SensorNodeProject

ESP32 Low-Power Weather Node

A custom ESP32-based weather sensor node targeting six months of battery life and less than 50 µA system current during deep sleep. The project was motivated by gaps observed while mapping environmental data across the Pacific Northwest during a National Weather Service internship.

Planned firmware will use ESP-IDF and statically allocated FreeRTOS tasks for sensor sampling, local logging, wireless transmission, and deep-sleep control. Later revisions may add transmission integrity checks and adaptive sampling based on changing weather conditions.

Status: In development. The PCB has been assembled and confirmed working. BME280 I2C communication is being tested. Chip ID was verified. Compensated temperature/pressure/humidity values are in progress. Project Status.

Why this project

A low-cost, low-power sensor node could make denser local observations more practical. This project explores the engineering requirements for that type of device.

Hardware

The following values are estimates based on component datasheets and haven't been verified on assembled hardware.

Target: <50 µA average system sleep current.

Design decisions

ESP-IDF over Arduino. ESP-IDF was selected to provide direct control over FreeRTOS task creation, static memory allocation, peripheral configuration, and ESP32 sleep/wake behavior.

LDO instead of a buck-boost regulator. An LDO offers a simpler design and lower quiescent current, though reduced usable battery voltage as the LiPo approaches the regulator’s dropout region. The RT9080-33GJ5 was selected for its ~2 µA quiescent current and 600 mA output rating. Did not use MCP1700 because of its 250 mA rating, which isn't enough for WiFi transmission.

No onboard current-sense IC. It needs to measure sleep current in the microamps as well as much larger Wi-Fi transmission currents. Instead of a dedicated IC, test points are placed across a removable jumper. The jumper can be replaced by a larger resistor for controlled sleep measurements or a lower one for active measurements. An INA219 did not provide sufficient dynamic range for both measurements with a single resistor, while the INA228 drew too much current.

Adaptive sampling instead of a fixed interval. The planned firmware will have a rolling window of pressure measurements, estimate the rate of change, and adjust the next sampling interval. Faster changes, such as a cold front or wind shift, will mean more frequent sampling; stable conditions mean longer sleep intervals.

Fully static FreeRTOS allocation. Every task and queue will be created statically to keep predictable memory usage.

Repo structure

├── firmware/       # ESP-IDF project
├── hardware/       # KiCad schematic + PCB layout
└── README.md

Project status

License

MIT — see LICENSE.