A tool to calibrate and measure pwEC with generic TDR sensors and Arduino.
Leia em outros idiomas: Português Brasileiro
This project arose from the need to use generic TDR sensors for use with crop steering.
In the practice of crop steering, the use of pore water EC (pwEC) measurements is very common, which is calculated from the measurement of dielectric permittivity (read more), a functionality present only in more expensive professional sensors, like MeterGroup Teros 12.
In our case we would like to use more accessible soil sensors and compare the measurements with a Teros 12 in order to try to find some correlation in the measurements.
For this, we recorded a series of 30 measurements of each sensor in a spreadsheet for approximately 10 days in order to find some pattern.
After some analysis and based on some papers, we made quadratic and cubic approximations with the data using Matlab.
So it looks like we managed to find a calibration. With a precision difference of VWC < 3% and bulk EC < 100 ppm, what we consider a satisfactory result:
Generic sensor readings with Arduino
Teros 12 sensor readings with ZSC and Zentra app
Generic sensor readings with Arduino
Teros 12 sensor readings with ZSC and Zentra app
You will need a TDR soil sensor that measures soil temperature, moisture, and conductivity (EC).
We used a soil sensor from ComWinTop, model THC-S with RS485 interface. So take into account that all the calibration was done with this sensor. If you use other sensor models or from another manufacturers, it may be necessary to recalibrate and adjust the source code (soon we will release a method and tools for this).
- Sensor on AliExpress (buy the THC-S model)
- Manual and reader software
- USB/RS485 Adapter (Optional. If you want to take sensor measurements with a Windows PC via USB. Use that specific adapter. Others tested adapter models do not work!)
- More resources about CWT sensor
You will also need:
- Arduino of your choice (we use the Arduino Uno).
- TTL to RS485 (MAX485) bidirectional module for Arduino.
- Breadboard.
- Jumper/dupont cables.
- OLED Display module (optional)
- The sensor yellow wire goes in the
Aconnector of RS485 module. - The senor blue wire go in the
Bconnector of RS485 module. - The sensor black wire go in Arduino GND.
- Th sensor red/brown wire go in Arduino 5V.
Just open Arduino IDE, load the soil_sensor.ino to your Arduino and watch
Monitor serial (default 115200 baud).
If you want a spot check that doesn't need a PC to take readings, use a display module. Here I used an Adafruit-like OLED Display module with 128x64 pixels. See soil_sensor_display.ino code.
If you want to use a Wi-Fi microcontroller like NodeMCU, see @danielfppps blog post about using these sensor with a NodeMCUv3.
As stated earlier, all calculations and software presented here were based on two specific sensors, ComWinTop THC-S and Metergroup Teros 12.
It is known that different sensors of the same type tend to have very close measurements, at least in the most accurate professional models. So it's possible that these calibrations will work with many generic TDR sensors similar to CWT's.
However, given the lower degree of accuracy, in addition to being generic sensors without a high manufacturing standard, it can happen that measurements are very different from expected. In that case, try to take your measurements and find a new calibration setting, send us a PR, and we'll be happy to test it. Or get in touch so that we can better explore and try to solve specific problems for each sensor.
The main issue (perhaps with all sensors) is sensor stability in soil/substrate.
The sensor needs to be inserted into the compacted substrate (perhaps that's why it works better in rockwool) and avoid moving or removing it from the site as much as possible.
Once moved or removed and subsequently replaced, there is a high chance that the measurements will change. So the biggest challenge here is to position all the sensors so that the measurements are very close together. It's boring work, but possible. So be patient.
Also, different sensors may show different measurements depending on the position. In the case of the CWT THC-S and MG Teros 12, sensors used in this project, measurements were taken in the same pot/substrate with the sensors positioned at the same height and alignment. So if you take these precautions, you will probably have very close measurements between the sensors (see more below).
The measurements will vary depending on sensor positioning/height, that will vary depending on substrate type and volume. To solve this problem, Aroya developed a sensor alignment tool:
However, it is not necessary to have the tool. Below are the position recommendations:
| 4-7 Gallon O.T. Bag (Tall & Narrow) | 4-7 Gallon O.T. Bag (Standard) | 2-3 Gallon O.T. Bag | Slabs, 3"-6" Blocks, and 1 Gal O.T. Bag |
|---|---|---|---|
| 3.5 inches | 2.75 inches | 2.0 inches | 1.25 inches |
Simply measure with a rule from the base of substrate (not necessarily the base of pot or bag).
Coco is not a uniform substrate. Even rockwool has no that uniformity. That's the main factor on measurements differences caused by sensor positioning.
Watch Ramsey Nubani from Aroya talking more about this:
https://aroya.io/en/knowledge-base/crash-courses/uniformity
- [Meter Group] Soil moisture sensors - How they work
- [Meter Group] Teros 11/12 Manual
- [Kameyama] Measurement of solid phase permittivity for volcanic soils by time domain reflectometry
- [Gonçalves] Influence of soil bulk density in a dystroferric red ultisoil moisture estimated by TDR
- [Topp] Electromagnetic Determination of Soil Water Content
- [Hamed] Evaluation of the WET sensor compared to time domain reflectometry
- [Hilhorst] A Pore Water Conductivity Sensor
- [Gaskin] Measurement of Soil Water Content Using a Simplified Impedance Measuring Technique