# Uncle Magic's Liquid Turntable: Hardware Overview

How the turntable is built: what the parts are, how they are connected, how power flows, and what
protects what. For operating it, see the [User Guide](USER_GUIDE.md). For exact pin-by-pin detail,
see the [README](files/README.md); for the reasoning behind design decisions, the
[detailed README](files/README_DETAILED.md); for their history, the
[changelog](files/CHANGELOG.md); the KiCad files in `kicad_mainboard/` are the authority on
connections.

## Contents

1. [System overview](#1-system-overview)
2. [The mainboard](#2-the-mainboard)
3. [Power](#3-power)
4. [Signals and control](#4-signals-and-control)
5. [Connector reference](#5-connector-reference)
6. [The LED ring board](#6-the-led-ring-board)
7. [Motor and mechanics](#7-motor-and-mechanics)
8. [Current, heat and protection notes](#8-current-heat-and-protection-notes)
9. [Manufacturing and files](#9-manufacturing-and-files)
10. [Verification status and limits](#10-verification-status-and-limits)

---

## 1. System overview

The turntable is built from **two printed circuit boards, a stepper motor and the panel controls**,
fed by one external 5 V / 12 V supply.

| Part | What it is |
|---|---|
| **Mainboard** | 100 × 100 mm board that carries three plug-in modules, the motor driver, the power protection and every connector. It is the brain of the unit. |
| **LED ring board** | A second board in this repository: 178 addressable LEDs on an 8 inch (203 mm) board with a centre hole. It is fixed near the top of the case and **does not turn** with the platter (section 6). |
| **Stepper motor** | A NEMA 17 bipolar stepper, joined directly to the platter by a flexible shaft coupler. No gearbox. |
| **Panel controls** | Four rotary encoders (with push buttons) and a tap button, wired to the mainboard through small headers. There is no separate panel board. |
| **Power supply** | One 5 V / 12 V supply (Mean Well RD-85A): 12 V for the motor, fan and electronics, 5 V for the LEDs. Outside the board. |

**Three computers share the mainboard**, all running at 3.3 V:

| Module | Role |
|---|---|
| **Waveshare ESP32-S3-ETH** | The main controller. Runs this project's firmware: tempo, speed control, web pages, OSC, encoders. Has wired Ethernet (W5500) and Wi-Fi. |
| **BigTreeTech TMC2209** (stepper driver module) | Drives the motor coils. Configured over a one-wire serial link. |

All three are **socketed**: they plug into female headers, so each can be replaced or reflashed.

### Power flow

![Power flow: 12 V and 5 V supplies through fuses and protection to the rails and loads](diagrams/power-flow.svg)

### Signal flow

![Signal flow: main controller, panel, motor driver, expander and lighting](diagrams/signal-flow.svg)

### Platter support and bearings

The platter is a 10 in (254 mm) by 1/4 in acrylic disc, about 380 g, turning at 17 RPM at most. The motor only drives it: the platter's weight is carried on
three inexpensive flanged ball units pressed into the lid (flange on the lid top, ball about 3 mm above it), and the drive is a sliding joint. A 5 mm D-flat stub
is clamped in the flexible coupler's top end, and the platter's hub slides on it with at least 10 mm engaged, so the shaft's height never holds the platter up
(it can be about 4 mm too high or 8 mm too low). The enclosure notes (`enclosure/MOTOR_MOUNT.md`) have the dimensions and the sketch; the ball units, the
hub and the stub are untested.

## 2. The mainboard

![Mainboard, top view](files/mainboard/Mainboard_render_top.png)

| | |
|---|---|
| Size | 100 × 100 mm, two copper layers, 1.6 mm, **1 oz copper** (JLCPCB's default; see section 8). |
| Mounting | Four M3 holes (3.2 mm) at the corners. |
| Design tool | KiCad 10 (`kicad_mainboard/Mainboard.kicad_pcb`) |
| Schematic | A one-page index plus six function sheets, below. PDF: [`Mainboard_schematic.pdf`](files/mainboard/Mainboard_schematic.pdf) |

### Schematic sheets

| Sheet | Contains |
|---|---|
| **Power** | The two power inputs `J-PWR-5V` and `J-PWR-12V`, the 12 V resettable fuse `F1`, diode `D1`, TVS diode `D5`, bulk capacitors and the reverse-polarity FET `Q5`. There is no 3.3 V regulator on the board: the ESP32-S3-ETH module makes `+3V3`. |
| **Controller** | Sockets for the ESP32-S3-ETH (`J1-H1`, `J1-H2`; `VSYS` fed from `+5V_LED_LOGIC`, `+3V3` taken from the module) and the boot-strap pull-up `R23`. |
| **Motor** | Sockets for the TMC2209 (`J2-MODE`, `J2-MOTOR`), the UART resistor, and the motor connector `J3`. |
| **LED** | The ring data and clock buffers `U3` and `U5` (3.3 V to 5 V), their supply diode `D33`, the ring signal connector `J-LED-SIG` (which also brings in the ring's temperature sensor: `D34`, `R65`, `R66`, `C16` to GPIO3) and power connector `J5` with its TVS `D32`, and the fan header `J-FAN` with its PWM inverter `Q6`. |
| **Panel IO** | The TCA9555 I/O expander `U1`, its I2C pull-ups, and every encoder and button header. |
| **Test points** | `TP1` to `TP13`, the probe pads listed in section 3. |

### Module orientation

Each module fits its socket only one way up. The board is silkscreened to match, but a module
plugged in backwards would put power on the wrong pins.

| Module | How it goes in |
|---|---|
| ESP32-S3-ETH | **RJ45 side up**, with the RJ45 and USB-C end at the top edge of the board. The 72.8 × 21 mm module overhangs the board edge by about 16 mm, which is where the network socket and USB port are reached. |
| TMC2209 | Rotated 180° from the usual drawing: EN and VM at the **bottom** (panel-header edge), EN column on the right, VM column on the left. Check the module's own silkscreen first; a reversed module will be destroyed. |

