Working on a Buckling Spring 40%

User avatar
FletchINKy

06 Sep 2017, 14:48

Posted to GH earlier, sorry for the X-post, but want to get more input;

I've been making use of out of spec parts to hack together a 4x12 buckling spring board.

The goal is to create a moderately repeatable NKRO build with a Teensy so my fingers can gently caress buckling springs without horizontal stagger tunneling my carpals.

So far I've proofed two concepts; firmware reads the signal off the membrane contact without the weird paddle debounce becoming an issue (even if the contact has a little resistance from the carbon contact coating), and the membrane makes reliable contact with metal (I was concerned it would be harder with non silver epoxy).

From here, I've started a POC build on a standard steel backplate using the original mylar (so no nkro), and testing if I can reliably connect to the pressure-contact leads that normally go to the controller board. After MUCH fiddling, the top half works so long as I hold my face just right. Getting even pressure for the contacts was terrible. Binder clips are strong and look straight, but they're not :lol: . I think I'm just going to hijack the connections off the circuit board using the standard compression fitting within the case.

The bottom half is (technically) easier to connect to, because the connection points are further apart, and don't have that carbon contact coating on them.

The endgame is to design a circuit board that will serve as the bottom layer of mylar, to receive the connections from the top layer. By shorting two top 2x12 layers of mylar, I can get two logical rows out of them, and then make a 2x24 matrix that will fit onto a Teensy++ 2.0. Shorting the mylar also simplifies making reliable connection to the sheet. The circuit board will also allow diodes to be easily mounted for NKRO, while lending rigidity to the frame barrel assemblies to be screwed into.
Attachments
20170905_001417.jpg
20170905_001417.jpg (2.63 MiB) Viewed 11727 times
20170902_230114.jpg
20170902_230114.jpg (2.09 MiB) Viewed 11727 times
20170902_224849.jpg
20170902_224849.jpg (2 MiB) Viewed 11727 times
20170902_224834.jpg
20170902_224834.jpg (1.94 MiB) Viewed 11727 times
20170902_163435.jpg
20170902_163435.jpg (2.02 MiB) Viewed 11727 times
20170902_163347.jpg
20170902_163347.jpg (1.78 MiB) Viewed 11727 times
20170902_163206.jpg
20170902_163206.jpg (1.8 MiB) Viewed 11727 times
20170902_163107.jpg
20170902_163107.jpg (1.97 MiB) Viewed 11727 times
20170902_162937.jpg
20170902_162937.jpg (1.84 MiB) Viewed 11727 times
20170830_140405.jpg
20170830_140405.jpg (1.86 MiB) Viewed 11727 times

User avatar
Techno Trousers
100,000,000 actuations

07 Sep 2017, 03:27

Interesting! Why did you decide to do membrane BS instead of capacitive? It seems like it would be simpler to start with idollar's FSSK PCB drawing, reduce it to 40%, have a plate made, and put it all together with an Xwhatsit or CommonSense controller.

User avatar
emdude
Model M Apologist

07 Sep 2017, 03:53

I'm curious too. The capacitive matrix would be very simple, just a 4x12 grid; that would work with the Model F Xwhatsit. You could even incorporate an edge connector and use the beam spring version instead, with a firmware change to switch polarity. Either way, you would still get NKRO without having to worry about diodes.

If you intend to continue with your design, take a look at wcass's 65% Model M project. He did the same thing essentially: A custom Model M with a PCB serving as the bottom membrane.

User avatar
FletchINKy

07 Sep 2017, 14:26

Techno Trousers wrote: Interesting! Why did you decide to do membrane BS instead of capacitive? It seems like it would be simpler to start with idollar's FSSK PCB drawing, reduce it to 40%, have a plate made, and put it all together with an Xwhatsit or CommonSense controller.
I had access to frames and membranes, and I wasn't sure if the firmwares I was familiar with could sense the capacitance change. Unfortunately, I'm still a novice at the EE side of programming. I'd have to buy parts (individual barrels from the model F project, I guess) too.
emdude wrote: I'm curious too. The capacitive matrix would be very simple, just a 4x12 grid; that would work with the Model F Xwhatsit. You could even incorporate an edge connector and use the beam spring version instead, with a firmware change to switch polarity. Either way, you would still get NKRO without having to worry about diodes.

