From Load Cell to Cloud User Interface: End-to-End Engineering of a Modular, Multi-Channel Pot-Scale IoT Weighing Lysimeter

This thesis presents the design, implementation, and experimental validation of an IoT, high-precision, modular weighing lysimeter to be used inside a plant growth chamber. Weighing lysimeters are used to precisely quantify plant water losses (evapotranspiration) and growth by monitoring the changes in the weight of the pots used to grow them. The requirements called for a system capable of monitoring four pots concurrently and transmitting the data to a central server using a Raspberry Pi single-board computer. The proposed architecture couples a commercial mechanical structure (CELMI LP7611 single-cell load frame) with two custom analog front-end alternatives, based respectively on the Nau7802 IC and the ZSSC3224 IC, which digitize the Wheatstone-bridge load-cell analog signals. Both proposals feature input filtering, dedicated regulated power for the analog section, and a PCA9615 differential I2C transceiver IC that reliably transmits the digitized data over Cat5 Ethernet cabling to the Pi_master custom Raspberry Pi shield. This shield, using a MAX7320 I2C port expander IC, multiplexes all the scale signals to a single Raspberry Pi I2C port. The Pi_master also features two STEMMA-QT ports, a reliable and proven standard that can be used to expand the capabilities of the system thanks to the wide commercial availability of sensor modules and actuators featuring this interface. The flexibility of the system was demonstrated when it was expanded to accommodate an additional irrigation control system based on a commercially sourced, inexpensive relay board. This board is used to control small pumps to selectively irrigate each pot. It was adapted using an MCP23017 I2C GPIO expander breakout board through the STEMMA-QT port. On the software side, an edge stack written in Python and running on the Raspberry Pi handles sensor polling, calibration, channel selection, irrigation actuation, and MQTT publishing. A central IoT server, running Eclipse Mosquitto, Telegraf, InfluxDB, and Grafana, is installed in Docker on an Amazon Lightsail cloud Virtual Private Server. This stack ingests, stores, and visualizes the data generated by each scale system. Performance validation demonstrates stable and precise weight acquisition, robust communication, and ease of use, proving the excellent capabilities of the system, which rivals equivalent commercial solutions at a much higher cost. These results highlight its potential as an enabling tool for the testing and validation of wearable plant sensors and for developing irrigation algorithms.