HULL/20 Spaceways

How I built it. Engineer's edition.

The saucer with the panels off

The same flight record as the Passenger edition, for makers: the cut list, the four data lines, the controller, the show's state machine, the flash math and the update that wouldn't land. Built for fun, by me and Linux, my tiny flight engineer. If you build one, send pictures.

Prefer the story? Passenger edition

Drawing package (frame, electrical, circuit boards): /build/drawings

The saucer standing on its edge in the garage at night with all eight spokes lit cyan around a white light in the middle.
Photo First light, in the garage.
Revisions
ZoneRevDescriptionDateApproved
ALLAFIRST ISSUE: ENGINEER'S EDITION, SPLIT FROM THE BUILD STORY2026-10-30R. DIXON
SH 14BSHEET XIII ADDED: THE SPACEPORT GETS A VOICE2026-10-30R. DIXON
ALLCNEW HULL/20 EMBLEM2026-10-30R. DIXON
—DFOOTER: NO COOKIES, NO TRACKERS, NO FORMS2026-10-30R. DIXON
ALLEREDRAWN AS A DRAWING SET. FRAME PLAN, SECTION AND ABDUCTION TIMING DRAWN FROM THE NUMBERS. DRAWING HULL-20-UFO-027 ADDED2026-11-01R. DIXON

Notes

  1. Millimetres, with inches in brackets, unless a sheet says otherwise. Drawing HULL-20-UFO-027 is inch-first, like its cut files.
  2. Letters A to X are the frame pieces in the cut list, sheet 5. Find numbers 1 to 29 are the 2027 parts list, drawing HULL-20-UFO-027 sheet 1.
  3. Figures 1, 2 and 6 and every view on HULL-20-UFO-027 are drawn by a script from the numbers: the frame drawing, the cut files and the firmware's phase table. Dimensions govern; don't scale from the screen.
  4. Cut members are hatched, hidden edges dashed, removed parts in phantom (long, short, short), centre lines long and short.
Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-11-01
Checked
R. DIXON
2026-11-01
Approved
R. DIXON
2026-11-01
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
MM [IN]
Projection
THIRD ANGLE
PROJECTION
Title
SAUCER, ACCESS PANELS OFF
Size
B
Scale
NTS
Sheet
1 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

IThe fleet

3,418 LEDs, five craft

The saucer. An octagon of 2020 aluminum extrusion, 87⅛ inches across the flats, a clear 24-inch dome, an alien pilot who has never asked for a raise, and 1,406 LEDs on four data lines from an ESP32-S3 board.

The banner. 1,500 WS2812B LEDs, 300 x 5, serpentine, on a thumbnail-size ESP32-C6 board. A quiet show all year so the neighbors don't get suspicious; a departures board in October.

The garage eyes. Two 16x16 WS2812B panels behind two garage door windows, on their own ESP32-C6 board. They blink. They look around. They know.

Gate 20. A single-board computer and a 15.6-inch USB monitor, held up by an inflatable alien.

Mission control. A web page and a phone app I run the yard from: pin a scene, set the brightness, change the floodlights, feel powerful.

Banner 1,500; saucer 440 + 115 + 480 + 241 + 128 + 2 = 1,406; eyes 2 x 256 = 512.

Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-30
Checked
R. DIXON
2026-10-30
Approved
R. DIXON
2026-10-30
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
—
Projection
THIRD ANGLE
PROJECTION
Title
FLEET, GENERAL ARRANGEMENT
Size
B
Scale
NTS
Sheet
2 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

IIBefore the ladder

A simulator first

Before touching the banner again I built a desktop simulator that compiles the real firmware engine against a stub of FastLED and draws the strip in a browser. It models the wire time too. A WS2812B bit is 1.25 µs, so 24 bits is 30 µs per LED, and 1,500 LEDs on one data line is 45 ms plus the latch: about 21 frames a second, flat out. Better to learn that from a chair than from a ladder.

The banner's engine runs in three places: on the board, in the simulator and in the phone app's preview. The phone preview caught a real bug before it reached the wall: after waking up at night, the autopilot stayed asleep. Relatable.

The saucer's show went the other way round. I wrote it first as pure functions in JavaScript, in a 3D browser sim of the saucer, then ported it line for line to C++ for the Nano ESP32 and checked the two against each other pixel by pixel. The sim is where the look got decided. The port is where it got real.

Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-30
Checked
R. DIXON
2026-10-30
Approved
R. DIXON
2026-10-30
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
—
Projection
THIRD ANGLE
PROJECTION
Title
SIMULATOR, FIRMWARE ENGINE
Size
B
Scale
NTS
Sheet
3 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

III12 October

The frame

2020 T-slot: eight rim bars (A) in an octagon, eight spokes (B) to a hub ring (C) that stands 6⅜ inches up on eight posts (E), and a small core ring (D) at the center. 87⅛ inches (2,213 mm) across the flats, 94⅜ across the points.

There are no sloped joints. Everything meets square or at 135 degrees, with corner brackets at the rim, and the slope of the top comes from the skin pulled tight from the rim to the hub. Flat rings and straight spokes are stiffer than hinges, there's nothing to set by eye, and it's cheaper. Under the deck, an IP67 box sits on two rails across the core ring; the dome sits on the deck plate. Four of the spokes carry pad eyes top and bottom as lift points, and holes for the legs it stands on when it's landed.

I drew all of it before I cut anything. This is the sensible part of the story.

