Buying, Programming, and Testing Capacitance Soil Moisture Sensors

[@byronap5814]

I made a similar system and it works great. I used ESP now to create a mesh network. In order to overcome the limitations of the sensor I wrote a calibration based on readings from where exactly I felt the soil was just moist enough. Since I am only working with grass my sensors only reach down about 3-4 inches. This made it easy to detect moisture penetration and simulate volume. The sensors have a small solar cell to help keep them charged. They would last about 2 years even without the help of solar. The data from the mesh network is captured retransmitted over wifi then logged. Since I made the sprinkler controllers it wound up working great and never has to be touched since it won't water if it's not needed or it's too cold or hot since the controllers grab weekly weather forecast data and adjusts as needed to keep the moisture optimal.

[@ModestMaker]

Thanks for sharing. Yes - I've been wondering how to handle the sensor calibration in a way that yields something meaningful. It can get pretty complicated fast so your approach of feeling the soil might be a better place to start so that I don't get bogged down. I have to learn more about ESP- I see lots of comments regarding the same and suspect I'm missing out.

[@tomran247]

Greetings from Australia, and thank you for another very interesting video! I've used capacitive soil moisture sensors in a similar application. I think your hypothesis is right. They give highly variable readings depending on soil composition and sensor location, and require calibration for their specific point of installation. I took such a sensor, applied a thick conformal coating over the electronics, buried it at 100mm and logged the readings every six minutes over six months. I found that over time calibration is inclined to drift, probably caused in part by changes to the board's electrical characteristics from temperature, component tolerances, ingress of moisture and corrosion. Additionally, I believe that the capacitive characteristics of the soil change because soil tends to have a mobile composition. Changes to chemical composition, biological motility, or organic structure would all alter the capacitive characteristics of the soil to some extent. An unexpected effect that I noticed was that the sensors reading changed dramatically depending on temperature and barometric pressure when soil moisture levels should be static. There was a direct relationship in these readings with my set-up. This requires some further investigation. I haven't given up on capacitive, but the variables are troublesome so I went back to a resistive sensor with which i can limit the variables more easily. It gives generally satisfactory results for irrigation.

[@ModestMaker]

It's wonderful to hear your experience and feedback from Australia. Your observations regarding temperature (and barometric pressure) have merit since I noticed some drift in my own logged tests when moving my tabletop experiments from my heated home into my cold workshop. This does raise some challenges in regards to developing a sensor rating curve for volumetric water content since workshop conditions will not be the same as field conditions. In fact, I'd like to use temperature to compliment these sensors given water has a high specific heat and buffers soil heating during the day. Further details on using temperature as an indicator of soil moisture are summarized here: ecorestore.arizona.edu/news/2020/10/using-mulch-prevent-heat-stroke-desert-landscapes. Your other comments regarding soil texture and movement are also of interest-- I've downloaded a few papers that look into these facets which I hope to summarize in a future video. My sincerest thanks for taking the time to share.

[@tomran247]

@@ModestMaker Thats a really interesting article! Our water problems aren't usually too different to Arizona where I live, in Adelaide, South Australia. Although perhaps not at the moment because of the unusual weather patterns here. I've buried the capacitive water sensor in proximity to a Dallas one-wire temperature sensor and that gives useful feedback for soil temperature. Although I haven't given up on capacitive sensors, I've been experimenting with measuring water levels in ollas (the buried terracotta pot irrigation method). In principle the measurements aren't too different from tensiometers but as I'm only interested in the rate of water transit through my soil medium rather than the specific volume, I don't make any special effort to calibrate for the usual variables and its good enough. The arrangement for automatic control and observation is quite simple, I just put one of my automatic drippers over an olla and chart the rate that the water level drops when the dripper is turned off. Lately I've departed from trying to measure the volumetric quantity of water in the soil, instead focusing on the soil temperature and the rate of water loss (say to evaporation, drainage, use by plants). I'm of the opinion that purely observing trends in water loss dispenses with the variables problem of electronic soil measurement for specific water volumes and gives a more usable metric for irrigation because its quite easy to gather good data and map trends for predictive watering.

[@ModestMaker]

​@@tomran247 I love the idea of using terra cotta pots for measuring water potential based on infiltration rate. That's fascinating and something I'm going to have to look into as an alternative-- it's so simple! I assume if the infiltration rate is relatively higher, your soils are relatively drier and warmer. (I also have a few Dallas sensors I've been playing with.) There's lot's to think about in regards to future designs and experiments. Thank you for sharing all this!

[@PeteHoffswell]

Great work. Thanks for documenting your project so well. Keep up the good work!

[@ModestMaker]

Thanks, will do!