If you intend to continue with your design, take a look at wcass's 65% Model M project. He did the same thing essentially: A custom Model M with a PCB serving as the bottom membrane.
See above for the capacitive answer. That project is really interesting to me! Particularly that he made an acrylic barrel frame! I have access to a laser cutter and most of my builds have been in acrylic, so making that barrel frame would be easy and inexpensive.

I also thought he was doing a surface mount on the membrane, which I recently found is possible, but expensive.

So mainly an answer of familiarity. I like being able to change my layout quickly and easily, because I make frequent updates to it to try new things, or as my workflow adjusts. Switching to a new controller/firmware would be hard for me, but might be worth it.

User avatar
emdude
Model M Apologist

07 Sep 2017, 17:57

Take a look at Xwhatsit's thread. And if you're interested, I think orihalcon is still selling assembled controllers with the firmware already flashed here. The controller is very easy to set up and use. If you're interested, I also recommend reading DMA's CommonSense controller thread, as that seems poised to supersede the Xwhatsit.

You would not have to buy individual barrels, just Model F flippers; they fit in Model M barrel frames just fine for your purposes. i$'s FSSK is also relevant here: workshop-f7/f-ssk-t10744.html

On a sidenote, I've seen your recent reddit posts. Do you happen to be a Unicomp employee? It would be awesome to get some insight with what's going on with them.

User avatar
FletchINKy

07 Sep 2017, 18:29

emdude wrote: Take a look at Xwhatsit's thread. And if you're interested, I think orihalcon is still selling assembled controllers with the firmware already flashed here. The controller is very easy to set up and use. If you're interested, I also recommend reading DMA's CommonSense controller thread, as that seems poised to supersede the Xwhatsit.

You would not have to buy individual barrels, just Model F flippers; they fit in Model M barrel frames just fine for your purposes. i$'s FSSK is also relevant here: workshop-f7/f-ssk-t10744.html

On a sidenote, I've seen your recent reddit posts. Do you happen to be a Unicomp employee? It would be awesome to get some insight with what's going on with them.
Lots to read there! Goodness, sensing high and low gets better and better. Glad to know about the flippers, would make the barrier to entry much easier. Still, I plan on proceeding with the PCB/diode design to make sure it works. Not sure how viable it will be long term given the membrane will be contacting bare copper to make contact. I expect oxidation and other issues, but there's probably something adhesive and conductive that isn't $40 for a 4oz tube like the silver epoxy used on the membrane sheet.

I am a new Unicomp employee, but I can't say how much insidght I have. I'm mostly working on weird manufacturing production issues and answering the phone. :oops:

User avatar
emdude
Model M Apologist

07 Sep 2017, 18:46

FletchINKy wrote: Lots to read there! Goodness, sensing high and low gets better and better. Glad to know about the flippers, would make the barrier to entry much easier. Still, I plan on proceeding with the PCB/diode design to make sure it works. Not sure how viable it will be long term given the membrane will be contacting bare copper to make contact. I expect oxidation and other issues, but there's probably something adhesive and conductive that isn't $40 for a 4oz tube like the silver epoxy used on the membrane sheet.
Okay, best of luck then, looking forward to your updates! Deskthority is home to many fun buckling spring projects, so please stick around! :D
FletchINKy wrote: I am a new Unicomp employee, but I can't say how much insidght I have. I'm mostly working on weird manufacturing production issues and answering the phone. :oops:
Still very neat, I don't think we've ever had a member who was also a Unicomp employee. Tell them to hurry it up with the Unicomp SSK! ;) :twisted:

User avatar
depletedvespene

07 Sep 2017, 19:17

FletchINKy wrote: I am a new Unicomp employee, but I can't say how much insidght I have. I'm mostly working on weird manufacturing production issues and answering the phone. :oops:
Cool! Now we'll all be tempted to ask you the questions we don't dare to ask in the official contact page. :mrgreen: :mrgreen: :mrgreen: :mrgreen: :mrgreen:

That said, keep up the good work!

