Build my own N-key rollover keyboard

[inornate]

I really want to get a mechanical version of n-key rollover keyboards, but commercial ones are too expensive, so I decided to build it up myself. :D

The idea was starting from this article.

http://www.dribin.org/dave/keyboard/one_html/

This article is verified with this little experiment - using arduino (simple embedded system) + 2x2 matrix setup

http://www.youtube.com/watch?v=4-QCBUkoldE

you can see this accepts any combination of the key stroke without ghosting.

Now it's time to roll!



Sorry for Koreans in the image. It is -
   1. Cutoff the patterns on the row-side circuit.
2. Scratch the coating over the copper so that naked copper surface is exposed.
3. Surface mount the Diode (the direction should be considered depend on how controller is built. In this case, the diode direction is row->col). In lower one case, additional rewiring needed.



diodes. - SMD type 1N4148


Pattern cutoff and fluxed


Mount the diodes


Mounting done

I used the LIMKB controller(A custom keyboard controller built in korean keyboard community, http://otd.kr) without ghost check logic.

As the result, I could get the USB version N-key rollover Apple Standard I. It works very well. (Although I can input only 6+1 keys due to USB keyboard protocol)

[keyb_gr]

That looks like a good bit of work there. One should be handy with a scalpel and of course a soldering iron for such a mod. Great to hear it works.

I'd try it with a modern-day Cherry PCB though. Seems like one would only need to swap the wire bridges in the switches for diodes (plus controller swap) - still fussy enough, but at least that only involves soldering and opening switches. While the 102-key G80-3000 boards I've seen had the wire bridges unconnected for the most part, they are very much in use in 105-key samples with Win keys (definitely confirmed for my G80-3000HEMDE here, would need to check some newer boards to be 100% certain).

[lowpoly]

Great mod, inornate. Like an instruction manual to convert any switch pcb 'board to NKRO.

I'd try it with a modern-day Cherry PCB though. Seems like one would only need to swap the wire bridges in the switches for diodes (plus controller swap) - still fussy enough, but at least that only involves soldering and opening switches. While the 102-key G80-3000 boards I've seen had the wire bridges unconnected for the most part, they are very much in use in 105-key samples with Win keys (definitely confirmed for my G80-3000HEMDE here, would need to check some newer boards to be 100% certain).

On current Cherry pcbs the wire bridges are not used as a placeholder for diodes. They are used to lead traces over others to avoid a double sided pcb. For ex. some keys from the bottom row are led to pcb traces running along the F-key row. Crossing the pcb vertically like that is done by 'jumping' from wire brigde to wire bridge. The wires have no logical connetion with the switches they are placed in.

Edit:spelling

[keyb_gr]

I just inspected my G80-3000HEMDE PCB closely, and it seems you're right. Several switches don't have their wire bridges connected at all, on a few others they connect to unrelated traces.

Too bad - that would have been neat. The old 102-key layout used far fewer of these bridges, but then it had another matrix as well.

[sixty]

I am surprised to see that this is possible. Good job on this mod.

PS: I am wondering since a while.. how much does one ready-made LIMKB PCB cost? I see this controller used so often in Korean keyboard mods.

Sadly there is no real "open controller" for the English community, I am kinda of jealous there!

[lowpoly]

Is there any info on the LIMKB that would not require me to go through google translate?

[inornate]

I am surprised to see that this is possible. Good job on this mod.

PS: I am wondering since a while.. how much does one ready-made LIMKB PCB cost? I see this controller used so often in Korean keyboard mods.

Sadly there is no real "open controller" for the English community, I am kinda of jealous there!

one PCB + all parts costs around 7,000 KRW (around 6 USD). To achieve this price, the pcb moq(minimum order quantity) should be large enough(at least more than 100 plates). Anyway you can build this on any general purpose board and I think that's cheaper for individual use.

Is there any info on the LIMKB that would not require me to go through google translate?

Not at now. I'm preparing official and international site for the Aikon (LIMKB is now Aikon), so please wait

[rdh]

Not at now. I'm preparing official and international site for the Aikon (LIMKB is now Aikon), so please wait

Excellent news -  I'm looking forward to it!

[sixty]

Excellent news -  I'm looking forward to it!

Second that. Awesome news. I have been looking around for info on the LIMKB for a while, finally somehting is coming up

[sixty]

I just checked our your other video on youtube (about programming the Aikon).. that looks too good to be true. I am really looking forward to seeing an English site now. By the way.. you are now the 2nd official developer for this controller? Even better to have you around here!

[Meyvn]

Could this design be made to work with BS keyswitches?

[itlnstln]

No - the hammers hit a membrane. No place to put the diodes.

It's a shame, too, since this kinda precludes you from doing a bunch of cool mods.
 
And prevents Unicomp from creating new keyboards.
 
And would prevent lowpoly from using BSs in his keyboard.
 
And... well, you get the picture.

[itlnstln]

... or try and jam a subwoofer into it. I like the fun mods better.

You should do this with the Boscom.  You would have more air space to work with, unless you plan to work with IB drivers (but I think IB sucks).

[itlnstln]

If you have room, you should try a passive radiator.

[Meyvn]

No - the hammers hit a membrane.  No place to put the diodes.

Ah.  So I'd be better off stealing materials from a rubber dome NKRO board?

[itlnstln]

Ah. So I'd be better off stealing materials from a rubber dome NKRO board?

I would think that might work, but I don't know how a BS membrane might differ from a rubber dome one.

[Meyvn]

What's different about the two membranes that would preclude their compatibility?

[Meyvn]

See the Boscom pic above and the membranes.  You'd have to get a diode onto every switch.  Now there IS a conductive glue - you could try scraping the traces with a Xacto knife and gluing a diode to each pad.  

Please report back if it works.

Thanks for the link.  I'll definitely be experimenting soon.

[lowpoly]

The membrane traces are on the inside between the membranes. This will make it even more difficult.

[roadblock2thesun]

Wouldn't attaching diodes to the membrane distort it, causing unwanted key presses? The membrane needs to be flat to work properly, correct?

[JBert]

Hmmm... You would have to get quite inventive to solder those diodes (or their wired equivalents) to the key matrix without getting the membrane to smolder in the process.

[lowpoly]

Not solder, glue. And as it has to be glued to the inside of the membrane, the wire has to be very, very thin.

[JBert]

Alright, I forgot about conductive glue.

Thin wires should not be impossible, but still...

[lowpoly]

The mounting plates have to be drilled anyway. So it might be possible to not just drill the plate but the two unaffected membranes too and glue the smd diode directly to the inside of the membrane. Didn't think of that before. It should be the most elegant way if the membrane layout allows this.

There will still wires be necessary (really?) if you look at inornate's sketches but a lot less.