AutoFilm-ESP32/docs/JC4827W543_PIN_AUDIT_VERIFIED.md

# JC4827W543 pin audit

Comparison of the GPIO assignments in this tree against the vendor
documentation at <https://github.com/lsdlsd88/JC4827W543> (GUITION
factory docs: spec PDF, IO pin distribution xlsx, Arduino demos).

Board variants: `JC4827W543N` (no touch), `...R` (resistive XPT2046),
`...C` (capacitive GT911). Ours is the **C** model.

## 1. Vendor-documented pin allocation (authoritative)

From `5-IO pin distribution/4.3 inches IO pin distribution.xlsx`,
cross-checked against the demo sketches
(`1-Demo/Demo_Arduino/3_3-*` and `4_*`):

| GPIO | Vendor function | Notes |
| --- | --- | --- |
| 0 | BOOT button | also wired to `LCD_TE` |
| 1 | `BL_CTRL` | LCD backlight |
| 2 | `SPECK_LRCLK` | onboard speaker amp I2S WS |
| 3 | `CTP_INT` / `RTP_IRQ` | touch interrupt (both models) |
| 4 | `CTP_SCL` | GT911 I2C SCL |
| 5 | free | P2 header (IO46/9/14/5) |
| 6 | free | P3 header (IO6/7/15/16) |
| 7 | free | P3 header (IO6/7/15/16) |
| 8 | `CTP_SDA` | GT911 I2C SDA |
| 9 | free | P2 header (IO46/9/14/5) |
| 10 | `TF_CS` | microSD slot |
| 11 | `RTP_DIN` / `TF_MISO` | shared SPI: resistive touch / SD |
| 12 | `RTP_CLK` / `TF_CLK` | shared SPI |
| 13 | `RTP_DIO` / `TF_MOSI` | shared SPI |
| 14 | free | P2 header (IO46/9/14/5) |
| 15 | free | P3 header (IO6/7/15/16) |
| 16 | free | P3 header (IO6/7/15/16); used by vendor LED demo |
| 17 | `U1TXD` | P4 header (GND/3V3/17/18), usable if UART1 unused |
| 18 | `U1RXD` | P4 header (GND/3V3/17/18), usable if UART1 unused |
| 19 | `USB+` | USB D+ |
| 20 | `USB-` | USB D- |
| 21 | `LCD_A0` | QSPI data 0 |
| 22–25 | — | do not exist on ESP32-S3 |
| 26–34 | — | flash/PSRAM or non-existent |
| 35 | not available | octal PSRAM (xlsx marks explicitly) |
| 36, 37 | free per xlsx | **but** consumed by octal PSRAM on N8R8/R8 modules — treat as unavailable until module variant confirmed |
| 38 | `RTP_CS` (resistive model) | on the **C** model this is the GT911 reset line — see §3 |
| 39 | `LCD_A3` | QSPI data 3 |
| 40 | `LCD_A2` | QSPI data 2 |
| 41 | `SPECK_DIN` | speaker amp I2S data in |
| 42 | `SPECK_BCLK` | speaker amp I2S BCLK |
| 43 | `U0TXD` | console |
| 44 | `U0RXD` | console |
| 45 | `LCD_CS` | QSPI chip select |
| 46 | free per xlsx | P2 header; strapping pin (`LOG`), **input-only** on ESP32-S3 — never an output |
| 47 | `LCD_CLK` | QSPI clock |
| 48 | `LCD_A1` | QSPI data 1 |

The board carries an onboard I2S speaker amplifier (datasheets for
NS4168 / AX9835 ship in `4-Driver_IC_Data_Sheet`), a `Speak` connector
(P7), `BAT` connector (P6), TF slot, UART0 console, and free-IO
headers: **P2** exposes IO46/9/14/5, **P3** exposes IO6/7/15/16, and
**P4** is GND/3V3/17/18 (the UART1 header). Vendor demos drive LEDs on GPIO 16/17 and a
DHT11 on GPIO 27 — note GPIO 27 is **not** in the vendor pin table,
so that demo value is suspect/generic.

## 2. What the code assigns today

### `board_jc4827w543` (`components/board_jc4827w543/board.c`)

| Signal | Code GPIO | Vendor | Match |
| --- | --- | --- | --- |
| LCD CS | 45 | 45 `LCD_CS` | yes |
| LCD SCK | 47 | 47 `LCD_CLK` | yes |
| LCD D0 | 21 | 21 `LCD_A0` | yes |
| LCD D1 | 48 | 48 `LCD_A1` | yes |
| LCD D2 | 40 | 40 `LCD_A2` | yes |
| LCD D3 | 39 | 39 `LCD_A3` | yes |
| LCD BL | 1 | 1 `BL_CTRL` | yes |
| TP SDA | 8 | 8 `CTP_SDA` | yes |
| TP SCL | 4 | 4 `CTP_SCL` | yes |
| TP INT | 3 | 3 `CTP_INT` | yes |
| TP RST | 38 | 38 (demo: `TOUCH_RES 38`) | yes — see §3 |
| Motor EN | 7 | 7 free (P3) | yes — locked |
| Motor STEP | 16 | 16 free (P3) | yes — locked |
| Motor DIR | 15 | 15 free (P3) | yes — locked |
| DS18B20 temp | 17 | 17 `U1TXD` (P4) | yes — UART1 unused |
| Speaker BCLK | 42 | 42 `SPECK_BCLK` | yes |
| Speaker LRCLK | 2 | 2 `SPECK_LRCLK` | yes |
| Speaker DIN | 41 | 41 `SPECK_DIN` | yes |

### `hal_display_nv3041a.c`

Driver **NV3041A** over QSPI via `esp_lcd` SPI host, 480×272, quad
mode, 40 MHz. Matches the vendor demo
(`Arduino_ESP32QSPI(45,47,21,48,40,39)` + `Arduino_NV3041A`,
`Arduino_Canvas(480,272)`).

