The Living 3D Printing Thread

[vvp]

system, which has positive tolerance towards the left on the left side and toward the right on the right side

This is the part which I do not see how it could apply to corexy.
It is not like the two corexy steppers work like the opposite sides of an anti-backslash nut. The steppers do not somehow apply a pretension on the belt forcing the belt to be always aligned to one side of the pulley teeth.

The two steppers are independent. Really (from the point of view of carriage movement and belt alignment to pulley errors) corexy looks only like a mechanism which transforms the coordinate system. In corexy it is rotated by 45° compared to a traditional cartesian. If you would not mind that the build area is diamond shaped (and not square shaped) then you can drive corexy just like a traditional cartesian. In corexy, driving only one stepper just moves carriage along a diagonal and changing direction of this one stepper (the second one still standing) will experience backslash just like in a traditional cartesian. The only difference is that the backslash will be along diagonal.

If corexy would somehow eliminate pulley to belt alignment error (the backslash) then you probably can also claim that traditional cartesian can eliminate this error too when it moves only diagonally (drives both steppers at the same time).

[mkawa]

system, which has positive tolerance towards the left on the left side and toward the right on the right side

This is the part which I do not see how it could apply to corexy.
It is not like the two corexy steppers work like the opposite sides of an anti-backslash nut. The steppers do not somehow apply a pretension on the belt forcing the belt to be always aligned to one side of the pulley teeth.

The two steppers are independent. Really (from the point of view of carriage movement and belt alignment to pulley errors) corexy looks only like a mechanism which transforms the coordinate system. In corexy it is rotated by 45° compared to a traditional cartesian. If you would not mind that the build area is diamond shaped (and not square shaped) then you can drive corexy just like a traditional cartesian. In corexy, driving only one stepper just moves carriage along a diagonal and changing direction of this one stepper (the second one still standing) will experience backslash just like in a traditional cartesian. The only difference is that the backslash will be along diagonal.

If corexy would somehow eliminate pulley to belt alignment error (the backslash) then you probably can also claim that traditional cartesian can eliminate this error too when it moves only diagonally (drives both steppers at the same time).

let's forget about bias and polarity of forces on the carriage here for a second. assuming iid error on the A and B steppers and respective belts, dY = c(dA - dB). error cancellation.

you may be right about the dX motion. i may have been over/underthinking myself on that one.

[gcb]

If anyone want to have a go on my just designed arcade buttons that use cherry MX (how come nobody thought of that before?!) here you go:

print one of these http://www.thingiverse.com/thing:421598

and one of these http://www.thingiverse.com/thing:417509

just pay attention to the 2 paragraph instructions on each "thing" and you should be fine.

I don't have a printer yet. And i'm dying to know if the plastic "plate" will survive one match of street fighter

[vvp]

assuming iid error on the A and B steppers and respective belts, dY = c(dA - dB). error cancellation.

I thought about it but the only result is that that I more believe that corexy does not have any built-in error cancellation for stepper pulley / belt meshing errors. That equation is just a mechanical coordinate transformation thing.
Here is the picture:

Lets assume the carriage is moving up first. Both steppers are moving the red/blue arrow directions, i.e. red (left) stepper counter clockwise, blue (right) stepper clockwise.

• The bottom teeth of the red stepper pulley will be aligned to the left side of the belt teeth.
• The bottom teeth of the blue stepper pulley will be aligned to the right side of the belt teeth.
• Now we change the direction of both motors so that the carriage is moving down.
• First both steppers move a little bit without moving the belts/carriage.
• Then finally the red stepper pulley will engage the right side of the belt teeth and move the red belt.
• At about the same time the blue stepper pulley will engage the left side of the belt teeth and move the blue belt.

There you go. There is full play even in the Y direction. No cancellation.

CoreXY is cool. If I would built a cartesian bot it would be corexy. But I do not think it has any error cancellation features related to the play between pulleys and belts.