### Key parts

| Ref | Part | Job |
|---|---|---|
| `U3` | 74AHCT1G125 | Lifts the LED data signal from 3.3 V to 5 V |
| `U1` | TCA9555PWR (TSSOP-24) | I2C input/output expander (address 0x20), fixed 100 k pull-ups on its inputs (the firmware also drives an MCP23017, for older boards) |
| `D1` | SS34 Schottky (SMA) | Stops reverse 12 V and stops motor energy flowing back |
| `D5` | SMBJ15A | Clamps voltage spikes on the 12 V rail (about 24 V at its rated surge current) |
| `D4` | PESD3V3S1UL | Static-discharge protection on 3.3 V |
| `Q5` | SP30P03NK (P-channel FET) | Blocks a reversed 5 V feed on the ring line |
| `D32` | SMBJ5.0A | Clamps spikes on the ring's 5 V |
| `D33` | SS34 Schottky | Feeds `+5V_LED_LOGIC` (the two LED buffers and the module's `VSYS`, so all of the 3.3 V logic) from the 5 V input and blocks a reversed supply |
| `D34`, `R65`, `R66`, `C16` | PESD3V3S1UL, 10 k 1 %, 1 k, 100 nF | Ring temperature input: ESD clamp and pull-up at `J-LED-SIG` pin 5, then a 1 k / 100 nF filter into GPIO3 |
| `Q6` | 2N7002 | Inverts the fan PWM signal |
| `F1` | 2 A hold resettable fuse (PTC), 1812 | Soldered on; trips at about 3.5 A and resets once the fault is gone and it has cooled |
| `C6` | 100 µF 35 V (SMD, same part as `C8`/`C11`) | Bulk capacitor on the 5 V input; the ring has its own 1000 µF `CP1` at its connector |
| `C8`, `C11` | 2 × 100 µF, 35 V | Bulk capacitance for the motor driver |
| `R23` | 1 k | Holds the motor-driver enable pin high (driver off) during boot |
| `R26`, `R62` | 10 k | Keep the LED data and clock lines low while the ESP32 starts, so the ring doesn't flash |
| `U3`, `U5` | 74AHCT1G125 | Lift the ring's data (GPIO18) and clock (GPIO2) from 3.3 V to 5 V |
| `R63`, `R64` | 33 Ω | Source resistors at the buffer outputs, to damp the signal cable |
| `R27` | 4.7 k, 1206 | Bleeder across the 12 V capacitors so the rail drains after power-off |

## 3. Power

### Rails

| Rail | Source | Feeds | Protection |
|---|---|---|---|
| **+12V_MOTOR** | 12 V supply via `J-PWR-12V` | Motor driver supply; fan header `J-FAN` (about 0.1 A) | `F1` 2 A fuse; `D1`; 200 µF bulk; `D5` TVS; `R27` bleeder |
| **+3V3** | The ESP32-S3-ETH's own JW5060 buck, from `VSYS` = `+5V_LED_LOGIC` | ESP32-S3-ETH, TMC2209 logic, TCA9555, encoder pull-ups, the ring NTC pull-up | The buck limits its own current; `D4` is static protection only |
| **+5V_LED_RING** | 5 V supply via `J-PWR-5V` pins 1 and 3 | LED ring; through `D33` (from before `Q5`) the LED buffers `U3`/`U5` and the module's `VSYS`, so all of the 3.3 V logic | 7.5 A blade fuse in the harness at the supply; `Q5` reverse-polarity FET; `D32` TVS |

The three rails do not share a protection path: a fault on one cannot ride through into another.
The 12 V supply feeds only the motor driver and the fan. The 5 V supply feeds the LED ring and, through `D33`, the
logic: the two signal buffers and the module, whose own regulator makes `+3V3` for everything else.

### Design choices worth knowing

- **3.3 V comes from the ESP32-S3-ETH module itself.** On the module, `VSYS` feeds a JoulWatt JW5060
  buck (4-18 V in, 3 A, fixed output); its enable (`3V3_EN`) is pulled up to `VSYS` by the module's `R67`. The board feeds
  `VSYS` (`J1-H1` pin 2) from `+5V_LED_LOGIC`, about 4.7 V after `D33`, leaves `3V3_EN` (pin 4) open, and takes `+3V3`
  from the module's `3V3` pin (pin 5). This replaced the board's own 12 V to 3.3 V converter (`U2`, K7803-2000R3) and its
  capacitors `C12`-`C14`: about $2.40 to $3.40 less per board, the part with the least stock gone, and no 17.5 mm tall part.
- **The 3.3 V rail has headroom.** The ESP32, the W5500 and a few milliamps of logic run from it, well under 1 A against the buck's 3 A; the main board's Wi-Fi power is still capped at 15 dBm as a conservative setting.
- **USB-C powers the logic.** USB VBUS reaches `VSYS` through the module's 1N4148, so the module's USB-C can run the
  ESP32, the expander and the encoders on its own for flashing and setup (not the motor or the ring), and our 5 V never
  reaches a USB host. With the 5 V supply on as well, it wins (about 4.7 V against USB's 4.3 V after the 1N4148).
- **The logic needs the 5 V supply.** With 12 V on and 5 V off, the TMC2209 has motor power but no VIO; the datasheet holds
  its logic in reset with the bridges off below the VIO threshold. Check it once on the bench.
- **Switch-off behaviour.** Without `R27` the 12 V capacitors would linger at a few volts and
  leave the processors in brown-out. With it the rail is under 1 V in about 2.5 seconds.
- **Switch-on order.** 3.3 V comes up with the 5 V rail; 12 V may arrive at any time; the
  worst case is a few milliseconds of 3.3 V on the data buffer's input before it has power,
  limited to about 10 mA by `R18`.

### Power budget

| Rail | Typical | Worst case |
|---|---|---|
| 3.3 V logic | 0.3 to 0.45 A, Wi-Fi off | About 0.7 A, Wi-Fi on (2 A available) |
| 12 V motor and fan | The motor runs at 800 mA per phase; about 0.5 A from the supply | `F1` holds 2 A (less when hot) |
| 5 V ring | Depends on colour and brightness; the LED engine holds each LED to 36.6 mA and the ring to about 4.5 A | A single colour at full brightness about 3.2 A (18 mA × 178); white would be 9.6 A at the full current field (31) without the limits, which must never be lifted |

Mains side: two switch-mode supplies together draw about 100 to 120 W in steady state, with a
20 to 60 A inrush for a few milliseconds at switch-on. A slow-blow fuse and a switch rated at
10 A or more are recommended.

### Power connectors `J-PWR-5V` and `J-PWR-12V`

The two supplies come in on **two different connectors**, so a 12 V plug cannot be put into the 5 V
header.

**`J-PWR-5V`**: Molex Mini-Fit Jr, 4 positions (2 × 2). The 5 V is on two pins because of the ring current
(about 3.3 A each at full white, 2.3 A at the LED engine's limit).

| Pin | Net | Pin | Net |
|---|---|---|---|
| 1 | +5V | 2 | GND |
| 3 | +5V | 4 | GND |

**`J-PWR-12V`**: JST XH, 2 positions (`B2B-XH-A`, 3 A; plug housing XHP-2 with SXH-001T-P0.6 terminals, wire 22 to 28 AWG).
Pin 1 is +12 V and pin 2 is GND, as printed on the board. A different connector family from the 5 V input, so the two
cannot be swapped.

Both connect **ahead of the fuses**. Never plug or unplug either with power on.

