Soil moisture Prototypes and Experiments

languages: ennl

Here we will document the development of the Meet Je Stad soil moisture prototypes

The last version of the PCB and arduino firmware can be found on github:

A list with ideas for improvement and experiments can be read and completed here:

Test setup (8-11-2020)

To better test the functioning of the soil moisture prototypes we had the idea to build a special test setup. The setup consists of a smaller version of the latest soil moisture prototoype with one electrode at 10 cm depth. The sensor is inserted into a round PVC (15cm diameter) cylinder with at the bottom a mesh and a filter paper on top.


Soil is inserted into the PVC cylinder, with the sensor in the middle. The PVC cylinder is then placed into a bucket which has a drain and a tap for removing excess water. The entire construction is placed on a load cell connected to an ADS1115 analog to digital converter (ADC). The load cell output is read as the differential between channel 0 and 1 of the ADC, with gain set to 16x.

As a first experiment the sensor is placed in the PVC cylinder with dry soil (mix of soil from 10-40cm depth from the Schothorst suburb in Amersfoort). Water is then added on top of the soil untill the water starts flowing out from the mesh+filter paper at the bottom of the PVC cylinder. At this point the soil is considered saturated. The soil is now left to dry out, while continiously measuring the output of the soil moisture prototype and the output from the load cell.

The data of the experimental setup is logged by two Meet Je Stad Nodes: Node 417 records the output of the soil moisture sensor and node 203 records the output from the load cell.

Prototype sensors for pilot

For the pilot with soil moisture sensors in Amersfoort, we choose to change the design of the sensor so that the bottom is closed and no water or soil ga go into the inner tube. This was done to make the interpretation of data and the calibration of the sensors less difficult, because we do not need to correct for water inside the tube.

The sensors were placed using a manual auger, which is quite job, especially when
- The soil is very dry, the soil then hardly stays within the chamber of the auger
- There are thick roots in the soil
- There is a soil layer with pebbels or rocks

When we are sufficiently experiences with calibrating these 'simple' sensors, we want to re-explore the design option with an open inner tube and placing the sensor using water pressure from a garden tap.


In a number of build-sessions with participants of Meet Je Stad we build 10 of these prototype sensors. On the figure above you can see the inner tube of the sensors with the electrodes (copper strips) at 10, 40, 80 and 120cm from the surface. The copper strips are 10cm long, 2.7cm wide and are placed within 1cm of each other at one side. At the other side small temperature sensors (NTCs) are placed as the height of the electrodes. With respect to the previous design the PCB has been redesigned to fit within the outer tube, so that all electronics are inside the outer tube. From the PCB 16 cables run onto the inner tube to the four pares of electrodes and the NTCs. On top of the PCB a cable with 6 cores is connected. 4 of these are used for the I2C connection to the Meet Je Stad Node and 2 can be used to reprogram the microcontroller without need to open the sensor.


The outer tube slides over the inner tube and a cap is glued at the bottom. On the top the outer tube is closed with a cap with a swivel through which the cable runs. On most locations a Meet Je Stad node is placed on a pole next to the soil moisture sensor.  On the PCb of the Meet Je Stad node an extra pin header is soldered for the I2C connection.

The greatest challenge for building the sensor is running the cables from the PCB to the elctrodes and NTCs on the inner tube. This requires quite a lot of patience, precision and takes the most time of the assembly process. To create a DIY sensor that can be easily assembled we need to further improve this part of the assembly.  During the building & experimenting sessions we experimted with the use of flat calble instead on running individual wires. To make room for the thicker flat cables the inner tubes were planed down on one side. Running the wires to the electrodes and NTC becomes much easier, but subsequently soldering the wires onto the PCB takes much more time.

Bodemvocht_regenboog sensor.png

From the building & experimenting sessions came a number of next steps and ideas for improvement:
- Place a flat cable connector onto the PCB so that connecting the flat cable becomes easier
- Design a flexibel PCB that can be sticked to the inner tube, eliminating the need for running wires
- Placing copper tape on the entire surface of the inner tube and cut lanes onto it.
- Buils a test setup to better evaluate the soil moisture sensor, while controlling soil type and soil moisture