[vvp]

If anyone want to have a go on my just designed arcade buttons that use cherry MX (how come nobody thought of that before?!) here you go:

print one of these http://www.thingiverse.com/thing:421598

and one of these http://www.thingiverse.com/thing:417509

just pay attention to the 2 paragraph instructions on each "thing" and you should be fine.

I don't have a printer yet. And i'm dying to know if the plastic "plate" will survive one match of street fighter

What printer are you planing to get? As you designed it, it will be very hard to print with an FDM printer.
The problems to fix if you want to use FDM:

• Remove the top rim of the housing so that the cylinder wall ends at the top suface of the Cherry MX mounting plate.
• Align the tops of the clips to the top of the mounting plate.
• You may need to make clips thicker so that they are stronger.
• Turn the whole thing upside down.
• If the inner overhang is wider than nozzle diameter then make it smaller.
• Print with support touching build plate enabled in slicer so that the outer overhang does not sag.

The button itself should be printable as you have it but the stem may not be strong enough.

[gcb]

If anyone want to have a go on my just designed arcade buttons that use cherry MX (how come nobody thought of that before?!) here you go:

print one of these http://www.thingiverse.com/thing:421598

and one of these http://www.thingiverse.com/thing:417509

just pay attention to the 2 paragraph instructions on each "thing" and you should be fine.

I don't have a printer yet. And i'm dying to know if the plastic "plate" will survive one match of street fighter

What printer are you planing to get? As you designed it, it will be very hard to print with an FDM printer.
The problems to fix if you want to use FDM:

• Remove the top rim of the housing so that the cylinder wall ends at the top suface of the Cherry MX mounting plate.
• Align the tops of the clips to the top of the mounting plate.
• You may need to make clips thicker so that they are stronger.
• Turn the whole thing upside down.
• If the inner overhang is wider than nozzle diameter then make it smaller.
• Print with support touching build plate enabled in slicer so that the outer overhang does not sag.

The button itself should be printable as you have it but the stem may not be strong enough.

to be honest, i was thinking about visiting a hackerspace and was going to print the threaded version of the housing split in two... so that i'd sandwich the switch and hold everything thogheter with a nut   so i'd be printing 2 tunnels.... but yeah, i gave up that idea when i start to consider metal printing on shapeways...

but i think i can improve it for FDM with your suggestions...

1. i was planning on adding a few trapezoids on the side so the switch could take a heavier beating... but yeah, without the support i think i should do that. (that is actually the bottom (which cover your number 4... i still have to get used to the community preference for inverted Z)

2. the part is parametric and people will probably adapt it to several sizes. made it that way to fit it with most 80's cabinet panel wood depth. but i will adapt it to sanwa spec when i get my switches in the mail. then i will probably fine tune it more.

3. what do you suggest? the part actually has one variable wall_thickness that control the desired minimal wall of the whole part. it is currently 1 or 2mm...

4. have to get used to everyone's upside down world

5. are you talking about the button plunger (keycap) hole vs the main cylinder body size? yeah, that is a mistake that i will fix if not, no clue

6. i understood no word from this one.

[vvp]

If you are going to print it in metal using SLS or EBM then it is good just as it is.

add 1,2) Removing of the rim at the top was intended so that if you flip it upside down then the whole MX switch mounting plate will be touching the build plate. No support needed there and it will stick well to the build plate. The alignment of the tops of the clip id done for the same reason.

add 3) Thin short walls (like in the clips) are not very strong when they come from an FDM printer. FDM printed plastics will split (de-laminate) along layer boundaries more easily. The short vertical walls start to be usable at about 2 * nozzleSize. If you need to print thing cuboids then it is good you if you can orient them in such a way so that filament is laid down along the longest dimension of the cuboid. This is not possible in the case of your button housing so other option is to make the wall thicker.

add 5) Yes. If the overhang should not be there than even better.

add 6) You will when you will actually print and play with slicers for FDM printing. The point is that outer overhang (that is the overhang opposite of the inner one you are going to remove) cannot be printed in air. It is too big for that. It would sag down. It must have some support from the bottom. Slicers can generate such a support automatically. I was telling you that you should ask slicer to "generate support touching build plate" automatically.

