45 Degree Overhangs

There is currently a moratorium on overhangs of greater than 45 Degrees when printing layers due to lack of support.

Current method for printing a layer is as I understand it (and could well be wrong) that the outlines or edges are printed first and then the infill second.

I guess not too surprisingly printing an outline for an over hang greater than a certain amount. means printing the outline onto thin air and clearly will fail. Gravity whilst sometimes useful is a bitch like that.

Here's the question....

If the infill of the part of a layer that was to overhang was printed first (so didn't over hang by more than 45 degrees, or if it did was to some extent cantilevered). WOuld it then be possible to print the outline after efectively hanginin it from or sticking it to the infill and therefore increase the angle of the overhangs we can currently print at?????

I do realize that this is probably not how the current slice n dice utilities work, is there any mileage in trying it though do we think????

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

With some tricks you might be able to increase the angle a bit, but this method isn't going to allow you to print at extreme angles approaching 90 degrees as eventually gravity is going to bend the edge downwards when the iverhang extends far enough from the main body of the part.

Maybe being able to build at, say, 60 degrees is useful enough? A suitably organic-looking part with curved/angled surfaces or buttresses rather than lots of orthogonal flat faces might be self supporting in this case, even if it was a really overhangy shape.

Nophead has nicely shown that ABS can fairly easily build bridges which are supported at either end, and small protrusions and recesses can also be built without any additional tricks.

I think the way to go is using ABS as scaffolding with a fairly sparse grid of pillars supporting overhangs. The trick there is to work out how sparse the scaffold can be, and the optimal extrusion parameters to build it so that it can be separated from the part without too much hassle. I'm keen to try this myself, but without access to a working printer it is tricky A recent blog post implied that Dr Bowyer was working on this problem, so it might just be solved soon without having to resort to overhang hacks.

I think I am with you on the fact that support is going to be necessary somewhere or other.

Given the lack of success in identifying an ideal support material and the benefits of being able to print a whole item without tool change I am also in agreement that the support may as well be what you are printing with.

Bio mimetics in geometry and architecture is something as an art nouveau and Antoni Gaudi fan I am rather keen on.

Given the above, the less plastic that is wasted/recycled the better. Don't get me wrong, recycling is better than waste and therefore good but is better still where not necessary. It's an embodied energy etc thing.

I guess where we can do more than 45 degrees then less support material will be needed.

It follows therefore that there is complementary benefit to pursuing the viability of both options.

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Now I think about it, scaffold would be most efficient in the case where the surface it supports is parallel to the printer bed, given that we know that bridging between supports can work quite well. Angled surfaces, unless they're reasonably self supporting, would need more dense scaffolding to hold them up as they're being built.

Perhaps it would be easier if the bridges between scaffold points did not have to be parallel to the bed, but that would require the firmware to support coordinates movement with the Z axis, and not building in layers means the toolpath generator would have to avoid collisions with parts of the scaffold higher than the current layer of the part... argh. All too complex

maybe I should just try my hand at writing a scaffold generating script, and sticking out here on the offchance that someone with a printer might play with it. I'm hesitant to replicate work that Dr Bowyer might have already done, however.

scaffolding shouldnt realy be a major issue at any angle to be honest (though i still dont have my printer done so cant do tests). with being able to do bridging a point of contact scaffold should work at each edge point where a turn of the extuder head would be required. the base scaffold could be a cross hatch alternating up to the point of contact # and a ^ crown where it will meet the build to maintain the bridging and support the sag as neccessary. for internal structure support im not sure if this would work as it still might be difficult to remove the scaffold afterwards. but each ftl file should be optemized by the designer with the scaffolds in place. (I Think )

\.....
x\....
  \...
  #\..
  # \.
  #  \

Here's a demonstration of what I meant. As you pointed out, for flat surfaces parallel to the bed, scaffolding is fine. But for angled surfaces, like the example above, it is not.

The ... is the part. the # is some scaffolding, \ is of course the outer surface.

Look at point x. If the surface between the top of the scaffolding and point x slopes outwards at greater than 45 degrees, it will not be possible to print x, as it will be unsupported. So lets add another scaffold.

\.....
#\....
#y\...
# #\..
# # \.
# #  \

Now look at point y. If the surface between the top of the first scaffold and y slopes outwards at greater than 45 degrees...

I'm sure you get my point.

Scaffold density is going to be affected by two factors... how far we can stretch and unsupported bridge between two points, and the angle at which a surface slopes.

It isn't practical to completely cover a sloping surface in scaffolding... it would use up lot sof plastic and would require lots of after work to carefully clean it up. Small surface features might be lost entirely, unless scaffolding is built in such a way that it will easily part from the printed object's surface...something easier said than done, I suspect!