[@Mythobeast]

Lots of good information here, thanks. It gives me a better idea of how much effort it would be to put something like this in my yard for irrigation guidance.

[@ModestMaker]

Glad it was helpful. I think the setup is much easier than understanding the data. I'm hoping to do a deeper dive soon as it relates to understanding soil texture influences on sensor response and interpretation.

[@jurgislll]

For measuring in scale of 5, these cheap sensors probably will just work fine. Since the 555 timer on the sensors outputs only around 15kHz it's nowhere near to the high(>40MHz) frequencies needed reduce the effect of soils salinity on the moisture reading, that means anytime there are tons of rain that washes the salts from soil the sensor will show that the soil are dryer than usually, opposite case is when fertilizing, fertilizer is just a salt that will also affect the moisture reading in this case show that the soils are wetter than baseline. This also implies sensor will show different readings in different soils with the same moisture content. Another BIG factor is temperature, these sensors vary around couple percents per degrees C if I remember correctly, that mean if the day/night temperature varies with 10 degrees C it will result in 20% error, I think that is completely unacceptable for it to be useful. I strongly recommend testing and calibrating with high performance sensor, best performance sensor that I've tested is METER Teros 12(350$) moisture sensor, super reliable accurate from my testing. And from the cheap end the best sensors I've found to be PinoTech SoilWatch10(25$) sensors, with some calibration, digital signal filtering and temperature compensation they also can be quite accurate and reliable.

[@ModestMaker]

Thank you for that insight. Yes, I've read temperature can impact the readings. Also, we have high mineral content in our municipal water so that may pose challenges for informing common area watering in our Association (I use rainwater on my own landscape). More experiments are needed. Thank you for the references on the alternatives for calibration- that's good information to have. Overall, your feedback suggest these low-cost sensors are best used for qualitative measurements and perhaps best calibrated using "feel" to give a general idea of soil moisture content as others have suggested-- but for a couple of dollars per sensor, I'm okay with that.

[@AlexSouthwell]

Great work. Would be good to see if you could make the pvc house everything including the microcontroller so you could auto log over time then report back to somewhere where you can graph all the data. ESPhome should have a bunch of the stuff you need to make this happen. Probably don't need it but you could add a small solar panel on to the top of the pipe to keep the microcontroller battery topped up. But there are some ESP stuff you can do to make them run for a VERY long time.

[@ModestMaker]

Yes- definitely going to have to look into ESP as an alternative and yes- your suggestion of encasing everything in PVC with a small panel makes sense. Okay- ESP on my to-do list. Thanks!

[@dsz75]

In my experience this sensors can't last for more than a half year on the field. The biggest problem is that the soldermask layers absorbs water then starts to come down. Afterwards it starts to produce false readings.

[@ModestMaker]

The commercial variants seem to have a beefed up amount of insulation around the electronics-- I'll have to watch that.

[@carstenolsen3641]

Thanks for an interesting video. I saw the exact same problem with the same kind of sensors I used for plants in my living room. The soldermask would start to get bubbles (from moisture) and the copper would corrode under the bubbles. When this happened, the readings would be off and the sensors would not detect when I watered the plants. I would love to see a follow-up video with your long-term experience when you have used the system for a year og two.

[@ModestMaker]

Absolutely- I will be posting updates. Thanks for your interest😊.

[@chrisstott8580]

Sensor placement is really tricky. Your method looks good but I see an issue with surface water seeping down the outside of the pvc tube. Water wells generally have a concrete pad to prevent this happening. A better system might be to completely bury the sensor (electronics suitably encapsulated) and just bring out the cable. Lay the sensor flat, run the cable horizontally for about 6 inches and the to the surface. Just my 2 cents worth.

[@ModestMaker]

Good point. I agree and will revisit the backyard installation. More importantly, I have to see if these sensors will actually work for predicting volumetric water content. From my preliminary tests, they seem to be extremely sensitive to the presence of the smallest amount of moisture in soil, and thus may suggest “moist” readings when available water is well below wilting point (not helpful for managing irrigation). I need to study these sensors a little more.

[@davidkruger8758]

Great vid and project :) Just one observation.. did the PVC not prevent water from getting closer to the probe than it could have? Edit: I see you responded to someone else on that.

[@ModestMaker]

You are right- I am redesigning to diminish bias from water flowing down the sides and include a datalogger. I will update once it's set up and improved. Thanks for checking it out!

[@iz11smh]

Oooo great video! What good about this is it rained the past 2 days :) Im making a new vid before december, and i bet you will like it.

[@ModestMaker]

Oh man - looking forward to that IZ! Thanks for checking in - subscribed!