Being absolutely in love with diffuse and unnatural lighting, I have of course been an avid user of both Phillips hue and Nanoleaf products. However, I’ve always been disgusted at their exorbitant cost.
Because, really, how expensive can a piece of molded plastic with some wires, LEDs, (shitty) connectors, a diffuser and a mediocre app really be? Apparently very expensive – and apparently sell very well despite of it.
There are quite a few DIY versions of Nanoleaf products floating around on the internet, however most of these are using breakout boards and take up much more space than needed. With the extremely easy accessibility of ordering PCBs through numerous online vendors nowadays, there’s no real reason NOT to dive into the brutish and hellish landscape that is EDA/ECAD.
For real though, once you pick up the core concepts of circuit/PCB design and you’re willing to fail a few hundred times until you learn to use circuit simulators, it is a blessing to be able to streamline prototyping and product design in one go.
The Nanoleaf product line has a rather brilliant modular design (at least in theory) and what i especially like is the ability to attach the controller unit to any available unity in the array.
Now the actual implementation of this, mainly using thin PCB’s(?) with exposed copper to connect between one another and it being the only source of contact and weight distribution – other than whatever fastening method you use to fix them in place – is not ideal. Especially not if you enjoy rearranging things often. The markup on these things are so insane that I can easily justify paying for and using god’s gift to mankind for this project: spring loaded magnetic pogo connectors.
Granted this means that every other shape will have to be either all male or all female contacts, which limits the amount of customization somewhat but that is not an issue necessarily – as i will get into later.(This approach is obviously not as economically feasible as the printed boards)
For starters, i want to test and see that my concept is at least somewhat sound, so what better way to test than to make a preliminary prototype that will inevitably crash and burn?
Starting with simple rectangular and rectriangle(?) shaped models to make it easy to both cut out strips of diffuse material as well as look for potential hotspots etc. For now i will power them with a regular benchtop PSU and since i only have 3-pin pogo connectors left for some reason, i’ll have to make do with a simple MOSFET driver circuit and control it via some MCU i have lying around (happened to be ESP32 in this case)
But before all of that, let’s have a quick look at a miniature version of the LED “strip” in action!
I can’t really describe how painful it was to cut out appropriate-sized diffuser sheets by hand with a scalpel, but it was very unfun.
Wow! with a single red diode and 2 layers of diffuser sheets, the bleeding and hotspot is very clearly visible.
Trying a different approach with flexible led filaments – it’s better but the diffuser is still not holding up
I tried adding a sheet of paper in between each diffuser, which gave it a very unique look but not the smooth diffuse lighting i wanted and it made me feel like i was doing some 5-minute-craft tier bullshit.
Unfortunately I have not yet gotten my new (proper) diffuser sheets which is apparently named Plastazote LD45 and is some kind of foam widely used for light diffusion. I also ordered some acrylic to see if dispersing the light across the entire width of said acrylic with a diffuser on top might produce a far more uniform light.
In the meantime i wanted to draw up the initial schematic for a very basic led controller – this will easily handle ~10 meters of a normal LED strip at full power, whereas the LEDs in the modules i plan to make will draw significantly less power than that. Another downside to this is that each module will not be individually addressable for v1 of this, that’ll have to wait for when i get a hold of proper strips or PCBs and 4 pin pogo’s after the initial testing.
This circuit will allow for simple lighting control via PWM on each of the gates on the MOSFETs and will be able to deliver plenty of power (especially for the insane price of power MOSFETS, sheesh). For now it’ll be fed via a benchtop or chinesium/commercial PSU, but in the future i would like to try my hand at designing a proper SMPSU, since power circuitry is one of many fields I haven’t really touched on in any major way yet.
After a quick and dirty routing job, the first prototype board is ready to be ordered and assembled by JLCPCB (not sponsored). For the first prototype I’ll just solder directly to the drainpad because I’m lazy and I am sure I’ll run into random issues i had not foreseen – thereby rendering this board practically useless. Same goes for the pins on the MCU, but in reality i only need the EN pin and the UART for this initial testboard. I was thinking of adding a socket to plop an ESP32 into instead of having the entire module on board, but I kind of forgot about it – it’s a planned feature for v0.2 and not a mishap, trust me.