GH-122 Making a Stupidly Large MX (or Alps) Keyboard

[samwisekoi]

PC-122 case from Unicomp came in today, and I took some measurements.

First off, the valley is approximately 0.325" between banks of CHerry MX switches.  It looks like it will be tight side-to-side and loose top-to-bottom using standard 0.750" spacing.  The cruciform nav cluster looks worse than it is, especially if tall keycaps with sloping sides are used.  (I would not use DSA keycaps.)

Second, there is a convenient place to put the PCB in the bottom section of the case, and even a couple of non-horrible spots to secure it.

The main changes to the existing PCB design are an extra .125" in the three valleys plus some additional height to fit into the existing PCB guides.  I will make those changes and see where that leads in terms of other expansion opportunities.

TL;DR The GH-122 PCB can be adapted to fit into the Unicomp PC-122 case, which can be puchased for only $20.

 - Ron | samwisekoi

p.s.  @SpecTP, if you'd care to donate one of those keyboards on the shelf, I would be happy to send you a pre-paid mailer.  I also used to use them to write some RPG, although it was on an older S/38!

[AKmalamute]

There's a linux distro for the S/390 IBM systems ... is that at all related to what you're talking about?

And SpecTP: The fact those keyboards are nigh indestructible, and their heft and horizontal stature warn you ahead of time, makes them quite desirable to a certain crowd.

I'd take one off you're hands if you're running out of storage space for 'em :-)

[0100010]

Could you fit the extra key locations on the PCB anyway?  That way for someone who wants to put it in a Unicomp case, they can populate only the switch positions needed.

For others, who would like some of those additional keys - the traces and key positions are available.

Edit : Something like this - http://www.keyboard-layout-editor.com/#/layouts/28751652c7dd637dd2edf475c784e6db

[0100010]

Also, thinking about the useful and not-so-useful expansion proposals, I have come up with some simple changes that don't impact the size of the master PCB.

I added:

 - Four keys in the upper left, with an optional 2x position.
 - Four keys in the Fn block, to enable expansion, movement, or re-insertion of gaps between blocks of eight.
 - Six keys above the arrow cluster to enable some or all of the PRINT/PAUSE section to be added.

Let me know your thoughts!

 - Ron | samwisekoi

Or even like this : http://www.keyboard-layout-editor.com/#/layouts/dc9f05bf859742bb4a02d3d27fb76905

[samwisekoi]

As long as they fit among the thousand traces, yes.

Extra switch positions don't add rows or columns, so as long as they don't interfere or add cost, why not?

 - Ron | samwisekoi

[jacobolus]

If someone wants a stupidly large Alps keyboard, and has some money to burn, this is pretty ridiculous:
http://www.ebay.com/itm/231329823834
Ortek MCK-142PRO

I知 guessing complicated white Alps, but who knows.

[samwisekoi]

If someone wants a stupidly large Alps keyboard, and has some money to burn, this is pretty ridiculous:
http://www.ebay.com/itm/231329823834
Ortek MCK-142PRO

I知 guessing complicated white Alps, but who knows.

Show Image

Now THAT is a layout I could get behind!

[0100010]

If someone wants a stupidly large Alps keyboard, and has some money to burn, this is pretty ridiculous:
http://www.ebay.com/itm/231329823834
Ortek MCK-142PRO

I知 guessing complicated white Alps, but who knows.

Show Image

That.  Is. Almost. Perfect.

Change that ISO enter to ANSI and add some 1x keys between the Alt / Ctrl.

Amazing.

[Melvang]

If someone wants a stupidly large Alps keyboard, and has some money to burn, this is pretty ridiculous:
http://www.ebay.com/itm/231329823834
Ortek MCK-142PRO

I知 guessing complicated white Alps, but who knows.

Show Image

That.  Is. Almost. Perfect.

Change that ISO enter to ANSI and add some 1x keys between the Alt / Ctrl.

Amazing.

I concur for prototyping at least.  If this makes it to a group buy, do this or have one option for Iso and that's it.

[Melvang]

If someone wants a stupidly large Alps keyboard, and has some money to burn, this is pretty ridiculous:
http://www.ebay.com/itm/231329823834
Ortek MCK-142PRO

I知 guessing complicated white Alps, but who knows.

Show Image

That.  Is. Almost. Perfect.

Change that ISO enter to ANSI and add some 1x keys between the Alt / Ctrl.

Amazing.

I concur for prototyping at least.  If this makes it to a group buy, do this or have one option for Iso and that's it.

