Levandor

KrakenPi CarBox Activity Log

6 min read

For Agents

Reverse-chronological session log. Newest entries at top, grouped by date (## YYYY-MM-DD). Each bullet: one piece of work, short summary, wikilinks to docs touched. Updated by obsidian-documenter on every project doc write. Read by historian at bootstrap.

2026-08-26

v2 — three-part split (later session)

  • Split the enclosure into 3 printed parts because the printer has only a 0.4 mm hotend and v1 was a ~15 h single print: floor 312 x 145 x 3 (bosses to Z=7), frame 312 x 145 x 44, lid 312 x 145 x 2.4, joined by 10x M3 up through the floor into 10 x 8 x 7 bosses inside the frame. Floor prints first (~2.5 h) so the hardware can be bolted in and wired while the frame prints — krakenpi-carbox, carbox-enclosure-geometry-spec
  • Built the frame as its own body (block → fillet → shell removing both the top and bottom faces) rather than splitting a shelled box. That decouples WALL from FLOOR and is where most of the material saving comes from — carbox-enclosure-geometry-spec
  • Cut material 23 %: 384 → 291.6 cm3. WALL 3 → 2.5, LID_T 3 → 2.4, lid grilles widened to 6 mm slots at 10 mm pitch across three zones, and a floor lightening grid added under both units (doubles as airflow, hidden by the devices). FLOOR stays 3.0 — it is the structural member and the M3 heads recess into it — krakenpi-carbox, carbox-enclosure-geometry-spec
  • 🔴 The cable corridor was the real find. Building the wiring mockup exposed that the socket ribs walled off the back cable channel and the mid-span lid posts sat in it — the Kraken’s PWR and DATA leads had no route to the adaptor or the Pi. Three coordinated fixes: lid posts sink 2 mm INTO the wall (_PC = WALL + POST_D/2 - 2, also necessary because the split frame has no floor to stand them on), SOCKET_Y 115 → 108 and RIB_Y0 92 → 88, and a RIB_NOTCH at Y 125..136, Z 8..32 through both ribs. CABLE_Y = 131 is now a named constant — carbox-enclosure-geometry-spec
  • ⚠️ A spot check passed; a full line scan is what would have caught it. The build now scans X 40..250 in 0.5 mm steps at three Y offsets around CABLE_Y and fails if any point is solid. Generalised as a rule: anything a cable, hand, tool or airflow must pass through gets a swept scan, not a point probe — fusion360-mcp-scripting, carbox-enclosure-geometry-spec
  • Corrected the LED. v1 modelled a 24.4 x 6.5 mm rectangular window; the part is actually a singular pre-wired 12 V LED with a 6.5 mm sleeve, and 24.4 mm was its overall length including leads, not a panel dimension. Now a straight D6.6 bore with a D14 x 4 mm round collar behind the wall — 6.5 mm of grip — krakensdr-pi5-reference-dimensions, carbox-enclosure-geometry-spec
  • Recorded the wiring topology: adaptor → Pi power, adaptor → Kraken power, Pi ←> Kraken data; 12 V positive straight to the adaptor, 12 V ground through the switch, and the status LED tapped BEFORE the adaptor so it follows the switch — krakenpi-carbox, carbox-enclosure-geometry-spec
  • Added a mockup + labels pass to build_carbox.py: KrakenSDR, Pi 5 case, D28 x 40.5 adaptor and 8 coloured wires (red 12V+, black 12V-, white USB power, blue USB data) built from quadratic-bezier-rounded polylines unioned via TemporaryBRepManager, plus a COMPONENT_LABELS sketch of 14 sketch-text labels at Z=52. Mockup bodies are built after the export loop and the export iterates (floor, frame, lid) explicitly, so preview solids can never reach an STL/3MF — krakenpi-carbox
  • Added 6 new Fusion API gotchas (21–26): appearances live in the 'Fusion Appearance Library' and not in material libraries (silent no-op if you look in the wrong one); Part Design docs allow only ONE component; TemporaryBRepManager beats sketch+sweep for decoration but ~300 unions times out the MCP call while still completing; BaseFeature bodies can be renamed/styled after finishEdit() if matched by index/name rather than the returned proxies; sketchTexts.createInput2 + setAsMultiLine needs a generously sized diagonal; and a spot check is not a scan — fusion360-mcp-scripting
  • Status: 102 self-checks pass. Exports verified sitting flat on Z=0 (floor 0..7, frame 0..44, lid flipped so its top face is on the plate, 0..4.4). The Fusion doc is now SAVED as rpi_kraken_car_case in the Default Project cloud folder — it is no longer Untitled. v1 kept as build_carbox_v1_onepiece.py, its exports moved to old_onepiece/ — krakenpi-carbox

v1 — one-piece box (earlier session, superseded)

Historical record — figures below are v1 and no longer true

The one-piece box, 3 mm walls, 158 checks, the unsaved Fusion doc and the 24.4 mm LED window were all superseded by the v2 entries above. Kept for provenance.

  • Project created. Documented the completed KrakenSDR + Pi 5 car enclosure — 312 x 145 x 50 mm, 3 mm walls, two printable parts, Kraken at X 13–190.3 on 4 mm standoffs, cased Pi 5 at X 201–299.5 on 3 mm standoffs, L-shaped 12 V power bay, 15.5 mm back cable channel — krakenpi-carbox, carbox-enclosure-geometry-spec
  • Recorded the verified reference dimensions that were hard to find: KrakenSDR 177.3 x 113.5 x 25.86 (+4.7 fan guard) with its 8x M3 pattern 162.6 x 88.8 from drawing KH-ASSEMBLYC1, the SMA bank’s −1.7 mm offset from centre, official Pi 5 case 98.5 x 70.3 x 33, Pi 5 board 58 x 49 M2.5 pattern, H2S build volume 340 x 320 x 340 — krakensdr-pi5-reference-dimensions
  • Captured the four buy/print blockers: PLA will sag at a parked car’s 70–80 °C (ASA/PETG/PC only); the Kraken needs M3x16 not its stock M3x12 because the fixing reuses its own case screws; the 12 V wire must be threaded bare before crimping (3.4 x 6.7 mm slot); and the switch cutout leaves only a 0.3 mm web to each screw hole — the weakest feature in the print — krakenpi-carbox, carbox-enclosure-geometry-spec
  • Documented the build-script-as-model workflow: build_carbox.py wipes the active Fusion design and rebuilds from a single CONFIG block, self-checks, re-exports. The Fusion doc is unsaved/Untitled (no cloud project on this machine) so the .step + the script are the only durable artifacts — krakenpi-carbox
  • Recorded the verification method — 158 pointContainment assertions plus line scans, including negative asserts that the wall between two SMA slots, the 0.3 mm switch web, the post bodies and the counterbore floors are still solid. Renders were explicitly rejected as evidence: they see through an opening to the far wall — carbox-enclosure-geometry-spec
  • Added 9 new Fusion 360 API gotchas (12–20) from this build — unitless userParameters, un-renameable root component, the PointOutsidePointContainment enum name, unpredictable construction-plane offset direction, modelToSketchSpace, counterbore plane choice, obround profiles, MCP timeouts that still complete, and verify-by-assertion — fusion360-mcp-scripting
  • Initialized activity log.

Graph View