over them. (k) One G10 strip, 1/8 x 3/4 in, under the lid edge all round but the back edge: outer half glued under the wing skin, inner half sticking out over the pocket and taking #10-24 rivet nuts; about 14 #10-24 316 pan-head screws about 3 in apart; the cut line is on the pocket wall. About 0.3 lb, against about 0.1 lb for separate pads. (l) Found: the spring link's caps are 5/8-18 coupling nuts (Trim Guide sec 6), 1.08 in across the corners, link about 6 in end to end; fully out (2 in stroke) its back cap runs from about 25 to 27 in back, where the plain wing is only 0.8 to 0.1 in thick inside, so it would push through both skins. Fix: a build-out, the wing's last 4.6 in (23.1 in to the back edge, 42 to 48.55 in up) built out on both faces to the boom's 1-1/2 in width, 0.15 in to spare each side; its near half is part of the lid. This replaces the boom-face hole and the raised lid strip proposed earlier the same day. Explained on one page, 2026-10-28 (Mike: "Explain and diagram the fillet idea"; then, 2026-10-28: "What is a fillet? I thought it filled a square corner with a curve", so the word is now build-out): Wing_Boom_Build_Out_2026-10-28 (.html, .pdf; built by Reference/_make_Wing_Boom_Build_Out_2026-10-28.py): the build-out is flat at 3/4 in each side of the centerline from 42 to 48.55 in up and tapers back to the plain wing over 1 in above and below; foam blocks glued on before skinning; about 0.1 to 0.2 lb (a guess). Mike's yes on the build-out asked 2026-10-28. (m) The link's own 1/4 in 316 rod runs on aft as the pushrod (one rod, no link-to-pushrod joint), replacing the threaded joint asked 2026-10-28; the actuator end of the link takes an eye for the 10 mm pin on cap A's stud (part to pick). (n) The blade is 8 x 3 in at 23.3 to 31.3 in back (4.4 in in the pocket). (o) The lid's front hook is dropped (saved 3 screws, needed a thin skin lip over a gap); the lid's back edge has no screws and the boom root's front face carries a gasket. (p) Two drains through the far skin at the two low corners; the wing cable gets a slack loop tied to the top wall. Weight, rough guesses: about 2.5 to 3 lb of parts against about 2 to 3 lb for the original plan. Simple version, drawn 2026-10-28 (Mike: "Way to complicated to build. Why not rectangle lid? What foam stays? Is inside foam glassed? Boom shape difficult. Build out hard."): Wing_Boom_Pocket_Simple_2026-10-28 (one page; built by Reference/_make_Wing_Boom_Pocket_Simple_2026-10-28.py). (q) One rectangle pocket, 10.3 to 22.4 in back, 42.6 to 48.9 in up: all foam inside it comes out, walls and floor get 2 layers of biax; all foam outside it stays. One rectangle lid, cut from the skin, on four straight 1/8 x 3/4 in G10 strips, about 12 #10-24 screws into rivet nuts. (r) No blade, no build-out, no shaped seat. The boom is the guide's plain tapered box (1-1/2 in wide), only 3 in deep at the root instead of 2-1/2, and starts 18.7 in back, inside the pocket, passing out through a plain rectangle notch cut in the wing's back edge from 22.4 in to the edge, cut faces glassed, gasket round the boom. (s) The spring link, unchanged, rides in a 1-1/4 in fiberglass tube set in the boom's core (18.7 to 27.6 in); its rod runs on as the pushrod in a 3/8 PEX liner. Fully out the back cap reaches 27 in, inside the tube. (t) Four 1/4-20 316 bolts come in through the far skin (flat heads, flush) into threaded inserts in solid epoxy plugs above and below the tube in the boom root, 2.2 in apart along the wing; nothing on the lid side; the boom root is bedded on the far skin in thickened epoxy (boom wrapped in release tape, bolted down while it sets). Worst case about 570 lb per bolt against about 2,000 lb yield (inferred). (u) To service the link: lid off, pull the actuator's front pin, take the four bolts out from outside, slide the boom aft out of the notch, then slide the link out of the boom's front end. (v) Open: the lightning jumper that went to the guide's boom-root clamping plate (Wing Construction Guide Stage 10) needs a new end, likely one of the four bolts. Weight, rough guesses: about 2.2 lb of parts, about even with the original plan. Mike's yes on the simple version asked 2026-10-28. Mike, 2026-10-28, on the simple version: "What does boom bolt to? Not looking sturdy. Why a vent? Where do actuator pins go? Getting closer I think." Claude: right, it bolts only to the far skin, a thin panel over the pocket; worked out roughly, a hard hit (the 60 lb breakaway load) puts about 1,100 lb on the two bolt rows 2.2 in apart and bends that skin far past what glass takes (inferred). Proposed fix, not yet drawn: (w) the pocket's top and bottom walls become the two load-carrying ribs: 3 layers of biax on the foam (or 1/8 in G10), glued to the spar's back face and to both skins; (x) two 3/16 in G10 cross walls between those ribs, at 18.7 and 22.4 in back, each with a rectangle hole the boom slides through snugly, bedded in thickened epoxy; the boom is held at two points 3.7 in apart, about 700 lb on each in the worst case and about 125 lb in a normal gust; the cross walls hand it to the ribs, the ribs to the spar; (y) two bolts through the far skin only stop the boom sliding out. Vent: dropped (Claude, 2026-10-28): the pocket is not sealed, the drain at its low corner lets it breathe. Actuator pins, both 10 mm with R-clips: rear pin through the actuator's rear fork and a G10 tab on a plate glued and bolted to the spar's back face, 11 in back, 47 in up; front pin through the actuator's front fork and the eye on the link's cap A stud, 18.2 in back, half an inch ahead of the front cross wall. The strap (Mike, 2026-10-28: "Ss strap? Specs?"), proposed: two polyester-coated 316 stainless cable ties, 14 in, 300 lb (Supreme Cable Ties, Amazon B084T58GGZ, , 100 pack, price not checked), each round the motor housing and through two slots in the G10 cradle, about 2 in apart, over a 1/16 in rubber strip where they touch the housing. The wrap is about 12-1/4 in round housing and cradle (2.13 x 3.63 in housing, Thomson), so the 14 in tie fits with about 1-1/2 in to spare. Load: the 2.4 lb actuator at even 10 g is 24 lb, against 600 lb for the pair (inferred). One-use: snip off to take the actuator out, fit new ones. The housing's material is not given in what was found (Thomson calls it corrosion resistant, 500 hour salt spray); the coating and the rubber keep bare stainless off it. Not on the tracker; nothing bought. Build sheet, 2026-10-28 (Mike: "Getting complicated again. Not clear at all what foam to cut out, where to bias, where for g10, what is a cross wall, etc"): Wing_Boom_Pocket_Build_2026-10-29 (two pages: what to cut, glass and G10 in colour; steps and parts; built by Reference/_make_Wing_Boom_Pocket_Build_2026-10-29.py). This replaces the simple version's joint and the cross walls and rib walls proposed the same day (items w and x above): (z) the pocket and lid cover the actuator only: 10.3 to 19 in back, 42.6 to 48.9 in up, all foam out down to the far skin, 2 layers of biax on the floor and walls, four straight G10 lid strips. (aa) The boom root sits in its own slot, 1-1/2 x 3 in, 45.5 to 48.5 in up, from the pocket to the back edge, bedded in thickened epoxy (boom wrapped in tape); both skins stay whole over it from 19 to 22.4 in, and past 22.4 in it opens both faces (the notch). (bb) Four 1/4-20 316 bolts go through both skins, the bedding and the boom (above and below the link tube, 2.3 in apart along the wing), with 1/8 x 3/4 in G10 strips under the heads and nuts and three biax doublers on both faces (8, 7 and 6 in tall, from 19.5 in to the back edge, cut round the notch, off the lid): the boom bolts where both skins and the foam between them are whole, as the guide's boom always did. (cc) G10 pieces: spar plate 1/4 x 2 x 3 in with a rear tab 1/4 x 1 x 1.2 in (10 mm hole); motor shelf 1/4 x 4.3 x about 3 in; lid strips 1/8 x 3/4 in (two 9.4 in, two 5.5 in); bolt strips 1/8 x 3/4 x 2.8 in (four). Build sheet revised 2026-10-28 (Mike: "Much better. Makes sense and steps help. Why so much open space under actuator? How to build boom? Not convinced boom mount is sturdy enough … please calc carefully."), now three pages: (dd) the pocket floor rises to 43.7 in up (was 42.6), just under the motor shelf, which stands on two small G10 blocks so the ties pass under it; the actuator lead's plug moves above the motor; the pocket is 8.7 x 5.2 in. (ee) Boom build (page 3): core from two strips of the wing's 3/4 in H80 glued face to face with half-round grooves for the 1-1/4 in fiberglass link tube (front 8.6 in) and the 3/8 PEX liner; straight taper 3 in to 1-1/4 in over 50 in (19 to 69 in back of the nose); the front 3.4 in made solid with 406-thickened epoxy round the tube; edges rounded; the guide's 3 layers biax front third, 2 beyond, cloth cut so the threads run along and across the boom (about twice as strong in bending as 45 degrees, inferred); weight guess about 5 lb. (ff) Bolts back to the guide's four 5/16-18 316 with Bellevilles, nyloc, 130 inch-pounds, witness marks. (gg) Strength worked out in Reference/Boom_Root_Strength_Calc_2026-10-28.md (numbers from Reference/_boom_root_strength_calc_2026-10-28.py): the weakest part is the boom tube just behind the wing, about 2.2 times the 60 lb breakaway case with 45-degree cloth; the joint holds only with all four of: the solid boom root (bearing 470 psi at the limit against H80's 200), the three doublers both faces (skin over the slot 8,500 psi with them, 136,000 without), preloaded bolts, a snug bedded slot. (hh) Mike, 2026-10-28: "I'm worried about the 4 mounting bolts … would it help to extend extra biax above and below the bolts maybe 4 inches or so, spread the load?" Claude: the bolts themselves carry little (about 30 lb each at the 60 lb breakaway load once clamped, about 200 lb at 1.5 times, against about 3,900 lb); the extra biax helps the wing panel take the boom's push beyond the solid zone and clears the lid opening; put in: the outer doubler 10 in tall (42 to 52 in up), with 7 and 6 in under it, both faces. The lid opening gets 1/2 in round corners. With the boom's stronger glass (item 465) the weakest part is now the wing skin over the slot, about 2.3 times the breakaway case with a 1.5 margin counting one skin alone. (ii) Mike, 2026-10-28: "Not the bolts themselves, the skin that holds them." Claude found its own mistake: the strength check had counted the doublers as 9 layers, but the guide's doubler is 3 layers in all, so skin plus doublers is about 0.25 in, and on its own it cannot take the boom pushing it out over the slot (about 33,500 psi at the 60 lb breakaway load against about 30,000). Clamp friction normally carries nearly all the load and the holes themselves are fine (about 2,250 psi bearing at 60 lb). Fix put in the build sheet: one 1/4 in G10 clamping plate on each face, 3 x 4.5 in (19.4 to 22.4 in back, 44.75 to 49.25 in up), in place of the four small strips (the guide's own clamping plate); front bolt row moved to 20 in for edge room. With the plate alone (no friction at all) it holds about 1.6 times the breakaway case with a 1.5 margin; about 2.6 times sharing with the skin. About 0.4 lb plus about 0.4 lb of nose lead (a guess). Working: Reference/Boom_Root_Strength_Calc_2026-10-28.md. (jj) Mike, 2026-10-28: "1/4 clamping? 1/8 glued on should be plenty. I still think the extra biax should be extended 4 inches above and below the clamps"; "The epoxy filler supports bolts at the clamping area too. Want the support extended beyond the clamping area." Put in: the clamping plates are 1/8 in G10, 3 x 3.4 in (19.4 to 22.4 in back, 45.3 to 48.7 in up), glued on each face over the doublers; glued, the stack holds about 1.8 times the breakaway case with a 1.5 margin between the bolt rows even with no clamp friction (worked out). The doublers are 11.4, 9.4 and 7.4 in tall, the outer one 4 in above and below the plates (41.3 to 52.7 in up). The foam comes out of a clamp block 19 to 22.9 in back and 44.8 to 49.2 in up, skin to skin, filled with 406-thickened epoxy when the boom is bedded, 1/2 in past the plates all round: the clamp presses about 816 psi behind the skin and H80 foam crushes at about 200, so Mike's point is needed; about 0.5 lb more (a guess). (kk) Deep review, 2026-10-28 (Mike: "Please do a super deep review of this. Look for mistakes, vulnerabilities, and OFIs. And anything that could make it easier to build."): Reference/Boom_Pocket_Review_2026-10-28.md, two passes (Claude, and a second Claude working from the files alone). Biggest mistakes found, not yet fixed in the build sheet: the finished boom (about 1.84 x 3.25 in with its glass) is bigger than the 1.5 x 3 in slot; the boom's taper locks it in its molded socket (make the first 9 in parallel); at mid-stroke the actuator's front fork runs into the boom's tube mouth (let it, with a short e-clip pin; pull the pin fully retracted); the actuator's rear fork overlaps the spar plate by 0.03 in (rear pin to about 11.2 in, glue the plate, no bolts into the hollow spar); the bottom PEX mouth would drain the pocket into the wiring tube (standpipes); one drain, in the wrong place (two, flush); bolts need to be about 3-3/4 and 3-1/4 in with wedge plates for the 7-1/2 degree faces; lid strips too narrow for rivet nuts (1-1/4 in, one-piece frame); the link's rod hits cap A's stud (rod stops short, cap A solid). Weak spots: actuator over-constrained; link and 52 in rod serviceable only with the boom off and the 48 in rod on hand too short (screw joint behind cap B); exotherm in the big pours; lightning jumper has no metal plate to land on; Trim Guide's backlash spring has no rod to reach. Strength numbers to correct (all still pass). Mike's yes on folding the fixes into the build sheet asked 2026-10-28. (ll) Second edition of the build sheet, 2026-10-28 (Mike: "Yes, make sure it is easy to understand … many things just listed don't make sense."): all the review's fixes folded in, rewritten plain, four pages (1 what to cut, glass and G10; 2 how the actuator, link and boom fit, with the three pin positions and how to take it apart; 3 steps and parts; 4 building the boom and the strength check). Main changes: rear pin 11.2 in; spar plate glued and taped, no drilling the spar; actuator held only by its pins, shelf a padded rest; link caps turned round to 0.95 in, rod stops short inside the link, screw joint (1/4-28 coupling nut) behind the link so it unscrews, pushrod one 48 in rod; boom finished 1.5 x 3 in (core 1.16 x 2.74), straight for its front 9.5 in, 1-1/8 x 1 in fiberglass link tube; front pin 10 mm with e-clips, enters the tube at mid-stroke, pulled with the actuator fully in (Thomson's fork sizes to confirm the 0.95 in fit); pocket 43.5 to 48.75 in up and clamp block 19 to 22.9 in back, 44.25 to 49.5 in up, on rib glue lines; one-piece G10 lid frame with glued-in #10-24 nuts (back edge 3/4 in, no screws, so the front pin stays reachable); PEX ends as standpipes; two floor drains plus a lid notch; bolts about 3-3/4 and 3-1/4 in, plates on epoxy wedges; slow hardener and pours in layers; first doubler layer carried forward to the spar; lightning jumper to the actuator's rear pin; anti-rattle spring to the pocket's back wall; build between Stage 8 and Stage 9, boom built first, Stage 10's solid zone, doublers and metal plate left out. Strength (worked out): boom about 6-1/2 times the 60 lb breakaway case with a 1.5 margin, weakest spot the glued stack between the bolts at about 1.8 times. (mm) A review from another chat, 2026-10-29 (Mike pasted it: "thoughts?"), folded into the build sheet the same day, which is now dated 2026-10-29 (Wing_Boom_Pocket_Build_2026-10-29; Mike: "how about todays date on todays file?"). Put in: the rear tab is sanded to about 0.22 in where the actuator's rear fork goes over it (Thomson's fork slot is 0.236 in, per that review; Claude has not seen it on Thomson's drawing); the spring link's front end is a flat tongue about 0.22 in thick (G10 or stainless), since the actuator's front end is a fork too; the plug on the actuator lead is a 4-pin Deutsch DT pair (DT04-4P and DT06-4S with wedgelocks and contacts, e.g. , price not checked), above the motor; the boom's width is one rule, under 1.65 in to fit the wing's foam at 22.4 in, set by the cloth, since the socket is molded on the boom; the boom is clamped down in its jig for the pour, since it floats in wet epoxy; PEX is scuffed and flamed before gluing and held mainly by being buried. Changed instead of a fit test (that review said to try the real parts; Mike does not want tests): the boom's solid front is bored out to 1-1/4 in for its first 2 in, the farthest the actuator's front fork reaches (1.8 in at full stroke), giving the 0.87 in fork and its e-clip pin about 0.19 in a side. The strength table now counts no clamp friction at all, since bolt squeeze cannot be measured once built: each bolt then carries about 110, 450 and 670 lb (storm gust, 60 lb breakaway, breakaway x 1.5) against about 3,900 lb. New check, tear-out: a bolt pushed toward the boom's top or bottom edge (0.45 in to the edge) carries about 160 lb at 5 g against about 2,600 lb (worked out, Reference/Boom_Root_Strength_Calc_2026-10-28.md). Answered, no change: the actuator's end play (up to about 0.047 in, per that review) is taken up by the Trim Guide's anti-rattle spring, which pulls every clearance one way; the sheet uses no 5200 on PEX. Error found in the Trim Guide: sec 2's table says the actuator is "IP66 and IP69K"; the Electrak MD is IP67 and IP69K standing still, IP66 while moving (distributor listings of Thomson's data, e.g. ). Still unknown: the depth of the actuator's fork slots, the length of the cable's boot at the motor, and the force at which the actuator's overload trips. The plug, 2026-10-29 (Mike: "does the thomson come with deutsch connector or do we need to buy both male and female?"): every Electrak MD listing on Thomson's site found comes with bare wires ("cable with flying leads"), 300 mm or 1 m (about 12 or 39 in) by the full part number, which Thomson's missing acknowledgement would show; no plug option was found. So both halves are bought: DT04-4P and DT06-4S, their wedgelocks (W4P, W4S), 4 pins and 4 sockets, plus a crimper for Deutsch solid contacts if none is on hand (the tracker's "crimping tools" line does not say which). The slip ring's DT plugs (Wing Power System sec 1A) are not bought either; buy them together. Not on the tracker. Decided 2026-10-29: a plug, not the splice. The Trim Guide sec 3 says no plug at the actuator (splice its lead to the 18/4 for good and pot it; it allows a plug, greased and booted, for refit convenience). The build sheet has a plug above the motor so the actuator lifts out through the lid; Mike, asked plug or splice, answered "how about a smaller plug, that one is BIG", taken as yes to a plug. Decided 2026-10-29 (Mike: "Yes On the plug"): Deutsch DTM 4-pin (DTM04-4P and DTM06-4S, size 20 contacts) in place of the DT: same Deutsch sealing, the DTM06-4S about 0.60 x 0.69 x 1.19 in (distributor listing) against about 0.83 x 0.83 in across for the DT06-4S (TE drawing table); 7.5 A a pin against the actuator's 2.6 A at full load. Stamped DTM contacts take 16 to 22 AWG, so the 18/4 fits; the solid DTM contacts are 20 AWG only. Kit, e.g. MUYI DTM 4-pin, 5 sets, pins 22 to 16 AWG, (price not checked). The crimper must match the kit's contacts: use size 20 stamped contacts and the JRready open-barrel crimper kit that comes with 50 pairs of them (Amazon B0DZX8723C, price not checked). Both are on the order tracker's Buy list, and the five old spring link rows there are marked on hold. (nn) Servicing the pushrod, 2026-10-29 (Mike: "Looking at boom guide, how to remove and service pushrod? Can everything else be replaced and serviced?"). Claude found its own mistake in the build sheet: "unscrew the link from the pushrod" could not be done in place, because the link's rod turns freely inside the link (round rod in round bores, Trim Guide sec 6), so turning the link does not turn the joint. Fixed in the sheet: a #6-32 316 flat-point set screw in cap B rides on a flat filed 1-1/4 in long on the link's rod, so the rod cannot turn in the link (the flat stays clear of cap B's O-ring over the full travel); the front tongue is pinned to its stud; the coupling nut is locked on the link's rod with red thread locker and the pushrod screws in snug with anti-seize and no locker. It cannot work loose because neither side can turn (the actuator holds the front, the clevis on the tab horn holds the back), and the low unscrewing torque keeps the set screw's load small (inferred). To service: actuator out; turn the link counterclockwise seen from the front, about 8 turns (0.3 in of thread at 28 a inch), and lift it out through the lid; pull the clevis pin at the tab horn and slide the pushrod out the back of the boom, its clevis left set so the flap's straight setting comes back by itself. Rule for the tail build (Stage 12): leave the pushrod a straight way out behind the boom. Built in for good: the link tube and PEX liner inside the boom (a new boom replaces them), spar plate and tab, motor shelf, lid frame and nuts (a stripped nut is ground off and a new one glued on from inside), the PEX tubes (their cables pull). Link tube in PVC, proposed, Mike's yes asked 2026-10-29 (Mike: "Would it be better to put pvc pipe for the link section of boom inside?"): Claude could not find the sheet's 1-1/8 x 1 in fiberglass tube for sale (McMaster's nearest fiberglass tube with a 1 in bore is 1-1/4 in outside, too wide for the 1.16 in core). Proposed: 3/4 in Schedule 40 PVC (1.050 in outside, about 0.80 to 0.82 in inside), with the link's caps turned to 0.75 in instead of 0.95 (about 0.03 in a side; 0.06 in of 316 left outside the 5/8-18 thread); sanded and wiped before gluing, held mainly by being buried. 1 in Schedule 40 (1.315 in outside) is too wide for the boom. PVC softens at about the same heat as the H80 foam (inferred), so the slow pours already protect it. Replaced the same day (Mike: "what are caps", "can we buy caps that fit?", "this seems kinda jerry-rigged"): bought round caps nearly exist: McMaster 93282A540, 316 round coupling nut, 5/8-18, 55/64 in (0.86 in) round, 1-1/4 in long, so no turning; but each cap still needs its outer end closed (the stud on cap A; the rod hole, O-ring and set screw on cap B), so it is shop work either way. Proposed instead, Mike's yes asked 2026-10-29: the machine shop makes both caps from 316 round bar to a one-page drawing (round from the start, about 0.86 in across), and the link slides in 1 in CPVC hot-water pipe, copper tube size (1.125 in outside, about 0.92 in inside by its standard wall of 0.102 in), the same outside size the sheet already draws, so about 0.03 in a side. CPVC is made to tight sizes and rated for hotter water than PVC. The 3/4 in PVC proposal above is dropped. If the set screw key is left out, the link and pushrod come out only with the boom off (four bolts). Also looked at, 2026-10-29 (Mike: "maybe use 1\" pvc and make the boom bigger up front?", "use 1\" pex?"): 1 in Schedule 40 PVC (1.315 in outside) makes the finished boom about 1.65 in wide, while the wing's foam at the socket's back end (22.4 in, where the boom also bends hardest) is about 1.75 in thick (worked out from the NACA 0015 shape), so about 0.05 in a side is left for bedding, with no room for the pipe's or the glass's size variation; dropping the extra side layer to make room undoes the boom glass decision. Not proposed. 