I normally don't play much with hardware, mainly because there isn't/wasn't much I want to do that tends to require hardware that's not a regular PC or maybe a phone or tablet. This one is different, because no self-respecting geek would want the usual rotary control "programmable" timer to run their sprinkler system, would they?
We do live at the edge of the desert and we have pretty strict watering restrictions here. I'm all for it - water being a finite resource and all that - and I want to improve our existing sprinkler system at the same time. It doesn't help that the people who set up the sprinklers were probably among the lower bidders, to put it politely. OK, to be blunt they seem to have failed the "giving a shit" test when they put the system together. I've spent a lot of last year's "gardening hours" just trying to make it work somewhat. Not well, just "somewhat". Time to fix that.
First step was researching hardware. I'm comfortable with Unix type OSs (obviously) and with seemingly the world and their dogs releasing small, low power consumption embedded Linux devices I figured one of them would be perfect. The original plan was to get a Raspberry Pi or a BeagleBone with relay shield/cape and drive the sprinkler valves that way. A bit more poking around the web led me to the various OpenSprinkler modules (standalone, Raspberry Pi shield and BeagleBone cape) and they look ideal for what I have in mind. I'm planning to order the Raspberry Pi version as one of the nice touches is that the Raspbian repository has packages for the Java JDK, which gives me bad ideas of hacking parts of the sprinkler system in Clojure or Armed Bear Common Lisp. I'm not sure that the system is powerful to run either, but one can dream.
The good thing about the various OpenSprinkler systems is that they have the 24V to 5V converter on board so the power supply isn't a problem. There is already open source software for them that covers the normal requirements and either of them can control enough valves for our current needs without resorting to genius solutions like running two valves off the same controller output because someone installed a wiring loom that is one wire short of being able to control all valves individually. Apparently the fact that the water pressure wasn't high enough to run two zones at the same time fell in the category of "not giving a shit".
The next step after getting the hardware is to run convert the existing system to run off the new controller with some additional wiring to be able to control all zones individually. This will require fixing up some of the wiring issues and will also have to tie in with my project of running some Ethernet wiring around the house unless I decide to go wireless for the sprinkler controller. Haven't figured that part out yet. Given that the controller is "headless" I'm tempted to hide it away out of sight and just run Ethernet and 24V power to it.
Once it's all up and running I'll look into adding some sensors for a bit more fine-grained control over the system. Rain sensors are not really helpful out here as it hardly ever rains during irrigation season. I'm thinking about adding at least a couple of moisture sensors for some of the more sensitive plants to ensure that they get the appropriate amount of water but not more than necessary. Not sure I'll get around to that part this year, first the system needs to be up and running reliably before I go and break it again.