[Niomosy]

More and useful keys?  I like it.

Now we just need a split/ergo version, too!

/runs

Go see the GH36 project in my sig.

 - Ron | samwisekoi

I would honestly only use that as a standalone 10-key pad type solution.  A tented 122 ANSI layout keyboard (MS Ergo style, not Ergodox style) would be perfect for me.  It keeps all keys in their standard location, is tented, and gives me a whole slew of keys I can use for volume and media without the need of having to hit more than 1 key at a time.

[SpecTP]

 - Ron | samwisekoi

p.s.  @SpecTP, if you'd care to donate one of those keyboards on the shelf, I would be happy to send you a pre-paid mailer.  I also used to use them to write some RPG, although it was on an older S/38!

I'll keep you in mind if we tag the inventory for removal.  They're all a mix of IBM and Unicomp model M's.

[samwisekoi]

Here is the matrix I am going to use as a base.

I'll add ANSI, ISO, Winkeyless and possibly JIS and/or a split spacebar.

 - Ron | samwisekoi

[Krogenar]

Here is the matrix I am going to use as a base.

I'll add ANSI, ISO, Winkeyless and possibly JIS and/or a split spacebar.

 - Ron | samwisekoi

You could put a handle on the short end and use it as a broadsword.

[0100010]

Here is the matrix I am going to use as a base.

I'll add ANSI, ISO, Winkeyless and possibly JIS and/or a split spacebar.

 - Ron | samwisekoi

[insert dancing banana gif]

[jacobolus]

If someone wants a stupidly large Alps keyboard, and has some money to burn, this is pretty ridiculous:
http://www.ebay.com/itm/231329823834
Ortek MCK-142PRO

I知 guessing complicated white Alps, but who knows.

Show Image

Looks like someone bought this monster. Was it someone here? If so, let us know what it痴 like when it arrives. (Is it as ridiculous in person as it looks from the picture? :p )

[0100010]

This layout would almost work on that PCB - except for the F row :

http://www.keyboard-layout-editor.com/#/layouts/2a48b060fef5b575e0d9d4b468fdd807

[samwisekoi]

This layout would almost work on that PCB - except for the F row :

http://www.keyboard-layout-editor.com/#/layouts/2a48b060fef5b575e0d9d4b468fdd807

Fine.  This is the matrix...

 - Ron | samwisekoi

p.s.  Although this makes me want to add just one more row above...

[0100010]

I would be fine with an additional row up top.  After seeing that Ortek 142 Pro, that's hard to beat as far as stupidly large layouts go.

[kalrand]

Might we be moving far away from "using the unicomp cases" plan with all these extra keys?

[samwisekoi]

Might we be moving far away from "using the unicomp cases" plan with all these extra keys?

Yep.  I was just about to sit down and make some choices.  Practical and aesthetic choices.

I'd like to be able to support the Unicomp PC-122 case, as well as a standard 104. And then add extras for looks, function, and differentation.

No sword handles, however.

 - Ron | samwisekoi

[Hak Foo]

That new "F row" needs to be spaced differently if you're thinking about having it support a conventional 101-104-108 case and plate.   The gap between Esc and F1 seems to be about 1.25x, while the ones between F4 and F5 and F8 and F9 are closer to 0.5x

[samwisekoi]

That new "F row" needs to be spaced differently if you're thinking about having it support a conventional 101-104-108 case and plate.   The gap between Esc and F1 seems to be about 1.25x, while the ones between F4 and F5 and F8 and F9 are closer to 0.5x

So for a conventional 104, Escape is 1x above Tilde.  For the matrix I am working on, that position would be R2C3, not R1C1.

The vertical gap between the main body and the upper row will be 0.750".  The horizontal gaps between blocks will be 0.325" to match the PC-122 case.  I measured many cases, and the horizontal gap seems very much not standardized, with Rosewell coming closest to the Unicomp spacing.  The range I found was 0.250" to 0.375".  (The Leopold TKL I am using at the moment seems to have a horizontal gap of 0.3", measured without disassemly.)

 - Ron | samwisekoi

[samwisekoi]

Final matrix except for mods and bottom row.

Vertical gaps are 0.750".  Horizontal gaps are 0.325" to fit Unicomp PC-122 case.

Majority of upper "PF" block enables shifting left or right by one-half switch unit (0.375") to enable separated blocks of four or eight keys as in a standard ANSI 104 keyboard.  Thus both PC-122 and ANSI 104 compatibility is retained.