1 in PEX (1.125 in outside, about 0.875 in inside) fits the boom, but its bore will not take the 0.86 in round caps, nothing sticks to it, and it comes coiled; CPVC is better. Decided 2026-10-29: no set screw key; the link and pushrod are serviced with the boom off (Mike: "I think taking off the boom to service is better than a weak link"). The set screw in cap B and the flat on the link's rod are dropped from the build sheet, and with them the in-place steps above and the rule for a straight pushrod exit behind the boom. To service: actuator out, four boom bolts out, boom slid back with tail, link and pushrod; on the bench pull the clevis pin at the tab horn, draw the link out of the boom's front end with the pushrod following, hold the coupling nut and turn the pushrod by its clevis to unscrew; clevis left set, so the flap's straight setting comes back by itself. Back together: bolts bedded in sealant again, 130 inch-pounds, new paint marks, sealant bead at the notch. The screw joint stays (it lets one 48 in rod do), with anti-seize: it cannot work loose, since neither side can turn in use. (oo) Spring link drawn and listed, 2026-10-29 (Mike: "I'm kind of confused about the caps and the link, the whole thing. Can you give me links, hopefully Amazon links, to these parts and what they look like?"): Reference/Spring_Link_Parts_2026-10-29 (.html, .pdf; built by Reference/_make_Spring_Link_Parts_2026-10-29.py), two pages: the link cut in half at full size, how it gives both ways, and every part with where to buy it. Proposed, not decided. Drawing it found four things the Trim Guide's link (sec 6) gets wrong: (1) bought 1/4 in shaft collars are 1/2 in across, as wide as the 1/2 in washers, so they cannot pass into the caps' pockets: use 1/16 x 3/8 in 316 dowel pins through the rod instead (the guide allows pins); (2) a 5/8-18 jam nut is 1.08 in across its corners, too wide for the tube in the boom: a jam nut turned round to 0.84 in as a lock ring; (3) a tongue 0.22 in thick cannot take a 1/4-28 thread: the tongue is made as the front of cap A, one piece; (4) the tube is about 1-1/2 in long, not 3.5, once the spring's working length (about 1-1/2 in at 4 lb) is set between the caps' faces (worked out). The three shop parts, all 316 from about 6 in of 7/8 in round bar, 0.84 in round: cap A with tongue (1.875 in long; tapped 5/8-18 5/8 in deep at the back; pocket 0.42 in across, 0.80 in deep; tongue 0.22 in thick, 0.60 in wide, 10 mm hole 0.80 in ahead), cap B (1.875 in; tapped 3/4 in deep at the front; pocket 0.42 in, 0.60 in deep; 0.26 in rod hole with an O-ring groove at the back), lock ring. Bought, links in the sheet: tube (Amazon B0F6TTK8YX), washers AN960C416 (18-8; no 316 found), spring (closest: Penn Tool 03308525, 0.48 in, 2 in free, 4.2 lb/in, probably 302; no 17-7 PH found), dowel pins (Olander 6R37DOW3), O-ring size 010 Viton (Amazon B01CUR4LHS), 1/4-28 316 round coupling nut (McMaster 93282A160, 0.36 in round, 1 in long), CPVC 1 in (Home Depot), clevis at the flap (McMaster 6071K47, 1/4 in pin). Still to find: a 48 in 1/4 in 316 rod for the pushrod (on hand: two 24 in), and a 10 mm front pin short enough (under about 1.2 in overall) for the boom's 1-1/4 in bore; McMaster's shortest stainless 10 mm grooved pin found is 24 mm between grooves. Mike's yes on this link design asked 2026-10-29. (pp) No custom-made parts, decided 2026-10-29 (Mike: "What are the shop parts? We'd like to buy everything ready to go. Maybe we could assemble it, but we don't want to have any custom-made parts. It's very undesirable."). So the three shop parts in (oo) are out, and the spring link has to be bought ready-made or put together from bought parts. Found so far (Claude, 2026-10-29, web search): no ready-made spring link at this size and setting (about 4 lb, 1/2 in each way, under 0.9 in across to fit in the boom). The nearest ready-made items are aircraft control load limiters (no seller found) and an eBay "compliant link" with about 40 lb (18 kg) of preload, 20 in long. The Trim Guide's link (sec 6) always needed shop work ("a local shop turns and bores the two caps"). An all-bought version looks possible only where it can be about 1 in across, so not inside the boom; not yet worked out. Mike's call asked 2026-10-29. (qq) All-bought spring link worked out, 2026-10-29 (Mike: "Yes work"): Reference/Spring_Link_Bought_2026-10-29 (.html, .pdf; built by Reference/_make_Spring_Link_Bought_2026-10-29.py). Two Bimba Original Line 3/4 in bore air cylinders, 1 in stroke, used with no air, only for the spring in each: SR-041-P (spring return: holds its rod in, so it gives in pull) and SR-041-RP (spring extend: holds its rod out, so it gives in push), joined rear tab to rear tab by two G10 side plates and two 1/4 in pins, inside the boom's CPVC tube where the made link was (body 0.86 in in a 0.92 in tube). Front: a Bimba-size 1/4-28 rod clevis on cylinder A's rod, pinned to a G10 tongue 0.22 in thick that takes the actuator's 10 mm front pin. Back: cylinder B's 1/4-28 rod into a 316 round coupling nut (McMaster 93282A160), then the pushrod. Bimba catalog (BIM-PFL-0119, 2019, pages 17, 28, 65): springs 3 lb relaxed, 6 lb at full 1 in stroke (heavy option 4 to 10 lb, too stiff once it gives); 303 stainless rod with the SR prefix; 304 stainless body; heads aluminum (seller listing); spring material and seal friction not given. So it gives at about 3 to 4 lb (3 lb plus the seal's rub, guessed under 1 lb), and at the 0.4 in the flap needs to go straight it pushes about 4.2 lb, under 1-1/2 times the start. Front pin to the end of the coupling nut about 12.9 in, against 7.5 for the made link, so the CPVC tube grows to about 15 in; both cylinders together about 0.46 lb. Prices: TSI Solutions, SR-041-P $37.60, SR-041-RP $48.98. Still to pick: a stainless 1/4-28 rod clevis with a 1/4 in pin and 0.25 in slot (Bimba's D-166-3 is plated steel), and two 1/4 in 316 clevis pins. Weaker than the made link: aluminum heads and a spring of unstated material in damp boom air; fixed setting (change by changing cylinders); both air ports left open to breathe; the G10 plates break the metal path to the pushrod, so a short braided wire would cross each plate if the lightning plan needs the pushrod tied in. Also looked at: SMC NCMC075-0100S and -0100T (same sizes and 3 lb springs, music wire spring, $24.60 and $32.70) and Clippard USR-12-1 and URR-12-1 (3 lb, heavy 5 lb; price and source not confirmed). No ready-made linear part that springs back both ways was found. Mike's yes asked 2026-10-29. (rr) Two better ways, 2026-10-29 (Mike: "please look for a better solution"; then "Can you find parts to buy for the original plan? We can cut and drill, but lathing and fashioning, not so much."). Both use only bought parts that are cut, drilled or glued, plus G10 Mike cuts. A. Two-spring link, in the boom (Reference/Spring_Link_Two_Spring_2026-10-29, built by Reference/_make_Spring_Link_Two_Spring_2026-10-29.py): the Trim Guide's link with two springs instead of one, one for push and one for pull, so each washer stops on a plain glued-in piece of tube and no end cap needs a pocket. Body: 5/8 x 0.049 in 316 tube, about 4 in (Amazon B0F6TTK8YX). Middle and back stops: 1/2 in long pieces of 1/2 x 0.049 in 316 tube (0.402 in inside), glued in. Four AN960C416 washers (0.265 x 0.50 in, 18-8). Two springs, 0.48 in across, 2 in free, 4.2 lb/in (Penn Tool 03308525), each squeezed to 1.05 in for 4 lb; at the flap's 0.4 in it pushes about 5.7 lb. Two 1/16 x 3/8 in 316 dowel pins pressed across the rod behind the washers (Amazon B00R1UOOJ0). Front: the G10 tongue slides into the tube and is held by a 1/8 x 1 in 316 dowel pin (Amazon B00R1UV7QS). Back: the 1/4 in rod joins the pushrod with a 1/4 to 1/4 in set-screw coupling (Climax RC-025-S, 303 stainless), so no threads are cut. About 5.7 in from the front pin to the coupling's end, 5/8 in across, so the boom's link tube shrinks to 3/4 in PEX (about 0.68 in inside) from the 1 in CPVC. B. One torsion spring at the flap's hinge (Reference/Flap_Spring_Fuse_2026-10-29, built by Reference/_make_Flap_Spring_Fuse_2026-10-29.py), the way radio-control models protect their steering servos: the flap's arm turns on the hinge pin by itself; one Lee Spring LTR070M 02 S (302 stainless, 0.625 in across, 2 in legs, 7.0 in-lb at 180 deg, $13.40) sits on the hinge pin with its legs pressing outward on two pins in the arm and two in a G10 plate on the flap's end. Wound 80 deg at rest: 3.1 in-lb, so it gives at about 4.1 lb on the pushrod with a 0.75 in arm; at the full 30 deg of give 4.3 in-lb (1.38 times); 61 percent of its rated twist then (the Trim Guide asks for half; only while giving). Nothing gives inside the boom, the boom's link tube goes, and the actuator can no longer force the flap against its stops; it sits out in the weather at the tail, and the tail is not drawn yet. Claude's opinion: B is simpler (one spring, four pins, two G10 plates). Mike's choice asked 2026-10-29. Mike, 2026-10-29: every extra store costs $10 to $20 shipping and account setup, returns are out of the question, and "This looks like a very time-consuming project for what i would like to be a simple thing." So nothing is to be bought for the link yet; once the choice is made, Amazon unless it is not competitive on price or spec, then the cheapest seller counting shipping, as few carts as possible, every size checked against the seller's specs (Mike, 2026-10-29, correcting a one-cart rule: "over-reaction not helpful"). Checked so far at McMaster: torsion spring 9287K173 (302 stainless, 0.672 in across, 0.075 in wire, 2 in legs, 7.07 in-lb at 180 deg, right-hand; 9287K174 left-hand), close to the Lee spring; rod clevis 6071K47 (316, 1/4-28, 1/4 in slot and pin, cotter pin included). Open before buying: the 10 mm front pin's length (fork width from Thomson's drawing) and the tail end, which is not drawn. (ss) Hinge spring drawn in detail, 2026-10-29 (Mike: "PLease draw it with details for me to eval"): Reference/Flap_Spring_Fuse_2026-10-29 is now two pages: seen from below and from the side at full size, how it gives, how it goes together, numbers and parts. Layout: a 3/8 in 316 axle glued 1 in up into the flap's bottom end on the hinge line; under it, a 1/4 in G10 flap plate glued to the flap with two 1/8 in pins hanging down 1.45 in out from the axle, the spring (McMaster 9287K173) on the axle, a 1/4 in G10 arm turning on the axle with two 1/8 in pins standing up 1.75 in out and the clevis pin 0.75 in out (the plate is 1/4 in, the clevis slot's width), and a 3/8 in shaft collar under it. The pins sit well out along the legs because the legs leave the coil off-center and the pins nearer in would hit the clevis. Legs 100 deg apart at rest (wound 80 deg): 3.1 in-lb, so it gives at 4.2 lb on the pushrod; 4.3 in-lb at the full 30 deg (1.38 times). Weight about 0.3 lb at the tail (worked out), about 1.4 lb of nose lead, less about 0.6 lb saved by taking the link out of the boom root (a guess). Taken as given, since the tail is not drawn: the tail stands on the boom's end with the flap's bottom end free, the boom ending about 2 in ahead of the hinge line. Parts: spring and clevis from McMaster; 3/8 in 316 rod (Amazon B076CMP5J5), 1/8 x 1 in 316 dowel pins (Amazon B00R1UV7QS), 3/8 in 304 shaft collar (Amazon B0D5RK2GGJ); G10 on hand. (uu) Whole tail end drawn, 2026-10-29 (Mike: "ok the idea looks appealing, how to build it, attach to the boom, tail, and tab? in detail and in the simplest, easiest, and most reliable way"): Reference/Tail_Assembly_2026-10-29 (.html, .pdf; built by Reference/_make_Tail_Assembly_2026-10-29.py), three pages. Proposed, not decided. (1) The tail stands on the boom's end; the boom stops 4.25 in ahead of the hinge axis. (2) Breakaway: a 1/8 in G10 root plate glued under the tail (skin lapped onto it) sits on a 1/8 in G10 saddle glued and glassed on the boom's last 4.2 in, both 2.3 in wide; two 1/8 in 316 dowels locate it; four nylon #10-24 bolts through the flanges outside the boom (0.95 in each side of center, at 0.8 and 3.4 in back of the tail's front) hold it. A side push tips the tail over the saddle's edge against the two far bolts, 2.1 in away: about 210 lb each dry (12,000 psi, CDE's figure), about 150 lb wet, so it lets go at about 45 to 65 lb at mid-span, the Stage 12 band; a 15 lb gust puts about 50 lb on a bolt. Nylon bolts replaced at every refit. (3) Hinges: three 2 in pieces of 316 piano hinge (0.04 in leaf, 3/4 in open, Amazon B09J23T1VM) glued into 3/8 in deep saw cuts on the center line of both cut faces; the flap's front edge sanded to a 35 deg point so it swings 30 deg; hinge axis in the 0.05 in gap. (4) Flap spring at the flap's bottom end, changed from (ss): the spring is McMaster 9287K47 (9.75 in-lb at 180 deg) wound 74 deg, so it holds 4.0 in-lb, the Trim Guide's 4 lb on a 1 in arm (the earlier sheet's 3.1 in-lb was low); 5.6 in-lb at 30 deg (1.41 times); the axle is glued into a 1/4 in G10 flap plate, not into the flap; the arm points forward and to one side, 0.75 in out, and a 1/4 in G10 drag link 1.6 in long joins it to the pushrod's clevis, so the pushrod stays straight down the boom's middle (Z -0.875, the middle of the boom's end); the drag link's pivot on the arm is a nylon #4-40 screw that breaks if the tail is shed. (5) Left out: the centering spring (a free flap trails straight in the wind, and the flap spring is far stronger) and the flap stops (the actuator's limits set the travel, and the flap spring gives first). Weight at the tail beyond the bare tail about 0.6 lb (a guess from sizes), each pound there about 4-3/4 lb of nose lead. Mike's yes asked 2026-10-29. (vv) Pushrod straight to the horn, 2026-10-29 (Mike: "simpler if push rod went directly to tab horn. need to offset (bend) the last 12-18 inches of pex/rod to the side"; "horn should 90 degree from longitude"; "maybe ~88 deg to optimize if desired"). Worked out: a smooth S-bend that moves the rod 3/4 in sideways over a length L bends the rod to a radius of L squared over 3 in, so E times half the rod's width over that radius: 1/4 in rod over 18 in about 32 ksi, over 15 in 47 ksi, over 12 in 73 ksi; 3/16 in rod over 18 in about 24 ksi, over 15 in 35 ksi; 1/4 in rod over 24 in about 18 ksi. Annealed 316 rod takes a set at about 35 ksi (a seller's yield figure), so: 3/16 in 316 rod with an 18 in bend (or 1/4 in rod with a 24 in bend). The rod is held in the PEX the whole way, so 3/16 in cannot buckle at under 8 lb of push (inferred); the bend adds a few lb of rub at the actuator (a guess), which does not change the flap spring's setting since the spring is at the flap. The 3/4 in offset puts the PEX at the boom's side face, so its last part slants out through the side wall and runs along the side face, glued and glassed over (the Stage 12 conduit along the boom's side). Horn at 90 deg to the boom at flap-straight (Mike); the S-bend ends parallel to the boom, so 90 is right, and if the rod ends at a small angle the horn is set square to the rod (Mike's about 88). The drag link and its nylon pivot drop out; a shed tail would then hang on the pushrod, the wing still free to feather, and the tail too far back to reach the wing. Clevises for 3/16 in rod: McMaster 6071K61, 316, 10-24 thread, 3/16 in jaw, 1-9/16 in long, pin and cotter included; so the horn's tip and the G10 tongue at the actuator are 3/16 in thick, and the rod's ends get a 10-24 thread (a hand die). Rod: a 6 ft piece of 3/16 in 316 (the Amazon listing found is out of stock; McMaster, part to confirm). Decided 2026-10-29 (Mike: "yes, draw it with details"), with one change found while drawing it: a smooth S (no kink, so the rod's bend barely changes as it slides and it does not wear out from bending back and forth) that also drops the rod 1/4 in to the horn's level peaks higher than the two-arc estimate above: about 32 thousand psi over 18 in for 3/16 in rod, too close to annealed 316's 35. So the bend is 24 in long: about 18 thousand psi (worked out). Drawn in Reference/Tail_Assembly_2026-10-29 (three pages, rebuilt): PEX in the boom's middle, then over the boom's last 24 in a smooth S out 3/4 in to the side face and down 1/4 in; for its last 13 in it cuts into the core's side and for its last 9 in it stands up to 1/4 in proud of the side face, faired in with thickened epoxy and glassed with one extra layer 3 in past each end; horn at 90 deg; McMaster 6071K61 clevises at both ends (horn tip and actuator tongue 3/16 in thick); rod ends threaded 10-24 with a hand die (Amazon B005ENUDLE). The rest of the tail end (root joint, hinges, flap spring) as (uu), still proposed. Pushrod back to 1/4 in, decided 2026-10-29 (Mike: "would feel better with 1/4 pushrod. is there not enough play in the pex to allow movement?"): yes, the play helps a little. The 1/4 in rod has about 0.11 in of room in 3/8 PEX (0.36 in inside), so it cuts the curve by about that much: over the 24 in bend it peaks at about 20 thousand psi instead of 24 (worked out), well under annealed 316's about 35. (Over 18 in it would be about 36 to 42: no.) Parts back to the 1/4 in set: McMaster 6071K47 clevises (316, 1/4-28, 1/4 in jaw and pin, 2 in pin to end) at both ends, the horn stays 1/4 in G10 with no sanding, the G10 tongue at the actuator stays about 0.22 in for the actuator's fork with a thin washer in the clevis's jaw, rod ends threaded 1/4-28 (die with handle, Amazon B0FS1RN49Q), a 6 ft piece of 1/4 in 316 rod, cold drawn (the two 24 in pieces on hand are too short). The stiffer rod rubs a little more in the bend, a few lb more for the actuator (a guess), against its 56 lb. Tail drawing rebuilt with the 1/4 in rod. *Documents: Wing Construction Guide Stage 12 (tail mount, hinges, breakaway, centering spring and stops); Trim Actuator Guide sec 6 (the fuse moves to the flap), sec 7 (tests).* (ww) Front end of the pushrod, and the tail's height on the boom, 2026-10-29 (Mike: "clevis on the front? how to connect to thomson?" and "your tail assembly looks like tail is all above boom? is that a change?"). Front: the Thomson MD's front end is a fork, the only end Thomson sells (catalog pp. 72 and 73: 6.0 mm (0.236 in) slot, 10.1 mm (0.398 in) hole for a 10 mm pin, hole 11.4 mm (0.449 in) from the tip, 22 mm (0.866 in) round; the A025 screw is 250 N (56 lb)). A fork cannot hold a fork, so a flat G10 tongue goes between: the Thomson's 10 mm pin through its front hole, the pushrod clevis's 1/4 in pin (McMaster 6071K47, the same clevis as at the horn) through its back hole. Proposed: the tongue cut from the 4 mm (0.157 in) G10 sheet of the Aug 9 Amazon order, 5/8 in wide by 1-5/8 in long, holes 0.9 in apart (10 mm front, 1/4 in back), thinner than both slots so no sanding and no washer; it may slide a little along the pins, which adds no play along the rod, so no tab wobble. This replaces "about 0.22 in with a thin washer" in (vv). Two parallel pins make a knee that could fold sideways under push; it cannot here if the PEX starts right behind the clevis's travel (worked out: with about 2.2 in of bare rod in front of the PEX mouth and the 2 in clevis, the rod holds the clevis pin sideways at about 240 lb per inch, against about 62 lb per inch that the actuator's full 56 lb push over the 0.9 in tongue tries to fold it; about 11 at normal loads). So the boom's front changes: the old spring link's 1 in CPVC tube shrinks from about 9-1/2 in to about 2-1/2 in, just room for the tongue and clevis to travel (clevis back end at about 21.3 in fully in, 23.3 fully out, on the build sheet's measure), and the PEX starts at its back end. Still to pick: the 10 mm front pin, under about 1.2 in long overall so it fits inside the boom's 1-1/4 in front hole (an R-clip or cotter pin is too long; it needs a ring-grooved pin; not yet found and checked). The rear pin stays McMaster 93890A742. To redraw on the boom pocket build sheet once the tail end is decided. Tail height: the tail drawing of (uu) stands the tail on top of the boom, which is a change from the July Wing Rig Drawing Set (rev H, sheet 3) and the Oct 27 boom root drawing, both with the tail centered on the boom, half above and half below; the guides' words do not say which. Not flagged to Mike when drawn; flagged now. Why on top (Claude's opinion): the boom's end is 1-1/2 in wide but the tail is only 1.32 in thick, so a centered tail must be built in two halves around the boom (two root joints, two breakaways, a split flap); on top it is one plate, one breakaway, one flap, with the flap spring and horn at the boom, and the tail sits 13-1/2 in higher off the waves. The cost: the tail's side push acts about 13-1/2 in above the boom, so it twists the boom a little (estimate: about 2 degrees at the tail in a 50 knot gust, which only leans the tail and does not change its angle to the wind), and the breakaway bolts see that lever, already counted in (uu)'s 45 to 65 lb at mid-span. The tail's top is then about 75 in above the wing root, under the 78 in tip. Mike's yes or no asked 2026-10-29: keep the tail on top. (xx) Tail kept at its designed height, decided 2026-10-29 (Mike: "If the tail needs to go on top or on bottom of the boom, either way, then I would think the boom should be raised or lowered half of the tallness of the tail to keep the tail where it was designed to be anyway, because that screws up the design. And you making these changes without even mentioning it ... it's bad."). The tail's middle stays at about 47 in up the wing, as the July drawing set and Stage 10 have it (the tail 9 to 10 ft above the water). Claude's pick under Mike's "either way": the tail on top of the boom, as drawn, and the boom with its hardpoint zone, clamp and actuator pocket moved down 14-1/4 in, from 47 to about 32-3/4 in: half the tail's 27 in, plus the boom's half depth at the tail end (5/8 in) and the saddle and root plate (1/4 in). 14-1/4 in is 19 rib slices of 3/4 in, so the pocket edges stay on glue lines (43.5 to 48.75 becomes 29.25 to 34.5). Why on top rather than hanging below: the pocket comes 14 in closer to the deck, easier to reach; the wing is stiffer nearer the mast top (about 25 in); the flap spring and horn stay at the flap's bottom end as drawn. Checked: the hardpoint zone (about 29-3/4 to 35-3/4 in) is above the mast socket (24 in) and only its bottom 2-1/4 in is in the 24 to 32 in blend band, which changes only the outline; static balance is unchanged (the twist depends only on how far back things are, not how high); the boom now sweeps at about the height the centered tail's bottom already swept, and the guides keep the post and antennas out of that circle (Build Guide; Instrument Post Guide: more than 7 ft clear). Every "47 in" for the boom root and actuator moves to about 32-3/4 in: Wing Construction Guide Stage 10 (boom root zone; "roughly 60 percent of the way up" becomes about 42) and Stage 12 (boom bolting; the breakaway line's "47 in mounting height" becomes the tail's height), Trim Actuator Guide sec 3, Wing Power System (two places), Build Guide sec 6.5, Design Decisions (the lightning reason), Lightning Guide (the strap paragraph); the boom pocket build sheet and its drawings to redraw at the new height. Claude did not flag the move of the tail on top of the boom when drawing it; Mike was right to call that out. (yy) Pushrod front end with no knee, proposed 2026-10-29 (Mike: "Is there something that could screw onto the push rod that would connect to the Thompson that we don't have to build from scratch and fashion? ... we don't want a knee"). Searched: nothing sold screws onto a rod and fits the MD's fork (6.0 mm slot, 10 mm pin). Closest found: McMaster's 316 gas-spring eyelet, M6, 0.24 in thick with a 0.24 in hole, too thick for the slot and too small for the pin; rod-end bearings are far too wide in the ball. The fork is made for a flat plate. Proposed: the front clevis (McMaster 6071K47, the same as at the horn) with a flat tab of 4 mm (0.157 in) G10 glued into its jaw with thickened epoxy, its 1/4 in pin in place, so clevis and tab are one stiff rod end that screws onto the pushrod and takes the Thomson's 10 mm pin. One pin at the front, so no knee: the only other pin is the actuator's rear one, 6.5 in back, and the actuator is far too long a lever to fold (worked out: about 9 lb per inch of folding push against about 240 that the rod holds). Tab 5/8 x 1-1/2 in, holes 0.9 in apart, the 10 mm hole 5/16 in from the tab's front end (pull tears out at about 700 lb, worked out, against the actuator's 56); drill the holes in a larger piece first, then cut the tab round them. The tab sits about 0.08 in loose side to side in the fork, which adds no play along the rod. The pushrod cannot unscrew: the tab in the fork holds the front clevis and the horn holds the back one. Replaces (ww)'s loose tongue. Mike's yes or no asked 2026-10-29. Still to pick: the 10 mm front pin, under about 1.2 in long overall. (zz) Pushrod front end looked at again, 2026-10-30 (Mike: "Actuator to rod ? Surely you can do better."). Searched again: cylinder rod eyes, gas-spring eyelets, ball rod ends (stainless 1/4-28 ones are about 0.375 in wide), rigging eyes, aircraft rod ends, actuator brackets. None screws onto a rod and fits the MD's 6.0 mm slot with a 10 mm pin; Thomson makes both ends forks for a flat plate the builder supplies. Worked through and set aside: (1) the actuator out at the tail, its forks straight onto the horn and a tab on the boom, no pushrod, PEX bend or wing pocket at all; set aside because Stage 10's 270 in-lb of nose twist (about 21-1/2 lb of lead at 12-1/2 in) would grow by about 55 to 85 in-lb, to roughly 26 to 28 