Caveat: the spec sheet PDF lists the driver chip as "ST3401A" — that
string appears nowhere else; every demo and the burn files use
NV3041A. Treat the PDF label as a typo.

### `hal_input_gt911.c`

GT911 at I2C addr `0x5D` (`GT911_SLAVE_ADDRESS1` in the vendor
`touch.h`), I2C_NUM_0 @ 400 kHz on SDA 8 / SCL 4. Reset sequence
(drive INT low, pulse RST, release INT to input) selects the 0x5D
address per the GT911 datasheet — matches the vendor demo's
`TOUCH_RES 38` / `TOUCH_INT 3` wiring.

### `board_wroom` + `hal_*` (classic ESP32, unchanged)

| Signal | GPIO | Source |
| --- | --- | --- |
| Stepper STEP | 12 | `config.h`, `hal_motor.c` |
| Stepper DIR | 14 | `config.h`, `hal_motor.c` |
| Stepper EN (active LOW) | 27 | `config.h`, `hal_motor.c`, `board_wroom` |
| DS18B20 | 13 | `config.h`, `hal_temp.c` |
| Beeper | 25 | `config.cpp`, `hal_audio.c` |
| LCD 2004 I2C | SDA 21 / SCL 22, addr 0x27 | `config.cpp`, `hal_display_lcd2004.c` |
| Keypad rows | 19, 18, 5, 17, 16 | `config.cpp`, `hal_input_keypad.c` |
| Keypad cols | 15, 2, 0, 4 | `config.cpp`, `hal_input_keypad.c` |

Self-consistent with `docs/CURRENT_STATE.md`. Not related to the S3
panel.

## 3. Findings / discrepancies

1. **Display and touch pins are all correct.** Every QSPI, backlight
   and GT911 pin in `board_jc4827w543/board.c` matches both the
   vendor xlsx and the vendor Arduino demos. No changes needed.

2. **TP_RST on GPIO 38 is undocumented for the C model.** The xlsx
   labels IO38 `RTP_CS` (its resistive-touch function). The capacitive
   demos (`LvglWidgets/touch.h`) set `TOUCH_RES 38`, so on the C
   variant IO38 is clearly the GT911 reset. Our code is right, but the
   vendor table alone would not tell you that — the demo code is the
   authority here.

3. **Motor EN/STEP/DIR and DS18B20 are now locked on the S3 board.**
   Header labels confirmed: P2 exposes IO46/9/14/5, P3 exposes
   IO6/7/15/16, P4 is GND/3V3/17/18 (the UART1 header).
   - Motor STEP **16**, DIR **15**, EN **7** (active LOW) — all on P3;
     IO6 stays spare.
   - DS18B20 on **17** — P4 puts GND, 3V3 and data on one header;
     UART1 is unused.
   - Speaker amp I2S stays on the vendor pins: BCLK **42**,
     LRCLK **2**, DIN **41** (P7 `Speak`).
   - **Avoid**: 0 (BOOT/LCD_TE), 1–4/8 (LCD+touch+speaker), 10–13
     (TF/RTP SPI), 19/20 (USB), 21 (LCD D0), 35–37 (octal PSRAM),
     38–48 (used/strapping), 43/44 (console).
   - GPIO 46 is xlsx-free and on P2 but is **input-only** on the
     ESP32-S3 — never use it for outputs.
   - Octal PSRAM (`sdkconfig.s3`: `CONFIG_SPIRAM_MODE_OCT=y`) means
     33–37 are off-limits even though the xlsx only flags 35.

4. **Audio on S3 should use I2S, not a beeper GPIO.** The board has a
     speaker amp on IO2 (LRCLK), IO41 (DIN), IO42 (BCLK) — P7 `Speak`
     connector. `hal_audio` currently builds the LEDC buzzer only for
     esp32 and a stub elsewhere; an S3 implementation should drive
     I2S on those pins rather than allocating a GPIO.

5. **DS18B20 on S3 needs one free GPIO** (any of §3's candidates; it
   only needs input+open-drain drive). Hal currently stubs temp on
   non-esp32.

6. **Reserved/bus-shared notes**: GPIO 0 carries `LCD_TE` in addition
   to BOOT — do not use it for anything else. TF card SPI (10–13) is
   shared with resistive-touch signals on R models; on our C model
   it's purely the SD slot.

7. **No conflicts found** between the code's S3 selections and the
   vendor map — every consumed pin is one the vendor assigns to that
   same function.

## Motor / temp / speaker wiring as locked (phase I01)

| Signal | GPIO | Rationale |
| --- | --- | --- |
| STEP | 16 | free, P3 header, no boot function |
| DIR | 15 | free, P3 header |
| EN (active LOW) | 7 | free, P3 header |
| DS18B20 | 17 | P4 header carries GND/3V3/17/18 — one connector for the sensor |
| Speaker BCLK / LRCLK / DIN | 42 / 2 / 41 | onboard amp wired to P7 `Speak` |

Keep motor EN's disable-first behaviour (drive inactive at boot before
enabling output mode).