[mkawa]

assuming iid error on the A and B steppers and respective belts, dY = c(dA - dB). error cancellation.

I thought about it but the only result is that that I more believe that corexy does not have any built-in error cancellation for stepper pulley / belt meshing errors. That equation is just a mechanical coordinate transformation thing.
Here is the picture:
(Attachment Link)
Lets assume the carriage is moving up first. Both steppers are moving the red/blue arrow directions, i.e. red (left) stepper counter clockwise, blue (right) stepper clockwise.

• The bottom teeth of the red stepper pulley will be aligned to the left side of the belt teeth.
• The bottom teeth of the blue stepper pulley will be aligned to the right side of the belt teeth.
• Now we change the direction of both motors so that the carriage is moving down.
• First both steppers move a little bit without moving the belts/carriage.
• Then finally the red stepper pulley will engage the right side of the belt teeth and move the red belt.
• At about the same time the blue stepper pulley will engage the left side of the belt teeth and move the blue belt.

There you go. There is full play even in the Y direction. No cancellation.

CoreXY is cool. If I would built a cartesian bot it would be corexy. But I do not think it has any error cancellation features related to the play between pulleys and belts.

what you're saying is true regardless of which direction the steppers move. however, look at the next pulley. in each system. we're looking at the error _at the carriage head_. in y-motion, the force of the two belts oppose each other. that's where the error would otherwise be coming from. you have to calculate ultimate positioning error due to red and blue forces.

[vvp]

what you're saying is true regardless of which direction the steppers move. however, look at the next pulley. in each system. we're looking at the error _at the carriage head_. in y-motion, the force of the two belts oppose each other. that's where the error would otherwise be coming from. you have to calculate ultimate positioning error due to red and blue forces.

If you mean there are no moments of force on the big carriage (and therefore it should not twist a bit while it is changing direction) then I agree. That is the advantage against hbot.

If you mean that the forces will not move the small carriage because they compensate each other then they better should. If they would not then it would be worse than a simple cartesian because what was intended only a Y motion would cause a small X motions too.

If you mean something else than I do not understand.

[vvp]

There may be something wrong with Makerbot Gen 5 machines:
http://www.hacknorway.com/project.php?id=28

[damorgue]

There may be something wrong with Makerbot Gen 5 machines:
http://www.hacknorway.com/project.php?id=28

Can't help but be a little happy that Stratasys seems to fail at everything they try.

[salcan]

There may be something wrong with Makerbot Gen 5 machines:
http://www.hacknorway.com/project.php?id=28

Can't help but be a little happy that Stratasys seems to fail at everything they try.

Hm, kinda interesting. I don't have a z18, but we have a replicator 2x at work and it won't even get close to printing a passable keycap.

[mkawa]

ssys is headed downhill quickly.

unfortunately i seemed to have acquired a lot of insider baseball on the company and the future does not look particularly good for them.

other than AHEM someone's ridiculously incredible single product company, i don't really know much about the rest of the industrial market, but i do know that all the consumer efforts seem to be failing on top of that.

[mkawa]

what you're saying is true regardless of which direction the steppers move. however, look at the next pulley. in each system. we're looking at the error _at the carriage head_. in y-motion, the force of the two belts oppose each other. that's where the error would otherwise be coming from. you have to calculate ultimate positioning error due to red and blue forces.

If you mean there are no moments of force on the big carriage (and therefore it should not twist a bit while it is changing direction) then I agree. That is the advantage against hbot.

If you mean that the forces will not move the small carriage because they compensate each other then they better should. If they would not then it would be worse than a simple cartesian because what was intended only a Y motion would cause a small X motions too.

If you mean something else than I do not understand.

yes, i'm talking about the former, specifically twist and ringing. obviously during a steady state feed, the thing is just moving and that's about that.

