bc250-console/WIRING.md

# BC-250 Console — Power Wiring Guide

Wiring a **Mean Well LRS-350-12** (12 V / 29 A / 350 W) to the BC-250's
**J1000** PCIe 8-pin power input, with an AC-side power switch.

> ⚠️ **The AC side of this build is mains voltage. It can kill you.**
> Work with the cord unplugged, never trust a switch to isolate, keep AC
> terminals covered when done (the LRS-350 ships with a terminal cover —
> use it), and if you are not comfortable with mains wiring, have someone
> qualified do §2.

## 1. Parts

| Item | Notes |
|---|---|
| Mean Well LRS-350-12 | Screw-terminal PSU: `L`, `N`, `FG` (AC side) + 3× `+V`, 3× `−V` (DC side) |
| PCIe 8-pin pigtail, 16 AWG | The **female receptacle** end mates with J1000. Must be **PCIe**, not EPS/CPU — the keying differs and EPS swaps 12 V/GND. Never force a connector. |
| Fused + switched IEC C14 inlet module | The cleanest "power switch": cord socket, rocker, and fuse in one panel part. Rated ≥ 250 VAC / 6 A. Fit a **5 A slow-blow** fuse (PSU pulls ~4 A max at 115 V, half that at 230 V). |
| Grounded (3-wire) IEC mains cord | Earth is not optional — the PSU chassis and `FG` must be earthed. |
| Ring or fork crimp terminals + ferrules | For the PSU screw terminals. Bare stranded wire under screws loosens over time. |
| Heat shrink, zip ties | Insulate every AC joint; strain-relieve everything. |
| Multimeter | Non-negotiable for §4. |

## 2. AC side (mains → PSU)

```
 IEC inlet module                          LRS-350-12
┌───────────────────┐
│ L ──── FUSE ──── SWITCH ────────────────► L
│ N ──────────────────────────────────────► N
│ E (earth) ───────────────────────────┬──► FG (ground symbol)
└───────────────────┘                  └──► case earth stud (if separate)
```

- Fuse and switch go in the **Live** line only (a double-pole switch that
  also breaks N is even better; never switch only N).
- Insulate the inlet's spade terminals with heat shrink or boots — they're
  live whenever the cord is plugged in, switch position irrelevant.
- Earth goes straight through, unswitched and unfused, to `FG`.
- Fit the PSU's plastic terminal cover after wiring.

## 3. DC side (PSU → J1000)

J1000 pinout (from the BC-250 hardware docs):

```
[ GND GND GND GND ]      3 × 12 V pins, 5 × GND pins
[ GND 12V 12V 12V ]
```

- Run **each of the three 12 V wires to its own `+V` terminal** and the
  ground wires to the `−V` terminals (bundle grounds 2-per-terminal if the
  pigtail has five). The LRS-350 has three terminal pairs precisely so you
  can split high current — at the board's ~220 W TDP that's ~18 A total,
  ~6 A per 16 AWG wire. Comfortable.
- Keep the run short — under ~1 m limits voltage drop.
- Ferrules or ring terminals on every wire; screw them down hard and tug-test.
- Don't extend or thin the wires: 16 AWG minimum, as delivered.
- No inline DC fuse needed — the LRS-350 has over-current, over-voltage,
  and short-circuit protection built in.

## 4. Verification — BEFORE the board ever sees power

With the **pigtail connected to the PSU but NOT to the board**:

1. **Continuity (AC off, cord unplugged):** beep out each connector pin to
   its PSU terminal. Confirm the three pins on J1000's `12V` positions trace
   to `+V` and all others to `−V`. Match the pinout above against the actual
   connector — identify which row is which by following the wires, not by
   assuming latch orientation.
2. **Voltage (AC on, board disconnected):** switch on. Measure across the
   connector's 12 V/GND pins: expect **12.0–12.2 V** (trim with the PSU's
   `+V ADJ` pot if needed). Then probe **every pin pair** — any 12 V on a
   pin that maps to GND means STOP and re-check.
3. Switch off, plug into J1000 (it only seats one way — if it doesn't seat,
   it's the wrong connector, do not force), then proceed to §5.

## 5. The power switch (how the console turns on and off)

The BC-250 has no case power button — it was a rack blade. Instead:

**`AUTO_PWRON1` jumper → pins 1-2** (factory default; also matches the BIOS
`AUTO_PWRON` setting from the install checklist). The board now boots
automatically the moment 12 V appears — so **the rocker on the IEC inlet *is*
the console's power switch**, exactly like a TV:

| Action | Result |
|---|---|
| Rocker ON | PSU powers up → board auto-boots → GNOME login |
| Shut down from GNOME / `sudo systemctl poweroff` | Board off; PSU still energized (standby). Stays off until AC is cycled. |
| Rocker OFF → ON | Full power cycle → boots fresh |
| Rocker OFF | Everything dead. Fine to do after a soft shutdown; the OS is btrfs and the image is atomic, but teach the kid "menu shutdown first" anyway. |

Notes:
- Jumper position 2-3 instead makes the board wait for a power-button press —
  useless in a custom case (there's no accessible button; don't use this mode).
- `J2001`'s `PGD` pin powers the board **off** when shorted to ground — that's
  a blade-chassis signal, not a power button. Leave J2000/J2001 unconnected.
- Leave the console rocker-ON overnight sometimes: automatic OS/app updates
  install while it's powered.

## 6. Fans (they also need 12 V)

- `CPU_FAN1` is a standard 4-pin PWM header: `[ PWM Tach 12V GND ]` — BIOS
  fan curve applies. Put the primary heatsink fan here.
- More fans: a PWM splitter on `CPU_FAN1` (fine for 2–3 standard fans), or
  run extras directly off a spare `+V`/`−V` pair (they'll run full speed,
  always — crude but effective for case airflow).
- Don't forget airflow over the **backplate** — the GDDR6 underneath runs
  hot and has no temperature sensor.

## 7. First power-on

1. J1000 seated, fans connected, DisplayPort to TV, keyboard for BIOS.
2. Rocker ON. Expect: fans spin, DisplayPort shows POST within ~30 s.
3. No signs of life in 30 s → rocker OFF, re-do §4. Do not keep retrying.
4. First stop: BIOS — confirm the firmware/VRAM settings from the install
   checklist, then boot the installer USB.