CadProd compact

Music player PCB: next revision (rev 2)

Everything to change, check and test for rev 2. The main goal is power saving while the player is off:

The estimates come from datasheet typical values. Measure the real numbers at bring-up (section 6).

1. Power rails (the new design)

BAT ── BQ24074 ── +REG ──┬── TPS7A2033 "Digital_Vol" (EN = VIN, always on) ── AwO_D3v3 ──┬── ESP32-S3
                         │                                                                ├── BQ24074 EN2, pull-ups R9 R10 R11 R12
                         │                                                                └── Q2 AO3401A ── D3v3 (switched) ──┬── SD card + R6 R7
                         │                                                                                                    └── PCM5102 DVDD + CPVDD
                         └── TPS7A2033 "AnaVol" (EN = EN_ANAVOL) ── A3v3 (switched) ──┬── PCM5102 AVDD
                                                                                        └── MAX9722 amp

2. Changes already in the schematic

• Digital regulator changed from MCP1825 to TPS7A2033 (Iq ~6.5 µA instead of ~120 µA), with EN tied to VIN
• C24 1 µF at the digital regulator's VIN
• C21/C22 (10 µF + 100 nF) at its output
• Always-on rail AwO_D3v3 split from the switched D3v3
• Q2 AO3401A high-side switch, with R25 100k gate pull-up (off at boot and in sleep) and R26 10k to the GPIO
• R5 (SD Det pull-up, ~330 µA with a card in) removed. The firmware uses the internal pull-up while awake.
• R6 and R7 (SD pull-ups) moved to switched D3v3, so they can't back-power the card
• PCM5102 DVDD/CPVDD moved to switched D3v3
• R9, R10, R11, R12 and BQ24074 EN2 on AwO_D3v3. EN on the switched rail would hold the ESP32 in reset forever.

3. Still to do in the schematic

Must fix

• C23 (47 nF) from Q2 gate to Q2 source, side by side with R25. This gives a ~0.5 ms turn-on (10k × 47 nF), so the inrush is about 75 mA instead of a dip that could reset the ESP32. (Gate-to-GND would switch D3v3 on briefly at every power-up.)
• Connect the GPIO label at R26 to a real pin: IO21 (module pad 23). IO11 (pad 19), IO12 (pad 20) and IO18 (pad 11) are also free. Avoid the strapping pins IO0, IO3, IO45 and IO46.
• Give the two GPIO labels separate names. Same-named labels are one net in KiCad. Use BAT_SENSE → IO8 (pad 12) and D3V3_EN → IO21 (pad 23).
• Battery divider on IO8 (module pad 12) (R5/R27 1 MΩ, C25 100 nF). It's needed for battery %, LED4 (above 60 %) and the low-battery warnings:

  BAT+ ──[1 MΩ]──┬── IO8
                 ├──[1 MΩ]── GND
                 └──[100 nF]── GND

  It draws about 2 µA. IO8 is on ADC1.
• Make thermal refes for the GND on LEDs and Battery
• Battery connector Make the battery connector fit a JST connection

Check against datasheets

• BQ24074 TMR (pin 14) and ITERM (pin 15) are unconnected. Check what that means for the safety timer and the termination current, and tie them as intended.
• BQ24074 TS: R8 is a fixed 10k to GND, so temperature protection is off. For a LiPo in a closed case, consider a pack with a 10k NTC wired to TS.
• TPS7A2033 output capacitance: confirm that 10 µF + 100 nF on each output is within the stable range. It should be.
• AO3401A pinout in the footprint: 1 = Gate, 2 = Source, 3 = Drain (SOT-23).

LEDs

• Make the schematic and firmware agree on the colours. Right now they don't:

  GPIO	Net	Schematic	Firmware config.h
  38	LED1	D1 red	green
  39	LED2	D2 orange	purple
  40	LED3	D3 green	orange
  41	LED4	D4 purple	blue
  42	LED5	D5 blue	red

  Decide which colour goes where, then make the schematic, the silkscreen and config.h match.

• Optional: change R13–R21 from 1 kΩ to 2.2–4.7 kΩ. That's about 1 mA per LED now; higher values are usually still bright enough behind a case.

Optional

• External 1 MΩ pull-up on PLAY to AwO_D3v3. The firmware could then power down the RTC peripherals in deep sleep, saving a few µA.
• MAX17048 fuel gauge on I2C (e.g. IO11/IO12), if the divider's ~±10 % isn't good enough. It stays accurate while charging and under load, and draws ~4 µA in hibernate. It needs a firmware driver.
• Remove any leftover symbols that have no footprint (the old AP2112K symbol).

4. Footprints and BOM

• Digital_Vol: SOT-23-5 for the TPS7A2033 (it was the MCP1825's SOT-223/DDPAK). Pins: 1 VIN, 2 GND, 3 EN, 4 NC, 5 VOUT.
• Q2: SOT-23. Add C23 and C24 footprints.
• The BOM says µF, not "yF". KiCad shows µ as y on some values.
• TPS7A2033 ×2 (digital + analog). The MCP1825 is no longer used.
• Battery divider: 2 × 1 MΩ, 1 × 100 nF (0402/0603).

5. Layout

• C24 right next to the digital TPS7A2033's VIN pin. C21/C22 right next to its VOUT.
• C23 close to Q2, between its gate and source pads.
• Keep the ESP32's decoupling (C6, C7, C8) right at the module's 3V3 pin.
• C1/C2 (SD decoupling) close to the SD socket, on D3v3.
• Battery divider close to IO8, away from the I2S and SD clock traces. The 1 MΩ node is high-impedance and picks up noise.
• Short, wide traces from Q2 to the SD card VDD.

6. Bring-up tests

1. Rails: power from the battery only. AwO_D3v3 should be 3.3 V. D3v3 and A3v3 should be 0 V until the firmware turns them on.
2. Boot: the ESP32 starts every time (if it doesn't, EN/R11 is on the wrong rail). Hold Play for 3 s to turn on.
3. Inrush: with a scope on AwO_D3v3, the firmware switches D3v3 on. The dip must stay above about 3.0 V, and the ESP32 must not reset.
4. SD and audio: cards mount, music plays, and there's no pop at power-on.
5. Off current: power off, then put a multimeter (µA range) in series with the battery. Target ~25–30 µA. If it's much higher:
   • Is D3v3 really 0 V?
   • Is any ESP32 pin driving high into the SD card or DAC?
   • Are the LEDs off?
6. Charging: plug in USB. The orange LED lights while charging, also while the player is off. Unplugged, there's no wake loop.
7. Battery %: compare Settings > Battery with a multimeter on BAT+ (around 4.2 V full, 3.7 V about 50 %).
8. LEDs: check each colour against the table in section 3.

7. Firmware changes needed for rev 2

These go into flac_player_4/flac_player/ once the board exists. They're not done yet; ask when the GPIO is final.

• config.h: new PIN_D3V3_EN 21 (or whichever pin you choose). LOW = D3v3 on.
• Boot: drive PIN_D3V3_EN LOW, wait ~5 ms, then start the SD card.
• Before deep sleep: drive all SD pins (CS, CLK, MOSI, MISO) and I2S pins LOW, then set PIN_D3V3_EN HIGH and hold it.
  • Rev 1 holds SD CS high in sleep. On rev 2 that would push current into the unpowered card through its protection diodes.
• PIN_BAT_ADC 8. This turns on Settings > Battery, LED4 above 60 %, the red low-battery LED and the "Battery low" / "Battery critical" warnings. That code is already written.
• LED pin colours in config.h to match section 3.