[damorgue]

The consumer SLA machines which are reaching the market appear good. They cost a bit more than the FDM ones, and the material is a bit more special and expensive. On the other hand though, with a more unique and better controlled material comes the advantage of higher consistency among batches and manufacturers. The technique itself is less prone to problems too. There is no chance of clogging, over/under heating or damaging other combonents and it has less mechanical moving parts. I think it can provide the reliability and consistent prints which the consumer market is lacking. SLA also allows for finer detail than FDM so the added price can be somewhat justified by that too.

The only thing which needs to happen is a decrease in price and I think it could grow big. Hopefully people's initial expectations on cheaper FDM doesn't throw them off the idea of at-home additive manufacturing all together.

[mkawa]

i haven't been following SLA. i do know that the FFM machine market is starting to become dominated by inexpensive chinese clone designs whose prices are inexpensive enough to match their utility (not all that high). a lot of the early home SLA efforts that raised initial funds on kickstarter, etc. don't look all that promising, but the market is pretty much wide open, so i'm quite interested to hear if someone with competent manufacturing and engineering resources has come out with something nice.

imo, nothing driven by an 8-bit micro is competent, so just toss all of those out immediately

[gcb]

Do you guys know a good reference for designing for FDM? i can't find anything.

[vvp]

If you search my posts in this thread I was giving some recommendations.
I do not know about some better design guide for FDM. Googling seems to return only some guides for commercial 3d printing which may be applicable if you have an excellent slicer but home made reprap machines seem to have more limits than what is mentioned. It probably depends a lot whether you have a separate dissolveable support material.

[mkawa]

Do you guys know a good reference for designing for FDM? i can't find anything.

if you're doing true FDM and not filament fusing, you can use the shapeways references. they're pretty good for the majority of machines. they have specific wall tolerances to their machines, but generally you want to stay much larger than those anyway.

[vvp]

Based on wiki there FFF (fused filament fabrication) is just another name for FDM. The difference is that Stratasys has a trademark on FDM.
So the question probably is whether it is going to be  printed on a commercial machine or a reprap. The process is the same. The differences are in firmwares and slicers and whether you have a separate nozzle for support material.

[mkawa]

i can't really say much more, but FDM doesn't even require thermoplastics to be used. the stratasys patent is limited, but there are a large number of techniques that are considered FDM.

we're mincing terms at this point though. it's kind of like how AHEM arcam calls their process Electron Beam Melting (EBM) instead of Selective Laser Sintering (SLS), because their laser is so thin and there are specific proprietary magics going on inside the vacuum chamber that constitute their IP protected process.

[vvp]

OK, so the process is not the same (or the names FDM/FFF are not specific enough). Thanks.

[damorgue]

I honestly see little future for FDM. They aren't being developed fast enough and struggle a bit as it stands. Their biggest advantage is that they are less limited in how big they can get, something which isn't a concern to the hobbyists. The example I posted earlier where it is used to build enormous building components is an interesting example which SLA or SLS would have had a hard time accomplishing.

There are machines which use the better technologies for additive manufacturing which and approaching consumer level pricing. Here is a new cheap-ish SLS printer:
https://www.kickstarter.com/projects/1812935123/ice1-and-ice9-the-first-low-budget-sls-3d-printers?ref=card

I recon the first wave of "at-home-3d-printers" has hit and passed leaving many wanting more, and that FDM didn't turn out as good as many people had hoped. I think it will soon be the time for better 3D printers to find their way to hobbyists.

[vvp]

Hopefully they can deliver. I would buy an SLS printer for 5k if the material price would be comparable to what we have for FDM. If they would want 10 more for material (as leslieann mentioned it is with SLA) then I'll pass.

[damorgue]

Don't forget to  account for less material use. Instead of using more filament to build a support structure and waste more material the part is suspended in reusable powder which supports it instead, and you will likely have less failed parts which will have to be reprinted. Material prices depend a lot on bulk volume and I think they can be decreased somewhat for low volume hobbyists if this thing takes of more stores dedicated towards hobbyists begin to offer it.