### The supply (Mean Well RD-85A)

The chosen supply is the **Mean Well RD-85A**, one dual-output unit that gives both rails. The board needs 5 V for the LED ring only (about 6.5 A at full white, 7.3 A at the LEDs' maximum drive current, 4.5 A with the LED engine's current limit) and 12 V for
everything else (about 2.5 A worst case), about 70 W together. Prices below are from retailer listings on
2026-10-07 and were not verified; check stock and the datasheet before ordering.

| Option | 5 V | 12 V | Approx. price | Notes |
|---|---|---|---|---|
| **Mean Well RD-85A** (chosen) | 8 A rated, 10 A range | 4 A rated, 5 A range | $27-41 | One box, one mains connection. The only limit on the ranges is total power (88 W), and the board's worst case is about 70 W; the ring's full white (6.5 A) is inside the 8 A rating. |
| Alternative: Mean Well LRS-75-5 plus LRS-35-12 (two supplies) | 14 A | 3 A | about $30 | Full ring current with margin; two boxes and a split mains feed behind the single switch. |
| Alternative: Mean Well MD-100S (about 100 W dual output) | 10 A | 4.2 A | not found | Similar to the RD-85A with a little more headroom. |

**Do not use the RD-85B**: it is 5 V / 24 V. The board's motor rail is 12 V.

RD-85A points from Mean Well's specification (RD-85-SPEC, 2024-11-22):

- **No minimum load.** The spec only recommends loading 5 V above 10 % (0.8 A) to hold its +/-2 % tolerance. With the ring dark the
  5 V output may sit slightly high, inside its 4.75 to 5.5 V range. The ring does not mind.
- **Tolerance and ripple:** 5 V +/-2 %, 80 mVp-p; 12 V +/-5 % (11.4 to 12.6 V), 120 mVp-p. The 12 V rail is well inside `D5` and the TMC2209 limits.
- **Protection:** overload (110 to 150 % of rated power) and over-voltage on 5 V (5.75 to 6.75 V) both hiccup and recover by themselves.
  A wrong current limit on the **LEDs** tab will make the ring flicker, not latch the supply off.
- **Start-up and ride-through:** set-up 0.5 s (230 VAC) to 1.2 s (115 VAC), both rails together. Hold-up is only 18 ms at 115 VAC at
  full load, so a short mains dip will reset the processors.
- **Inrush:** 50 A cold start at 230 VAC. Use the slow-blow inlet fuse described in the README.
- **Heat:** 78 % efficient, about 24 W at full load, convection cooled, derating from 40 C ambient. Give it airflow; do not seal it in.
- **Size:** 159 x 97 x 38 mm.
- **Unverified:** the spec text was machine-extracted. Check the terminal numbering against the drawing before wiring.

### Wiring the RD-85A to `J-PWR-5V` and `J-PWR-12V`

The RD-85A has a 7-position screw terminal block: AC L, AC N, FG (earth), then the DC side with one common (COM), +V2 (12 V) and +V1 (5 V).
The two outputs share COM, which matches the board's single ground.

| RD-85A terminal | Goes to | Wire |
|---|---|---|
| +V1 (5 V) | the 7.5 A inline fuse, then `J-PWR-5V` pins 1 and 3 (two wires after the fuse, one per pin) | 14 or 16 AWG to the fuse, 16 AWG each after it |
| COM | `J-PWR-5V` pins 2 and 4 (one wire each) **and** `J-PWR-12V` pin 2 | 16 AWG for the 5 V pair, 22 AWG for the 12 V lead (or one 16 AWG per terminal screw) |
| +V2 (12 V) | `J-PWR-12V` pin 1 | 22 AWG |
| FG | mains earth, and the enclosure | 18 AWG |

- **5 V fuse:** one 7.5 A ATO/ATC blade fuse in a sealed inline holder (14 or 16 AWG leads), on a single red lead from
  +V1, as close to the terminal as it will go. After the fuse, split into the two 16 AWG wires to pins 1 and 3 (a lever
  connector rated 20 A or more, or a crimped splice). Use one fuse before the split, not one per wire: two in parallel act
  as a 15 A fuse. The normal load is about 4.6 A (about 60 % of the rating, so it won't age-trip). It blows on a short in
  the harness, the connectors or either board, protects the 16 AWG and 18 AWG wires, and caps each Mini-Fit pin at about
  3.75 A. It is not the ring's current limit (the LED engine is), and it does not see a 12 V wire crimped into the 5 V plug.
  About $1 to $2 for the holder, fuse and splice.
- Crimp the 5 V harness to Mini-Fit Jr terminals (5557 housing) and the 12 V lead to JST SXH-001T-P0.6 terminals (XHP-2 housing). Use ring or fork terminals at the supply's M3 screws, so no strands are exposed on a mains-adjacent block.
- Keep the 5 V and COM wires to the ring side short and twisted together; at 10 A, 100 mV of drop is only 10 mohm of wire and crimps.
- **Set the 5 V trim to about 5.1 V** with a meter on the supply terminals, then check at `J5` with the ring at full white. The trim range is 4.75 to 5.5 V.
- Run the 12 V and 5 V with the ring dark first: confirm 12 V on `J-PWR-12V` pin 1 before connecting the board.
- Slow-blow fuse at the inlet: the RD-85A draws 2.5 A (115 VAC) or 1.5 A (230 VAC) at full load, so use about T3.15A at 120 V or T2A at 230 V (a sizing estimate, not from the datasheet; the 50 A inrush rules out a fast fuse).
- Ground the supply's FG terminal to the enclosure and mains earth; the spec lists 0.5 kVAC output-to-FG isolation only.

### Harness build and test (before first connection)

`Q5` (a P-channel FET in the 5 V ring line) blocks a reversed 5 V feed, and the 5 V and 12 V
inputs are separate connectors of different families, so a 12 V plug cannot go into the 5 V header. The remaining mistake is a wire crimped into the wrong cavity of the 5 V plug (12 V on a 5 V pin), and nothing on the board catches it. `D32`, the SMBJ5.0A on the ring rail, conducts from 6.4 to 7.0 V and clamps at 9.2 V at its rated surge; a board crowbar was tried and dropped, because it could not clamp much lower without tripping on a healthy supply (see the CHANGELOG). A 12 V mis-wire from a supply that can deliver several amps would damage the ring and probably `D32`. So the harness is tested **before** it meets the board, every time one is made or modified.

**Colour code and labels.** Use the same colours at both ends and put a printed heat-shrink label on each wire:

| Net | Colour | Pins | Wire |
|---|---|---|---|
| +5V | Red | the fuse holder, then `J-PWR-5V` 1 and 3 | 14 or 16 AWG to the fuse, then 16 AWG each |
| GND (COM), 5 V side | Black | `J-PWR-5V` 2 and 4 | 16 AWG each |
| +12V | Yellow | `J-PWR-12V` 1 | 22 AWG |
| GND (COM), 12 V side | Black | `J-PWR-12V` 2 | 22 AWG |
| Mains earth (FG) | Green/yellow | not on these connectors | 18 AWG |

Never use red for anything but +5V. Keep the harness short and twist each supply wire with a ground.

**Indicator jig (optional, about 5 USD, an evening).** A small perfboard with a mating 2x2 Mini-Fit Jr header and a JST XH 2-pin
header, and four indicator circuits, each drawing only a few milliamps:

| Indicator | Wiring | Lights when |
|---|---|---|
| 5 V good (green) | 5 V pin 1 and pin 3 (one diode each) to a 1 kΩ and a green LED to GND | pins 1 and 3 carry about 5 V |
| 12 V good (green) | XH pin 1 to 3.3 kΩ and a green LED to GND | XH pin 1 carries about 12 V |
| 12 V on a 5 V pin (red) | 5 V pin 1 and pin 3 (one diode each) to a 6.8 V zener, 470 Ω and a red LED to GND | a 5 V pin sees more than about 8.5 V |
| Ground continuity | GND pins (5 V pins 2 and 4, XH pin 2) each to a test pad, meter reads below 0.1 Ω between them | |

Reversed polarity lights nothing (the LEDs block it), so **both green LEDs must be on and the red LED off**.
Add 4 mm test pads for +5V, +12V and GND so a meter can read the voltages. The jig is not a substitute for the meter
readings below; it makes a wrong harness obvious at a glance.

