A simple prototype of a simple idea: slide-together hexagonal pixel tiles, input on one side and split outputs on 5 sides. Use case is prototyping/quicky props. EDIT I’m thinking of these as a 5cm spaced LED string, chopped up into 1-pixel sections, with easy splicing/forking. More limited in layout than an LED string, but with other advantages.
I’m going to play around with these (another 45 boards are on the way) and, if they seem useful, add interlocking notches.
HT @Benn for prompting me to assemble and test PCBs I’d already made.
It would be a good idea to add other dummy pins on each side so they can be held to one another on more than just the middle connection or maybe magnets if you could work it out so that the N/S always aligned correctly.
This is a proof of concept. As I mentioned, a next version will have interlocking notches, mainly to prevent connection misalignment, but which will also help hold the tiles together. If the concept proves viable, I’d look at a way of addressing all the edges. Magnets are a possibility, and should be doable by alternating N/S orientations on each tile. Until then, tape or hot glue.
These are 5cm pixel-to-pixel, and I’m also considering a half-size version.
This is a prop, so I don’t really want the full pixel lighting up (or maybe I’ve just gotten used to that look). I was considering multiple LED beads per tile, but it is already bright enough. Right now, I’ve just put some white plastic cardboard on top of the pixels just to try it out.
I think I’ll do the smaller tile, because I’ll get better resolution for the same PCB cost (if I tile them myself?) although connector hardware cost scales with the number of tiles. Also considering 12-gons.
I may forgo the pluggability (saving on connectors) and go for assembling with copper tape and solder. I’ve another project using that approach.
I ordered some 120 degree plastic diffusers (3 yen a piece delivered, about the same as the 5050 LEDs), and I’ll see how they look.
@TokyoDave up to you
for a real good WLED projekt i woudt go with 7 pixels per Hex as of 2-3-2 arangment
this gives you so many options to make a outstanding display
I’m thinking of these as a 5cm spaced LED string, chopped up into 1-pixel sections, with easy splicing/forking. More limited in layout than an LED string, but with other advantages.
This is not an attempt to connect 2W lighting elements together; that’s a fine project, but that’s not my goal.
Version two: 12-gons for more angles, smaller for better resolution, interlocking for safe connections, mounting holes, and uber-simple on the connectors - simple U pin headers pushed into place. I only ordered five, in order to check the fit.
They fit snugly enough (I can pick this up and it remains a flat sheet), and they should be paddable to add tension with a thin coating of something rubbery, etc. or soldered if necessary. The JLCPCB standard scoring process is +/-0.2 mm, so I am not sure how repeatable this is. But I’m going to order a hundred or so and see.
I like the design idea overall, it looks like a good adaptable solution.
If you’re suggesting you might not need to solder the U-pin headers - that’s not a good idea at all. Flexing power connections over time will be guaranteed to cause you grief.
If you’re thinking about adding some additional surface pads that can be used to create solder joints for stability, that may help. Alternatively, along those lines, you could think about larger ground pads closer to the center of the disc. Those could be bridged by something like a heavier gauge straight piece (18AWG or larger) to give an anchor point as well as improved grounding.
The other option I might suggest would be to add silk screen labeling for the single input DI and dual +/- marking for all the other output points (think about both top and bottom sides). That should make it much easier to see that things are laid out correctly in a larger sheet.
@divsys
Thanks for the feedback. I always seem to miss some labeling in the rush to get something out. In this case, I think marking the input should be enough since the geometry constrains everything else. (For real idiot-proofing, I’d make the notches asymmetric)
If I had my druthers, I’d use wider pins, with more spring and a slight detent, and hard plated holes. That would make a nice product. Failing that, I have some ideas like retaining plates that clip on over the pins on the backside of the joints, and thicker PCBs, both to increase mechanical stability and to improve electrical contact. And of course, soldering the input side, as it is needed by every tile in use. Probably I’ll insert the U pins from the back instead of the front.
The use case for this version would be prototyping, which could be ‘permanentized’ by soldering the pin headers in place, and then mounting to a substrate through the mounting holes or on the edges. I’ve ordered a couple hundred at about $0.10 a piece. We’ll see how they stand up before I tweak it some more.
On soldering with wire, I’m using that on other projects e.g. the below prototype. A somewhat larger dodecahedron enclosing a battery and controller module with accelerometer motion detection is in the works using the wire construction.
The basic issue with not soldering connections comes down to your guaranteed “surface area of connection”. If you look at the design of pretty much any removable pin connections, you’ll see something squeezing a tube against a cylinder and/or a springy plate against a flat surface. The idea being, keep as much surface area as you can in contact.
Your doing that with the spring tension in your U-jumpers against solder plated holes.
You’ve also got square pins in round holes which will probably give you 2 points of contact and not a lot of surface area.
Your experiences will be the ultimate test of course.
I’ve just chased too many flakey issues that turned out to be a bad connection when I thought “that should be fine…”
Certainly I agreed with all your points and decided it is worth trying. There are precedents, including those FlexyPin spring clips used for (testing, one presumes) castellated modules.
After a day of searching, I did find co-planar connectors (not a popular part or design), about $0.45 a piece in bulk or over $1 for my quantities. That’s $5-$12 a tile in materials, vs $0.15 for the clip version. I’ll try my hillbilly approach and see how far I get. Starting from simple and improve often beats the other way around. But not something one typically recommends to others. My $50
I’m talking about connector costs. The bare boards are $0.10 each at small batch quantity, delivered. Up in the thousands of units, that goes down maybe two cents more. The pins about $0.02 for a set of three. I buy 5050s for $0.02 in reels of 1,000. Triple that gives a pack of 50 for $25. Manufacturing assembly is simple, and a heat table makes attaching the LED beads easy and clean.
A fun lego toy that can be permanentized with simple soldering skills, is what I’m thinking.
@TokyoDave this woudt be gread and packed in 50Pcs without any solder Pin around 0,30Cent woudt sold alot as 50mm is the space most Holiday displays use in WS2811 Plugs that now are up to 10USD per 50 Nodes
That’s along the lines of what I’m thinking. I have an online store on a couple of Japan handmade marketplaces for gourd and woodworking items, and I’m thinking of doing another for seasonal lighting.
Here’s my larger experiment with the daisy style, around 45 tiles. I solder the pins on the input side (insert from back/underneath), so assembly is just clipping them together. They are under a bit of tension (on purpose), so they tend to bow up a bit, which isn’t necessarily a problem. Electrical connection seems to be excellent. With the smaller size, I don’t think they need the 120 lens.
I’m still not quite there with releasing it to my local market; packaging, sales and instruction copy needs writing and pricing confirmed. But I’m close now.
