A Zephyr RTOS application that reads frames from an MLX90640 32×24 thermal camera and renders them as an ASCII heat map on the RTT/UART log.
| Component | Details |
|---|---|
| MCU board | Nordic Semiconductor nRF54L15 DK |
| Sensor | Melexis MLX90640 — 32×24 IR thermal array |
| Sensor I2C address | 0x33 (factory default) |
| Sensor supply voltage | 3.3 V |
Important — DK voltage configuration: The MLX90640 module requires a 3.3 V supply. The nRF54L15 DK's GPIO header voltage must be set accordingly before powering the sensor. By default the DK is shipped with a VDD rail setting of 1.8V, so you'll need to do the follwing change to supply the camera module properly! Open nRF Connect for Desktop → Board Configurator, select your DK, and set VDD (nPM VOUT1) to 3.3 V. Configure your board by clicking "Write config". Please note that this changes the VDD rail of all GPIO ports, just in case you have other HW connected to it make sure that it is 3.3V compliant.
The sensor is connected to the I2C21 (TWIM1) peripheral of the nRF54L15 DK.
| Signal | nRF54L15 DK pin | Header |
|---|---|---|
| SCL | P1.11 | General-purpose header (no board hardware) |
| SDA | P1.12 | General-purpose header (no board hardware) |
| VDD | 3.3 V | Use Board Configurator to set VDD to 3.3V |
| GND | GND |
Note: Both pins are free general-purpose header pins on the nRF54L15 DK — no PCB modification is required. Pull-up resistors are enabled in firmware via
bias-pull-upin the device tree overlay. The I2C bus runs at 400 kHz (Fast Mode).
- Visual Studio Code with the nRF Connect for VS Code extension pack installed.
- nRF Connect SDK v3.2.4 — the application was built and tested against this version. Select it in the nRF Connect extension's Toolchain Manager before building.
- Open VS Code in this folder.
- In the nRF Connect side-panel, click Add Build Configuration.
- Choose the board
nrf54l15dk/nrf54l15/cpuapp. - Leave all other settings at their defaults and click Build Configuration.
- After a successful build, click Flash to program the DK.
The application overlay (app.overlay) and project configuration (prj.conf) are picked up automatically by the build system — no extra CMake variables are required.
The application prints a statistics header followed by a 24-row ASCII heat map every 5 seconds:
[00:00:27.514,721] <inf> main: +--------------------------------+
[00:00:27.521,183] <inf> main: | MLX90640 32 x 24 greyscale |
[00:00:27.527,722] <inf> main: | min 22.7 C |
[00:00:27.534,423] <inf> main: | max 35.7 C |
[00:00:27.541,128] <inf> main: | mean 26.9 C |
[00:00:27.547,831] <inf> main: | span 12.9 C |
[00:00:27.554,529] <inf> main: +--------------------------------+
[00:00:27.561,074] <inf> main: |+ .. . . .. .|
[00:00:27.567,603] <inf> main: |+ . . ....... .. ..|
[00:00:27.574,131] <inf> main: | ..****+.. .. |
[00:00:27.580,662] <inf> main: | . ..+oooo++.. ... .|
[00:00:27.587,194] <inf> main: | . . .*oo##ooo... . |
[00:00:27.593,722] <inf> main: | . . .ooo#o##o+..... |
[00:00:27.600,253] <inf> main: |. . .+o#######++... .|
[00:00:27.606,785] <inf> main: | .*#o##o#o#*..... .|
[00:00:27.613,313] <inf> main: | . .+**o*++++++ . ..|
[00:00:27.619,843] <inf> main: |.. . ..+++oo+++**+.... .|
[00:00:27.626,384] <inf> main: | . .+**##***oo*. .. |
[00:00:27.632,911] <inf> main: | .*ooo##oooo*......|
[00:00:27.639,439] <inf> main: |. . .*######o#o+.. ..|
[00:00:27.645,970] <inf> main: | . . .*oo######*+.. .|