Full backlighting is provided.  Sub-block backlight zones are not provided, but I may add WASD backlighting.  Individual block backlighting will be supported in hardware.  I am probably going to locate CAPS and NUMLOCK LEDs above the numpad as done on IBM keyboards.  I *may* add a secondary CAPS LED position as seen on some boards.

All blocks will be removable and connectable.  The controller section will be in the upper RH corner, and will be useable as-is or detached.

Mods To Do List:
 - CAPS stepped or centered
 - ANSI, including PCB-mount stabilizers
 - ISO, probably requiring plate-mount stabs
 - Winkeyless (1.5-1-1.5-7-1.5-1-1.5 bottom row, probably with support for PCB-mount stabilizers
 - JIS and/or split spacebar support, probably requiring plate-mount stabs
 - Probably support for long spacebar(s) using plate-mount stabs
 - Possibly support for one or two "Big-Ass" Return keys using plate-mount stabs.

Clearly I've abandoned the 208-switch matrix.  That was for both practical and aesthetic reasons.  This board is still 20 inches wide and 7 inches tall before you add a case!  Big enough, IMHO.

Anyhow, I'd appreciate feedback, especially from the list of members signed up for the initial run.

Have I covered all the bases needed for GeekHack's first mega-board?

 - Ron | samwisekoi

[AKmalamute]

It'll be a long time before I can get in on this, but I like everything I see, assuming you find some way to make the WKL bottom row of the day sit beside the 1.25 mod layout that "standard" keycap kits provide (will this require two full kits, now?)

[0100010]

I think that will work.  Support both the PC122 case and have the extra switch positions for additional keys if desired.

[samwisekoi]

WTF is the JIS Bottom Row?

I have not found a definitive reference, so I'd like confirmation that the JIS bottom row is:

1.25 - 1.25 - 1.25 - 1.25 - 3.25 spacebar - 1.25 - 1.25 - 1 - 1 - 1 - 1.25  http://deskthority.net/wiki/ANSI_vs_JIS

I have also seen:

1.5 - 1 - 1.5 - 1.5 - 2.5 spacebar - 1.5 - 1 - 1 - 1 - 1 - 1.5  http://en.wikipedia.org/wiki/Keyboard_layout

and

1.5 - 1.25 - 1.25 - 1.5 - 2.5 spacebar - 1.5 - 1.25 - 1 - 1 - 1 - 1.25   http://www.win.tue.nl/~aeb/linux/kbd/scancodes-8.html

I cannot find a written spec in English, but it appears to be Japanese Industrial Standard JIS X 6002:1980.

Lysol Ivan, you asked for it, so please tell me what you think it should be!

Thanks all,

 - Ron | samwisekoi

[bianco]

looks good!
personally, i prefer the off-center caps, and the 3x1.25--6.25--4x1.25 for the bottom row

[MOZ]

Haha, the JIS layout is confusing, I don't think there is a standard, for various projects of mine, whenever I tried to find a JIS layout, there were too many variants.

[IvanIvanovich]

Whatever key sizes doesn't really matter as long as they are all normal sizes that are easy to source from anywhere. 2.75 space is good. I like the one Cherry used fine, but I wondered what was the point of using a 1.5 key when could have made right win or menu 1.25 instead.


Or something like this is fine

[samwisekoi]

Ok, unless someone (Ivan?) has strong opinions otherwise, I am going to go with the following layout for JIS:

1.5 - 1 - 1.5 - 1.5 - 2.75 spacebar* - 1.25 - 1 - 1 - 1 - 1 - 1.5

*I went with a 2.75x spacebar because it is an attainable keycap size.

In addition to being made up from attainable keycaps, it also has lots of direct overlap with ANSI and Winkeyless, so it fits well.

Exploded diagram of target bottom rows attached.  (1x15 row is for reference only!)

 - Ron | samwisekoi

p.s.  Ok, Ivan replied while I was typing, so it looks like we are in sync.  Thank you sir!

p.p.s.  Alps-compatibility may be compromised by all the overlap.  As a result I may have to restrict Alps switches to a single bottom row layout, probably ANSI.

[samwisekoi]

And here is where we are.  Looks like no fatal overlap between the various bottom rows.

 - Ron | samwisekoi

[berserkfan]

I don't personally use a 2.75x (I prefer 1x), but yes, I support any variant with a commonly available keycap size. Right shift sizes are common and lots of people have them and they use 2x stabilizers which are also common. And right shift is often the same profile as space bar which beats using a Row 4 relegendable 1x on the space bar. So this is a practical solution.

