Ultimaker - Black Edition

[aviphysics 9]

I did that in my first sketch. But then I realized it would become very difficult to get the gantry aligned when you have every single piece for itself. With these long brackets, the shaft spacing is defined and very accurate (this is a 5mm thick, lasered aluminum part!). On top of that, you just place the first two brackets flush on one side, getting them perfectly aligned. Then you place the other two brackeds flush onto the first two and have them perfectly aligned as well.

The only thing that you still need to care about is to make sure that the aluminum extrusion profiles form perfect 90° angles.

Yes, it means you need to calculate the spacing for your belts carefully. I'll probably have to re-order these parts for better belt tension, but the second batch should be just right for the belts I'm using. It's very easy to adapt these brackets for longer or shorter belts. I'll probably make different sizes anyways as soon as the "standard" size is finished.

The first sketch here has a bad gantry sizing anyways - it can move further than necessary in the x and y directions. There is no use in being able to move the hotend beyond the edges of the build platform... (of course, a few millimeters more is good. But I have about 1-3cm surplus just for nothing.

I did this because I had already ordered all the expensive precision shafts from Misumi and then redesigned the whole frame. Now the first prototype will be a bit off in dimensions, but it should work well. And it'll be easy to adjust the spacing once I can re-order a new set entirely.

Oh and by the way, I "should" have taken into account all of the aluminum profiles' length tolerances (+-0.5 mm !) so that there won't be any problems with that. There will be a very small error introduced by that (the 90° angles will be slightly off), but that should be negligible.

 

I might consider having an adjustable stop on one bracket and, so that you have a way of fine tuning them to be a perfect 90 degrees.

 

[aviphysics 9]

Grrr. Fedex says my parts from Misumi were delivered today, but they clearly were not (wife was home all day and says there were no deliveries. Also, no one at FedEx is answering the phone.

Update: Driver says it was delivered to the wrong house. Will be here by 8 pm tonight.

 

[jonnybischof 60]

I see you can just replace them... you could also cut them in 2 and use precice spacers between each to get good alignment.

do you have any good link to learn about tolerances & material quality etc... was looking at link below, how to compare with your misumi, and the quality of f.e the UM2 shafts ?

http://www.dold-mechatronik.de/10mm-h6

http://www.misumi-europe.com/en/online-shop/proxy/precision-linear-shafts

 

Possible... I would still just make new brackets if I needed a different length - KISS

I don't have any references for mechanical tolerances & quality. When you're able to buy stuff from Misumi, you have two major advantages:

1. You get the best quality there is

2. Lots of information in their catalogue, like "use these kinds of shafts with these bearings". Also, you can generate STEP files for the parts and their mechanical drawings are very well readable.

So, I basicaly don't have to worry about these things anymore

 

I might consider having an adjustable stop on one bracket and, so that you have a way of fine tuning them to be a perfect 90 degrees.

 

No go - the profiles are what makes up the rigidity of the machine. It is absolutely critical that they form a rock solid entitiy. I will measure the angles when I receive the parts (5 days left..).

I suppose that "+-0.5mm" length tolerance refers mainly to "how well they dial their cutting machine". Meaning when they cut several parts of equal length, these parts can be +-0.5mm in respect to the specified length. But less than +-0.5mm difference between the parts.

If that's true, then the error in my frame becomes even smaller.

 

[aviphysics 9]

No go - the profiles are what makes up the rigidity of the machine. It is absolutely critical that they form a rock solid entitiy. I will measure the angles when I receive the parts (5 days left..).

I suppose that "+-0.5mm" length tolerance refers mainly to "how well they dial their cutting machine". Meaning when they cut several parts of equal length, these parts can be +-0.5mm in respect to the specified length. But less than +-0.5mm difference between the parts.

If that's true, then the error in my frame becomes even smaller.

 

I wasn't meaning for the adjustment to be on the aluminum frame, but attached to the plate that holds the rods (it looks like it could slide and be firmly screwed down against the extrusions along a range of points. Currently, it looks like you plan to just slide both plates to one side of the frame, but you could have a screw that is mounted to the frame or to the plate that allows you to have an adjustable stop. The adjustment would probably be small, so the plate wouldn't have to tilt away from the frame. It would just be slid a little one direction or the other. Then you don't have to rely on the frame to by dead on perfect.

