typewriter/hardware/case/README.md

Enclosure — typewriter body (concept)

Part of Typoena — the distraction-free DIY writing machine. This page covers the enclosure only; the project root README covers the whole appliance (hardware, software stack, roadmap).

A 3D-printable case for Typoena. The e-paper strip sits on a reclined deck where a typewriter's sheet of paper would be; the keyboard you bring rests in front. There is no platen part (it complicates the print) — the rounded back-top edge is a subtle roll that nods to one for free.

Status: v0 concept, not yet printed. Outer form and the screen-retention / board-mounting strategy are worked out and render cleanly. Board hole positions and port offsets are placeholders marked << MEASURE >> in the .scad — confirm them against the real board before printing a final.

Files

File	What
typoena-case.scad	The parametric model. All dimensions live at the top.
concept.html	Dimensioned side/front/top drawing (open in a browser).
renders/	PNG previews (regenerated by the commands below).

Render / preview

Needs OpenSCAD. Open typoena-case.scad in the GUI and flip the show variable, or from the CLI:

cd hardware/case
# assembled, coloured, screen ghosted in
openscad -o renders/assembled.png --imgsize=1100,825 --colorscheme=Tomorrow \
  --camera=0,0,0,62,0,22,0 --viewall --autocenter \
  -D 'show="assembled"' typoena-case.scad

# export a printable part to STL (body | bracket | baseplate)
openscad -o body.stl -D 'show="body"' typoena-case.scad

show accepts assembled · body · bracket · baseplate · print_plate · section (vertical cut) · plan (exploded horizontal) · plan_up / plan_down (each half on its own).

Dimensions

From the datasheets, baked into the model:

• Panel — GDEY0579T93: glass outline 150.92 × 56.94 × 1.0 mm, active area 139.00 × 47.74 mm, pitch 0.1755 mm. Strip aspect ~2.9:1.
• Board — ESP32-S3-DevKitC-1: ~70 × 28 mm, USB-C ×2 + reset/boot on one short edge (that edge faces the back wall).
• Body (default): 176 W × 104 D, 24 mm front → 58 mm back, deck reclined ~21°. Walls 2.4 mm, deck 2.6 mm, corner radius 8 mm.

The deck angle is the one knob worth tuning first — see below.

How the hardware goes in — glueless

The whole design avoids glue on the fragile 1 mm glass and keeps every part serviceable.

Screen (bezel lip + foam + screwed bracket)

 front face                     the sandwich, front → back:
 ┌───────────────┐              1. deck BEZEL LIP  (overlaps ~4–5 mm of the
 │  ┌─────────┐  │                 glass's inactive border only, never the
 │  │ active  │  │ ← lip           active area — lip_t = 1.4 mm of material)
 │  └─────────┘  │              2. GLASS drops into the recess from behind;
 └───────────────┘                 the recess walls locate it in X/Y
                                 3. FOAM gasket (non-adhesive closed-cell,
   [lip][glass][foam][bracket]      foam_t ≈ 1 mm) spreads the clamp load
        ↑ screwed to 4 bosses       so you never point-crack the glass
                                 4. printed BRACKET, screwed to 4 bosses,
                                    presses the stack forward

• The through-aperture is a hair larger than the active area and stays inside the glass-minus-lip envelope, so the lip covers only dead border.
• The recess opens into the cavity, and an internal FPC clearance on the left (the user's left as they face the screen) — kept below the bezel lip so it's invisible from outside — lets the flex fold back to the DESPI-C579 breakout. Nothing breaks the external bezel.
• Foam does three jobs: cushions the glass, takes up print tolerance (±0.2–0.5 mm), and removes any need for adhesive. Cut it from a plain EVA/ PORON sheet — no sticky backing.
• Alternative if you want no screws: replace the bracket with printed cantilever clips. It works, but clips point-load the glass edge; the foam+bracket route is gentler and I'd default to it.

The lip alone can't hold the glass — it only stops it falling out the front. The glass is trapped at both edges between the front lip and the rear foam+bracket; the bracket is what stops it dropping into the cavity (show="section"):

Boards (the baseplate is the chassis)

Mount everything to the baseplate on the bench, then drop it in and close from below — far easier than fishing screws inside a shell.

