Extreme 80° "Atom 3D" overhang, printed on a Prusa Mk3
 

Extreme 80° "Atom 3D" overhang, printed on a Prusa Mk3  

  RSS
Vojtěch
(@vojtech-3)
Honorable Member

Ever since I saw this video:

 I've been wondering how they could have done it. A 80° overhang is bordering on the impossible.

This is the same model, on a Prusa Mk3. You can find all the gcode and project files HERE.

This topic was modified 2 years ago 3 times by Vojtěch
Posted : 20/05/2019 9:53 am
bobstro
(@bobstro)
Illustrious Member

I can't really tell from the pic what the underside looks like at 70+. Can you post some "less artful" pics showing the undersides clearly?

Clever use of the modifiers. I wonder if that's something that could be done programmatically according to overhang? Would also be interested to see if you can modify extrusion width selectively with other parts of the print on the same layer not changed. Unfortunately, I'm stuck traveling for a few days.

With a basic profile I have that print looking pretty good up to 60 degrees, with a bit of underside distortion at 70 and hairy dangling bits at 80. An improved fan helped slightly with the rear bits.

My notes and disclaimers on 3D printing and miscellaneous other tech projects
He is intelligent, but not experienced. His pattern indicates two dimensional thinking. -- Spock in Star Trek: The Wrath of Khan...
Posted : 20/05/2019 12:58 pm
Vojtěch
(@vojtech-3)
Honorable Member
Posted by: bobstro

I can't really tell from the pic what the underside looks like at 70+. Can you post some "less artful" pics showing the undersides clearly?

When I get back home, I'll try to make some photos that show the printed object in a worse light. 🙂 The 70° overhang is perfect, same quality as all preceding angles, at 80°, there is artifacts at the edges caused by the very thin PLA flexing under the force of the freshly extruded filament. But even the 80° looks okay. No significant deformation and no loose hair or loops.

Clever use of the modifiers. I wonder if that's something that could be done programmatically according to overhang?

I'd love the new PrusaSlicer to eventually acquire this ability. It requires to be able to dynamically change layer height and extrusion width for specific part of the print. I'm afraid that breaks some basic assumption made in the slicer core.

One approach that could be rather easy would be to automatically halve (or quarter) layer height on steep overhangs, basically draw the overhangs in two or four passes depending on slope instead of one. Certainly something to discuss when I meet the PrusaSlicer developers.

Would also be interested to see if you can modify extrusion width selectively with other parts of the print on the same layer not changed. Unfortunately, I'm stuck traveling for a few days.

That doesn't work currently. When there is a vertical boundary in parameters, PrusaSlicer models the layer as two separate parts with non-connected perimeters. If done on this model, it just breaks apart.

There is more that I had to work around in the project, like the fact that PrusaSlicer has several types of infill (top, solid, normal, ...) and when multiple are used in the same layer, even when they have the same parameters (same with, same 100% infill percentage), they're not merged and so the part ends up not as stiff as it could. That's why the model is set to have a solid infill layer each 1 layers.

With a basic profile I have that print looking pretty good up to 60 degrees, with a bit of underside distortion at 70 and hairy dangling bits at 80. An improved fan helped slightly with the rear bits.

A better fan would allow printing the overhang faster, 2h is still a lot for a small model like this. The most problematic aspect of the model is the flexing of the model, though, it's the reason why I print it at 100% infill - to make it as stiff as possible to be able to carry the pressure of the extruded filament under the nozzle without noticeable flexing.

Posted : 20/05/2019 2:50 pm
Vojtěch
(@vojtech-3)
Honorable Member

Ok, here is the picture. It took a while with a polarization filter, a macro lens and a light at exactly the wrong angle to make the layers well visible. This is WTL PLA, after all, where WTL stands for "Where's The Layer?". A PLA that's Low-Warp, but also makes the layers hard to see due to a nacre effect. You're seeing a little bit of the 60° step and the whole 70° and 80° steps.

This post was modified 2 years ago 6 times by Vojtěch
Posted : 20/05/2019 5:27 pm
VM00000
(@vm00000)
Active Member

Low layer height & slow printing speed & powerful cooling fan.

Your attempt does not look that bad...

Posted : 20/05/2019 6:53 pm
lord-carlos
(@lord-carlos)
Trusted Member

At 0.2 mm layer height? Impressive. Did you change the stock profile much?

Last time I tried to print this test, the filament which was printed in the air would be dragged on top of the model at the end, resulting in curling up. The next layer the nozzle would bump into it.