So yes, in principle it is easy. In practice, there will be lots of little issues to sort ot.

AFAIK comercial fabbers use a tree-like morphology of the support as sketched here:


So you need only small volumes of support and through forking at the top you can define any surface as support for the fabbing.

I have a chunk of support-material from a 3D-printer - it's a brownish, brittle hollow tube-structure made of some sort of shugar ...

Viktor

Interesting scaffold idea... I hadn't thought of that. But it doesn't have any significant advantages over plain old vertical scaffolding poles aside from saving material... the previous problems of scaffold density and detaching scaffold from the part after printing still arise.

Of course, if you have a nice, brittle, easily removed material everything is fine. BUt in the absense of such a thing, we're stuck with ABS, and will just have to try fiddling with print parameters and hoping everything comes away cleanly.

There is still the problem of extruding fresh layers where the edges are between scaffold tops, and so over thin air. I guess it just needs someone to test something like Andy's idea to see how much we can get away with before a part starts collapsing around the edges.

Parting looks like it is on the way to being solved.

Nophead has done a bunch of work on this with his detachable rafts etc and I think Adrian is also doing a bunch of work on this (I think he tweaked some of the firm/soft ware to add some degree of functionality) and making some progress.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

In steel welding is is not uncommon to :-

Weld overhead pipe/pressure vessel seams from inside the pipe/vessel. ie upside down. It takes skill and the right venting/burn proofing but is regularly done in the petro & chemical industries.

Create a "Lace Curtain" to infill holes in metal work using torch and welding rod (maybe possible with mig/tig too, I don't know). Starting along a top metal edge and working towards the lower edge taking in the sides as the infill is built up in layers. Again a job of skill but regularly done.

It all relies on the viscosity of the weld puddle and it's adhesive capability being greater (or made greater) than the effect of gravity.

Thoughts for what they are worth.

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

If you are doing support material for an overhang in the Y direction (call it front to back) then your support must be dense in that direction, but it can be sparse in the X direction (left to right in this case) as dictated by the bridgeable length and the number of turns to be made. Thus a curved object would probably need to be dense in both dimensions, but a straight sided one could be relatively sparse. This, combined with the tree formation Viktor proposes would probably reduce the actual amount of material used by a great deal, but I worry about the software demands of such a system. It seems to me that the software required would need to A- recognize the overhang (trivial) B- analyze the overhanging slice with reference to material properties like bridging strength and shrinkage as well as the angles, curves and segments of the edges of the slice and the amount (length) of the overhang(tricky), and then "build down" from the highest overhanging layer to the floor taking into account the lower overhanging layers while minimizing material use(also tough). This is without addressing the problems of hollow objects and other hard to detach areas.
This is why I am still a fan of finding a soluble or otherwise flowable support material which could be recycled.

Quote

analyze the overhanging slice with reference to material properties like bridging strength

Given that we make single material prints, things like 'max bridging span' can be yet another build parameter, and should be calculable without too much hassle. One experimental print would do.

Quote

and shrinkage

We make no allowance for this anywhere else, nor is it a trivial parameter to calculate or make use of. Why is it required here, particularly?

Quote

and the amount (length) of the overhang(tricky)

Only tricky if you have complex undulating overhangs. And even then, it inly needs to be calculated if you're being efficient with the scaffolds... you could always just give up being clever and liberally print dense scaffolding over the problem area and go back to sensible and efficient strategies elsewhere where the angle of slope is easy to calulate.

Quote

This is without addressing the problems of hollow objects and other hard to detach areas.
This is why I am still a fan of finding a soluble or otherwise flowable support material which could be recycled.

Evidently such a material exists, but absoluetly no-one has any idea what on earth it could possibly be, and we aren't really short of clever people on the project.

Not only can we do ABS support *right now*, it has the additional advantage of requiring only a single build head, and all the simplicity that implies.

I'd rather have a reasonable solution right now that vastly expands the sort of objects that reprap can print, than a perfect solution in a year's time.

Ru Wrote:
-------------------------------------------------------
> Evidently such a material exists, but absoluetly
> no-one has any idea what on earth it could
> possibly be, and we aren't really short of clever
> people on the project.

Say what? What about PVOH [forums.reprap.org]? You even replied to the post about it!

There is a question as to whether it will set fast enough to work as a support (though, after quite a bit of thought, I really don't think this will be a problem). In any case, the statement that "absoluetly no-one has any idea what on earth it could possibly be" is quite unbelievable.

Quote

Say what? What about PVOH

What about it? It may, possibly, be useful. Only No-one knows if it sets fast enough to be a useful support material, nor if filament will stick to it.

