ADR-0017: Print practice from the community, and the fasteners in the model

# ADR-0017: Print practice from the community, and the fasteners in the model

- **Status:** accepted (2026-10-13)
- **Date:** 2026-10-13
- **Requirements touched:** D-01 (survives a 1.2 m drop), the kit's buildability
- **Builds on:** ADR-0015 (buildable shell), ADR-0016 (flared post roots)

## Context

Rafa's requests, verbatim:

> "point yourself also to 3D printer feedback forums online as well to integrate best-practice from ongoing discussions as well into our design"

> "And the bom and model have the screws / fastenings as well?"

A research pass read community and maker sources: CNC Kitchen's tests and guides, the Prusa knowledge base and forum, the Bambu Lab forum, FTC's tolerancing notes, and a Xometry snap-fit thread. Reddit could not be read. Each finding was mapped to this design. The ones that changed something are below, each with its source. The research found the posts' flares and the TPU band's zero-clearance fit already consistent with the advice.

On the second question: the BOM listed the inserts, the screws and the glues. The model had only the insert bores and the TPU washers. Nothing checked that the screw heads clear the parts under the board.

## Decisions

1. **The insert bores match the insert's datasheet.** The BOM's insert, the ruthex RX-M2.5x5.7, has this table on 3DJake (opened 2026-10-13): "⌀ D1 4,6 · ⌀ D2 3,9 · ⌀ D3 4,0 · L 5,7 · W (mind.) 1,6". CNC Kitchen's table for the same insert names D3 the hole and W the minimum wall. CNC Kitchen also advises: "A blind hole should be slightly deeper than the length of the insert (approx. 1 mm)".
   - **Bore:** 3.6 → 4.0 mm, and 6.0 → 6.7 mm deep.
   - **Post:** 6.4 → 7.2 mm across, for the 1.6 mm wall.
   - **Flare:** narrowed from 2.0 to 1.6 mm, so the root stays 10.4 mm across. The tail mic and the tap icon keep their clearances.
   - **Why it mattered:** the 3.6 mm bore would have printed at about 3.35 mm. CNC Kitchen found "every hole measured about 0.25 mm smaller than designed". That is roughly twice the interference the insert is made for, with about 0.9 mm of wall left.
   - **Re-pack:** the wider post pushed the NFC tag 0.5 mm toward the tail and the icon LED 1.5 mm with it.
2. **The fasteners are in the model.** Each post now has, as layout boxes with role `fastener`:
   - `insert of post …`: Ø4.6 × 5.7 mm, in the post's free end;
   - `screw head of post …`: Ø4.5 × 1.85 mm under the board. 3DJake's page for the CNC Kitchen M2.5 ISO 14580 screw gives "Kopfdurchmesser (dk) 4,5 mm", "Kopfhöhe (k1) 1,85 mm", TX8.

   The heads clear the battery bay by 2.1 mm. The 6 mm screw engages 3.4 mm of thread through the 1.6 mm board and the 1 mm washer. The checks and the 3D view see them. The BOM gives the screws' dimensions and price, and the assembly guide adds a TX8 driver.
3. **Inserts go in at 245 °C.** CNC Kitchen: "Set your soldering iron 10 to 20 °C above your printing temperature … for PETG 245 °C", and "melt the inserts only to about 90% of the depth", then press flush with a flat tool. The guide said 220 °C, which is a PLA setting.
4. **The tap icon's skin is 0.36 mm.** The icon was pocketed to 0.4 mm from inside and debossed 0.2 mm from outside over the same outline, which left 0.2 mm: one or two layers. A Bambu forum thread notes "PETG is significantly more opaque than most TPU", so the skin can't be thick either. It is now three 0.12 mm layers under the deboss. A check measures it in the printed mesh.
5. **A lead-in on the top shell's lip.** A 0.4 × 45° chamfer runs round the inner lip at the parting line. The top shell prints with its seam edge on the bed, so the first layers squash inward and would catch the tongue. Prusa's guidance is that printers are "accurate to at least 0.2 mm".
6. **The TPU foot is glued with a flexible contact adhesive, not CA.** A Prusa forum user reported that with superglue "the TPU separated from the PETG"; others reported flexible adhesives holding. The shop for the adhesive is not confirmed yet.
7. **A test coupon is printed first** (`test_coupons.stl`, BOM P11, assembly step 1.0). Each row has the design value in the middle:
   - insert bosses Ø3.8 / 4.0 / 4.2 mm;
   - tongue slots for a 0.10 / 0.15 / 0.20 mm fit, with a loose tongue printed upright;
   - icon skins 0.24 / 0.36 / 0.48 mm;
   - switch-cap sockets 2.0–2.3 mm.

   The community advice on fits splits: about 0.15–0.2 mm on tuned printers, and Prusa's "at least 0.3 mm of clearance for movable parts". So the 0.15 mm fit stays until a coupon says otherwise. The coupon is generated from the same constants as the parts.
8. **Print notes:**
   - **Nozzle:** 240–245 °C for PETG. CNC Kitchen measured layer adhesion strongest there; upright PETG "only reaches about 60% of the base material strength".
   - **Supports:** PETG fuses to PETG supports, so start with a top Z distance of 2 layers and 2 interface layers. The Bambu forum doesn't agree on PLA as an interface material, so the notes don't ask for it.

## Not changed, and why

- **Tongue root fillet.** It is suggested for snap cantilevers. Our tongue's root sits where the top shell's lip lands, so a fillet would need a matching relief. The lead-in chamfer and the 50-cycle test (B-46) come first.
- **ASA instead of PETG.** CNC Kitchen measured about twice PETG's notched impact strength (18 vs 8.6 kJ/m²). But ASA needs an enclosed printer, which a kit can't assume.
- **The eye cradle's pins** stay as in ADR-0016.

## Checks

New in `review_geometry.py`:
- **Heat-set insert bosses:** bore, wall, depth and thread engagement against the datasheet.
- **Tap icon skin:** measured with a ray through the printed mesh.
- **Lead-in on the top shell's lip.**
- **Keep-outs and fasteners clear of parts:** extends ADR-0016's keep-out check to the inserts and screw heads.