I'd like to set up a capacitive soil measuring system, inspired by products like
I'm thinking it would be great to separate the electronics from the actual capacitive probe for long-term outdoors deployment.
So, as a first experiment, I've removed the electronics from a device similar to the linked one above, soldered short leads to the ends of the traces of the sensor part, and used it as the capacitor in a simple 555 astable oscillator circuit. This generates a really nice frequency response between about 500kHz and 1.5MHz between wet and dry conditions.
Rather than buying and chopping up existing sensors, I'll just design a PCB with big traces appropriate to being the soil probe. Does anyone have any thoughts about the design of this? My first thought is to make it much like the widely avaiable probes. But there is also an opportunity here to improve/experiment. I'm thinking I might make it a bit bigger so as to integrate the reading over a slightly larger area of soil. I wonder if there would be sense in any different pattern/shape of the traces?
Also, does anyone have any thoughts on the length or other aspects of the leads between the circuitry to the probe itself? It seems to me I certainly want to keep them pretty short (and consistent length among any devices that I want to compare between). I'll have to experiment, but any thoughts on what to consider would be appreciated!
Thanks. I see that there are various forum discussions about this. I'm not entirely clear what disadvantage of the capacitive approach is being solved. I suppose it is potentially simpler. But, it would be harder to compare among multiple sensors (which is going to be inevitably crude in any case), due to the practical challenge of putting a pair in real soil, keeping them parallel at a specific distance.
Thanks jim-p! They note in their dicussion that at the time of writing they were working on characterising the the active volume of influence - it may be worth contacting them for their thoughts on whether an alternate shape-size might be useful. I've browsed some more recent papers by the senior author and it doens't seem that they have published on that aspect yet.
There is another capacitive unit called Chirp!
The original designer did alot of experimentation with different electrodes and posted his findings but I can't seem to find them.
You might want to Google Chirp soil moisture and see if you can find it
I'll have to put some time in to fully understand, but I think I get the basic approach. I'm slighty wondering if my original approach of counting pulses might be simpler, but I'll try to properly understand both.
Now the shape is a multifaceted task - for one, I was able to achieve a comparable performance with all the shapes by choosing frequency and known resistance, but in general you want water in dielectric to have as much influence as possible, but still have capacitance in a place where you can comfortably measure it.
Speaking of separating electrodes from electronics - for this particular measurement method it's disadvantageous to move electrodes away, as capacitance and thus physical configuration of interconnect from electronics to probes starts to play a significant role. In one project I had to move electronics even closer to probe to avoid variation caused by manufacturing process.
I can't comment more on your pulse counting idea as that's a different method than used in my probes. Note, that to avoid influence of salinity you want frequency as high as practically possible, ideally over 80MHz.
Blockquote
Speaking of separating electrodes from electronics [...] had to move electronics even closer to probe to avoid variation caused by manufacturing process.
Thanks - I'm hoping that I can come up with something practical with even a couple of cm, and still have to do tests, but I'll definitely keep that in mind. But I'll keep in mind that they just might have to be right on the same PCB. Plan B is to play around with various liquid plastic dips and so on, which I gather lots of people will be able to give further feedback on.
Blockquote
that to avoid influence of salinity you want frequency as high as practically possible, ideally over 80MHz.
Most inexpensive sensors don't operate at anywhere close to that, right? I think that's a bit beyond a 555 timer anynow. But certainly point taken! (n.b., I did just notice a typo in my original post, the range I've been getting in my initial proof of priniple set up is about 0.5 to 1.5 MHz).
Yes, that's why they are all crap also there is a lot variation from sensor to sensor due to capacitor/resistor variation, that's why I generate 16MHz from MCU and use the crystal as internal RC oscillator gave too much variance and also was more sensitive to temperature.
The bigger picture is that I'm doing pilot studies working up to potentially needing to deploy ~200 sensors. So, cost could become an consideration. I'd obviously like to make the most meaningful measurements possible, but what really matters is being able to identify periods of very low moisture availability, and possibly also of outright waterlogging. So, for the partiuclar purpose, being able to distinguish small variation in soil moisture content may not be all that important (at least, probably isn't, but in any big deployment effort it is worth getting as much resolution as you can without trading off too severely against cost, complexity, power consumption, or any of various other considerations).
So, what next:
experimenting with my current mock-up and different wire lengths
make probe PCBs with different patterns (though from this thread it doesn't look like my biggest practical concern at this point)
get a couple of miceuz's units
empirical calibrations in soils from the study areas, using my proof-of-principle pulse measuring system, the super cheap sensors, and miceuz's units (also pimoroni.com seems to have a pulse-based one that is much like what I'm thinking about, might try it too)
learn about generating higher frequency signals, though I supsect that there are good practical reasons why miceuz settled on the frequency that he did
Further thoughts on any of these or other aspects still very appreciated!
also there is a lot variation from sensor to sensor due to capacitor/resistor variation, that's why I generate 16MHz from MCU and use the crystal as internal RC oscillator gave too much variance and also was more sensitive to temperature.