lb against the rack's 24 lb capacity (estimate; Reference/Wing_Actuator_Options_2026-09-08 sec 25 gives 14 to 20 lb fitted, 24 capacity), the swing inertia would rise roughly a fifth (estimate), and the actuator would sit in the spray at the tail, the most lightning-exposed spot, with its cable back into the wing; (2) a 2 to 1 lever at the front (halves the actuator's end play at the tab, but adds a pivot, an axle and its mount). Proposed: the rod ends in a flat blade, the (yy) clevis and 4 mm G10 plate, now locked by the clevis pin plus one 1/8 in 316 dowel (the pack already listed, Amazon B00R1UV7QS), drilled through clevis jaw and plate together with the clevis pin in; no glue. Mike's yes or no asked 2026-10-30. Also found on the MD datasheet (Electrak MD technical features): end play up to 1.2 mm (0.047 in), internal anti-rotation, restraining torque 0. At the 3/4 in horn that end play is about 3.6 deg of flap free play in all, about 1.8 each way, unless the linkage is kept pushed one way; Wing Construction Guide Stage 12 already asks for "a light one-direction preload so it never sits unloaded", and the (uu) tail drawing has no preload (its centering spring and stops were left out). Part 1 work: put a light one-way preload back in the tail drawing. (aaa) Looks, Saildrone style, proposed 2026-10-30 (Mike: "Let's think about looks. We want to sell this as cool usv, what looks best for wing and tail. Look at saildrone ... very cool."). Found: US patent 9,371,121 "High-tailed wing sail" (Jenkins and Seyfang; Google Patents lists Saildrone as the current owner) puts the tail behind the wing with, in one form, its top higher than the wing tip, placed to use the wing-tip swirl, which it says can cut the tail area needed. Saildrone's USVs carry high-visibility wing colors. Claude's opinion: line the tail's top up with the wing tip, so tip, boom and tail read as one shape: boom at about 50 in (tail on top, its top at about 78 in, the wing's span), the hardpoint zone and pocket 3 in above the guides' 47 in (four rib slices), replacing (xx)'s 32-3/4 in if Mike says yes. Height does not change the balance (only distance back counts) or, as far as worked out, the tail's steering power, except that the tip swirl may add some (the patent's claim; not worked out here). The cost: the tail's top 17 in higher than (xx), so more exposed to lightning (rough rolling-sphere estimate: both tail heights are outside the tip rod's cover, the higher one by more); the tail is the part the design can lose. Going above the tip as Saildrone does would make the tail the lightning target: not proposed. Paint: the Wing Construction Guide picks white to keep the foam cool (sec on finishing: white takes 20 to 25 percent of the sun, dark 90 to 95; the foam softens at about 158 F), so color only as stripes or lettering. Mike's yes or no asked 2026-10-30: tail top level with the wing tip. (bbb) Withdrawn 2026-10-30: nothing here is decided. A "yes to all three" (tail end as drawn, pinned end on the pushrod, tail top level with the wing tip) was sent by the app's auto reply, not Mike (Mike: "Didn't mean to say yes to 3, the auto thing did it."). The Buy list rows added for it are marked ON HOLD. (ccc) Mike's points, 2026-10-30: "Nylon is too flimsy"; "No idea how pin keeps blade straight. Blade is stupid name" (say "flat tab"); "Does tail above boom have enough support? Less cantilevered stress when centered." Worked out: centered, each half of the tail sticks out half as far with half the push, so each root bends about a quarter as much, and the boom is not twisted; on top, the root takes the full bend and the boom twists (estimate about 2 deg in a 50 knot gust), though the tail's root glass is still at roughly a ninth of its strength then (rough estimate). Nylon's weak points besides its planned break: it soaks up water and loses about a third of its strength, creeps under steady load and tires under years of gusts. Proposed: the tail centered on the boom as the July drawing set had it, the boom's end in a slot across the tail's front middle, the tail bolted to the boom's top and bottom faces, the flap one piece behind the boom's end; the breakaway bolts in PEEK (a plastic that takes up very little water and creeps far less than nylon, at about the same strength). Mike's yes or no asked 2026-10-30 to draw it that way. (ddd) No breakaway, decided 2026-10-30 (Mike: "I don't want to build in weakness which will increase chance of failure. Whether boom or tail breaks, unmanned who cares. The goal is nothing breaks."). No designed weak link anywhere in the tail mount: the tail bolts on with 316 stainless, sized to the biggest load it will really see, and the boom root is sized to match. Read with Mike's "Less cantilevered stress when centered", the tail goes back centered on the boom as (ccc) describes (told to Mike in plain words the same day); PEEK and nylon bolts are dropped and the nylon rows on the Buy list are marked not needed. To change: Wing Construction Guide Stage 12 (the "breakaway link that snaps on purpose at roughly 40 to 60 lb" line) and Stage 10 (the boom root "In a tail strike the cheap breakaway link at the tail lets go, never this joint"); Design Decisions if it names the breakaway. Still to work out with the drawing: the design loads (a gale gust with the tab hard over, and what a knockdown that dunks the tail does). Mike also asked whether the clevis has room for the flat tab's second pin, far enough back to be strong. McMaster's table for 6071K47 reads 2 in pin to end, 3/4 in thread, 1/4 in pin, 1/4 in jaw, 5/8 in wide, and 1-1/4 in, which by the arithmetic (1-1/4 plus 3/4 is 2) is the slot's depth behind the pin (inferred; drawing not seen). So the second pin goes about 5/8 in behind the first, with the tab running on about 1/4 in past it. Its only job is to stop the tab turning, a few lb at most (estimate); a 1/8 in 316 pin takes several hundred lb in shear and the 4 mm G10 about the same in bearing (worked out roughly), so the margin is large. (eee) Centered tail drawn, 2026-10-30 (Mike: "Yes draw"): Reference/Tail_Centered_2026-10-30 (.html, .pdf; built by Reference/_make_Tail_Centered_2026-10-30.py), three pages, proposed. The tail in two halves of 18 slices each (13-1/2 in), one on the boom's top face and one on its bottom, each inner face 3/4 in from the boom's center line; a 1/8 in G10 root plate, 3 x 4-1/2 in, glued to each half's inner face; four 1/4-20 316 bolts with 316 tube spacers and nyloc nuts, 1-1/4 in each side of center, clamp the plates onto the boom, whose last 4-1/2 in is filled solid; two 1/8 in 316 locating pins per plate. The flap in two halves tied by the 3/8 in axle, glued into a 1/4 in G10 end plate on each flap's inner end (set 1/32 in clear of the tail); two piano hinge pieces per half. The flap spring and the 3/4 in horn sit on the axle in the gap between the flaps, level with the boom: upper end plate, spring, horn, thin washer, lower end plate. The horn stays 3/4 in because a shorter one puts the clevis into the spring's coil (a 1/2 in horn was drawn first and collided). Pushrod S over 24 in, out 3/4 in and down about 1/3 in; about 21 thousand psi with the rod's room in the PEX. Worked out: each half's root bends about 3.8 times less than the tail on top; with no weak link, the boom root's weakest spot (about 162 lb at the tail, inferred from the Oct 28 calc) is the limit, and there each far tail bolt pulls about 150 lb against about 2,400 for a 1/4-20 316 bolt. Still to add: the light one-way preload. Replaces the on-top tail of Reference/Tail_Assembly_2026-10-29 if Mike says yes; his yes or no asked 2026-10-30. (fff) Tail stays one piece, decided 2026-10-30 (Mike: "Making the tail 2 pieces looks horrible and loses strength of being joined. Why the weird change? ... First tail all above, now split??"). The split in (eee) came from keeping the flap spring (the give-way) at the horn, level with the boom; it was not flagged before drawing, and Reference/Tail_Centered_2026-10-30 is set aside. Mike also: "Consider a 1\" horn. Less force needed." Worked out: a 1 in horn needs 3/4 of the pushrod force of a 3/4 in one, uses about 1 in of the actuator's 2 in travel instead of 3/4, and cuts the wobble from the actuator's up to 0.047 in of play to about 1.3 deg each way; the cost is that the pushrod must end 1 in out to the side, so its S-bend grows to about 30 in to keep the rod near 17 thousand psi (about 27 over 24 in, with its room in the PEX). Proposed, Mike's yes or no asked 2026-10-30: back to the July drawing set's layout (sheet 7, tab hinge detail): the tail one piece, centered on the boom; the 1 in horn fixed on the flap at the boom's height, standing out of its side; the pushrod along the tail's side to it; the give-way back in the pushrod as the guides have it (which spring link to pick next), and the light centering spring at the hinge. (ggg) One-piece tail and its attachment drawn, 2026-10-30 (Mike: "We want to make things better, not worse"; "Yes. How to attach tail to boom?"): Reference/Tail_One_Piece_2026-10-30 (.html, .pdf; built by Reference/_make_Tail_One_Piece_2026-10-30.py), three pages, proposed, replacing (fff)'s "give-way back in the pushrod" (that needed the made spring link). Its first page lists every change from the guides: (1) a slot in the tail's front, 1-1/2 in tall (the two middle slices) and 4.6 in deep, lined top and bottom with 1/8 in G10, for the boom's end, whose last 4-1/2 in is filled solid; (2) two 1/8 in G10 clamp plates, 3.2 x 4.5 in, bedded in thickened epoxy on the tail's sides over the slot, flat against the boom's sides (release tape on the boom so the tail can come off), and three 1/4-20 316 bolts through plates and boom, 0.3 in above its middle, heads on the horn's side, nyloc nuts; (3) no breakaway; (4) the give-way spring at the flap, in a pocket 1-1/2 in tall and 1.9 in deep in the flap's front edge at the boom's height, with G10 end plates glued in its top and bottom and the axle glued in both, plus a 1/2 in notch in the tail's back edge for the spring; flap one piece; (5) horn 1-1/8 in, so the pushrod clears the clamp plate by 1/8 in and runs under the bolt heads; (6) the PEX's last 16 in run out along the boom's side, up to 5/8 in proud, faired, ending 0.2 in ahead of the clamp plates; S-bend 30 in, about 20 thousand psi with the rod's room; (7) four hinge pieces, two above the pocket and two below. Worked out: worst roll on the joint, all of the boom root's about 162 lb limit on one half of the tail, about 1,150 in-lb, puts about 420 lb on each bolt against about 2,400. Mike's yes or no asked 2026-10-30. (hhh) Tail 2/3 above the boom, 1/3 below, decided 2026-10-30 (Mike: "Would 2/3 above and 1/3 below look better (cooler)?" then "2/3 yes"). The one-piece tail of (ggg) moves up: 12 slices below the slot and 22 above, so it reaches 9-3/4 in below the boom's center line and 17-1/4 in above; the slot, clamp plates and the spring's pocket stay level with the boom. Worked out: the upper part's root bends about 1.6 times more than centered, about 2-1/2 times less than the tail on top; the boom gets a small twist again (the tail's middle 3-3/4 in above the boom); worst case each clamp bolt pulls about 525 lb against about 2,400; the tail's top is about 14 in under the wing tip. Mike also asked that the drawing show parts at their true size (the rod was drawn as a hairline; the clevis is really 5/8 in thick and about 2-1/4 in long). Redraw waits on the hinge question asked the same day: one full-length 316 piano hinge in two long pieces (above and below the pocket), glued on one straight pin so they line up, instead of four 2 in pieces (the guides say three pin hinges). (iii) Hinges: four short pieces, decided 2026-10-30 (Mike: "No to long hinge, don't need to overkill"). Four 2 in pieces of the 316 piano hinge (Amazon B09J23T1VM), two above the spring's pocket and two below, each leaf glued with thickened epoxy into a 3/8 in deep saw cut on the center line of the tail's back edge and the flap's front edge, the epoxy squeezing through the leaves' screw holes; the gap between tail and flap as wide as the hinge's round middle (about 1/8 in, an estimate). Reference/Tail_One_Piece_2026-10-30 redrawn the same day: tail 2/3 above the boom (hhh), every part at its true size (rod, its tube and the clevis), hinges at about 3-3/4 and 14-1/2 in above the boom's center and 2-3/4 and 8 in below; its first page lists every change from the guides. Mike's yes or no on the whole tail asked 2026-10-30. (jjj) Is epoxy enough for the hinges and the flap? 2026-10-30 (Mike asked). Worked out: the flap carries a few lb in normal sailing; the give-way spring caps it near full deflection at about 6 lb (5.6 in-lb at about 0.9 in behind the hinge); a hard gust with the whole tail at an angle might put 30 to 40 lb on the flap (a guess), about 10 lb a hinge. Each leaf is glued over about 1-1/2 sq in (2 in long, 3/8 in deep, both faces): at a cautious 500 psi for epoxy on sanded stainless after years wet (an estimate) that is about 750 lb a leaf. The weak spot is the foam the slot is cut into: H80 crushes at roughly 200 psi (the maker's figure as remembered, not checked today), about 150 lb a leaf, still about 15 times the load. Part 1 work, added to Reference/Tail_One_Piece_2026-10-30 build steps: a thickened-epoxy hard spot, about 1/2 in deep and 2-1/2 in long, at each hinge in the tail's back edge and the flap's front edge, and under the pocket's end plates, before glassing; leaves sanded 80 grit and wiped with acetone; epoxy through the leaves' screw holes locks them even if the glue let go of the steel. (kkk) One-piece tail decided 2026-10-30 (Mike: "yes 1 piece"): the tail as Reference/Tail_One_Piece_2026-10-30 draws it, with every change its first page lists (2/3 above the boom; the slot and two bedded G10 clamp plates with three 1/4-20 316 bolts; no breakaway; the give-way spring in a pocket in the flap's front edge; horn 1-1/8 in; the PEX out along the boom's side for its last 16 in; four short hinge pieces, with epoxy hard spots). Reference/Tail_Assembly_2026-10-29 (tail on top) and Reference/Tail_Centered_2026-10-30 (split tail) are superseded. Order tracker Buy list updated the same day (hinge, axle, die, spring, clevis, rod and the clamp bolts decided; collar not needed; the 1/8 in dowel and the second clevis wait on the actuator end). Still owed by Mike: the pinned flat tab at the actuator end (zz). Part 1 work: write the tail into Wing Construction Guide Stage 12 and the Trim Actuator Guide. (lll) Cat 3 check, 2026-10-30 (Mike: "needs to hold in cat 3 wind"). Cat 3 is 111 to 129 mph sustained; gusts to about 160 mph (a guess at the gust factor). Tail push, estimated with the Oct 28 calc's lift slope: about 80 lb in a 10 deg gust at 160 mph, about 135 lb at 20 deg, about 160 lb if the gust hits it broadside (a sudden 90 deg shift in the eyewall, the feathered rig not yet swung). The tail's clamp holds it: about 525 lb on a bolt against about 2,400. The flap's share broadside, about 40 lb, is about 10 lb a hinge against about 150 in bare foam and far more in the hard spots. The boom itself is good for about 590 lb at the tail. The weak spot is the boom root: the Oct 28 calc's glued stack (skin, doublers and the 1/8 in G10 plate) pushed out between the bolt rows gives at about 160 lb at the tail with no clamp friction counted, so a broadside Cat 3 gust takes it to its limit. Proposed: 1/4 in G10 plates there instead of 1/8 (Mike chose 1/8 on Oct 28 as "plenty"); the stack goes from about 0.38 to 0.5 in thick and its bending strength up about 1.8 times, to about 285 lb at the tail (worked out roughly from that calc). Mike's yes or no asked 2026-10-30. (mmm) Actuator end drawn, 2026-10-30 (Mike: "clevis drawing or pic please"): Reference/Actuator_End_2026-10-30 (.html, .pdf; built by Reference/_make_Actuator_End_2026-10-30.py), one page, full size, two views. The flat tab of (zz) and (ddd): 4 mm G10, 5/8 x 2-1/8 in, holes 10 mm (5/16 in from the front end), 1/4 in (0.9 in back) and 1/8 in (5/8 in behind that); held in the McMaster 6071K47 clevis by its own pin and a 1/8 in 316 pin cut to 5/8 in, drilled through clevis and tab together; into the Thomson's fork on its 10 mm pin. The clevis jaw past its pin and the Thomson's slot depth are guesses (not given by the sellers). Mike's yes or no still owed. (nnn) Two decided 2026-10-30 (Mike: "yes flat tab", "yes 1/4"): (1) the actuator end as Reference/Actuator_End_2026-10-30 draws it, the 4 mm G10 flat tab held in the 6071K47 clevis by its own pin and a 1/8 in 316 pin, on the Thomson's 10 mm pin; (2) the boom root's two glued G10 plates go from 1/8 in to 1/4 in (from the sheet on hand), so the glued stack holds about 285 lb at the tail, past a broadside Cat 3 gust's about 160 lb (lll). Buy list updated (the 1/8 in pin and both clevises decided; the 10 mm front pin and the 1/4 in rod still to pick). Part 1 work: redo Wing_Boom_Pocket_Build_2026-10-29 (top of the folder) for what is now decided: 1/4 in root plates, no spring link (the give-way is at the flap), the flat tab end, the pushrod straight from the actuator, and the boom root strength numbers in Reference/Boom_Root_Strength_Calc_2026-10-28 with no breakaway; then write the boom, tail and trim drive into the Wing Construction Guide (Stages 10 and 12) and the Trim Actuator Guide. (ooo) Boom pocket build sheet, third edition, 2026-10-30 (Mike: "yes" to redoing it and picking the two parts): Wing_Boom_Pocket_Build_2026-10-30 (.html, .pdf, top of the folder; built by Reference/_make_Wing_Boom_Pocket_Build_2026-10-30.py), renamed from the Oct 29 edition. Its first page lists each change: no spring link (the boom's front 1 in tube cut to 2-1/2 in, 21 to 23.5 in back, just room for the clevis and flat tab; the PEX starts behind it); the flat-tab actuator end; 1/4 in root plates (bolts 1/4 in longer: front pair about 4 in, back pair about 3-1/2); no breakaway, the strength table on Cat 3 loads (broadside gust about 160 lb, x 1.5 = 240), scaled from the Oct 28 calc: weakest the skin stack between the bolts, about 285 lb, 1.8 times; boom 48-1/2 in long (not 50), so with the tail's slot its lift center is the guides' 58.8 in behind the pivot; pushrod about 49-1/2 in (inferred); the boom root at the guides' 47 in up (the 32-3/4 in of (xx) is void since the tail went centered, then 2/3 above). New and needing Mike's yes: a short tinned copper braid from the actuator's rear pin to the front clevis pin, because the G10 tab breaks the metal path the lightning jumper relied on. Parts picked: front pin McMaster 93890A741 (10 mm, 24 mm between grooves, rings included), both ends cut back to about 1/16 in past the grooves so it is about 1-1/8 in long inside the 1-1/4 in bore; pushrod McMaster 89325K113 (1/4 in 316, cut to order 13 to 90 in; order 54 in), part number from McMaster's listing as found in a search, its page not opened. Order tracker rows updated. The one-way preload is the Trim Guide's anti-rattle spring, already on the sheet (pocket back wall to the actuator's front end). (ppp) Lightning braid past the G10 tab, decided 2026-10-30 (Mike: "yes"): about 12 in of the 1 in tinned copper flat braid on hand (Amazon, Sept 2, 1 in x 5 ft, Lightning Guide parts list "jumpers only") from the actuator's rear pin (end folded, drilled and slipped on beside the fork under the ring clip, with the guide's wing-strap jumper) to the pushrod's front clevis (held round the clevis body by a small 316 hose clamp), with a loop for the clevis's 2 in of travel; the G10 tab breaks the metal path otherwise. Nothing to buy. On Wing_Boom_Pocket_Build_2026-10-30 page 3. To write into the Lightning Guide sec 5.1 (step 4, the braid jumper) with the rest of the boom write-in. (qqq) The braid in (ppp) does not work; metal tab proposed in place of the G10 tab, NOT decided (Mike and Claude, 2026-10-30, late). Mike: "That braid seems to large for the purpose"; then asked about 8, 10 or 12 gauge wire with lugs, then "what about using metal there and avoiding the braid?". Found, worked out from the build sheet sizes: the pushrod clevis pin never leaves the boom's 1-1/4 in front bore (0.3 to 2.3 in inside it over the stroke), so any jumper must squeeze through the 0.19 in gap round the actuator's 0.87 in round front end and rub there; the hose clamp in (ppp) cannot fit either. Also, the G10 tab leaves only about 0.15 in of air between the clevis jaws and the actuator's front tip (jaw length a guess), which a strike jumps, so the G10 does not really keep a strike out of the actuator. Claude's pick, waiting on Mike's yes: a 316 stainless tab, 3/16 in thick, same holes and length as the G10 tab, and no jumper at all. A tail strike then runs pushrod, tab, actuator, rear pin, the guide's strap jumper, strap, and likely kills the actuator (about $415, tracker), which the Lightning Guide sec 5.1 already accepts for a direct hit. Width: 5/8 or 3/4 in fits; 1 in does not (its corners foul the 1 in tube the tab's back end enters at full stroke). Mike has no 4 mm G10 to hand (he has 1/8 and 1/4 in; the 4 mm came from the Aug 9 Amazon order in the order log), so if the G10 tab stays it is the 1/8 in sheet. Stock not picked: no 3/16 in 316 bar found on Amazon; Online Metals lists 3/16 x 3/4 in 316 (their part 649), price not checked. Actuator rear hole, asked by Mike: 10.1 mm for a 10 mm pin, 0.236 in slot (Thomson catalog). Once decided, fix Wing_Boom_Pocket_Build_2026-10-30 (page 3 lightning paragraph and CHANGES item 7 still say the braid and hose clamp) and Reference/Actuator_End_2026-10-30 (says 4 mm G10). (rrr) Protecting the actuator so the wing keeps working: two proposals, NOT decided (Claude, 2026-10-30; Mike asked "what is the best way to protect the actuator from lightning so the boat can keep using the wing"). (1) A tip-rod strike, the common case, is the bigger threat to the actuator: its case is tied to the strap but its wires are tied to the hull about 10 ft below, so a strike puts tens of kV between case and wires (inferred). Proposal: a tinned copper tubular braid sleeve, 3/8 in, about 8 ft, over the actuator cable, tied to the strap jumper at the actuator and to the mast where the cable enters it, so the wires ride with the strap and the wing box surge parts take the difference. Fits the 3/4 in PEX beside the wing cable only if the wing cable (not bought) stays about 0.3 in across. It makes a loop in the lightning copper. Mike, on the link Claude gave (a 50 ft Jameco roll, about $140): "$140 for a few in copper strap ... from a vendor website ive never heard of?" Any pick must be a short length, Amazon first. (2) Tip rod about 12 in taller (24 in above the tip instead of the 12 in Mike picked 2026-10-03): rough 100 ft rolling sphere, tail top about 26 in below the rod top and 54 to 66 in aft of it, puts the tail about 7 to 18 in outside the rod's cover now and inside it with the extra foot; needs a new 30 in bar, about 2/3 lb more at the tip. Mike was asked "work both into the guides?" and has not answered. (sss) Stock found for the metal tab, and the cable sleeve priced, 2026-10-30 (Claude, cloud session; nothing decided). Searched again for 3/16 x 3/4 in 316 bar: still none on Amazon. McMaster, where the pushrod rod, front pin, clevis and give-way spring order already goes, lists 316 flat bar 1/8 x 3/4 in as 8992K794 and 3/16 x 1/2 in as 8992K136 (part numbers read from a web search of McMaster's catalog; McMaster's own pages would not open from this session, so confirm size in the cart; prices unchecked); no 3/16 x 3/4 in found there. A 1/8 in 316 tab is plenty: tear-out through the tab's front end at the 10 mm hole about 1,200 lb against the actuator's 56 lb push (worked out), and its extra looseness in the Thomson's 0.236 in slot (about 0.11 in, against the 0.08 in already accepted for the 4 mm G10) is sideways only and adds no play along the rod. So if Mike says metal, Claude's pick narrows to a tab cut from the 1/8 x 3/4 in 316 bar on the McMaster order, 3/4 x 2-1/8 in, same holes as Reference/Actuator_End_2026-10-30. Online Metals part 649 (3/16 x 3/4 in 316) stays the fallback; its page would not open, price still unknown. For (rrr) proposal 1, the sleeve over the actuator cable: Electriduct 1/2 in tinned copper braid sleeve, 10 ft, Amazon B01BIBQ940 (nominal 0.32 in across, expands to slide over an about 0.3 in cable; 10 ft covers the about 8 ft run); Amazon's price not readable from this session, Walmart's matching Electriduct listing shows about $23. With the sleeve snug on the actuator cable, the 3/4 in PEX still only takes it beside a wing cable near 0.3 in across, as (rrr) says. Claude's opinion, told to Mike the same day: yes to the sleeve (cheap, no weight aloft, guards the common case, a tip-rod strike reaching the actuator through its wires), no to the 12 in taller tip rod: it guards only the rarer direct tail hit, which now costs only the actuator the Lightning Guide already accepts losing; it adds about 2/3 lb at the very top; a 24 in free length of the 1/4 x 3/4 in bar takes a permanent bend at about an 11 lb fore-or-aft push at its top, half the 12 in rod's figure (worked out from the bar's section, 304's yield taken as about 35,000 psi), so green water in a knockdown could fold it; and no 36 in bar of the stock was found on Amazon, only Shapiro's 24 in bars, too short for a one-piece 27-1/2 in rod. Mike's three yes-or-nos asked 2026-10-30: metal tab; the sleeve; the taller rod (Claude recommends yes, yes, no). (ttt) Taller tip rod: no, decided 2026-10-30 (Mike: "Don't need taller tip"). Ground wire from the roll on hand proposed in place of the bought sleeve; the McMaster bar is 304, not 316 (correction). Mike's screenshot of the McMaster cart: 8992K794 is "Multipurpose 304 Stainless Steel Bar, 0.125 in thick, 3/4 in wide, hot rolled", $3.39 the 1/2 ft piece; (sss) called it 316 in error. 