Fig. 1, frame planThe saucer frame from above with the skin and deck off: an octagon rim 2213 mm (87 1/8 in) across the flats, eight spokes to a small core ring, a hub ring on posts, porthole posts on every spoke and pad eyes on four. A A 2213 [87⅛] ACROSS FLATS C B A H D K DOME Ø24 IN, ON THE DECK PAD EYES TOP AND UNDER,R1000, 4 SPOKES (J12) CENTRE PLATE 18 IN ANDØ9½ IN BEAM PORT, BELOW K POST, Ø4 IN PORTHOLE,R744, EVERY SPOKE PLAN, SKIN AND DECK OFF. 94⅜ IN ACROSS POINTS. LETTERS: CUT LIST, SHEET 5
Fig. 1 The frame from above, skin off. 87⅛ inches across the flats. Original drawing, full size. E
Fig. 2, section A-AA cut through two opposite spokes: the spokes flush with the rim top, the hub ring 162 mm (6 3/8 in) up on posts, the deck and dome on the hub ring, the centre plate on four legs under the core ring, and the 2026 skin in phantom. 162 [6⅜ IN] 142 E 60 90 F 150 L Ø24 IN DOME, 12 IN HIGH ALUMINET SKIN, 2026 BELLY SAIL, 2026 PORTHOLE PLATE, 12.6° PAD EYE Ø9½ IN BEAM PORT H RAILS ON THE CORE RING SECTION A-A. CUT MEMBERS HATCHED. 2026 SKIN IN PHANTOM. MM UNLESS NOTED
Fig. 2 A cut through two spokes. The dome sits on short posts over the hub. Original drawing, full size. E
Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-12
Checked
R. DIXON
2026-10-12
Approved
R. DIXON
2026-10-12
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
MM [IN]
Projection
THIRD ANGLE
PROJECTION
Title
FRAME, OCTAGON, 2020 T-SLOT
Size
B
Scale
NTS
Sheet
4 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

IVMid October

Twenty bars and a stop block

Twenty bars of 1,220 mm extrusion, a 3 mm kerf, square cuts only. At the saw, one stop block setting per letter: cut every piece of that letter, then move the block, and tape a label on everything as it comes off. Matching lengths within a letter matter more than the exact number.

Then 88 ends tapped M6, 25 mm deep (every B, E, F, G, K, L and X, both ends), 6.5 mm clearance holes on the center line with a drill guide, and 5.5 mm bracket holes marked through the brackets. And a great deal of deburring. Aluminum chips get everywhere. Everywhere.

Drawing: the cut sheet. Each bar is a row showing the pieces cut from it, A and B at 900 mm, the offcuts used for the hub and core rings.
Fig. 3 The cut sheet I took to the saw.
Cut list: 20 bars of 1,220 mm
ItemPieceQtyLengthCut from
Arim8900 mm (35⅜")bars 1 to 8
Bspoke8900 mm (35⅜")bars 9 to 16
Chub ring8300 mm (11¾")offcuts of A
Dcore ring8137.8 mm (5⅜")offcuts of B
Ehub post8142 mm (5⅝")offcuts of B
Fplate leg490 mm (3½")bar 18
Griser post895 mm (3¾")bars 17, 18
Hbox rail2373 mm (14⅝")bar 19, cut last
Kporthole post860 mm (2⅜")bar 18
Lleg post4150 mm (5⅞")bars 17, 20
Xleg extension4400 mm (15¾")bars 17, 19, 20
The bare frame on the driveway: black aluminum spokes meeting the hub, with the clear dome sitting on it.
Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10 MID
Checked
R. DIXON
2026-10 MID
Approved
R. DIXON
2026-10 MID
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
MM [IN]
Projection
THIRD ANGLE
PROJECTION
Title
CUT LIST, 20 BARS OF 1220
Size
B
Scale
NTS
Sheet
5 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

V14 October

Last year's lights, four lines

No new LEDs. I had four 5 m strips of 60/m RGB and a warm white RGBW one from last year, so each got a bench test and every cut was planned around what was on the reel. One had a bad seal and got resealed with heat shrink. Recycling, spaceport edition.

The plan called for 56 LEDs per spoke. A meter is 60, so the spokes got 60 and the firmware was changed to match the ship rather than the drawing. The rim is one strip running all the way around through eight aluminum channels with milky diffusers. The spokes are cut pieces joined with short wires, so one data line snakes out along a spoke, across to the next, and back in, then carries on into the big light in the belly.

Drawing: five LED strips as bars with numbered cuts, showing which part of the saucer each piece goes to and what's left on the reel.
Fig. 4 Five strips I already had, cut by cut. The hatching stays on the reel.
Drawing: the belly from below, eight spokes with arrows showing the data running out and back in, ending at the light in the middle.
Fig. 5 Line 3 snakes through all eight spokes and ends at the belly light.
Four data lines
LineRunsLEDsWire time
1Rim, one strip through all eight channels440 RGB13.2 ms
2Dome lip, warm white115 RGBW4.6 ms (32 bits each)
3Eight spokes (8 x 60), then the belly disc (241, nine rings)721 RGB21.6 ms
4Eight portholes (8 x 16), then the pilot's two eyes130 RGB3.9 ms

Line 3 is the long one. Putting the belly light at the end of the spokes saved a cable to the center and cost nothing that shows.

Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-14
Checked
BENCH TEST
2026-10-14
Approved
R. DIXON
2026-10-14
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
LEDS
Projection
THIRD ANGLE
PROJECTION
Title
LED LINES, FOUR DATA RUNS
Size
B
Scale
NTS
Sheet
6 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

VI25 October

Final flight check

The frame came together on the driveway. Linux, my tiny flight engineer, did the final flight check on the dome. The skin went on the same week: Aluminet, an aluminized shade cloth, top and belly, pulled tight from the rim to the hub.

Linux reaching for the clear dome on the bare saucer frame, me standing beside him.
The saucer's underside with the Aluminet skin stretched over the frame, standing on its edge in the garage.
Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-25
Checked
LINUX
2026-10-25
Approved
R. DIXON
2026-10-25
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
—
Projection
THIRD ANGLE
PROJECTION
Title
FRAME, FINAL FLIGHT CHECK
Size
B
Scale
NTS
Sheet
7 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

VII25 October

The controller

The fog machine and the beam pods (with their 12 V supply and MOSFET) came off the manifest the same day, and so did the screw terminals. Every part you take out is a part that can't fail at 8 pm with a line of kids watching.