[mkawa]

i think one thing that people forget when thinking about printing in general is how freaking hard it is in any number of dimensions. a HUGE number of extremely hard engineering problems had to be solved to turn computable objects in 2-space into pieces of paper and even though some devices are basically commoditized now, there are many that aren't: very high resolution, solid inks, very soft and very hard printed mediums, perceived color correspondence between source and output, extremely large format, extremely small format, etc. etc. etc.

butler lampson is one of the smartest people to have lived in the last 30 years, and he spent like 5 good years trying to make graphite fusing laser printing on paper happen. that crap was hard!

although i hate to say it, because i think it's kind of silly, FFM will almost certainly settle into life as a commodity where the printed medium is a starchy polymer, the machine designs will differentiate based only on whether the bed moves horizontally or vertically, and the support is just more water soluble than the print. there's absolutely nothing wrong with this, as long as the machines reach commodity pricing. see: idealab's new mod-t printer. longevity of this machine will be about the same as a typical 30$ inkjet printer, but frankly, that's enough for the vast majority of users. ie, they'll get it to make one or two objects either to demonstrate a point or to replicate a neat design they saw, but their interest will mostly end there.

zzzz.... more later.

[vvp]

makwa: 3dPrinting does not sound hard to me. How it could be hard when people build printers themselves at home and they actually work quite well. There is a lot of work to make them better (especially better firmwares and slicers) but it is just a simple implementation. Medium level code monkeys can do it. Strengthening printer structure is not hard either. No new research is needed.

damorgue: Material wasted in FFF printers on support and failed prints is about 30% at most. That is my experience. You cannot claim much savings there. You can claim better savings in time when designing parts (less constrains for SLS/SLA) and post processing (when support is not soluble). And of course, some things are well doable in SLS/SLA but not practically doable in FFF.

[gcb]

so the cheap filament printer i will find in the nearby hackerspace is not exactly an FDM?  what keyword should i look for when researching design guides?


PS: it is freaky as hell having fred mercury staring at you while you read the nice informative posts about 3d printing.

[vvp]

It is probably FDM. FFF is just a different name for FDM.
http://en.wikipedia.org/wiki/Fused_deposition_modeling

[vvp]

I was talked into trying to print a keycap (including the stem).
Doing it the way an original stem looks like is not possible with 0.5 mm nozzle.
So this is out of the question:

But after removing a small sprue on the switch stem it is possible to make the keycap stem bigger and then it is printable. Quite a bit of clean up is needed around the stem. I painted it with acetone to make sure the fibres fuse better. It is seems to be strong enough.

The top view. There is some error in the layer alignment on the keycap sides. Not sure why that happens. The common reasons are fluctuating hotend or bed temperature or uneven extrusion. But that is probably not my case. I would blame it on slicer :-)

It fits OK.


So I need to correct my original opinion. It is possible to print a keycap with a strong enough stem using FFF. It is just a lot of work.

[NeedAFix]

This is proof of concept though, now all that musts needs be done is Improvement, Dramatic improvements!

Innovation FTW.

[vvp]

Version 2. The good about it is that only post processing needed was to cut away the remains of the brim. And I painted it with acetone. It may not be needed. But it is quick, makes surface smoother, and helps layer adhesion.
I changed the stem dimensions slightly. Before I needed to "scrub"  the inside of the stem too. Now it fits Cherry MX like a butt on a chamber-pot  )

Ok, this is only for fun. People here clamour for double shot PBT keycap magic so FFF printed ABS will not do. But still, it was fun to try it.

[gcb]

now you got me wondering... why not print the whole thing? all the way down!

imagine printing the cap, steam (not sure about naming. the brown/blue/black/red/etc thing that pushes the contacts.)

would also have to cut one side of the switch cove to slide everything in.

maybe it will be harder because that lower part requires even less clearance on the end...  but should be interesting.

