Snips controller

Snips Controller PCB (V2)

A wireless handheld controller for R2-D2 droid operation, designed for use at public events and conventions. Communicates with the droid over Zigbee, in a packetized state-report architecture.

The electrical design is in Stackup KDL, using the pinned Stackup library plus a local thumbstick model. The existing KiCad PCB is linked through snips_controller.stackup_sch; sync it with the Stackup KiCad plugin. Run stackup check PCB/v2.x/snips/stackup/board.kdl from the repository root before syncing.

A STEP model of the assembled board is in step/v2.0/snips_controller.step.

V2 board

70 x 30 mm, 4 layers: F.Cu signals, In1 solid 3V3, In2 solid GND, B.Cu signals. 1.6 mm thick. Routed, not yet fabricated.

• Module: an ESP32-S3-MINI-1U-N8 on the top, between the macro buttons and USB, turned 270 degrees. The 19.2 mm WROOM-1U does not fit the 17.5 mm there. The turn puts the pinless GND edge on the side and the signal pins toward the stick, the OLED and the carrier; pins are assigned by position, as GPIO_table.md explains. USB runs about 15 mm as a result, which is well within full-speed limits. Bluetooth and ESP-NOW work with no other radio.
• XBee: optional. U_XBEE1 is a row of 12 hand-solder pads on the bottom for a 22 x 17 mm castellated daughter board, which carries the XBee 3 Micro, its decoupling, the RESET pull-up and RC, and the ATTN/BOOTMODE pull-up. The link is UART with flow control, not SPI. RTS and DTR also allow XBee serial firmware recovery through the ESP32. Leave the area inside the pads free of parts; vias there stay tented. See GPIO_table.md.
• Reflow: the bottom side goes first and the top side last, because Espressif requires a single reflow for the module (datasheet section 12.3). The XBee reflows on its own carrier and never goes upside down, which is what stopped the two modules from sharing this board before.
• Radio link: Amidala's droids use XBee Zigbee 3.0, so a Zigbee controller needs the carrier. The firmware still has to move from the XBee's SPI interface to UART, and needs a V2 pin map.
• Programming: USB, through the module's native USB. To force download mode, hold SW_BOOT1 while pressing the power button. RXD0's pull-up holds the latch with no firmware, so a blank board flashes after a 1 s press, with no button held during the upload.
• No reset button: the power button turns the board on, and only firmware turns it off. PWR_HOLD is on RXD0, whose pull-up holds the latch through any reset or USB upload, so a blank or hung board stays on until USB and the battery are removed. Keep a watchdog running to recover from a hang. C_LATCH_FB1 gives the latch hysteresis, so that pull-up cannot restart the board at power-off, and R_PWR_EN_PD1 is 22 kΩ so a rising gate cannot couple through it and lift EN.
• Buttons: all 13 connect directly to module GPIOs. See GPIO_table.md.
• Power blocks: the buck, charger and power latch follow V1's relative placement, all on the bottom. Two charger GND vias were moved 0.1-0.75 mm in V1's layout to clear earlier button positions. C_BUCK_IN1 turned 90 degrees with its VSYS pad straight above the VIN pin. BUCK_FB runs around the input cap and hops to F.Cu past the inductor, so both planes shield it; keep it out from under L_BUCK1. The latch is too long to fit whole, so it is split at PWR_BTN_TRIGGER: the switch and PWR_LATCH_G node sit under the stick, turned 270 degrees, and the inverter sits beside the stick's lower pins. See diagrams/ for why they are arranged this way.
• OLED cable: U_OLED1's opening faces USB, and a Qwiic plug is 7.0 mm wide with its protrusions, so keep top-side parts out of x 0.9-7.9 mm from the mouth at y 52.7 mm down to the MH3 post. Check that a real cable goes in before committing the shell.
• BOM: the Value column carries each part's full details, including a capacitor's rated voltage, because fabs read it and the BOM has no voltage column. Keep rail names out of designators: PCBWay queried the V1.2 caps named C_3V3_* as 3.3 V parts, so V2 uses C_MCU_*.
• Hall sensor: U_HALL1 is on the bottom at (6.5, 4.75) mm, rotated 180 degrees, forward and left of the stick, 7.8 mm from its nearest pot pin in plan view against about 15.4 mm on V1. A 2 x 2 mm N52 magnet swings up to sit directly under the package centre. Keep at least 1.8 mm from the top of the package to the magnet at full press, or the output clips at the sensor's ±44 mT range. The stick reading against trigger travel still needs a bench test.
• Mounting: four Ø2.2 mm unplated holes, MH1-4, at (2.5, 2.6), (27.4, 2.8), (3, 62.5) and (27, 67) from the top-left corner, for M2 self-tapping screws into shell posts. MH3 sits above the power button, which holds the true corner, so a Ø4 post clears C_DBNC1 by only about 0.15 mm. The carrier has a long shallow chamfer at that corner, which also clears a Ø4 post on the bottom.
• Battery: BT1 is a 2-pin JST PH, the common connector on pouch cells this size: pin 1 VBAT, pin 2 GND. R_TS1 ties the charger's TS pin to a fixed 10 kΩ, so there is no pack temperature monitoring. Check the polarity of the cell's lead before plugging it in.
• USB and status LED: J_USB1 sits 2 mm right of centre, with its CC resistors and TVS moved with it. D_RGB1 sits between SW_PWR1 and J_USB1, turned 90 degrees.
• Through-board parts: the Alps pegs, stick pins, USB stakes and BT1 pins all reach the other side.