 

[jonnybischof 60]

Okay, now I get what you mean

It is indeed possible to slide the brackets if necessary. But that would make the assembly too difficult. So I want to avoid any kind of adjustment wherever possible. There are already some parts that need careful adjusting (build plate levelling, z-limit switch, z-stage linear bearing alignment). We don't want to overpower the user by giving him 1000 screws to fiddle. In the end, he'll just mess everything up...

If the construction is not simple enough - I will have to redesign it.

By the way:

The Zurich FabLab is closed over the whole holiday (w0000t??? :eek:), so I can't make the CNC milled parts in time. I'll be able to put most of the gantry together, though, so there will at least be some pictures and maybe some insight in how good the design looks in reality - compared to the paper design.

 

[aviphysics 9]

Makes sense in that context. I was thinking of it as a professional tool that you were building for yourself.

 

[drayson 75]

Jonny, where have you ordered the precision shafts for x/y??

I get rid of mine as they'r not 100% straight... Should not be an issue but start to suck viewing it from a mechanica engineers position.... :-)

Btw... I ordered some Festo couplings for my bowden - maybe also usefull for your setup...

[jonnybischof 60]

I order my stuff from Misumi. Premium price - but they do not disappoint

But you have to be a business customer, they don't accept private customers. I'm lucky enough to be able to order stuff via my workplace.

I'm using E3D's embedded couplings:

http://e3d-online.com/Mechanical/Bowden

Very elegant solution. Once the prototypes are there, I'll probably try leaving the metal part away completely (integrating it into the printed part), so I can order any cheap coupling and just use it's "clamp" part (the inside thingy that bites into the bowden tube).

 

[jonnybischof 60]

Makes sense in that context. I was thinking of it as a professional tool that you were building for yourself.

 

I am - but it will be completely open source and I want it to be of some use to other people. It is already a derivative of a proven product - so it needs to be "matured" and simplified, not made even more complicate than it's predecessor.

I don't need much modularity, nor does it need to be easily modifiable - because it is good the way it is. I won't rest until that condition is fulfilled

And, as I said, I'm probably going to sell some of these some time.

/edit:

Heated build platform is partially assembled (pictures are still on the cam, *d'ooh*), and I'm printing the parts for the gantry.

I won't be able to do much more than that as long as the aluminum extrusion profiles are missing, because everything mounts to them.

At least december 30th isn't that far away anymore

 

[jonnybischof 60]

Here's the pictures I mentioned earlier:

 

 

 

 

I also put together the parts of the z-stage which already exist:

Works very nicely so far. But it's too early to celebrate.

One more thing for today:

This has been laying around for some time, but today I finally assembled one. I'll have to do some more tests, but it seems like it works just fine (of course it does, I made it ).

 

[zumfab 3]

That looks awfully nice and robust! I am very curious to see where this is going

 

[jonnybischof 60]

Stupid mail didn't arrive on the 30th...

Now I don't have the materials to continue working on the main parts. I've made some progress on the hotend mount and material feeder however, but not much to show off yet.

[aviphysics 9]

I know what you mean about the mail. Currently waiting on replacement parts from Misumi after FedEx messed up the last delivery.

FedEx really ought to put GPS on their package scanners to tell them when they try to delivers package to the wrong address.

[zumfab 3]

FedEx really ought to put GPS on their package scanners to tell them when they try to delivers package to the wrong address.

 

The technical state of affairs does not really seem to be the problem. The company culture and work ethics sometimes often leave room for improvement though.

 

[jonnybischof 60]

Small update:

I got my shipment and assembled the frame. The corner pieces are very strong - but not suitable for setting up accurate 90° angles.

I've adapted my design and am now going to try using lasered steel plates for the corners. This solution will have quite an impact on the frame's weight, but should deliver the rigidity I demand.

I've also ditched the Trespa bottom plate and will be using aluminum profiles in the bottom, too. That makes the frame another 24mm higher (without getting more build height), but the machine is already huge so it doesn't matter anymore

I'm also making good progress on my material feeder design. Got my first prototype "working" (didn't print with it yet, but it grips the filament well and the quick-release works, too. Rest should be a piece of cake). More to show soon...

 

[thirty6 3]

Just started laying out the design for 500x500mm with a standard frame height due to only having a 600x400mm laser cutter, the plan is to move the z stage to the side wall and mirror on the other side. Construction is 5mm acylic as i would like it in white and frosted. I would like to sneak out more Z height as per the ultimaker 2 so what do people think is the best way?

 

[jonnybischof 60]

Acrylic is not the best material for making an Ultimaker frame. It's quite wobbly and warpy. I'd choose the thickest kind of acrylic that you can process on your laser cutter (6-10mm), or prefer good plywood over acrylic. It will also be rather difficult to get a good fit for the ball bearings as acrylic tends to get strong "V" shapes when lasered.

