Low-cost Open-source Microcontrollers to Build Automated Irrigation and Fertigation Systems Using Soil Moisture and Electrical Conductivity Sensors

Low-cost, open-source, embedded computers and microcontrollers can be used to build sensor-based automated systems to supply plants with on-demand water and fertilizer. The Raspberry Pi is a credit-card size embedded computer capable of running Linux and software written in Python, C, or C++. The Arduino is a prototyping platform centered on a microcontroller board and open-source development environment. We designed, built, and tested two independent systems using soil moisture and EC sensors to control irrigation and fertigation in geranium (Pelargonium ×hortorum Bailey) ‘Maverick Violet’ grown in 6-inch pots with peat:perlite substrate. Pore water electrical conductivity (EC), temperature, and volumetric water content (VWC) data were collected, added to a database, and made available online. The Raspberry Pi-based controller was designed to be self-contained, hosting its database and communicating with the internet by Wi-Fi. The Arduino-based controller was designed to be energy-efficient, and the data was transmitted to a database using XBee radios. A web application using twitter/BootStrap was written to query the databases and provided graphs online. Water and fertilizer solution with 600 mg L^-1 nitrogen (15-5-15 Cal-Mg) were provided by different drip irrigation lines, based on sensor measurements taken every 15 minutes. When the VWC dropped below the tested thresholds (from 0.15 to 0.5 m³ m^-3), the systems compared the EC to a threshold (from 0.5 to 3.0 dS m^-1). If the measured EC was higher than the threshold EC, a 24-VAC solenoid valve was opened and the plants were irrigated for 30 seconds; if the measured EC was lower than the threshold EC, another valve was opened and the plants were fertigated for 30 seconds. Our systems did not control irrigation and fertigation properly. The connection with the database was unstable due to the authentication protocol used by the network and the communication between sensors and microcontrollers was problematic, resulting in random loss of data and storage of unrealistic values. As a result, valves opened at the wrong times, not allowing a precise control. More studies are needed before these microcontrollers can be recommended for agricultural applications.