3D Printed/Milled Switch Plates with Inbuilt Spill Resistance: Check My Idea?

[VoteForDavid]

It is theoretically possible I could be considering making custom keyboards for sale.  After my recent experience, I don't really want to make switch mounting plates by hand so I've started playing with 3D models. 

One of the goals I missed with my DK1 board was spill resistance.  This is an idea for a *small* amount of spill resistance to be designed into a switch mounting plate.  The top side of the plate has all the holes elevated by (e.g.) 1/2mm, creating 1/2mm-deep grooves around all the switch mounting locations.  In larger gaps between switches, the grooves could be made deeper.  These would lead to drain holes.



The plate is thick enough to be strong, maybe 3-4mm.  This is too thick to snap a switch in place.  So the switch holes are all recessed UP into the bottom of the plate.



It's past my bedtime and my mind may be playing tricks.  Does this seem to you like it could work?

[vvp]

Only if you 3d print with SLS or SLA 3d printer.
FFF/FDM 3d printers will have the parts of the model which are printed in the air (the bridges) less precise. So you want to avoid it on the parts of the model where precision and nice smooth surface is important.
Check KHAANNN's thread here:
https://geekhack.org/index.php?topic=72183.0

[Mad_Maxx]

I have no idea on the effectiveness of the design, but I do know that the only way to plausibly make something like that is 3d printing. It would take for freaking ever to mill that, with not a lot of added value. Flat plates are great because they can be so quickly cut in 2D with a laser or waterjet, but to make this out of steel would only be worth it for a one-off for yourself, or other buyers who have little regard for cost.

[VoteForDavid]

vvp thanks for that.  It seems like I should be able to FFF print supports for the overhanging parts as necessaryand trim them out, which may be worth the effort depending on relative costs.  That's waaaaay down the road from here and there's plenty of time to iron it out.

Value is what the market is willing to bear!  With a CNC mill, the ideal is to clamp the piece in place and press "Go" and come back when it's done.  Milling this by hand in metal would be almost as crazy as what I just did on the DK1 board (which is to say, not entirely out of the realm of possibility).  If I went for metal I'd rather make it of aluminum for no particular reason, but I like working with it more than steel. But yes I was thinking of 3D printing as a primary means of production. 

3D printing is a whole new ball of wax for me.  The place I work is perfect for someone looking to DIY a 3D printer, which I just might do.   And I'd want to go big - like, 25" X/Y and maybe half that Z capacity. But it would cost, even at employee pricing.  Have to put in some overtime at the lemonade stand to afford a project like that!  I think it would be funny to print two full sized keyboard cases - press "go" and come back in a week!

[Mad_Maxx]

With a CNC mill, the ideal is to clamp the piece in place and press "Go" and come back when it's done.

Well yeah thats fine if you have a CNC mill with no time constraints, and know how to use it. Its just that its just going to take a while to CNC that, as opposed to the other plate making methods. And this would require flipping the workpiece, so thats adding a lot of time onto the process.

I'm just talking if your making more than one. Anything is possible, as long as your willing to throw time and money at it.

[kurplop]

In a large enough scale, it would be commercially feasible to stamp the plate in a punch press. The design would lend itself to the process for both the holes and the drawn 'curbs'. It could be made with 1/16" stock.

Of course the dies would be cost prohibitive for a small run.

[Leslieann]

Good luck using FFF or FDM.
At this size that falls into large format, and usually that means large nozzles. The printers capable of a print this big can probably be counted on one hand. Then there is the actual object. Something most people don't understand with FDM is that like other forms of manufacturing it has it's strengths and minuses and you have to design for it to get the most out of it. On subtractive manufacturing, which is how most people think, you want a hole here, punch a hole. On FDM, you have to think more like a medieval bricklayer, planning ahead for every hole and overhang. Yes, you could print this and then break everything out, I recently tried this on just such a printer, it sucks, and the result was sub-par.

Forging/stamping is simply not practical unless you have very deep pockets and/or need a lot of them. Or happen to own a 20 ton press and a mill to match. How many do you expect to sell, 20? 100? Unless you plan on thousands, it won't be practical. Think of the size of this and how much pressure that will require.

