Try to never try resistive sensors. It is reaaly wrong way. I try to bult few resistive. No way.
I try to buld some inductive. Yes, it's possible, but lot of analog parts, difficult to calibrate. No way too.
Capacitive senors is most reliable and has a simple digital schematics.

Good luck you on your way)

PS: here is my own sensor http://vegimatics.com/products/current/
want do discuss - wellcome)

RF_CH bits are used to select channel. Channel number and frequence calculated like this: F = 2400 + RF_CH.

And some usefull table.

@thucar
Hello, i'm not sure it is to good idea to public links here, but there is my personal website where is some sensors examples in sensors history page http://vegimatics.com/products/history/

And of course, we can discuss here some interesting for you things)

Sure. At first, I'm very surprised that your resistive sensor in soil still not destroyed under corrosion especially when the current flows. All my experiments with such sensors have been stuck many years ago. May be a miss sometihing, just show me your sensors after three years in soil? And again, it's a basic phisical things, that the conductivity of the soil depends on the mineral composition. I have already said more than once that there is possible to get a "negative" conductivity of the soil, something like effect of the battery.

@mfalkvidd It depends from mineral composition of soil. I buld capacitive sensor with additional electrodes to measure salinity of soil with resistive method (it measures the resistance of soil to alternating current). So, while value of capacitive sensor stable and let say 50%, value of resistive can be critically changed by adding few milliliters of water with fertilizer. Capacitive value will be chaged only to 60% e.g.

It's not a good idea to use resistive sensor, especcialy measuring resistance to direct current.

@mikaeljones
I think the most suitable device for monitoring processes is a tablet PC based on Intel and Windows. There are simple and cheap models costing less than $ 200. This is much simpler and more versatile than developing advanced controllers with a display. In addition, you can get remote access in a simple way, for example, TeamViewer. Well, the simplicity of developing a program is also important.

Since I use the RS-485 protocol and Modbus, it's easy for me to connect all sensors and pumps via USB-RS485 adapter.

I do not use wireless sensors, as I believe that if a pipe comes to the plant for irrigation, it is not difficult to add a cable.

However, from time to time, I return to developing my own simplified watering controller

@thucar
Hello, i'm not sure it is to good idea to public links here, but there is my personal website where is some sensors examples in sensors history page http://vegimatics.com/products/history/

And of course, we can discuss here some interesting for you things)

@mikaeljones
it's depend of how often sensor will send data. I suppose that much more then one year if exchange data interval will be 5 min. The problem is that nrf24L not reliable especially in "noisy" environment. And finding a free channel will take a time that affect battery lifetime .

I suppose for Canada. Anyway, is better to avoid to use 0-82 channels. It's really noisy.

RF_CH bits are used to select channel. Channel number and frequence calculated like this: F = 2400 + RF_CH.

And some usefull table.

Catia V5 is really good.

No reason to reprogramm. There is I2C wired outside of sensor instead of UART. And some IC necessary to drive RS485.

Would be easy to use custom adapter over the sensor with only two IC on board, some MCU and MAX485 (or analog).