304 serves for the tab: the tip rod lives fully exposed at the very top in the same 304, the tab sits sheltered inside the pocket, and strength is unchanged. Shipping: the bar alone shows $13.88 shipping against the $3.39 part (Mike: "Amazon works better for small items"), but it would not ship alone: the clevis (6071K47), the 10 mm pin (93890A741), the give-way spring (9287K47) and the cut-to-order pushrod (89325K113) are all McMaster-only picks still to buy, so the bar rides in that box; no usable Amazon piece was found in the searches of (qqq) and (sss). Mike's "Why not use existing wire on hand?": agreed, and the sleeve of (sss) is dropped. Proposed instead: about 8 ft of one core from the 12 AWG duplex roll on hand (Wire list A7, 100 ft), run beside the actuator cable, a ring end screwed to the strap jumper at the actuator and to the gusset bolt at the mast base; costs nothing; the wing-box surge parts take what is left. The wire only carries the difference between the cable and the strap, the strap inches away takes the strike itself, so 12 AWG serves though bonding wire is usually 8 AWG or fatter (opinion). The actuator's bought cable (Wire list C4, 18 AWG 4-core shielded) has only a thin foil shield and drain, which cannot carry a surge on its own. Add the 8 ft to A7's measure-before-cutting list at write-in. Told to Mike in plain words, settling the apparent contradiction: the metal tab and the ground wire guard against different strikes. A tip-rod hit, the common case, never touches the actuator but lifts the wing top tens of kV above the hull for an instant, and the actuator dies through its own wires unless they are tied to ride with the strap, which is the wire's job. A direct tail hit, the rare case, runs through the pushrod and the actuator whatever the tab is made of (G10's 0.15 in gap stops nothing), and the metal tab only gives that current a clean path; the actuator's loss there, about $415, stays accepted, and with the taller rod now declined nothing else shrinks that case. Mike's two yes-or-nos asked 2026-10-30: the 304 tab bar in the McMaster box; the ground wire from the roll on hand. (uuu) Would a G10 tab with a larger gap protect the actuator in a tail strike? No, worked out 2026-10-30 (Mike asked). Two reasons. (1) The gap cannot be made big enough. Inside the boom's front the clevis and the actuator's fork are at most 2 to 3 in apart at any design, and a lightning arc crosses air at very roughly 100 to 300 kV per foot (rule of thumb, labeled a rough figure); a strike on the tail stands at millions of volts, so any gap buildable there is jumped for free. Even with a fat copper bypass wire run along the boom so the gap only had to stand the voltage drop along the bypass, that drop is roughly 150 kV for an instant (worked out from about 1.5 microhenries of run and lightning's rate of current rise), still more than a 2 in gap stands. (2) A blocked path does not stop the current, it moves the arc somewhere unchosen: through the PEX and the glued foam and glass of the boom, where an arc flashes trapped gas to steam and blows the structure open, the same reason the Lightning Guide runs the strap outside the skin. The metal tab is choosing the path on purpose: the $415 actuator is the sacrifice, the boom and tail survive. So G10 at the tab buys nothing in a strike and costs the clean path. The only real way to shrink the tail-strike case was the taller tip rod, declined in (ttt); the case stays rare and accepted. (vvv) Do middling strikes exist where a 1/2 or 1 in gap saves the actuator? No, worked out 2026-10-30 (Mike: "Are there not many cases where 1/2 gap would help? or a 1\" gap? Even with damaged glass, if the actuator works, the wing should function"). Strikes do come in all sizes, and a gap does block the small ones; the question is whether any size exists that the gap blocks AND that would have hurt. Walked up the ladder: a 1/2 in air gap stands very roughly 20,000 to 40,000 volts against a lightning-fast pulse, a 1 in gap roughly 50,000 to 75,000 (sphere-gap figures, rough). Events under that (induced surges on the pushrod from a tip-rod strike or a near miss, the pushrod and strap sitting nearly at right angles so they couple weakly) are amps-level for microseconds; through a metal tab they pass along the actuator's steel body and screw and do no harm. Events over that are direct attachments, and even the weakest direct strikes on record, a few thousand amps, stand at hundreds of thousands of volts (from strike records and physics, rough), so they jump 1 in of air without noticing; after the jump the current path is the same as the metal tab's, through the screw to the rear pin, and the actuator's fate is the same. So the gap only ever blocks events that were already harmless, and passes every event that kills: no middle band exists. Correction to (uuu): with the gap between the clevis and the fork the arc jumps metal to metal through air inside the bore, a crude spark gap, so the burst-boom risk was overstated; the real extra cost of G10 is the tab itself. An arc chars and cracks G10 (epoxy-glass carbonizes under arc heat), so after a big strike a G10 tab is likely burnt or split and the pushrod link broken, the flap free, the wing without trim until a refit, on top of the same actuator damage. The metal tab comes through the same event still working. For Mike's goal, the wing keeps functioning, metal is the better bet in every size of strike. Tab and ground-wire yes-or-nos still open. (www) Why no middle band, said sharper, 2026-10-30 (Mike: "why no middle band? That does not sound correct"). A middle-band event would have to be rich in amps (enough current to weld the actuator's screw and bearings) but poor in volts (too few to jump a 1/2 or 1 in gap). Lightning does not make that combination. Serious current flows into the tail only when a channel actually attaches, and an attached channel is the cloud's own wire, standing at millions of volts however small its current: the weakest attached strikes on record differ from the strongest in amps (a few thousand against two hundred thousand) but all carry volts enough to jump a foot of air, let alone an inch (physics and strike records, rough figures). Everything below attachment, charge-ups and near-miss pulses on the pushrod, is starved of energy: doorknob-spark class, amps for microseconds, harmless through a steel path. So strikes vary in how hard they hit, never in whether they can cross a small gap, and the gap's threshold sits uselessly between two populations with nothing in it. Standard lightning doctrine agrees: insulation thicknesses of inches are not counted as protection anywhere on a strike path. (xxx) If the tab stays G10: the wiring that gives a gap a real job, worked out 2026-10-30 (Mike: "If i choose to use g10 with a 1\" gap, please explain how to best wire the lightning protection"). A bare gap has no middle band (www), but a bypass path changes the gap's job: with a copper run from the tail's metal to the boom-root plate (already jumpered to the strap), a tail strike no longer needs to cross the gap to reach ground, so the gap stands only the voltage difference between the two parallel paths, roughly 30 to 40 kV for a typical first stroke (about 1.2 to 1.5 microhenries of run at the median current rise; rough), not the cloud's millions. The scheme: (1) about 50 in of the 1 in flat tinned braid on hand, from the flap axle and horn cluster at the tail forward along the boom's side, faired like the PEX, to the boom-root clamping plate; (2) the G10 tab lengthened so the clevis's metal and the fork's metal sit about 2 in apart along the tab's surface, because an arc creeps along an insulator's surface at about half open air's strength, so 2 in of surface is about 1 in of air, standing roughly 50 to 75 kV dry; a salt-wet surface tracks lower still, so at sea even that discounts; boom front re-laid (tube about 1 in longer, PEX starts later, pushrod about 1 in shorter; tab buckling at the longer length checked, about 630 lb against the 56 lb push); (3) the rear-pin case jumper, the 12 AWG ground wire beside the actuator cable and the wing-box surge parts stay regardless, since the common case does not change. Honest performance (rough): the gap holds a typical flash's first stroke, but most flashes carry fast follow-up strokes pushing 120 to 150 kV across it, which flash over, feed the actuator anyway and char the tab (vvv); so the actuator is saved in perhaps one tail strike in five, and the link risks breaking in the rest. Cost: about 0.4 lb of braid on the boom aft of the pivot plus roughly 1 lb more nose lead on a rack already carrying 14 to 20 of its 24 lb (Stage 10), the longer tab, and the front re-layout. Claude's opinion stays the metal tab: the partial save in the rare case is small and the boom's balance pays for it; but the scheme above is the right wiring if Mike picks G10. His pick owed: metal tab, or G10 with gap and braid run. The ground wire yes-or-no stands either way. (yyy) Actuator lightning rethought, 2026-10-30 (Mike: "Why 50 inches? And why 1\" braid? Why not from the front of the rod to the grounding strap with 6awg wire? What size g10? ... Please rethink it all", with a screenshot of McMaster 6071K74). (1) A short jumper from the rod's front end is right in principle: the gap holds only the voltage set by the loop between the copper path round it and the actuator path, about 0.25 microhenry for a 6 to 8 in jumper against about 1 to 1.3 for a run from the tail, so the gap would save the actuator in roughly 9 tail hits in 10 instead of about 1 in 3 (rough). It cannot be built: the clevis is inside the boom root at every point of its stroke (its jaw tips reach the bore mouth at 19.0 in only fully retracted, and sit about 1-3/4 in deeper with a 2 in gap); a side slot would cut the skin stack between the bolt rows, the Cat 3 weak spot, and the bore mouth leaves 0.19 in round the actuator's nose, where a ground wire would hand the strike a 0.19 in hop into the actuator. So any copper path round a G10 gap starts at the tail. (2) Correction to (xxx): there is no metal boom-root plate any more (the root plates are G10), so the run goes on along the wing face to the strap, about 60 in in all, not 50. (3) Braid was only what was on hand; the Kenhihi 6 AWG tinned silicone wire (Sep 27, 10 ft red and 10 ft black) serves, likely spare after the backup battery board's two short 6 AWG leads (inferred). (4) G10: 1/4 in, full thickness in the clevis (1/4 in jaw, McMaster drawing; may need a light sand), sanded to about 0.22 in where it enters the Thomson's 0.236 in slot, as the rear tab is. Mike is right that it survives a strike better: thickness does not change the flashover distance (set by the gap's length along the tab) but leaves more after charring and is twice as strong. (5) Gap, if G10: 2 in between the clevis's and the fork's metal (with the long run, 1 in holds only a weak first stroke, and only dry). Layout change, Mike's yes needed before drawing: the clevis sits about 1-3/4 in deeper, the boom's front tube about 1-3/4 in longer, the pushrod about 1-3/4 in shorter. (6) What matters (rolling-sphere method, rough): with the tip rod 25-3/4 in above the tail's top and the tail 54 to 66 in aft, only strikes under roughly 7,000 to 12,000 amps can reach the tail; under 10 percent of first strokes are that weak (IEEE distribution), so tail hits are about 1 strike in 20 or fewer. Mike's middle band is partly right: tail hits are the weak strikes. But G10 with the 60 in run changes the outcome in about 1 strike in 60; the ground wire beside the actuator cable covers the other 19 in 20. Claude's opinion stays the metal tab. (7) Ground wire: a 12 AWG core from the duplex on hand, not the 6 AWG (about 0.3 in across), since it shares the root PEX with both cables. Found: the build sheet's 1/2 in root PEX (about 0.475 in inside) cannot take the actuator cable (0.25 in) and the wing cable (about 0.3 in) together; (rrr) assumed 3/4 in (about 0.68 in inside, room for both plus the ground wire). Part 1 work: make the root PEX 3/4 in on the build sheet. (8) The screenshot's 6071K74 is the 1/4-20 clevis; the plan's 6071K47 is 1/4-28, matching the 1/4-28 die on the Buy list; otherwise the same (1/4 in jaw and pin, 2 in pin to end, 1-1/4 in jaw depth, 5/8 in wide). Picks owed: metal tab or G10 with the 2 in gap and the 6 AWG run; the ground wire. (zzz) Pushrod length, threads and seller; how lightning kills the actuator, 2026-10-30 (Mike: "how long is pushrod in both cases, where to buy, which threads?" and "how does lightning kill the actuator, wiring or mechanical?"). Length, worked out from the drawings with the flap straight at mid-stroke: tail's front edge 63.0 in back of the wing's nose (boom 48-1/2 in from 19 in, 4-1/2 in in the slot), hinge 8.3 in back of that (Tail drawing: fixed part 8.25 in plus the gap), so the horn's clevis pin is at about 71.3 in; front clevis pin at 20.3 in (front pin 19.4 plus 0.9); pin to pin 51.0 in plus about 0.1 in for the 30 in S-bend; each clevis is 2 in pin to end with 3/4 in of thread, so the rod is about 48-1/2 in with a metal tab, about 46-3/4 in with a G10 tab and a 2 in gap (its pins about 1.85 in farther apart). This corrects (ooo)'s "about 49-1/2 in (inferred)", which was about an inch long; the 54 in order stands and the rod is cut at fitting. Threads: 1/4-28 at both ends, cut about 3/4 in long with the 1/4-28 die on the Buy list, since the clevis's thread runs 3/4 in deep straight into its jaw slot (McMaster drawing) and a longer thread would poke into the jaw; at least about 1/2 in screwed in at each end. Seller: McMaster 89325K113, in the one box the clevises (6071K47), the 10 mm pin (93890A741) and the give-way spring (9287K47) already need; no 1/4 in 316 rod 48 in or longer found on Amazon, and the two 24 in pieces on hand are too short. Order tracker Buy list row updated. How lightning kills it: mostly electrically. The MD's CNO option carries its own control board with a CAN link (J1939; Compute Bay Plan), and chips like its CAN transceiver die at tens to a few hundred volts, while a strike puts thousands between the case and the wires unless they are tied to ride together: that is the tip-rod case and the ground wire's job. A direct tail hit through a metal tab adds current through the moving parts: it arcs where the screw, nut and bearings touch and can pit or weld them; whether a weak strike leaves it working with the ground wire in is a guess either way (the MD's nut material not found). (aaaa) Outside wire along the boom; all the inside work dropped except the ground wire, proposed 2026-10-30 (Mike: "Because of your new and changing language, i don't know where wires go and where they don't. Would it help to run a strap on the outside of the boom tip directly to the ground strap? Skip all this inside stuff?"). Yes, and it works with a metal tab too, which (yy) to (zzz) missed. In a tail hit the outside wire and the pushrod-and-actuator path share the current: the fast first microseconds split by inductance, so part goes through the actuator; the slow part that follows (continuing current, which does the pitting and welding) splits by resistance, and 6 AWG copper (about 2 milliohms over 5 ft) against the stainless pushrod (about 30 milliohms) plus sparking contacts inside the actuator (tens of volts) sends nearly all of it by the copper (worked out, rough). The ground wire covers the electronics either way. So the G10 gap adds little and is dropped from the proposal, with its re-layout and the 60 in braid: metal tab (the 304 bar, 1/8 x 3/4 in, McMaster 8992K794), pushrod about 48-1/2 in (zzz). Route: about 5 ft of the Kenhihi 6 AWG tinned silicone wire on hand (if the backup battery board's two 6 AWG leads leave 5 ft of one color; inferred), a 6 AWG 1/4 in lug (on hand, Sep 22) under one of the tail's three clamp bolts (the pushrod runs about 0.1 in under the bolt heads, so for a strike rod and bolts act as one; a fixed bolt needs no slack loop), along the boom's side and forward on the wing face to the wing strap at about 47 in up, joined by a 6 AWG #10 lug (on hand, Aug 31) under a #10 silicon bronze screw through the wing strap, No-Ox-Id, bedded in 5200; held by glass straps across it, since 5200 and epoxy grip silicone poorly, and faired. Both straps on the wing's far face (the one without the pocket lid), so the lid still comes off; that settles the wing strap face the Lightning Guide sec 5.1 step 3 leaves open (proposed). Weight about 1/2 lb, about 1-1/4 lb more nose lead (rack 14 to 20 of 24 lb). Nothing to buy. The whole list of lightning conductors at the actuator, written for Mike in plain words the same day: outside, the tip rod, the wing strap (both in the guides) and the new boom wire; inside, only the short jumper from the actuator's back pin to the wing strap (in the guides) and the 12 AWG ground wire in the actuator cable's tube from the back pin to the mast-base gusset bolt, which cannot move outside because it works only lying right beside the cable. Dropped: the braid sleeve, the braid jumper in the boom, the hose clamp, the G10 gap, the 60 in braid. Mike's yes-or-nos asked: metal tab; the outside boom wire; the ground wire in the cable's tube. (bbbb) Threading the pushrod, 2026-10-30 (Mike: "Will normal tap and die (like mine from amazon) put threads in stainless?" and "why 1/4 28 vs 20?"). Die: a set die (the 60-piece SAE and metric set, Sep 3; its steel not recorded) will usually cut two short 316 ends, each about 3/4 in, with the end chamfered, Tap Magic (on hand), slow half turns forward and quarter turns back; it dulls the die and can tear a thread (inferred from shop practice). The 54 in rod leaves about 5 in over the 48-1/2 in pushrod, so a torn end can be cut off and redone. So the separate 1/4-28 die on the Buy list (Amazon B0FS1RN49Q, listing not readable from the cloud session) is only needed if the set has no 1/4-28 die. Thread: either works at the pushrod's loads (56 lb at most). 1/4-28 kept because its thread is about 2/3 as deep (easier to cut by hand in stainless), it leaves about 14 percent more steel at the thread (stress area 0.0364 against 0.0318 sq in), and a half turn of the clevis moves the flap about 0.9 deg instead of 1.3 when setting it straight. Stainless threads gall either way: anti-seize (Tef-Gel) on assembly. If 1/4-20 is wanted, both clevises become 6071K74 and the die 1/4-20. (cccc) If the tab is G10: what ties what, 2026-10-30 (Mike: "with g10, nothing to equalize the actuator body?"). The actuator body is tied either way: the short jumper from its back pin to the wing strap (Lightning Guide sec 5.1 step 4, landed on the rear pin per the build sheet), with the ground wire in the cable's tube on the same pin. What G10 leaves without a metal tie is the pushrod: with a metal tab it is tied to the actuator body through the tab; with G10 it touches nothing metal (G10 tab at the front, G10 horn at the tail, PEX between) and reaches the boom wire only by an air hop of about 0.1 in to the tail's clamp bolts. Enough to carry a strike, but otherwise it floats, and standard practice (ABYC E-4, as remembered) bonds long metal near a strike path so it cannot side-flash. So with G10 only, one addition: the boom wire's end also on the pushrod's back end, a 6 AWG 1/4 in lug (on hand) held by a 1/4-28 316 jam nut against the tail clevis (which also locks the clevis), with a small slack loop for the rod's about 1.1 in of travel. Jam nut: none in 316 found on Amazon; add to the McMaster box, part number not yet picked. Not needed with a metal tab. (dddd) Mike leaning to G10 with a 1 in gap, 2026-10-30 ("i'm leaning toward g10 with 1\" gap"); not decided. With the boom wire as the path round the gap (loop about 1 to 1.3 microhenries), a 1 in gap (about 1 in of tab surface, which flashes at roughly half of air's strength) holds the first stroke of a weak tail strike only when dry, and flashes on the faster follow-up strokes most flashes carry: it keeps the spike out of the actuator in roughly 1 tail hit in 6; a 2 in gap in roughly 1 in 3 (rough guesses). Layout change either way, Mike's yes needed before drawing: tab pins about 1.75 in apart for 1 in (about 2.75 for 2 in) instead of 0.9, so the clevis sits about 7/8 in deeper (1-3/4 for 2 in), the boom's front tube that much longer, the pushrod about 47-3/4 in (46-3/4 for 2 in). Tab 1/4 in G10, 3/4 in wide, sanded to about 0.22 in where it enters the Thomson's fork. Wires with G10: the five of (aaaa) plus the pushrod tie of (cccc). Claude's opinion: if G10, make the gap 2 in (twice the benefit for about another inch of the same change); the metal tab remains the simplest. Asked: G10 or metal; if G10, 1 or 2 in. (eeee) G10 tab with a 2 in gap, decided 2026-10-30 (Mike: "ok 2\" gap. Please make detailed drawings with very clear instructions"). The actuator end is a 1/4 in G10 tab, 3/4 x 4 in, front 1 in sanded to about 0.22 in for the Thomson's 0.236 in slot; 10 mm hole 5/16 in from the front; the clevis placed so its jaw tips sit 2-1/2 in behind the 10 mm hole's center (2 in of bare G10 past the fork's tips), its 1/4 in pin about 2-3/4 in back; the 1/8 in pin 5/8 in behind that. So the clevis sits about 1-7/8 in deeper in the boom, the short tube is 4-1/2 in long (21 to 25-1/2 in back), and the pushrod is about 46-3/4 in. Taken as decided with it, since the gap does nothing without them: the outside boom wire of (aaaa) (Mike's own idea) and the pushrod tie of (cccc). Still owed by Mike: the ground wire beside the actuator cable. Worked out while drawing, and corrections: (1) (yyy)'s "make the root PEX 3/4 in" is withdrawn: the wing cable is Cat6, 0.225 in (Mike's call, recorded earlier in this item), so 1/2 in PEX takes both cables as Mike worked out; but the 12 AWG ground wire cannot also fit, so it is laid in the groove beside the PEX, built in (build sheet step 3). (2) The ground wire's bottom end cannot reach the mast-base gusset bolt from inside the mast: it runs down the mast with the cables and lands on the slip ring's drive bar (turning side, on the strike path), on a #10-24 316 screw in a tapped hole, not under a slewing ring bolt. (3) The actuator jumper must cross the far skin (the old jumper went to an outside metal plate that is gone): it leaves through a 5/16 in hole in the pocket's far wall about 11 in back, 46 in up, with the ground wire in a 3/16 in hole beside it, both filled with 5200, to two #10 silicon bronze landing screws set under the wing strap at 45 and 46 in up when the strap is glued. (4) The boom wire crosses the joint where the boom slides out, so it gets a bolted splice (two 6 AWG 1/4 in lugs, one 1/4-20 316 bolt, nyloc) on the far face just ahead of the wing's back edge, below the boom at 44-1/2 in up; the boom wire's tail end and the pushrod tie share the tail's rear clamp bolt (lugs under its head). (5) All outside copper, the tail's horn and the PEX S-bend go on the far face (the face without the pocket lid); this settles the strap face the Lightning Guide sec 5.1 step 3 leaves open; flagged to Mike. (6) Tail weight: every pound at the tail costs about 4-3/4 lb of nose lead (Stage 10); the boom wire, about 0.6 lb in all with its lugs, needs