There are a few ways to do this though, some more practical than others. The cheapest method would be a combination of milling and waterjet, followed by milling alone.  You may be thinking but plastic is cheaper, not always. The machines that can do this needs specially prepped plastic and it's not cheap, neither is the printer that uses it, so companies want their return on investment. Most large printers are also under expensive service contracts help together by patents and a massive team of lawyers. At the size of this, to SLS print, odds are this will be four digits per plate, and with only a small bulk discount. I would imagine you could waterjet it, and then mill it for at least a few hundred less. In bulk, probably low 3 digits.


However, there is a much simpler method.
Why does it need to be part of the plate? You could cast or print a tray that sits on top of the plate. You have to remember, the only place fluid gets in, is between the keycaps. A thin plate with lips going up along side the switches, and tucks under the keycap would more than suffice. Again, printing this for anything over a 60%or 75% though, is going to be a rare or expensive printer, especially at the tolerances, angles and thicknesses involved. Also, it will not be fast, a FAST FDM printer, if it could do the size, would still be looking at a dozen hours, machine hours cost more than minimum wage.


None of these solve everything and I don't think the op has thought this through.
A small spill has almost zero chance of getting into the switch housing anyway, think about how it's constructed. It cannot get in through the top without being completely flooded and rises up into what would become an air bubble inside the key cap. The only other ways in, is through the side where it has to flow horizontal across an slope or by complete flooding from below, which at that point you have other problems. Basically, they are already fluid resistant to a pretty good degree, but, I understand, it could be better.

So let's say you do seal the switch, what then? What happens after it runs down the plate? Or under the edges of the switch onto the pcb? Even a tray above the plate cannot stop it from having to go somewhere. This is what usually kills a keyboard. You can always swap out a switch, boards, not so much.



I think this is being attacked from the wrong direction.
Why not just run a very thin silicone, wax or grease bead around the inside of each switch top before assembly, and then seal the switch to the plate in a similar fashion. Then you just need drain holes at the bottom edge of the case. This is used in all sorts of other applications, no reason it cannot work here too.

[vvp]

I second Leslieann. If resistance to a spill is the main reason then think rather how to water seal the switch and the PCB with some kind of water resistant enamel or paint ... and use simple shape for the plate which you can laser cut.

There is a reason big RepRap FFF/FDM printers are not done. With bigger bed size the problem of warping is bigger too. So if the part printed is not very thin then you may not be able to use big build area of the printer. Well, at least not with the standard ABS. PLA may work. Or not. I do not have much experience with PLA. Although if the only thing you want to print is a rather thin plate then you may be able to do it even with ABS.

Another problem is printing on support. You cannot really do this for things you want to have precise. From this point of view bridges do not differ much from printing on support. The surface (which is touching the support) will be very rough. Think something like a 1 mm error in the surface precision. So you do not want anything visible or anything precise to be printed on support or using bridges.

Here is a picture of a keywell printed on support. In this case the imperfections do not matter since this is a bottom side of the plate which is hidden inside the case but it gives you an idea how bad it is. The print time was 3 h 37 m on medium speed Rostock. There are already some switches in the plate but look only at the plate.



The only way how you can get somewhat acceptable surface on support is printing the support using water soluble material (or something like that) with full infill on the top of the support object. And no gap between the support object and the actual object you want to print. That means dual extruder printer. I do not have ane experience with dual extruders.

[VoteForDavid]

That was an excellent post Leslieann.  I think I have not explained myself well.  Have a wall of text as a reward!

None of these solve everything and I don't think the op has thought this through . . . . I think this is being attacked from the wrong direction.

Probably, which is why I am here presenting my idea instead of my finished product

The MX switch housing has a relatively accessible set of holes for liquid ingress by the diode location, as well as around all the fingers holding the halves of the switch together.  If liquid gets around down around the edges of the switch and under the mounting plate, there are even-bigger slots that would probably wick liquid in by capillary action.  I had thought to elevate the switch, but it may be a better idea to put a ridge around the outside of the switch as you suggest.