So it is a wild guess, that might be slightly better than the other few dozen wild guesses we've had,

Quote

In any case, the statement that "absoluetly no-one has any idea what on earth it could possibly be" is quite unbelievable

So, commercial extruded plastic prototyping machines use a support material. It evidently does the job very well.

Does anyone know what this material actually is? If so, why have you chosen to hide it from us all over the last year people have been asking about support materials?

when it comes to straight runs of angled material 45 deg or more in angle doing a cross hatch bridge ############# for the length of the overhang where you print several short lines in the direction of print and two or more parralel lines in the direction of the overhang moving up the build area and caping the point of contact with a ^ point of contact is very similar to the tree concept victor pointed out and would also use very little plastic. and should also be reasonably easy to remove.

I also believe that the support structures should be IN the stl file rather than calculated by the software. that takes one time effort of the objects designer and no longer is a real issue (other than shrinkage which is again another subject).

Quote

I also believe that the support structures should be IN the stl file rather than calculated by the software

I think this is bad for a few reasons:

1) Hassle. The designer is going to have to sit down and add in each and every bit of scaffold himself, which may be very tiresome for more complex parts. The more difficult it is to design and print your own parts, the less people will be keen on doing so.

2) No material choice. It requires you to make all scaffold with the extruded plastic, never allowing you to make use of other materials. Multi material file support is a large project in itself.

3) No geometry choice. Different scaffolding strategies such as the crosshatching you just mentioned, or the tree shapes Viktor described would require a complex and time consuming redesign of the part. Intricate tree scaffolds would be awkward to place manually.

4) No deposition parameter choice. Improvement in scaffolding strategies with different extrusion parameters (speed, temperature, etc. look at nophead's rafts for examples) cannot be used with pregenerated scaffolds, unless (again) a much more complex and expressive design file type is used.

5) Scaffolding is a largely machine or technology dependent thing. Right now, STL files can be printed on many different kinds of device... printing scaffolds on a commerical rapid prototyper would be pointess and wasteful. Needing different designs for different systems is tedious at best.

A tool for adding scaffolds to an existing part avoids all these issues, and it is not a hugely complex task. It would be applicable to all designs and materials. Finally, it is a process which lends itself to computer automation... let the computers do the dull and repetetive jobs, and leave the designers to concentrate on what they do best!

editted for lousy spelling

Edited 1 time(s). Last edit at 09/13/2008 10:49AM by Ru.

6) The shape may be printed more than one way up.

Ru Wrote:
-------------------------------------------------------
> So, commercial extruded plastic prototyping
> machines use a support material. It evidently does
> the job very well.
>
> Does anyone know what this material actually is?
> If so, why have you chosen to hide it from us all
> over the last year people have been asking about
> support materials?
Because it's hard to pronounce the actual components of the materials.

Actually, it's probably because the focus of the people with RepRaps who would be able to find/understand the information are busy with other things, like making the basic extrusion work properly (and the people without RepRaps are flailing around trying to get them ). Or something like that.
Then again, maybe they just want to come up with something that's not patented.

Anyway, there seem to be at least a few different support materials researched by the commercial companies. Stratasys seems to use two support systems: BASS breakaway supports, and WaterWorks.

According to patent # 6869559, patent application # 10/019,160 discloses the WaterWorks material (see [www.google.com]). Application # 10/019,160 is now patent #6790403 (see [www.google.com]).

So it seems that the support material is a polymer of methacrylic acid and methyl methacrylate mixed with the plasticizer (the plasticizer is named in claim 9) and can be dissolved in an alkaline (basic) bath.


The patent system was designed to encourage the release of research to allow further scientific and economic progress based on that research. It's one of the main reasons it exists, and it's one of the main reasons we don't have to sneak into Stratasys and reverse engineer a spool of their soluble thermoplastic to figure out what it is... not that we would, uh, ever do something like that...


The main difficulty with such support materials, then, will not so much be figuring out what can be used as a support material, but more at figuring out how to manufacture these support materials at such small scales and with the available equipment. I don't know how much of a help I can personally be in this matter, as I am more of an inorganics guy.


In any case, I do think that we should get scaffolding working (using the build material) for a few reasons (which all seem to blend together and are really just repeats of what Ru said earlier):
1) We can do it now. Any water soluble thermoplastic is at least a few weeks/months away (if not years, but let's not be pessimistic).
2) Not all materials will work with a single soluble thermoplastic, so for certain materials we will need scaffolding anyway (if not of the build material, then of a breakaway support material like in Stratasys' BASS). This is one of the reasons why Stratasys has two support systems.
3) Scaffolding techniques will transfer over even after we get a soluble thermoplastic, as they can minimize waste and warp (that is, we can build the scaffolds out of the soluble thermoplastic).
4) Scaffolding with the build material only uses one head (hopefully), which I suppose has a minor advantage in simplicity... which doesn't really matter all that much.

