Flight data logging system using an Adafruit RP2040 microcontroller to capture IMU, GPS, barometric, and magnetometer data at up to 200Hz with Mahony AHRS filtering.
| Component | Description |
|---|---|
| Adafruit Feather RP2040 Adalogger | Main board with SD card slot |
| Adafruit BMP390 | Barometric altimeter (50Hz) |
| Adafruit LSM6DSOX | 6-DOF accelerometer/gyroscope (208Hz) |
| Adafruit LIS3MDL | 3-axis magnetometer (155Hz) |
| Adafruit Mini GPS PA1010D | GPS module (10Hz) |
- Adafruit Feather RP2040 Adalogger
- Adafruit BMP390
- Adafruit Mini GPS PA1010D
- Adafruit LSM6DSOX + LIS3MDL 9-DOF
The second enclosure revision features a full case for better protection and improved battery mounting.
| Part | Quantity |
|---|---|
| M2.5 heat inserts | 14 |
| M2.5 6mm standoffs | 12 |
| M2.5 6mm screws | 12 |
| M3 heat inserts | 4 |
| M3 6mm standoffs | 4 |
| M3 6mm screws | 4 |
| 50mm STEMMA QT cable | 3 |
| 32GB SD card (FAT32) | 1 |
| 3.7V 2000mAh LiPo battery (2-pin JST-PH) | 1 |
| Hot glue | - |
| Item | Est. Price |
|---|---|
| Adafruit Feather RP2040 Adalogger | $14.95 |
| Adafruit BMP390 | $10.95 |
| Adafruit LSM6DSOX + LIS3MDL 9-DOF IMU | $19.95 |
| Adafruit Mini GPS PA1010D | $29.95 |
| STEMMA QT 50mm cables (×3) | $2.85 |
| 32GB microSD card | ~$8 |
| 3.7V 2000mAh LiPo battery | $12.50 |
| 3D printed enclosure (PLA filament) | ~$3 |
| M2.5/M3 hardware (heat inserts, standoffs, screws) | ~$10 |
| Hot glue | ~$1 |
| Total | ~$113 |
Prices based on Adafruit retail (single-unit) as of early 2026. SD card and hardware prices vary by vendor.
rp2040-black-box/
├── firmware/
│ ├── quat_datalogger/ # Dual-core Mahony AHRS + SD card logging
│ ├── calibration/
│ │ ├── magnetometer_calibration/ # MotionCal-compatible mag streaming
│ │ └── accelerometer_calibration/ # Interactive 6-position accel cal
│ └── test/
│ ├── diagnostic/ # Serial & sensor diagnostic tool
│ └── sd_card_tester/ # SD card diagnostic tool
├── software/
│ ├── flight_visualizer/ # Python Streamlit web app
│ └── matlab-visualizer/ # MATLAB trajectory plotter & live orientation viewer
├── data/
│ └── samples/ # KITTI-derived test datasets
└── hardware/
└── boards/ # STEP files and fab prints
Add this URL to Arduino IDE > File > Preferences > Additional Boards Manager URLs:
https://github.com/earlephilhower/arduino-pico/releases/download/global/package_rp2040_index.json
Install Raspberry Pi Pico/RP2040/RP2350 by Earle F. Philhower, III
Select board: Tools > Boards > Raspberry Pi Pico RP2040/RP2350 > Adafruit Feather RP2040 Adalogger
Note: If no serial port appears, that's normal - RP2040 uses USB bootloader. A port will appear after first upload.
Install via Library Manager (Tools > Manage Libraries):
| Library | Purpose |
|---|---|
| SdFat | SD card operations |
| Adafruit BMP3XX | BMP390 altimeter |
| Adafruit GPS Library | PA1010D GPS |
| Adafruit LIS3MDL | Magnetometer |
| Adafruit LSM6DSOX | Accelerometer/Gyroscope |
| Adafruit AHRS | Mahony sensor fusion |
| Adafruit Unified Sensor | Common sensor interface (dependency) |
| Adafruit BusIO | I2C/SPI helper (dependency) |
Important: This project uses the
SdFatlibrary (not the standard ArduinoSD.h) for proper RP2040 Adalogger hardware support.
