Oct 27, 2026 ~8:45 AM ET: RUN 9 — flip-up door, lower-bay panels, components assembly, 10 W 355 nm UV note
Queue done on the OSSO DIY laser-ablated PCB machine CAD (head design untouched; public site/Cloudflare untouched):
- Exhaust: D100 90° elbow at the front port, outlet UP (replaces the front-down flex). Elbow top under the top panel (zmax 615 vs top 656.5).
- Embossed CAD text labels removed forever (
label() is a no-op); never model embossed labels going forward.
- STEP: purchased/off-the-shelf parts go into a top-level
components assembly. Raw extrusions and sheet stay with their made parts. panels stays its own top-level assembly (all external 1.5/3 mm sheet, including lower-bay close-in and the door).
- Lower cart bay closed with 1.5 mm aluminium panels (front/back/left/right) in
panels, Z_RB (-511) to -TB (-3). Front/back have clamp-upright cutouts so the chiller and laser-base fixtures pass through.
- External panels render light grey in GLB/USDZ (not white); review translucency kept where needed.
- Front door: slide-up replaced by flip-up on a top piano hinge (leaf 18), open catch at 95° holding the handle. Tracks, shoes, pulleys, counterweights, cords, and cord-break stops removed. Interlock and brush light seal kept. Gravity closes (fail-safe). Niche/hood clearance OK; open top under the top panel; front reach ~181 mm. Catch hold ~2 N.
- Upper-chamber floor: removed bay_lid + 6 standoffs + lid grommets (electronics left the bay in run 6/7; lid not needed for UV shielding). Kept bay_floor_plate (shields the bottom-panel laser-slot and coolant hole from stray 355 nm).
- Docs/notes: laser is a 10 W 355 nm UV source (site copy to be updated separately).
Final checks (America/Detroit): size 993 × 763 × 1233.5 mm (top panel; exhaust elbow top ~12 mm above), mass 125.9 kg, castor push SF 1.24 / feet 1.41. Interference self clear; harness REAL 0; boundary ok. STEP/GLB/USDZ 585 parts. Delta at review_sync_run9_20261027_0852.
-
2026-10-27 17:42: OSSO blank v3b PCBs arrived (qty 5, 140x140 mm, 1.6 mm, black, 2-layer). Logged only.
-
2026-10-27 17:58: Version snap v19 (run 9 + cam head concept, station strategies, drawings, cam_head_v3).
-
2026-10-27 ~9:00 PM ET: OSSO blank v3c generator updated (items 2-5 from NEXT_BLANK_CHANGES, plus a version mark). v3b untouched. Not ordered.
-
2 Fiducials moved to the frame: A(22,4), B(118,4) are Ø0.4 plated with rings on both sides; C(118,136) is a top surface dot and C'(22,136) a bottom surface dot (same right=top, left=bottom scheme as before). Ø4 mask, Ø5 silk keep-out. Edge vias were checked against the fiducials, with the via pad required to sit at least 1.0 mm outside the Ø5 keep-out. The closest via is 1.38 mm away, so none were moved or dropped. C1's 21.74/118.26 positions were wrong; the real nearest vias are at x 19.20/24.28 and 115.72/120.80. Script now asserts frame symmetry about x=70 (orientation marks excepted).
-
3 Slot routed X70Y135.5 to X70Y136.5 with a Ø2.0 tool, giving 2.0x3.0. The pour clearance is 0.3 mm all round with no nick.
-
4 Clamp zones (27,1)-(63,6), (77,1)-(113,6), (27,134)-(63,139), (77,134)-(113,139). They are masked pour with no vias, pads or holes, so no copper change was needed. Written up in pcb/fab/v3c_blank/JIG_HOLD.md.
-
5 Via text corrected to Ø0.3 drill / Ø0.6 pad. New drawing: project/drawing_src/base_v3c.py -> osso_base_v3c.png.
- Added an 'OSSO v3c' silk-void version mark, 1.2 mm high, on the bottom rail x 10.3-18.7 (bottom silk mirrored at x 121.3-129.7). It is silk only. FAB_NOTES 8, 9 and 11 updated.
- Review: tools/check_v3c.py rendered all layers and ran 63 checks; all pass (connectivity: only the fiducials isolated). Sheet: pcb/fab/render/v3c_review/review_sheet_v3c.png. Re-zipped OSSO_blank_v3c_gerbers.zip.