**Test points on the board.** Thirteen pads and loops are silkscreened with what they carry; none is fitted as a part (they are not in the BOM or placement files).

| Label | Net | Where | Use |
|---|---|---|---|
| GND (three loops) | `GND` | top left, right of the `J1-H1` socket, and beside `EN` | scope or meter ground; a hook probe or a wire loop clips on |
| 12V | `+12V_MOTOR` | left, beside `D1` | supply after the input connector |
| 12V F1 | `+12V_MOTOR_F1` | right of `D1` | after `F1`: 12 V here and not at `12V` means `F1` has tripped (switch off, remove the fault, let it cool) |
| 3V3 | `+3V3` | right of the ESP32-S3-ETH | logic supply |
| 5V | `+5V_LED_RING` | left, above `J5` | ring supply after `Q5` |
| DAT 3V3 | `LED_DATA_3V3` | right of the module | GPIO18, the ring data before `R18` |
| DAT, CLK | `LED_DATA`, `LED_CLK` | in the LED block right of the module | the 3.3 V data and clock at the `U3`/`U5` inputs |
| SCL, SDA | `I2C_SCL`, `I2C_SDA` | SCL below the DAT 3V3 pad, SDA at the right edge | panel expander bus |
| EN | `ENABLE` | beside `R23` | TMC2209 enable (low = driver on) |
| LED | `LED_DATA` | beside `EN` | 3.3 V ring data before the level shifter |

`STEP`, `DIR` and the TMC2209 UART lines run only under the plugged-in modules, so there is no pad for them: probe them on the pins of the module itself.

**Checklist** (supply unplugged from the mains for steps 1 and 2; use a multimeter):

1. **Harness, supply end disconnected.**
   - [ ] Continuity (below 0.1 Ω) from each plug pin to the far end of its wire: `J-PWR-5V` pins 1 and 3 to the +5V lugs,
     `J-PWR-12V` pin 1 to the +12V lug, `J-PWR-5V` pins 2 and 4 and `J-PWR-12V` pin 2 to the COM lugs.
   - [ ] The 12 V plug is an XH 2-pin and the 5 V plug a Mini-Fit 4-pin: neither fits the other's header (try it, unpowered).
   - [ ] No continuity (no beep) between any two of: +5V group, +12V group, COM group.
   - [ ] Each crimp tugs without moving (pull each wire firmly).
2. **Harness on the supply, mains still off.**
   - [ ] Terminals are tight and no strand is exposed; FG is on the enclosure and mains earth.
   - [ ] Still no beep between +5V, +12V and COM at the plug.
3. **Supply on, plug not connected to the board.** Read each at the plug, black lead on a COM pin:
   - [ ] `J-PWR-5V` pins 1 and 3: 4.75 to 5.5 V, then **trim to 5.10 V**. Positive on the meter, not negative.
   - [ ] `J-PWR-12V` pin 1: 11.4 to 12.6 V, positive.
   - [ ] Jig: both green LEDs on, the red LED off.
4. **Supply off. Board connected, ring cable not connected.**
   - [ ] Plug in `J-PWR-5V` and `J-PWR-12V` (each keyed, fully seated, the Mini-Fit latch clicked).
   - [ ] Resistance between the +12V rail and GND at `C8` is not a short (it rises from a low reading as the
     capacitors charge, then climbs above 1 kΩ).
5. **Supply on, `F1` is soldered on, ring cable still off.**
   - [ ] 12 V at `C8`, 3.3 V on the `+3V3` rail (a pad of `C9`, or any 3.3 V socket pin).
   - [ ] Nothing hot after a minute (fingertip on `D33`, the ESP32-S3-ETH module and the TMC2209 carrier).
6. **Ring cable check, supply off.** Ring end of the cable:
   - [ ] Pin-for-pin continuity from `J5` to the ring's `J1` (straight through, no crossed wires). Pin 2 is +5 V on the mainboard but not connected on the ring, so it should show no continuity at the ring end. Pin 1 is unused at both ends.
   - [ ] Signal cable: `J-LED-SIG` to the ring's `J2`, pin for pin (1 data, 2 ground, 3 clock, 4 ground), no crossed wires.
   - [ ] +5V and GND at the ring end are not shorted (no beep, and no steady reading below about 100 Ω).
7. **Supply on, ring cable still off.**
   - [ ] 5.1 V on `J5` pins 5 and 6 (and pin 2, which is tied to +5 V), GND on pins 3 and 4; pin 1 is unconnected. About 0 V on `J-LED-SIG` pins 1 and 3 before the firmware starts.
8. **Switch off, connect both ring cables, switch on.** Check the limits on the **LEDs** tab first (below).
   - [ ] The ring stays dark at power-on (`R26`), then shows the set colour.
   - [ ] 5 V at `J5` pins 5 and 6 with the ring at full white (briefly) is above about 4.7 V.
9. **Tag the harness** with the date and your initials. Any change to the harness voids the tag.

**Before the first power-up with the ring**, check on the **LEDs** tab of the settings page that the ring limit is 4500 mA and the per-LED limit 36.6 mA (the defaults, with the SK9822 current at 31).
Without them full white would draw about 9.6 A, over the RD-85A's 8 A.
The 4500 mA limit holds the ring to about 4.5 A, the 70% of full scale the LEDs' datasheet allows for three-colour use.

**If a green LED is off or a red LED is on, stop.** Do not connect the board. Re-check the crimps and which wire
went into which terminal; a swapped or mis-pinned wire is far cheaper to fix on the harness than on the board.

## 4. Signals and control

### Main controller pin use

| GPIO | Signal | Goes to |
|---|---|---|
| 46 | STEP | TMC2209 step input |
| 48 | DIR | TMC2209 direction input |
| 0 | ENABLE | TMC2209 enable (active low); `R23` holds it high through boot |
| 47 / 15 | UART TX / RX | TMC2209 one-wire serial, via `R9` (1 k) |
| 18 / 2 | LED data / clock | Ring data and clock (SPI for SK9822) through `U3`/`U5` to `J-LED-SIG` |
| 17 / 16 | I2C SDA / SCL | TCA9555 expander |
| 38 / 39 | Main encoder A / B | Panel (its button is on the expander, `P0_3`) |
| 1 / 40 | Hue encoder A / B | Panel |
| 3 | Ring temperature | `J-LED-SIG` pin 5 through `R66` (1 k) and `C16`, pulled up by `R65` (10 k 1 %); ADC1_CH2 |
| 43 / 44 | Saturation encoder A / B | Panel |
| 41 / 42 | Value encoder A / B | Panel |
| 45 | Fan PWM | `Q6` pulls the fan's PWM input low (inverted); undriven = fan at full speed |
| 21 | Tap button | Panel |
| 9 to 14 | Ethernet chip | Inside the ESP32-S3-ETH module |
| 19 / 20 | USB | Programming and log port |

Every pin on the ESP32-S3-ETH socket was checked against Waveshare's published pinout, and the
TMC2209 sockets against their makers' pinouts.