But I also think that we should work on getting a soluble support system working, more than we are already.
Ru Wrote:
-------------------------------------------------------
> I'd rather have a reasonable solution right now
> that vastly expands the sort of objects that
> reprap can print, than a perfect solution in a
> year's time.
I'd rather have both. We should continue working on both software/firmware scaffold generation and soluble supports.

So, any progress on the scaffolds?

Edited 1 time(s). Last edit at 09/17/2008 12:58PM by Joshua Merchant.

Why not use another type of plastic as support material? Such that the print plastic sticks but does not weld to the support plastic, so that it is easier to breakaway. Or is there even such a plastic?


Hears some detail on the support material(fast forward to 2:08):
[www.youtube.com]

Apparently Dimension BST printers are capable of printing ball-bearings:
[www.youtube.com]

Quote

Why not use another type of plastic as support material?

The problem there might be finding two materials that have similar melting temperatures (cos you don't want the build to melt the scaffold, or vice versa) which don't fuse together nicely.

I don't doubt that such materials exist, however we have only managed to extrude a fairly small number of plastics to date so I don't think that anyone has come across a suitable candidate just yet.

Quote

I'd rather have both. We should continue working on both software/firmware scaffold generation and soluble supports.

There isn't really any separate work to do. The same principles apply regardless of the support material; the only difference is the extrusion parameters (which will need tuning on a per-material basis) and the small matter of toolhead swapping.

Quote

So, any progress on the scaffolds?

I believe Dr Bowyer is doing some work on using a single plastic as both main and support material (though I can't for the life of me find the blog post about that. maybe I'm going mad ) so I'm in no particular rush to duplicate his efforts.

Not having an extruder to experiment with makes doing practical work in this area a little challenging , to say the least.

What about building >45 degree overhangs by rotating the entire reprap?
If the reprap is mounted on a 45 degree slope, it should be possible to print any maze and put a ceiling on top.

-Geert

hey! That's a nice idea...

Just an extra axis or two to rotate the whole thing...hmmm. Would be at least partially functional.

Demented

why rotate the entire reprap?

Some new bed corners for the z axis and a four motor conversion (just like nophead for example) and one could tilt the build table at least some degrees.
The tilt depends mostly on the bedcorners so having a good design for that we could fab overhangs to 30 maybe 25 degrees that should be enough I guess.

On the other hand, I have no idea how to controll the four z motors independently.. but I'm sure Viktor knows about that, because here we're getting close to parallel kinematics

'sid

Rotating just the bed is far harder than it seems. For starters, this will mess with positioning. Even if we know the angle of the bed very precisely, we suddenly depend not just on relative positioning to some arbitrary zero, but need to know absolute distance from the rotation axis. Also, assuming rotation around the Y axis, we'd have to adjust the Z axis when we want to move in the X direction.

While any rotation of the reprap during printing may be too tricky to deal with as a result of changed forces, it might make sense for some kind of objects to work with a rotated reprap. Basically, just change the direction of gravity. Of course there is the slight issue of deposited material adhering to a rotated work surface...

The intriguing part is that experimenting with a rotated reprap is very easy and it might be the case that only software needs to be changed.

GeertB, the software change is what I was wondering about...I think we are probably making a lot of assumptions about which way gravity is going and that might be hard to change...


Demented

I am truly amazed at what seems to be a total, long-term allergy amongst reprappers from the core team all the way down to the most casual builder to tackle the support material problem in some way more substantive than merely verbally.

hehe, true! We will go so far as to rotate the whole machine to avoid it...

Demented

LOL! It makes no sense whatsoever to me.

Similarly, nobody has seriously tackled the problem of grinding recycled plastic or extruding 3 mm filament in well over a year, either.

Edited 1 time(s). Last edit at 10/22/2008 01:34PM by Forrest Higgs.

They aren't glamorous! Everyone wants to be a Rock Star with the extruder or with Printing stuff...

We'll get around to it I'm sure...

Demented

Demented Chihuahua Wrote:
-------------------------------------------------------
>
> They aren't glamorous! Everyone wants to be a Rock Star
> with the extruder or with Printing stuff...
>
I think you've put your finger on the problem.

I know I'm trying to get the software for milling with Tommelise 2.0 going. The job is jam-packed full of tedious little problems that I have to solve. I've noticed that when I see what other people are doing on the blogs, I'm strongly tempted to go do something a bit easier that has more flash to it.
>
> We'll get around to it I'm sure...

I wonder. It's literally been years, now.