Reliable way of measuring soil moisture

[conducteur]

I want to make my own "automated garden".

I tried first to measure te humidity of the soil with a resistive divider network (where one of the resistors are two copper wires inserted in the ground). Less water in it should give a higher resistance.
With the 10 bit ADC, i got readings about 450 before i gave water to my PUT (Plant Under Test) . After giving water i got readings of 810 but after 3 days the ADC gave me results of about 800 while the soil was really dry.

Next i tried a concept based on a galvanic cell. I replaced one of the copper wires with a galvanized steel nail. (and removed the pullup resistor). Substances in the soil generate electricity when dissolved in water. When the amount of water decreases in the soil, this voltage should drop....

A first test generated about 1.1V. I ran two tests each spanning more than 8 hours. I gave the PUT water and started logging the measurement. (applied a moving-average filter).  A few hours have passed after the first test before i started the second test (without the moving average filter).

I think i can't rely on one of both sensors, so if anyone has experience with this topic on sensors. You could help me a lot.

[rmacintosh]

Take a look at the Vinduino project. I believe it was one of the Hackaday Prize winners, or runner up ....

Moisture sensor: http://vanderleevineyard.com/vineyard-blog/the-vinduino-project-2-making-and-installing-gypsum-soil-moisture-sensors
Sensor Interface: http://vanderleevineyard.com/vineyard-blog/-the-vinduino-project-3-make-a-low-cost-soil-moisture-sensor-reader

I had plans to build test this out last spring but didn't get around to it, might end up doing it this spring.

[calexanian]

There are many off the shelf soil moisture sensors out there of varying qualities. Sparkfun have one I believe. For a more professional I have seen the Irrometer Watermark sensor used. Just google search soil moisture sensors and there you go.  The proper term in tensionometer.

[Ian.M]

Professional quality sensors use AC excitation, non-corroding materials for the probe, and the electronics are potted or epoxy dipped to protect against corrosion.   The Sparkfun sensor is strictly hobby grade and will die fairly quickly as it uses DC excitation.

Google: fringing capacitor "soil moisture sensor" design
to find articles about capacitive sensors.

https://eesensors.com/media/wysiwyg/docs-pdfs/ESP06_moistP.pdf looks promising - AC coupled and built within a rugged stainless pipe.  It could be simplified by omitting the wireless link and second MCU, replacing them with a PIC and an analog switch to select between the sensor capacitor and the reference capacitor.    The PIC program would be very simple and would easily fit in any PIC12 device - every N instructions, toggle an output pin.  The output pin would directly control the analog switch, and also provide a varying duty cycle waveform to your main MCU.     Its possible to do it without the PIC, but you need to build a RC oscillator that runs reliably with a timing cap <100 pF, with one side grounded, then divide its output down to get a slow enough output pulse to run over long wires without problems.  N.B.  If you are powering it externally, put a low leakage 1uF capacitor || 10nF ceramic, between circuit 0V and the stainless pipe (sensor cap earth electrode) to block DC and avoid corrosion!

[helius]

The very old and cheap technology for soil moisture is the electrochemical cell of copper and lead. It is attached to a galvanometer and is self-powering.
It works fine for momentary use, but can't be left in the soil due to corrosion.

[5321488]

Not sure about accuracy, but I think Silicon Chip did a project like that at some point.

[DaJMasta]

I don't know if it's a method that's been tried, but what about a permeable vessel housing a weatherized humidity sensor inside the vessel's air chamber?  The idea would be to have a small pocket of air that's sealed underground at the depth you're hoping to measure so you can read how much moisture is in the air pocket.


No idea if it's been done or studied, but if you were willing to try, it seems to me like it may provide usable results.  The trick would be preventing liquid water from reaching the sensor, so I'd propose something like what you'd use to keep out weather in an above ground sensor, so even drips into the chamber wouldn't get on the sensor.

[calexanian]

The super duper ones use Time Domain Reflectometry. Before we got out of the landscape irrigation controller business a couple years ago there were some good quality ones at reasonable prices coming on the market. Well reasonable for a piece of test gear anyways.

[Zero999]

I want to make my own "automated garden".

I tried first to measure te humidity of the soil with a resistive divider network (where one of the resistors are two copper wires inserted in the ground). Less water in it should give a higher resistance.
With the 10 bit ADC, i got readings about 450 before i gave water to my PUT (Plant Under Test) . After giving water i got readings of 810 but after 3 days the ADC gave me results of about 800 while the soil was really dry.

Next i tried a concept based on a galvanic cell. I replaced one of the copper wires with a galvanized steel nail. (and removed the pullup resistor). Substances in the soil generate electricity when dissolved in water. When the amount of water decreases in the soil, this voltage should drop....

A first test generated about 1.1V. I ran two tests each spanning more than 8 hours. I gave the PUT water and started logging the measurement. (applied a moving-average filter).  A few hours have passed after the first test before i started the second test (without the moving average filter).

I think i can't rely on one of both sensors, so if anyone has experience with this topic on sensors. You could help me a lot.

The trouble is the salts dissolved in the soil affect its conductivity, as much as water does. And it isn't the chemicals in the soil which generate the electricity in the galvanic cell, but oxidation reactions in the electrodes.

Sorry, that's probably not very helpful because I don't know what to suggest. You could take a sample of soil, weigh it, then bake it at >100^oC to dry it out, weigh it again and the difference in weight should tell you the moisture content, assuming the sample isn't heavilly contaminated with other volatile compounds. The trouble is, it would be difficult/impossible to automate.

[conducteur]

But it's not easy to automate that principle with a microcontroller . So the voltage i measure now (with the copper wire and the galvanized steel nail) isn't in related to the moisture level?

[Arjan Emm]

Yes, it's related to moisture content, but also with soil type, amount of fertilizer and the level of corrosion of the electrodes.

I am working on the same thing at the moment. Trying to grow a banana plant inside with an automated watering system.
I ordered these:
https://www.plantcaretools.com/nl/meten-vochtgehalte/vochtsensor-met-7-led-s
I believe they deliver in belgium as well. It's a capacitive moisture sensor with no metal contacts to the soil. 7 leds indicating the moisture content. My plan is to tap off the signal to the leds to activate my watering system. I'll report back when i did some tests with it, but that's gonna take a week.