### The motor driver

- **Mode:** step and direction, configured over the UART at start-up. The driver's
  address pins and clock are tied to ground; the carrier's trimpot is ignored because the firmware
  sets the current.
- **Settings:** 800 mA RMS (the motor is rated about 1 A), 1/32 microstepping with the driver
  interpolating to 256, and the quiet StealthChop mode. At 32 microsteps a full turn is 6,400
  steps.
- **Health checks:** every 5 seconds the firmware checks the serial link and the driver's
  over-temperature and short-circuit flags, and restores the settings if the driver has reset.
- **Boot safety:** three of the control pins (GPIO 0, 45 and 46) double as the processor's boot
  "strapping" pins. `R23` (1 k to 3.3 V) holds the enable pin (GPIO 0) high, which keeps the
  driver disabled from power-on until the firmware is ready. The other two, STEP and the fan PWM
  pin, are low at reset. This makes boot independent of whatever pull-ups the driver module has.

### The lighting path

1. The main controller's LED engine draws each frame (colour, effect, beat flash, current limit), about 100 a second,
   and sends it to the ring as SK9822 data and clock on GPIO18 and GPIO2 (SPI, 2 MHz).
2. `R18` and `R61` (330 Ω) feed the two buffers.
3. `U3` (data) and `U5` (clock) lift them from 3.3 V to 5 V (they must
   be the **AHCT** part, whose inputs accept 3.3 V as a "high"), and `R63`/`R64` (33 Ω) feed `J-LED-SIG` and the signal cable.
4. `R26` and `R62` hold both lines low while the ESP32 starts, so the ring stays dark rather than flashing. On the ring, `R1` and `R2` (470 Ω) take the lines into `D1`.

The VAL knob picks one of the engine's **presets** (a press of its button makes it the brightness) and EXP1 a **palette**; HUE and SAT shift and fade the colours of whatever the preset draws (the User Guide's *Presets and the knobs*).

### Panel controls and the expander

- The **four encoders** and the **tap button** go straight to processor pins, and the encoders are
  read by interrupts so fast turning isn't missed.
- The **main, hue, saturation and value push buttons** are on the **TCA9555** expander, because the
  socket has no spare pins (the main encoder's button is there so GPIO3 can read the ring thermistor). They are read by polling every 20 ms.
- The expander also carries **three spare encoder headers** (`J-ENC-EXP1` to `3`; the third has
  no push-button channel). The current firmware does not read them. The expander's spare pins (`P0_4`, `P0_7`, `P1_7`)
  are inputs, not brought out.
- **Static protection.** Every panel signal pin goes through a 1 k resistor, with a small TVS diode
  to ground on the connector side, so a spark from a person touching the panel wiring is clamped
  before it reaches a processor pin. Each encoder's 3.3 V supply pin has a 100 Ω resistor in series,
  and the LED ring's 5 V supply has its own TVS (`D32`) beside `J5`.
- **Encoder modules** are the common KY-040 type. Their pull-ups are fed from **3.3 V**, never 5 V,
  because the lines go straight into 3.3 V-only inputs.
- GPIO 43 (saturation encoder A) is also the processor's boot-log output; at worst a closed encoder
  contact shorts it to ground for a few milliseconds during a reset, which the pin tolerates.

### Encoder header pin order

All seven encoder headers use the same five pins, in this order:

| Pin | 1 | 2 | 3 | 4 | 5 |
|---|---|---|---|---|---|
| Signal | +3V3 | SW (push) | DT | CLK | GND |

(The board is silkscreened V, S, D, C, G.) The tap button header has two pins: GND and TAP.
Suggested panel wire colours: main encoder blue, hue red, saturation green, value yellow, buttons
orange, ground black.

### Network

Wired Ethernet comes from the W5500 chip on the ESP32-S3-ETH module. Wi-Fi is the module's own
2.4 GHz radio. Both are active at once and share the same OSC port and web pages.

## 5. Connector reference

| Ref | Connector | Use |
|---|---|---|
| `J-PWR-5V` | Molex Mini-Fit Jr, 4 pin (2×2) | 5 V in: pins 1 and 3 +5 V, pins 2 and 4 GND (see above) |
| `J-PWR-12V` | JST XH, 2 pin (`B2B-XH-A`) | 12 V in: pin 1 +12 V, pin 2 GND (see above) |
| `J3` | JST-XH, 4 pin | Motor: **1 A+, 2 A−, 3 B+, 4 B−** (pins 1+2 are one coil, 3+4 the other) |
| `J5` | JST VH, 4 pin, vertical (B4P-VH), 3.96 mm | LED ring power: 1 and 2 +5 V, 3 and 4 ground |
| `J-LED-SIG` | JST XH, 5 pin (B5B-XH-A), 2.5 mm | LED ring signal: 1 data, 2 ground, 3 clock, 4 ground (5 V), 5 ring thermistor |
| `J1-H1`, `J1-H2` | Two 20-pin sockets (1 × 20 each), 2.54 mm | ESP32-S3-ETH |
| `J2-MODE`, `J2-MOTOR` | 8-pin sockets, 2.54 mm | TMC2209 driver module |
| `J-ENC-MAIN`, `-HUE`, `-SAT`, `-VAL` | 5-pin headers | The four panel encoders |
| `J-ENC-EXP1` to `3` | 5-pin headers | Expansion encoders: EXP1 palette, EXP2 effect speed, EXP3 spare |
| `J-BTN-TAP` | 2-pin header | Tap button |
| `J-FAN` | 4-pin keyed header (KK254 style) | **4-wire 12 V** fan (not 5 V): 1 ground, 2 +12 V, 3 tach (to the expander, `P0_6`, through 4.7 k), 4 PWM. Powered straight from the 12 V motor rail |

The motor connector follows the common stepper wiring, so a motor cable wired black, green, red,
blue (A+, A−, B+, B−) plugs straight in. If the platter turns the wrong way, swap two wires of one
coil, or tick **Reverse motor direction** on the settings page.