about 1-1/2 lb more lead (rack 14 to 20 of 24 lb). Drawings: Reference/Actuator_End_2026-10-30 (2nd edition, with a full-size tab template), Wing_Boom_Pocket_Build_2026-10-30 (4th edition), Reference/Tail_One_Piece_2026-10-30 (2nd edition), and new Reference/Wing_Lightning_Wiring_2026-10-30 (every lightning wire on the wing, with build steps and details). Order tracker: pushrod row updated, the die row marked buy-only-if-the-set-lacks-it, a 1/4-28 316 jam nut row added. (ffff) Review of the boom, tail and trim drive, 2026-10-30 (Mike: "Please review and evaluate the entire boom, actuator, tail plan and mechanism for any errors, vulnerabilities, and OFIs"). Full list: Reference/Boom_Tail_Actuator_Review_2026-10-30.md (two independent reviews plus Claude's). Part 1, done in the drawings the same day: the boom's last 4-1/2 in straight at 1-1/4 in (tapered, it was 0.2 in too tall for the tail's slot); all four root bolts 5/16-18 x 4-1/4 in (every bolt grips about 3-1/2 in; the 4 and 3-1/2 in were short); bore the front after glassing (the 1-1/4 in bore is wider than the 1.16 in core); the short tube is 1 in CPVC, about 0.9 in inside; soft limits 0.56 in each side, a must (the actuator's own travel would drive the horn to about 61 deg; the tail clevis meets the spring and end plate at about 40 to 50 deg); take-apart steps for the front pin (tail clevis off, soft limit widened by laptop); the tab's nose rounded, its sanded faces sealed with epoxy, its 1/8 in pin glued not pressed; tail clamp bolts 2-1/2 in with 316 Bellevilles, thread locker, about 60 in-lb (a guess), witness marks; G10 plates and horn painted; hinge pieces lined up on a straight rod while the glue cures; marine grease on the rod, Tef-Gel on pins and threads, a boot on the PEX's tail mouth (part to pick), silicone in the root PEX's mouth; glue-lined heat shrink where silicone wire passes the skin or is faired; landing screw heads glued; face run clear of the drains. Order tracker: root bolts and rear pin (McMaster 93890A742) rows added; tail bolts 2-1/2 in with Bellevilles; hinge row text four pieces; the Kenhihi wire's job written in. Part 2, Mike's yes or no asked 2026-10-30: (A) flap stops (316 pins at about 30 deg) and dropping the centering spring and Test 2, both missing since the one-piece tail; (B) the one-way preload moves to a small 316 pull spring at the tail, about 1 lb (it cannot reach the moving fork where drawn; the MD's end play alone is about 0.115 in at the flap's back edge against the Trim Guide's 1/16 in); (C) a flush 1/4 x 5/8 in 316 roll pin in place of the front clevis pin and cotter, for the 0.9 in tube; (D) 1/4 in tail clamp plates (the 1/8 in plate is near its limit at the Cat 3 gust by the reviewer's prying model); (E) 1/2 in root plates against a knockdown (the root, 285 lb at the tail, is weaker than the boom, 590; the tail swept broadside through water above about 6-1/2 knots reaches 285); (F) the ground wire: it cancels only the drop above the mast base, not the 10 to 40 kV drop from the can to the junction (rough), and buried in foam it carries part of a strike; Claude's opinion: drop it and accept that a tip strike may kill the actuator; (G) take the pushrod tie back out and keep a bare jam nut with the back thread 1-1/4 in long (the lug loosens the jam nut; the crimp flexes; the rod already sits about 1/4 in under the clamp bolt that carries the boom wire); (H) aim the PEX's S to end about 1.05 in out, mid-swing, so the rod does not bend at the PEX mouth on big swings; (I) a 316 or 17-7 PH give-way spring if one is sold (the 9287K47 is 302). Optional: a 3/16 in pushrod (saves about 1.2 lb with its lead), a lightning wire up the tail's front edge, a surge diode across the tip's 24 V pair. Software (Part 1): soft limits into the Trim Guide and System Settings Sheet; SocketCAN restart-ms about 100 and an alarm when the actuator's status stops; a deadband on the damper; hold the actuator still above about 45 deg of heel (IP66 moving); the CAN terminating resistor at the actuator if the MD lacks one. Write-ins (Part 1): the spring link as the fail-safe in the Trim Guide, Wing Power, Lightning Guide, Wing Construction Stages 10 and 12 and System Settings; the braid jumper to a metal plate in Lightning 5.1 and 5.3 and Stages 10 and 11; one trailing-edge conduit against two 1/2 in PEX tubes; the potted pigtail against the DTM plug; pushrod stock; horn 1 in against 1-1/8; hinge count; Stage 12's breakaway; Stage 10's doublers and metal plate. The Thomson actuator is not logged as received (ordered Sep 16). *Documents: Wing Construction Guide Stages 10 and 12; Wing Trim Actuator Guide secs 2, 4 and 6 (sec 2: the sealing rating).*
462. Wing boom root and trim actuator drawing: what it found wrong in the guides, and the fixes it proposes (Claude, 2026-10-27, from Mike's "draw the entire actuator, boom, etc area in lots of detail" and "need to be able to remove actuator without cutting"). Drawing: Wing_Boom_Root_and_Actuator_2026-10-27 (.html master, .pdf to print, five pages; built by Reference/_make_Wing_Boom_Root_and_Actuator_2026-10-27.py). Black is from the guides, the A4 rib table or Thomson's Electrak MD catalog (pp. 71 to 73); orange is proposed. Found: (a) fork meets fork: both actuator ends are forks for a 10 mm pin (Thomson option N), but the guide puts a clevis on the link's cap A to take it, and clevises on both pushrod ends, so two joints are fork to fork; each joint needs one fork and one eye. (b) The guide's clevis pins are 3/16 in; the actuator's holes are 10 mm (0.39 in): about 0.2 in of play, far past the proof-test free-play limit. (c) The pins turn only in the plane of the wing face, so the actuator's rod and the pushrod must sit the same distance out from the boom; the rod is about 1.3 in out, a conduit glued flat on the boom about 1/4 in; the guide does not say how they meet. (d) The wing face falls about 0.63 in over the back 3.8 in the boom covers, and the boom is flat: a filler wedge or a shaped boom root is needed on the boom face and under the clamping plate. (e) Actuator plus link is about 15-3/4 in pin to pin; the fairing that covers them and the cable mouths is about 22-1/2 in long, not "small" (Trim Guide sec 2 already says it was drawn for a smaller part). (f) The guide fits one gland where the cable enters the fairing floor, but Josh's lower mouth brings out both cables of different size, which never share a gland (Mike, 2026-10-13): seal that mouth with silicone. (g) An L-shaped fairing has two low points; each needs a drain. (h) The guide's ordering code stops after CNO; the lead length (0.3 or 1 m) is in the part of the code not recorded. Proposed, for Mike's yes: 10 mm pins with R-clips at both actuator ends, a G10 lug at the rear and an eye on the link's cap A; a 1/4-28 rod eye on the pushrod's link end; the pushrod conduit raised on a strip to the actuator's rod line; filler wedges at the boom root; a plug (greased, booted, which the guide allows) on the actuator lead so it comes out with no cutting; a lid over the whole outer face; a third drain. Also drawn but not decided: the boom's end positions, the bolt pattern, the plate size, the actuator's spot (rear pin 28.2 in back; the guide gives only "about 20 in aft of the pivot") and the tail centered on the boom. *Documents: Trim Actuator Guide secs 2, 4 and 6 (joints, pins, fairing, lead plug, shopping list); Wing Construction Guide Stages 10 and 12 (boom seat, conduit height, fairing size).*
461. How the wing's two cables leave the wing can: through the mast plate, not the mast wall? (Claude, 2026-10-27, from Mike's "need a better plan".) The guides run both cables up the mast bore and out a hole in the mast wall inside the wing root, sealed with a rubber grommet (Wing Power System 1A). Mike, 2026-10-27: tapping threads in stainless is not easy, no gland on the market is long enough for a 3/4 or 1 in wall (the longest long-thread glands are about 15 mm, just under 5/8 in; checked 2026-10-27), and PEX barely bends. A grommet does not seat in the mast's 0.600 in wall. Claude's plan, no tapping, no PEX bend, no holes in the mast: (a) one plain drilled hole in the mast plate (304, 3/4 in) about 5 in aft of center on the wing's centerline, inside the bearing's 11.97 in inner bore (Rig Base 3A), clear of the drive bar (set the bar across the boat); a short G10 tube glued in, standing about 1 in above the plate; both cables through it, silicone filled at its top (silicone, not a gland, so the different-size rule does not bite). (b) The PEX root length runs straight but slanted: its bottom end about 5 in aft of the mast axis at the root closeout, its top 24 in from the nose where it meets the fairing, about 43 in up, so about 15 degrees off vertical and no bend; the fairing-to-tip length unchanged. The NACA 0015 section is about 4 in thick 12 in from the nose (inferred), so the slant has room; lay it out on the foam stack clear of the spar, the socket and the two mast bolts before the channel is cut. (c) The cables cross the about 3 in gap under the water-shedding disc from the PEX bottom end, silicone filled as already planned, to the plate tube. (d) In the can they hang from the plate tube to their plugs at the slip ring, outside the drip cup, tied off. (e) The mast gets no wall hole and no center hole, so the bore stays sealed by its weld. Refit: cut the silicone, unplug, lift the wing; the cables come up through the plate tube. Refined 2026-10-27 (Mike: "maybe run the pex vertical from there"): two straight PEX tubes from the same root spot, one cable each. The wing cable's tube runs vertical, about 12 in from the nose, root to tip: the tip gear (air terminal on the spar line, wind meter ahead of the nose, compass on the tip cap) sits forward, so it arrives about 4 in behind the spar line instead of 16 in aft of it. The actuator cable's tube keeps the slant to the fairing, since the actuator sits at the boom root about 20 in behind the pivot. The fairing-to-tip length goes. The two come to about 121 in (a guess), so a second 10 ft PEX stick (about $5). Lay both out on the foam stack clear of the spar and the socket. Changed again (Mike, 2026-10-27: one vertical tube with a side branch back to the actuator, "less intrusive than a 2nd tube"): one vertical PEX from the root to the tip carries both cables; at about 43 in up, just under the boom-root solid band (44 to 50 in), a short straight PEX side branch runs aft to the fairing footprint, about 24 in from the nose, and ends in Josh's filleted mouth. The corner is not a fitting (PEX elbows and tees are narrower inside, inferred): it is a small curved pocket cut in the foam where both tubes end, lined smooth with epoxy, so the actuator cable turns on a sweep of about 1 in radius (Mike, 2026-10-27: the 18/4 bends at 0.75 in radius easily and pulls through a 1 in radius) and a pull string stays in each tube. One 10 ft PEX stick covers it (about 88 in). Claude's pick for the vertical tube: about 5 in behind the mast center (12 in from the nose), clear of the socket fill and the spar, unless the layout on the foam shows the spar ends closer. Drawn with the actuator moved forward (Mike, 2026-10-27: "quick option moving it forward 10-12 inches", "should hang"): Wing_Actuator_Forward_Option_2026-10-27 (three pages: layout, enlarged, three views). The actuator hangs, motor below the rod line, rear pin 17.2 in back of the nose, on a G10 cradle with a stainless strap (so it cannot swing down round the rear pin, Mike's question); vertical PEX 12 in back; corner pocket and a 4-1/2 in branch at 43 in up to a mouth inside the fairing at 16.6 in. To build into the wing before the boom face is skinned: a 3 x 3 in solid hardpoint centered 17.2 in back, 47 in up, and a firm base 15.8 to 23.9 in back, 42.2 to 48.4 in up, plus the PEX, pocket and branch. Rod line 3.0 in out from the chord line so the link's caps clear the boom bolt heads; lid about 4.4 in out. Proposed, from Thomson's sizes and the rib table. Mike's yes still needed on the whole route (mast plate hole, no mast holes, this tube layout) before the channel is cut and before the mast weldment drawing. PEX size, 2026-10-29 (Mike: "why not 1/2 pex for the vertical run?"; "18/4 is .25, cat6 is .225 = .475, should fit fine in 1/2\" pex"). The wing cable is now Cat6 at 0.225 in across (Mike; shielded, all copper, stranded per Buy List C8's rule, to confirm); with the 0.25 in 18/4 the two add to 0.475 in. 1/2 in PEX is 0.625 in outside with a 0.070 in minimum wall, so its bore is about 0.485 in at nominal and smaller where the wall runs thick (known). Claude: below the pocket that leaves about 0.01 in, two cables side by side plus a pull string in a sealed wing that cannot be reopened (the guide asks for 5/8 in bore or more), so keep 3/4 in PEX (about 0.68 in bore) from the root to the pocket; above the pocket the Cat6 runs alone (0.225 in in 0.485 in), so 1/2 in PEX fits well there. DECIDED: 1/2 in PEX for the whole vertical run, root to tip (Mike, 2026-10-29: "consequence of a tight fit is basically, 'hey, its tight', especially with the slick walls of pex"; "cables are on my desk right now"; "plenty of room for the string, they don't line up inside the pex side by side"). The pull string rides in the gaps beside the two cables. Claude's note: Mike can pull both cables through the stick before it is glued in, since they are on hand. One 10 ft stick of 1/2 in PEX covers both lengths. Build sheet updated. *Documents: Wing Power System 1A, 8.6 and 11; Wing Construction Guide Stages 4, 6, 10 and 11; Wing Trim Actuator Guide sec 4; Rig Base Two Plate Design (the mast plate hole); Lightning Ground System Guide 5.1.*
440. Questions from the v10.3 accuracy check (Claude, 2026-10-19). Each has an Open line in the guides named. (a) The cellular units' control output: Build Guide 8.3 and Design Decisions R-8.3 give the Primelinks one output that power-cycles the active Pi, but Compute Bay 4.3 says the watchdogs have no remote trigger, on purpose, and no hole list carries the wire. Keep it (then Compute Bay and the Electronics plan get the wire) or drop it (Claude's view: drop it; the watchdog already restarts a hung computer). (b) The rudder stock bellows boot: the one bought went back (refunded 2026-10-02). Buy another, and which, or go without? (Build Guide 3.8, Steering Handbook, BOM 15.10.) (c) Science on the load-shed ladder: the Settings Sheet says science runs full to the 10 percent floor in the tropics, and also that it is off from 30 to 50 percent. Does the tropics exception stand? (Settings Sheet sec 4, Routing Guide 2.3.) (d) Boat 2 at Cape Horn: Routing Guide 7.3 says stay 50-plus nm south of the cape, but 7.1 aims the drogue at about 56.05 to 56.4 S, only about 4 to 25 nm south. Which holds? (e) The post lights' cable: Instrument Post Guide 10.1 runs them on one four-core cable, which is on no buy list; buy a four-core, or run two 16/2 cables through a second gland? (Wire List B2, Electronics plan A6.) (f) Surge parts on the post's wind and light runs: Instrument Post Guide 10.3 calls for surge diodes at the junction and the central node, no other guide has them, and the 33 V diode on hand does nothing on a 5 V line. Fit them (a 5 V part for the wind line) or drop the line? (g) The shaft thrust block's bond wire (Lightning Guide 6.2 item 2): its ball bearing conducts, so the wire ties the shaft to the copper, the very path the spark gap is there to stop. Drop the wire (Claude's view) or keep it? (h) The sound archive: plain 48 kHz recording fills the 4 TB drive in about 16 months, not 18 (Compute Bay 4.1); record as FLAC, or accept about 16 months between refits? (i) Two dated sheets at the top of the folder, Box_Parts_With_Photos_2026-10-15 and USB_Hub_Box_Wiring_2026-10-14, show dropped parts (the Iridium isolators, the returned TICONN Iridium boxes, the hub on a DIN rail). Move both to Archive? (j) Moot: the coolant loop is dropped and the motor's ports plugged, so no M16 barbs are used (item 466, 2026-10-29). *Documents: as named.*
427. Box parts check, every box: what is missing (Claude, 2026-10-14, from Mike's "review ALL boxes and see what parts are missing"). The full list, box by box: Reference/Box_Parts_Missing_2026-10-14 (.md and .docx). Three causes of the waits: small parts no guide ever counted, parts that shipped with no delivery mail, and parts that wait on a choice. (a) To buy now, nothing to decide (added to the tracker Buy list 2026-10-14): the wing box (the TICONN 5.9 x 5.9 x 3.5 in picked 2026-10-12 was never ordered); a second and third 100 ft roll of 16/2 (need about 220 to 260 ft against 100); an SMA bulkhead-to-male lead 2-pack for the two GPS boards (one inside lead on record, two needed); 30 AWG tinned copper wire for the 22 fuse links (Wire List E4 says make them from 30 AWG wire, none was ever bought); 3/8 in ring lugs for the VESC's 8 AWG and 12 AWG leads; neutral-cure silicone for the VESC capacitors (item 417); M4 316 pan-head screws for the cord puller motors (item 276); plus the rows already on the Buy list. (b) Count, then buy only the shortfall: sealed inline fuse holders (about 13 needed, 9 on hand; item 401 p, less the hub box fuse dropped in item 426), membrane vents (about 15, 9), crimp butt splices 22 to 18 AWG (17, type on hand unknown), WAGO 221-415 (8: 2 at the VESC, 6 in the battery box), Micro-C 2 m cords (about 11, 8 on record), DIN fuse holders (8), glands by size (PG16 x7 first). (c) Mike's calls that block buying: the USB hub box, wing tip box, light driver box and second drogue box (pick each); the wet lab's reader and switch board (none picked); the slip ring's Deutsch DT plugs (pin plan not written); the release trunks' third gland (DECIDED 2026-10-26: yes, a third gland at each end, item 456); whether Pi A keeps the INA226 (Claude: strike it); Pi B's metal SMA bulkhead; the 5 V TVS on the detector lead (not picked); the emergency bank strip's fuse sizes; the VESC's kill, temperature and switch plug types. *Documents: order tracker Buy list; the guides named in the Reference file as each choice is made.*
410. SBMS0 stream: swap the micro USB lead for a serial adapter on the add-on board's green push-in block? (asked by Mike, 2026-10-13). The WiFi/USB add-on board has a six-pin green push-in block (GND, 3.3Vdc, TX, RX, SCL, SDA: the same serial stream as the micro USB socket, on the isolated side; manual page 15). Bare wires pushed into it hold better than a micro USB plug, whose small socket and solder tabs are the weakest plug in the watchman box. The block is plain 3.3 V serial, not USB, so reaching the hub takes a USB-to-serial adapter: the DSD TECH SH-U09C (Amazon B07BBPX8B8, $12.49, FTDI FT232RL, jumper set to 3.3 V so its VCC pin gives 3.3 V; checked 2026-10-13) in the watchman box, four wires to the block (GND to GND, VCC to 3.3Vdc, the adapter's TXD to the block's RX and RXD to TX, crossed), its USB-A plug into a USB-A extension lead to the hub, the joint tied down and staked. Not the SH-U09G cable or an FTDI TTL-232R-3V3-WE: both give 5 V on their power wire, which must not go to the 3.3Vdc pin. Pi A's own serial pins are taken by watchdog WD-A, so the adapter route is the only one. Claude's opinion: a modest but real gain for $12.49, do it. If yes: Battery Box Guide Part 8 and 10, Electronics plan A4 entry 4 and A1a, Settings Sheet sec 1, BOM 15.4, Box Wiring Sheets A4, order tracker Buy list. If no: nothing changes; the micro USB lead stays, tied down outside the plug.
405. Where to run the offshore trial: Brunswick, a spot nearer the sound, or the Gulf? (asked by Mike, 2026-10-08). Brunswick's hard parts: an 18 nm harbor run through the ship channel from Brunswick Landing Marina; a trial area 34 to 46 nm past Buoy STS, put there to stay outside right whale critical habitat in calving season (November 15 to April 15); winter fronts; the Gulf Stream beyond. Options: (a) launch from a marina nearer St. Simons Sound (Jekyll Harbor Marina or the marinas at the St. Simons causeway), which roughly halves the harbor run (guess: about 10 nm to Buoy STS; Jekyll Creek is a known shoaling spot on the waterway, fine at mid-tide for 3.5 ft draft, inferred); (b) pull the trial box in to about 15 to 20 nm offshore, inside the critical habitat, which is not closed to a 2 to 3 kt boat (inferred: the speed and approach rules bind larger or faster vessels and approaches to whales), halving the chase boat run; (c) run the trial on the Gulf coast (Pensacola, Panama City) before trucking to Brunswick: closer to Tulsa, no Gulf Stream, no right whales, deep water close off Destin, but a second haul, truck and launch, and less time before a December 26 departure. Claude's opinion: (a) plus (b) at Brunswick; the Gulf only if the Brunswick trial slips. Mike's call. *Documents: Boat 1 Route sec 4; Build Guide sec 11.3; the Coast Guard answers (d).*
408. The drogue U-bolt collars that came are carbon steel: keep them or swap for stainless? (found 2026-10-12). Mike bought one Sinoblu "45 steel" double split shaft collar listing on Oct 8, an hour after deciding on collars (item 407): $9.99 ($10.84), Amazon 113-7231241-9142661, here Oct 11. 45 steel is plain medium carbon steel (known), so it will rust on the transom and streak the topsides. The bore and the count are not in the mail; eight are needed (two per U-bolt in use, 7/16 in bore). Claude's opinion: a slipped or rusted collar costs nothing structurally (item 407), so keeping them works for the trial, but rust streaks under each U-bolt are certain within weeks in salt water; stainless Summit CSS2-43-1 on eBay is $11.99 each, about $96 for eight. Mike to choose. *Documents: Transom Drogue Box Guide 4A; BOM 15.7; the order tracker.*
406. Leave from Galveston instead of Brunswick for Leg 1A? (asked by Mike, 2026-10-08). Worked in Reference/Galveston_Start_Leg1A_2026-10-08.md. The track: Galveston ESE across the Gulf to south of the Dry Tortugas (about 680 nm), then east and north through the Florida Straits riding the Florida Current to off Cape Canaveral (about 410 nm), then east north of the Bahamas to join the Brunswick track 50 nm south of Bermuda; the rest is the same. About 1,050 nm longer (4,680 nm) and about 3 weeks longer: 80 / 93 / 113 days against 64 / 72 / 85, counting a wait in the eastern Gulf for a gap between fronts (inferred). Leaving December 26 reaches the Canaries about March 29 instead of March 8; leaving about December 5 keeps the March 8 arrival. Against: about 4 days in the Florida Current instead of a 1-day Stream crossing, in the choke point the Routing Guide avoids (sec 1.2) and Boat 1 Route sec 3 rejects for a winter start (a norther in the Stream with no room to turn back); more Stream lightning; 1,311 oil and gas structures still standing in the Gulf (BSEE file of 2026-10-08), 38 within 10 nm of the straight track, all in its first 270 nm; Galveston's tanker traffic; between fronts the southern Gulf's NE to E wind sits forward of the beam (inferred); the Coast Guard answers would go to Sector Houston-Galveston instead of Savannah; Cuba on the south side of the Straits (sanctions rules not researched). For: about 550 road miles from Tulsa against about 1,100; the offshore trial and departure in one place with no right whale rules (item 405 option c without the second haul and truck); two weeks in the Gulf with Tampa and Key West close, a shakedown before the Atlantic. Leg 1B still lands at Brunswick: home to Galveston means the Straits against a 2 to 4 kt current, or the Caribbean in October and November, where the late storms form. Mike, 2026-10-08: the Gulf leg is 3 more weeks of trial before leaving US shores, in case surprise repairs come up. Claude agrees it changes the answer: from Brunswick the boat is near home about a day and a half before the Stream, then cannot come back; from Galveston it spends about 2 weeks never more than about 240 nm from a US port, meets a Gulf norther within reach of help, and Key West is a last stop before the Straits (the Reference file, sec 4). Claude's opinion now: Galveston, keeping December 26 and landing in late March (refit about 5 months; the September return does not move). Mike also asked about Port Arthur or Cameron, LA: not easier (sec 5). About 50 road miles and a day at sea shorter, but 2 to 3 times the platforms along the track, a 20 to 30 mile channel shared with LNG carriers under moving security zones, and few yards for a masted sailboat; Galveston is the best of the three (opinion). Whatever the port, the Spain research filing should give an arrival window through late April 2027. Mike's call: switch the start to Galveston? *Documents: Boat 1 Route secs 2, 3, 4; Routing Guide sec 1.2.*