It's two stacked protoboards. Board A holds the Nano ESP32 on headers and a 74AHCT125 in a socket (so the chip is never soldered), with a 0.1 µF ceramic beside it. Board B is where the outside world lands: four 330 Ω series resistors, one per data line, a 100 µF can across the 5 V input, and solder points for four two-wire cables (DATA plus a GND reference) and the 5 V feed from the power box's fuse. Six short wires join the two boards. No screw terminals: every wire soldered straight into its hole, one wire per step, 35 steps.

Why the level shifter: the ESP32 drives 3.3 V logic, and a WS2812B powered at 5 V wants 0.7 x VDD, 3.5 V, for a clean high. Sometimes it reads 3.3 V anyway. Sometimes is not a word you want in a flight record. The 74AHCT125 has TTL input thresholds, so 3.3 V is a solid high, and its outputs swing the full 5 V. The 330 Ω resistors take the ringing off the edges on the way down the cables.

Pixel power doesn't go through the controller at all. The power box feeds each run through its own inline fuse, with a 100 µF capacitor at each strip start. The controller only sends data, plus a ground for the data to be measured against.

The bench test runs test firmware that lights all four lines at once, each in its own color (rim green, dome blue, spokes purple, portholes cyan), with a short test strip moved from cable to cable. Three of the four lit the first time. The rim stayed dark. I went down the chain with a meter, found it at the first resistor, and then everything lit.

The protoboard before the controller, with a reel of last year's LED strip behind it.

I also drew the controller as a proper circuit board, HULL/20 on the silkscreen and all. That drawing is from 17 October, before the screw terminals came off, and it stayed a drawing. The one flying is the hand-wired stack.

A render of the UFO controller drawn as a circuit board, with headers for the Nano, two chip sockets, resistors and green screw terminals. A design, not the board in the saucer.
The controller as a circuit board: a design, not the one in the saucer.
Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-25
Checked
BENCH TEST
2026-10-25
Approved
R. DIXON
2026-10-25
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
—
Projection
THIRD ANGLE
PROJECTION
Title
CONTROLLER, TWO-BOARD STACK
Size
B
Scale
NTS
Sheet
8 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

VIII26 October

First light, and the show

The next night it lit up in the garage: 1,406 LEDs. I counted. So did the firmware. We agree.

The show is a small state machine, the same one in the sim and on the board:

The show
ModeHow longWhat it does
Idle150 to 240 s, randomThe rim turns like a turbine. Rare events: comms (4.6 s), search (6 s), sputter (6.6 s)
Approach3.5 sThe portholes light in sequence from one side, like the crew rushing to the window
Abduction11.2 sCharge 2.0 s, beam 4.5 s, lift 8.0 s, flash 9.6 s, black 9.75 s, release 10.05 s, end 11.2 s
Cooldown3.0 sBack to idle
The abduction, timedA time axis of the show: approach 3.5 s, abduction 11.2 s, cooldown 3.0 s, with the firmware's phase constants marked from the start of the abduction: charge 2.0, beam 4.5, lift 8.0, flash 9.6, black 9.75, release 10.05, end 11.2. APPROACH 3.5 ABDUCTION 11.2 COOLDOWN 3.0 S 0 2.0 CHARGE 4.5 BEAM 8.0 LIFT 9.6 FLASH 9.75 BLACK 10.05 RELEASE 11.2 END PASSENGER COUNT +1 ORDINATES: SECONDS FROM THE START OF THE ABDUCTION (PH_CHARGE ... PH_END)
Fig. 6 The abduction against the clock: the three modes, and the phase table from the firmware, in seconds. E

The abduction, in order: the portholes lock on, cyan fills the spokes toward the white center, violet at full speed, a lift of soft white on the rim, a flash, a moment of black, and a ring of light runs back out. Abductions are scheduled, not strobed. The saucer keeps a passenger count, incremented at the flash. Nobody has been taken. The count is a formality.

The phase table and the rings of the belly light, as they are in the firmware:

static const float PH_CHARGE = 2.0f, PH_BEAM = 4.5f, PH_LIFT = 8.0f,
  PH_FLASH = 9.6f, PH_BLACK = 9.75f, PH_RELEASE = 10.05f, PH_END = 11.2f;
static const uint8_t IRIS_RINGS[9] =
  { 1, 8, 12, 16, 24, 32, 40, 48, 60 };   // 241 LEDs

A few details I'm fond of. The turbine has inertia: a first-order lag that's quick to spin up (0.9 s) and slow to spin down (2.2 s), capped at 2 turns a second. Idle events come 20 to 60 seconds apart and only start once the turbine has calmed down, because comets plus an event went over the flash limit. The big light in the belly is drawn as an eye: a pupil, a gaze that drifts on slow noise, and a blink every 5 to 14 seconds in which the rings close from the outside in. On lock-on the gaze stops drifting and snaps. There's an input for a camera to say where the person is; there's no camera on the saucer, so it snaps at whoever it imagines. And the show clock wraps around in a quiet moment, so float precision doesn't wander over a long night.

Looking up under the saucer: the spokes lit cyan and violet against the silver skin, the light glowing in the center.
Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-26
Checked
FIRMWARE
2026-10-26
Approved
R. DIXON
2026-10-26
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
S
Projection
THIRD ANGLE
PROJECTION
Title
SHOW, STATE MACHINE
Size
B
Scale
NTS
Sheet
9 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

IXPower

The supply and I have an understanding

Full white on 1,406 LEDs is roughly 87 A on paper at 20 mA a channel (1,291 RGB at 60 mA, 115 RGBW at 80 mA). The supply is a 350 W enclosed supply, 5 V and 60 A, and I'd rather not ask it for everything, so every frame is summed before it goes out and scaled down if it would go over 30 A. From the simulator; the firmware does the same thing, with less commentary:

// ~20 mA per channel at 255
export function frameAmps(buf) { let s = 0;
  for (let k = 0; k < buf.length; k++) s += buf[k];
  return (s / 255) * 0.020; }
export function limitFrame(saucerBuf, cap, limitA) {
  const raw = frameAmps(saucerBuf) * cap;
  const k = raw > limitA ? (cap * limitA) / raw : cap;
  return { scale: k, rawAmps: raw, amps: Math.min(raw, limitA) };
}
Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-30
Checked
R. DIXON
2026-10-30
Approved
R. DIXON
2026-10-30
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
A
Projection
THIRD ANGLE
PROJECTION
Title
POWER, 5 V 60 A SUPPLY
Size
B
Scale
NTS
Sheet
10 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

X27 October

No faster than three a second

The saucer faces the street, so it holds one hard line: no part of it flashes more than 3 times a second, the usual photosensitive safety line. My first version flickered 16 to 30 times a second in places, which is less "visitor from another world" and more "nightclub". Every flicker now steps on one shared clock at 4 changes a second, sparkle became fades, and the turbine tops out at 2 turns a second.