## 6. The LED ring board

The ring board is the second of the project's two printed circuit boards and lives in this repository
(`kicad_led_ring_offset/`, its own KiCad project), so the repository holds every part of the physical build. It is
fixed near the top of the case and **does not turn** with the platter.

| | |
|---|---|
| Size | 203.2 x 203.2 mm (8 inch), two layers; **1 oz copper** (JLCPCB's default), with the 5 V harness fuse |
| LEDs | 178 x SK9822-A (OPSCO, APA102-type with data and clock, 5050 6-pin package, RGB, 5 V with a **5.3 V maximum**, 27 kHz PWM so it films without flicker), on a 12.7 mm (0.5 inch) pitch in an offset "brick" pattern, wired as one long chain (`D1` at the top to `D178` at the bottom) |
| Per LED | One 100 nF capacitor each (178 in total), on the top side beside its LED: every part is on the top, so JLCPCB assembles one side only |
| Other parts | `CP1`, a 1,000 uF 16 V surface-mount bulk capacitor (D10 x L10.2); `R1` and `R2`, 470 ohm in series with the data and clock inputs (the datasheet asks for both); `TH1`, a 10 k NTC (Murata NCP18XH103F03RB, B 3434 K, 0603) on the top side just inside the inner LEDs, for the ring temperature; four M3 mounting holes |
| Connectors | `J1`, JST VH 4 pin, side entry (B4PS-VH), in the bottom-right corner about 4 mm in from the bottom edge, with its opening toward the left: power, pin-for-pin the same as the mainboard's `J5`. `J2`, JST XH 5 pin (B5B-XH-A): data, clock and the thermistor, pin-for-pin the same as the mainboard's `J-LED-SIG` |
| Current | 18 mA per colour at the full 5-bit current field (31, the firmware default), 12.2 mA at the maker's recommended 22. The firmware holds each LED to 36.6 mA (full white at field 22, the maker's heat limit), so single colours run at the full 18 mA and white at about 68%; full white on every LED would be 9.6 A at 31 and 6.5 A at 22. The LED engine's limiter holds it to about 4.5 A, the 70% of full scale the datasheet allows for three-colour use |
| Mounting | Four M3 holes on a 183.8 mm square, fixed near the top of the case; does not rotate with the platter. An aluminium heat plate goes under it (see the enclosure notes, `enclosure/MOTOR_MOUNT.md`): only `J1` and `J2` have pins through the board, so the plate needs one closed window there |
| Check status | Design-rule check with schematic parity: 0 violations, 0 unconnected, 0 parity issues. Electrical-rule check: 0 |

### Cables to the mainboard

Two straight-through cables: a four-wire power harness (JST VHR-4N housings and SVH-41T-P1.1 crimps, 16-18 AWG, at both ends) from `J5` on the mainboard to `J1` on the ring,
and a five-wire signal cable (JST XHP-5 housings) from `J-LED-SIG` to `J2`, each signal beside a ground and the thermistor on pin 5. No crossed wiring or adapter is needed. Keep the signal cable away from the motor cable.

| Pin | Mainboard `J5` | Ring `J1` |
|---|---|---|
| 1, 2 | +5 V ring supply | +5 V |
| 3, 4 | ground | ground |

| Pin | Mainboard `J-LED-SIG` | Ring `J2` |
|---|---|---|
| 1 | `LED_DATA_OUT` (5 V data from `U3` through `R63`) | `LED_DATA_RAW`, through `R1` (470 ohm) to `D1` SDI |
| 2, 4 | ground | ground |
| 3 | `LED_CLK_OUT` (5 V clock from `U5` through `R64`) | `LED_CLK_RAW`, through `R2` (470 ohm) to `D1` CKI |
| 5 | `NTC_RING_J`: `R65` 10 k 1 % pull-up to +3V3, `D34` ESD clamp, then `R66` 1 k / `C16` to GPIO3 | `NTC_RING`, `TH1` (10 k NTC) to ground |

**Wire sizes.**

- **Power harness (`J5` to `J1`):** 16 or 18 AWG, all four positions, as short as the case allows (about 0.5 m or less). JST VH is rated 10 A per pin with 16 AWG; at the 4.5 A limit each pin carries about 2.3 A.
- **Signal cable (`J-LED-SIG` to `J2`):** 22 to 24 AWG, all five wires. Twist each signal with its ground, or keep the four wires together, and route it away from the motor cable.

The signal cable's two grounds sit in parallel with the power harness's grounds, so they carry part of the ring's return current: about 15 to 25 %, or about 1 A at the 4.5 A limit. That is harmless at 22 to 24 AWG. With thinner wire, a power plug whose ground pins are not fully home would push the whole ring current through the signal cable. Always plug and unplug both cables with the power off.

### Voltage drop and brightness at the platter

These figures are simulated or estimated, not measured.

**Ring copper.** The ring's own copper was modelled from the filled pours, with every LED drawing the same current. At 1 oz the worst LED (`D1`, farthest from `J1`) sees about 8 mV of drop at 4.5 A, counting +5 V and ground together (2 oz would halve it for about $3 more per board at 50 to 100 boards, roughly double the bare board). The busiest copper carries about 1.4 A per mm of width at 9.6 A, and 1.4 A per mm is still inside 1 oz (IPC-2221 gives about 2.4 A for a 1 mm external 1 oz track at a 10 °C rise). `J1`'s pads join the pours solidly, with no thermal-relief spokes, so there is no narrow neck at the connector. A fault's current is capped by the 7.5 A harness fuse (section 3).

**Whole path at the 4.5 A limit.** Most of the loss is in contacts and wire, not copper:

| Part of the path | At 4.5 A |
|---|---|
| Mainboard copper | about 12 mV |
| `Q5` | about 27 mV |
| Mated connector pairs (the Mini-Fit input, two VH pins each way on the ring cable) | up to about 135 mV at maximum contact resistance (about half that typically) |
| 0.5 m of 18 AWG harness | about 47 mV |

From a 5.1 V supply the LEDs see 4.86 V or more. Do not raise the supply to make up the drop: the SK9822's limit is 5.3 V, and it would see the full supply whenever the ring is dark.

**Brightness at the platter.** These are estimates from the datasheets' typical luminous intensity, which varies by about ±30 % between parts. The 1/4 in acrylic platter is 1 in above the LEDs; 69 % transmission at 1/8 in works out to about 52 % at 1/4 in. About 92 % of the ring's light reaches the platter. Inside 90 mm of the centre, the platter top is lit evenly to within about 0.64 (darkest to brightest); outside about 110 mm it gets only about 18 % of the centre's light, because the LEDs end at a 96 mm radius.

| Platter top, cd/m² (r < 90 mm) | White | Red | Green | Blue |
|---|---|---|---|---|
| SK9822-A, field 31, per-LED limit 36.6 mA (default) | about 2,460 (4.5 A) | about 1,350 | about 3,180 | about 730 |
| SK9822-A, field 22 | about 2,460 (4.5 A) | about 915 | about 2,150 | about 490 |
| SKC6812RV ring (archived), for comparison | about 3,150 (4.5 A) | about 990 | about 2,890 | about 600 |

For scale, a phone screen is 500 to 1,000 cd/m² and an indoor LED video wall 800 to 1,500. White is set by the 4.5 A limit, and the SK9822-A gives about 20 % less light per amp than the SKC6812RV. Single colours draw a third of white's current (3.2 A at field 31), so the ring limit does not hold them, and they are where the SK9822-A is brighter.

The right-angle header was chosen for the ring so the cable enters from the side and fits the low mounting location; the mainboard uses a
vertical header.

