Design of soil moisture sensor based on phase-frequency characteristics of RC networks.

Abstract

Dielectric-based methods are widely used due to their non-destruction, efficiency and accuracy. The capacitance of the probe on the sensor is affected by the soil moisture. Therefore the mathematical model can be built between the capacitance of the sensor and the soil moisture. In this paper, a new soil water content sensor based on the phase-frequency characteristic of RC network is proposed. The sensor consists of four parts, that is a VHF oscillator, a phase-detecting circuit, a first-order RC low-pass circuit, and a probe. The VHF oscillator outputs a frequency-specified f* signal to drive the RC network, and the capacitor C of the first-order RC low-pass network is replaced by the capacitance of the probe of the sensor. Moreover, the changes of capacitance of the probe brought by the change of the soil moisture will cause a significant change in the phase-frequency response of the RC network. The AD8302 phase-detector is used to measure the change of the phase-frequency response of the RC network by converting the phase angle of the RC network to a voltage signal. Thus, the relationship between the soil moisture content and the output voltage signal can be built to estimate water content in soil. Compared with existing published works on the theoretical implementation which has low accuracy and sensitivity of the sensor, the proposed sensor is optimized by the following steps: 1) The measurement equivalent circuit model of the first-order RC low-pass circuit along with the input equivalent circuit of AD8302 is built; 2) The relationship between the output voltage signal of AD8302 with the phase-frequency response of the measurement equivalent circuit with a specified frequency f and the resistor R of RC network is derived; 3) Formulating the optimization problem by maximizing the integration of change of the output voltage of AD8302 in the entire predefined variation range of the capacitor C of the RC circuit, 1×10-12 F<C< 1×10-8 F, subjecting to f and R; 4) Solving the objective function by Genetic Algorithm (GA) to obtain the optimal f*=1.9412×108 Hz and R*=13.1 Ω, making the sensor achieve the highest sensitivity and accuracy of the measurement of the changes of C due to the variations of the water content in soil. Experiments on the sensor are divided in the following two steps. First, the sensor is calibrated in a series of tested solution with different equivalent soil gravimetric water content, and the gravimetric water content prediction model is built as y=-79133x3-18141x2-1418x+0.5926 with the coefficients of determination R2=0.9889. Second, the sensor is evaluated in the soil samples with different gravimetric water content. The maximum prediction and average errors are 4.58% and 1.63%, respectively.