Don't forget that the UM2 only uses acrylic for the side panels - which means they're not that much of a weakness because they're supported on all sides by other material (Dibond, which is - tbh - also not the greatest material...). Also, UM's acrylic panels are CNC-milled, not lasered. That's why they get a good fit for their ball bearings.

Progress on my UM builds is slow because I have a lot of work to do. But there's a lot going on. I now have all the parts laying around for the second attempt at my "aluminum extrusion profile frame".

I've also started to re-think that strategy because it's going a bit too far in the "heavy, difficult to assemble, and expensive" direction.

 

[zumfab 3]

Acrylic is a bit of a distaster from an engineering point of view. It is rather fragile and has some other drawbacks. If you want to go with plastics you might go with something that is a little more high grade, like POM, but unfortunately that also means things become rather expensive quickly.

Mind you, I never did a full evaluation of POM, but it tends to be superior in most applications.

 

[jonnybischof 60]

POM is great. But afaik you can't lasercut it (I may be wrong here?).

But POM is VERY expensive compared to acrylic. Would be easier to use composite materials such as HPL or Dibond. However - these can't be lasercut as well, and they're generally a bit more difficult to process.

I found Dibond too weak for my taste. The UM2 seems to do well enough, though. HPL is a better option, but more difficult to machine and slightly more expensive.

My next approach will be to have a frame lasercut from 5mm aluminum. Very similar to the UMO in design, but just made of aluminum. It's not an ideal solution imho, but I found that to be the cheapest solution I've evaluated until now. The drawback of aluminum is it's high thermal conductivity. If you want to have your UM enclosed for a heated bed, then you'll either have to insulate the chamber or you'll need much more energy to keep the chamber hot.

My idea for the aluminum frame is to make a plywood (or cork) inlay for the chamber. Design isn't finished yet, but it might be in 2-3 weeks.

 

[zumfab 3]

POM is great. But afaik you can't lasercut it (I may be wrong here?).

 

I never did, but I think you can: http://www.eurolaser.com/materials/polyoxymethylene-pom/ But indeed, POM is not cheap at all, but it is a real structural material, whereas acrylic always seems a huge compromise. HPL seems a great material too, though unfortunately I cannot figure out a way of processing that without changing the design quite a bit. You really need to router that, rather than laser it.

Looking at the different materials I felt that plain old steel has a lot going for it. Thermal stability is pretty good, especially compared to aluminum, and it's very strong and fairly cheap. You will need to protect it against rust and it is not the lightest, but as long as you do not try to make 6mm panels out of steel that will not be as bad as it sounds. The weight helps to reduce vibrations and noise too, often a problem with aluminum. Cutting is a bit harder though, you will need to pay someone to do that.

The resulting printer is a rather different device than the Ultimaker we know, but you should end up with a very robust printer.

 

[jonnybischof 60]

I've thought about using steel instead of aluminum (prices are pretty much the same, because aluminum is more difficult to lasercut than steel), but I was concerned about the weight. The Ultimaker frame's stability suffers when you decrease wall thickness (the screws and nuts don't hold that well anymore).

And 3mm steel is still about 170% the weight of 5mm aluminum.

I'm thinking about making my smaller printer for flexible material from steel.

I will change a lot of details about the frame, making it more suitable for thin-sheet material. One idea is making "composite" sheets from two 1mm thin sheets of steel and one 4mm plywood panel sandwiched in between (*). Basically a better Dibond, because both aluminum and PE used in Dibond are balls...

There are several waterjet cutting specialists around where I live. Once I'm happy enough with my (large printer) design, I'll have it waterjet-cut from HPL. That would be pretty much the ultimate way to do it.

HPL is thermally (and electrically) insulating. It's extremely strong, much stronger than aluminum. It's heavy, but not nearly as heavy as steel. Tensile strength is great, thermal elongation is minimal. It can take temperatures beyond 100°C without problems and usually has a very high quality surface (Trespa is great).

When waterjet-cut, you get very smooth and brow-free edges. It's also not as expensive as you'd think. But of course you need to make more than one frame.

//edit

*: By the way, if you do that "composite" thing, there's one more advantage: You need double the amount of lasered parts, making them much cheaper per-part. Also, 1mm steel is much cheaper and lighter than thicker steel or aluminum. Thinner sheets also come out of the lasercutter very nicely.

