Baby steps, but I think this should work. Should it?

[VoteForDavid]

Here is a plate, probably to be 3D printed in ABS, 3.5mm thick.  It has small reliefs printed where the key switches and stabilizers clip in.  You are looking at the top when it's being printed, but with keys mounted this would be the bottom.

The relief cuts are *only* where a piece clips in.  The rest is full thickness all the way around the keys.  I think this should be plenty strong enough, and still able to withstand being bent along its width for tenting in the middle with the outer keys flat.

What says GH?



This is practice on my way to making a better mousetrap.  Before getting silly with angles I wanted some solid familiarization time in Sketchup modeling a regular switch mounting plate.  This is the far-right end of a keyboard BTW.

[CPTBadAss]

Where is the bend? I'm not sure I'm following your plan.

[VoteForDavid]

Ugh.  First week of kids back in school dulling my brains sorry for not being clear.

Think along the same lines as my DK1 board (or the Microsoft Natural 4000): two half-keyboard switch mounting plates, bent up in the middle maybe an inch or so.  Tented in the middle, flat along the outsides.

ETA: the bend is currently made with your imagination this is a drawing of a flat plate

[njbair]

This looks cool. I've bent ABS for non-keyboard things before. I just used a heat gun and bent the sheet around a form. Held pretty well.

One recommendation, invest some time learning FreeCAD for 3D modeling instead of Sketchup. I think it's much easier to use once you get the concepts.

[VoteForDavid]

I clear my throat quietly
http://www.autodesk.com/education/free-software/fusion-360

I don't like cloud based computing -at.all- but this is interesting.  I use sketchup because it's what I've used for ages now.  CAD software was in horrible command-line-programming-with-math hell when I looked around the last time, and sketchup had a GUI.  I think that's changed for the much-better recently, and I've considered looking into CAD software again.  Two of the thirty tabs open on my workstation browser window are there for this purpose.  But I know sketchup and it works quickly enough to get the a point like this across.  Time is short these days so for the time being, you guys have to suffer looking at sketchup drawings.

I don't know why this small of a bend on this thin a sheet of ABS wouldn't work, but I don't know everything, so I thought I'd ask.  Bending the sheet over a form, with heat, seems like a good idea.  I was thinking of just forcing it to adhere to a curved mounting surface, but heat forming seems much better.

[Oobly]

If you make it 5mm thick you don't need the relief areas since the plate will be held solidly between the switches and PCB. 5mm is harder to bend, but if you bend the outside edges down and don't mind about having the PCB exposed underneath, you don't need anything more to complete the case.

[Melvang]

I would agree, get away from sketup as quick as you can.  There really isn't any good way to export the files to anything that a fab shop will use.  At least, I haven't heard of any shops using those files.  Those reliefs should work as well.

Are you planning on making a PCB or hand wire?  If you are using a PCB, you can skip the reliefs, as they switches will be held in by the PCB anyway.  If you are going to hand wire, the reliefs will be necessary, unless you plan on gluing the switches into the plate.

Also, good luck with 3d printing this size of a piece with enough accuracy to work as a functional plate.  Before you get pieces made, get quotes from places for laser cut acrylic.

As far as software, I use SolidWorks student edition.  Normally it usually costs $150, but I get the veterans discount for $20.  Plus it actually saves time to draw it in a proper CAD environment first, because you still have to draw it there for getting quotes as well.

[Oobly]

Also, +1 for SolidWorks if you can afford it. It's very intuitive (although there is still a bit of a learning curve due to the inherent complexities 3D design and all the features supported) and can import / export practically any format. Also includes simulation / analysis in the same package. It's super expensive if you don't qualify for the student package, though.

I'd also go for laser cut acrylic rather than 3D printed anything. It would be slow and use a lot of material doing a 3.5mm or thicker 3D print and the material contraction during cooling can either warp the part or introduce inacuracies in the hole dimensions. Should also be a lot cheaper. Acryilic is easy to bend with some applied heat and accepts form molding well, too.

If you have the skills I recommend designing and using a PCB, too. Makes the switches mount to more contact surfaces and provides a really solid feeling, although for prototyping it's convenient to handwire to a Teensy, but then you need the relief holes for the switches to clip in. This can be achieved by using a 1.5mm acrylic plate without the reliefs and 2mm or more 2nd plate with the relief areas. You can dissolve some acrylic shavings in acetone as a glue to mate the plates into one.

[VoteForDavid]

I've seen reports from people whose shops work entirely with sketchup, having ditched $$$$/seat licensing years ago and never looked back.

