Grabbed a few of those type c breakout boards you linked to.
Somewhere in the thread you mentioned checking the threading on the soldering iron to see if it was still fully screwed in, mine was a bit loose (hadn’t thought about the heat/cooling expansion/contraction cycles loosening the tip, but should have thinking about it retroactively).
My switches are the last bit I’m waiting on, as instead of scavenging enough box royals off a hot swap board in another office to start, I thought I’d order some more and let curiosity get the better of me, ordering some of the new gateron silver blizzard switches and some blueberries to give a try. It seemed like if I’m doing this one, I might as well try it twice since it won’t take as much time or energy when if this works I want to make another for work, and the v3 mf108 is available again. For future cases where someone gets curious about the differences, I thought I’d post some more observations that might help someone else down the line with these things.
Also, if anyone here around 5/15/21 likes clicky switches that are beyond loud, novelkeys has a sale on the non-box thick click (navy?) switches - they are listed as speed switches, but with a weighting of 100/100 (box navy’s are listed as 60/90 on their site), I’m wondering if they are really the same thing. I feel like they might be good on a wood or acrylic plate, but the steel plate I have without a pcb seems like it would be super pingy (the bm43a i used them on is a bit pingy itself and the plate doesn’t have near as much space for reverberation). But I’m almost certainly talking out of the wrong end, so who knows. I do have a sneaking suspicion that once you get a board layout deep into muscle memory and up the wpm, taste in switches changes away from “i can’t believe how loud/pronounced this is.” That being said, I’m still smitten by how satisfyingly like a stress toy typing on heavier switches is. The old mf108’s browns were like sliding around on slushy ice and really highlighted the crappiness in my typing skills or lack thereof.
First off, the v3 seems to run different stock firmware, which fixed the most annoying part of the v1 I have. The upper nav cluster originally operated a bizarre countdown timer by default, requiring fn+ the keys to send the normal scancodes like prt scr. I thought I was going to record a video of the thing going crazy like my old one, but on plugging in on MacOS, the keys do what they should, with the “timer” dohicky hidden behind fn+those keys. Good on IKBC for fixing that. Unfortunately they didn’t add any way to remap keys other than a few modifier swaps. Nothing to make capslock into shift+tab. If mac os just had a slightly larger set of remapping options at this point to get me there, I’d just swap out the key switches and plug this into my old case.
Regarding hardware - the two PCBs are close to identical, and the plates if different, are immeasurably so. They can swap with one another. So can the connector plugs going from the pcb to either mounted mini b (v1) or type c daughter board (v3). Oddly the v3 has no insulation between the pcb and back of the case. Either they dug it out a little more or were confident the pcb and steel plate don’t flex enough to ever cause an issue. The v1 also had an additional thick piece of clear plastic over the uncovered section with some light adhesive, really not taking any chances. I’m guessing anyone wanting to build using the case of the v3 wouldn’t need a daughterboard at all, just remove the plug mount or splice the wires from it. The port clearances are also different, as the v1s have a rectangular cutout, relying on the straight up rubber mini b connector to seal up the excess. The v3 has a daughter board I don’t see why someone wanting to move, also good on them for fixing their connection method as the v1s were almost all ripped from being screwed in too tightly (mine was, but I just hot glued it back in place).
There’s also a totally different siding to the v1 and v3, as the latter’s bottom is one solid piece, the former seeming to be cnc’ed differently, or extruded and later cnc’ed only in a few spots. While the plates and pcbs are interchangeable between the two, it’s dicey whether one could mix and match as the screw holes seem ever so slightly off, and placing the top of the v1 above the bottom of v3, there seems to be all of a millimeter in difference, enough I wouldn’t advise trying lest some screw holes strip.
This second board, being in fine condition and fixing one of the worst issues almost makes me want to just swap the switches and try one of those hasu usb to usb remapper dongles. I’m thinking at minimum first I’ll work on the plate file in wood and preserve this pcb. I should warn anyone else on here that as of 5/15/2021, I need to tweak the plate .svg file, as the switch sockets are slightly too small, owing likely to the wood being several times thicker than the steel plate (like the way wider rings fit more tightly at the same size than thinner ones). Either that, or just the imperfect nature of using illustrator and piecemeal photocopier scans. Probably not a surprise to anyone. The switches could maybe be made to fit with some filing and/or coaxing, but as the burn didn’t get all the way through on some of the cutouts, I have to reburn another anyway and might as well do it the right way.
Had done a good bit of research repetitive aimless google searching on whether the pcb on the mf could be reprogrammed via qmk or anything else. In case someone thinking the same dumb thing I was but stumbling onto these posts before others, apparently the controller is something totally outside of the realm of qmk compatibility, not on any list to figure out. The only hope is if it shares enough dna to the poker boards that use the same controller, but that’s so far beyond me. It may also be locked, which I can guess means it’s either read-only, or is just otherwise as averse to flashing as a beach-walker with photographic memory realizing they just stumbled into a geriatric nudist colony on learn-how-to-macarena-day. Wondering if it’s possible to use an existing pcb for the wiring traces and just have it run to a different controller. Figure if this wasn’t dumb for a reason of which I’m ignorant, I’d see it happening more in these forums. I know about the )replacement controllers? Is that the right term for ghost squid etc.?) for really popular boards, just not exactly how they function. Figure it has more to do with either the traces for microcontrollers being too tiny for novice solderers, or maybe just that reverse engineering the wiring layout along a pcb is harder than handwiring for less popular/common boards like this one.
I’ll dig through the thread again and find the link to the pcb layout tool and use the calipers to maybe generate a better file others can use without as many fit issues.