Need help with support for ABS

[dorgeman]

Hello,

I am having problems printing ABS filament on ABS supports. The supports remove pretty easily, which is good. The bottom surface of the print, however, is very ugly. Below is a photo of what the part looks like.

bottom surface of a sphere printed on support - ABS

It actually looks better in that photo than in real life. The whole lower surface is messed up. Here is another photo to try to show what is wrong.

Another view of the bottom surface

And here is an image of the support structure that I removed from this part.

removed support structure - ABS

I am printing 1.75mm ABS through a Bowden tube using an E3Dv6 hotend. I am printing on a heated bed, and can print ABS parts cleanly when not on support. The printer can handle overhangs pretty well - getting a perfect score on Makezine's overhang test.

I have tried multiple things to try to improve this, but so far the bottom surface always looks the same. Some things I have tried:

1. Add cooling fan. Although this is generally bad with ABS, I find a low power (15%) often works wonders. I have a custom circuit that allow low PWM rates to be translated to a low voltage to drive the fan so that I can run it at that kind of low speed.
2. Changed Support Pillar Resolution from 4mm to 2mm.
3. Changed Horizontal Offset From Part from 0.30mm to 0.20mm
4. Tried adding 1 Dense Support Layer
5. Tried Max Overhang Angles of 45, 40, and 35.
6. Slowed the default speed from 45mm/s to 20mm/s.
7. Changed perimeters from 2 to 3.

Although the removed support material sometimes looked different after the changes, the actual parts are indistinguishable from one another.

I am pretty convinced that it is possible to do better than this. Any guidance is appreciated.


-David

[dorgeman]

After trying to print the sphere dozens of times without success, I decided to give up and go print something else. Since I didn't hear from anyone on the list, I thought maybe it just wasn't possible to do much better for this shape. Given that, I decided to try to print the Makezine articulated robot from their printer tests. Much to my surprise, the exact same defects seen on the bottom of the sphere showed up in some places on that model.

I came to the conclusion that the problem WAS overhang, but only under certain conditions. I had multiple good overhang tests, but none of them involved overhang while making a tight curve. I made a small test object with a 50 degree overhang to test this theory. That printed like so:

overhang in a tight curve - before

Note especially how it looks at the smaller diameters. I would say that looks almost identical to what I had been seeing when trying to print the sphere.

After a bit of testing, I realized two things. First, the board that allows my cooling fan to be used at a low speed had burned out - so the cooling fan was never running. Second, my firmware acceleration, although OK for most prints, was probably too high for this specific case. I made a new circuit board (using a slightly higher power resistor this time) and lowered the acceleration. That resulted in the following:

overhang in a tight curve - after

Given that I now had dramatic improvement on those overhangs, I set out to try the sphere again. Now the following was the result:

sphere - after acceleration and cooling fixes

The sphere is much better now. Now the bottom is not nearly as messed up. It could still be better, but it isn't anywhere near as bad. Now I think it is just a matter of tuning the support settings a bit more. Maybe also something like additional perimeters...

If you have any input on how this could be further improved, I would appreciate it.


-David

[magicwiser]

I have also struggled with this. My experience has been that printing rounded overhangs that curve upwards (like the bottom of a sphere) always come out pretty ugly (like the first photo shows). Up until now I have just tried to avoid them but its nice to know there might be hope afterall!

[dkightley]

If you have any input on how this could be further improved, I would appreciate it.

What about using a separate process for the very bottom of the sphere and make the layer thickness thinner. Then step up to the thicker layer thickness in your main process. The thinner layers will probably come out muck better....

Adding a dense support layer for the lower process may also help.

[dorgeman]

I made it a bit better. I liked the idea of a different process for the lower layers with a lower height - mostly because it would make the extrusion width greater, which should handle overhangs better. Given that, I set the width manually to 0.6mm versus the default of 0.48mm. I also made it three perimeters to make it more likely that perimeters would cover the gaps at the very top and very bottom. I made the support 2 layers of dense at 75% and the sparse at 25% - angles of 45 degrees and -45 degrees. I have an offset of 0.2 and one layer between top and bottom. I also double clicked on the part and elevated is 2mm so that support is completely beneath it. Previously, the lowest point was in contact with the build plate. That gave me this sphere:

with cooling, low accel, improved support settings

I am pretty happy with that result. The bottom is not badly scarred by the support. The support was still pretty easy to remove. The bottom is pretty clean. You will notice that there are still gaps between the lowest perimeters. I am not sure why that is still happening. Based on the top, the width should completely close the surface. My best guess is that the lowest overhang is really a free extrusion above the support, and reduces to the free extrusion diameter - can't be squished out to the higher width. As we get to higher layers with less severe overhang there is at least something beneath the layers so that they become more able to hit the desired width. I don't think that could be fixed without allowing the lowest layers to have greater contact with the support, but then it will be harder to remove with ABS support. I think that gets more into soluble support. Although it can't be seen on this picture, there is a slight narrowing from a perfect sphere just above where the support stops and it becomes a pure overhang. I started support at 50 degrees. If I started at 45, it might help that slight narrowing.

Although I can just barely see it, there is one place where there is a small layer split (delamination). That is a common problem with ABS. It is one of the arguments against cooling with ABS. From what I have seen, however, cooling is essential to print some parts - so I think it is a balancing act between how low you can get your cooling fan and between how high you can have the extrusion temperature. I have seen others who have pushed the extrusion temp on ABS much higher than expected because it solves the delamination issue. If I go too much higher, however, the ABS melts inside the hot end before it is extruded and then there is a jam. I might take one more more shot - slight raise of extrusion temp, slight reduction of cooling fan power, slight decrease of degrees in unsupported overhang angle... - but I think the current results are satisfactory.


-David