They’re 3D printed.
Here’s the files that I used:
Thanks for the link. I could retrofit the diffuser panels on my cube using that technique.
He could get away with much less wire than he did. If you arrange the panels corner-to-corner, you need only very short loops of wire for the bridge–except for the very last panel, which ends up in the wrong place. But that’s also the least important one from a voltage drop perspective, so you can use thinner wires there.
The panels are likely wired that way for a few reasons. It may have been easier for him to run the wires that way than to remap the LEDs to get them in the order he wants when some of the panels are rotated. I don’t know your experience with LEDs but those are directional. The data can only be sent in from a specific side.
But on top of that, the extra slack is helpful for getting everything in the cube. If there’s no, or not enough, slack, it’s harder to assemble.
Of course he may have also just done it to make the layout more obvious to us. Something the author of my current project could use some help with.
You can also run power to each panel or back feed it from the end (if those panels allow it) to help with that issue.
In any case, it was a fun project and I’m thrilled with how well it turned out.
Here’s the one I’m currently working on:
I like it and I can tell the author put a lot of time and effort into it. It’s clear there’s multiple iterations in both the model and the directions. The problem is that the directions aren’t so good. I’m guessing the author’s written and rewritten them enough times that they either don’t care at this point are don’t realize how much is left out. For example, there’s multiple pictures of the back and front of multiple iterations of the electronics, but no schematics and it’s not always obvious which front goes with which back so you can try to work it out yourself.
Anyways, that’s just me venting about it. I have all the parts printed and I think I got that circuit running (hard to tell without any lights).
These are the types of projects I start and then halfway through I realize I should probably document everything I’m buying and the things I’m figuring out so I can post it for everyone else that’s going to have the same questions.
Yes, they’re the same panels I have. You can still arrange the panels so that the corners are all directly adjacent, with the exception of one of them (it’s easiest to make this the last one). Though you’re probably right about getting them into the cube. That one puts the panels on the inside of the frame, while mine are mounted to the outside. Easier to fold up when they’re on the outside. Aside from the diagonals making the 3D print easier, leaving the corners free meant leaving space for the jumper wires from panel to panel.
That was it. Though in a sort of tricky way. GPIO9 is the BOOT pin as well as the default I2C SCL pin. It should be pulled high for a normal boot. Being I2C, I already had a 10k resistor pulling that line to 3v3 (and it has a 45k internal pullup). So it should boot normally, right?
Well, the I2C acc/gyro board was connected as well. And for whatever reason, I decided to power it through another GPIO pin, so that I could shut it off if necessary (power saving, say). But it turns out that when the board is unpowered (as it is at boot time when all the GPIOs are high impedance), it pulls the SCL pin low. Maybe there’s some resistance in there also, but whatever it was, it overpowered the 10k pullup I had. Therefore the chip was put in bootloader mode.
Anyway, I changed it to just keep the accelerometer powered all the time, and it’s all working.
For a while I thought the problem was the 3v3 rail. Turns out the LDOs on these boards aren’t all that L. Powered at 3.8v, the 3v3 rail was at about 3.16v. A little on the sketchy side. Not the actual problem, but I might replace the LDO at some point for more margin.
Glad you got it figured out.
Those configuration pins are kind of annoying.
Debugging is kinda fun, really. Satisfying to go through the detective process.
I’m still sorta curious what’s going on with the accelerometer unit. Nothing obvious showed up when probing it out–the SCL pin wasn’t just clamped to ground or anything (I initially wondered if their output driver had a P-channel mosfet to ground). I’m wondering now if the unit got semi-powered up through the data pins themselves, which itself propagated some nastiness back through them.
Lesson learned, regardless. Watch out for that BOOT pin.
Incidentally, since it still fits the DYI theme, here’s the breakout board I’m using:
Mount an ESP32C3 Supermini to the right-hand side, and the rest of it provides a number of common functions, like current measurement, high-side and low-side power switching, optoisolation, status LEDs, input protection, and some other things. Simple but has already worked well in multiple projects. And uses only basic components at JLCPCB, so I could get the PCB made and assembled on the cheap (like <$2 each shipped).
I really like the use of pads rather than thru-holes for most of the I/O. Makes prototyping and assembly so much simpler. Will undoubtedly build a v2 at some point with the lessons learned.
Minor project for smallish garden:
Much progress with dirt turned, topsoil added and fence put up and gate made.
Gate is a simple frame made of materials I had in my shop and some very old fencing. The fencing is mostly just stapled on. Trying a simple hinge of 4 binder rings.
First role of fencing up.
All fencing up and gate made and hung.
What’s the fence made to keep out (or in)? Most wildlife won’t have any difficulty at all with one that short.
Can’t speak for WE?, but it looks high enough to keep out the bunnies in our neighborhood (North Shore of Boston).
That’s what I would guess.
Our deer fencing is 7’. They’d barely need to jump to clear that fencing.
Just bunnies. No deer here and no way to keep out squirrels.
So I wouldn’t really consider myself much of a DIYer, but since we bought our house, I’ve found some enjoyment in fixing up/improving small (and not so small) issues all over. Latest one was our upper bathroom door, which was a rickety accordion-type of contraption that’s been falling apart, and even in pristine condition never was what I’d call nice:
So I built a replacement from pinewood, opting for a mid-folding/sliding kind of design:
https://i.ibb.co/vxdBTfb8/IMG-20250321-120942.jpg
Thanks to the upper rail and a lower recessed guide, it runs super smooth and stable…
Now, I used the wrong finish for the side beams, and the black spray paint I used for the hinges isn’t as durable as the packaging boasted, but all in all, I think it’s a definite improvement.
A huge improvement! It’s lovely.
Joanna wants some shiplap in there somewhere. 
Really nice job.
Definite improvement! Well done!
Thanks, everybody! I am kind of happy with how this turned out. Also with not sacrificing any fingers to the circular saw!
Really is a vast improvement.
Until the other day, I had dozens of comic book boxes. Thanks to advice from my local Comic Book Guy (and Viewers Like You…) I sold virtually ALL of them.
Now I have to figure out what’s going on the shelves… QUICK, before my wife sees “Room For More Fabric and Xmas Decorations”!
Lego sets? Board games?
-Brian