[vvp]

I do not think it makes sense to try to print the switch itself using a reprap:
* you cannot print the metal parts anyway (so one would need to buy a switch for the metal parts)
* ABS/PLA have low glass transition temperature - the switch housing would soften and deform when soldering the switch
* its lifetime would be low since ABS/PLA are less abrasion resistant compared to delrin
* some parts of the switch are too thin (notice how I needed to increase the size of the keycap stem compared to the original to make it printable)

[vvp]

As for as the research and 3dPrinters, this was recommended for people who want to improve delta printers.
http://books.google.sk/books/about/Parallel_Robots.html?id=78DHjrzNt9oC

[Krogenar]

I was going to put this in 'Great Finds' but figured this might be the more appropriate place.
A company called USCutter that is located in Redmond, Washington is having an Open House on December 12th. Why is that important? They sell a 3d printer and various vinyl cutting machines - and they're giving away a 3d printer and a vinyl cutter as door prizes. I received an email with this test:

Redmond Washington Open House on 12/12!

If you live in or will be visiting the Northwest December 12, consider this your invitation to come see us at our big Open House at our USCutter location in Redmond, WA. There will be food and drinks galore. We’re giving tours of our warehouse and showrooms. And we’ll be showing off the latest new Mimaki, our full line of 3D printers from Leapfrog, Phoenix, and other equipment that can help make your 2015 more prosperous. We’ll also be giving away a Mimaki 28” CG-SRIII and a Phoenix 3D Printer as door prizes and other goodies as well.
The fun starts at 12 Noon and we’ll be open until 7PM. Mark your calendars now and plan to attend. We'd love to see you there!

Now, I'm in New York - but I know a lot of Geekhackers live in the California area, so if I were you guys --- ROAD TRIP! Get as many people together as possible, and work the numbers on the door prize. The Phoenix 3d printer is here: http://www.uscutter.com/Phoenix-3D-Printer-EZ3D

Hope this maybe helps someone get into their first 3D printer.

EDIT: Ok, I realize now that California is actually VERY FAR from Seattle. 

[Asininity]

I was inspired by others and tried 3D printing a keycap. I didn't use my own design but from someone on Thingivesrse The results were mediocre (not due to the file but the 3D printer.)




The greatest difficulty was the Stem. With a little work I was able to clean it out.


I'm hoping to solve these issues by using a uPrint that's available to me. It uses a proprietary cartridge and costs $1 per gram, which is a lot. However, the keycaps I made (using a different 3D printer) weighed only 1g.

[KHAANNN]

I was just thinking of printing/designing a simple keycap, upside-down printed so there are no supports, therefore having a flat surface (as a down-side)
(and consider using 3d printed keycaps from now on : )

PLA comes easy to me, but I'm not sure how it will act as a keycap, yet I'm sure the texture from the 3d printing will make the cap feel great

Any inspiration / existing designs? (I checked the very few thingiverse designs, yet they all concentrate on design/accuracy rather than printability)

I guess it's not a very good idea, better buy dsa/pbt blanks instead

[vvp]

You do not need to print it upside down. FFF printers are good enough at bridging to print a keycap. Just design the bottom side of keycap top to be horizontal and it will print just fine without support. See here:
https://geekhack.org/index.php?topic=43362.msg1507680#msg1507680

[KHAANNN]

You do not need to print it upside down. FFF printers are good enough at bridging to print a keycap. Just design the bottom side of keycap top to be horizontal and it will print just fine without support. See here:
https://geekhack.org/index.php?topic=43362.msg1507680#msg1507680

Theoretically yes, but practically it's much easier, robust to print without bridges and supports, also the end results is usually much more pleasing

[tp4tissue]

Hey guys..  I'm about to pull the trigger on a Form 1+  ..  any other recs that I should look at  @ this price range?

[BlueNalgene]

Hey guys.  I made a thing.  I originally posted the .stl files in the CAD thread, but I figured I should cross post it here in case anyone finds it useful.

I'm teaching myself Fusion 360 so I have some experience with CAD which is based on a GUI rather than scripting.  My practice item is a Row 3 spherical keycap.  This should be similar to the SA keycap dimensions but not quite the same.





I've sent off to have Sculpteo print a copy of it off, so I can see how it looks in the real world.  My first time using Sculpteo over i.Materialise, so I am interested to see how their stuff looks in comparison.

