9 phases
RF-1 · Phase 1 — Drivetrain Coupon
The locomotion de-risk for the rolling face (../brief.md).
v0.2software-verified
rendered + manifold), awaiting print/test-fit
RF-1 · Phase 1b — 4-Wheel Belt-Drive Drivetrain Coupon
The 4-wheel evolution of the locomotion de-risk for the rolling face (../brief.md), per the design doc.
v0.4print-ready
three-compartment symmetric cradle on caliper-confirmed N20 dims; manifold + collision/clearance-checked (awaiting test-fit
RF-1 · Phase 2 — Powered Rover
The integrated, untethered, phone-driven rover — drivetrain + electronics + power, layered and wired, before the OLED face.
v0.9in progress
v0.9 second-print feedback applied: motor clip reworked to a narrow free-flexing finger (lip 0.8→0.6), inter-bay ribs lowered (z=18.5), wire chase cut to the bay floor; manifold + collision-checked, awaiting re-test
RF-1 · Phase 3 — The Face
The two-part enclosure that finally gives RF-1 a face: a press-fit cover closes the powered-rover body, and a separate tilted head module at the front holds a 0.96″ OLED behind a masked visor slot.
v0.3software-verified
full cover + face (contract-driven, gated): cover clears the Ø34 wheels, tail USB×2 + switch cutouts, front face-module socket; head wedge with masked visor. 3 gates PASS, form_lint 0 FAIL, form-resolved (one 5 mm house radius across body+cover+head). Next: seam/retention coupon → dry-fit → print.
RF-02 · The Monocoque Loaf
Why RF-02 instead of patching RF-1/P3: RF-1 grew function-first with form patched on at the end (three corner radii reconciled after the fact, a chamfer-vs-round-over residual, a body-align override).
v0.14in progress
v0.14 ARCHITECTURE PIVOT (board-in-cover) + version-stamped parts; all 7 gates GREEN, both parts 1-shell manifold, print-orientation-correct. The custom perfboard moved from the tub deck INTO the COVER: it mounts component-side-UP on 4 wall-webbed ceiling standoffs (PCB top z33.6), pushed rear (pcb_cy −4) so its front edge clears the nose OLED. The on/off switch is soldered to the perfboard and its actuator pokes UP through the Ø11.9 crown hole — the captive-lip-in-the-crown holder is RETIRED (the lip rode 5.7 mm up the actuator, so making it captive drove the body into the component zone; soldering the switch to the board ends that whole fight). The TUB is now pure chassis (drivetrain + battery + the seam locator); its board-mount idler-corner standoffs were REMOVED, so the tub is reprint-different too. Seam crossing = 3× 2-pin JSTs (battery keyed distinct so it can't seat in a motor socket). RETAINED: WIDENED body 48 (track 58); 5×2 N35 magnet cover retention (3 stations, friction-fit Ø5.1 pockets, flush Ø7 pads); FLUSH seam; rigid KEEPER-BAR N20 retention; OLED variant-C (cover bosses + tub chin); crown 49; flat-top visor (prints with supports). VERSION STAMPED (repo-standard debossed lib/stamp.scad, −D-injected from parts.json so the glyph can't disagree with the manifest): C v0.14 debossed on the cover's INTERIOR rear wall, T v0.14 on the tub interior floor — both INTERNAL only (repo rule: no external stamps; the version is read at assembly, with the part open). PENDING (physical): REPRINT cover + tub (both geometry-changed; motor_keeper geometry-identical — vol+bbox unchanged, STL only re-tessellated — so no reprint) and bench the board-in-cover stack, the 5×2 press-fit magnets, the flush seam, the switch-through-crown fit, the keeper + wheel running clearance; calipered PROVISIONALs in (2026-06-29: OLED 1.8, USB ±9.75 / z≈34, comp_tall 8, Gikfun 12×12 switch).