Levinson keyboard

This project is highly experimental:

• Split keyboards sounded like an interesting concept I hadn’t tried out.

• Originally I got into mechanical keyboards out of desperation while looking for a non-sucking external keyboard for my crippled laptop, and would have been happy with scissor-switch keyboards otherwise.

  This is my first keyboard with low-profile Kailh Choc switches, after the Cherry ML fiasco.

• The FR4 case I ordered together with the PCB turned out to be completely incompatible with the switches I wanted, so I had to design and make a custom case.

Parts

I was collecting these parts over a longer period of time:

• Keebio Nyquist revision 3 with the bottom bits broken off, making it a Levinson, which is basically a Planck split in half.

• Kailh Choc Burnt Orange switches, a heavier variant of Brown. Soldered into the PCB, as there is no space for hotswap sockets.

• MBK keycaps—​there is hardly anything better available. Minor quality issues when ordered from splitkb.com—​mainly specks of dirt in the plastics.

  It is possible to 3D print keycaps, preferably in PETG, as it’s the closest material to the proven and desired PBT. The textured PEI sheet is much better here than the smooth one, although the resulting top surface texture will still be suboptimal.

• Custom-made, very minimalistic 3D printed case on rubber feet. I designed a front cover for it, too, but it turned to be incompatible with the keycaps and just didn’t hold well onto the bottom part. I’d have to position it underneath the switches before soldering, making it a flexible plate.

  Originally I fucked up by assuming the metric system, and used rounded values measured by hand with a digital caliper; however, the keys are actually 3/4" apart, making it 19.05 mm. This had real consequences in undesired warping of the whole assembly. Eventually I found the "official" PCB outline in SVG format, which confirmed my suspicion.

Putting it together

There are some surprising aspects to this build:

• It needed to be programmed as a Nyquist rev3, using a 4x12 layout. The default_4x12 QMK keymap is fairly WTF and I’ve had to port my Planck file almost in its entirety, which just worked and I gained RGB controls.

• The half that you connect a USB cable to will take over the other half. Both can be configured differently. Both need to be flashed separately.

• It takes about 0.44 W and with underlighting up to as much as about 1.85 W when doing the gay and annoying shifting rainbow animation.

Evaluation

In general, the project has been a success, and I’ll keep using this keyboard:

• Kailh Choc is absolutely nothing like Cherry ML. It’s a good switch. But the Burnt Orange variant might be a bit too heavy with such a short amount of travel, even when coming from Cherry MX Clear. The Brown variant would certainly be more than adequate.

• The concept of split keyboards seems alright. It feels rather good to have that distance between the halves.

• The smaller size of the keyboard is rather pleasant, albeit the cable connecting the two halves somewhat ruins it. I haven’t been able to find any decent replacement cable at all—​it becomes even worse with USB C.

The negatives, coming from my Planck keyboards, are as follows:

• For a while it’s harder to locate keys visually, because there isn’t that 2u spacebar, and none of my MBK keycaps are labelled. But I’ve got used to it already.

• Typing single-handed is much, much harder. I used to do that a lot, e.g., while eating, and holding various other stuff in one hand.

• It hurts my fingers, mostly my thumbs on the bottom edge. This might be resolvable by greatly increasing the slant of the keyboard. Though my thumbs seem to have adapted.

• The underlighting is useless with its sparseness of LEDs, and draws an insane amount of power at its peak. I haven’t yet looked at diffusers.

• I somewhat miss the Planck’s speaker and start-up melody.

Future

Reprint the case with increased slant, the farther edge could be about 1 cm higher. Unfortunately, this makes the USB cable stick out into the air.

Should I make another keyboard like this, get one that doesn’t use TRRS jack connectors to connect the two parts, such as the rev3.5 that came out later. The TRRS jack socket is a big and awkward part, the same goes for the plug.

It also seems desirable to put the switches on a steel plate, which requires a different design than with Cherry MX style switches. I’m not missing the plate at all, however, although the MX plate was useful to position the switches before soldering.

It’d also be good to try out making a PCB for the lower profile version of Kailh Choc, or the regular one but with hotswap sockets.