Shobo - thinnest wireless split

Portability has always been an important factor in the design of ergonomic keyboards.
If you have the best keyboard in the world tailored perfectly to your needs, but it's bulky and the majority of the work is still done on a laptop board - the ergonomic benefits of your home board do not really matter.

A small split keyboard is inherently compact, but thickness can still be an issue.
One of my first split keyboards was the Jian:

It looks sleek and compact, but the controller was mounted underneath the PCB, similar to an ErgoTravel, increasing the bulk of the board.
Wires were another annoyance - more to carry, more to drag on the table as the board moves around, but integrating a wireless controller into an existing design is not always clean or easy.
Eventually, I moved on to modifying PCBs, and then to my own designs - I described those in an article a few years ago.

Most of my designs were focused on regular height switches, but I built quite a few boards with (almost) the lowest profile ones available - Kailh Choc V1.
Those offered a much lower profile, but still did not go far enough.

With the mechanical butterfly switches from Cherry (MX ULP) and Kailh (PG1316S), a new leap in portability became possible.
However, even those switches are 3.5 mm. in height by themselves, with keycaps adding more to that, and they need a PCB underneath.
The One Key Modules from Framework offer a design challenge beyond that of even the thinnest mechanical offerings, but offer a total height of 3.65 mm. in return.
Making a board around them while maintaining usability and (hopefully) durability was not going to be easy.

Naming

The Jian was the keyboard that inspired me to get into this hobby, to start using a split.
It was originally named after a mythical bird in Chinese legends, but that is also a name of a straight Chinese sword.

As the thumb cluster on the Jian is straight (save for the stagger), when a friend suggested a design with the same key layout on the main keys, but a curved thumb cluster, I suggested the Dao - a curved Chinese sword.
That board is also an excellent Choc design.

For my own designs, I decided to use that same naming convention.
My main project is called the Skean - a very configurable wireless split that can use anything from hotswap MX to at least ten variants of vintage switches that are no longer in production.

For this project, I picked the name "Shobo" - a small, concealable spike, perfect to underline the emphasis on portability.

Constraints

The One Key Module was originally meant for custom layouts for the Framework 16 laptop.
That purpose has informed the design of the modules in many ways, with the most relevant being the thickness - the overall keyboard module may only be 3.7 mm. in height for the laptop lid to still close safely.
With the module itself at 3.65 mm., that does not leave a lot of room.

While I do not have a Framework 16 myself (hopefully, yet), I wanted to design the split in a way that could be stowed in a laptop, with future iterations hopefully being able to charge from the internal keyboard connectors.
I wanted to keep to the 3.7 mm. height, which was going to prove very challenging.

The PCB for these switches is also the mounting plate.
Even with no PCB underneath, however, the need to add a wireless module restricts possible PCB thickness.
The thinnest (reasonably hand-solderable) wireless modules are about 2 mm. thick, and with the PCB sitting a bit higher than the bottom of the switches a 1.6 mm. PCB would not be tenable.
With thinner PCBs, however, only 0.8 mm. would support a mid-mount USB-C port - an essential component, as a top-mount port would exceed the height constraints at 3.6 mm. by itself.

An 0.8 mm. PCB with so many holes would probably not prove very stiff, so a 3D-printed case is planned for the top as a structural support.
It would also cover the 3 mm. thick battery and add a 4 mm. thick rim around the board to house magnets.

The bottom is, however, the difficult part.

The current plan has two options:

• Use an 0.2 mm. adhesive rubber sheet to cover the entire bottom of the board, removing the 0.15 mm. Mylar bottoms of the switches.
  This would be the simpler option, but the board may feel uneven with the switches sticking out slightly.
• Use an 0.5 mm. adhesive sheet to only cover the solid areas of the PCB (under the controller/battery) and maybe the rim of the board.
  This would provide a somewhat better protection for the battery, but might leave the board feeling slightly uneven as well if the rubber backing sticks out further than the Mylar switch bottoms.

Both of these options are imperfect - only protecting the battery with a thin rubber sheet worries me - but a solid bottom and traditional rubber feet would not be possible with the height required here.

The modules also add constraints on the possible columnar stagger offsets.
If you look at the Jian or Sweep photos above, you can see that the columns for the index and pinky fingers (the two innermost/outermost columns) are offset vertically from one another.
The pad layout of the One Key modules makes that very difficult - 7.5 mm. is the only offset I found that works in both a right-to-left and a left-to-right direction.
With other offsets, mounting pads start to overlap contact pads.
Even the current design only works because I omitted the LED pads, as those usually have a significant impact on wireless efficiency.

Overall, the design feels more geared towards ortholinear boards, or perhaps row stagger.
The fact the mounting tabs are scooted closer to the center line on the top and bottom, plus the relative lack of pads there allows for the rows of switches to be offset horizontally more flexibly than what is possible vertically.
Nevertheless, I am reasonably happy with the stagger I managed to get working, and more flexibility could be achieved by increasing the switch spacing.

Magnets

A curious reader might have noticed a detail in the Sweep photo above: each rubber foot has a depression next to it.
That depression is for the matching foot on the other half - the case for the board has magnets embedded in it, so the halves can be snapped together.

Snapping the halves back-to-back never sat right with me, however, as this configuration leaves the keycaps exposed.
The alternative would require a thick border around the board to embed the magnets in, and I could never arrive at a design that felt right... Until the Shobo.

The relatively wide spacing of the mounting tabs on the One Key modules leaves a lot of space to embed rectangular magnets next to the switches.

The magnets would be glued into the 3D printed top case, which would sit flat with the keycap tops.
The halves should then snap together top-to-top, with the keycaps securely in.
The resulting transportation solution should look like the Jarne Blade, but without the external case.

Schematic

Finally, a note on schematics, as power efficiency is incredibly important in such limited designs.

All of my schematics, just as the majority of the open source wireless keyboard movement started by re-arranging bits of the nrfmicro - these boards would not be as far along as they are today if not for Joric's incredible work.

However, the field has also evolved beyond that due to the work of the ZMK project and Pete Johanson in particular, ebastler and many others.

The Shobo utilizes the nRF52840 in its power-efficient high voltage mode.
The integrated design allows for the omission of unnecessary components - there is no need to include an external LDO if there is nothing for it to power, for example - while including nice-to-haves like ESD protection.

The design is meant for a 100 mAh battery, which should last for about a week.
Please consult the ZMK Power Profiler (choose the nice!nano V2, the schematic's power efficiency is equivalent) for a more precise estimate.

The schematic also eschews a power switch in lieu of a dedicated soft off button - allowing the keyboard to be turned off just like a phone would be.
This mode ensures the board does not accidentally wake up on an errant keypress and drain the battery during transportation.

Future

The current revision of the Shobo is very early - an alpha at best.
The schematic has been tested on my previous designs, but the case is yet to be finished and the rigidity of the board as a whole has me worried.

Still, when it is all assembled, the next version would ideally include some method to power/charge the halves inside the Framework 16.
The current vision is a carrier board with a power connector for the input module interface that would reside in the 16 permanently, with the split halves snapping into it magnetically, charging via a magnetic connector.
Please give the project a star to follow its future!