GPIO Assignment

See GPIO_table.md for every module pin.

Hardware Overview

Layout Diagrams

Suggested top-copper floorplans for the three sheets with real layout stakes — a switching regulator, a linear charger with a thermal pad, and the soft latch that switches the buck's enable. Each is checked against the actual component datasheet, not just the schematic. See diagrams/README.md for the full list and how the files are organized.

• Buck converter — /Power_Buck/* nets in Stackup
• Charger — /Power_Charger/* nets in Stackup
• Power latch — /Power_Control/* nets in Stackup

Passives & Support Components

Decoupling Caps

Resistors

Buttons

No external parts. Every button wires to GND and a module GPIO, using the internal pull-ups; see GPIO_table.md.

---

Communication Architecture

The droids use XBee Zigbee 3.0, so the controller joins that network through the XBee on the carrier, over UART in API mode. Each droid has its own coordinator and PAN ID, and all coordinators share one AES key. Without the carrier the ESP32 can use Bluetooth or ESP-NOW.

---

Bringup Sequence

Recommended bringup order:

1. Current-limited supply on VSYS, with no carrier fitted: press power for 1 s, then check 3V3 and idle current.
2. Flash over USB with no button held, with SW_BOOT1 held at power-on if the chip does not enumerate. Confirm PWR_HOLD reads high.
3. Time the shortest press that stays on, at 3.3 V and 4.2 V. The simulation gives under 0.4 s at worst, so a 1 s press is well clear. Run a USB upload with no button held, then confirm a watchdog reset reboots the board. Power off at least 20 times, on USB and on battery, and scope PWR_EN and PWR_LATCH_G: the board must stay off every time, since the simulation's margin has not been measured.
4. OLED and ICM-42670-P on I2C.
5. SK6812 RGB LED.
6. Buttons, using the pin table.
7. Charge status on STAT1 and STAT2, and VSYS sense.
8. DRV5055 and thumbstick on the ADC. Sweep the stick and the trigger together to measure how much each moves the other.
9. Solder the carrier on. Check UART traffic, CTS, SLEEP_RQ and ON/SLEEP, then XBee reset, then join a droid's network.

---

Power Architecture

• Battery → bq25185 (charger and power path, 4.2V, 500mA) → TLV62569 (3.3V buck) → 3V3 for everything, including the carrier.
• Power switch: soft latch (AO3401A PFET + AO3400A NFET) controls the TLV62569 EN pin. USB does not hold it on.
• Boot: press the power button for about a second → latch enables 3V3 → RXD0's pull-up holds the latch through D_OR2 as soon as the chip has power → firmware drives RXD0 high.
• Shutdown: firmware sees a hold on IO21 → graceful shutdown → RXD0 low → the gate decays through R_LATCH_G1 → C_LATCH_FB1 kicks it clear of threshold as PWR_EN falls → power cut after about 0.8-2.6 s, counted from release if the button is still held.
• bq25185 STAT1 and STAT2 go to IO15 and IO14 (open drain, 10kΩ pull-ups to 3V3).
• Battery voltage: VSYS through R_VSYS1/R_VSYS2 to IO7, reading VSYS/3.
• Below about 3.3 V on the battery the buck passes it through, and the DRV5055's minimum is 3.0 V, so the firmware cutoff should sit above that.

Power Management

Firmware implements tiered power saving:

• Active: full brightness OLED, normal polling rate
• Idle (short timeout): OLED dimmed via SSD1306 contrast command
• Idle (long timeout): OLED off
• Sleep: MCU sleep, wake on any button press via GPIO interrupt