I wrote a flash counter that renders the show at 64 frames a second, measures every rim face, porthole and spoke, and counts flashes in every one-second window, across 300 random seeds and every command the show accepts. It found one I couldn't see: during the charge, a bright comet ran around the eight portholes. Each window flashed slowly, but the ring as a whole pulsed 12 times a second.

The fix is a lobe of ((1 + cos φ) / 2)⁴, the porthole's brightness as a function of its angle from the sweep. Expand it and it's a polynomial of degree 4 in cos φ, so it has no harmonic above the 4th. Sum it over eight evenly spaced windows and every harmonic from the 1st to the 7th cancels, which leaves only the constant: the ring's total brightness stays flat while the lobe goes round. It still looks like it's spinning. The math just stopped it from shouting. Now 0 of 300 seeds go over the line.

// one sweep at 1.5 a second. A cos^4 lobe has no harmonic above the 4th,
// so over 8 evenly spaced windows the ring's total stays constant.
float b = powf(0.5f + 0.5f * cosf(2 * (float)M_PI
  * (frac(st.t * 1.5f) - aw)), 4);
set(o, CYAN, 0.08f + 0.4f * sstep(PH_CHARGE, PH_BEAM, s));
add(o, lerpC(CYAN, HOT, sp), b);
Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-27
Checked
FLASH COUNTER
2026-10-27
Approved
R. DIXON
2026-10-27
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
HZ
Projection
THIRD ANGLE
PROJECTION
Title
FLASH LIMIT, 3 PER SECOND
Size
B
Scale
NTS
Sheet
11 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

XI29 October, morning

Eleven tries in the driveway

The saucer takes updates over Wi-Fi with a password. In the driveway, inside an aluminum frame, an aluminum box and shade cloth, 11 tries in a row stopped partway, each at a different point. It looks like the board restarts mid-transfer. The one update that ever landed out there had USB power plugged in as well. So updates go over the cable now, and the current firmware turns the Wi-Fi modem's power saving off. In space, no one can hear your Wi-Fi.

Every update lands in the other of two flash slots, so the slot doesn't tell you what's running. Each build now prints its own build id on the serial line at boot, so I always know which version of me wrote the bug.

Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-29
Checked
R. DIXON
2026-10-29
Approved
R. DIXON
2026-10-29
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
—
Projection
THIRD ANGLE
PROJECTION
Title
FIRMWARE UPDATES
Size
B
Scale
NTS
Sheet
12 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

XIIThe yard

How the yard talks

Every board runs its own show. If the Wi-Fi goes down, the saucer, the banner and the eyes carry on like professionals. A small server in the house sits on top: the control panel and the phone app talk to it, and it talks to the banner and the eyes through a message broker. It feeds the banner live data, like planes and ISS passes overhead, and it runs the four floodlights. A person detector on an outside camera tells the banner when someone walks by, and the banner answers with a soft glow that walks along with them. The saucer doesn't listen to the network at all yet. It runs its show and minds its own business, like a good saucer.

Gate 20 is a single-board computer running an airport departures loop. From 15 to 31 October, every evening from 6 to 11, it showed a code for that hour's planet, and scanning it put the stamp in a Planet Hopper passport on the visitor's own phone. Six stamps made a Grand Tour. The red flag on the 16th switched it to RED FLAG WARNING, FLIGHTS HELD until 8 pm on the 17th; passport control stayed open, because paperwork waits for no weather.

Diagram: how the yard talks. The phone app and the control page talk to a small server in the house, which talks to the banner, the eyes and the floodlights; each board runs its own show.
Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-30
Checked
R. DIXON
2026-10-30
Approved
R. DIXON
2026-10-30
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
—
Projection
THIRD ANGLE
PROJECTION
Title
YARD NETWORK
Size
B
Scale
NTS
Sheet
13 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

XIII30 October

The spaceport gets a voice

A spaceport without announcements is just a driveway with ambitions.

The theme. An original 27-second HULL/20 theme, synthesised from scratch in Python with numpy and scipy, no samples: a theremin lead, a walking upright bass, brushes, an FM electric piano and airline chimes, at 100 BPM and cut to the reel's beat grid. Three styles were on the table: lounge, abduction and surf-rocket. Lounge won. It's a spaceport, not a getaway.

The voices. Kokoro text-to-speech, run locally, one voice per role: gate agent, Captain Vrell, passport control, purser, adverts, paging, NAV-20 and the tower. NAV-20 gets ring modulation and a metallic comb filter; the tower is band-limited like a radio. Then every line goes through the same 1960s terminal PA: a narrow band, a horn honk and a slap-back echo off the far wall. Each role has its own chime: a ding-dong for the gate, a bing-bong for the captain, three notes up for passport control, one bell for the purser, a jingle for the adverts, a low two-tone for paging, robot blips for NAV-20 and a squelch for the tower.

The script. 36 announcements. Pluto is still a planet. Flights to Andromeda are cancelled, because Andromeda is coming to us. Lost and found is holding one towel. Abductions this evening are scheduled, not strobed.

Can you hear it? Every line goes through Whisper speech-to-text with the music under it, and anything it mishears gets fixed. Two real catches: "HULL twenty" came back as "Hall 20", fixed by writing it as "Hull, twenty,"; and "Spaceways" lost its final s under the music, fixed by slowing the line, ducking the music deeper and holding the duck past the last word. The spelling you hand a speech engine is part of the mix.

