Here's a quick test of the PCB electromagnet, snapping the hour hand into position. It works!
This is just with a lab supply.
I designed a controller PCB for this too, and even got it manufactured. I don't think I'm going to bother populating it, though. If I ever come back to this, I'll have to do some serious thought about how to get a powerful enough battery into that tiny package.
Shelved! I have lost interest in this, for now. I'd love to get some better photos and videos, but it's all currently buried in a box now.
This is likely to be the last log in a while. With the close of the contest, real-life stuff is going to get in the way of this project for a bit, and I obviously didn't finish.
UPS appears to have lost my package for two full weeks or something, I dunno. My PCBs still haven't arrived. I ordered them on the 23rd of December, so that's annoying! It's not OSHPark's fault for sure. They shipped on the 6th/7th, I think that turnaround is normal for a 4-layer board, particularly during Christmas break. Definitely going to blame UPS, although they're not helped by the holidays either.
Everything was 3D printed to test fit. Surprisingly enough, threads worked pretty well. There was a ton of stringing in the internal threads, though.
Initially, I was using a knife to clean those up, but that got old quick. Additionally, oozing from the print of the outer threads made it a little tough to mate for the first few cycles. So I printed a tap. You can see the taper on the leading threads, and the notch to clear out the chips.
That worked great! Passing it through the system a couple times made the threads work easily.
Here are all the components:
The black plate on the left is a stand-in for the faceplate PCB. More on that in a bit.
I tossed a magnet on there and was able make it jump around the coils using a small bench power supply. The resistance at each phase was about 12 ohm, the same value that Carl Bugeja measured. Each phase does go through three entire coils, however, so assume 4 ohms per coil.
The coil pattern is illustrated quite well with Carl's coloured gerber export:
Application of the right hand rule shows that as current goes in a clockwise spiral direction (curl your right-hand fingers in that direction), your thumb will point in the direction of the magnetic field, out of the screen. Interestingly, the ordering of the coils on the layers isn't from top to bottom. Carl ordered it from layer 1(top)->layer 3->layer 2->layer 4(bottom). Probably to (very very slightly) increase the length of the current path and therefore the resistance of the coil.
To drive a coil with any polarity we like, we need a standard H-bridge circuit. A future version might do some interesting stuff with that, so the fine control that an H-bridge provides is kinda cool. Generally an H-bridge consists of a couple matched pairs of transistors of opposite substrate. Poking around, there's a single chip MOSFET solution that looks suitable, the VT6M1.
So here's a quick PCB layout of how that would work:
But of course, just dropping that into a board is boring! We're going to make a pretty PCB, we might as well go all the way! I've spoken before about PCBModE. Using my tool to convert the above PCB into a compatible format, I pull it into PCBModE, cleaned it up, added some electromagnetic spirals, and a little bit of flair. There are solder pads to hook this PCB up to another PCB, containing all the brains.
Figuring out a compact way to drive all those connections is somewhat challenging. With six coils around the perimeter and one in the centre, a "drive high" and "drive low" connection for each side, there are 28 signals, plus power and ground. All of those need to be transferred out to a "controller" PCB, so I added those green teardrop solder pads. The idea being that pins are to be soldered onto them, and then pushed into through-holes on the controller PCB.
There were many more, with very little polish in any of them. They're just to get the mechanicals down and direct the overall thrust of the design.
Watch crystals are quite inexpensive from China, so I picked up a selection of sizes, and while waiting for shipping, I modeled and 3D printed some tests for the three most likely candidates. Shown are 30mm, 34mm, and 38mm, but you really can't tell. The 38mm one seemed to fit best on my wrist, so that's what I went with.
Rolling my sketches into the CAD, I came up with this:
The main body is just a shell with internal threads, and inside there's a stack-up consisting of:
1.2mm crystal face
37mm OD x 1.5mm o-ring
Double sided o-ring retainer
15x3x1mm neodymium magnet
Coin cell battery
It's a very tight system. As designed, it doesn't really fit with the battery.