Yes, No 3D printing filament will be 'good enough' in heat transfer to not cause it to melt or diode to overheat, I have tried it with a 505 at 5mW and that still overheated in less than 30s and its power decreased a lot.
Tbf just the hours spent machining the host themselves warrant about double that, not to mention soldering and initial hours it took to code the software along with glue ing and aligning the optics, easily looking at 4-8 hours per minimum.
Yeah, Its completely different for high power blues, they don't really care, a nubm anything surviving a bad driver is nothing, one of my friends has blowtorched an m140 till red hot and it still lazed afterwards. => 'thousands of times per second, so my guess is, ripple probably won't do any...
Its not just that, dielectric and voltage rating for the caps he used matters a lot, when you get close to the voltage rating capacitance can drop to half of the rated capacity, besides that with those switching frequencies he does not need bigger for the most part, he needs properly calculated...
They are 25 bux, not for the open source ones though, 25 is for the improved 4A capable 10mm X 9.8mm bucks with reverse polarity protection. And you can go stronger with a green, 9mm 2w green diodes do exists, I also have 9mm copper modules in stock in case you don't have one, that will be 3 USD...
For a driver you can build up one of my open source bucks or have me set one up for you, https://sites.google.com/view/giannislasershack/open-source-drivers?authuser=0
In the meanwhile (since the boards for more are stuck in Bulgaria for whatever reason), 12 800mA boost drivers up for dibs, freshly made and organic :P
Update lads, in about 2 weeks time tops I will have around 10 adjustable boosts, 20 fixed smaller boosts and 6 up to 4.5A bucks(in the 10mm x 10mm format).