[0100010]

On that 3rd modifier from the right (with the 1x / 1.25x / 1.5x positions all next to each other); does that one through hole where the Alps pin touches the Cherry pin present any issue?

[samwisekoi]

On that 3rd modifier from the right (with the 1x / 1.25x / 1.5x positions all next to each other); does that one through hole where the Alps pin touches the Cherry pin present any issue?

Yes, any touching of a pin_1 and a pin_2 would be fatal.  But take a look at the multi-position CAPS switch mounts to see my solution.

But the proof will be in the pudding; or in this case the traces.  We'll see what I can do when I get there.

Thanks for pointing that spot out.  It is very helpful!

 - Ron | samwisekoi

[0100010]

On your caps lock layout, looks like the through hole is only partially plated on the top and bottom 'sides', so one could solder the appropriate pin to only needed 'side' of the hole?  Side in this case is probably not the best word though, since one would only be soldering from the back side of the PCB.

[samwisekoi]

On your caps lock layout, looks like the through hole is only partially plated on the top and bottom 'sides', so one could solder the appropriate pin to only needed 'side' of the hole?  Side in this case is probably not the best word though, since one would only be soldering from the back side of the PCB.

Yes, exactly.  Conductor and soldermask are only present on the lower half of the Alps pads, although the Alps holes remain to spec.

This will make it tricky but possible to solder an Alps switch in that position, and there is no short created between any of the eight switch pads.  I'll do the same as needed for the bottom row.

 - Ron | samwisekoi

p.s.  I've attached a close-up of the mod made to the pads on the impacted bottom-row switch positions.

[0100010]

One other thought.  Maybe have a small area around the through hole that doesn't have a solder mask over the trace, so one could cut the un-needed trace with an exacto knife?

[samwisekoi]

One other thought.  Maybe have a small area around the through hole that doesn't have a solder mask over the trace, so one could cut the un-needed trace with an exacto knife?

That is an interesting idea.  I'd worry about long-term reliability (and noob soldering) but I'll take a look and see if that can work.

It would certainly make the LED-chooser circuit on the GH36 easier to use.

Thanks!

 - Ron | samwisekoi

[samwisekoi]

Finally an update here...

This is a monster PCB, so I am trying to get in 30 minutes of layout time each morning to keep progress moving and keep mistakes to a minimum.  I've hit a design milestone:  All switches, rows, and columns have traces run.  All holes and pads are done, and I am ready to do LED and controller wiring.

FYI, there will be a total of 45 pins going to the controller!

I am planning to have individual LED control for each block, plus WASD (all PWM.)  Including non-PWM indicator LED circuits, here is the planned list of LED circuits:

NUM
CAPS
FN
FN2
LED1
LED2
LED3
LED4
LED5
LED6
LED7
WASD

I have not worked out cut lines yet.  I can't support TKL, and anyhow, what is the point?  I will support full-size and full-width. 

Does anyone anticipate building a tall TKL?  That is the one that will be hardest to support.

Anyhow, here is a screen-shot of the PCB so far.  I have a full-resolution version, but that is 3M compressed, so I have not posted it.  Ask and ye shall receive!

Anyhow.  That is the news from Lake Wobegon for today.

 - Ron | samwisekoi

[0100010]

Looking good.  Only see some nitpicky stuff, nothing function impacting.

[Puddsy]

i unhid this forum just to say that i love this idea and will certainly be buying one

[HoffmanMyster]

i unhid this forum just to say that i love this idea and will certainly be buying one

Why do you hide the best subforum...?   

[bianco]

Finally an update here...

This is a monster PCB, so I am trying to get in 30 minutes of layout time each morning to keep progress moving and keep mistakes to a minimum.  I've hit a design milestone:  All switches, rows, and columns have traces run.  All holes and pads are done, and I am ready to do LED and controller wiring.

FYI, there will be a total of 45 pins going to the controller!

I am planning to have individual LED control for each block, plus WASD (all PWM.)  Including non-PWM indicator LED circuits, here is the planned list of LED circuits:

NUM
CAPS
FN
FN2
LED1
LED2
LED3
LED4
LED5
LED6
LED7
WASD

I have not worked out cut lines yet.  I can't support TKL, and anyhow, what is the point?  I will support full-size and full-width. 

Does anyone anticipate building a tall TKL?  That is the one that will be hardest to support.

Anyhow, here is a screen-shot of the PCB so far.  I have a full-resolution version, but that is 3M compressed, so I have not posted it.  Ask and ye shall receive!

Anyhow.  That is the news from Lake Wobegon for today.

