Support material for "mid air" objects

I'm trying to print out these cool Gimbal Bearings: [www.thingiverse.com]
The problem I'm hitting is that the balls in the bearings appear in "mid air" in the STL file. I figured that I could just turn on "support material" and it would put support material under the start of the balls, but it doesn't appear to do that.

The slicer setting I've found for support material has a setting for the biggest angle where support will be added (90 is vertical) called "overhang threshold". So if I put it at 80, it will generate support for the sides of the balls, but still prints the start (center) of the balls in mid air. I've played a bit with the "Enforce support for the first "n" layers", but it sounds like this is a different feature - I don't need support for other pieces, but for the ones that are created in mid air I'd like it to generate support.

Any suggestions for settings? Thanks.

Have you tried printing with no support to see how it turns out? I'm often amazed at what the printer achieves if it is adjusted just right. When an STL has objects "in mid air" but only a very small (1 layer) gap beneath and above, the extruded plastic of the bottom layer drops the 1 layer gap to be supported by the surface beneath but without bonding (fusing) to it very well, and the plastic of the layer above the shape behaves similarly, so it is a technique that ends up producing a part with some weakly bonded, easily broken internals, and can be used to print movable hinges, cogs and bearings etc. This guitar capo for example [www.thingiverse.com] has a printed-in hinge that would be impossible to manufacture as one piece any other way (I have printed it and it worked well).

It is also what Cura does when it adds support or a raft - it leaves a 1 layer gap between the top of the support columns or raft and the surfaces they are supporting, so the bond is weak and the support is easily broken away (It is an option with a raft because the weak bond will also be broken by any warping that takes place). To be successful however, the printer must be printing really well, with the right extrusion rate and temperatures, print at just the right speed and have no loose belts etc. Such prints are far less forgiving of badly adjusted mechanics or non-optimum slicer settings than most other types of print.

Dave

Interesting Dave. I had originally sliced it with 0.1mm layer height and so it came out to be quite a few layers (5 if I recall correctly and the "in air" print didn't really work. The little dots that were the start of the "balls" just stuck to the extruder and drug around. I just sliced it at 0.3mm layer height (usually works more easily for the printer) and the dot for the start of the balls is on the second layer. So there will be a single layer of the races for the bearings and then the first dots of the balls will start on the second layer. But with single dots, will they stick to the print plate? I'll give it a shot and see. If not, is there a way to force support for that layer? If the "mid air" objects are higher in air will slic3r generate support? (I code something up in openscad and check later).

OK, I ran a few tests just to see how the default "support" switch affected things. I created a simple scad ile with just a couple blocks stacked with a good 5mm between them. Without the support switch enabled it was just sliced as printing in mid air. With the support switch enabled slic3r added support for the upper block which seemed pretty reasonable I sliced the bearing again with and without the support switch at 0.3mm layer height. Without support the start of the balls was on the second layer (no support). With support enabled the balls were still printed without support, but there was a second layer of "support" material added in between the first layer and the layer where the balls started. It was support for some other random overhangs and it really didn't make any sense. Effectively every other layer was support material only, but that means that all the "normal" pieces of the bearings were printed with 0.3mm gaps in between the layers. Seems like something went wrong there.

I'll give it a shot with the "no support" slicing and see what happens. Thanks for the tips.

Well, I have tried to slice that model with overhang threshold of 80 degrees. That took a few hours, ugh!

All the model is almost completely soaked with the support material (expectedly =), including right beneath the balls. The SM is separated from the model by quite some distance. See the attached picture, where I've selected layers 5 to 18 for rendering, just where the interesting stuff is happening.

Slic3r v1.1.6. Layer height 0.07 mm, extrusion width 0.5 mm.

The bad news about my slice is that some of the balls got fused together. They intersect in the original STL file, and Slic3r has joined some perimeters between balls (not surprising!).

That is a HUGE thing (in geometry sense, 20 MB worth of triangles!), so it is not comfortable to experiment with. Consider cutting a bit out of it for experimenting. Once you are happy with the result, go for the whole model.

BTW, did you notice that the author recommends to print it on a raft (presumably, with no support material =)?

Quite a few of the designs published on thingiverse are unprintable or will not work as they should. It seems there are quite a few people who go ahead and publish designs that have never actually been printed or tested. If the balls in the design in question are touching they will be bonded together and I don't rate the chances of breaking them apart to be very good. Besides which an STL of touching spheres is not manifold (simple). Objects should either be intersecting or have mating faces (rather than mating lines or points), or not touching at all.

Dave