Applying lessons learned - two more keyboard iterations

The journey continues (see previous post for more info...)...  This post is really just for a few folks who had asked for updates.

I have learned a few more lessons:
  • Expect it to take 3-4 months (100-200 hours) per full board build - a long feedback loop is just the nature of the game, especially with my availability
  • Expect to get some crucial feature wrong - design for design failure
  • Expect to break parts or have bad solder joints - design for repair
  • Expect to be tempted to rush things
If you want to follow along or look for more detail or laughs or horror, I keep a build log in a google doc here: It's super long, but often updated daily

What follows is a brief summary of the keyboard development work between July 2022 and April 2023.  I'll leave out the business analogies (continuous improvement aside).  For the curious, the 'mk13' (mark = old British engineering thing, I think?) indicates the thirteenth time I started from scratch - and the maths-inclined amongst whoever reads this might note that 3-4 months times 16 doesn't equal 18 months of me walking these paths - some designs got thrown out real quick-like.

mk13 - the daily driver

I'm typing this post using this board - the ergonomics of it seem to work for me - and indeed it's made it possible to continue working (as I can no longer use a mouse).
  • Thumb trackball positioning improved from mk9, along with auto-mouse
  • Many more keys: 7x6 + 6 thumb keys per side
  • Thumb clusters that don't destroy my thumb tendons
  • Larger palm rests
  • Better use of layers and combinations
  • Screens - why? peer pressure, that's why.  Also was an interesting option for making visible the invisible workings of the firmware.

The outcome?
  • Some teething issues around solder joints...ok a lot of issues with solder joints...
  • The design, in practice, was not 'operable'
  • There was a qmk bug, maybe?, that brought performance issues when using RP2040 and split transport, and/or my firmware/hardware had issues; had to turn off the screens eventually to recoup enough performance
  • But finally, something that would allow me to keep working!  Success.
Except until the one day I messed up a firmware update for it, and realized how fragile my situation was!  So now we need a backup...

mk15 - the 'definitely going to be better' improvement/backup

End of November 2022 through April 2023 produced this...
  • 30-degree tent (BKB's Charybdis as a tracing aid)
  • 4x6 + 6 thumb keys / side
  • Dual trackballs (left = vert/horizontal scroll, right = mouse)
  • Indicator strip at back (colour changing based on stuff like 'shift-hold is on')
  • Encoders change per-layer (left = functions, right = layer select+hold)

The outcome?
  • My symptoms have progressed since the design of this board started; I'm physically unable to use it.
  • Dual encoders (Falcon pcb is great!!) and trackballs are pretty super duper.
  • Dpad is where it's at!! (5-way switch giving directional arrow keys)
  • Learned some lessons around design-for-repair....again...
  • Learned what does NOT work for me! \o/
In the meantime, I have mk13 and mk15 together on my desk, pulling functionality from both! mk17 - seventeenth time's the charm, right?

mk16 was fun, but didn't work out.  Between issues with finding the right lofting look and packaging and realizing that I should use dual encoders/side...well, lessons learned.  Some of the design aspects will transfer over.

mk17 is going to be an improved and more maintainable mk13, with all the lessons of the in-between...we'll see how it pans out.


Popular posts from this blog

Learning through failure - a keyboard creation journey

Learning Opportunities - Watching/listening list

DFSR - eventid 4312 - replication just won't work