A PCB seems like it might only hold switches tight if it's shaped properly and secured properly.  Otherwise it will prevent them coming out altogether, but they could still be loose, especially if this turns into a project with compound curves and strips of PCB instead of a flat PCB.  http://cdn-ak.f.st-hatena.com/images/fotolife/s/squidapache/20130921/20130921132905.jpg

One of the projects where I work is "one and a half calendar weeks" away from being a 3D printer with more than enough size capacity to do all three switch mounting plates in one job.  It's so bleeping hot in the shop right now we'd have to TRY to get the parts to warp! If the accuracy proves too bad, it wouldn't be the first time I finished switch holes by hand (  The lead dude on that project is keen to try his machine on my parts.  We'll see, I guess!

I've secured a license for AutoCAD 2016 and it appears to be  . . . complicated.  I've also got hold of Fusion 360 but haven't looked into it just yet.  I thought Photoshop had lots of options, but that was before AutoCAD showed up on my monitors.

The layered acrylic is an idea with merit.  I'm (needlessly?) afraid of intersheet voids looking terrible, and I worry also (needlessly?) about flexibility and strength with the bending I might want.  I've got to have cuts or thicknesses somewhere for clipping-in switches because this is going to be hand-wired at least initially.  I've switched out caps on my DK1 board and a few of them were loose enough to pop out.  Some n00b must have finished those holes by hand or something. 

A PCB is nice for a finished project.  I prefer hand-wiring for prototypes.  The impression I have is, for those of us without a CNC router and a solid PCB design workflow already in place, it would take as long to design and make a PCB as to hand wire, with less flexibility for the inevitable revisions. 

ETA: I just realized we've all lost the plot.  I started this thread to have someone hold my hand and tell me just these small reliefs would be all that's required to mount switches and stabilizers.  I'm fairly certain it is, but I still have to go measure some stabilizers to be 100% sure.

[Oobly]

If you can get the accuracy and you already have a workflow in place (Sketchup to 3D print), then go for it. AFAIK, the recesses should do the trick to allow the switches to clip in, but even then be prepared for switches to pull out when pulling caps. I used a solid steel 1.5mm plate in my build and ALSO secured the switches with a glue gun and they still pop out when pulling caps.

Instead of doing bends you may consider doing it in separate parts with connections with the meeting surfaces angled. Increases repeatability accuracy since it's hard to repeatedly bend accurately.

The advantages of a PCB are that it holds all the switches to each other, so they're connected to each other at the plate as well as the PCB. Which means an individual switch can't pull out the plate since the combined force of all the clips of the surrounding switches comes into play. If you use a 5mm plate, the whole assembly is one solid chunk: switches, plate and PCB all held together in a tensioned unit.

I agree that for prototyping, a PCB is overkill unless you're not planning to change the main area layout much and / or are planning to productise it.

And complex curves require hand wiring unless you're using shallow curves. Even then, finding a shop that will make thin / flexible PCB's is not that easy. This is one area that Kinesis compromised on compared to Maltron and I think the curves are too shallow on their boards in order to accomadate using a PCB.

Bending thick acrylic can be a PITA and sometimes requires relief holes that reduce the strength at the bend, but if you're using heat molding it can work fine. But then you're risking the precision-cut holes distorting. These are all just factors to take into consideration at the design stage.

AutoCAD is not as intuitive as SolidWorks and not as powerful or flexible either. But go with what you've got.

[VoteForDavid]

I suppose if I hadn't seen stems pulled right out of switches from too-tight caps being pulled I'd object, but I have seen it so . . .

I hadn't thought about the keys holding each other through a PCB.  If you pull one switch, it pulls the PCB, which resists flexion - and as much as it does NOT resist, the switches nearby are also being pulled against their plate-mounting clips. 

If I'm 3D printing and doing angles in more than one plane, I might as well go for printed-in angles instead of separate parts.  I anticipate using bent plates with curved supports underneath, as I'm pretty happy with the angles i have already.  The small amount of curvature I'd like, a standard set of key caps with different heights is close enough.

AutoCAD is something I've been wanting to look into for a long time, and this is as good an excuse as any.  Being able to manipulate items in an array by changing just the original shape (which is new in this release) should help a whole lot for something with as many repeated parts as a keyboard.

[vvp]

I used a solid steel 1.5mm plate in my build and ALSO secured the switches with a glue gun and they still pop out when pulling caps.

This is my experience too. At the end I just glued the switches in with a regular glue. After that, they hold well and one can still pry the switch out if it breaks (which is very improbable anyway).