deskthority

Edit: The original post is about the handwired ISO50. There is a PCB version of the keyboard; see below. Links to fabrication files for either version are listed here.

After a few fruitful weeks of lurking, I think it's time to show what's come out of it.

While I like typing on the DASK3 I've been using for the past seven years, I find the voluminous tumours keyboards came to suffer from since their invention rather annoying. The Pok3r dissolves those swellings very nicely only to render itself useless (to me) by means of its "programmability" which involves putting a no-op "Pn" key where my muscle memory expects something else.

I would have definitely bought a MiniVan if English was my main language, but my native alphabet needs a few additional keys if not going mad is a requirement. Here is my attempt to build something of comparable beauty.

Modifications to hasu's TMK firmware and everything necessary to manufacture the hardware parts can be found at https://github.com/trebb/iso50 under an open-source licence.



Symmetry!

The whole thing weighs 600 g. With aluminium plates instead of steel, it would be 400 g.



As boring as possible: the layout

The navigation layer is inspired by the Pok3r; the mouse emulation layer comes from hasu's hhkb keymap.

Parts of the non-existing number row of the Alt Gr layer have been merged into the existing keys; the rest of the Alt Gr layer is standard. This part of the layout is probably a bit language dependent. Here, TZUIOP have non-standard bindings that are usually found in the number row.



Bottom

No visible screws. Non-rectangular foot arrangement precludes wobbling.



Gateron PBT keycaps, Cherry profile

In terms of a standard keyboard, only keys from space bar row, bottom row, home row and tab row are used. There are lots of uniform leftovers from the number row and function row.



Preparing the switches

Putting the diodes into the switches saves some space and improves insulation and mechanical stability.



Preparing the switches

Diodes shouldn't be completely naked if in close proximity to their LED cousins.



Switches in place

Some hot glue will be needed due to the poor quality of the laser cutting.

The two white dots denote switches with flipped diodes.



Shoehorning the Teensy onto the switches

Some of the holes and pins are off by half a raster step.



Spaghetti



Minimizing height

Nothing should exceed the height of the plastic parts of the switches. Clearance between plates is 6.3 mm.



Reset hole (top left); strain relief cable strap (centre right).

The stabilized enter key couldn't be divided up as standard key sets lack a 1.25u home row key.



The most expensive part of the project has arrived.

https://www.shapeways.com/model/3dtools ... fd8e4e31ed

In total, parts were about 140 € (key caps excluded).



The insulating layer on the bottom plate has been cut from a sheet of transparency.

The 3D print could probably be replaced by a nice wooden frame.



Using a stepped Tab key (CapsLock on other keyboards) improves symmetry and leaves a bit more room for the controller.



Kink protection

To replace the cable, a new cable strap is needed and a knife to trim the USB mini plug.



Initiating firmware flash

Specialized tool required.



Comparing heights: Pok3r



Comparing heights: DASK3

Das Keyboard uses up 345 % of the ISO50's desk space.



Comparing heights: a popular rubber dome keyboard

The Novatouch TKL needs a bit more than twice the desk space of the ISO50.



The ISO50 lacks a number row

But, given appropriate equipment, one can type digits nonetheless.



The ISO50 next to a shoe

Presenting its NumLock and CapsLock LEDs. ScrollLock is off, but does exist.

What a great keyboard!

And what a great first post

Pretty impressive work on the keyboard!

Thank you!

Currently, I'm cargo-cult designing a PCB version whose overall height will be reduced by 1.2 mm, I hope.

That's a really cool project. What's also cool is the Lamy 2000 in the back.

So cool... how long did it take to plan and build?

I would switch Enter and Backspace key though - for more symmetry.

t!ng wrote: ↑So cool... how long did it take to plan and build?

Building takes the best part of a day. Planning took a couple of weeks, but it involved wrapping my head around most of the technology involved.

I would switch Enter and Backspace key though - for more symmetry.

I'd love to have Backspace above Enter, but I'm not sure how to do it with keys from a standard set.

ISO50 (PCB version)
Preview:

Finally, I've come up with a design that looked sufficiently like a PCB. I ordered everything and started assembling the board. Trying to solder the IC, I ended up with this:Ok, that's not going to work. The correct (QFN) package fortunately seemed to match my PCB footprint.It took two attempts to place the tiny thing without shorts.
I wasn't entirely sure if I needed a programmer at all, but the legless variant of the Atmega32u4 comes completely empty, lacking the USB bootloader. Installing one and setting the fuses went surprisingly smoothly. Due to the coarse tolerances of the laser cutter, the steel top plate of the handwired ISO50 required lots of hot glue to become usable. The experimental top plate shown below is actually a 1.6 mm PCB. The PCB proper is 0.6 mm and the bottom plate is 1.5 mm steel. Each of them was about 5 € plus VAT and shipping.


The PCB-style top plate seems to work quite well. Diodes go into the switches as with the handwired ISO50, but this time without unclipping the switch tops. You can just push them through. Still missing: 3D-printed frame and standoffs. The finished product will be as thin as is possible for a keyboard with Cherry MX-style switches.

Really interesting board!!

Ordered the 3D-printed part of the casing. It is a bit cheaper than the hand-wired version due to its reduced height.

The ISO50 now has an official VID/PID USB identifier, thanks to http://pid.codes/.