The yard. A wired speaker on the Gate 20 computer's headphone jack. Not Bluetooth: on that board, Bluetooth shares a radio with the Wi-Fi, and the Wi-Fi already has a job. A small Python program plays a quiet ten-minute bed (pads, hall air, the saucer idling overhead) with an announcement every one to two minutes from 9 am to 11 pm, then signs off with "Lights down at eleven." Local files only; nothing goes over the network. The film on the Passenger edition has the sound.

Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-30
Checked
WHISPER
2026-10-30
Approved
R. DIXON
2026-10-30
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
—
Projection
THIRD ANGLE
PROJECTION
Title
PUBLIC ADDRESS, GATE 20
Size
B
Scale
NTS
Sheet
14 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

XIVManifest

What went into the saucer

ItemPartQtyNotes
12020 aluminum extrusion, 1,220 mm20 barsCut list above. 135-degree corner brackets, M6 into tapped ends
2WS2812B strip, 60/m, IP67920 LEDs440 on the rim, 480 on the spokes. Reused from last year
3RGBW strip, warm white115 LEDsThe lip under the dome. Also reused
4241-LED WS2812B disc, 9 rings1The big light in the belly
516-LED WS2812B rings8Portholes, behind frosted 4-inch acrylic discs
6Aluminum LED channel, milky diffuser8The rim
7ESP32-S3 module board, Nano form factor1On headers, on protoboard
874AHCT125 quad buffer, socketed13.3 V logic to 5 V data, four lines
9330 Ω resistors4Series, one per data line
10Capacitors0.1 µF at the chip, 100 µF at the controller input, 100 µF at each strip start
11Enclosed supply, 5 V, 60 A, 350 W1Pixel power, an inline fuse on every feed, limited to 30 A in firmware
12IP67 enclosure1Under the deck, on two rails
13Clear 24-inch dome1Cockpit
14Aluminet (aluminized shade cloth)The skin, top and belly

Banner: 1,500 WS2812B LEDs (300 x 5, serpentine), an ESP32-C6 board, a 5 V 60 A supply with power injected along the run. Eyes: two 16x16 WS2812B panels, an ESP32-C6 board. Gate 20: a single-board computer, a 15.6-inch USB monitor. Floodlights: four color floods on my home automation.

Tools: a miter saw with a metal blade, an M6 drill-tap bit, a drill guide, a soldering station, a multimeter, heat shrink and more zip ties than I'd like to admit. Software: C++ on the ESP32 core with FastLED, the desktop simulator, Bun and TypeScript for the control server, SwiftUI for the phone app, Blender for the Plan A renders. The code lives in the hangar for now. If enough of you ask, we'll open the doors.

Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-30
Checked
R. DIXON
2026-10-30
Approved
R. DIXON
2026-10-30
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
—
Projection
THIRD ANGLE
PROJECTION
Title
MANIFEST, SAUCER
Size
B
Scale
NTS
Sheet
15 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

XVUpcoming departures

What boards next

Boarding passes you can hold, for Gate 20's passengers: a circuit board cut like a ticket. Touch both halves and a Darlington pair, running on what leaks through your fingers, switches on the beam. A hundred assembled boards are on order, which makes us either a souvenir shop or a very small airline.

The Gate 20 boarding pass as a circuit board: black, with a yellow saucer and beam, HULL/20 and Gate 20 printed on it, and TOUCH BOTH HALVES under the beam.

Saucer control from the browser. It's built; it boards with the next flash over the cable.

Cat eyes. After the season, the big light in the belly becomes one of a pair of cat eyes. It already has a pupil and a blink, so the saucer took the news well.

Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-30
Checked
R. DIXON
2026-10-30
Approved
R. DIXON
2026-10-30
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
—
Projection
THIRD ANGLE
PROJECTION
Title
UPCOMING DEPARTURES
Size
B
Scale
NTS
Sheet
16 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

XVI29 October, night

Maiden flight

The fleet's newest ship took its maiden flight. Acid rain or shine. It hung over the garage, lit from below, with the banner and the eyes behind it, and every few minutes it abducted nobody in particular.

That night I held Linux up under the saucer and he reached up and touched it.

First contact.

He has since been abducted by the maker bug. Flight record closed. The ship is still up there.

Me holding Linux up under the lit saucer at night, seen from behind; he reaches up and touches its glowing center, a giant inflatable alien beside us.

Prefer the story? Passenger edition

Unless otherwise specified
DIMENSIONS IN MM [IN]
BUILD RECORD: AS BUILT,
NOT FOR MANUFACTURE
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-10-29
Checked
LINUX
2026-10-29
Approved
R. DIXON
2026-10-29
Owner
HULL/20
SPACEWAYS
Document type
BUILD RECORD
Status
RELEASED
Units
—
Projection
THIRD ANGLE
PROJECTION
Title
MAIDEN FLIGHT
Size
B
Scale
NTS
Sheet
17 OF 17
Dwg no.
HULL-20-UFO-001
Rev
E

The 2027 skin

Next year the Aluminet, the belly sail and the ¾ inch plywood plates come off. In their place go sixteen floating panels, top and belly, and aluminum plates for the 2020 frame, all cut from files a script writes. The script draws every part from the frame drawing's numbers, and the same numbers draw these sheets and the 3D model, so the three always agree.

The seams between panels are open: an 8 mm reveal, about 5⁄16 inch, with every panel edge free. It reads as deliberate, and it absorbs ±4 mm of build error at every edge. Each panel sits on three lines of the frame: the rim, the middle of its spoke, and the hub or centre plate. A top panel rests on two standoffs on the rim and one on the porthole plate, and its inner edge is clamped under the deck plate. A belly panel hangs from the rim and from a 2 inch standoff under its spoke, and its inner edge is clamped between the centre plate and the bezel. The reveals under the belly fall right under the eight LED channels, so from below it reads as eight lit lines between glowing panels.