I'm not looking for IP65 here, more like oh****mycoffee grade resistance.  A few tens of milliliters at most.  I have thought of something that goes up under key caps, but I'll need to see what the clearances look like under there.  This is not exactly to scale, but it seems like the ideal spill resisting solution for a board using non-siliconed-shut switches:



It would need to fit under a cap, but have a big enough opening to allow the switch to be mounted in the plate.   That's cutting things pretty close.  A switch tester-scale thing where I can see all sides of a switch AND up under the cap would be handy, and I may have to cut a cap or three in half, to see how much space is available.

I hadn't thought about adding a second part as the liquid guide, but that could be brilliant.  A nicely textured flat plate printed normally, with a little layer of empty squares to make ridges around the switches, glued on out of sight under the caps.  That could work. Maybe. 

I'm not afraid of a bunch of drain holes on the frame between switches, it just has to be designed properly.  There would be holes that join to empty cylinders (pipes) going right down through holes in the PCB and out the bottom of the case.  Spills go down between the caps, into the drain holes, through the drain pipes and out the bottom of the case.  Holes in JUST the top would be dumb and you would end up with a bottom keyboard case full of Sprite.  It would have to be drains through the plate, and outlets in the bottom.

If (big If) I were to start making keyboards, it would likely at first be my preferred flavor, which I've seen moderate demand from programmers who can't find what I also can't:  Full size, split/spread, tented middle, with something like the Microsoft Natural 4000 layout but without the rubber switches.  The switch plates in particular (because there are three or more) would be like 8"x6" on the largest, which is quite do-able.  The case parts could be more easily printed in one piece as they are both flat and thick.  The cases could also be milled out of aluminum WAY easier/cheaper than printed. 

Regarding a big DIY printer: When I say I'm in an ideal location, I'm not kidding.  I could probably swing out a thousand dollars cash and walk away with all the linear rails, ballscrews, motors, and T-slot to have the size of printer I'm talking about, with the precision I would want.  We take apart industrial machinery as part of what we do, and the cost of parts to employees is something like a few percent of retail.  It's crazy, but in a good way.  And in the big rock candy mountain where I have built this thing and it prints great, I can defray the costs by selling print time in between my own hobby/production use.  Now THAT would be a kickass side job.

vvp, I would be happy to see the top side of that part if you care to post a picture here. The undersides can be much uglier than tops IMO because the user should never see them.  If I were building a bigass printer, it would also have a bigass temperature-controlled, heated, vacuum-capable bed - maybe with a temperature-controlled closed airspace over the whole thing.  And I like me some thick case parts.  Thick = heavy = feels like quality IMO.  From this side of experience* I tell myself I could do it my way and make it work.

*This side of course is ignorance.  The other side of experience is where I look back at my naivete and shake my head

[squishygnomes]

I have a feeling you guys might be interested in the Bloody B740, it uses a spill guard thing pretty much exactly like VoteForDavid modeled.

While it's using one of those fancy new infrared switches, they are the same shape as MX so I can't imagine it wouldn't work with them or at least be modified to work with MX switches

Here are some crappy pics of it ripped from the promotional page and a video where they pour crap on it.

[Leslieann]

It would need to fit under a cap, but have a big enough opening to allow the switch to be mounted in the plate.   That's cutting things pretty close.  A switch tester-scale thing where I can see all sides of a switch AND up under the cap would be handy, and I may have to cut a cap or three in half, to see how much space is available.


Regarding a big DIY printer: When I say I'm in an ideal location, I'm not kidding.  I could probably swing out a thousand dollars cash and walk away with all the linear rails, ballscrews, motors, and T-slot to have the size of printer I'm talking about, with the precision I would want.  We take apart industrial machinery as part of what we do, and the cost of parts to employees is something like a few percent of retail.  It's crazy, but in a good way.  And in the big rock candy mountain where I have built this thing and it prints great, I can defray the costs by selling print time in between my own hobby/production use.  Now THAT would be a kickass side job.

Like you said, it's mostly capillary other than a few bits.  Seal the LED with silicone, seal the leg holes the same way. Doesn't change the rest of what I said.


