Skip to content

Hardware — bring-up procedure & log

The procedure to take a freshly-fabricated rev-B board from "just came back from the SMT line" to "validated, can hit a desk and start shipping units off it." Run every check on every new board before declaring it shippable; the procedure surfaces solder defects, BOM swaps and firmware/hardware mismatches early.

The page also serves as the running engineering log — each board that goes through bring-up appends one entry to the log table at the bottom.

Time budget: ~45 minutes per board on the bench, including the first power-on, USB enumeration, every bus loopback, RGB calibration and the entry write-up.

Pre-power-on (DMM only, ~10 min)

Before connecting USB at all — these checks catch the kind of solder bridge that destroys an LDO in 2 ms.

  1. Visual inspection under 10× loupe:
  2. QFN-80 RP2354B fillets even on all four sides
  3. No bridges between USB-C VBUS / D+ / D- / GND
  4. LDO IC orientation matches silkscreen (pin-1 dot)
  5. Electrolytics polarity correct

  6. No solder balls on the QSPI flash side

  7. DMM continuity (USB cable disconnected, no power):

Probe between Expected If wrong
VBUS pad ↔ GND open (≫ 100 kΩ) short — stop, find it
+3V3 net ↔ GND open (≫ 100 kΩ) short — LDO output
RP2350 VDD_CORE ↔ GND open (≫ 1 MΩ) short — die-side bridge
USB D+ ↔ D- open (≫ 1 MΩ) bridge across pair
RP2350 VDD_IO pins ↔ each other shorted (same rail) one trace open
  1. Mechanical jumpers are seated in their factory positions (CAN 120 Ω closed, RS485 120 Ω closed).

If any pre-power check fails → stop, fix, re-run pre-power. Do not power up a board with a known short to ground on a power rail.

First power-on (~5 min)

Cable: USB-C from a current-limited supply or a host that doesn't mind a brown-out. Reasonable bench setup is a USB-C power meter inline so you can read VBUS and current draw at the same time.

Plug in →  measure within first second:

  VBUS        4.85 – 5.15 V       (USB-C nominal, no PD)
  IBUS_in     50 – 120 mA         (boot + USB enumeration burst)

After the firmware boots (~150 ms):

  +3V3        3.27 – 3.33 V       (LDO output, idle)
  IBUS_in     35 – 80 mA          (steady state, LED dimmed)
  LDO surface 25 – 35 °C          (room-temp + 5–10 °C, no bus load)

If IBUS_in > 200 mA at idle → unplug immediately, recheck for short / wrong LDO / wrong RP2350 brick.

If +3V3 reads outside 3.27–3.33 V → either the LDO is wrong, the sense divider is wrong, or VBUS is sagging (try a different cable).

The RGB LED should run a heartbeat scene shortly after boot (slow pulse on green during the watchdog grace window). If it stays dark → SWD attach to find out where the firmware crashed.

SWD attach + serial stdio (~5 min, optional)

Skip this section if power-on + USB enumeration both work the first time — most boards do, and the SWD step is for diagnosing boards that don't.

Tools: any picoprobe-class debugger on the 5-pin debug header (SWDIO + SWCLK + GND), plus any 3.3 V serial adapter on the same header's UART1 stdio pair (GP4/GP5 = TX/RX).

# OpenOCD via picoprobe at /dev/ttyACM0:
openocd -f interface/cmsis-dap.cfg -c 'adapter speed 5000' \
        -f target/rp2350.cfg

# In a second terminal — serial console for printf() output:
picocom -b 115200 /dev/ttyUSB0

The firmware's printf() output names every subsystem as it inits; a halted board reveals where it stopped.

USB enumeration check (~2 min)

lsusb -d 0545:0101 -v | head -40

Expected highlights:

idVendor           0x0545 BAUER GROUP
idProduct          0x0101
bcdDevice            1.00
iManufacturer       1 BAUER GROUP
iProduct            2 RPBridge
iSerialNumber       3 RPB-<16 hex>
bNumConfigurations  1

dmesg | tail should show cdc_acm binding the four CDC interfaces plus the gs_usb binding the CAN interface (no rpbridge.ko on the host yet — that's the next step).

If a CDC fails to enumerate → check the per-interface descriptor strings via lsusb -v; if descriptors look fine, the firmware shipped wrong USB-side endpoint allocation.

Driver bind verification (~3 min)

Install the kernel module (see build.md) and replug:

sudo modprobe rpbridge
sudo udevadm trigger
dmesg | tail -20

Expected dmesg entries (in order):

rpbridge 1-x:1.4: probing matched USB interface
rpbridge 1-x:1.4: capability map: 0x... bytes
rpbridge_gpio rpbridge.gpio.0: registered (32 lines)
rpbridge_i2c rpbridge.i2c.0: registered as i2c-N
rpbridge_i2c rpbridge.i2c.1: registered as i2c-N
rpbridge_spi rpbridge.spi.0: registered as spidevN.M
…

The auxiliary children spawn in capability-map order — use the same map to verify all expected subsystems registered.

Per-bus smoke tests (~15 min, the bulk of bring-up)

Each test exercises one signal path end-to-end. Run them in this order — earlier tests rule out classes of failure (USB-stack, power, MCU itself) before you blame later results on a single bus.

LED + button (no test gear required)

rpbridge-ctl led set --rgb FF0000          # solid red
rpbridge-ctl led set --rgb 00FF00          # solid green
rpbridge-ctl led set --rgb 0000FF          # solid blue
rpbridge-ctl led set --rgb FFFFFF          # full white

While the LED is on white, note the brightness of each channel visually — they should look balanced. Wide imbalance (e.g. green overpowering red and blue) means the per-channel calibration in firmware/src/main.c (RPBRIDGE_LED_CAL_R/G/B) needs adjustment for this board's LED batch.