-
2026-10-28 ~10:45 AM ET: Run 10. linear_cam_short_v3 built (r3 feeder, FR4 wheel, OPTIONAL coin vibration motor, enclosure panels, cheap fixes; 2.20 kg, 289 x 148.5 x 175.5) and integrated on the X/Z carriage in place of the 5-station head. Down camera moved inside the head. Soft limits X >= 160, Z >= 316.4, tray T_BACK interlock Z >= 320. Head-related clashes 0. Machine 126.9 kg, castor SF 1.24. Open: no hot-air nozzle on v3; N6 cannot reach board X 200-215; feeder bundle bend R/D 1.6.
-
2026-10-28 ~4:50 PM ET: Reflow v6 (test baseline): ONE ball size, Ø0.25, for every part (replaces Ø0.30); one size per job, count tuned ±1 ball per coupon round (BALLS_OVERRIDE, calc_v5.py default D=0.25). Balls per terminal at 15 µm tin: 0402 2 (116%), 0603 4 (101%), 0805 8 (109%), 1206 14 in 3 toe rows (100%; 0.85 mm toe needed, 0.88 exposed, no extension), SOT-23 4 (116%, +0.325 toe), SOIC 6 (110%, +0.12 toe), QFN 1 (103%). Ø0.30 rule and QFN fallbacks withdrawn. Docs process_reflow.md, process_v3.md; drawing drawings/2026_10_28_two_step_reflow_pad_guidance_v6.png; site latest.html updated. All subject to testing.
-
2026-10-28 ~7:00 PM ET: Component rotation plan rev 2 (project/docs/plan_component_rotation.md). Decisions: rotation plate is the only rotation method; nozzle theta REJECTED (no extra motor or movement on the head, ever); cycle time not a concern (multiple plate cycles OK); textured pocket, no vacuum; global-shutter OV9281 up camera + Ø30 4-zone ring now; vacuum sensors on N4/N5 now (H20/H21, W60/W61); unclear polarity → auto-reject; tape + keyed nests. Plate module CAD cad/rotation_plate_v1/ (NEMA 8 + AS5600, mSLA plate, laser-cut alu bracket) on the tray carrier front-right X 420, Y = T+14: 0 clashes (min gap 0.68 mm N4 approach); left-beam X111/Y440 position failed (16 hits, unreachable). Found: the existing up camera at X111 is out of N4 reach (X ≥ 171); proposed moving it onto the plate bracket. Keyed nest cassette concept cad/nest_tray_v1/. Per-part pickup/rotation tunables + pickup_overrides.yaml + one-parameter-per-round coupon procedure. INVENTORY lines 33/40 fixed. cam_head_v3 run_fov() re-run: 0 hits, 46.2 mm. Not deployed.
-
2026-10-28 ~7:45 PM ET: Run 12 baseline. Up camera moved onto the rotation plate bracket on the tray carrier (X 111 camera, ring and left-beam bracket retired). 2 M3 countersunk holes added to the tray carrier. Tray cable route: igus E2.10 class drag chain 10x15 R48 L285 along the tray axis (fixed end at mid travel, lower run in a 1.5 mm alu channel), AS5600 in PWM mode, chain-rated USB (W44/W45/W62/W63). Tray motor margin with +0.19 kg and 2 N chain drag: SF 4.9 (was 6.1). Dedicated cam_head_v3 sweep (run12_sweep.py): the head frames sit 4.3 mm below the needle tip. As built: 82 hits. With a 1 in needle: 17 hits, all on tray-axis hardware (keep-outs needed). New interlocks 368/356 (or 355/343). Diagram decision box now points to per-part limits. STEP/GLB/USDZ, renders and cycle MP4/USDZ re-exported. Not deployed.
-
2026-10-28 ~8:50 PM ET: Run 13 study (the designer): a 4 in (101.6 mm) needle (+88.9 over the ½ in) with the cam_head_v3 body bottom raised by the same 88.9 mm, so the body sits 173.5 mm above the extended tip. Fine tray sweep (T 322–540 every 10 mm plus key positions, X 160/270/425, 194 poses; run13_needle_raise.py): with needles up at board Z 316.4 there are 0 hits at every T and every X, including over the fixed left rail. The tray no longer has to fully retract while the head is at board height. The N4 needle can go down to board Z from T ≥ 462 (tray off the needle line). Plate, camera, tape pick and board-touch poses: 0 hits. Min gap 5.2 mm with needles up, 1.16 mm overall. Side effects: the lowest point at HOME is now the retracted needle tips (128.5 mm below the body; stroke still 45 mm); 16 ga tip deflection is about 0.8 mm at 0.1 N side load and the first bending mode is about 130 Hz, so a tip guide or a stiffer needle is needed for 0402. Chart drawings/2026_10_28_needle_raise_sweep_v1.png. Proposed as the working baseline; head CAD not yet changed. Other extensions skipped. Not deployed.