[00:00:27.652,499] <inf> main: |. . . .+########++.... .|
[00:00:27.659,029] <inf> main: |. +o######o........|
[00:00:27.665,561] <inf> main: |.. .. . oo#####o..... . |
[00:00:27.672,090] <inf> main: |... . . .. .+oo####o.. .... |
[00:00:27.678,618] <inf> main: |.. . . . .+o######+.... ..|
[00:00:27.685,148] <inf> main: | ....... ... ..*#@####++......|
[00:00:27.691,675] <inf> main: |.. ... . .+.*o####****++....|
[00:00:27.698,206] <inf> main: |..... . . ..+*oo##oo*oo***++.+|
[00:00:27.704,739] <inf> main: |.. . . ...++***##oooooooooooo++|
[00:00:27.711,267] <inf> main: |+..... ...+*****ooooo****o*o****|
[00:00:27.717,785] <inf> main: +--------------------------------+
| Character | Meaning |
|---|---|
(space) |
Background / very cold |
. |
Cool |
+ |
Below ambient |
* |
Near ambient |
o |
Warm |
# |
Hot |
@ |
Very hot |
In addition to the UART/RTT log output, the firmware continuously streams live thermal frames over Bluetooth LE using the Nordic UART Service (NUS). This allows a PC or phone to receive and display the thermal image in real time.
| Parameter | Value |
|---|---|
| Role | Peripheral (advertises as "ThermalCam") |
| Service | Nordic UART Service (NUS) |
| NUS TX characteristic UUID | 6e400003-b5a3-f393-e0a9-e50e24dcca9e |
| LL data length | 251 bytes (DLE enabled) |
| ATT MTU | 247 bytes |
| Frame rate | One frame every 5 seconds (see MLX90640_READ_INTERVAL_S) |
Each frame is a fixed 776-byte packet transmitted as a series of NUS TX notifications. All multi-byte integers are big-endian.
| Offset | Size | Type | Content |
|---|---|---|---|
| 0 | 2 | uint8[2] |
Start-of-frame marker: 0xFF 0xFE |
| 2 | 2 | int16 |
t_min × 10 — coldest pixel temperature × 10 (e.g. 227 = 22.7 °C) |
| 4 | 2 | int16 |
t_max × 10 — hottest pixel temperature × 10 |
| 6 | 768 | uint8[768] |
Pixels in row-major order (24 rows × 32 cols). 0 = coldest, 255 = hottest. |
| 774 | 2 | uint8[2] |
End-of-frame marker: 0xFF 0xFD |
The pixel values are linearly scaled between t_min and t_max within each frame, so the full 0–255 range is always used regardless of absolute temperature.
thermal_viewer.py is a Python script that connects to the device over BLE and displays the thermal image in a real-time matplotlib window.
- Python 3.9+
- Packages:
bleak,numpy,matplotlib >= 3.9
Create a virtual environment and install the dependencies (run once):
python -m venv .venv
.venv\Scripts\Activate.ps1 # PowerShell
# or: .venv\Scripts\activate.bat # cmd.exe
pip install "bleak" "numpy" "matplotlib>=3.9"Note — PowerShell execution policy: If
Activate.ps1is blocked, run once:Set-ExecutionPolicy -Scope CurrentUser RemoteSigned
- Power the nRF54L15 DK and make sure the firmware is running (the UART log should show
Advertising as "ThermalCam"). - Activate the virtual environment (if not already active):
.venv\Scripts\Activate.ps1
- Run the viewer:
python thermal_viewer.py
The script will scan for a BLE device named ThermalCam, connect automatically, start receiving notifications, and open a live colour-mapped plot window (inferno colour map, cold = dark, hot = bright). The title bar of the plot updates with the current min / max / span temperatures for each received frame.
Close the plot window to disconnect and exit.
The BLE stack (bleak / asyncio) runs in a background thread; the matplotlib GUI runs on the main thread. This prevents the GUI event loop from starving the asyncio event loop, which is a common pitfall on Windows.