# button:
rpbridge-ctl gpio --bus 0 watch 25         # press the front-panel button
                                            # — should print "GP25 = 0"
                                            # while held, "GP25 = 1" released

I²C buses (Qwiic + 4-pin headers)

Plug a known-good I²C device (BME280, AHT20, anything with a fixed address) into Qwiic 1 (bus 0) and run:

i2cdetect -y 0
# expected: device address visible at the documented row/col

Repeat for bus 1 (Qwiic 3 connector). On rev-B both buses should respond; on a Pico 2 variant only bus 0 exists.

SPI loopback

Wire MOSI ↔ MISO on the PMOD (J8) — no external device needed:

spidev_test -D /dev/spidev0.0 -s 1000000 -p 0x12,0x34,0x56,0x78
# expected RX: 12 34 56 78  (loopback echo)

Repeat with -s 24000000 to confirm the high-speed path.

CAN

Connect the CAN port to a CANable / candle-bus reference adapter, both with 120 Ω terminators in (jumpers closed on the rev-B board).

sudo ip link set can0 type can bitrate 500000
sudo ip link set can0 up
candump can0 &
cansend can0 123#DEADBEEF                  # send from the bridge
                                            # CANable should echo it

Frames should appear on both ends. Bus errors (error-active status in ip -details link show can0) usually mean a mismatched or absent terminator on one end.

UART transceivers (rev-B only)

# RS232 — loopback DE-9 pin 2 ↔ 3:
picocom -b 115200 /dev/ttyACM0
# type chars; you should see them echoed.

# RS485 — loopback A ↔ B with 120 Ω across, or use a USB-RS485 dongle:
picocom -b 115200 /dev/ttyACM1

For each phy bank, the firmware's auto-DTR-switch routes UART0 HW to whichever CDC currently has DTR=1. picocom raises DTR by default → opening the device picks that bank.

ADC

# four channels A0..A3 — connect a known voltage (e.g. divider on +3V3):
iio_info | grep -A 4 'rpbridge'

# or one-shot read:
rpbridge-ctl adc read 0
rpbridge-ctl adc read 1

Compare the returned value to a DMM measurement on the same pin — within ±2 % at room temperature.

PIO applets (1-Wire, WS2812, DMX, I²S)

These need specific external hardware:

  • 1-Wire: DS18B20 sensor on the 1-Wire header with the 4.7 kΩ pull-up. cat /sys/bus/w1/devices/28-*/temperature.
  • WS2812: 8-pixel strip on the LED-strip header. rpbridge-ctl led strip set --rgb 00FF00.
  • DMX: external RS-485 transceiver (MAX485) wired to the DMX XLR, into a DMX fixture (LED par, fog machine). rpbridge-ctl dmx set 1 255 (channel 1 → full).
  • I²S: an I²S DAC (UDA1334 / PCM5102) on the I²S header. Pipe a sine wave: see the rpbridge-ctl i2s subcommand.

Final checks (~5 min)

  1. Hot-soak: leave the board running for 30 min with the CAN bus echoing traffic and an I²C scan loop. LDO surface should stay below 60 °C, IBUS_in shouldn't drift more than 10 mA.
  2. Hot-replug: yank the USB cable, plug it into a different port, confirm enumeration recovers without dmesg errors.
  3. Sleep / wake: suspend the host, wait 30 s, resume. The device should re-enumerate cleanly. (Some hosts don't VBUS-cycle on resume — that's a host quirk, not a board defect.)
  4. rpbridge-ctl info snapshot: capture for the log entry below.

Bring-up log

Each row corresponds to one board. Fill from top to bottom; entries that don't apply (no I²S DAC on the bench today) get n/a.

Template

## YYYY-MM-DD — RPB-<serial>

| Check                   | Result                              |
| ----------------------- | ----------------------------------- |
| Pre-power continuity    | clean / detail                      |
| IBUS_in idle            | … mA                                |
| +3V3 idle               | … V                                 |
| LDO surface 30 min soak | … °C                                |
| USB enumeration         | clean / dmesg detail                |
| Driver bind             | all subsystems / list missing       |
| LED RGB calibration     | R=…% G=…% B=…% for neutral white    |
| Button GP25             | press / release events seen         |
| I²C bus 0 (Qwiic 1)     | device at 0x… / "no devices"        |
| I²C bus 1 (Qwiic 3)     | device at 0x… / "no devices"        |
| SPI loopback @ 1 MHz    | OK / mismatch                       |
| SPI loopback @ 24 MHz   | OK / mismatch                       |
| CAN echo @ 500 kbit/s   | both directions / one-way / none    |
| RS232 loopback          | OK / no echo / framing errors       |
| RS485 loopback          | OK / no echo / DE-timing issue      |
| UART2 PIO loopback      | OK / no echo                        |
| ADC channel sweep       | within ±… % of DMM                  |
| 1-Wire DS18B20          | … °C / not detected                 |
| WS2812 strip            | colours render / pixel offset / n/a |
| DMX fixture             | level changes / n/a                 |
| I²S DAC                 | tone audible / silent / n/a         |
| Hot-replug              | recovers / errors                   |
| Suspend/resume          | recovers / errors                   |
| Issues observed         | …                                   |
| Changes planned         | …                                   |

Entries

No entries yet — the first rev-B run is pending. Append above with a copy of the template, one section per board.