1. Documentation
  2. Engineering
  3. USB housing R4 candidate
  4. R4 mechanical review

Hardware

R4 independent mechanical review

Scope: PROVISIONAL_MECHANICAL_REVIEW. Nominal geometry checks PASS for the catalogue RX-M3x5.7 profile. Overall result and print release BLOCKED: actual stock identity/fit remain unconfirmed; every physical observation is NOT RUN. No historical R1/R3 approval is inherited.

Reviewed final native source, actual component meshes, five coupon/fixture meshes and all five rendered PDF pages. Final PDF SHA256: 3a91c27b9285c120ab8562c516784aacd6e0a3d34a97df8cc6c8af326294c47e. No open mechanical instruction finding remains. Mesh/source hashes and direct measurements are in result.json, frozen-input.json, coupon-frozen-input.json and the four geometry JSON reports.

  • The 76.8 × 35.9 × 20 mm exterior is rectangular; all bosses, recessed heads and retention features remain inside it. The loaded official board is enclosed without using M3 through its roughly 2.75–2.82 mm PCB bores.
  • Assembly is feasible at the tested poses: heat-set inserts in the empty supported base and cool; lower the unplugged PCB onto pads/pegs; lower lid 1.2 mm left; slide +1.2 mm; then seat both M3×8 screws. Three surviving tongues have 1 mm width, 1.5 mm slots, 0.85 mm slot walls and 0.3 mm roof clearance. Reverse service removes screws, slides −1.2 mm and then lifts.
  • Nominal screw stack: 3 mm grip, 5 mm axial insert overlap, 0.7 mm insert-end reserve, 1.7 mm plastic-bottom reserve and 0.5 mm head recess. Catalogue OD4.6 inserts leave 1.9 mm boss wall; entry countersink leaves 1.75 mm. Screws clamp printed hard stops and bypass the PCB, whose annuli have 0.18 mm capture play. Actual useful threads, head bearing and torque remain physical questions.
  • Rear tabs have 0.75 mm overlap, 3.9 mm width, matched sloping roofs with 0.3 mm vertical clearance, 4.4 mm pockets, ≥1.5 mm roofs and 1.2 mm rear walls. Screws positively block reverse slide near 0.2 mm; native tab/roof interference blocks straight lift at +0.4 mm, including −0.2 mm X backlash. These are geometric capture checks, not strength results.
  • Independent regeneration exactly reproduced both full-case meshes and the final 3 mm-wall jig. Full-case solids clear 13 slide and 10 lowering poses. The nonmanifold official board has zero transverse edge crossings or vertices inside plastic at 26 checked poses; two valid USB-C solids independently clear, with minimum lid gap 0.80024 mm. Finite samples and surface checks can miss untested or face-only contacts.
  • Four coupon solids exactly equal native full-case crops (zero symmetric-difference volume). Their closure and closed-pair jig insertion paths clear. The tall X55.5 stop restrains the lid, X67.7 restrains the base, and the jig permits vertical motion. Declared heater and driver envelope proxies clear; actual tool geometry is unmeasured. Cable boots/latch finger access are unmodeled and require actual service cables.

The final physical protocol correctly keeps the front screw fitted for its 5 N separating test, then performs service cycles. The rear pair closes outside the jig, lowers into it, and is pulled +Z at the exposed Y9.1 roof edge. Both trials use 5 N for 10 seconds three times and 10 service cycles. Rear fixture drag must be below 0.5 N; wall deflection ≤0.1 mm and loaded relative X travel ≤0.3 mm prevent fixture compliance from being mistaken for joint failure. Residual gap must be ≤0.2 mm. Full-case acceptance permits board play ≤0.3 mm and tests actual cables, orientation, retention and thermal behavior separately.

Both exact nominal plates pass mesh preflight; the advisory sampled reports have zero collisions. Open passages, bridge shelves and supported pocket roofs have no cleanup membranes. Slicing, toolpaths, physical bridges, pilot heat-setting, actual stock/tool fit, strength, cable/latch operation and thermal performance remain NOT RUN. This record is neither PREPRINT_ASSEMBLY_REVIEW nor dispatch permission. Ordinary exact-job preparation/review and existing owner authorization still govern any future print.

Subject sources were read-only. Reviewer changes are confined to .scratch/r4-mechanical-review; no commit, slicing, approval or printer action occurred.