User avatar
FletchINKy

07 Sep 2017, 20:22

depletedvespene wrote:
FletchINKy wrote: I am a new Unicomp employee, but I can't say how much insidght I have. I'm mostly working on weird manufacturing production issues and answering the phone. :oops:
Cool! Now we'll all be tempted to ask you the questions we don't dare to ask in the official contact page. :mrgreen: :mrgreen: :mrgreen: :mrgreen: :mrgreen:

That said, keep up the good work!
I already got some OK to offer defective or non-defective parts for makers or modders, so if you want to experiment but don't want to sacrifice a Model M frame assembly, drop me a line; tfletcher at pc keyboard dot com. Or call! Extension 206

User avatar
depletedvespene

07 Sep 2017, 20:37

FletchINKy wrote: I already got some OK to offer defective or non-defective parts for makers or modders, so if you want to experiment but don't want to sacrifice a Model M frame assembly, drop me a line; tfletcher at pc keyboard dot com. Or call! Extension 206
WOW! I don't think that :shock: actually has the eyes open wide enough. :D

All the possibilities I'm thinking of right now...

User avatar
0100010

07 Sep 2017, 20:49

Convince your co-workers / mgmt to start offering greater than 1x width custom cap top printing, and side printing using existing templates // would be a GREAT help.

User avatar
FletchINKy

07 Sep 2017, 20:51

0100010 wrote: Convince your co-workers / mgmt to start offering greater than 1x width custom cap top printing, and side printing using existing templates // would be a GREAT help.
Not sure I follow, send me a PM or email!

User avatar
wcass

07 Sep 2017, 23:34

Yep, this is a nice group of folks here. If you need anything, i would be happy to help (though i might loose power or internet service starting about Saturday). I too would say go with capacitive sensing as the cost for small run is a LOT cheaper. You can even use Model M barrel frame with model F flippers (you just need the F flippers!) If you want to go this route, i know some guys that might help you find some.

User avatar
FletchINKy

08 Sep 2017, 12:03

wcass wrote: Yep, this is a nice group of folks here. If you need anything, i would be happy to help (though i might loose power or internet service starting about Saturday). I too would say go with capacitive sensing as the cost for small run is a LOT cheaper. You can even use Model M barrel frame with model F flippers (you just need the F flippers!) If you want to go this route, i know some guys that might help you find some.
Is the capacitive sensing robust enough to work with modified layouts of circuit board? It seemed like there was a lot of tweaking to get the sensing to work in the "real world" of the actual PCB. Could I really just apply my novice skills of pcb design and come up with something that the controller will be able to drive cleanly?

xueyao

08 Sep 2017, 12:25

FletchINKy wrote:
wcass wrote: Yep, this is a nice group of folks here. If you need anything, i would be happy to help (though i might loose power or internet service starting about Saturday). I too would say go with capacitive sensing as the cost for small run is a LOT cheaper. You can even use Model M barrel frame with model F flippers (you just need the F flippers!) If you want to go this route, i know some guys that might help you find some.
Is the capacitive sensing robust enough to work with modified layouts of circuit board? It seemed like there was a lot of tweaking to get the sensing to work in the "real world" of the actual PCB. Could I really just apply my novice skills of pcb design and come up with something that the controller will be able to drive cleanly?
I think it's not exactly cheaper, but much easier in terms of work needed to put in. The capacitive switch is also revered more compared to membrane sensing. You can reference the FSSK project in terms of PCB design, slap on an xwhatsit and you'll be good to go.