3D model: /build/3d

Drawing HULL-20-UFO-027 sheet 1: skin, floating panels, assembly, parts list and notes

view=exploded-iso ; drawing=HULL-20-UFO-011 ; scale=none (isometric)10987653421
EXPLODED ISOMETRIC, RIM STANDOFF STACKSCALE NONE
Revisions
ZoneRevDescriptionDateApproved
ALLACUT FILES: SIXTEEN FLOATING PANELS WITH AN 8 MM REVEAL; DECK, CENTRE PLATE, BEZEL, PORTHOLE PLATES2026-10-31R. DIXON
C6 D6 E6BNOTE 6: UV-PROTECTED SHEET, OUTDOORS ALL YEAR. FIND 19 M5 × 25 (WAS × 20); FIND 18 5.6 FT (WAS 4); FIND 17 16 FT (WAS 12)2026-11-01R. DIXON
B6 D6 E6CRIM STACK AS BOUGHT SETS SLOPE 10.59°; 3/16 PANELS AND PLATES; M5 × 302026-11-01R. DIXON
C6 D6 F6DSTACK INCLUDES ANGLED WASHER; SLOPE 10.26° TOP, 4.39° BELLY; MID PADS DELETED2026-11-01R. DIXON
Exploded view key, rim standoff stack
BalloonPartFind no.
1RIM A, 2020, TOP SLOTDRAWING 010, A
2T-NUT, DROP-IN, M611
3SCREW, M6 × 1212
4STANDOFF, GLASS, Ø.75 × 1.009
5, 9WASHER, ANGLED, 10.26°, LOWER AND UPPER28
6, 8WASHER, EPDM, 3.015
7PANEL, TOP1, 2
10CAP, M6 (WITH THE STANDOFF)9
Parts list
Find no.Qty reqdPart or identifying no.Nomenclature or descriptionMaterial or specificationNotesBuy
14HULL-20-UFO-027-P01PANEL, TOP, LIFT SPOKE, BRIDLE SLOTPOLYCARBONATE 3/16 [4.76], UV-PROTECTED6 74
24HULL-20-UFO-027-P02PANEL, TOPPOLYCARBONATE 3/16 [4.76], UV-PROTECTED6 74
34HULL-20-UFO-027-P03PANEL, BELLY, LIFT SPOKE, LEG AND PAD EYE SLOTPOLYCARBONATE 3/16 [4.76], UV-PROTECTED6 74
44HULL-20-UFO-027-P04PANEL, BELLYPOLYCARBONATE 3/16 [4.76], UV-PROTECTED6 74
51HULL-20-UFO-027-P05PLATE, DECKAL 5052-H32 3/16 [4.76], ASTM B20951
61HULL-20-UFO-027-P06PLATE, CENTRE, Ø9.50 BEAM PORTAL 5052-H32 3/16 [4.76], ASTM B20951
71HULL-20-UFO-027-P07BEZEL, BELLYAL 5052-H32 .125, ASTM B20951
88HULL-20-UFO-027-P08PLATE, PORTHOLEAL 5052-H32 .125, ASTM B20958
940HULL-20-UFO-027-H01STANDOFF, GLASS, Ø.75 × 1.00, M6 CAPSTAINLESS44
108HULL-20-UFO-027-H02STANDOFF, GLASS, Ø.75 × 2.00, M6 CAPSTAINLESS1210
1140HULL-20-UFO-027-H03T-NUT, DROP-IN, M6, 2020 SLOT 6, SPRING BALLSTAINLESS50
1240HULL-20-UFO-027-H04SCREW, M6 × 12, HEAD TO SUIT STANDOFF BASEA2 STAINLESS1250
13ARHULL-20-UFO-027-H05SCREW, M6 × 16, SAME HEAD (IF BASE IS THICKER)A2 STAINLESS1225
148HULL-20-UFO-027-H06SCREW, BUTTON HEAD, M6 × 20 (TOP MID)A2 STAINLESS10
1596HULL-20-UFO-027-H07WASHER, 3.0 THK, M6, Ø20 TO 25 ODEPDM12120
16—HULL-20-UFO-027-H08DELETED, REV D (WAS PAD, EPDM, 4.0 THK)——
1716 FTHULL-20-UFO-027-H09STRIP, SPONGE, .50 × .125, ADHESIVE BACKEPDM25 FT
185.6 FTHULL-20-UFO-027-H10TRIM, U EDGE, FOR 3/16 SHEETRUBBER810 FT
198HULL-20-UFO-027-H11SCREW, BUTTON HEAD, M5 × 30, WITH NUT, NYLOCK, M5A2 STAINLESS10
2016HULL-20-UFO-027-H12WASHER, FLAT, M5A2 STAINLESS20
218HULL-20-UFO-027-H13SCREW, BUTTON HEAD, M6 × 50 (J5), REUSESTAINLESS—
228HULL-20-UFO-027-H14SCREW, BUTTON HEAD, M6 × 20 (DOME LIP)A2 STAINLESS10
238HULL-20-UFO-027-H15WASHER, FENDER, M6, Ø1.00 ODSTAINLESS10
248HULL-20-UFO-027-H16WASHER, FLAT, M6, WITH NUT, NYLOCK, M6STAINLESS10
251HULL-20-UFO-027-H17GROMMET, Ø1.50 HOLE, .125 PANELRUBBER2
264HULL-20-UFO-027-H18SCREW, BUTTON HEAD, M6 × 16, WITH WASHERA2 STAINLESS6
27ARHULL-20-UFO-027-H19DISC, SANDING, Ø5, 220 AND 320 GRIT, OR FILM, FROSTED—725 EA
2896HULL-20-UFO-027-H20WASHER, ANGLED, Ø25, M6 CLEARANCE, 2.0 THIN EDGE: 10.26° TOP × 48, 4.39° BELLY × 48; 8 TOP MID +0.6 THICKASA, PRINTED12104
29ARHULL-20-UFO-027-H21THREADLOCKER, MEDIUM STRENGTH—11—

Notes

Unless otherwise specified:

  1. Dimensions are in millimetres [inches]. Tolerances: X.X ±0.25 [.XX ±.01], X.XX ±0.13 [.XXX ±.005], angles ±0.5°. Dimensioning and tolerancing per ASME Y14.5.
  2. Profiles per the cut files (*.dxf), 1:1 in inches, one part per file. The cut file governs the profile.
  3. Break sharp edges 0.25 [.010] max. Deburr all holes.
  4. Aluminum parts: 5052-H32 per ASTM B209. Find 5, 6: 3/16 [4.76] thk. Find 7, 8: .125 [3.2] thk.
  5. Finish, aluminum: black powder coat, mask holes. Find 5 is over 23 in: powder coat, not anodize.
  6. Polycarbonate parts, find 1 to 4: 3/16 [4.76] thk, UV-protected grade, coated face out; cast acrylic acceptable. Leave the protective film on until install.
  7. Frost the inner face only (underside of find 1, 2; upper face of find 3, 4): wet-sand 220 then 320 grit, or frosted film. Gloss face out.
  8. Panel holes Ø.394 [10] for M6, clearance for thermal movement. Line the slots with U edge trim, find 18.
  9. Reveal between panels .31 [8], absorbs ±.16 [4] build error at each edge.
  10. If a 36.6 in belly panel exceeds the cutting bed, split each belly facet on its centre line (16-panel belly).
  11. M6 into T-nuts and tapped posts: medium-strength threadlocker, find 29.
  12. Stack as built: standoff barrel 25.4 [1.00], printed angled washer (find 28) 4.26 [.168] at the axis, EPDM washer 3.0 [.12]: 32.66 [1.286] from the rim to the panel underside, which sets the top slope 10.26°. Belly stack 31.36 [1.235], slope 4.39°. An angled washer goes on each side of the panel at every standoff; the 8 top mid-support washers print 0.6 thicker. Test-fit one standoff before ordering the rest. See section A-A, sheet 4.
Unless otherwise specified
DIMENSIONS IN MM [IN]
TOLERANCES: X.X ±0.25 X.XX ±0.13
ANGLES ±0.5° BREAK EDGES 0.25
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-11-01
Checked
R. DIXON
2026-11-01
Approved
R. DIXON
2026-11-01
Owner
HULL/20
SPACEWAYS
Document type
ASSEMBLY DRAWING
Status
IN WORK
Units
MM [IN]
Projection
THIRD ANGLE
PROJECTION
Title
SKIN, FLOATING PANELS, 2027
Size
B
Scale
AS SHOWN
Sheet
1 OF 4
Dwg no.
HULL-20-UFO-027
Rev
D

Drawing HULL-20-UFO-027 sheet 2: cut parts

Aluminum plates, find 5 to 8, at 1:12The deck plate, centre plate, belly bezel and porthole plate drawn from their cut files at one scale, with centre marks on every hole. DECK-PLATE.DXF QTY 1 5 CENTRE-PLATE.DXF QTY 1 6 BELLY-BEZEL.DXF QTY 1 7 PORTHOLE-PLATE.DXF QTY 8 8 5
PLATES, FIND 5 TO 8SCALE 1:12
Polycarbonate panels, find 1 to 4, at 1:12The four polycarbonate panels drawn from their cut files at one scale: top panels with and without the bridle slot, belly panels with and without the leg slot. Outer edge down. TOP-PANEL-LIFT.DXF QTY 4 1 TOP-PANEL.DXF QTY 4 2 BELLY-PANEL-LEG.DXF QTY 4 3 BELLY-PANEL.DXF QTY 4 4 6 7
PANELS, FIND 1 TO 4SCALE 1:12
Unless otherwise specified
DIMENSIONS IN MM [IN]
TOLERANCES: X.X ±0.25 X.XX ±0.13
ANGLES ±0.5° BREAK EDGES 0.25
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-11-01
Checked
R. DIXON
2026-11-01
Approved
R. DIXON
2026-11-01
Owner
HULL/20
SPACEWAYS
Document type
DETAIL DRAWING
Status
IN WORK
Units
MM [IN]
Projection
THIRD ANGLE
PROJECTION
Title
SKIN, CUT PARTS
Size
B
Scale
1:12
Sheet
2 OF 4
Dwg no.
HULL-20-UFO-027
Rev
D

Drawing HULL-20-UFO-027 sheet 3: deck plate, find 5

view=deck-plate-top ; scale=1:5 ; units=mm (paper) ; part=deck plate ; source=make_dxf.py deck() 764.2 (30.09 in) A/F764.2 (30.09 in) A/FPCD Ø635.5 , B1-B8 8× Ø6.6PCD Ø744.2 , A1-A8 8× Ø6.6C1 Ø38.18× R3 A1A2A3A4A5A6A7A8B1B2B3B4B5B6B7B8B holes on 22.5° from A. Origin at plate centre. Coordinates: HOLE TABLE. DECK PLATE, TOP (1:5)
Hole table, find 5
TagXY[X][Y]SizeDescription
A1+372.100.00[+14.650][0.000]Ø6.6 [.260]J5 DECK + HUB RING TO E POST
A2+263.11+263.11[+10.359][+10.359]Ø6.6 [.260]J5 DECK + HUB RING TO E POST
A30.00+372.10[0.000][+14.650]Ø6.6 [.260]J5 DECK + HUB RING TO E POST
A4-263.11+263.11[-10.359][+10.359]Ø6.6 [.260]J5 DECK + HUB RING TO E POST
A5-372.100.00[-14.650][0.000]Ø6.6 [.260]J5 DECK + HUB RING TO E POST
A6-263.11-263.11[-10.359][-10.359]Ø6.6 [.260]J5 DECK + HUB RING TO E POST
A70.00-372.10[0.000][-14.650]Ø6.6 [.260]J5 DECK + HUB RING TO E POST
A8+263.11-263.11[+10.359][-10.359]Ø6.6 [.260]J5 DECK + HUB RING TO E POST
B1+293.56+121.60[+11.558][+4.787]Ø6.6 [.260]DOME LIP
B2+121.60+293.56[+4.787][+11.558]Ø6.6 [.260]DOME LIP
B3-121.60+293.56[-4.787][+11.558]Ø6.6 [.260]DOME LIP
B4-293.56+121.60[-11.558][+4.787]Ø6.6 [.260]DOME LIP
B5-293.56-121.60[-11.558][-4.787]Ø6.6 [.260]DOME LIP
B6-121.60-293.56[-4.787][-11.558]Ø6.6 [.260]DOME LIP
B7+121.60-293.56[+4.787][-11.558]Ø6.6 [.260]DOME LIP
B8+293.56-121.60[+11.558][-4.787]Ø6.6 [.260]DOME LIP
C10.000.00[0.000][0.000]Ø38.1 [1.500]CABLE PASS
view=deck-plate-edge ; scale=1:5 ; units=mm (paper) ; part=deck plate ; source=make_dxf.py deck() THK 4.76 (0.1875 in)5052-H32, black powder coat if offered DECK PLATE, EDGE (1:5)
Unless otherwise specified
DIMENSIONS IN MM [IN]
TOLERANCES: X.X ±0.25 X.XX ±0.13
ANGLES ±0.5° BREAK EDGES 0.25
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-11-01
Checked
R. DIXON
2026-11-01
Approved
R. DIXON
2026-11-01
Owner
HULL/20
SPACEWAYS
Document type
DETAIL DRAWING
Status
IN WORK
Units
MM [IN]
Projection
THIRD ANGLE
PROJECTION
Title
PLATE, DECK
Size
B
Scale
1:5
Sheet
3 OF 4
Dwg no.
HULL-20-UFO-027
Rev
D

