Enhanced Gas Monitoring for Automated Greenhouses

This paper presents enhanced methods for monitoring gas concentration in an automated greenhouse, addressing the shortcomings of a previous system that uses a MICS-6814 sensor paired with a 10-bit Successive Approximation Register (SAR) Analog-to-Digital Converter (ADC). Although the original design was cost-effective, it faced several limitations, including leakage currents, sequential measurement delays, and vulnerability to electrical noise. To overcome these issues, two improved designs were proposed based on MAX11060, a 16-bit Sigma-Delta ADC. The first design introduces current injection through transistors, while the second, preferred, approach employs a single voltage reference and takes advantage of the ADC’s simultaneous sampling capability, resulting in improved measurement accuracy with a modest increase in system cost. Additionally, a machine learning model built using TensorFlow Keras was trained to classify air quality, achieving a test accuracy of $\mathbf{8 8 \%}$ on real-word sensor data. The last solution was implemented on a printed circuit board (PCB) and tested in parallel with the original system under identical conditions to evaluate and compare their performance.