How to integrate a soil sensor and a rain sensor – what is the best way?

[tmittelstaedt]

I have an existing Rainbird controller that works OK. It does not have a rain sensor just has the jumper on it’s terminals for one. It uses a “normally closed” rain sensor from Rainbird that uses a felt washer that supposedly when it gets wet it expands and clicks off the rain sensor to normally open. That seems to be to be a bass-ackwards way of sensing rain but OK whatever.

I have a Pi and bought the O.S. pi version for it I have yet to plug any of it in.

The research I’ve done says that sensing rain is an exercise in futility you really want to determine watering based on soil wetness sensing. But that seems to be more complex for a number of reasons:

1) Most soil sensors are garbage and fail due to oxidization within a few months however there are capacitance sensing ones that do not pass a current through the soil that do work but require some tuning.

2) The soil sensor outputs are analog and the computer needs a “go/no go” determination on whether or not to water

3) Many people say “screw the sensors use the national weather service weather reports to determine if your going to sprinkle”

I am wondering what the best course of action is? If I go with sensors then it sounds like I really need to get both a soil and a rain sensor and write some software that maybe uses a table to calculate whether the rain supplied enough water or not. If I go with a “cloud” solution for weather than what is the best way?

This is located 2 blocks from the Pacific Ocean in Oregon and the landscape is basically all based on a few inches of “topsoil” covering sand. Terrible water retention and if you don’t water regularly the plants start shooting roots down down down whereupon they hit the water table which is loaded with salt that kills them. Either you plant salt tolerant stuff and let nature water it (essentially everything dies in the summer except for coastal pines that are adapted) or you water it. And when it rains the rain is lightly salted. I always thought the stories about high corrosion at the coast were a myth until we bought the house out there and I saw it for myself. Even stainless steel that is out in the rain rusts. So getting the watering dialed in properly is a must. Too much and your going to be leeching all the nutrients out of the soil and the stuff dies unless you spend a mint on fertilizer, too little and you will be letting too much salt build up and the stuff dies. Most people out here have ridiculous water and fertilizer bills and great landscape, and can afford it, or they have drab blah landscape with brown dead grass all summer and coastal pines. It’s 2019 I’d kind of like to bring some real technology into the picture to avoid the water waste. What should I do?

[Ray]

I’ve been hesitant on adding support for soil moisture sensor. To be fair, I did some research on existing soil moisture sensors and even reverse engineered a few (https://rayshobby.net/wordpress/reverse-engineer-a-cheap-wireless-soil-moisture-sensor/). Your point 1 is definitely valid: there are many soil sensors on the market and many are quite rubbish. Those that are good also don’t use any common standard, some are analog, some are digital, so it’s quite difficult to pick one and say that’s the one we are going to support. Moreover, ideally the soil sensor should be wireless (dragging a long wire from the lawn to the controller is very inconvenient), and for robust sensing it should probably be sampled at multiple locations instead of just one location (e.g. what if the spot the sensor is plugged into is particularly wet or dry compared to the average?). Introducing multiple, wireless sensors increases the complexity of the system and decreases its reliability.

Your point 2 is also valid but not a show stopper — we could implement some sort of scaling so that the watering percentage gradually reduces when the soil moisture increases. Also, there are threshold switches which basically convert analog signals to digital based on a preset threshold.

Your point 3 is basically what we ended up with.

So I am still unsure what to do about soil sensor. We would like to add some initial support, but I also don’t want to introduce a solution that ends up wasting people’s times and turns out garbage. Probably the first thing to decide on is a small set of soil sensors that are of good quality and commonly used, so that we can start taking a look and see what needs to be done to get them supported.

[Author]

Posts