Hardware configuration:
- CS Pin: 23 (NOT pin 10!)
- SPI Bus: SPI1 @ 16MHz
- Library: SdFat (configured automatically in firmware)
Requirements:
- Format: FAT32 with 4KB allocation unit size
- Works with most SD/SDHC cards
Troubleshooting:
If SD initialization fails, check that the card is inserted and formatted as FAT32 with 4KB allocation units. Try cleaning the contacts or using a different card. The firmware retries 5 times automatically. Use firmware/test/sd_card_tester/ to diagnose issues.
Dual-core flight data logger with real-time AHRS filtering and signal processing.
Core 0 reads sensors and logs to SD at 100Hz. Core 1 runs the Mahony AHRS filter at 200Hz.
Signal processing pipeline:
- Adaptive Mahony filter (Kp ramps 10.0 → 1.0, Ki = 0.008 for drift correction)
- Sensor calibration: magnetometer hard/soft iron, gyro auto-cal at boot, accelerometer offsets
- Altitude: BMP390 hardware IIR → software low-pass filter → GPS complementary filter
- Auto sea-level pressure calibration from first good GPS fix
- GPS quality rejection (min satellites, max HDOP) with low-pass filtered speed/heading
- Automatic file splitting every ~1 GB to stay under FAT32 4 GB limit
Calibration constants are loaded from calibration.cfg on the SD card at boot (see Calibration). Filter tuning parameters are editable at the top of the firmware file.
Two standalone sketches in firmware/calibration/ for computing sensor calibration values:
- magnetometer_calibration — Streams
Raw:/Uni:data over serial in the format expected by MotionCal. Rotate the device in all orientations until the sphere fills in, then copy the hard/soft iron values intocalibration.cfg. - accelerometer_calibration — Interactive serial-guided 6-position calibration. Prompts you through each orientation, collects samples, and writes the computed offsets directly to
calibration.cfgon the SD card.
Serial and sensor diagnostic tool that tests each I2C sensor individually with detailed bus diagnostics. Upload and open Serial Monitor at 115200 baud to verify all sensors are detected and responding correctly.
Diagnostic tool for verifying SD card compatibility. Tests initialization, read/write, and speed benchmarking.
Output: 20-column CSV at 100Hz
| Column | Description | Units |
|---|---|---|
| ms | Timestamp | milliseconds |
| qw, qx, qy, qz | Orientation quaternion | - |
| ax, ay, az | Acceleration | m/s² |
| gx, gy, gz | Angular velocity | rad/s |
| mx, my, mz | Magnetic field | uT |
| alt | Filtered altitude | meters |
| gps_fix | GPS lock status | 0/1 |
| lat, lon | GPS coordinates | decimal degrees |
| speed | Filtered GPS speed | knots |
| heading | Filtered GPS heading | degrees |
File naming: Timestamps use UTC from the GPS module's battery-backed RTC.
- GPS time available:
20260130_143022_0.csv,..._1.csv,..._2.csv - GPS time unavailable:
log0_0.csv,log0_1.csv, ... - Header:
# Start: YYYY-MM-DDTHH:MM:SS UTC
Calibration values are stored in a calibration.cfg file on the SD card. The firmware reads this file at boot and applies the values automatically. If the file is missing, compiled defaults (zeros / identity) are used.
The firmware averages 500 samples at boot while stationary. Keep the device still for 5 seconds after power-on. The computed bias is printed to serial.
To skip the wait, note the printed values and hardcode them in the firmware:
static const bool GYRO_AUTO_CAL = false;
static float gyro_bias[3] = { -0.0023f, 0.0041f, -0.0012f };This is the most important calibration. Without it, heading (yaw) will be unreliable.
