A dumping ground for all of the one-off breakout boards I build.
I haven't touched a breadboard in years, when it's more worth my time to design a PCB and send it off to @oshpark. A few bucks for a board saves me an hour or more messing around with a breadboard, or protoboard, and also proves out my component footprint at the same time.
From the fab, it connects VCC from the STLink from the board. Things IRL are more complex though, so you can cut the VCC trace in between those centre header pins. So like, when you're powering the STM32 externally, you don't power from the programmer.
A joule thief is a circuit that pulls as much energy out of a battery is possible, by slowly adding energy into an inductor (to be stored in its magnetic field) and then pulsed out into an LED suddenly when it reaches a threshold.
I wanted to use photodiodes as the energy source, however.
On the right, I stacked up some BPW34 / BW34 photodiodes, (ab)using them as solar cells until they could very slightly light up an LED while directly against some extremely bright indoor lights.
Each cell is about 0.24v open circuit voltage, so I totally guessed and tossed two of them in series on a board.
One more spin on the concept is that I used a surface mount coupled inductor. Honestly not sure how that's going to go, most examples out there are huge toroids and I just don't want that.
Also used a MOSFET instead of BJT, because, efficiency?
With a 5k resistor, it seems to just turn on at 0.55v. It seems to be pulsing slightly, as it should, but it's hard to tell without digging out the oscilloscope, and that's too much work for today. I'm still not convinced it's not powering the LED directly through the inductor, though.
With a 1.3k resistor in there instead, it turns on at closer to 0.47v.
Soldering on the photodiodes, nothing happens in the dim indoor light. I'll check it out again when there's sunlight in, oh, about three months.
Behold, the schematic and PCB that took maaaaybe twenty minutes to lay out, as fast as humanly possible. That's why nothing is straight!
For posterity, I tried to find the most efficient parts possible for this design.
LED: Digkey 475-2709-1-ND
So... I also have some proper toroids coming, just to see if they work. I'm open to suggestions for other components/values to play around with.
This is basically just the power section from the upcoming version of #uMesh
With the changes, I kinda wanted to test it all in isolation, in addition to this probably being useful in its own application. There aren't really any dedicated battery charger / power supply with discharge protection going to 3.3v that I've seen. Most of them seem to output 5v.
It's using two different styles to connect the two halves. A set battery connector leaf contacts on one, and PCB copper as a target, connected via a 2mm pitch right angle header, abused to be SMD on both boards.
You can see I was doing some logo experimentation, too. Can't waste good real estate!
Due to postal strikes in my area, I wasn't actually able to get 2mm right angled headers, so I made do with dual straight headers, and bent/snipped contacts as necessary.
Then I taped up one side:
And then they snap together quite well:
They have a pretty satisfying feel, and seem kinda sturdy. I might try using some cardstock or plastic on the front instead of tape, and a jig to keep the things at 90 degrees while soldering.
Also, I used three magnets only in this one, because I was having a dumb moment and didn't see that you can't have symmetrical polarities with three magnets! Silly.
Anyway, I more or less like the mechanical design, will have to test the reliability now.
I soldered on an LCMXO2-256HC, which is, hands-down, the cheapest fpga you can buy right now, at about $3 in singles. I bought a couple, and also an LCMXO2-1200HC, in case I need something beefier in the future. Same footprint, but more like $11.
I think these footprints might be 0805, but fuck tha police! I got 0603 components! They work fine.
Currently, I have no way of testing this breakout board, though! I don't have a lot of evenings to sit down and figure out the toolchain / programming strategies for a bit.
They really don't work all that well. The plastic in them melts like butter if they get so much as a peek at a soldering iron, the spring contacts are too stiff to be compressed by any small magnet, and the females sticking out like that result in huge problems trying to get alignment right.
Shown with a Digikey ruler. I'm liking this for scale, I might do it all the time.
That hole for the microphone wasn't wide enough to actually be a hole. It was filled up with dried soldermask. A small drill bit cleared that right up. Note to self, make the hole bigger in my footprint. It's currently not a via, and it's probably best to keep it that way so as to not encourage solder to flow into the hole.