LaserWandsUWB + TDMA plan · 12–16 players · ~450 m²

Hub / Hardware / LaserWand

hardware in progress

LaserWand

Player controller — UWB tag, IMU, laser tip, retro bands.

LaserWand concept
Product concept — not a photograph

Wand hardware — layout & markers

Matches wand_sw/main/wand_config.h. Button is GPIO15 active-low (switch to GND, MCU pull-up). Optical tracking = passive retroreflective tape on the shaft (zero electronics).

Three wands with retroreflective tape band patterns: tip double-band, unique spacing ID bands, and IR camera view showing collinear bright blobs
Approach A v2 — retro band markers. 3–4 Scotchlite-class bands over ~250–300 mm; tip double band resolves orientation; spacing cross-ratio = wand ID. Tip forward = RGB spell VFX only. v1 launches cameraless; bands ~50 CZK/wand.

Electronics (grip-heavy prototype) hw pending

  • Grip: Esp32-C6-Bug, 18650, USB-C pommel, LPKF carrier (BMI160, AO3400As, haptic, SW1 GPIO15 active-low)
  • UWB shelf (above fist): DWM3001CTR13 (UART), antenna toward tip, ≥5 mm keep-out; J3 SWD
  • Tip: KY-008 laser + WS2812 RGB VFX only
  • Shaft: Scotchlite-class retro bands only — no electronics

Optical markers (v2)

  • Passive Scotchlite bands — applied with fleet spacing jig
  • Ceiling: OV9281 + continuous 850 nm illuminator (v2 only)
  • Coloured grip wrap = staff ID only (cameras are mono IR)
LaserWand physical placement to scale: grip stack, UWB shelf above fist, hollow shaft with retro bands, tip laser and RGB
Physical placement (prototype) hw pending — 1 mm = 2 px. Mass (~95%) in the grip; BMI160 on the LPKF grip carrier; DWM3001CTR13 on a shelf just above the fist; tip = laser + WS2812 only; shaft = retro bands. LPKF ProtoMat E33 grip carrier.
Wand block schema: ESP32-C6, BMI160, DWM3001C, power chain, tip RGB and laser
Block schema. Solid blocks = current hardware (matches wand_config.h). Optical path = passive retro bands. C6-Bug integrates charge / protect / 3V3. Vector companion: wand-electronics-overview.svg.

DWM3001C — pinout, J-Link, UART

Same Qorvo module on the wand and every anchor. The ESP32-C6 talks to it over UART (115200 8N1, CRLF). The on-module nRF52833 is programmed over SWD from a SEGGER J-Link 10-pin socket. Canonical write-up: HARDWARE.md.

DWM3001C module pin diagram: 48 castellations, antenna at the top, pins 1–48 counter-clockwise
Figure 1 — DWM3001C pin diagram (Qorvo datasheet). Antenna at the top; pins 1–48 counter-clockwise from top-left. J-Link: SWD_CLK pin 2, SWD_DIO pin 3, VDD pin 12, RESET pin 47 (GND = 1 / 11 / 21 / 38 / 48). UART is not labeled TX/RX on the drawing — our firmware uses the right-side pads 32 (P0.15, UWB RX) and 34 (P0.19, UWB TX).

J-Link 10-pin socket · pin face

The programmer cable is a 10-pin header in two rows of five. Only five wires are used; unused holes stay open. Socket pin 1 is top-left (key notch on that side).

Flash: cd DWM3001C-starter-firmware-main && make build && make flash — see RUNBOOK A2.

UART — ESP32-C6 ↔ DWM

Cross-wired. Firmware: custom_board.h RX_PIN_NUMBER=15, TX_PIN_NUMBER=19.

ESP32-C6DWM3001CnRF
TX GPIO 4pin 32 UWB RXP0.15
RX GPIO 5pin 34 UWB TXP0.19
GNDGND—

ESP GPIO 4 / 5 — not DWM pin 3 (SWD_DIO) or DevKit header positions 3 / 4.

