Skip to content

← Docs index

08 - Troubleshooting

XIAO resets when motors start

Likely causes:

  1. battery voltage sag;
  2. insufficient bulk capacitance;
  3. long/thin battery wiring;
  4. motor noise coupling into the XIAO supply;
  5. weak/aged LiPo.

Actions: inspect VBAT with an oscilloscope if available, add/relocate bulk capacitor, shorten wiring, test with a known-good high-C battery, and reduce motor duty during diagnosis.

Camera works on USB but fails on battery

Measure battery rail during capture. The camera/Wi-Fi load produces substantial transient current. Check antenna, PSRAM settings and power integrity. Current Seeed documentation lists roughly 110 mA typical Wi-Fi-active consumption for the Sense expansion and much higher short peaks in camera operation.

One motor starts at boot

Immediately disconnect power. Check AO3400 pin orientation, solder bridges and missing gate pulldown. Never troubleshoot this with a prop fitted.

Craft flips instantly

Do not tune PID. Check, in order:

  1. motor numbering;
  2. prop location CW/CCW;
  3. motor rotation direction;
  4. IMU axis orientation;
  5. mixer configuration.

Strong vibration / jello

  • replace bent/damaged prop;
  • check motor shaft runout;
  • make sure the motor is tightly retained;
  • use TPU camera cradle;
  • verify frame arms are not cracked;
  • try another motor if one has worn bearings/bushings.

Wi-Fi control becomes poor when streaming

  • lock video to QVGA;
  • one viewer only;
  • lower camera frame rate/JPEG quality;
  • keep control/telemetry UDP traffic prioritized;
  • use the supplied U.FL antenna with clear orientation;
  • for a later V2, consider independent RC/SBUS so flight control is not dependent on the video WLAN link.

Camera initializes but every captured frame is NULL

esp_camera_init() returns ESP_OK (the sensor answers on SCCB/I2C), but esp_camera_fb_get() reliably returns NULL afterward.

Observed on a real XIAO Sense unit at the default 20 MHz XCLK. Two independent fixes, both worth trying:

  1. Lower xclk_freq_hz to 10 MHz and discard one warm-up frame right after init (see firmware/bench_test/camera.cpp and firmware/camera_wifi_test/main.cpp) — this alone made capture reliable on the affected unit.
  2. If frames are still intermittently NULL after the XCLK fix, suspect a marginal FPC ribbon-cable seat on the Sense camera module rather than a firmware issue. Re-seating or replacing the camera module resolved a case where roughly 1 in 10–20 frames still failed even at 10 MHz.

CAMERA_GRAB_WHEN_EMPTY instead of CAMERA_GRAB_LATEST also avoids returning a buffer that a slow consumer is still holding, but did not by itself fix the NULL-frame issue described here.

ESP32 build breaks after library/core update

Return to the last recorded tested toolchain. Flight firmware should be treated like embedded production software: pin the framework and library versions and re-run the full validation plan after any update.

← 07 - Safety and Legal | Docs index | Next: 09 - Amazon.de Order List →