[chbot] Just when I thought I had a good soil moisture solution...

[Robin Gilks]

> ... I discovered that rather than resistive I should be looking at high
> frequency capacitive sensors.

Looking into this some more and getting my calculator out, I have come up
with the following figures (perhaps someone has a feel for how close these
are likely to be!!) based on some of the ideas in the 'tindie' sensor I
quoted.

A pair of probes, 100mm long, 20mm apart will exhibit a capacitance of
about 1pF in air (using 2 methods, I get .4pF and 2pF so in the same
ballpark!!).

The dielectric constant of water is 80 (compared to just over 1 for air)
so saturated soil (50% water) would look like 40pF and dry soil about
10pF.

The STM32F103 has a MCU clock output that can run at half the core clock
speed, i.e. 36MHz so I'll be looking to measure a change of impedance from
110 to 440ohms

Just playing with the idea of having the CPU driving one end of a length
of CAT5 twisted pair with a differential signal and putting a bridge of
low capacitance schotttky diodes between the probes and feeding the output
back on another of the twisted pairs. That keeps all the underground parts
passive and puts the CPU next to where I want the 2.4GHz module to return
the results!!

If I can cut/etch some PCB material then I'll knock something up for the
next meeting. This is all becoming quite interesting!



-- 
Robin Gilks