Why those pads

Pins 32 and 34 sit on the right edge of the figure, just above the bottom corner — generic P0.15 / P0.19 labels, not “TX/RX”. The DW3000 radio is driven inside the module firmware; the ESP never speaks SPI to it.

Same UART pins on wand (wand_config.h) and anchor (anchor_sw/main/main.c UART1).

Circuit schematic hw pending

Hierarchical KiCad design under hw_schematics/laserwand/v1/. Four sheets: root hierarchy, POWER, SENSORS, DRIVERS.

LaserWand root schematic: POWER, SENSORS, DRIVERS hierarchical sheets
Root — hierarchical blocks POWER → SENSORS → DRIVERS.
POWER sheet: battery and Esp32-C6-Bug
POWER — 18650 + Esp32-C6-Bug, 3V3 / VBAT / GPIO fan-out.
SENSORS sheet: BMI160 and DWM3001CTR13
SENSORS — BMI160 (I²C) + DWM3001CTR13 (UART) + SWD/ESD.
DRIVERS sheet: button, laser, haptic, WS2812, tip harness
DRIVERS — SW1, laser, haptic, WS2812, tip harness.

Wand build — placement rationale & soldering guide hw pending

Why every part sits where it sits, and how to solder/assemble the wand so it survives being swung all day. Full engineering detail lives in docs/WAND_BEST_PRACTICES.md — this is the field summary.

Esp32-C6-Bug integrates USB-C charging, cell protection, and 3V3 regulation; the discrete TP4056 + DW01/FS8205 + AP2112K guidance applies only to non-Bug / bare-module builds.

Placement — the physics of a swung stick

To-scale wand placement for the grip-heavy LPKF prototype
Same figure as Hardware — LPKF grip carrier prototype. hw pending

Mass lives in the grip

A wand pivots at the wrist; the effective swing weight of a part grows with the square of its distance from the pivot. Grams at the tip feel like a hammer, grams in the grip are nearly free. So the 18650 cell and control boards sit in the grip, and the tip carries only what must be there.

Discrete charger: next to USB-C, beside the cell

For non-Bug / bare-module builds, USB-C → TP4056 → 18650 is the highest-DC-current loop in the wand (up to ~1 A at the default Rprog; the sealed tube prefers 0.5 A — Rprog ≈ 2.4 kΩ — for less heat). Keeping the three adjacent at the pommel means short fat traces, no charge current routed down the shaft, less heat and noise near the emitters. Leave a few mm air gap between the charger and the cell wall — the linear TP4056 warms up while charging and Li-ion likes staying cool. The module must carry the DW01 + FS8205 protection pair (B+/B− pads).

IMU on the LPKF grip carrier

BMI160 lives in the grip with the mass stack — not at the tip. Hard casts no longer hammer the gyro at full lever arm; swing inertia stays low. Retrain gesture models for grip radius (tip-trained TFLite models will not transfer cleanly). Mount rigidly on the carrier, axes aligned with the shaft.

UWB shelf just above the fist

DWM3001CTR13 sits on a shelf immediately above the fist so the antenna clears the hand without putting mass at the tip. Keep a ≥5 mm keep-out forward of the antenna. The 18650 sits immediately aft of the UWB shelf and must not shadow or cross the antenna face; no battery wires across it either. J3 SWD on the side for nRF52833 flash. Antenna end toward the tip.

Retro bands define the aim vector (v2 optical)

Apply 3–4 Scotchlite tape bands over ~250–300 mm using a fleet spacing jig. Tip double band marks orientation; unique spacing = wand ID. Bands wrap fully around the shaft (visible from any roll). No solder, no MOSFETs, no sync. Optional coloured grip wrap for staff ID only (cameras are mono IR). v1 ships without cameras — bands are cheap future-proofing.