390. Coast Guard answers for Savannah (the COTP risk-assessment topics a to l): suggested wording for the yellow gaps, written 2026-10-08. Mike's draft of 2026-10-07 had yellow gaps. Claude's suggested wording is in green in Reference/COTP_Savannah_Answers_Suggested_2026-10-08.docx (US letter); yellow is left only where the facts are Mike's: his FAA pilot certificate and ratings; the shore watch keepers (names, how many, experience); how often the Primelink units report position. Also Mike's to do: make watch@viatrax.com (or give the real address); book the escort and the offshore chase boat (Sea Tow Brunswick, which Sea Tow's 2021 article says is run by Brunswick bar pilots, phone 800-473-2869; or TowBoatUS Brunswick, Saltwater Cure LLC on Jekyll Island, (877) 283-1564; a BoatUS membership is $165 a year, and each TowBoatUS port must cover at least 25 miles from home, so the trial area 33 to 42 nm out is likely a paid job); get an FCC Restricted Radiotelephone Operator Permit (Form 605 in the FCC's online licensing system, $35, no test, a pilot certificate qualifies, issued in days); take a Rules of the Road course (National Seafarers Academy "Nautical Rules of the Road," 20 hours, Coast Guard approved, price not found). Where to get insurance quotes (2026-10-08): the Shipowners' Club in London has a liability policy written for autonomous and remotely run vessels (collision, wreck removal, towage, pollution; it covered about 80 small uncrewed craft under 79 ft (24 m) as of ABB's report), contact through ; Viatrax's own business insurance agent, who can place marine cover; a US ocean marine market such as Travelers Ocean Marine, through a broker, with custom wording. No US broker was found advertising a program for uncrewed boats, and ordinary yacht policies assume crew aboard. Liability is the must-have; hull cover is harder to find. Suggested, for Mike to confirm: $1,000,000 liability insurance including wreck removal (Claude's opinion; Spain's floor is about $390,000, items 261 and 139); the watch procedure ready by November 15, 2026. One fix found while filling it: the draft said an all-round white shows sunset to sunrise. The design shows the bow bi-color and the sternlight at night, and the two post whites only while the motor runs (Compute Bay Plan sec 11, channels 3 and 7); the suggested copy says so. New questions it raised: items 391 to 395. Watch procedure (2026-10-08): Claude's draft is Reference/Watch_Procedure_PRIME_RANGER_2026-10-08.docx (markdown beside it; US letter, 4 pages): who stands watch, what to check each shift, red and amber alerts and what to do, the quiet-boat time limits (2 hours near shore and in the trial, 12 hours at sea), who to call, notices, commands and who may send them, the log and drills. Still to fill: the watch keepers and rota, the Primelink report interval, Portugal's rescue center number; confirm the Sector Charleston Command Center number (833-453-1261 on the Coast Guard site; 843-740-7050 on an older page). Chase boat for the offshore trial (2026-10-08): the trial area is 34 to 46 nm from Buoy STS (computed, about 40 nm to its middle), 52 to 64 nm by water from Brunswick Landing Marina; at 15 to 20 kt that is 3 to 4 hours out, and a tow back at 5 kt about 12 hours. Needed: on call (not on station) for the about 5 trial days plus the harbor days; a boat fit for 60 nm offshore in winter; a licensed captain with an assistance towing endorsement for towing for hire; towline and bridle, VHF and AIS. The Coast Guard answers' "about 3 hours away" should read "about 3 to 4 hours away". Rules of the Road course (2026-10-08): NATSEA's is "blended" (part online; how the rest is done and the price are not posted); fully online choices: Maritime Institute, 16 hours, $400, final exam taken in person with a proctor; ASOS captains' course, self-paced online, $395, a completion certificate (Coast Guard approval not stated). Mike to pick; the Coast Guard answers name NATSEA until then. Mike's shorter version of 2026-10-08 (Reference/COTP_Savannah_Answers_MINIMUM_2026-10-08.docx) is now the working copy: it drops the offshore trial and the escort. Claude's notes on it: fill the hull color (white); drop "or take remote control" and "or cellular command" (no steering command is on the boat's command list, and the cellular units only listen); "computer or power failure" should be "computer failure" (a dead battery takes the lights with it); the cone row should say no cone carried (item 392); the horn row stands (item 391); the unescorted 6 to 9 hour channel run is the line most likely to draw questions. Trial site (asked by Mike 2026-10-08, answered in chat): easier spots near Brunswick (a marina nearer the sound, a trial box nearer shore) and the Gulf coast; see item 405. *Documents: none in the set; the Reference file above.*
393. No tow point on the boat (found 2026-10-08). Tows are planned in and out at Las Palmas and off St. Simons Sound at the end (Boat 1 Route secs 4 to 6), and the Coast Guard answers count on an escort that can tow. No guide names a bow tow eye or how a towline is made fast; the cleats and pulpit were stripped (Build Guide sec 2). Needed: a bow tow point a towboat crew can reach from their own deck with nobody aboard, for example a bow eye through the stem, low, with a backing plate, and a short pendant led up to the deck. Mike's call on the kind. *Documents: Build Guide; Boat 1 Route.*
394. Brunswick harbor: tow the boat between the marina and Buoy STS instead of running it under escort? (Claude's opinion, 2026-10-08). The run is about 18 nm: 6 to 9 hours at 2 to 3 kt, against about 10 hours of late-December daylight at Brunswick (inferred). The ebb at the sound's mouth is strong (Waterway Guide; speed not found). Under tow at 4 to 5 kt, the speed the lake tow test already uses (Build Guide sec 11.2), it takes about 4 hours, the towboat carries the lookout and sound signals, and the Coast Guard answers on lookout, take-over and fog near shore get simpler. The boat would run on its own only seaward of Buoy STS. Cost (2026-10-08): towing companies bill non-members about $250 to $300 an hour or more, more at night or in bad weather (BoatUS; GEICO, Sept 2026, says $300 and up), timed from leaving their dock to getting back. Claude's estimates from that rate: one marina-to-Buoy STS tow about 5-1/2 hours, about $1,400 to $1,700; the three harbor tows (trial out, trial back, departure) about $4,000 to $5,000; a recovery from the trial area, if needed (corrected 2026-10-08: about 58 nm by water from the marina, about 4 hours out and 12 hours back under tow), about 16 hours, $4,000 to $5,000; the November 2027 tow in from off St. Simons Sound about $1,500 to $2,500. A $165 BoatUS membership is for breakdowns; a planned move is very likely billed at the hourly rate (inferred). A chase boat kept on station for the 5-day trial would cost far more than one on call; day rate not found. Needs the tow point (item 393). Mike, 2026-10-08: no need to decide today; do not ask again until he raises it or the escort is being booked. *Documents: Boat 1 Route sec 4; Build Guide sec 11.3.*
395. What the boat does when every link goes quiet, and how the Starlink link is secured: neither is written down (found 2026-10-08). The Coast Guard answers needed both. Suggested there: with no comms the boat keeps to its plan (inside the box during the trial, on its route at sea), lights and AIS as normal; shore calls the Coast Guard after 2 hours silent near shore and in the trial, 12 hours at sea (3 missed reports at the 4-hour rate). Link: commands over Starlink carry the same HMAC signature as Iridium ones; Starlink traffic inside an encrypted tunnel (WireGuard) with no open ports; shore server logins per person with two-factor sign-in. To confirm with the software side and write into the software repo; the guides need only the comms-loss line. *Documents: Build Guide secs 8.3 and 9.1; software repo.*
382. Pi B's converter while Pi B sleeps (found 2026-10-07; Mike's call). v10.2 writes in the watchdog design Mike ordered the relays for (item 375): each Pi's 24 V feed runs through the normally closed contact of its watchdog's relay, ahead of the Pi's converter, and each watchdog runs on its own Pololu D36V6F5. With the coil off nearly all the time, Pi B's potted Klnuoxj stays powered while Pi B sleeps (Pi B only shuts itself down), and its idle draw is not published: converters of this kind typically idle at 10 to 30 mA at 24 V, about 6 to 17 Wh a day (inferred) of the backup island's 25 to 32 Wh budget. The question: should WD-B hold its coil on while Pi B sleeps, so the converter is fully off (about 0.4 W of coil instead, a guess at a 5 V relay coil: roughly 10 Wh a day), or accept the converter's idle draw? Claude's opinion: accept it; holding the coil costs about as much as it saves and turns the fail-safe relay into one that must stay energized. No bench test (Mike, 2026-10-07). *Documents: Compute Bay 4.2, 4.3, 5.3.*
373. Starlink Mini on deck: what covers and holds it, and how its Ethernet comes out (asked by Mike, 2026-10-05). Mike is worried the bare Mini is not tough enough for the open sea. The docs say only "a sealed enclosure" (Build Guide comms section; Electronics plan A13) and name no part. Facts (Starlink spec): the Mini is IP67 with its cable plugged in, works in wind of 60 mph and more, -22 to 122 F. Below deck is out: the deck is 3/4 in glassed okoume, and wet wood soaks up its signal (Claude, inferred). Claude's recommendation: the Seaview APEX low-profile mount, marine aluminum, powder coated, 2 in tall, the cable through a sealed pass-through (Amazon B0GGD3L6SY, ), with its HDPE protective cover (APXMRMPC; Seaview says no measurable loss with it on; not yet found on Amazon). Bolt it through the deck with backing, low, where green water hits it least; check signal in the Starlink app with and without the cover before sailing. Price not checked. Ethernet (Mike, 2026-10-05, confirmed on the Starlink spec sheet): the Mini has one latching Ethernet port made for Starlink's own plug (Starlink spec sheet). A standard RJ45 plug fits, but the Mini is then no longer rated IP67, so seal that plug the same way as the box jacks: dielectric grease, then self-fusing silicone tape (item 370). The docs' "Starlink coax" is really its Ethernet cable and its DC power cable (12 to 48 V on a DC5521 barrel plug); the Mini has no coax. Mike, 2026-10-05: the Seaview mount and cover take too much room; wants something smaller, and thinks green water would be too hard on the Mini. Claude's idea: mount the Mini flat on a short G10 arm off the instrument post's starboard face, about 40 to 44 in up, below the wind sensor (head at 51-1/2 in) so it does not foul the sensor's clean air, and inside the air terminal's protected zone. Up there green water rarely reaches it; flat, it shows only its 1-1/2 in edge to wind and spray. The G10 arm is the part meant to break first in a rollover, per the post's weakest-first rule. Check against the post's starboard face layout (the radar reflector and the cable exit are there) and whether the post's top 6 in blocks any of its view. Mounting part not picked. Mike, 2026-10-05: no good room on the post; the Mini should lie flat on or under the deck. Under the deck: the wood is sealed in epoxy and glass, so it stays dry. Claude's estimate (inferred, not measured): sealed, dry 3/4 in okoume with its glass and paint costs the signal roughly 1 to 3 dB; a film of salt water on top costs more, but it does that to a bare dish too. Test before deciding (next step): glass and paint a scrap of the deck plywood the same way as the deck, lay it over the Mini outdoors, and compare the Starlink app's speed test and signal with and without it, then again with the scrap wet from a hose or spray bottle (tap water shows most of the effect; for salt, about 1/2 cup of table salt per gallon is sea strength). If plywood costs too much: cut a window in the deck panel over the Mini and fill it with closed-cell foam core glassed both sides, the way radomes are built (foam barely touches the signal). Heat is the bigger risk under the deck: the spec sheet gives -22 to 122 F operating (it does not say air or case) and 25 to 40 W of its own heat; owners report it shutting itself off in full sun with no air moving, reportedly near 130 F (forum reports, unverified). Under a sunny deck it would sit in a hot pocket with no breeze. So under the deck it needs a real air path, and an on-deck cover that shades it with an air gap beneath may run cooler than either bare sun or under the deck. Mike, 2026-10-06: under the deck is the cabin, one open space heated by the sun and cooled by the sea, not a closed pocket. Claude: agreed, so the Mini sees cabin air, but the top of the cabin is its warmest air and nothing moves it; estimate 100 to 115 F there on a sunny tropical day (a guess), against the 122 F rating. A small fan (1 to 2 W) blowing cabin air across the Mini's back would likely give enough margin. Have the Pi log the Mini's heat alerts (the Starlink app shows thermal throttle and shutdown alerts). If the deck route fails: a low, rugged cover on deck, not yet picked. *Documents: Build Guide comms section; Electronics plan A13 and sec 6; Instrument Post Guide 7 and 8; BOM 15.8.*
293. Where the hatch and the post go, since the panel layout hangs on both. Mike, 2026-09-26: the hatch can go anywhere, and the post is in the last 12 in of deck and about 6 in wide. The best layout found puts the hatch across the boat at 7.4 to 9.7 ft, just ahead of the pylon, with a 50 W panel each side of it, which costs about 100 W less than any other spot. Four deck measurements decide the rest, and all four want checking before the panels are ordered: deck width at 24 ft (two 200 W panels side by side need 62 in clear inside the 2.5 in edge band, or the stern loses 100 W), the pylon keep out width (36 in or less), deck width at 3.8 ft for the bow pair (about 1 in to spare each side), and confirmation the hatch may sit over that forward compartment. No panel crosses the centerline peak in this layout, which avoids rounding the ridge. The layout is written into Build Guide 4.4 at the best case (2026-09-28). Also to settle with it: the Instrument Post Guide's 30 in keep-out square round the post may reach forward onto the two 200 W stern panels, which run to 24.0 ft. *Documents: Build Guide 3.1, 3.2 and 4.4; Instrument Post Guide 3.2 and 11.*
283. Boat 2's route plan and year. Two plans exist: the Routing Guide (Brunswick, early January 2028) and San Diego Routing (mid-August 2027). A mid-August 2027 start cannot follow Boat 1's mid-November 2027 return, so the San Diego plan's first workable start is mid-August 2028, written in as such. Recommendation: keep the Routing Guide's Brunswick plan as the starting point and move San Diego Routing to Reference once Boat 1's date is firm. *Documents: Routing Guide 1.3, San Diego Routing 1, Science Payload Strategy 1.* Mike, 2026-09-26: wait and see. Not a question again until Boat 1's date is firm. San Diego Routing moved to Reference at v10.0super (2026-09-27), on hold with it.
322. Tabbing widths for the pylon bulkheads and the frame grid (2026-09-28). Build Guide 3.1 ("Tab the grid") says three layers of 17 oz biax "4 in wide, lapping 3 in onto frame and 3 in onto hull". Both cannot be true: a 4 in strip laps 2 in each side; 3 in each side takes a 6 in strip. Mike asked whether 4 in and 6 in biax do enough on the pylon bulkheads (the three wing support bulkheads, Rig Base Two Plate Design sec 5), which meet each hull side over about 20 in (19 3/4 in deep at the ends). Yes. First pick (since revised, below): stepped, 4 in, then 6 in, then 6 in, on both faces at both ends, over a thickened-epoxy fillet, the hull ground to bare glass about 4 in out. Load, if the three bulkheads alone carried a sideways tip to the hull sides (the pylon box also carries it, so this is high): about 50 lb per inch of joint at the governing load (16,200 ft-lb at the deck) and about 150 lb per inch if the mast bends to breaking (46,800 ft-lb, Wing Load Path Recheck finding 2); allow two to three times that at the joint ends. Hold, estimated (the plywood's face veneer tearing is usually the weak part, not the glass): about 800 to 1,200 lb per inch with 2 in laps both faces, about 1,200 to 1,800 with 3 in laps. The Wing Load Path Recheck got the same answer (78 lb per inch against over 1,000). Wider than 3 in adds little: the load goes in near the lap ends. The grind, the fillet and no dry spots matter more than the width. The same 4, 6, 6 schedule fits every grid frame. Outside rules, checked 2026-09-28: Lloyd's small craft rule, first layer at least 2 in each side (as quoted by a naval architect on Boat Design Net); ISO 12215-6, flange 3 times the bulkhead thickness, capped at 3 in, so about 2-1/4 in for 3/4 in plywood (from a summary of the standard, text not read); common practice 2 in minimum, 3 in better. So 2 in and 3 in laps meet them. The weak way is peel: the bulkhead end pulling straight off the hull, worst at the bottom end of each 20 in joint. Width helps little there; a full fillet and more layers help. Peel strength of hand-laid tabbing is not known here. Revised the same day after Mike asked whether one layer is enough: by the numbers, yes, for the load along the joint. One layer of 17 oz biax on each face holds about 300 to 500 lb per inch (the plywood face tears first, so extra layers barely raise this) against about 150 needed at mast break, so about 4 times. The fourth layer above had no number behind it and is withdrawn. Claude's pick now: two layers each face, 4 in then 6 in, wrapped round the bulkhead's bottom corner onto the hull. The second layer is for what the numbers do not cover: a dry spot or bubble in a single 1/32 in layer is a hole straight through the joint, and a thicker tab resists peel better. Mike to say one, two or three layers; the Build Guide 3.1 wording follows his answer. Nothing to do unless Mike says otherwise; then Build Guide 3.1 gets the stepped wording. *Documents: Build Guide 3.1 ("Tab the grid"); Rig Base Two Plate Design sec 5 ("17 oz biax both faces" gets the same schedule).*
324. The rig base under the main bolts: decided 2026-09-30, written into the guides the same day; only the buying is left (tracker). 2026-10-12 (Mike): four more 3/8 in edge bolts, one at each end of each middle angle where plate 1's side edge crosses it; the top view drawing (Reference/Rig_Base_New_Plan_Top_View_2026-10-12) and Rig Base sec 5 now show 36 edge bolts; 12 of them are the 2 in hatch bolts on hand (tracker). Same day (Mike): the stainless angle sticks come with 12 holes, 3/16 in, 3 in on center in the upright leg; they are the sideways bolt holes, drilled out to 17/32 in. 2026-10-19: all 36 edge bolts are the 3/8-16 x 2 in bolts (30 on hand, 6 to buy); the long answer to "what size bolts everywhere" is in Rig Base sec 5 and the top view's cut list. Every hole in the stainless angles, unfolded, and where to cut the 48 in sticks (four sticks for the 36 in angles, a fifth for the side pieces): Reference/Stainless_Angle_Holes_2026-10-19.pdf (2026-10-19). The 12 head bars from the 36 in bar, cuts, holes (5/8 in from the upright edge) and how one sits: Reference/Head_Bars_2026-10-19.pdf (2026-10-19). Middle angles: 10 bolts, the two holes at 10-1/2 in each side of center left undrilled (they sit behind the head bars); front and back angles: all 12. Rig Base sec 5; top view drawing; which holes: Reference/Middle_Angle_Bolt_Holes_2026-10-12. Mike: long angles 3 x 3 x 1/2 in 6061 aluminum, 2 x 2 x 1/4 in stainless (on hand) everywhere else. Nuts: two bronze nuts per bolt, no jam nuts; so 4 in main bolts at the eight front, back and side bolts and 4-1/2 in at the four corners (thread length sets it), a 316 washer under the first nut at the corners and sides; back row slots 13/16 x 1-5/8 in. Priced list in the recount file: about $590 before shipping and tax. Written in 2026-09-30: Rig Base secs 3, 4, 5 and 7; Build Guide 3.3; Design Decisions R-3.3A and the bolt decision; BOM 15.2; Wing Can Spec Sheet (plate 3 holes and pads). Left: order the parts (tracker row BUY) and release the shop sheet after Mike reads it (item 343). Earlier text kept below. Decided 2026-09-30 (Mike): the middle bulkhead angles are the 2 x 2 x 1/4 in stainless on hand, with the head bars under the four side bolts, which move to 2 in from the middle bulkhead's center; long angles 11-3/8 in; the bolt through both middle uprights 1/2 x 3-1/2 in; jam nuts silicon bronze. Drawings, shop sheet and recount redrawn (all 2026-09-30). Still open: the long angle size (Claude: 3 x 3 x 1/2 in aluminum, one 72 in and one 24 in, $229) and whether the front and back angles stay stainless (0.9 at mast break) or go 3 x 3 x 1/2 in aluminum (2.0, $153 plus 24 edge bolts). About $570 all in with stainless outer angles, $745 with aluminum. v10.1 writes in what Mike decided (Rig Base Two Plate Design secs 3 to 8): the pylon bulkheads glued in, the outer two 36 in apart outside to outside; the middle one tripled with an angle on each face; plate 1 bolted to 2 x 2 x 1/4 in stainless angles on the bulkheads and doublers (no angle rail on a pylon box rim, no webs, no brackets); plates 1 and 3 both 1/2 in; angle under every main bolt, the long angles butted fore and aft between the outer and middle angles; new angle is aluminum; end bolts on every long angle, the inner one up from below into a hole tapped in plate 1; 12 of the 3/8 in edge bolts along each outer angle; the twelve 3/4-10 x 3-1/2 in 316 main bolts stay, with no bolt test; heads at the bottom spot welded to a 2 x 3 in stainless head bar against the angle, nuts on top of plate 3 (this also answers how the heads are held once the rail is gone); only the back row of plate 3's holes are fold slots. Waiting on Mike: (a) which of the three buys in Reference/Rig_Base_Strength_Recount_2026-09-29.md: A (2-1/2 in x 3/8 aluminum, 1/2 in bar, 4 in bolts, $447), B (3 x 3 x 1/2 in, 1/4 in bar, bolts on hand, $481) or C (3 x 3 x 1/2 in, 1/2 in bar, 4 in bolts with a bronze nut and a jam nut, $615; Claude's pick); the jam nut in C is 316, which clashes with the Rig Base rule "never 316 nuts": use a silicon bronze jam nut (made in 3/4-10, 27/64 in thick, same height as the 316 one, so the bolt length figures stand; Albany County Fasteners lists them, 2026-09-30); (b) the 1/4 in 316 corner pads on plate 3; (c) the plate 3 hole pattern as drawn in Reference/Plate_3_Shop_Drawing_2026-10-12 (four across front and back, corners included; two each side of the middle bulkhead); (d) 3M 5200 in the clearance hole round each main bolt. The shop sheet draft shows (b) and (c) as firm: settle them before it is released. *Documents: Rig Base Two Plate Design secs 3 to 8; Wing Can Spec Sheet; BOM 15.2 and 15.6; order tracker (new line, Your call).* Notes 2026-09-30 (Claude, checking v10.1 against the session): with the XXS mast every recount figure is 0.8x, so in option C the outer stainless flat legs come out at about 0.9 at mast break, the one part under 1 (aluminum there gives 2.0); the recount file now says so at its top. The stainless angle on hand for the rig base is the two Amazon sticks (the six McMaster sticks are the wing pylon's, item 239); the new layout needs about 114 in of it (two 36 in outer angles and four side pieces of about 10-1/2 in), checked 2026-09-30: the Amazon sticks are 4 ft each (Amazon B0F1KNLHDR, 48 in), so 96 in, about 18 in short on their own; the six 4 ft McMaster sticks (the BOM's "wing pylon angle", and the pylon bulkheads are these bulkheads) are 288 in more. Both lots are for this job, 32 ft on hand against about 9-1/2 ft needed; Rig Base sec 5 now says so. Parts list and cost for C with aluminum outer angles too: about $995 with guesses ($708 for the metal and main bolts), in the recount file, 2026-09-30. Rig Base sec 3 "Mounting it" says "no tapped holes or inserts" (about the bearing bolts) while sec 5 threads the long angles' inner end bolts into plate 1; word it so they do not read as a clash. Bought 2026-09-30: Speedy Metals 620450 ($524.49: two 72 in sticks of the 3 x 3 x 1/2 in angle, 53 in more than the 91 in needed; the 1/2 x 2 in bar; the 1/4 x 4 in pad stock) and Bolt Depot 2630993 ($328.39: the main bolts, the 1/2-13 bolts including the 24 wood bolts, nuts and washers). Left to buy: the 1/2-13 tap and 27/64 in drill (tracker Buy list).