I have caps and switches to play with, with the key depressed on black switches:
SP caps leave you about 2mm per side, and about 2.5mm across the bottom.
Zinc caps (original ones) leave you 1.3mm up the sides, and 1mm across the bottom
Filco and WASD stock caps leave you 1.2mm up the sides and about .75mm across the bottom.
Vortex PBT, forget it. It actually has to step to get clear and only clears at most by .2mm, if that. It's seriously close.

To fit all of those, you need .5mm across the plate, and .1 up the walls. 100 micron walls are darn thin.


As for the printer, you have it all backwards.
It's not taking time off from the plate to mess with the printer, you will be working on the plate as an escape from working on such a printer. The printer you want and need is a MUCH larger project than the part you are trying to make here. The printer is not a means to an end for the part, the printer is a massive time and money investment all in itself that will draw you away from the plate. This is not a 1 month break from working on your plate design, this is all your free time for a full year or more by the time you are happy with it and capable of printing something this large even remotely reliably. That $1000, that's a drop in the bucket.

Not that it matters now that we have those measurements. FDM will simply not make thin enough walls, you need a resin printer. I have never seen an open source one remotely large enough for this (they are still extremely cutting edge partly due to patents), in fact, there really isn't many open source ones even. Don't take that as a challenge, when I say no one remotely close, I mean, most resin printers are in the 6inch range and are extremely slow compared to FDM. I know 3D Systems makes one capable and fast enough, but you're probably talking a couple hundred grand. It's not just a printer, but a support system as well.

By the way,
There are old Alps mechanicals that had drains in them, I think Dell made it (even had a pointerstick). My brother loved his, he may even still have it.

[vvp]

vvp, I would be happy to see the top side of that part if you care to post a picture here. The undersides can be much uglier than tops IMO because the user should never see them.  If I were building a bigass printer, it would also have a bigass temperature-controlled, heated, vacuum-capable bed - maybe with a temperature-controlled closed airspace over the whole thing.  And I like me some thick case parts.  Thick = heavy = feels like quality IMO.  From this side of experience* I tell myself I could do it my way and make it work.

With a heat chamber you possibly can make it work even for big parts. It will be a lot of tinkering but probably doable. I do not want to discourage you from building an FFF 3dPrinter. But as Leslieann indicated it is a major undertaking. A custom keyboard project is tiny compared to a custom 3dPrinter.  But if you enjoy tinkering with stuff and tweaking it then you will enjoy building your 3dPritner.

You can find a lot of pictures of the keyboard plate (both top and bottom) in this thread:
http://deskthority.net/workshop-f7/katy-keyboard-or-k80cs-key80-contoured-split-t8524.html
Everything on the pictures there was printed with  Ø0.5 mm nozzle, 0.2 mm layer height and at speeds in the range of 80 - 120 mm/s. Accelerations about 5000 - 76000 mm/s². You can get better quality by printing slower and with smaller layer height and maybe even smaller nozzle. But then the print will take much longer.

If you want higher resolution pictures of some specific parts then ask for it. I'll take the pictures for you.

[VoteForDavid]

Slow down people let's not start writing contracts yet! ) I know a big DIY printer would be a huge PITA undertaking and I expect it would take a long time.  I've also got a bunch of things like roof, siding, fence etc. standing in line first so don't look for that particular design thread anytime soon.  Big parts like full sized cases don't HAVE to be printed, just that would be super cool IMO.

squishygnome that's along the same lines as I was thinking.  It seems like something with that much coverage would have to be a part that was installed OVER the switch plate and it would have to be a thin part indeed according to the measurements supplied by leslieann - or just use switches that would clear the thickness of the splash shields.  plastic thin enough to clear all the different kinds of key caps would be scary-thin and I'd be afraid to use it even if I could.

The cap clearance measurements take me right round again to the idea of an elevated mounting surface for the switches, as in the OP of this thread.  BUT I guess there is no reason the switch plate would have to be one piece.  Make the bottom with the drain pipes and reinforcements, and a flat top.  Then make the top with the switch mounting bosses sticking out, and glue/screw the two parts together.  It could work but I don't like it much.  It seems inelegant in a way that makes me feel there must be a better way.  The best way, of course, is injection molded plastic or some additive metal printed process but let me tell you who doesn't have money for THAT in his budget.

vvp the bottoms are not pretty but the top surface looks accpetable for my idea of a switch plate.  The idea currently knocking around in my head is three plates screwed or sandwiched into a top and bottom case.  These would be mostly covered by the keys and out of sight.  The model in the OP is of the tenkey and arrows etc. plate.  That's the simple one.  There are two other plates for the main left and right hand sets of keys still to be finalized.  I would LIKE to be able to use a standard set of key caps except for the space bar(s?) (on which I am undecided). 