If you want the .stl file, I attached it to this post.  If somebody wants a different format, let me know, and I will try to hook you up.

I got the real world version in today.  My 3D printed cap. 


The height is somewhere between Signature Plastic's DSA and SA row 3 profiles



I'm glad I did this test run.  The stem shrank in a way I didn't expect, and the cruciform is a bit too big.  It slips on and off the stems too easily.  This is an easily fixable problem.

[CPTBadAss]

Nice cap bluenalgene

[MOZ]

Nice design.

[TheJonas]

Don't mean to bump a dead thread, but didn't wanna make my own if it's already there. I've got access to one of those Cube 3D printers and a MakerBot Replicator 2. I've tried printing a cherry R2 keycap and it turned out alright, needs a bit of filing, but I recently found out I can change the filament to be much thinner. I've also tried printing a topre keycap, but had no support and the edges turned into spaghetti. I don't have much time to tinker around with 3D files at the moment, but once I do I would want to make MX compatible sliders next. Anyone have any experience with these?

[sinusoid]

Don't mean to bump a dead thread, but didn't wanna make my own if it's already there. I've got access to one of those Cube 3D printers and a MakerBot Replicator 2. I've tried printing a cherry R2 keycap and it turned out alright, needs a bit of filing, but I recently found out I can change the filament to be much thinner. I've also tried printing a topre keycap, but had no support and the edges turned into spaghetti. I don't have much time to tinker around with 3D files at the moment, but once I do I would want to make MX compatible sliders next. Anyone have any experience with these?

This thread shouldn't be dead imvho

Makerbot Replicators 2 are solid machines.

What do you mean by changing the filament to much thinner? This shouldn't affect your prints. What you want to change is the print head diameter, preferably to something like 0.2.

If you want to avoid spaghetti, cut the models into two parts that will print nicely and glue them together, then file.
You can also use Replicator's two heads to print support from a dissolvable plastic. for PLA use PVA, for ABS use HIPS.
PVA dissolves in water, HIPS in Limonene.
That's overkill in most cases, though.

As for 3d printing keycaps: you need calipers, some 3d software, patience and time. Print, make amends to the model, print again. FFF printers tend to mess up the dimensions on a submilimeter scale, there are several factors that stack up on each other:
- calibration
- imperfections in the filament
- plastic properties (flow, shrinkage)
- approximations in the slicer software
- oscillation from mechanical parts
etc.

From my experience slicers mess up the most, both commercial and opensource. Stick to one slicer, get to know it, you'll do fine.

[TheJonas]

Thanks for the reply and excellent advice. I indeed didn't mean change the filament but instead the printing head. Cutting the molds in half is a great idea, but I'd rather not do that. Getting to know one printer and make adjustments based on your prints is probably what I'll be doing now.

[KHAANNN]

Anyone else get a new spool for every new project?

Just got a spool of Colorfabb's PLA/PHA in black, can't wait to try it out - to print a keyboard plate

I'm just hoping it's more matte then regular black PLA, because that turns out pretty glossy

[TheJonas]

1 spool per project would drive up the costs, then yet again, running out halfway could be worse. Show us some pics of your 3D print once it's done

[KHAANNN]

1 spool per project would drive up the costs, then yet again, running out halfway could be worse. Show us some pics of your 3D print once it's done

For me it's a seldom hobby or occupation, so spools costs aren't much of a worry, haven't ran out of a spool yet - tho my current chinese black pla is about to be done

Here is my initial prototype


I was going to use this case with colorfabb bronzefill, yet exotic filaments are too risky, and I'm not sure how it would turn out, so decided to stick to black PLA instead, as the keycaps are Dolch, and gray + black + Dolch would probably go well

Here is a prototype bronze plate:


After clogging my initial hotend with bronzefill, then dislodging the replacement E3D with XT-CF20, sticking to PLA seems like the better choice, tho I have a hunch the E3D should handle bronzefill well, but can't risk it