• ESP32 + DESPI-C579 sit on printed standoffs (M2.5 self-tap). Positions in esp_holes / brk_holes are placeholders — set them to your board's holes. No mounting holes on your board rev? Switch to slide-in edge rails.
• The DESPI-C579 breakout sits in the cavity on the left, right under the FPC exit; short SPI jumpers (MOSI/SCLK/CS/DC/RST/BUSY + 3V3/GND) run across to the ESP32. Keep that left channel clear.
• A reset button — a small momentary switch, soldered to the board's EN and GND header pins — mounts through a hole in the back wall, out past the µSD slot so it's never hit while typing (rst_* params, << MEASURE >>). This is our own switch, not the DevKitC's on-board buttons: those are top-actuated and buried once the board lies flat. BOOT is deliberately omitted — on the S3, auto-download (UART-bridge DTR/RTS or the native USB-Serial-JTAG) makes it recovery-only, not worth a fat-fingerable button on a writing appliance.
• The baseplate screws up into 4 corner posts in the shell.
• A cable relief notch at the back lets the keyboard's USB-C cable exit and route around to the front.

Exploded horizontal section (show="plan") — the deck/screen half lifts off the cavity half so you see both at once: ESP32 standoffs centre, DESPI-C579 standoffs left under the FPC exit, the two back corner posts, and the three port openings in the back wall.

Or inspect each half alone — plan_up (the deck/lid, shown from below) and plan_down (the cavity):

Assembly order

1. Lay glass into the deck recess (from inside), add the foam gasket, screw the bracket down onto the 4 bosses.
2. Screw ESP32 + breakout to the baseplate standoffs.
3. Connect the FPC (screen → breakout) through the slot.
4. Screw the baseplate up into the corner posts.

Tune first

• Hb (back height) → deck angle. 18–22° is typewriter-shallow; raise Hb toward ~28–35° if the screen reads too edge-on when you're sitting close.
• << MEASURE >> items: esp_holes, brk_holes, port_x, port_z, rst_x/rst_z/rst_d (reset button), active_off_x/y (the panel's active area sits off-centre from the glass).

Print notes

• Material: PLA/PETG. Print the body in matte indigo (#130f40), the bracket/base in cream or brass — two-tone reads unmistakably "typewriter" for the price of a filament swap.
• Make the engrave read: on a body this dark the recessed TYPOENA is near-invisible until you give it contrast — a swipe of paint pen in the recess, or a 3–4 layer filament swap across the nameplate band mid-print.
• Shell, not solid: 2.4 mm walls + open bottom keep material low despite the chunky body form.
• Orientation: body deck-up (or on its back) needs little/no support; the bezel lip is a small overhang. Bracket and baseplate print flat.

Nameplate font

The TYPOENA engrave on the deck (recessed, faces the user) is set via the name_font parameter. Current pick: Monaspace Krypton (GitHub's mechanical mono). OpenSCAD only renders fonts installed on the system:

# current: Monaspace Krypton (installs the whole family)
brew install --cask font-monaspace

# alternative tried: Cutive Mono (typewriter slab-serif)
curl -sL -o ~/Library/Fonts/CutiveMono-Regular.ttf \
  https://github.com/google/fonts/raw/main/ofl/cutivemono/CutiveMono-Regular.ttf

fc-cache -f    # so the OpenSCAD CLI picks either up

To audition other faces: Google Fonts, filtered to monospace and previewing the name.

Open questions / TODO

• Confirm the GDEY0579T93 active-area offset (FPC confirmed on the left edge); adjust active_off_x/y.
• Real board mounting-hole + port coordinates. A custom carrier PCB is planned — when it lands, re-fix esp_holes, port_x/z, and the reset rst_* to the PCB (and decide there whether to expose BOOT as a test pad).
• Optional hinged lid over the deck (portable-typewriter-case echo, protects the glass in a bag) — docs/hardware.md calls for one; not yet modelled.
• Decide feet: printed (modelled) vs. stick-on rubber bumpers.