On a side note, have you seen the Model M50?

User avatar
FletchINKy

08 Sep 2017, 14:29

xueyao wrote: I think it's not exactly cheaper, but much easier in terms of work needed to put in. The capacitive switch is also revered more compared to membrane sensing. You can reference the FSSK project in terms of PCB design, slap on an xwhatsit and you'll be good to go.

On a side note, have you seen the Model M50?
I seem to be unique in this wild and strange keyboard land in that I have little reverence for anything expect productivity. ... That said... I am sure putting in a lot of work to get a BS keyboard I can use comfortably and with my layers/layout, sooo... :roll:

I'll chalk it up to having never clacked a model F. Maybe the clouds will part and angels will descend. But unless they do so in a firmware that will allow me to build my present layout and easily make modifications I might just turn around and leave.

Also, not sure I'm Googling right, can't find anything on an M50... Link please?

User avatar
0100010

08 Sep 2017, 14:55

Info on the xwhatsit Model F capsense USB controller : https://geekhack.org/index.php?topic=58138.0

4704 50 on the bottom (they came in both M and F variants if memory serves):

Image

Some info on the 4704 : http://kishy.ca/?p=648

User avatar
emdude
Model M Apologist

08 Sep 2017, 18:32

FletchINKy wrote: Is the capacitive sensing robust enough to work with modified layouts of circuit board? It seemed like there was a lot of tweaking to get the sensing to work in the "real world" of the actual PCB. Could I really just apply my novice skills of pcb design and come up with something that the controller will be able to drive cleanly?
You may have to first draw the capacitive matrix in some CAD software, but it is enough to copy IBM's implementation and follow some design rules to get a working PCB.

More links:

wcass' Xtant: workshop-f7/bringing-the-ibm-pc-xt-into ... t3047.html
i$'s FSSK: workshop-f7/f-ssk-t10744.html

Links relevant to matrix design:

workshop-f7/f-ssk-t10744-150.html#p248404
workshop-f7/f-ssk-t10744-240.html#p274800

It's not difficult either, I had never used an EDA software before pursuing my own capsense project. Most of the work was in the CAD software anyway.

q11q11

16 Sep 2017, 11:17

Oh man, I recognized you by photos here too :)
It was pain to see your video about carpals, how long you need to wear this stuff?
Hope your hand gets back to 100% as soon as possible.

User avatar
Phenix
-p

16 Sep 2017, 11:33

Just out of curiosity:
1) is a split layout (lets split style) possible with Model Fs?
2) if so, any Chance you are gonna design it with the split in mind? had be pretty cool to have a split model F...

__red__

17 Sep 2017, 08:30

FletchINKy wrote: Is the capacitive sensing robust enough to work with modified layouts of circuit board? It seemed like there was a lot of tweaking to get the sensing to work in the "real world" of the actual PCB. Could I really just apply my novice skills of pcb design and come up with something that the controller will be able to drive cleanly?
So, there are two schools of thought:
0. Make all pads swing the same amount so everything is consistent. (Original IBM design, xwhatsit etc...) IBM calibrates on boot, xwhatsit has you calibrate a setting which applies to all keys.
1. "I'll fix it in prod"-PCB design. Pioneered by DMA's CapSense controller. Each and every key can be calibrated independently which means that PCB design can be a lot more sloppy^Wfree.

I'm a fan of the latter approach altering a few numbers in software when I first build a device is trivial compared to the potential PCB iterations going back and fore to .cn.

ymmv.