I am not arguing that, for near-absolute spill resistance, a bit of RTV all the way around is actually the way to go.  It could be used to perfectly seal a switch right up to the point where the stem goes in the top and would both dampen noises a bit and help to hold switches securely in place.  It would also be heck to open for spring and stem swaps.  I would really like to not limit the end user (me, initially! ) to a single type of switch when the MX series is so easy to modify.  Hence the notion to raise the mounting surfaces.

I think it is an idea with merit.  I'm not sure how much, but you all are definitely raising good points for pondering.

[Leslieann]

If you use soft silicone to fill the holes, and then a thick grease to seal the switch halves, it will not change switch maintenance at all other than possibly a bit messy.
If you really want to get tricky, you could get laser or water jetted silicone switch "stickers". Done right, they alone could almost entirely seal the switch without the need for silicone, wax or grease.

From a manufacturing perspective, if you can make the plate from one piece, it's going to be far cheaper in the long run than doing 3 bolted together. You may be thinking screws are cheap, and you're right, but it's three sets of molds, and 3 runs per plate. The initial cost for those three injection molds will probably exceed the cost for the one big one (which would be considerable) and you would need a pretty good size injection machine to handle the size. However, in terms of actually producing them you would be talking 20cents in plastic and you could do hundreds per hour with just one machine. Ignoring cost of the mold, $2 each to manufacture if you rent the machine (which you can do)?



Let's attack this a different way.
Print negative molds, one of the top of the plate, one of the bottom, pull each half of the positive molds from those with silicone, and then finally cast in high strength resin. The silicone will need some internal framework to keep it rigid and aligned (toss some wood dowels in the mold), and will need to be replaced every so often. You do still need a VERY large printer with decent resolution, but because of how it all works as mold halves, there's no overhangs, so a big FDM printer could actually pull it off. With this, it would be the switch plate. It would be much thicker than a steel plate, probably 10mm in total height (see how VVP's plate, hugs the entire switch) which is actually good since it would add rigidity. Then you just design the case around that plate. You need a custom case anyhow.

Your cost per plate would be relatively cheap, probably $5-$10 each, however, manufacturing time would be around 12 hours per casting session (however many you can cram into the vessel or vessels), but you don't need to be there once you put it in. I thought about doing a case and plate this way myself, without the channels, unfortunately our vessel is too small for even a 60%.

[VoteForDavid]

That's interesting.  One could achieve a good quality surface finish with casting, it's all in the prep. work.  The manufacturing might be messy but the finished part could be quite nice indeed.  Hmm.  One more thing to think about. 

[VoteForDavid]

Great minds think alike, and fools seldom differ.  Sure enough, seems like these days if you've thought of an idea someone probably already beat you to it and there's a commercial offering out there.

I give you: the AZiO Levetron Mech4, as spill-resistant as a Cherry MX keyboard can get, without a sheet of plastic or a spooge of silicone over the switch:



I still think this idea has merit, but not enough to break my head over the geometry.  I think if any small amounts of liquid got between the keys of my hypothetical plate, surface tension would keep the spill well-contained.  If this much spill resistance failed, you would probably have been looking at a disassembly and cleaning anyway, because you've got coke all up in your key caps.

But I think it's not worth the loss of easy-to-print-pretty texture on the top plate, by printing the top as the bottom on the printer bed.  And if you print flat plates upside-down, the overhangs on the bottom of the key holes are just not-as-tall features on top of the printed item vs. 17mm x 1mm flying hopeitprintsright features.  Relatively no-sweat.  So after banging my head on the geometry in sketchup for many hours, I'm going to call this idea tabled for now.  Maybe later it gets revived.  maybe.