Second Field Test

I made several improvements to address the most pressing failures in the first field test.  I used a piece of PETE from a water bottle to form a cover for the gear train.  Because this plastic can be deformed when hot, I warmed it in hot water and then molded it to match the motor mount before hot gluing it in place.  I also changed the mounting of the limit switch to better align it with the drive wheel.

Clear plastic shield over the gear train and motor

I'd noticed some odd behavior from the circuit, where it would change states unexpectedly.  At first I suspected this was some sort of electrical noise in the light sensor circuit, and confirmed that the output from the Schmitt trigger was intermittently unstable.  After suspecting thermal noise, but finding that even with the 100K Ohm resistor I'm using to pull the light sensor to ground the effect of this noise is quite small, I found a cold connection on one of the jumpers on the board.  Resoldering this connection seems to have fixed the problem.

I returned the feeder to the yard, the chickens appeared nonplussed.

Nonplussed Chickens and Feeder

It worked great for the first day, but then it began to rain.  At first, everything appeared to be fine.  Sadly, after two days of rain I discovered a minor disaster.  The drive wheel had detached, and both guide-rails for the door had fallen off.

 The issue seems to be that the door absorbed enough water to warp, jamming the door open, and eventually popping off the guide rails.  This leads me to wonder just how much pressure a wet piece of wood can exert?  I expect this mechanism at work is that the wood fibers absorb water due to osmosis and act like little hydraulic jacks.  How would you model the total force a piece of wood could exert when it gets wet?

My guess is that this is a problem of computing osmotic pressure, which, for a incompressible solvent like water is a product of surface area and temperature.  This 1901 Nobel Prize Lecture gives a workable constant of 2/3 atm for a 1% sugar solution.  That's roughly 10 PSI, which over the approximately 5 square inches of contact area is plenty to break off the rails.  Interesting to note that you could get an approximate measure of the van 't Hoff factor for whatever the water is dissolving in the wood by measuring the exerted pressure at different temperatures.

Anyway, now I have to decide whether to protect the door from rain (which would help keep the feed dry too) or redesign the door mechanism so that it's unaffected by weather.