- Upload
firmware/calibration/magnetometer_calibration/to the board - Open MotionCal and select the serial port
- Slowly rotate the device through all orientations (figure-8 pattern) until the sphere fills in
- Read the values from MotionCal and add them to a new file called
calibration.cfgin the root of the SD card, using this format:
mag_hard_iron = <X>, <Y>, <Z>
mag_soft_iron = <M00>, <M01>, <M02>, <M10>, <M11>, <M12>, <M20>, <M21>, <M22>
Do this calibration with the device in its final enclosure. Nearby metal magnets, and mechanical stresses may affect the result.
- Upload
firmware/calibration/accelerometer_calibration/to the board - Open Serial Monitor at 115200 baud
- Follow the prompts: place the device in each of 6 orientations (X up/down, Y up/down, Z up/down) and press Enter
- Hold the device perfectly still during each 2-second measurement
- The computed offsets are automatically saved to
calibration.cfgon the SD card
The sketch preserves any existing entries (e.g., magnetometer values) in the config file.
The firmware auto-calibrates from the first good GPS fix. No action needed.
For manual override, add to calibration.cfg:
sea_level_hpa = 1018.5
And optionally disable auto-cal in the firmware:
static const bool BARO_GPS_AUTO_CAL = false;After calibrating, run these checks on the quat_datalogger.ino firmware.
| Test | Procedure | Expected Result |
|---|---|---|
| Static | Place flat and still for 30s after boot | Quaternion settles to ~[1, 0, 0, 0] |
| Rotation | Rotate 90° and return | Quaternion returns to original value |
| GPS walk | Walk a known rectangle outdoors | Lat/lon tracks your path, altitude is stable |
Interactive web-based visualization.
pip install streamlit pandas numpy plotly scipy
cd software/flight_visualizer
streamlit run app.pyLive demo: rp2040-fdr-visualizer.streamlit.app
Desktop visualization with satellite map overlays, dead reckoning, and sensor charts. Requires MATLAB with Navigation Toolbox (quaternion, rotateframe).
Open MATLAB → navigate to software/matlab-visualizer/ → run trajectoryvisualizer.m
A file picker dialog opens — select any 20-column CSV from data/samples/ or from the SD card.
Displays:
- Summary statistics popup (duration, distance, speed, altitude range)
- 2D GPS + dead-reckoned track on satellite basemap (time-gradient colored, circle markers for GPS, diamond markers for dead reckoned)
- 3D trajectory with gradient-colored line (North/East/Altitude in meters)
- 4-panel sensor charts: Speed & Altitude, Quaternion, Accelerometer, Gyroscope
When GPS fix is lost, the visualizer dead-reckons position by rotating body-frame acceleration to NED using the quaternion, subtracting gravity, and integrating velocity/position. Barometric altitude is used for the vertical axis throughout.
Real-time 3D orientation display from live serial data. Connects to the RP2040 over USB and renders orientation using poseplot. Requires MATLAB with Navigation Toolbox and R2019b+ (serialport).
Open MATLAB → navigate to software/matlab-visualizer/ → run liveorientationvisualizer.m
Select a serial port and baud rate (default 115200) when prompted. The figure updates in real time as the device streams quaternion data. Close the figure window to disconnect.
- Ruggedize with TPU secondary enclosure for crash survivability
- Upgrade GPS module to use a larger antenna for better signal
- Design custom PCB integrating RP2040 and all sensors onto a single board
- Add Bluetooth for wireless flight data streaming
- Consider BNO085 IMU: Replace LSM6DSOX + LIS3MDL with BNO085 for onboard hardware EKF fusion — reduces CPU load with hardware-accelerated quaternion output and automatic calibration
- Optimize battery usage and create power consumption metrics
- Add internal telemetry: battery life, board temperature
- Add Simulink visualization for real-time 3D flight simulation and analysis
- Transition to an AI-optimized microcontroller for on-device inference
- AI-enhanced dead reckoning to reduce position drift when GPS is unavailable
- Altitude fusion model to correct for local pressure variations
- Predict remaining battery life based on flight behavior patterns
- Auto-label flight phases (takeoff, cruise, landing, idle) from sensor signatures
- Anomaly detection to flag unusual vibration or orientation patterns
The first enclosure prototype — open-top design with standoff-mounted boards.
