A Compact SSK

wcass · 22 Jun 2018, 02:29

Two and a half years ago, I replied to the “post your ideal keyboard layout” thread with this:

A couple of months ago, I decided to make a CSSK and document how I did it for anyone interested in making a custom Model F. This is that.

: CSSK4.jpg (318.88 KiB) Viewed 4462 times

I decided to go with a “layered case” design because it is simple and inexpensive to prototype. It also let me experiment with the idea of a “Flat F” (or F♭ ) an idea floated a while back by my friend Lot Lizard. Without a curved back plate, the gap between rows of key caps is wider, so I would want to compensate for that.

Another thing I wanted to do was to reduce or eliminate the border around the keyboard edge. All previous Model F pad cards (the PCB with the capacitive pads) have rows that gather at one side and route to the controller, and columns that gather at the top and route to the controller. This dictates a need for a border on at least two sides. I wanted to try making my pad card from 4 layer PCB which would allow me to place rows on layer 1, columns on layer 2, use layer 3 as a shield, and run traces to the controller on the bottom – thus eliminating a need for much of a border. This also significantly reduced the material thickness between columns and rows – making the capacitive effect easier for the controller to read.

And a final thing I wanted was to fit the controller inside the keyboard. Xwhatsit is much too complicated for this and can only get so small. This was going to be my first use of DMA’s controller. Originally, I was going to put the controller directly on the pad card, but after some research, I changed my mind.

Over the next week or so, I’ll write an in-depth post on each … the case, pad card, and controller. Feel free to ask questions; I’ll try to answer in the write ups. I will be posting at GeekHack too (sorry reddit fans, I limit myself to two interest groups per subject and GH and DT were here first).

Techno Trousers · 22 Jun 2018, 04:59

Thanks again for letting me try your prototype, wcass! I'm looking forward to reading the whole write-up.

I've been really yearning for an MF SSK lately. Reading about custom capacitive buckling spring projects like this will help my patience.

Darkshado · 22 Jun 2018, 06:42

How much of a difference does the "flat" profile make? Is it weird to type on? Did you try with M2 caps?

wcass · 22 Jun 2018, 18:45

Darkshado wrote: How much of a difference does the "flat" profile make? Is it weird to type on? Did you try with M2 caps?

The first two questions are pretty subjective, so I can only answer as it applies to me.

If you have typed on a laptop recently, you have typed on a flat keyboard. I don't have an issue with that specifically. It is a bit unusual (but not bad) that the key tops are tilted back as much as they are. I hope that rsbseb's Spherical Buckling Spring caps project will offer something with less back tilt.

I did purchase an M2 specifically to try out the caps on F and original M barrels. Unfortunately, they do not fit F or M barrels properly - and have only slightly less back tilt. I think that all of these older IBM keyboards expect you to have the back legs fully extended which would put the key tops closer to level, but stepped like stairs.

I will make a graphic that explains this and the "gap issue" better.

green-squid · 23 Jun 2018, 22:44

Hey, great keyboard! I'm impressed with how good it turned out.
Also.. you said 'an idea floated a while back by my friend Lot Lizard'. Is he still OK? Does he plan to return?

wcass · 23 Jun 2018, 23:31

green-squid wrote: Hey, great keyboard! I'm impressed with how good it turned out.
Also.. you said 'an idea floated a while back by my friend Lot Lizard'. Is he still OK? Does he plan to return? :)

I have not heard from LL in quite a while. I wish him well and hope he comes back soon.

wcass · 28 Jun 2018, 02:39

I call my keyboard layout the Compact Space Saving Keyboard or CSSK because it uses the same key cap set as the original SSK (plus a 1u caps between both Ctrl and Alt set) and does it in a package just 12” x 7” (304 mm x 178 mm). The caps are placed logically near where you expect them to be (except for Print Screen). For me, it combines the best features of 60% and tenkeyless. I have been using a chopped off M122 derivative of it for quit some time and am very happy with it.

: hack14.JPG (68.51 KiB) Viewed 4257 times

The first thing I did to make my keyboard was to prototype the case. I had a feeling this would be the most troublesome and expensive part of the project, so I wanted to make sure that it was right before I continued on with the rest of the process. Acrylic sheets come in standard thicknesses – multiples of 1.5 mm. I decided to use a 3 mm back plate, 4.5 mm spacer, 1.5 mm barrel frame, and 9 mm trim – for a total thickness of 18 mm. During the design process, I knew I needed to adjust the barrel for the caps to look right.

A Model F buckling spring barrel is 19.0 mm wide, but about 20.6 mm deep while the key cap is 18.6 mm wide and 19.3 mm deep. This is not a problem when the back plate has a curve, but when you flatten it out, the keycap rows get further apart. You can see this in the graphic below. My solution to this was to file off the front side of some of the barrels so that they were only 20.0 mm deep. This does not seem to alter the functionality.

