Cavity resonator for soil water content

    For our final project in 6.013, Electromagnetic Waves and Applications, my group and I developed a proof of concept product for predicting the water content of soil. We wanted to make something to help in farming efficiency and conserving water, as both over and underwatering crops is detrimental.

Our experiment setup with soil, NanoVNA, scale, and cavity resonator can

    We had several ideas for measuring the water content using electromagnetic waves, such as measuring a reflected wave from top of the soil, but we finally settled on a cavity resonator. A reflected wave without a resonator would be more dependent on the distance between transmitter and soil than the actual material, and we thought the results from a resonator would be more accurate and consistent. We used an aluminum can as our resonator with an SMA connector and 3cm antenna inside, then filled the can with 300g soil and more water. For every 25g of water added we transmitted EM waves from the antenna and measured the resonance frequency of the can. We estimated a resonant frequency for the can by approximating the dielectric constant of soil, ε, and swept frequencies around it to be in range of resonance. For this data collection we used a NanoVNA (vector network analyzer) and the NanoVNA program.

    The video below shows the frequency response of adding progressively more water to the resonator. In the chart on the right, the x-axis is a range of frequencies from 500MHz to 3GHz and the red line S11 shows the reflection coefficient of the wave (how much power is reflected from the antenna). The peaks of S11 are the resonance frequencies of the can at each amount of water. Using those frequencies and the depth of the peak we could calculate the quality factor (how damped the resonator is).

Data from soil sample 1

    The next step was to analyze the data. We had results from two soil samples and wrote a linear regression model to map the Q-factors and amount of water. We then tested the model on a new can of soil with an unknown water quantity and found it accurate within a few grams.

Some of the code from our linear regression model

    Although the cavity resonator was effective at predicting soil water content, it's perhaps not the most practical application of them and they are better off in things like microwave ovens and lasers. However, I enjoyed working on the project and I learned a lot about frequency response and resonators!