Daishi

Let’s break down the nomenclature real quick…

  • Daishi: The name of this project, and the name of a complete keyboard using a CBC1 PCB and Daishi-X case.
  • Daishi-X: The name of the specific keyboard case – for example, the Daishi-Prime is the first version, a prototype using a plate and “floating switch” design, and the Daishi-A is the first production model; plateless using a top case & gasketed standoff-mount design, but made for the same layout & PCB.
  • CBC1: The first MetaMechs PCB. CBC because I call the layout “compact battlecruiser”. This PCB uses SMK 2nd gen switches.

Super weird, I know. Next project won’t be so mucked up.

The PCB

Back of the CBC1 PCB

This project started from the layout. I was interested in saving space on the x-axis, but didn’t want to give up top-layer functionality coming from a full-sized keyboard, plus I wanted to have macro keys. The obvious solution to me was to just add an extra row on top, which is certainly not a novel idea (the F-122 is one of my dream boards). And while I love the battleship design, and indeed I love the style of other “battlecruisers” (FK-5001 hype), I can’t help but see wasted space in those designs.

As a result, my CBC1 layout has a few quirks, and I’ll go over them here.

  • Inverted-inverted-T nav cluster: I love this thing. It’s intuitive, it mirrors the arrow keys, and it slots in perfectly to the gap left by the Num Lock (which is not a key that needs to be conveniently accessible).
  • Less-smushed F keys: I absolutely detest gapless designs – they look worse to me and are far less functional. But I also think that .25u gaps aren’t enough for the gaps between clusters of keys on the F row, while .5u just seems exorbitant to me. As a result, I try to stick between .33u and .4u for the F4-F5 and F8-F9 gap.
  • Top-left Esc key: This is hit and miss for me – I like keeping the escape in the top-left corner of the keyboard, and moving it there was necessary to fit the previous two features without losing easy access to the delete key (I tried a version with split backspace and wasn’t a fan), but on rare occasion I do still end up hitting F1 on accident. Happens less and less, though.
  • Volume knob: Every keeb should have one. Preferably a smooth one with no detents (the little click feedback you get as you turn the knob), but I might be alone in that preference. The knob itself is machined aluminum, and it feels so nice.
  • Num Lock: This key should be taken out back and shot. Unfortunately, Windows sometimes likes to turn num lock off, so I had to keep it around somewhere on the keyboard just in case (though I am considering re-mapping the numpad keys to their non-numpad equivalents…).

The rest should be fairly straightforward. Lots of macro keys, less-used keys shunted off to the edges, and nice chunky 1.5u keys between the spacebar and arrow keys.

For switches, I chose SMK 2nd gen clicky switches (also known as Monterey switches, but hereafter referred to as just SMKs). These are light switches, using an alps-style click mechanism. The result is a very “delicate” keyfeel, with a large amount of sharp tactility. The one downside is that they aren’t as smooth as some of their competition, but it’s not super scratchy, and the clickiness makes any scratch less noticeable.

Populated with switches

An important thing to note is that SMK MX-mount sliders are the perfect MX stem width in one dimension, but just a bit wider in the other. This means keycaps fit tightly, but I’ve never encountered any issues like the ones reported with old Kailh Box switches where stems would crack, and since it’s only wider in one dimension, any “stretching” should be a non-issue. So far I’ve tested them on GMK, SA, OEM, and MT3 profile caps, and have no complaints, except that they take a bit of extra effort to install. For reference, I have a board with original Box Jade switches and OEM keycaps and there has been noticeable stretching on those, but the keycaps that I’ve put on SMK switches are still tight on regular MX stems.

The CBC1 is designed to be used with a daughterboard for the USB connection, and of course it’s USB type-C, because it’s 2019. For stabilization, it uses screw-in GMK stabs, clipped and lubed with dielectric grease. And for the encoder, I’m using an Alps encoder with a switch and no detents. I’ve tested about a dozen rotary encoders and this is the one I liked best – it has no detents, which I feel is right for a volume knob, and it has 1.5mm of travel on the switch, which is much nicer than the typical ~0.5mm.

The AT90USB1286 mcu is programmed in QMK, and much like QMK, the PCB files and schematic are all open source!

The Case

Since the Prime variant is just a prototype, I’m only going to discuss the Daishi-A here.

There are a few notable features I want to discuss. First off, and my favorite feature of this case, is the per-key gasketed standoffs. Each key (with the exception of a couple macro keys that sit over the USB daughterboard) sits on a little pole that goes down into the case. This is a brute force approach to ensuring uniform keyfeel throughout the entire board, and a very rigid feel as well.

These standoffs are each topped with a thin layer of neoprene, making this sort of a gasket mount. I’m not gonna lie, I partly did this for the memes, but also to ensure perfect fit between the pcb and case, and for some vibration dampening. It works.

It kills me that PF8 and PF9 don’t get their own standoffs

This idea is what basically drove the entire case design, and I’m not aware of any other case that takes this approach. I love it, and I plan to use it to varying extents on future builds, but SMK (and Alps) switches are perfectly suited to it due to them not having a protrusion through the PCB, so the standoff can go directly beneath each switch, instead of offset like you’d need for MX compatibility.

There is also a gasketed ridge around the outside of the PCB cavity in the case so that the edges of the PCB are supported, and the top case has gasketing all over it where it touches the PCB. You can accuse this case of many things, but it’s certainly not under-gasketed.

The second feature is more of a…trait. Due to the rigidity and uniformity afforded me by the standoff mounting system, I found that there was really no need for a plate. It does make building the PCB a bit more of a pain, and I recommend taking keycaps off with a bit of an angle to the left (or up on the numpad enter) so that you don’t bend any pins, but it works fine.

Next we’ve got another “trait”. It’s all 3D printed plastic. Now, I actually don’t consider the plastic part to be a downside – plastic is probably my favorite case material. But…the aesthetics of 3D-printed parts aren’t quite to my liking, and it has to be printed in multiple pieces on any 3D printer with less than a 400mm bed. I also wasn’t able to print at 100% infill without having serious issues, but I managed to keep it over 50% on all parts. On another note, the quality of the print isn’t the best. It’s as good as I could get out of the machines I’ve been using and my current level of skill. I plan to fix up my personal 3D printer to get better quality prints for production models, but I can’t guarantee better quality than you can see in the photos below. And I encourage you to look closely at said photos if you are considering buying one of these.

Despite the compromised aesthetics, I’m still pretty proud of it in every other way. It feels solid, you don’t have to worry about cross-threading due to the use of brass threaded inserts, and the cable to the daughterboard is held down by sticky putty (I plan to replace this with velcro) to make cable management painless. I use transparent filament to serve as a light pipe for the lock lights, and sorbothane for the non-slip feet. Without keycaps, it weighs 1.5kg. I think the result sounds pretty great, but you can decide for yourself.

Typing test