User avatar
FletchINKy

10 Oct 2017, 21:14

Thanks for the continued interest! I'm just getting settled after being back from vacation, so I have a bit of work to catch up on and deliverables to deliver, but this project will see completion, one way or another!
q11q11 wrote: Oh man, I recognized you by photos here too :)
It was pain to see your video about carpals, how long you need to wear this stuff?
Hope your hand gets back to 100% as soon as possible.
I'm off it now, but I'm also just back from a two week vacation, and I've modified my Unicomp 122 firmware to bring things closer to my fingers; caps to escape, ctrl/alt swap, custom extended left alt key, and some other strangeness. So I'm ok now, and trying to take it easy. I have, in this time, become quite addicted to the BS switch though. Definitely noticing typing speed increase too. Thanks for your concern! The lack of pain and a contributor ;) helping make the project properly (instead of my hacking at silver epoxy traces!) has delayed things slightly, but I certainly still intend to complete it!
Phenix wrote: Just out of curiosity:
1) is a split layout (lets split style) possible with Model Fs?
2) if so, any Chance you are gonna design it with the split in mind? had be pretty cool to have a split model F...
My understanding of the sensitivity of capacitive sensing is that splits make the sensing uneven and inconsistent. I'd really like to be wrong on that, but for the first build, a split is unlikely. I do really like splits though, and if it is possible, I will be pursuing it!
__red__ wrote: So, there are two schools of thought:
0. Make all pads swing the same amount so everything is consistent. (Original IBM design, xwhatsit etc...) IBM calibrates on boot, xwhatsit has you calibrate a setting which applies to all keys.
1. "I'll fix it in prod"-PCB design. Pioneered by DMA's CapSense controller. Each and every key can be calibrated independently which means that PCB design can be a lot more sloppy^Wfree.

I'm a fan of the latter approach altering a few numbers in software when I first build a device is trivial compared to the potential PCB iterations going back and fore to .cn.

ymmv.
I think the fix it in prod seems like the best way, as it can account for production variations, or any weird little surprises that pop up, but my understanding is that controllers can support both firmwares, so potato/tomato! :mrgreen:

User avatar
DMA

11 Oct 2017, 05:11

FletchINKy wrote: but my understanding is that controllers can support both firmwares, so potato/tomato! :mrgreen:
Image
for xwhatsit you need to have a highly customized hardware.
CS, on the other hand, can be had on CY8CKIT-059, ask __red__ :)

__red__

26 Oct 2017, 05:06

... and that CY8CKIT is $10 in quantities of one. :-)

User avatar
FletchINKy

26 Oct 2017, 16:40

Whoops, yes, they are not interchangeable, I mean that both controllers could run off the same PCB design. Is that accurate?

User avatar
wcass

26 Oct 2017, 22:23

FletchINKy wrote: Whoops, yes, they are not interchangeable, I mean that both controllers could run off the same PCB design. Is that accurate?
if you are asking if both controllers could run the same "key switch PCB", the answer is yes.

User avatar
FletchINKy

25 Jan 2018, 16:46

Latest updates!

Got Wcass's PCB printed, and laser cut the case for it. This is a test fit, but it definitely sounds great. Modern Unicomp stems work perfectly.

Sound test and some other projects
https://www.youtube.com/watch?v=XRFf-oLy6Vc

Just need to decide on a controller, hook it up, and get to work! :D
Attachments
20180111_172251.jpg
20180111_172251.jpg (2.23 MiB) Viewed 8816 times
WIN_20180125_06_45_38_Pro.jpg
WIN_20180125_06_45_38_Pro.jpg (2.27 MiB) Viewed 8816 times
WIN_20180125_06_35_33_Pro.jpg
WIN_20180125_06_35_33_Pro.jpg (2.17 MiB) Viewed 8816 times
WIN_20180125_06_45_22_Pro.jpg
WIN_20180125_06_45_22_Pro.jpg (2.4 MiB) Viewed 8816 times

Vizir

25 Jan 2018, 17:49

Awesome progress!

Sent from my Pixel 2 XL using Tapatalk

User avatar
depletedvespene

25 Jan 2018, 18:18

I just can't wait to get my hands on THREE of these. :-D

(one to use as a regular 40% and two more to mash them into a split ortholinear keyboard)

codemonkeymike

25 Jan 2018, 20:01

Add this to the "things I didn't know I wanted" list

Post Reply

Return to “Workshop”