**Strain relief at `J1`.** The ring hangs from the lid on four standoffs, which easily carry its own weight (about 190 g). The load to guard against is the harness: a stiff four-wire cable carrying up to about 7 A pulls on a side-entry connector near the board edge, and plugging and unplugging it pushes there too. Clamp or tie the cable to the tray or the lid a few centimetres from `J1`, with a loop of slack between the clamp and the connector, so the board never carries the cable's weight or stiffness. When fitting the ring, put all four standoff screws in loosely and then tighten them evenly, so a slightly bowed board is not pulled flat against its 0603 capacitors. On the board, the data line (`J1` pin 1, the one thin trace at the connector) now leaves its pad straight up, away from the cable's pull, and tapers from 0.5 mm to 0.2 mm over about 12 mm, so a tug on the connector does not load a thin neck at the pad. (A straight connector on the back was considered and not adopted: its plug stands 15 to 20 mm off the board, which gives a sideways pull about three times the leverage, and through-hole parts on the bottom side are not part of JLCPCB's normal through-hole assembly.)

### Layout and power copper

- **Data chain:** one long chain of 178 LEDs. The jump from `D89` to `D90` goes around the centre hole on the bottom layer as a routed track.
- **Power:** +5 V is a copper pour on the top layer and ground a pour on the bottom layer, not traces; the board has 388 vias and 926 track
  segments, 0.2 mm wide except the 0.3 mm capacitor ground stubs and the tapered data entry at `J1`. A simulation of the voltage drop to the farthest LED came out under 12 mV (not measured).
- **Power in:** the mainboard side carries the same current through two connector pins each for +5 V and ground (fused at the supply, 7.5 A; see section 3).

### Parts and ordering

| Part | Reference | LCSC | JLCPCB tier | Qty |
|---|---|---|---|---|
| SK9822-A 5050 clocked RGB LED (OPSCO, 27 kHz PWM, pin 1 = SDI, 2 = CKI, 3 = GND, 4 = VDD, 5 = CKO, 6 = SDO, checked against the maker's datasheet on 2026-11-09) | `D1` to `D178` | C5378730 | Extended | 178 |
| 100 nF 0603 X7R, 50 V | `C1` to `C178` | C14663 | Basic | 178 |
| 470 ohm 0603, 1 % | `R1`, `R2` | C23179 | Basic | 2 |
| 1,000 uF 16 V SMD electrolytic, D10 x L10.2 (ROQANG RVT1C102M1010) | `CP1` | C72470 | not checked | 1 |
| 10 k NTC 0603, 1 %, B 3434 K (Murata NCP18XH103F03RB) | `TH1` | C13564 | Extended | 1 |
| JST VH 4 pin, side entry (B4PS-VH) | `J1` | C157996 | Extended | 1 |
| JST XH 5 pin, vertical (B5B-XH-A) | `J2` | C157991 | Extended | 1 |

- **Assembly scope:** 360 surface-mount parts (LEDs, capacitors including `CP1`, `R1`, `R2`, `TH1`) plus two through-hole parts, the connectors `J1` and `J2`. Both have
  matching JLCPCB wave-solder equivalents, so **full assembly needs no hand-soldering**. Upload `LEDRing_BOM_SMD.csv` and
  `LEDRing_CPL_SMD.csv` (360 rows) for surface-mount only, or the `_FULL` pair (362 rows) for everything.
- **Component cost:** about $15.58 a board at 1 board, $13.55 at 10 and $11.36 at 100 (LCSC prices of 2026-11-13,
  `jlc_export/ring_bom_cost.csv`; this excludes JLCPCB's per-order assembly and setup fee, and will drift).
- **Stock (2026-11-10):** the SK9822-A has about 17k at JLCPCB (enough for about 95 rings) from one maker; re-check at
  upload. OPSCO's plain SK9822 (C2829089, about 92k) is a fallback: same footprint and price, BGR colour order; film a
  sample first, as older SK9822s ran their PWM at 4.7 kHz instead of 27 kHz. The CPL turns every LED 180 degrees from KiCad's rotation to match JLCPCB's footprint (`tools/check_cpl_rotation.py`).
- **Before ordering:** leave the copper at 1 oz (the default); check the LED pinout above if the part is ever substituted; run JLCPCB's BOM upload for a final live
  stock check.

The Gerbers, BOM, placement files and cost roll-up are in `kicad_led_ring_offset/jlc_export/`, generated by
`kicad_led_ring_offset/tools/build_jlc_assembly.py` (re-run it after any board change).

Two earlier comparison layouts (a plain square grid and a true hexagonal packing, neither fully routed) remain in the separate MegaMagicTurntable
repository as design history; this offset layout is the one with real order files.

## 7. Motor and mechanics

- **Recommended motor:** StepperOnline **17HS08-1004S**: NEMA 17, 1.8° per step, 1 A per phase,
  3.7 Ω, 4.5 mH, 16 N·cm holding torque, 21.5 mm body, 400 mm leads (black, green, red, blue =
  A+, A−, B+, B−). It is a short ("pancake") motor, which is enough torque for a light platter.
  A heavy platter would justify a full-length NEMA 17 (40 to 60 N·cm); the electronics suit either.
- **Coupling:** a flexible shaft coupler joins motor and platter directly. There is no gearbox, so
  platter inertia is felt straight at the motor. This is why every speed change is ramped.
- **Speed:** one revolution per 12 beats; default top speed 17 RPM (adjustable). At 1/32
  microstepping the platter needs 6,400 steps per turn.
- **Layout constraints on the mainboard:** the bulk capacitors `C8` and `C11` are 7.7 mm tall and
  fit between the two plug-in modules, so check clearance if you substitute taller parts.