Emitters & button

The tip contains only the laser + forward RGB / WS2812 (spell colour only); its harness is VBAT/GND + laser gate + WS2812 data, with no I²C trunk to the tip. Button on grip → GPIO15 active-low with internal pull-up (switch-to-GND).

Soldering — the rules that matter

Never solder directly to a bare 18650. An iron hot enough to wet the nickel can boils the electrolyte under the cap. Use cells with factory-welded tabs/pigtails, or a proper cell holder, or a spot welder. Insulate the positive shoulder (kapton), and add the protection board before the pack ever meets the rest of the wand.
Every joint gets strain relief. The wand is a vibration machine; a soldered joint that carries mechanical load will fatigue and crack. Anchor each harness with a zip tie, adhesive pad or a dab of hot glue next to the joint so the wire, not the solder, takes the bending. Silicone-insulated stranded wire only — never solid core in the shaft.

Wire gauge by net

  • VBAT / GND trunk (battery → regulator): 24–22 AWG
  • Laser / WS2812 feed: 26–24 AWG, kept short
  • Signals (I²C, UART, GPIO, WS2812 data): 30–28 AWG
  • Colour code: red VBAT, black GND, yellow 3V3, then signals.

Route power and signals apart

  • Keep I²C in the grip; route the tip harness as VBAT/GND + laser gate + WS2812 data only.
  • Keep signals away from the laser/WS2812 feed; cross at 90° if needed.
  • WS2812 data: ~100 Ω series at the source end.
  • GND star at the grip — don't daisy-chain grounds through modules.

Decoupling

  • 100 µF + 0.1 µF at WS2812 / laser feed if needed.
  • 0.1 µF ceramic close to BMI160 VDD.
  • Retro bands are passive — zero tracking power on the wand.

Laser / optional haptic

  • Laser on GPIO20 (Class 1 — KY-008 attenuated to Class 1 AEL).
  • Haptic (GPIO 6 reserved): MOSFET + flyback diode if added.

Solder technique

  • Leaded 63/37 at ~330 °C, 1–2 s per joint; add flux, don't cook flux out.
  • Tin wire and pad first, then join — no "blob and hope".
  • Joint must be shiny and concave; grey/ball = cold joint, redo it.
  • Heat-shrink every splice (no bare tape); kapton over anything facing the cell.

Assembly & bring-up order

  1. Bench-test every module loose before anything enters the tube: flash ESP32-C6, talk to DWM3001CTR13 over UART (115200 8N1, CRLF), read BMI160 (0x68), blink WS2812.
  2. Power chain first, alone: verify Esp32-C6-Bug's integrated USB-C charge/protection/3V3 path. For non-Bug / bare-module builds, verify USB-C → TP4056 → protected cell → AP2112K-3.3, ≥600 mA (MCP1700 is undersized for WiFi + UWB peaks). Verify charge indication, charge termination, and 3.3 V under load before connecting any logic.
  3. Harness on the bench, full length, connectors on — test everything joined but outside the shell. Fix it now, not through a 15 mm hole.
  4. Slide in tip-first, anchor the emitter harness as you go; then seat the grip-carrier IMU rigidly and respect the UWB antenna keep-out. Align button and USB-C last.
  5. Apply retro bands from fleet jig; QC photo under bench IR camera (optional before v2).
  6. Re-run self-test after closing: I²C scan, UWB ranging, IMU stream, charge cycle. Label wand ID.
Use connectors, not solder-through joints, at module boundaries. JST-SH/PH pigtails on ESP32↔DWM3001CTR13, the tip's VBAT/GND + laser gate + WS2812 data harness, and battery→board mean a dead module is a 2-minute swap, not a desoldering session inside a tube. A fleet of wands lives and dies by serviceability.
Smoke-test rule: first power-up of any new board or harness happens on a current-limited bench supply (or USB with an inline meter), never straight from the Li-ion. The cell will happily deliver the failure current of any mistake.

Also see