I printed a similar model with stock mk3 + 0.2 fast profile VS skele + LDO blower + 0.2 stock profile: https://cdn.knightlab.com/libs/juxtapose/latest/embed/index.html?uid=5927cede-7fd5-11e9-8804-0edaf8f81e27

Posted : 27/05/2019 11:44 am
Vojtěch
(@vojtech-3)
Honorable Member
Posted by: lord-carlos

At 0.2 mm layer height? Impressive. Did you change the stock profile much?

Nah. The print uses 0.3mm, 0.20mm, 0.10mm and 0.05mm layer heights and layer widths in the 0.40 to 0.80mm range to achieve the results. And that, of course, aren't the only tweaks to the stock profile. But it works on a stock printer.

You can see all the tweaks when you download the 3MF project file.

I do not think printing it with a constant 0.20mm layer height is realistic with the stock cooling solution. Even with my profile adjustments the print would benefit from more cooling significantly (I tested that, it looked much better still, but that's for another post.)

Last time I tried to print this test, the filament which was printed in the air would be dragged on top of the model at the end, resulting in curling up. The next layer the nozzle would bump into it.

The trick employed here is avoiding overhangs entirely. Given the dynamic layer height and width, the print is fast (2hr), yet doesn't contain any actual overhangs - all passes are supported by underlying layers.

I have options for improved cooling (a Sanyo Denki SanAce B52 109BC24GC7-1 5cm 24V blower and then for going crazy a 70W CPAP-style BLDC micro-blower in my parts drawer), so I want to try printing it at 0.20mm layer height with unsupported overhangs and a lot of cooling, I will report once I get to it.

I printed a similar model with stock mk3 + 0.2 fast profile VS skele + LDO blower + 0.2 stock profile: https://cdn.knightlab.com/libs/juxtapose/latest/embed/index.html?uid=5927cede-7fd5-11e9-8804-0edaf8f81e27

Very nice! I'll try some of your prints once I finish my printer upgrades, to compare. 🙂

Posted : 27/05/2019 1:29 pm
bobstro
(@bobstro)
Illustrious Member

I have been using that same print for testing cooling fan alternatives for my original Prusa i3 Mk3 (R2) configuration. Here are the results printed in eSun PLA+ at 210C with the z-seam to the rear:

It's a bit hard to tell from the picture, but the 70 degree result is slightly better for the 3rd design (2nd from right). I was curious to see the different at 80 degrees with the 4th design (at right) and realized that orientation has a lot to do with results owing to the focus of the part cooling fans airflow. I left fan design 3 mounted, and printed the same part in 2 other orientations, again with the fan in each example pointing to the front. For good measure, I did a print with the filament printed at a much cooler 190C in the optimal orientation.

A real surprise was how much of a difference the slicer used makes. Here is the same part printed with MadMaker PLA at 200C using PrusaSlicer (left) and KISSlicer (right) in the original STL orientation with the arrow to the front.

While KISSlicer is a bit rougher at 70 degrees, the results look very different at 80C. The algorithm KISS uses for overhangs results in significantly less drooping. I'm going to tinker a bit more with KISS, but this does show that the ability to tune overhang parameters would be beneficial in PrusaSlicer.

Different part fan designs make a bit of an improvement in cooling to the rear, but optimizing the part orientation makes a huge different for printing overhangs.

Edit: (while I still can) - All prints were at fixed 0.20mm layer height.

This post was modified 2 years ago by bobstro
My notes and disclaimers on 3D printing and miscellaneous other tech projects
He is intelligent, but not experienced. His pattern indicates two dimensional thinking. -- Spock in Star Trek: The Wrath of Khan...
Posted : 27/05/2019 11:40 pm
gregory.b10
(@gregory-b10)
Trusted Member

Yea I got that too; 80 degrees.  3D printer test from Thingiverse.  Not sure of the link. 

Part orientation is important.  The part fan must be able to cool the underside of the arch or filament will hang excessively.

Posted : 13/06/2019 6:50 pm
gregory.b10
(@gregory-b10)
Trusted Member

Can anyone see the picture?  It won't show up for me.

Posted : 13/06/2019 6:55 pm
Neophyl
(@neophyl)
Famed Member

I can see it but its rotated 90 degrees anti clockwise.  Its aspect ratio is also off like most pics posted in the new forum 😉

Posted : 13/06/2019 8:17 pm
Tim
 Tim
(@tim-2)
Illustrious Member

Select Link To : Media File when posting ... then folks can zoom in and see the original, which is normally unaffected by WordPress preferences.

It is always wise to get more than one opinion...
Posted : 13/06/2019 8:55 pm
Share:

Please Login or Register