Something stuck me at the dojo tonight. The true irony of Karate is found in that moment when the Makiwara makes you so frustrated with your punching that you want to hit it. I swear that thing is mocking me.
I’m in the process of prototyping several IoT devices as part of evaluating a startup idea, and since the devices are going to go in people’s homes design matters – even for initial units for evaluating the idea. Actually, I think design is especially important for evaluation units because I am trying to find out if people want to bring technology like this into their homes.
One of the things that make me love 3D printers is that they make the black project boxes a thing of the past. Once I finished prototyping the electronics, firmware, and cloud side code – I was able to sketch up a few case designs on paper. Then pick the one I liked the best for initial testing.
At this point I should probably point out that I am not an industrial designer. When I showed my design to the designer I want to work with on the project, he said it reminded him of one of those bubble gum tape dispensers from when we were kids. (Ouch!) Still, I think my design is a hell of an improvement over testing with black project boxes, and it’s a pretty decent match for the design I liked best of the 15 or so I sketched through. Here is what it looks like sitting on the windowsill in my reading room.
I have the design split across three boards. A sensor board inside the dome, the eps8266 board with sensor breakout, and a DC/DC converter to power everything off of a AAA battery. The boards I made at home are super ugly, and hacked on, but functional. Enough that I am ready to put in an order for an initial run of boards to build a dozen or so test units.
I’m building some proof of concept prototypes for a startup product around the ESP’s, and they proved quick and easy to iterate around. This stuff just used to be so much harder! It is awesome to have an 80MHz 16 bit micro-controller with built in WiFI for $3. The ESP-03’s I have been using are also nicely packaged into an easy to use breakout board.
This picture pretty much captures my route so far. I had just heard about Radio Shack’s being broken up when my ESPs arrived. So I used a Rat Shack breakout to test the core circuit, then iterated my design on some quick and dirty home cut PCB boards. Forgive the nasty look of the soldering, each of those boards has been hacked on a bunch as I iterated through different sensor circuits. The end result turned out to be a nice proof of concept prototype, enough for me to evaluate the startup idea. Well, enough that I could figure out I want to make a run of ~100 of these units and get them in the hands of others to do more evaluation.
You know that saying that single men are bears with furniture? Well, for me at least its true. You can ask my house plants which I am always forgetting to water. So I wanted a network connected soil moisture sensor to remind me about watering specific plants. The other sensors on the board are for a project I am evaluating with some friends as a startup idea. That circuit sits, mostly, on a separate daughter board not shown.
Here you can see the soil moisture sensor being tested.
So while I love the ESP’s, and intend to blog more about them – it is a bit of a love / hate relationship. I have a pile of blown boards. The first one I blew I chalked up to my being the cause, but after a few of them I’ve noticed that the chips in the batch I bought seem to be highly sensitive to reprogramming cycles. It looks like the number of read / write cycles on some of the chips internal flash might be as low as 10-15 cycles. Once I have some more time I intend to try thrash testing them to nail down failure parameters better.