: the gap.png (16.58 KiB) Viewed 4257 times

I sent my design off to built-to-spec.com for a RFQ in early September of 2017 … just before hurricane Irma hit south Florida. I didn’t have power and internet for a few weeks, so wasn’t able to give the go-ahead until December. I had delivery a few days later but the first test fit showed that alignment between trim and caps was off by 1.0 mm. I updated my design, ordered a replacement 1.5 mm barrel frame layer, and had everything assembled by New Years.

So, what did I learn? Acrylic’s glossiness is a fingerprint magnet, it flexes and cracks, but it is good and comparatively cheap for checking how things will fit together. I plan to make a final case using a single block of aluminum.

wcass · 31 Jul 2018, 06:38

The last couple of weeks, I have been working on the 3D model for the case and messaging with CNC vendors. I checked out eMachineShop, Xometry, and Fictiv. All were well outside my budget. The Chinese CNC shops were in budget, but communication, quality, and import tariffs could be a problem. Then i stumbled upon 100kgarages. I posted an early version of my design and a target price of $250 which was more than SuNPe quoted, but less than the US shops. I started getting quotes from maker members by the next day. A couple of the of the folks i chatted with pointed out some pain points. The original design had the space bar stabilizers fully integrated into the body. I knew that kind of undercut would be difficult, but thought it would be possible if i made a hole in the barrel frame below the overhang - so technically not an undercut. I was informed that chatter would likely rip the overhang right up.

I just placed an order for the aluminum case with Dalas at Rival Machining. This is what it should looks like.

: CSSK6-body.step (1).png (187.95 KiB) Viewed 4175 times

Since undercuts are so difficult to do, i built only the uprights for the space bar stabilizers into the body. The top of the stabilizer (1 mm thick flat) will be attached with an M3 screw. The 3 mm (1/8") thick base plate (not shown) will be attached with 20 Torx drive flat head M3 screws. Despite the lack of alignment pins, the barrels will be unable to move out of alignment because i have them completely surrounded by a tight fitting ledge (or another barrel). Another ledge (and 4 screws) keeps the pad card PCB from moving.

: CSSK6-body.step (2).png (111.68 KiB) Viewed 4175 times

Techno Trousers · 31 Jul 2018, 07:14

I think it's a good solution to attach the tops of the space bar stabilizers with screws. It does kind of beg the question though: is it worth exploring building up the frame out of screwed together layers? Would that be appreciably less expensive than machining the whole thing together as one piece?

DMA · 31 Jul 2018, 08:14

Ow, that aluminium case is really cool! Can't wait to see it in metal!
I would not worry about aligning pins.

wcass · 09 Aug 2018, 04:14

Progress on the aluminum case.

The stabilizer tops were cut a couple of days ago on a brand new Tormach 1100.

wcass · 20 Aug 2018, 04:40

The pad card is the PCB that makes up the switch matrix. Each switch has two parts; a signal pad and a sense pad. The pads are arranged with all of the signal pads on columns and all the sense pads on rows (or the other way around). The traditional design rules for the pad card is that all rows and columns cross exactly one time. Notice in the XTant pad card example below how the bottom row trace extends so that it can cross the two rightmost columns exactly once. Red color is the top layer, blue is the bottom.

: XTantPCB.jpg (312.33 KiB) Viewed 4034 times

Also notice the wide right and top margins needed to route the columns and rows toward the controller. I wanted to see if I could make those top and side margins disappear. Compare that to the pad card of the CSSK. Red is again the top layer, blue is the second layer, green is the 4th layer. The 3rd layer is a ground flood (not shown).

: CSSK-PCB.jpg (502.65 KiB) Viewed 4034 times

And now some CNC action ...

wcass · 06 Oct 2018, 05:53

I finished programming the controller today. All done and it works great. Thanks DMA for the CommonSense.

: CSSK.jpg (586.3 KiB) Viewed 3916 times

By the way, the LEDs themselves are white. I use a speedlite gel to change the color. A pack of 20 different color gels is $2 on eBay.

: LEDs.jpg (760.04 KiB) Viewed 3916 times

Techno Trousers · 06 Oct 2018, 06:22

Congrats on finishing it up! It really looks nice, and is certainly a unique Model F. Are you completely satisfied with how it turned out?

wcass · 06 Oct 2018, 16:16

Techno Trousers wrote: Congrats on finishing it up! It really looks nice, and is certainly a unique Model F. Are you completely satisfied with how it turned out?

Oh yeah.

The top corners are very pleasant quarter hemispheres with 7 mm radius fillet around the top and side edges. The 3 mm bottom plate makes an invisible seam with the main body. Perhaps i over did it with 20 bolts holding this in place. The main body is ridiculously rigid - absolutely zero flex. LEDs are not too bright or too dim and i love that i can change the color easily without changing the chips. The space bar retaining clips work perfectly and do not interfere with C, V, comma or period as they did on the acrylic prototype.

I think this proves that we can do pretty much any layout with capacitive buckling springs fairly inexpensively even at relatively low volumes. The 4-layer pad card was $100 for 5; the controller also $100 for 5. The expensive part is the CNC case, but it is definitely worth it. With that too - cost goes down as quantity goes up. You do need a supply of barrels, flippers, and caps. It would be great if we could find these easily.