Plywood is dirt cheap, no need to even mention it

Thinking about it, I might even ditch the aluminum frame before even starting, and try the composite way from the start.

//edit:

I just realised that by making "composite" frame plates I can actually make the frame a lot more beautiful because I can hide most of the ugly cutouts by just not including them in the outer aluminum sheets.

That means I am building a composite frame now

 

[aviphysics 9]

Finally got my GT2 belts and quality rods from Misumi. Set x/y steps to 80 and printed a 20mm cube using default HQ print settings. Final cube came out to 20.005 along the x and y, according to my calipers, which is probably within the tolerance of the calipers.

I had previously been having a significant backlash problem in the x direction, which I had been struggling with since I got the machine.

Getting the shipping grease off the rods was a bit of a PITA. I tried using WD40, which left a gunky residue. Probably should have gone with simple green. I ended up needing to add some light silicone grease (like is used for bicycle chains) to get it moving smoothly; but IIRC, bronze bushings are self lubricating, so you should need anything. I think the problem was from a small amount of residue that the silicone grease spray helped disolve and dilute. The fit of the bushings on the rods alone was just about perfect. Move freely but no perceivable lateral wiggle.

One small issue is that the screw on reptar gt2 blocks doesnt reach the rear end stop. Almost a non issue, unless your not paying attention while jogging the machine.

[nick-foley 5]

I don't think POM is a great choice for a frame material (cost vs. stiffness isn't so ideal) but it does laser cut very well... it's probably the best thing to cut after acrylic. The part accuracy is great and the strength/slipperiness/toughness is fantastic. With our 90W laser we can get through 1/4", but that's probably around the limit. One thing to watch out for though - it has a clear flame! Burnt myself picking up some small pieces that I couldn't tell were on fire from the laser.

 

[zumfab 3]

I've thought about using steel instead of aluminum (prices are pretty much the same, because aluminum is more difficult to lasercut than steel), but I was concerned about the weight. The Ultimaker frame's stability suffers when you decrease wall thickness (the screws and nuts don't hold that well anymore).

And 3mm steel is still about 170% the weight of 5mm aluminum.

 

3 mm is already a huge wall thickness for steel Most machines I know from the workshop have 1, maybe 2 millimeter shells. Only heavy equipment seems to have more, but most strength comes from smart braces, folding or boxing the material, not pure wall thickness. I can imagine that a 3 mm Ultimaker is way too heavy and although I did not do the math, my gut tells me you can get away with less, since the surface area of an Ultimaker is actually quite small compared to other machines.

The sandwich idea is interesting, but I am a bit worried how the materials behave together, since they are not properly bonded. Vibrations might become a problem and any gaps will eat away strength. Do you have a plan how you are going to execute those panels?

 

[aviphysics 9]

3 mm is already a huge wall thickness for steel Most machines I know from the workshop have 1, maybe 2 millimeter shells. Only heavy equipment seems to have more, but most strength comes from smart braces, folding or boxing the material, not pure wall thickness. I can imagine that a 3 mm Ultimaker is way too heavy and although I did not do the math, my gut tells me you can get away with less, since the surface area of an Ultimaker is actually quite small compared to other machines.

The sandwich idea is interesting, but I am a bit worried how the materials behave together, since they are not properly bonded. Vibrations might become a problem and any gaps will eat away strength. Do you have a plan how you are going to execute those panels?

 

I think the issue is with the nuts not having a broad enough surface to balance against.

I can only imagine the complications that trying to get the pieces precision bent will cause. IME, bending is not nearly as precise as a cut line. Would make a lot more sense if he were going to be making a lot of them. Then the cost of working out the kinks would at least be distributed.

I wouldn't sandwich with plywood. The problem with wood isn't strength, but dimensional stability (mostly warping). Perhaps a plastic core material, like in Dibond, but with thicker plates. Still seems like a PITA.

Another alternative would be to add steal bands or thin rectangular rod laid flat along the edge, to increase thickness just where you need it. Of course, all that is a lot more work than Al, which would probably be perfectly good.

I had another idea for steel plates separated by a thin grating (like steel cardboard), but couldn't find anything appropriate. I feel like I have seen something like that before, but cant remember where (maybe armor plating or something like that).

Edit: this stuff looks super cool, but I bet it totally doesn't laser cut and probably costs a bundle. http://www.alsic.com/_blog/Structures_and_Armor/post/Lightweight_Aluminum_40_Percent_Lighter_than_Ordinary_Aluminum/