Here are a few resources I found useful for this project:

Keyboard-Layout-Editor
GH60 (PCB example)
ALPS64 (PCB example)
Open-Source ATMEGA32U4 breakout board (PCB example)
THKB (similar project)
How to solder QFN components
Overview on bootloaders
Flashing bootloaders on AVR
ATMEGA32U4 datasheet
How to build your very own keyboard firmware
TMK keyboard firmware (lots of examples and documentation)

Thanks for those resources.
Will there be support for switch top temoval+some others layouts as well? (if e g. split enter in two keys (more ANSI like)/different shift/bottom rows..)?

Phenix wrote: ↑Will there be support for switch top temoval+some others layouts as well? (if e g. split enter in two keys (more ANSI like)/different shift/bottom rows..)?

Switch-top removability could be achieved easily and I'm willing to accept contributions.

There is no 1.25u non-bottom-row key available that could be part of a split Enter. One could imagine an ANSI Enter, thus sacrificing a home row key in favour of a new key in the outermost corner of the board. While this doesn't look like a smart move to me, long-term ANSI users may find it more comfortable this way and I would, again, accept and maintain contributions.

As to the physical bottom row layout, the PCB looks a bit crowded to my untrained eye, so any modifications of that kind would amount to a complete redesign. One possible exception would be replacing the five keys around the center by a 7u space bar.

I just mentioned it because the option to do so is always good (as long as it is not that hard to include.

I for myself would maybe try out.
For keycaps: most have some spare caps, eventough profiles arent 100% accurate..

Standoffs in place:

The 3D print is made of Nylon; this time, it doesn't contain any aluminium powder.

The case parts were about 35 €.

Non-detachable, but easily replaceable cable.

Top and bottom:Apart from LEDs and reset button, the layout remains unchanged.

The PCB version next to a hand-wired ISO50. This one has been made of Alumide and shows the outcome of a dyeing experiment.

Done!

wow! chapeau my dear!
Now U got me really into wanting to try gateron greens for my next upcoming 42% project..

tent:wq

For those who wish to build the thing while avoiding setup of the tool chain, here are the fabrication files.

• Top plate
  Hand-wired version: DXF for laser cutting.
  PCB version: DXF for laser cutting or Gerber files for a PCB-style top plate.
• Bottom plate
  Hand-wired version: DXF for laser cutting.
  PCB version: DXF for laser cutting.
• Frame for 3D printing
  Hand-wired version: STL.
  PCB version: STL.
• PCB
  Gerber files.
• Firmware
  (Should be usable for some languages and not for others as 12 AltGr layer keys are missing; contributions welcome.)
  Hand-wired version: HEX file.
  PCB version: HEX file.
• Bottom label (SVG)
  Spoiler:

  (Currently German; contributions welcome.)
  Hand-wired version:
  PCB version:

Nice to see the fjles!
Could you also add the keymap file? (IIRC in .hex chaning layout is not possible, or?)

Phenix wrote: ↑Could you also add the keymap file? (IIRC in .hex chaning layout is not possible, or?)

Here you are:
https://github.com/trebb/iso50/blob/mas ... ap_iso50.c

Using it involves compilation of TMK, so you probably want the rest of the modified files, too.

Beautiful results and amazingly clean work you're showing!

But you know that for the ultimate flat board you can shave off 1.5mm of the MX center stem and use thin tape to prevent dust from creeping in, don't you ? Seriously; I can totally understand that obsession even though it is probably totally ridiculous - but hey, if the outcome is two such well done boards all is good!

suka wrote: ↑But you know that for the ultimate flat board you can shave off 1.5mm of the MX center stem and use thin tape to prevent dust from creeping in, don't you ?

Thank you.

And of course I couldn't resist to file off the bottom of a stem bearing, which seemed to have been the part the switch bottoms out with as it increased key travel by 0.6 mm.

trebb wrote: And of course I couldn't resist to file off the bottom of a stem bearing, which seemed to have been the part the switch bottoms out with as it increased key travel by 0.6 mm.

Just a warning: Filing could be a bad idea, as plastic dust will immediately get into the switch, and also rough inside edges of the resulting hole can lead to the switch stem locking up. Snipping them off with a good wire cutter is much safer and also faster

suka wrote: ↑

trebb wrote: And of course I couldn't resist to file off the bottom of a stem bearing, which seemed to have been the part the switch bottoms out with as it increased key travel by 0.6 mm.

Just a warning: Filing could be a bad idea, as plastic dust will immediately get into the switch, and also rough inside edges of the resulting hole can lead to the switch stem locking up. Snipping them off with a good wire cutter is much safer and also faster

Snipping off is indeed more elegant than filing, but I think I'd consider the switch destroyed after the treatment. More key travel means parts get hit that aren't designed to get hit, which can't be good.

This keyboard is great! Since I saw this beauty for the first time I'm thinking of ordering the parts, so thanks to op for the files!

Who of you ordered a PCB, at which manufacturer and how much did it cost? Would it be possible to use a bigger space key and a detachable cable?

Necromancer achievement unlocked!
This thread sleeps with the fishes. Well, unless you woke it back up again. Could be worth a chance, I guess.

I am sorry not to have read and questioned it already in 2016... /s

What else could I do, instead of writing down here? Asking the OP via PM? Doesn't help anyone. Create a new thread and link it? Leads to confusion. Tbh, this accusation of picking over the bones is nonsense imho. Hope you have a great day.

True. You’re not doing anything wrong, but the chances of a topic revival after 6 years are far from good. A bit of natural arcana skill and an existing familiarity with the undead may be helpful in your endeavour.

In your position I’d have probably done the same thing. One difference though: I’d mention the gulf of time at the start, so it’s clear to passers by this is not an active project… at least not yet.