# Chaitan: an Adom molecule LED nameplate

Chaitan's name, **etched out of the soldermask over the ground plane**, lit by
14 side-view LEDs that fire inward across the board, powered from any
USB-C phone charger, and mountable in a probing workcell by four large machine
pins on the 32 mm base-scaffold grid.

## The one idea

**The name is the ground plane.** A rectangular GND copper pour covers the top
layer, and the soldermask is opened in the shape of the letters so bare plated
copper shows through. There is deliberately **no second ground path** (no
bottom ground pour), so every milliamp those 14 LEDs draw physically
crosses the letters on its way back to the connector.

So the line you get to say to Chaitan, *"your name is part of the electrical
path that lights up the LEDs"*, is a statement about the netlist, not a
slogan. It is enforced by construction: the ground pour is sized to swallow
every ground pad on the board, so ground is bonded by geometry and the
autorouter is never given the chance to route a shortcut around the letters.

## The board

| | |
|---|---|
| Board | 236 x 76 mm, 2 layer, 4 mm corner radius |
| Pin box | **224 x 64 mm** = 7 x 2 on the 32 mm grid |
| Name | 194.0 x 40 mm, Familjen Grotesk Bold |
| Mask openings | 362 letter polygons |
| GND plate | 229.6 x 69.6 mm, top layer |
| LEDs | 14 x Würth WL-SMSW `155060GS75300`, 515 nm side-view |
| Current | 168 mA @ 5 V (14 x 12 mA) |
| Parts / traces / vias | 38 / 60 / 36 |

### Molecule compliance

Per [molecule-design-guide](https://wiki.adom.inc/adom/molecule-design-guide)
and [base-scaffold-system](https://wiki.adom.inc/adom/base-scaffold-system):

- A **large machine pin at every corner** (3.6 mm body press-fit into a 3.45 mm
  plated hole, 0.15 mm diametral interference).
- The pin box is **224 x 64 mm**, both exact multiples of 32 mm (the base
  scaffold's large-contact pitch), so it drops straight onto 8 x 3
  contact positions of the 18 x 18 field across the 576 x 576 mm workspace.
- **Minimal wings:** the board stops 6 mm outside each corner pin. That is as
  tight as the pin allows: a 5.2 mm annular ring eats 2.6 mm of it, leaving
  3.4 mm of solid laminate, about 2x board thickness, which is what a press-fit
  needs so insertion force does not crack the edge out.
- Corner pins carry power, not just mechanics: **MP1/MP4 are GND, MP2/MP3 are
  +5 V**, so the plate can run off a workcell control panel instead of USB-C.
  The GND pins land inside the plate, which means the scaffold's ground return
  also passes through the name.

## Why side-view LEDs

A normal chip LED is a top-view emitter: it fires straight up, away from the
board, and the copper you are trying to light sees nothing. This board uses a
true **side-view** part. The die sits on its side behind a window in the long
face of the package, so the beam leaves *parallel* to the board about 0.5 mm
above the soldermask. Ringed around the edge facing inward, the light rakes
across the plate at a grazing angle and washes the exposed-copper letters
instead of escaping to the sky.

Worth knowing if you go sourcing: Kingbright's `APTR3216` family is
*reverse-mount* (it fires down through a hole in the PCB) and Broadcom's
`HSMx-C265` is bottom-entry. Both come back in a "right angle SMD LED" search
and neither does this job.

- 1.6 x 0.8 x 1.1 mm; 2 pads 0.6 x 0.8 mm on 1.4 mm pitch
- 515 nm peak / 525 nm dominant, 260 to 560 mcd @ 20 mA, 140 deg 2θ½
- VF 3.2 V typ; run here at 12 mA through 150 R

## Power

USB-C VBUS (5 V) -> 150 R -> LED (3.2 V) = 12 mA each, **168 mA total**, well
inside the 500 mA a charger gives an unnegotiated sink, so any phone brick
works. `R_CC1`/`R_CC2` are the mandatory Rd pulldowns. Without them on *both*
CC lines a USB-C source never enables VBUS at all and the board is dead.

Layer plan is inverted from the usual convention on purpose:

- **Top** = GND, one rectangular pour. The name is cut out of it.
- **Bottom** = +5 V plane.

Everything is placed on the **top side only** (the line does single-sided
pick-and-place), so each LED string runs LED -> 150 R -> a via down to the +5 V
plane, while the cathode bonds straight into the plate.

## BOM

| Ref | Qty | Part | Package |
|---|---:|---|---|
| D* | 14 | Würth WL-SMSW `155060GS75300` side-view green 515 nm | 0603 side-view |
| R1 to R14 | 14 | 150 R ±1 % `CR0402-FX-1500GLF` | 0402 |
| R_CC1/2 | 2 | 4.7 k ±1 % `CR0402-FX-4701GLF` | 0402 |
| C1 | 1 | 10 µF X5R `0603X106K100CT` | 0603 |
| C2 | 1 | 100 nF X7R `0402B104J160CT` | 0402 |
| J1 | 1 | `TYPE-C-31-M-12` USB-C (JLC C165948) | SMD + 4 plated anchors |
| MP1 to MP4 | 4 | `MachinePinLargeStandard` | 3.45 mm plated hole |

Passives are all Adom Basic Parts. The LED is Mouser/DigiKey (~$0.40).

## Fab notes

- **Specify ENIG, not HASL.** The letters are 194 x 40 mm of deliberately
  bare copper and are the entire point of the board. HASL leaves them lumpy,
  ENIG leaves them mirror flat.
- Dark soldermask maximises contrast against the exposed letters.
- The four 3.45 mm machine-pin holes are an **interference fit** and are meant
  to be undersized. Do not let a fab house "correct" them to a clearance fit.
- Tented/plugged vias: there are vias whose rings lap onto LED resistor pads.

## Verifying the mask opens

Do not trust the render. Read the gerber. `F_Mask.gbr` is negative polarity
(`%TF.FilePolarity,Negative*%`), so what is *drawn* is where mask is *removed*:

- **362 region polygons**, bounding box **194.04 x 40.00 mm**, the letters
- the plate rectangle **absent** from F_Mask (present would mean inverted)
- **zero** letter regions in `F_Paste.gbr`. The letters are decoration, not
  solder joints, and must never be stencilled

## Reproducing it

Built with [adom-tsci](https://wiki.adom.inc/adom/adom-tsci) from a parametric
tscircuit source. One command per name:

```bash
scripts/make-nameplate.sh "Chaitan"
```

Board size, pin box, LED count and silkscreen are all derived from the name's
ink box, so a new name is a one-liner. The source bundle attached to this page
has everything.