340. The wet lab drip tray, now there is no leak sensor in it (found at v10.1). A leak caught in a walled tray never reaches the bilge sensors, so a slow wet lab leak shows only when the tray overflows. Options: a drain from the tray to the bilge, or a tray with no walls. Mike's call. *Documents: Flow-Through Wet Lab Design 7 and 12.*
343. Read the welding shop's sheet before it is released. Reference/Wing_Can_Welder_Instructions_2026-10-13_v13.pdf Cut files for the plasma table (2026-10-19, Mike's ask): Reference/Plate_3_Cut_File_2026-10-19.dxf (plate 3's outline, 8 round holes 15/16 in and 4 slots 15/16 x 1-3/4 in on layer CUT; can, post and pad outlines on LAYOUT, not cut) and Reference/Plate_3_Corner_Pads_Cut_File_2026-10-19.dxf (the four 4 x 4 pads laid out on Mike's 18 x 4 x 1/4 in bar with their holes and slots; script _make_Plate3_DXF_2026-10-19.py), each with a PDF and PNG; give them to the shop with the sheet. (seven US letter pages; sources and drawing script in Reference/_shop_instructions_src/; every earlier sheet is in Archive/). The welder's copy is rewritten in plain words (2026-10-02, Mike: no jargon), named Wing Can Welder Instructions; sources welder.html and rev3_drawings.py in Reference/_shop_instructions_src/. Rev5 (2026-10-01): no machining; every flat part, holes included, CNC plasma cut, the lip holes before welding; the bearing bedded by Mike on thickened epoxy (Mike: simplify so one shop can do it all). Rev4 used the 21 in tube uncut, so the lip face is about 21-1/4 in above plate 3, and the post is 9-1/4 in (Mike). Rev3 took in Josh's review (Reference/Shop_Instructions_Rev2_Review_2026-10-01.docx) and Mike's answers of 2026-10-01: Josh Johnston (918-606-0099) is the shop's contact; no base ring; slots 15/16 x 1-3/4 in; lip holes 3/4 in; bearing seat from 16.30 in dia; the mast plate disc cut out by plasma or waterjet before the bore is finished; a cut list, top view, dimensioned elevation, lip section, weld map, clamping and separate tools for steel and stainless. Not taken from the review: the mast plate stays 304 (decided 2026-09-29) and the corner bolts are already 4-1/2 in. Left: Mike's read-through before release. Hand-holes now two 7 x 9 in ovals with bolted covers, at port and 150 degrees (2026-10-02). The corner pads and the 12-hole pattern wait on the rig base buy (item 324). The three rig base drawings (Plate_3_Shop_Drawing_2026-10-12, the top and side views of the same date) now carry the 15/16 x 1-3/4 in slots and no base ring, matching v11 (2026-10-08). *Documents: the shop sheet; Wing Can Spec Sheet.*
348. Lightning: two questions left from the 2026-10-01 research. The research (Reference/reports/USV lightning risk and protection.md): a direct hit on Boat 1's loop is about 1 in 930 (computed); a strike most likely leaves a sound hull with dead electronics. Still open: (1) A tracker wired to nothing. Mike, 2026-10-02: not now; ASK HIM AGAIN ON OR AFTER 2026-11-02. Costs found 2026-10-02: Iridium Edge Solar $389 plus $15 to $65 activation and about $6 to $14 a month for daily reports (about $500 for Boat 1's 11 months); RockSTAR hardware price not published, $30 activation, $12 a month at the cheapest (every 6 hours), $3 a month paused. It gives a recovery location only, after any total loss of power or electronics, not just lightning. Chance it is ever the only thing reporting (Claude's guess, 2026-10-02): about 1 in 1,000 for Boat 1's loop, since both Iridium chains would have to die while the boat stays afloat (a sinking takes the tracker down too); the Viatrax units cover coastal waters but probably not mid-ocean (Starlink direct-to-cell is sold for land, lakes and coastal waters). At about $80,000 of boat that is about $80 of value against about $500, so on the numbers it does not pay for Boat 1; Claude's opinion: skip it for Boat 1, reconsider for Boat 2's longer, more remote voyage. Claude's pick, revised 2026-10-01: the Iridium Edge Solar (Iridium, sealed IP68, solar-charged with a backup battery good for about 5 years at two reports a day; $389 at Ground Control; on Amazon B096MDDJ2K, , stock not confirmed). It never needs charging, so it can sit on deck in the sun or under the deck on its backup battery, kept away from the lightning copper. Airtime plan price not found. Second choice: the RockSTAR (Ground Control, own battery, about 3 months hourly; $15 a month at every 4 hours; hardware price not found; not on Amazon). Whether the Viatrax units can report mid-Atlantic is not known. (2) Underwater fittings near the copper. Lightning leaving the hull jumps to underwater metal and can blow a fitting out; the science intake is bronze, 16 to 18 in down and kept off the lightning copper. Claude's answer to how close is too close (2026-10-01, estimate): the bar near the water may sit roughly 20,000 volts above the sea for a few millionths of a second, which jumps an inch or two of air and farther along a damp surface, so use the wiring clearance rule already in Lightning Guide 5.3, 12 in, for every underwater fitting too, inside the hull and outside. The junction and bars run at deck level, so it is probably met already; check it when the exit-bolt and strip station is set. The float switch idea is withdrawn (Claude, 2026-10-01): the backup pump has its own Water Witch, and the case it covered is too rare to be worth it. Dropped by Mike (2026-10-01): extra post-strike rechecks; the boat checks everything all the time anyway. *Documents: Build Guide (Viatrax, bilge); Boat 1 Route; Lightning Guide 5.3.*
371. Hydrophone fin: the channel cannot reach the point of the fin as cut (Claude, 2026-10-09). Written in at v10.2 (Mike, 2026-10-04 to 10-07): 3/8 in G10, about 4-1/4 in out from the hull, front edge swept back 45 degrees, back edge raked forward about 30 degrees as cut and tapered on one face over about 1 in; a 1/4 x 1/4 in groove routed in the flat face square to the hull, glassed over a greased 1/4 in rod that is pulled when cured; no tube. The groove must stay about 1-1/4 in clear of the tapered back edge and about 1/2 in from the front edge. On this fin those two lines meet about 2-3/4 to 2-7/8 in out from the hull, about 3-1/2 in behind the root's front corner (Claude's arithmetic, checked twice: root about 6.7 in), so the channel ends about 1-1/2 in short of the point. Mike's call: (a) keep this fin, and the cable leaves it about 2-3/4 in out, so the hydrophone trails about 3 in off the hull; or (b) cut a new fin from G10 offcut with a 2-1/2 in flat tip, so the channel runs the full 4-1/4 in (the 2026-10-04 shape: root about 6-3/4 in with a square back edge). Claude's pick: (b). The fin was set at 4-1/4 in to clear the slimed layer late in the trip (3 to 3-1/2 in, Claude's estimate), and (a) puts the hydrophone back inside it; a new fin costs an offcut and an afternoon. The drawings show the fin as cut, the clear zone and this open point: Reference/Hydrophone_Fin_Face_Groove_2026-10-04.png and Reference/Hydrophone_Fin_Root_Joint_2026-10-04.png (redrawn 2026-10-09). Not written in, Claude's advice only: the root joint method, the pre-curved sleeve, the edge finishing figures. The full exchange is in Archive/Pending_Amendments_v10_2.md item 371. *Documents: Science Payload Strategy 5; Design Decisions R-10.2B (Open lines in both).*
365. Root rib socket hole is 4.30 in, the socket tube is 3-11/16 in (found 2026-10-05, Brandon through Josh). The 32 root ribs, already cut, have a 4.30 in round hole (Wing Construction Guide Stage 1 and Appendix A; Foam Piece Key). The socket tube Brandon molded on the 3.500 in mast, four wraps of biax as written, is 3-11/16 in across, so the ribs leave a gap of about 5/16 in round the tube's exposed side. Claude's guess at the cause: 4.30 in fits a 4.000 in pipe (3-1/2 in nominal pipe size) plus a socket wall, and was never resized when the mast became 3.500 in OD. The 1.4 in center spacing still holds: the rib holes fix both centers (round hole 6.93 in back, spar 8.31 in back), so the tube sits 1.38 in ahead of the spar center and takes less off the spar than a 4.30 in tube would; Brandon was told to carry on (2026-10-05). Mike's call: fill the gap as the ribs go on with epoxy thickened with a light filler (Claude's pick: about 2 lb, nothing changes on the spar), or build the tube up to about 4-1/4 in with about six more wraps of biax over the bottom 24 in (about 3 lb, stronger socket, but the spar scallop must be cut deeper). *Documents: Wing Construction Guide Stages 1, 4 and 6, Appendix A; Foam Piece Key.*
374. Parts bought in early October that no document names (Claude, from the order mail, 2026-10-06 to 10-08). Mike to say what each is for: (1) a TICONN IP67 box ($68.39, with the Flipsky, Amazon 114-2434294-4539414; maybe the second drogue box, since the QILIPSU went back Oct 1); (3) two more Taidacent RS-12V/2S surge modules bought Oct 7 for the DO sensor and SBE 45 lines (no guide puts surge parts on those lines); (4) a second AS3935 lightning board (Playing With Fusion SEN-39003, $34.91, Oct 6): ANSWERED 2026-10-14, Mike is wiring it into the detector mini-box; see item 420; (5) two Pi GPIO screw-terminal boards ($31.76, Oct 6); (6) a Winford BRK2X6-L-FT board ($14.45, Oct 3), beside the BRK2X6-L-DIN the battery taps use; (7) a MECCANIXITY 12-pin 6 in ribbon ($6.19, here Oct 6): is it the 12-pin female-to-female ribbon the Battery Box guide still lists to buy (Amazon B07L9Q3R2K)? *Documents: BOM; Battery Box Building Guide Materials; order tracker.*
367. The Abireiv outdoor Cat6 cables look shielded; the docs say unshielded (Claude, 2026-10-03, from Mike's photo of a 10 ft one). The plug has a metal shell, which normally means a foil or braid shield inside (inferred, not confirmed). Wire and Cable Buy List B2 says outdoor Cat6 is unshielded so no shield brings a metal path into the hull; that was written before the surge protectors moved to the board near the deck, wired to the lightning copper (item 359). Claude's view: shielded is fine on the deck legs, since each one ends at its surge protector, which is where a shield is meant to drain; the inside legs down to the switch box stay unshielded (B3). Check: cut the jacket back an inch on scrap or read the jacket print for FTP, STP or S/FTP. Salt: the jacket (UV-rated, direct burial) is fine wet; the plug ends are the weak point, so every plug sits inside the hull or under the device's waterproof cap, never bare on deck. *Documents: Wire and Cable Buy List B2; Electronics plan A6a; Lightning Guide 7.4.*
318. Three parts bought 2026-09-26 and 27 that no document names. (1) Kenhihi 6 AWG tinned silicone wire, 10 ft red and 10 ft black, $35.98 (Amazon 112-2902556-7813006, with the release trunk 14/4). (2) A 169-piece bonded seal washer kit, M6 to M24, $20.61, here Sep 27; Claude's guess: for the M20 x 1.5 barbs bought Sep 22. (3) A 5-pack of 1-1/4 x 1-1/4 x 1 in PVC reducing tees, $16.65, bought with the DO chamber bushings and unions; the chamber as written uses the plain 1-1/4 in tees on hand. Also bought 2026-09-28 and 29 with no document line: (4) a YONICO 1/8 x 5/16 in slot cutter router bit, $16.95; (5) eight 1/2-13 x 2 in 18-8 bolts and 25 304 nuts, $23.80; (6) a US Stainless 316 5/16 in part, $9.89 (name cut off in the mail); (7) six VAMRONE 4 in aluminum DIN rails, $6.64; (8) a QILIPSU UL94-V0 16.4 in box, $82.15 (probably the second drogue box). Say what each is for, and it gets its line. *Documents: order tracker (Do next, Order log); BOM.*
441. The pad under the shaft thrust block is neoprene, not EPDM (Claude, 2026-10-20). The McMaster neoprene sheet that came 2026-10-20 (1290N252, 1/4 x 6 x 6 in, 50A, the last line of the 2026-08-18 order) is the pad under the thrust block: the thrust block parts list picked it (Reference/Thrust_Bearing_Selection_2026-08-11.md, item 4c), with the ribbed neoprene bushings and washers on the same order for its bolts. The guides had drifted to "EPDM pads". Neoprene stays: the block is packed with grease and sits in the bilge, and oil and grease swell EPDM, while neoprene stands them fairly well. Written into Build Guide 3.5, Design Decisions R-3.5C and BOM 15.5; the fastener list's washer line and the order tracker now say the washers and bushings are the thrust block's, not the motor bed bolts'. *Documents: Build Guide 3.5; Design Decisions R-3.5C; BOM 15.5; Reference/Fastener_Inventory_Bought_2026-10-03; order tracker.*
Nine reviewers read every v10.3 guide against the home guides, the order tracker and the decisions here; about 290 fixes went into the guides the same day. The main ones, so nobody re-opens them:
v10.3 is the v10.2a set with these written in. Left over: item 438.
v10.2a is the v10.2 set with these Part 1A items written in. Left over: item 436. The Iridium box itself is open again (item 435).
433. Cell terminal Belleville washers with flange nuts (Mike, 2026-10-15: "should you use belleview washers with flange nuts on batteries" and "with flange nut which way does belleview go?"; Claude's answer, standing unless Mike says otherwise). Yes: the M6 Belleville goes under the flange nut, the same as under any nut. The flange nut counts as the guide's "locking nut" when used with Loctite 242. Which way up: hollow side down, against the ring lug or busbar; the raised middle points up into the nut's flange, like a small hill. Tightening squashes the hill flat and the washer then pushes back as a spring. Upside down, it rocks on its outer rim and does nothing useful. Serrated flange nuts: the teeth bite the stainless washer, which is fine; never let a serrated flange sit straight on the busbar or lug. The washer sits on top of the stack, so no current passes through it (the stack order already in Battery Box Part 7 step 13). Torque stays 37 in-lb (the MB31 sheet check is in item 310). Loctite: keep it, blue (Mike, 2026-10-15: "ok keep loctite"). Blue 242 (on hand, two bottles, July) on the top threads only, threads wiped clean of grease, never on a face that carries current. Never red (262, 263, 271, 272): it needs about 500 F or a force that can snap the welded stud to come off. Torque wrench: the 1/4 in drive 20 to 200 in-lb click wrench on hand; 37 in-lb is just under its rated range (20 percent of full scale, 40 in-lb) but close enough, nothing to buy. Tape the wrench and any extension so only the socket tip is bare; cover the next terminals. The 24 hr re-torque is dropped (Mike, 2026-10-15: "yes drop the 24 hour retorque, was senseless after loctite"). Once the blue sets, the wrench clicks without turning the nut and proves nothing; the Belleville takes the settling. Torque once. *Documents: Battery Box Building Guide Part 7 step 13 (add the flange nut, the which-way-up line, blue only, the taped wrench; strike "torque again at 37 in-lb after a 24 hr settle"); Glossary (Belleville washer: which way up).*
432. How each Iridium modem connects to its Pi: its own USB cable, through a USB isolator (DECIDED, Mike, 2026-10-15: "ok, new diagram please", after Claude's pick of this plan; reached through "how does iridium connect to pi, with usb cable included?", "can we use the serial to usb in the rockblock cable?" and "i still don't understand why rockblock cant get power over usb"). The plan. The cable SparkFun ships with each RockBLOCK is an FTDI USB to serial cable (TTL-232R-3V3 with the RockBLOCK's 10-pin plug); its USB to serial chip sits in the USB plug, and its four wires at the small plug are black pin 10 ground, red pin 8 5 V in, orange pin 6 data into the modem, yellow pin 1 data out (Mike's photo of the plug, 2026-10-15). Cut each cable 8 in from the small plug, inside the Iridium box. Rejoin yellow to yellow and orange to orange in 2-way lever connectors; black to black in a 3-way lever connector with a short black wire to the Mean Well DDR-15G-5's −V; the small plug's red to the Mean Well +V; the long piece's red cut short and taped (brown or green at the cut, if any, the same). The long piece leaves the box through a gland, and its USB plug goes into a DSD TECH SH-G01A USB isolator (ADuM3160; 1,500 V power and 2,500 V signal isolation, 400 mA out; Amazon B093LKN3YH), which plugs into a port on Pi A's USB hub (Iridium A) or a USB port on Pi B (Iridium B). The cable enters the hub box or the Pi B box by a split gland like the battery watchman's USB lead (Pi B: its spare entry 7). One isolator on hand (bought Aug 23, unused since 2026-10-07); one to buy for Iridium B. If the cable is too short (its length is not known; FTDI's standard cable is about 6 ft, Claude's reading), a USB extension goes on the Pi side of the isolator, never between the cable and the isolator. Pin 7 (on/off) has no wire: Ground Control lists a floating on/off pin as on. Software: the modem is a USB serial port at 19,200 baud, flow control off from the factory. Why this way. The modem's ground is tied to the lightning copper (the antenna cable's shield goes to the stub arrestor), so a plain USB link would tie the Pi to the copper; the isolator stops that path, and only data crosses it. The modem cannot take its power through the isolator: Ground Control says the 9603 needs a supply of at least 500 mA ("a lower current source will cause brownouts during transmission"), and the isolator gives 400 mA, so the modem runs on the Mean Well (isolated, item 357) and the isolator powers only the chip in the USB plug, a few mA. The isolated Mean Well stays (Mike asked, 2026-10-15: "is there any reason we need the fancy converter instead of just the cheap 24 to 5 volt?"; Mike, 2026-10-16: "looks good"). The USB isolator breaks the path to the Pi; the Mean Well breaks the path to the battery. A potted Klnuoxj joins its 5 V minus to its 24 V minus, which would tie the main negative (A) or the emergency island's floating negative (B) to the lightning copper through the antenna stub. Do not re-propose. Not chosen: the four-wire link through an ADuM1201 isolator board in the Pi box, with a Pololu 3.3 V regulator (its modem side must run on 3.3 V to read the modem's 3.3 V signals) and a cut NMEA 2000 cord (rev 1 and rev 2 of the drawing, earlier on 2026-10-15); a Waveshare isolated USB to serial converter (FT232RNL, Amazon B0BTYKS8LK; isolation rating not published; Archive/Iridium_USB_Option_Wiring_2026-10-15); the Ground Control tails cable (US$17). So the ADuM1201 boards and the two 3.3 V Pololus have no Iridium job, and the Compute Bay Plan's "one ADuM1201 in each box for its Iridium serial line" goes. Logic level settled (Ground Control: 3.3 V logic, 5 V tolerant); closes the "Iridium modem logic level" open check. The drawing: Iridium_to_Pi_Wiring_2026-10-15 (rev 3; .html master, .pdf to print). *Documents: Electronics and Wiring Plan A11 (contents: the cut cable, three lever connectors; entries: the USB cable in place of the serial line) and A1a (hub port for Iridium A) and A2 (Pi B entry 7 for Iridium B's USB cable) and 6.5; Compute Bay 4.1, 4.2 (the ADuM1201 lines go; USB to the hub and Pi B), 14 (Iridium serial runs: the USB isolator); Instrument Post Guide 16.1 (logic level settled; the cable is used, cut); Wire List (no Iridium serial cords); Glossary (USB isolator); Box Wiring Sheets A11; BOM 15.8 and 15.12; order tracker.*
431. Backup battery board: one board beside the Dyness in place of the "small fused DIN strip" (Mike, 2026-10-15: "should we make a board and mount the board near the backup battery?", then "if so, please make board layout pdf"; the layout is Claude's pick, standing unless Mike says otherwise). The drawing, parts and wire list: Backup_Battery_Board_Layout_2026-10-15 (.html master, .pdf to print). A 16 x 12 in board (1/2 in marine plywood or 1/4 in G10, no bond wire) within a foot of the Dyness, standing up, holding the Orion-Tr and the SmartSolar 75/10 side by side with 2 in of air round each, the 30 A ANL at the corner nearest the + post, its 6 AWG from the post in split loom its whole length, as short as the board allows and 72 in at most (ABYC E-11 allows 72 in for a cable in a sheath, 7 in bare; Mike, 2026-10-15: "under 7 inch for pos? not realistic"; the Dyness BMS also cuts off a dead short; A10's "within a few inches of the positive post" changes to this), a red and black bus bar pair, and the load breakers. Loads on 5 A auto-reset stud breakers, not fuses, by the 2026-09-28 rule that put Iridium A on a breaker: Pi B 5 A and Iridium B 5 A, each on 16/2 (Pi B worst case about 1.2 A, Iridium B about 0.7 A). Chargers stay fused, as on the main board: the Orion-Tr output 10 A (as A10 says) and the 75/10 battery side 15 A (1-1/4 times its 10 A, next size up), sealed inline blade holders, 14/2. The Orion-Tr input (16/2 from main board breaker 13) and its remote wire (central node channel 10) keep to a marked strip on the left edge; nothing else on the board touches the main battery. A spare + and - stud and a spare breaker spot are left. Settles the strip's open fuse sizes (A10; item 427) and item 352 (b), the Pi B fuse: it is the breaker on this board, none in the box. The four DIN fuse holders the strip would have taken are free for the cord pullers. To buy: the bus bar pair and the sixth surge module (tracker Buy list, 2026-10-15). The two breakers are likely already on the shelf (the Sep 7 order bought about 30; 15 are on the main board). Mike, 2026-10-15: "yes, add the surge module": a sixth NanQue PV surge module ($15.50, Amazon B0DZ267ZV4) on the board for the two 50 W panels, its ground screw to its own PV minus, never the lightning copper (the main board gives each controller one; the backup string had none; on the tracker Buy list 2026-10-15). Found: Build Guide sec 8.3 puts Primelink unit B on the emergency bank, but A10 and Compute Bay 5.3 list only Pi B and Iridium B; the spare stud and breaker spot cover it if it lands here. All wire on the board 14/2, except the two 6 AWG battery leads (Mike, 2026-10-15: "make everything 14awg except for 6awg feed lines"). That covers the Orion-Tr feed from main board breaker 13 (was 16/2; stud table, Electronics plan 7.6), its remote pair from central node channel 10, the deck pair to the 75/10, the Pi B and Iridium B feeds (were 16/2), and the board jumpers. 14/2 does not fit a PG7 (about 1/8 to 1/4 in), so three entries go from PG7 in a 1/2 in hole to PG11 in a 3/4 in hole: Pi B box entry 1 (A2), the Iridium B box power entry (A11), central node entry 28 (A3). The Orion-Tr's 2-pin remote plug (5 mm spacing) takes 14 AWG (inferred from that plug type). About 60 ft of 14/2 (a guess); the Sep 4 roll is spoken for, so a second GS Power 100 ft roll is on the tracker Buy list. *Documents: Electronics and Wiring Plan A10 (the board replaces the DIN strip; A2, A3 entry 28 and A11 glands; Wire and Cable Buy List 14/2; fuse sizes; breaker rule) and sec 7.6 (the "Still fused" list); Compute Bay 4.2 (Pi B fed through the board's breaker); Build Guide 4.3; BOM 15.4; Box Wiring Sheets if a sheet is wanted; order tracker.*
412. The six DS18B20 pack probes: where they go and how each is named (Claude's proposal, Mike, 2026-10-15: "yes"). Three per 16-cell stack: one on each of the stack's two middle groups (the cells heat has the farthest to travel from) and one on an end group, each held against its cell face the way the XT1 thermistor is held (Battery Box guide Part 8), reading the metal, not the air. Each probe is named by its factory serial number, so its place on the chain does not matter. At setup, before mounting, read each probe's number one at a time with the spare Nano Every on a PC (Arduino IDE, the OneWire library's DS18x20_Temperature example, data on D10) through a DS18B20 screw-terminal adapter board with its 4.7 k resistor built in (HUABAN 2-pack kit, $9.99 on Amazon, shown by Mike 2026-10-15); tag each lead and write number-to-cell in the System Settings Sheet sec 9 and on the node lid. Probe wire colors differ by seller: go by each seller's listing. The trip itself does not care which probe is which. *Documents: Battery Box Building Guide Part 8; System Settings Sheet sec 9; pack wiring diagram.*
428. Wet lab box, a new box A18: a small node on the house bus that switches each wet lab part and reads its sensors (Mike, 2026-10-14: "yes, lay out the wet lab box"; the layout is Claude's pick, standing unless Mike says otherwise). The layout, parts, relay map, inputs, fuses and hole list: Reference/Wet_Lab_Box_2026-10-14 (.md and .docx); drawing with the plate layout, wiring and tables: Wet_Lab_Box_Layout_2026-10-15 (.html master, .pdf to print, top of the folder); every part with a photo on one page: Wet_Lab_Box_Parts_2026-10-15 (.html, .pdf). Why: the wet lab line has one on/off switch (central node relay channel 2), but the hourly cycle (Wet Lab Design sec 10) needs the Iwaki pump, the dosing pump, the Gulper, valve 1 (open and close lines) and valve 2 switched one at a time, and the pressure sensor, the Celsius probe, the flow meter and the valve position wires read; the guides name only "the board that reads them". In short: a NUCLEO-G431KB on an AITRIP carrier with an SN65HVD230 and a Taidacent RS-12V/2S on a house bus drop (one of the two bought Oct 7 with no job); a spare Klnuoxj; a spare HiLetgo 24 V 8-channel relay module (ch 1 Iwaki, 2 dosing pump, 3 Gulper, 4 and 5 valve 1 open and close, 6 valve 2, 7 and 8 spare; all normally open); a spare Fafeicy opto board for the flow pulses and the four valve position wires; the pressure sensor on the node's 5 V through a 10k/18k divider; the Celsius on I2C at 3.3 V; the XWST inside the box, five fused groups off it; a 1N5408 across each pump motor. The node turns everything off after 60 s with no word from Pi A, runs the pump only with valve 2 open and the Gulper only with valve 2 shut, and never powers valve 1's open and close lines together. Relay channel 2 on only around each burst (Claude's pick), so the wet lab draws nothing between bursts. The DO sensor and SBE 45 cables join their FTDI data cables in this box. Box: the TICONN IP67 box of 2026-10-03 if about 12 x 8 x 5 in or bigger (which leaves the second drogue box to fiberglass), else buy that size. To buy: one NUCLEO-G431KB (on the tracker Buy list 2026-10-14); an N2K tee if the shelf has none spare. Left out, with stand-ins: pump current (no sensor on the boat; flow rate stands in) and reservoir level (counted from doses). *Documents: Electronics and Wiring Plan (new A18; sec 5.2 and 5.4 house bus); Flow-Through Wet Lab Design secs 10 to 12; Box Wiring Sheets (new A18); BOM 15.11; System Settings Sheet (node ID, the silence rule, relay channel 2 timing).*
422. dAISy antenna: its socket is BNC, not SMA; one SO-239 to BNC jumper, no bulkhead (Claude, 2026-10-14, from the maker's page; Mike asked for the link, "yes link"). The docs run the Eifagur's coax through an SMA to SO-239 jumper and an SMA F-F bulkhead on the USB hub box (Electronics plan 6.2, 7.4 coax line and A1a; BOM 15.2 and 15.8; Wire and Cable Buy List D2; Design Decisions on the dAISy in the hub box; Box Wiring Sheet A1). The dAISy 2+ on hand (Wegmatt S15162) has a BNC antenna socket and a mini USB port (Tindie and Starpath listings). The Eifagur ends in a PL-259 plug on 16 ft of RG-58. Fix: the antenna coax, plug on, passes the hub box wall in the split gland (item 423); inside, one Superbat 1 ft RG-58 jumper, SO-239 female to BNC male (Amazon B087Q49GS1, $12.99 for a straight and a right-angle one), joins it to the dAISy. No SMA bulkhead and no SMA to SO-239 jumper; the 1/4 in SMA hole on the hub box goes. A flexible jumper, not a solid adapter, so the coax's weight does not lever on the dAISy's socket. On the order tracker Buy list 2026-10-14. *Documents: Electronics plan 6.2 and A1a, BOM 15.2 and 15.8, Wire and Cable Buy List D2, Design Decisions, Box Wiring Sheet A1.*
423. USB leads cross box walls in split-seal glands, plugs left on (Claude's pick, Mike 2026-10-14 "yes"). The Electronics plan gives each USB lead a PG7 gland in a 1/2 in hole (A1 entries S and 7, A4 entry 4); a USB-A plug with its molded grip is about 5/8 x 5/16 in and will not pass either. Use Altelix PG25 pre-terminated glands ("allows cables with existing connectors to pass through enclosures without cutting", Altelix listing, checked 2026-10-14). Hub box top wall: the 5-port on hand takes the hub's lead and the CAN adapter's lead to Pi A, the SBMS0 lead in, the dAISy's antenna coax (item 422) and the MOTU M2 lead if the MOTU sits outside (else plugged). Pi A box: entries S and 7 share the 4-port on hand, two ports plugged. BMS box entry 4: one more Altelix 4-port to buy, $6.99 (Amazon B0GYT3ZMDL; on the tracker Buy list 2026-10-14), three ports plugged. The M25 2-hole gland on hand is not a split gland as far as known; not used for USB. No cutting and rejoining USB leads (no solder), no panel-mount USB sockets (one more plug joint). Gland hole sizes come from the gland's own thread at fit-out. *Documents: Electronics plan A1, A1a, A4 and the gland totals; Box Wiring Sheets A1, A1a and A4 (drawn 2026-10-14, item 425).*
424. SBE 45 on RS-232: the FTDI USB-RS232-WE-1800-BT-0.0 cable (Mike, 2026-10-14: "SBE is rs232", then "get the 0v one?"). Settles the open RS-232 or RS-485 question (Electronics plan A1a; BOM 15.8 and 15.11; Science Payload Strategy serial link row). The 0 V version has no power out on its wires, so nothing to tape off; the SBE 45 takes its power from the boat. Amazon B00DEED3GM, $23.00, sold by Cableshopus, shipped by Amazon (checked 2026-10-14); on the tracker Buy list. It plugs into a hub port in the USB hub box; its bare wire end leaves by its own bottom gland. The Taidacent RS-12V/2S bought Oct 7 for "the SBE 45 pair if RS-485" is an RS-485 part and has no job on an RS-232 line: see the bought-but-unassigned list in 2B. *Documents: Electronics plan A1a, BOM 15.8 and 15.11, Science Payload Strategy, System Settings Sheet sec 9, Box Wiring Sheet A1a.*
425. USB hub box on its own drawing sheet, A1a (Mike, 2026-10-14: "Need usb box separate from pi a. Needs separate page(s)"). Done 2026-10-14 in Box_Wiring_Sheets_v10_2.html and its PDF: the hub box left Sheet A1 for a new Sheet A1a (drawing page, notes page) with items 422 to 424 drawn in; A1 keeps three stubs to it; the index, Sheet 0, A4 entry 4 and A14 now point to A1a. Still to write in the guides: Electronics plan A1a and A1 (items 422 to 424). *Documents: Box Wiring Sheets (done); Electronics plan.*
426. USB hub box: no fuse on the 12 V converter; the feed stays as it is (Mike, 2026-10-14: "no change on feed", "get rid of fuse"). The box's own 24 V feed (16/2 on its own 5 A breaker at the branch bar) runs straight to the third YRDZXG 24 to 12 V converter, with no fuse on its input or its 12 V output. Replaces "through one fused 24 V line inside (fuse size set at fit-out)" in Electronics plan A1a and the "1 A fuse" of the closed weather-sensor feed item (216). Claude had said a 1 A fuse on the 12 V output would guard the thin sensor wires, since a short there pulls only about 2.5 A from the 24 V side and would not trip the 5 A breaker; Mike chose none. Kept as is: the hub on Pi A's 5 V, the box's own 24 V feed for the converter (feeding the converter from Pi A was offered and declined). Sheet A1a redrawn 2026-10-14. *Documents: Electronics plan A1a, BOM 15.4, Box Wiring Sheet A1a (done).*
420. Lightning detector: the Playing With Fusion board at 5 V, no level shifter (Mike, 2026-10-14: "yes, use this board at 5 V"). The drawing with wire colors: Lightning_Detector_Wiring_2026-10-14 (.html master, .pdf to print). Colors on the Belden: red 5 V, black ground, blue IRQ to D2, green SCL to A5, white SDA to A4, brown spare cut off. The docs name the SparkFun SEN-15441 at 3.3 V with a 4-channel level shifter in the Pi B box (Compute Bay Plan 4.2 and 13.1; Electronics plan A2 entry 6 and A12; Box Wiring Sheet A12; Glossary). Mike has both boards and is wiring the Playing With Fusion SEN-39003 (bought Oct 6, here by Oct 14). Its datasheet gives 2.4 to 5.5 V for power and signals, and the maker's Arduino example powers it from the Uno's 5 V, so on the Nano Every's 5 V it needs no level shifter (Reference/AS3935_PWF_SEN-39003_Wiring_2026-10-14.md). It also comes tuned at the factory: this board's value is 88 pF (on the bag label; register value 11). The wiring: (a) wiring at the board: VDD to the lead's 5 V, GND to the lead's ground, MOSI/SDA to SDA, SCK/SCL to SCL, IRQ to IRQ; short jumpers on the board's pin row tie SI to VDD (picks I2C) and CS to GND; MISO left empty; the two Qwiic plugs are not used (they carry no IRQ); an 8-way 0.1 in screw terminal block soldered onto the pin row (Mike, 2026-10-14; one of the Hxchen 20-pack on hand, the same blocks as the G2 driver's logic row), the 6504FE's wires under its screws on ferrules, one wire per screw; the two jumpers (SI to VDD, CS to GND) are short wires soldered on the back of the board between those pads, not under the screws; a small dab of West epoxy or hot glue (high-temp sticks; the Compute Bay Plan already allows a hot glue dab on the GPS plug) over them is fine (Mike asked, 2026-10-14 and 10-15), flux cleaned off first, kept on the pin-row half of the board, off the antenna coil (MA5532) and the small parts beside it, so the factory tuning holds; then the conformal coat every board gets (Compute Bay Plan sec 3.2); screws checked tight once more a week or two after wiring; mounted on nylon 4-40 screws, no metal. (b) At WD-B: 5 V from WD-B's D36V6F5 rail, ground to the island's floating negative, SDA to A4, SCL to A5, IRQ to D2 (no WD-B pin map exists yet; D2 is Claude's pick). The lead (Mike asked, 2026-10-14): 5 cores are all it needs (5 V, ground, SDA, SCL, IRQ), not the 7 the docs say; the cable is Belden 6504FE (Mike's roll, Mouser 566-6504FE-500-877, 22 AWG, 6 cores, foil shield with a drain wire, plenum jacket; Mike, 2026-10-14, in place of a piece of the 22/12), under 3 ft, the one unused core cut back flush inside the jacket at both ends. About 0.175 in across (Allied Wire & Cable listing for the 6504FE), so a PG7 gland (grips about 1/8 to 1/4 in, 3 to 6.5 mm) fits at both ends. Bare copper by the same listing, as on the 22/12 already used inside the boat; fine for a short dry run between two boxes. Drain wire landed at the WD-B end only, to the island's floating negative; cut back and taped at the detector end. I2C address 0x03. (c) The TVS on the detector lead (Compute Bay Plan 16) is now on a 5 V line: pick a 5 V part. (d) Write the 88 pF tuning value into System Settings Sheet sec 10, and mark the board here on the order tracker. (e) The SparkFun board and the level shifter 10-pack go unused (the shifters stay as stock). *Documents: Compute Bay Plan 4.2, 13.1, 13.3, 16; Electronics plan A2 entry 6, A12, sec 15.8 parts list; Box Wiring Sheet A12; Glossary (Level shifter); BOM 15.8; System Settings Sheet 10; order tracker.*
418. VESC kill line on the Flipsky servo plug's PPM pin (Mike, 2026-10-14). No doc names the pin where the opto-isolated kill line from WD-A (and later Pi A) lands on the controller. ADC1 and ADC2 carry the motor's sin and cos, so Claude's pick: the servo plug's PPM pin, with the VESC Tool kill switch set to that pin, active low (inferred from the VESC 6.x firmware's kill switch choices; set with the rest of System Settings Sheet sec 2). The kill line shares the UART ground core in the Micro-C run from Pi A. DECIDED (Mike, 2026-10-14, "yes"): the PPM pin. To write in: Build Guide 5.1 (layer 3), Electronics plan A7 plugs, System Settings Sheet sec 2 (kill switch on PPM, active low), Box Wiring Sheet A7. Drawing: VESC_A_Box_Wiring_2026-10-14.