Drawing HULL-20-UFO-027 sheet 4: sections and details

view=section-A-A-rim-standoff ; scale=1:1 ; units=mm (paper) ; source=make_dxf.py, README.md, PARTS-2027.md, frame-2020 25.432.7Ø19.05EPDM 3.0GLASS STANDOFF Ø3/4 x 1 inM6 DROP-IN T-NUTM6 STUD ≤10 (or M6 x 12 through the base)RIM 2020, TOP SLOT (generic profile)M6 CAPPOLYCARBONATE 4.76 (3/16 in), 10.26°ANGLED WASHER, ASAANGLED WASHER, ASA Angled ASA washers both sides of the panel, EPDM 3.0 between washer and panel. Rim top topanel underside on the standoff axis: 32.7 (make_dxf STANDOFF 32.66, rev 65ccf283fc92).Cut up-slope through the x = +320 standoff, looking toward -x. SECTION A-A, TOP PANEL AT RIM STANDOFF (1:1)
view=detail-B-deck-clamp ; scale=2:1 ; units=mm (paper) ; source=make_dxf.py, README.md, PARTS-2027.md, frame-2020 15.0 THREAD IN POSTDECK 5052 4.76 (3/16 in)EPDM STRIP 1/2 x 1/8 inM6 x 50 BUTTON HEAD (J5)TOP PANEL 4.76 (3/16 in), notched round J5HUB RING 2020 (C), Ø6.5 THROUGHE POST 2020, END TAPPED M6 Thread in the E post 15.0 (M6 x 50). Section on the spoke line through J5. The panel's Unotch (10 wide) clears the screw; the deck bears on the EPDM strip at the panel's inneredge. DETAIL B, DECK CLAMP AT J5 (2:1)
view=detail-C-belly-clamp ; scale=2:1 ; units=mm (paper) ; source=make_dxf.py, README.md, PARTS-2027.md, frame-2020 19.1CENTRE PLATE 5052 4.76 (3/16 in)EPDM 1/8 inEPDM 1/8 inBEZEL 5052 3.17M5 NYLOCK + WASHERBELLY PANEL 4.76 (3/16 in)M5 x 30 BUTTON HEAD + WASHER Panel shown clamped flat; it takes its 4.39° slope outboard of the bezel. Screw tip +3.9past the nylock. DETAIL C, BELLY CLAMP (2:1)

Measure the ship before ordering

The cut files follow the frame drawing. The plates went on in ¾ in ply, so check these against the frame as built:

  1. Hub ring, outside across flats: about 30⅛ in (765 mm). This is the deck size.
  2. Hub ring top above the rim top: about 6⅜ in (162 mm). This sets the top panels' slope.
  3. Rim side, outside: about 36.1 in (916.6 mm).
  4. Centre plate underside below the rim top: the F legs (90 mm) under the core ring, plus the plate. This sets the belly slope.
  5. Centre plate leg holes: 4 on a 7⅛ in (181 mm) radius, each 1.57 in to one side of a flat's centre line, all to the same side. Lay the old plate over the print.
  6. Dome lip bolts: 8 × M6 on a 25 in circle between the spokes, through the lip into the deck. No G riser posts were built: the dome sits on the deck. Cable hole Ø1½ in, central.
  7. Lift pad eyes 1000 mm from centre; leg posts 940 mm, on the same 4 spokes.
  8. Porthole posts K: 744 mm from centre, 60 mm tall.
  9. Belly LED channels must sit above the belly panels. If a channel hangs lower than the panel at mid-span, the belly standoffs go to 1½ in and 2½ in.
  10. The 2020 profile, drop-in T-nut and standoff internals in the sections are generic: check them against the hardware in hand.
Unless otherwise specified
DIMENSIONS IN MM [IN]
TOLERANCES: X.X ±0.25 X.XX ±0.13
ANGLES ±0.5° BREAK EDGES 0.25
DO NOT SCALE DRAWING
Drawn
R. DIXON
2026-11-01
Checked
R. DIXON
2026-11-01
Approved
R. DIXON
2026-11-01
Owner
HULL/20
SPACEWAYS
Document type
DETAIL DRAWING
Status
IN WORK
Units
MM [IN]
Projection
THIRD ANGLE
PROJECTION
Title
SECTIONS AND DETAILS, SKIN 2027
Size
B
Scale
AS SHOWN
Sheet
4 OF 4
Dwg no.
HULL-20-UFO-027
Rev
D