# 1:1 router-bit case — two PLA parts, sized for FDM, not for CAD

Same product as before (Cleartec Grip Pak `GRP`): a translucent square tube with a closed
end, and an orange plug that carries three external ribs and grips the bit's shank.
Two printed parts, no cap, no threads, 3.175 mm shank.

What changed is that **every fit is now dimensioned as it will print**, not as it is drawn.
The first printed version failed twice and both failures are traceable to numbers I got wrong.

## The two failures, and the numbers behind them

**1. "the bit did not fit in"**

| | drawn | printed | result |
| --- | --- | --- | --- |
| mouth | 3.175 | ~3.04 | the shank could not start |
| grip throat | 3.050 | ~2.92 | 0.13 mm/side of interference before the leaves were even engaged |

A 3.175 hole for a 3.175 shank is not a fit, it is a line-to-line collision. Holes print
small (this print says ~0.13 small), so the hole was smaller than the shank at the mouth —
the one place that must be bigger.

**2. "only the first part fit in, and it takes too much strength"**

| | drawn | printed | result |
| --- | --- | --- | --- |
| rib crest vs bore | 11.10 in 11.00 → 0.05/side | ~0.11/side | heavy drag the moment the first rib entered |
| detent crest | 10.85 → 0.125/side on the rib | ~0.19/side | 0.22/side total at every detent, × 6 |

A square plug in a square bore has nowhere to go. The wall can bulge a little, but 0.22 mm/side
of forced interference is past PLA's elastic range: the first rib went in, the second seized.

## Print budget used for every dimension

Measured from that print — the only number in this file I did not choose:

```
hole  -0.13 mm   (a bore prints small)
outer +0.05 mm   (a boss prints big)
```

Every fit below is computed with those two offsets, printed in the run log, and stored in
`report.json` as `tolerance_budget`. `+` is interference, `-` is clearance, all per side.

| fit | drawn | as printed | target | |
| --- | --- | --- | --- | --- |
| socket entry 3.60 / shank 3.175 | -0.213 | **-0.148 clear** | ≤ -0.10 | the bit always starts |
| socket grip 3.12 / shank | +0.028 | **+0.093 press** | 0.05–0.13 | the leaves hold it |
| core 10.70 / bore 11.00 | -0.150 | **-0.060 clear** | ≤ -0.04 | the core never drags |
| rib 1 — 10.78 | -0.110 | **-0.020 clear** | -0.05–0.01 | entry is free |
| rib 2 — 10.85 | -0.075 | **+0.015 press** | 0.00–0.05 | you feel the bore |
| rib 3 — 10.90 | -0.050 | **+0.040 press** | 0.03–0.06 | firm, holds the depth |
| flange 15.00 / tube 13.20 | +0.900 | **0.90 lip** | ≥ 0.70 | there is something to pull |

## V6 — it went in, then would not come out

The second print fit: the shank entered and the plug pushed home. Withdrawing it took real effort,
and the cause was not the interference. The flange was **12.4 across a 13.2 tube and 3.6 tall** — inside
the tube's own silhouette, so once the plug was home there was no lip to get a finger behind.

So the handle changed, not the fit. Loosening a press to make a part easier to handle buys a part
that falls out.

- flange **15.00** across — **0.90 mm/side** proud of the tube OD
- **5.2** tall, with a **0.5** mm finger waist round the middle and a 45° × 1.0 underside bevel
- ribs narrowed 2.20 → **1.80** with 0.45 chamfers, so the break-out load sits slightly under the
  push-in load; the ladder's presses came down with it (10.85 / 10.90)

The bevel cost a round trip: cut as a lofted square subtraction it ate the flange's whole bottom
1.0 mm, because the loft was wider than the part at every height. `chamfer()` on the flange's bottom
edges removes 28.1 mm³ and leaves 13.1 at the edge, 15.0 above it.

## Mechanical changes that follow from the budget

**The six detents are gone.** At 0.1 mm engagement a detent sits inside FDM's own error band:
drawn shallow it vanishes, drawn deep it seizes — which is exactly what happened. What is
printable is a **ladder**: three ribs at 3.0 mm pitch whose crests climb 10.78 → 10.85 → 10.90,
so each rib a little firmer than the last. The plug stops wherever that friction balances, which
is a stepped friction fit with continuous depth (86.2–92.2 mm) instead of six fixed clicks.
One **seal ring** (crest 10.98, 1.2 mm above the rim) stays: drawn it is 0.01 mm/side — visually
flush — but printed it closes to 0.06/side on the last rib, so it becomes the click at the rim
without ever being able to jam.

**The socket is a funnel, not a hole.** 4.60 at the face → 3.60 straight entry (2.5 mm) →
3.12 throat (7.0 mm) → 5.2 relief. Printed, the entry is 3.47: the shank drops in with
0.15 mm/side to spare even if the printer runs fat, and it is the throat that grips. Two 0.90 mm
slits cut from 0.9 mm above the last rib to the face make four leaves 11.3 mm long.

**Nothing else touches.** The core is 10.70 in an 11.00 bore, so only the ribs ever contact the
wall — no long 30 mm press, no cumulative drag. The four corner vents run past every rib.

The tube's far end is a 2.0 mm ceiling printed with the tube (no cap), with a 1.0 mm × 30°
lead-in chamfer at the rim so the first rib starts on a ramp instead of a knife edge.

## Checks

49 measured checks, all passing, on the exported solids: closed end, two printed parts only,
rib crests measured at their section, throat measured under the shank, leaves counted, slits
clear of the ribs, vents open with a control probe on a flat so the check cannot pass vacuously,
bit tip clear of the ceiling, cutter clear of the tube. STEP round trip re-imports each part and
compares volume at 0 ppm.

`report.json` carries `printer_budget`, `tolerance_budget`, `params`, `volumes_mm3`, `checks`.

## Print

| | |
| --- | --- |
| parts | tube 5.35 g, plug 4.56 g — 9.9 g total |
| orientation | tube upright, rim down (bore stays round, ceiling bridges 11 mm) · plug flange down |
| supports | none |
| walls | 3 perimeters, 0.2 mm layers, 0.4 nozzle |
| material | PLA |

Elephant-foot matters here: a 0.2 mm skirt bulge at the tube's rim narrows the bore right where
the first rib enters. If the plug is still tight, drop `RIB_CREST[2]` by 0.05 in `bit_case.py`
and re-run — the budget table prints the new printed value before you slice.

## Files

- `bit_case.py` — the model, the budget, the 49 checks, the exports
- `proofs.py` — 2D sections rendered from the exported STEPs
- `out/bit_case_tube.step`, `out/bit_case_plug.step`, `out/bit_case_assembly.step`, `out/reference_bit.step`
- `out/section_axial.png`, `out/details.png`, `out/sections_plan.png`, `out/telescope.png`
- `out/report.json`