417. VESC A capacitors: two WAGO 221 lever nuts, one for plus and one for minus, no copper strips, no G10 board (Mike, 2026-10-14: "not ok with copper strips", "ordered wago", then "wago has 5 holes, one for wire and 4 for caps"). Replaces the capacitor board in Reference/Controller_Capacitor_Bank_2026-09-26 (a G10 offcut with two copper strips and the capacitors soldered to them) and the Electronics plan A7 wording "on a small G10 board bolted to two insulated studs". The same four Nichicon 1,000 uF 63 V (on hand) lie in a row on the box floor, stripes all one way. Each leg, left full length, gets a heat-shrink butt splice (22-18 AWG, crimped on the bare leg, then shrunk) and an 18 AWG tail, 3 to 4 in, all eight the same length: red on the plus legs, black on the minus (striped) legs. The plus WAGO 221-415 takes the four red tails and a 12 AWG red lead; the minus WAGO the four black tails and a 12 AWG black lead. A WAGO joins everything in it, so plus and minus never share one. Both WAGOs lie flat, levers up. The tails are there because the legs are short and stiff: four cannot reach one WAGO without bending bare plus and minus legs close together (Claude, inferred from the 5/8 in cans and the usual 0.6 to 0.8 in legs). Each lead, under 6 in with a crimped 3/8 in ring lug, goes to its MGI post; each post then holds three lugs, bottom up: 2 AWG, 8 AWG, 12 AWG. Capacitors and WAGOs glued to the box floor with neutral-cure silicone; nothing bare but the post lugs; no solder anywhere. The controller's leads stay the Flipsky's own 8 AWG, cut under 6 in (Mike asked about 6 AWG, 2026-10-14; at 180 A for seconds a 4 in run drops about 0.04 V, and changing them means soldering on the board). WAGO 221-415s ordered by Mike 2026-10-14 (listing and price not given). Drawings: Cap_Board_Drawing_2026-10-14.pdf and VESC_A_Box_Wiring_2026-10-14.pdf (top of the folder, HTML masters beside them). Not fitted, do not re-propose: copper strips or any bare bus copper at the capacitors; a bought RC capacitor pack (the Castle CapPacks on Amazon have solder posts; Jeti, APD and Scorpion boards have solder pads); stacked WAGOs (the lower one's levers get trapped). To write in: Electronics plan A7, Build Guide 5.1, BOM 15.5 (add 2 WAGOs, 8 butt splices, 18 AWG and 12 AWG red and black, two ring lugs; drop the G10 offcut and copper strips), Box Wiring Sheet A7 notes; archive the 2026-09-26 drawing.
429. Starlink on for red alerts and hands-on moments (Mike, 2026-10-15, "add 2 but not shipping lanes"). Besides land, AIS traffic and the camera (items 414, 415 and 416), Starlink comes on at once and goes back to its normal schedule by itself when the situation is over, not when shore clears it (Mike, 2026-10-15), after a 10 minute hold so it does not flick on and off (Claude's pick, standing unless Mike says otherwise). The same goes for items 415 and 416: off again 10 minutes after the boat leaves the 20 nm zone, the AIS vessel is past, or the camera no longer sees the boat. When each case is over: a red alert, when the fault reading is back to normal; a drogue release, 30 minutes after the release; a tow, when shore marks it done, since the boat cannot sense a towline (inferred), with a 4 hour limit in case shore forgets; the Canaries arrival, when shore marks the tow has the boat. (a) on any red alert: a steering fault, a bilge pump running, a battery temperature alarm or trip, both GPS lost, a computer failover; (b) during hands-on moments: a tow hooking up or letting go, a drogue release, and the Canaries arrival from 48 hours before Spain's 12-mile line until the tow has the boat. Not added (Mike): on near charted shipping lanes with no ship showing; AIS traffic and the camera cover those. Not added (Claude): low battery (Starlink is the first load shed, Build Guide sec 4.6) and storms (the lightning rule already runs it every 15 minutes). *Documents: Build Guide sec 8; System Settings Sheet.*
416. Starlink on when a camera spots a boat (Mike, 2026-10-14 and 15). Mike asked for Starlink on for any boat or object without AIS within about 2 miles; this is a trigger for low-delay control, not a promise to anyone. Rule: Starlink on whenever the AI (the Hailo-8L chip on Pi A) spots a boat on the foredeck fisheye, which Mike says will be on (2026-10-15), at whatever range it manages; the forward post camera can feed the same check when it runs. What the fisheye can do: the Hanwha XNF-8010RV is 2048 x 2048 over a 360 degree circle (Hanwha's product page), so at the horizon about 18 pixels per degree (computed), about half the forward post camera's 32 (the Reolink RLC-520A, 2560 x 1920 over 80 degrees, Reolink's page). A 25 ft boat is about 2 pixels wide at 2 nm and about 8 at half a mile (computed), so small boats are likely spotted at about a quarter mile in daylight (a guess); ships, about 50 pixels at 2 nm, are easy, but carry AIS. At night only lights show. Its view: it is aimed up and forward at the wing and rig (Electronics and Wiring Plan A6a), sits low on the foredeck, and the deck is flush (no cabin, Build Guide sec 3.1), so what blocks it is the wing and mast standing aft of it, and its upward aim, which puts the horizon astern at or past the edge of its picture; so it watches ahead and to the sides, less well astern (inferred). Cost: with the fisheye on anyway, the extra is the AI, about 1 to 2 W (inferred). Not yet in the power table: the fisheye running all the time; the table carries 20 Wh a day for all cameras, intermittent (Build Guide sec 4.6), and the camera's own draw is not found (it is a PoE Class 3 device, so 13 W at most; a guess of 5 to 8 W in use gives about 120 to 190 Wh a day). *Documents: Build Guide sec 4.6 (camera line) and sec 8; Electronics and Wiring Plan A6a; System Settings Sheet.*
415. When Starlink is on, beyond the harbor (Mike, 2026-10-14). Starlink stays on (a) within 20 nm of land, and (b) whenever the AIS receiver shows a vessel whose closest pass is forecast inside about 3 nm within the next 15 minutes. Any vessel sending AIS counts, big ship (Class A) or small boat (Class B); AIS beacons on buoys and shore stations do not (Claude's pick, standing unless Mike says otherwise; near port they would keep it on anyway, and the 20 nm rule covers port). An AIS rescue beacon (SART or man-overboard) turns it on and alerts shore. Why wider than the 15 nm first suggested: 15 nm covers Brunswick to Buoy STS (about 9 nm off the beaches, computed) and the Las Palmas tow meeting (10 nm, or 13 to 15 nm, right at the edge), but not the Canaries' one-way shipping lanes between the islands, which the boat crosses after the tow out east of Gran Canaria up to about 20 to 25 nm from either island (inferred), nor the coastal and ocean ship traffic. 15 minutes at a closing speed of about 23 kt (a 20 kt ship and the boat) is about 6 nm of warning; enough if Starlink connects in a few minutes (a guess). Power: small, since ships are seldom in sight offshore. A camera trigger for boats without AIS is a separate question (item 416). *Documents: Build Guide sec 8 (Starlink paragraph); System Settings Sheet (Starlink cadence); Routing Guide comms paragraph.*
414. Starlink stays on through the harbor (Mike, 2026-10-14, before the Coast Guard meeting at Savannah). From the dock at Brunswick Landing Marina to Buoy STS, and on the tow or run back in at the end, the Starlink Mini stays on the whole time instead of its hourly 5-minute window. Why: Iridium is the only always-on command path, and a command takes about 20 to 40 seconds by it (up to about a minute if a wave or the wing blocks the sky and it retries); over Starlink a command arrives in well under a second, but only if Starlink is already on, since it takes minutes to connect from off (a guess; no start-up time found for the Mini). The cellular Primelink units only report and listen, so they carry no commands. Cost: the Mini draws about 25 to 40 W, so a 6 to 9 hour channel run is about 300 Wh, about 1% of the 32 kWh bank. The "Starlink cadence override" on the command list (Build Guide sec 8) already sets it. To confirm in the software: the boat accepts signed commands over Starlink as well as Iridium (the Build Guide implies it, never says it). What Mike tells the Coast Guard: "usually under 30 seconds by satellite, always within about a minute; in the harbor, Starlink stays on and commands arrive in seconds." Also from the meeting prep: the command list has no steering or route command, so the shore can stop the boat (the rudder then holds) but cannot drive it; say "stop," not "take remote control." *Documents: Build Guide sec 8 (Starlink paragraph: add the harbor rule); Routing Guide comms paragraph; System Settings Sheet (Starlink cadence); Reference/Watch_Procedure_PRIME_RANGER_2026-10-08 (commands line).*
411. Battery box signal glands: one gland per cable, and the probe chain skips the BMS box (Mike, 2026-10-13, "yes, plan B"). Replaces item 274 (2026-09-25) and the one-sleeve bundle in the Battery Box guide Part 8. Mike's point: the balancer's 10-core is 10 mm (0.39 in) across and the SBMS0's 12-core 6 mm (0.24 in), and two cables of different size do not seal in one gland, sleeved or not. Written in: (a) The battery box aft wall gets three glands in place of the one bundle gland, all from the glands on hand: PG13.5 (13/16 in hole) for the ENERJOUR 18 AWG 10-core, PG9 (5/8 in hole) for the 22 AWG 12-core shielded, PG9 (5/8 in hole) for the DS18B20 chain's 20/3 cable (see (f)). Each cable with its own drip loop; dielectric grease on each insert as before. The 12-core's shield is still grounded at the BMS box end only. (b) The DS18B20 chain runs straight from the battery box to central node entry 18 and never enters the BMS box; the only reason it passed through was the one-hole bundle. (c) The BMS box keeps four glands plus the vent, all facing down: the 10-core in on PG13.5, the 12-core in on PG9, the two shunts' sense pairs on PG9, the USB lead out on PG7. Its entry 6 (chain out) is deleted and not drilled. (d) The 2-hole M25 gland on hand (Oct 5) does not suit this: its two holes are the same size. (e) The six probe leads join the chain's cable with adhesive heat-shrink crimp butt splices, no solder (Mike, 2026-10-13: no soldering on the boat) and no WAGO 221 (Mike, 2026-10-15: they are for large wire; the probe leads are thinner than the 24 AWG they take at the low end). Daisy chain, not a star: the cable runs along the stacks and each probe is spliced in where it sits, with the probe's own lead cut short. At each probe, each of the three wires is one splice: the cable coming in on one end, the cable going on plus the probe wire on the other. 6 probes x 3 wires = 18 splices; the last probe's three are plain end-to-end joins. Daisy chain is also the better layout for 1-wire signals. The smallest common heat-shrink butt splice (red, 22 to 18 AWG) needs each thin probe wire folded double before crimping; stagger the three splices at each probe so the cable stays slim, and tie the cable down clear of the terminal rows and the series links. Which butt splices are on hand (the Sep 16 order) is not checked. Added 2026-10-15 (Mike: "run 3 wires a length and tie the probes along them"): the probes' own leads, about 3 ft each, stay full length (Mike, 2026-10-15): six 3 ft branches plus the run to the node is well inside what one chain of a few probes handles on a 4.7 k resistor (Claude, inferred from common 1-wire practice, not from a datasheet figure); the full leads let the splices sit together on the cable away from the cell terminals; coil and tie the spare lead neatly. Added 2026-10-15 (Mike asked the trunk size and how to tie in): (f) The trunk is Mike's 3-conductor 20 AWG cable on hand (Mike, 2026-10-15; not on the tracker), all three wires used: red power, black ground, the third data. No shield is needed for a 1-wire chain this short. Its gland is PG9 (5/8 in hole) in the battery box wall; PG9 takes about 0.16 to 0.31 in (4 to 8 mm), which covers a 20/3 jacket. (g) Tie-in: the trunk runs along the stacks and passes each probe, the probes about 2 ft apart (Mike). At each probe, strip the trunk's jacket back about 3 in and cut its three wires; each wire gets one red heat-shrink butt splice (22 to 18 AWG): the trunk wire coming in on one end, the trunk wire going on plus that probe's matching wire (folded double) on the other end. A 20 AWG wire plus a folded thin probe wire fits the red splice's range. Stagger the three splices about 1 in apart, shrink them, then slide a glue-lined heat-shrink sleeve over the whole stripped section and the jacket ends. At the last probe the trunk ends: the trunk wire on one end, the probe wire on the other. The probes' 3 ft leads stay full length, coiled and tied; tie the trunk and each joint down clear of the terminal rows and the series links. (h) Central node entry 18 has both 1-wire chains in one PG9 gland (Electronics plan A3 entry 18): the same two-cables-in-one-gland problem. Fix (Claude, following Mike's 2026-10-13 rule): chain 1 alone in entry 18 (PG9), chain 2 in spare entry 17, drilled for it (PG7 or PG9 to suit chain 2's cable; Mike, 2026-09-26: drill entry 17 only when a cable needs it). *Documents: Battery Box Building Guide Part 8 (the bundle paragraph, the wall paragraph, the Materials rows for glands and the watchman box); Electronics and Wiring Plan sec 6.7, A4 entries 1 and 6, A9 entry 3, A3 entry 18, the gland totals; Reference/BMS_Box_Wiring_Diagram_v10_2 (html, pdf, png) and Box Wiring Sheets where the BMS box appears; Design Decisions if the bundle is named there.*
409. SBMS0 first-time settings, not in the Settings Sheet (Claude, 2026-10-13, from Mike's question "how should we set dip switches"). The one DIP switch is the six-position switch on the WiFi/USB add-on board, labeled MTX, STX, SRX, WRX, WTX and IO0, beside the ribbon header; the board's own printed table (Reference/SBMS0_v03d_Manual.pdf page 15, the photos) gives two settings: Normal = MTX, STX and SRX ON, WRX, WTX and IO0 OFF; Program ESP32 = the reverse, only for loading new firmware into the WiFi module. The boat runs on Normal, which is how the board ships (the manual's photo shows it). Claude first told Mike the board had no DIP switch (2026-10-13, wrong; Mike's photo found it). The main board has no switch or jumper: every other setting is made on its touch screen (manual pages 5, 6 and 29). What the Settings Sheet's SBMS0 table lacks is the Parameter Settings row: battery type 1 LiFePO4, number of cells 8, battery capacity 1256 Ah (the 8S4P pack's rated figure, 4 x 314 Ah; the typical 341 Ah would give 1364 Ah, but the rated figure makes the percentage run out a little early, the safe side for load shedding), then Store Parameters, then unplug and replug the 12-pin cell plug so the stored values take (manual page 6). Whether the capacity field takes a number over 999 Ah is not known; if it stops short, enter the top it allows and log it here. The rest is already in the Settings Sheet sec 1: shunts 0.2500 and 0.7500 milliohm, the EXTIO types, parameters 29 and 30, UART on with the open serial speed (item 401 c), Save device settings. *Documents: System Settings Sheet sec 1 (SBMS0 watchman settings table: add the Parameter Settings row and a DIP switch row); Battery Box Guide Part 10.*
434. Membrane vents: only the central node gets one (Mike, 2026-10-16: "the boxes are all inside a dry cabin", "less work", "maybe just central"). Every other box drops its vent and its 1/2 in vent hole: Pi A, Pi B, the network switch box, the aft junction, the BMS box (if it was to get one), Release Node B, the transom terminal box, both Iridium boxes, the detector box, the USB hub box, the light driver's dry box, the wing box and the Starlink enclosure. The central node keeps its vent because its BME280 barometer reads outside pressure through it. The battery box's relief stub into the cabin is not a membrane vent and stays. Reason: in a dry cabin the glands never get wet, so a vent adds little; a sealed box also keeps a desiccant pack working longer (Claude's reasoning, 2026-10-16). Desiccant packs stay where the guides already have them. Vent count to buy: one (nine were counted). *Documents: Electronics plan sec 3 (the "every plastic box carries a membrane vent" rule), sec 7, sec 8 buy line and every Appendix A hole schedule's V row and the totals; Compute Bay 3; BOM 15.8 and the compute box line; Box Wiring Sheets (every V entry); Reference/Box_Diagrams_v10_1 and Parts_By_Box_v10_1.*
430. No ferrite on the motor phase leads (Mike, 2026-10-15: "the motor wires are too short for one"). Changes item 421: the 1.0 in clamp over the three phase leads at VESC A is dropped. The leads stay tightly bundled and short, with no signal cable alongside, which is what keeps motor noise down. Ferrites boat wide now about 11. The Fair-Rite 1.0 in clamp bought 2026-09-13 has no job and is kept: Amazon's 30-day return window closed 2026-10-13 (Mike, 2026-10-15). Written in: Electronics and Wiring Plan 7.5 and the motor phase lead rule, the parts list line and App A totals; Compute Bay 9.4; BOM 15.8; VESC A Box Wiring sheet; order tracker Parts row. *Documents: as listed.*
421. Ferrites only on cables that leave the cabin (Mike, 2026-10-14, "yes"). Was one ferrite on every cable at every box wall, about 60. A clip-on core does nothing against a direct strike, which goes down the lightning copper; it helps only against the surge a long outside wire picks up from a nearby strike. Kept, about 12: GPS A and B coax, Iridium A and B antenna coax, the post wind meter, post lights and horn cables, the bow lantern feed, each slip ring hull cable at the wing box, and the 1.0 in clamp on the motor phase leads (for motor noise). Dropped: every cable that stays in the cabin, so the BMS box and battery box take none (this settles the BMS box tap cables' core size), and the cables that already have their own surge protector (cameras, Starlink Ethernet, solar). At least 20 cores on hand (order log), none to buy; the shelf count in item 126 no longer needs ferrites. Written in: Electronics and Wiring Plan 7.5, App A entry note and totals; Compute Bay 9.4; BOM 15.8; Box Wiring Sheets (general note, A1, A12, A15, A16); VESC A Box Wiring sheet; order tracker Parts row. *Documents: as listed.*