 - Ron | samwisekoi

it looks good! so good... i might get 2 of these. not unless you're planning a v2.
btw, as i'm new to pcb design my apologies in advance for asking --- what are those lines in the upper right part of the board?

[dorkvader]

it looks good! so good... i might get 2 of these. not unless you're planning a v2.
btw, as i'm new to pcb design my apologies in advance for asking --- what are those lines in the upper right part of the board?

It's a work in progress so they won't look like that on the finished board. I would expect they show trace spacing so he can check the amount of space those 45 incoming traces will take up to ensure he has enough space in the controller area.

i unhid this forum just to say that i love this idea and will certainly be buying one

Why do you hide the best subforum...?   

It's ok. I set him straight. Welcome to GH, tacticalstache.

[bianco]

it looks good! so good... i might get 2 of these. not unless you're planning a v2.
btw, as i'm new to pcb design my apologies in advance for asking --- what are those lines in the upper right part of the board?

It's a work in progress so they won't look like that on the finished board. I would expect they show trace spacing so he can check the amount of space those 45 incoming traces will take up to ensure he has enough space in the controller area.

i unhid this forum just to say that i love this idea and will certainly be buying one

Why do you hide the best subforum...?   

It's ok. I set him straight. Welcome to GH, tacticalstache.

understood. thank you sir

[MOZ]

i unhid this forum just to say that i love this idea and will certainly be buying one

Why do you hide the best subforum...?   

BEST.SUBFORUM.EVER

[samwisekoi]

Thanks all!

Looking good.  Only see some nitpicky stuff, nothing function impacting.

What? The wiggly traces that avoid invisible obstacles?  That was to enable standardized switch cells.  Anything else?  Tell me now while they are still just pixels.

(Don't count the between-block trace stubs.  Those will be used or removed.)

it looks good! so good... i might get 2 of these. not unless you're planning a v2.
btw, as i'm new to pcb design my apologies in advance for asking --- what are those lines in the upper right part of the board?

It's a work in progress so they won't look like that on the finished board. I would expect they show trace spacing so he can check the amount of space those 45 incoming traces will take up to ensure he has enough space in the controller area.

^ This.  0.006" can carry 0.59A and 0.010" can carry 1.14A.  Those test blocks show the difference in density between the two trace widths.  I used 0,010" in the blocks, and will use it for LED power distribution.  I'll use 0.006" for signal traces in the controller area and for signal distribution.  Ground and VCC will use 0.010" or larger.  In theory the board could pull 2.5A if all LEDs are populated and lit at 100%, but 2.5A far exceeds the USB spec.  Also, it would require 0.050" traces!  If I add an external power port, I'll use traces that big for ground and +5v.

Last call for Jumbo TKL support!

 - Ron | samwisekoi

[samwisekoi]

Name Change?  GH-132?

So this is way bigger than a 122-key keyboard.  I counted 138 max switches out of something like 200 switch positions available.  However, visually it is a 122 plus two extra blocks for a nominal one hundred and thirty two keys.

So I propose calling it, henceforth, the GH-132.

1:1 image attached in the White-on-Black the PCB will actually be.  I rendered the GH132 logo in silkscreened "traces" that represent the logo from many posts back.

Thoughts?

 - Ron | samwisekoi

[berserkfan]

Finally an update here...

This is a monster PCB, so I am trying to get in 30 minutes of layout time each morning to keep progress moving and keep mistakes to a minimum.  I've hit a design milestone:  All switches, rows, and columns have traces run.  All holes and pads are done, and I am ready to do LED and controller wiring.

FYI, there will be a total of 45 pins going to the controller!

I am planning to have individual LED control for each block, plus WASD (all PWM.)  Including non-PWM indicator LED circuits, here is the planned list of LED circuits:

NUM
CAPS
FN
FN2
LED1
LED2
LED3
LED4
LED5
LED6
LED7
WASD

I have not worked out cut lines yet.  I can't support TKL, and anyhow, what is the point?  I will support full-size and full-width. 

Does anyone anticipate building a tall TKL?  That is the one that will be hardest to support.

Anyhow, here is a screen-shot of the PCB so far.  I have a full-resolution version, but that is 3M compressed, so I have not posted it.  Ask and ye shall receive!

Anyhow.  That is the news from Lake Wobegon for today.

 - Ron | samwisekoi

Seriously, I can't tell whether to call you a genius or a madman when I see how complicated everything is. This is one awesome PCB, and it is only a shame that most people including me will never be able to fully appreciate the awesomeness of what you have done.