## 8. Current, heat and protection notes

- **Ring current path.** The 5 V route from the power connector through `Q5` to `J5` is
  carried by copper pours 6.5 mm wide or more at the narrowest point, with the return on a full
  bottom-layer ground plane. The power connector and ring connector each use two pins per
  supply, so each pin carries about 3.3 A at full white (2.3 A at the LED engine's limit).
- **TMC2209 heat.** Fit a heatsink on the module; the maker calls for heat dissipation.
- **Enclosure heat.** The LED ring is the main heat source: about 3.9 W per amp of ring current.
  The other parts add about 7 W (motor 3 to 4.5 W). For the earlier 14 × 17 × 2 in plywood box (not redone for the current 6 mm birch box, about 13.3 × 10.7 × 2.7 in inside, which probably runs hotter) in a 43 °C
  (110 °F) room, a closed box holds only about 2 A average ring current (3 A with generous vents);
  high brightness for hours needs about 6 to 15 CFM of forced air, which one slow 120 mm fan in the
  bottom panel can provide. The numbers are estimates, so measure first: the Status tab, the log
  and `/status` report the main processor's own temperature. The board has a fan header (`J-FAN`) whose speed
  follows that temperature between two settings you can change on the Motor tab; if the firmware stops, the fan
  runs at full speed.
- **Fuses.** `F1` (a 2 A hold resettable fuse, 1812, rated 16 V) protects the 12 V side; a 7.5 A blade fuse in the harness at the supply protects the 5 V side (section 3). `Q5` (SP30P03NK) blocks reverse polarity on the
  5 V side; it loses at most about 0.32 W at the ring's 7.3 A worst case (6 mΩ maximum at 4.5 V gate drive, hotter at temperature), and about 0.12 W at the LED engine's 4.5 A limit. There is no overvoltage crowbar: the connectors are keyed and of different families, and the harness is tested before it meets the board (section 3). The RD-85A's overload protection is 110 to 150 % of its total rated power (88 W), not a 5 V current limit: on the 5 V output that is roughly 20 A or more, and the current-range figures differ between datasheet copies ([Mean Well RD-85 datasheet](https://meanwellaustralia.com.au/sites/default/files/2025-01/MEAN%20WELL%20RD-85%20DATA%20SHEET.pdf)). That protects the supply, not the wiring, hence the harness fuse. The ring's normal limit is still the LED engine's current limiter (about 4.5 A): the 9.6 A of an unlimited full white is too little over 7.5 A to blow the fuse quickly, and the wiring, connectors and copper carry it.
- **Voltage spikes.** `D5` (SMBJ15A) starts to conduct at about 16.7 V and clamps at about 24.4 V
  at its rated surge current: below the TMC2209's 29 V limit, the 35 V bulk capacitors `C8` and
  `C11`. Running the motor rail at 12 V rather than 24 V is what makes this possible. Use a regulated 12 V supply; an adjustable one should
  stay between 12 and 13 V. Still never hot-plug the power connector.
- **Copper and voltage drop** (simulated, not measured). The mainboard is ordered at 1 oz.
  Ring +5 V path from the power connector to `J5`: about 2.8 mΩ of copper at 1 oz, 13 mV at the 4.5 A limit; ground
  return about 2.4 mΩ, 11 mV. IPC-2221 (external, 1 oz) gives the narrowest point, 6.5 mm wide, about 9.2 A at a 10 °C rise
  and 12.5 A at 20 °C, both above the 7.5 A harness fuse. 12 V motor path: about 50 mΩ, 0.1 V at 2 A. `+3V3` now comes out
  of the module at the socket, next to its main loads.
- **No hot-plugging.** Do not connect or disconnect the motor (`J3`), the ring cables (`J5`, `J-LED-SIG`), the
  power connector or any of the plug-in modules with the power on. A motor unplugged live produces a
  voltage spike that can destroy the TMC2209 (the module's maker says the same about the module
  itself), and the ring cable carries up to about 7 A. The `D5` clamp on the 12 V rail is not
  designed for this.
- **Logic supply.** `D33` (SS34) feeds `U3`/`U5` and the module's `VSYS` from `+5V_LED_PSU`, so they see about 4.7-4.8 V from a 5.1 V supply (the 74AHCT works from 4.5 V, the module's buck from 4 V) and a reversed supply is blocked. Not yet tested on hardware.
- **Ring temperature.** `TH1` on the ring (10 k NTC to ground) against `R65` (10 k 1 % to +3V3) on the mainboard: half of 3.3 V at 25 °C, read on GPIO3 through `R66`/`C16` every 2 s. The firmware runs the fan from it first and then lowers the ring current limit (README, Heat and ventilation); an open or shorted sensor is reported and handled.

## 9. Manufacturing and files

| Item | Where |
|---|---|
| Schematic (sources) | `kicad_mainboard/*.kicad_sch`, rendered to `Mainboard_schematic.pdf` |
| PCB | `kicad_mainboard/Mainboard.kicad_pcb`, renders `Mainboard_render_top.png` and `_bottom.png` |
| Manufacturing files | `kicad_mainboard/jlc_export/`: Gerbers and drill, plus bills of materials and part-placement files in two versions (surface-mount only, or full assembly) |
| Ring board | `kicad_led_ring_offset/` (KiCad project, PDFs, reports) with its own `jlc_export/` and `tools/build_jlc_assembly.py` (section 6) |
| Check reports | `erc_report.rpt` (electrical rules) and `drc_report.rpt` (layout rules), in each KiCad folder |
| Part-number tool | `kicad_mainboard/tools/build_jlc_assembly.py`, which rebuilds the bills of materials from the board |
| Test points | `kicad_mainboard/testpoints.kicad_sch`; `tools/add_test_points.py` places the pads on the board |

**Ordering from JLCPCB.** Surface-mount only: 86 parts, 254 solder joints. Full assembly
(through-hole parts too): 104 parts, 370 joints (counted from the files by `tools/build_jlc_assembly.py`; see `jlc_export/README.md`). Every part has an LCSC number (JLCPCB re-checks stock when you upload).
Notable ones: expander `U1` TCA9555PWR C465732, data buffer `U3` C7484, TVS `D5` C113988, bulk capacitors
`C6`, `C8` and `C11` C176683 (an "extended" part: confirm stock when ordering).

**Fuses.** The board has no blade-fuse clips. `F1` is a soldered resettable fuse (RUILON C20812, about $0.09) and the 5 V fuse is a standard 7.5 A ATO blade fuse in an inline holder in the harness (section 3), bought with the harness parts, not from JLCPCB. A through-hole ATO holder on the board was considered and not adopted: the only well-stocked one (Littelfuse 178.6165.0002, C207061) costs about $1.91 each at 100.

**Ring board.** 359 surface-mount parts plus three through-hole parts, about $18 of components (section 6).

**Before ordering the board**, check:

- Both boards at 1 oz copper (JLCPCB's default).
- In JLCPCB's placement preview, the orientation of `C11`, `Q5` and `U1` (TSSOP-24, CPL rotation set from JLCPCB's usual TSSOP drawing, not looked up) looks right.
- You have the inline holder and 7.5 A ATO fuses for the 5 V harness.

## 10. Verification status and limits

**Checked on the computer**

- Both boards pass design-rule checks (no violations, nothing unrouted) and each matches
  its schematic exactly. Both schematics pass electrical-rule checks.
- Socket pin assignments match the manufacturers' published pinouts.
- The board meets JLCPCB's two-layer manufacturing limits (smallest hole 0.3 mm, narrowest track
  0.2 mm).
- The speed-control logic has automated tests (`tests/`), and the firmware builds with no warnings.

**Checked on a bench unit.** A standalone ESP32-S3-ETH running the current firmware boots, joins
Wi-Fi, announces `liquidturntable.local` and serves its web pages. The automatic renaming of a
taken hostname was tested against a name owned by a PC on the network: the unit detected the clash
and announced itself as `-1`.

**Not yet verified.** No fully assembled mainboard has been powered up yet. Still to confirm on
real hardware: the motor driver link and motor behaviour, the boot straps with the driver module
fitted, the LED engine and ring, encoder feel, and the fuse clip fit with real fuses. The project
README lists these under "Open items".

**Known limits**

- The three spare encoder headers are wired but not used by the firmware.
- Wi-Fi shares the 3.3 V rail with the wired network and the main controller's other loads; its power is capped to keep the rail within budget.
- The panel and ring cabling are made by hand. Check the pin order on the silkscreen, not on the
  connectors' descriptions in text.
