Last winter, I built a device that I called “Mini Weather Station.” It is a small plastic enclosure which contains a thermometer, barometer and hygrometer. And of course, a microcontroller, realtime clock and a 2.4GHz radio module.
It worked reasonably well for a few months, even though it had a few minor design flaws.
Yeah, that little piece of red wire is a sign of one of those minor design flaws. The other problem isn’t as obvious from the photo. I had originally intended the microcontroller to run at 1MHz on its internal oscillator. But when I did, the barometer stopped playing nice with the microcontroller. While I am able to run everything at 8MHz for a little while, the µc starts to become unreliable as the battery depletes.
Last time, I thought the battery was dead. So I replaced it with a fresh one. But, no matter how fresh the battery, it’s no longer transmitting. The culprit seems to be some corrosion on a few of the radio module’s solder joints.
While I’m disappointed that the board has failed so soon, it’s given me an opportunity to address the design problems.
Here, I have a new board design with space reserved for a 4MHz ceramic resonator, a new expanded debugging header, and a completely redesigned board layout to better suit the orientation of the finished PCB in the plastic enclosure.
This still won’t solve the corrosion problem, though. I found something called a “conformal coating”: a nasty chemical which claims to protect against moisture, corrosion, fungus, thermal shock, and static discharges. I haven’t bothered with it before, mostly because I didn’t know that it was even a thing. But even though it’s starting to cost me far too much money, this whole fake engineering thing is about trying new things and learning, with eye protection and safety gloves, of course.
With any luck, this iteration will prove to be far more reliable than the last one!