After hours and hours of taxing my brain for what approach to use to streamline the “guts” of the HoneycombLight (for real – it deadened me for a bit from getting stuck several times and trying to overcome these obstacles. I lost sleep.), I finally came to a conclusion for the architecture of the matrix routing (and then I was ok).
So, I figured, I’ll just place daughter boards, matrix mixer boards, every 4 pixels and let the pixel driver boards run straight to it, 4 at a time, raw, (R, G, B, and cathode) instead of “mixing upstream” by soldering wires together right on the drivers (like I did last time). Mix as I go – 4 times downstream to the final MCU – it’ll all be prepared once it arrives there. …Because the entire thing is really messy if done with twisting or soldering wires together onto connectors at the pixels, believe me. It’s a “grid”, a Matrix, a mess. I really needed the double-sided planes, here. I did it in two “phases” (those 9 via holes in the center), alternating top and bottom copper. Reminds me of the butterflies in Fast Fourier Transforms, but physical.
This allows the entire path to the MCU boards to be nice and clean with dedicated plugs for easy service and replacement, and I admit, also for the cleanliness of the way it looks – fresh, not cluttered. I like that. 🙂 That was one of the tradeoffs I made. The way that “under the hood” looks is important to me.
And, as for the MCU board, most of its architecture is the same as the first through-hole rendition, but it’s just double-sided, solder-masked, and silk-screened, now. Loads of improvement! But, the concept is the same… However, I can’t have it looking like it’s still a prototype. (I have already built that and it worked (with wires all over the place), so now it’s time to do it for real). So, every component is now surface-mount, minus the row and column connectors (for rigidity). It’s not the *final*, final version – I have already thought of things that I’ll have to do to revise it, but it’s a good start.
I am yet to send it off to be manufactured, so I haven’t tested it yet, but I’m really excited to get a small batch back and run it through the ringer! I mean, I’m almost as excited about how it’s going to *look* as I am about how it’s going to work! Haha – to me, it’s like a work of art! Especially with that ground plane in there surrounding the traces – looks dope.
I don’t know if putting “BradOrmand.com” on there is “forced” or not, actually. I want a “maker’s mark” on there, but am still deciding if that’s the direction I wanna go or not. I might put in a bitmap with my logo or even transfer it over to the company name I’m thinking about using for the marketing of the lights, etc. I don’t know, yet. Decisions, decisions.
All-in-all, though, this board was weeks in the making and took a lot out of me. I didn’t think I could do it there for a while… But, I was also kind of stressed out with all of the Portfolio and job stuff, recently. But, I have it quite in-hand, now. Let’s hope the electrical connections all check out and that those tiny-ass traces can handle the current I’ll give it. It’ll be multiplexed, but still only milliseconds between blasts. We’ll see. This is all part of the learning. I’m determined to do this.
Last thing… The main thing I am concerned about right now with this is soldering the central ground pad underneath the 0.8mm pin pitch TQPF without a reflow oven or a good heat station. I don’t know how that’ll work. I’ll be thinking about it. In fact, I think it has solder-mask under it right now. I’m going to have to investigate, but the datasheet *does* recommend tying it to ground. Onward we go…