Poor horizontal-supported surfaces.

[CCVirginia]

On several models that have horizontal or near horizontal overhang I am getting an odd pattern of single threads, even with support (See picture - center of Guantlet). I know this is the "big challenge" of FFF, I am looking for advice to make it better.
The model I am using to illustrate this is the "Gauntlet" from here: http://www.thingiverse.com/thing:596966, from the e-nable initiative.

There are also some long slots that are a challenge and tend to have "fallen strings", even with Simplify supports (Under paper in picture) - in fact the supports did not seem to have much impact but I may try some different supports (the photo is without supports in the slots, modeled supports for the center).

odd pattern

GuantletTopAndSlots.jpg (51.06 KiB) Viewed 14074 times

I have tried it with the supports in the model as well as supports from Simplify3d. Simplify looked a bit better but not great. It seems like there is something odd going on with the way it is sliced that may be causing this problem. I have tried slowing bridging WAY down (20% and 80% fill) as well as the various support. Other than these fully horizontal sections, I have been surprised how well it did on the "almost vertical" parts which suggests overhangs as much as 65 Degrees can be done without support.

The other parts of the model look fine and the calibration has been done for the filament. (Well, almost fine - there are some small "divots" I am trying to get rid of as well).

Thoughts?

[insta]

Support inflation set to 0.50mm, dense support 2, dense density 70%, infill percentage 20%, bottom layers 1, top layers 2, horizontal separation 0.50mm, print speed > 80mm/sec, PLA temps @ 190-210C, fan at 100% speed, layer height 0.2 mm.

Give those a shot

[jsc]

If you look at the gcode preview for that file, you will see that the layers want to curve in and start doing a little wiggle over the support on each layer. Having any sort of curves over support is tricky, because in order to be able to remove it the slicer needs to put in a gap between support and model, and curves will tend to cut the corner. You may have seen this when you're not getting good first layer adhesion and trying to print holes or convex curves.

Suggestions: check your support settings to see what upper vertical separation layers is set to. If it's not 1, make it 1.

Slow the layers that contain the overhang way down. You said you slowed bridging down, but it's not a bridge, so if you just set the bridging settings, that won't take effect here. Use separate processes to slow just those layers down.

Print the whole thing upside down, which would take less total support anyway.

[CCVirginia]

If you look at the gcode preview for that file, you will see that the layers want to curve in and start doing a little wiggle over the support on each layer.

Wiggles over a support would seem a bad idea, is there any way to stop that? I think that is the pattern seen, even when printed over the "built in" supports. Any place you have partially or fully unsupported regions I would think you would want a straight line of minimum length - in this case it changes direction in the worse place. Also, it doesn't even seem to be putting in sufficient plastic to fill the void.

This video shows a similar part from the same source: https://www.youtube.com/watch?v=OLJJ0T801Qc
Here is a screen shot from that video (last minute or so):

Part printed in video

OtherGuysPart.PNG (304.17 KiB) Viewed 14056 times

Differences I noted: The built-in support was much more tightly attached than mine, which mostly dropped out like it was not attached. His resulting print was very smooth, note that he used Cura slicer. He used the model with built-in supports, so support options would not matter. Also, the model looks smoother - even with the same layer height - makes me wonder if layer width (.42) or extrusion isn't ideal. You can see every strand in my print, on all surfaces.

ps: I'm using Inland black PLA, MG-2 with V4 hot end. Hot end @215, bed @60.. Top and bottom support layers are both 1, other shell layers are 3.
I will try some of the suggested settings as well.

[jsc]

The wiggles are because that's what the model geometry requires. It's not a straight bridge, there are two subtle grooves being printed at an angle.

Did you run an extrusion calibration? viewtopic.php?f=3&t=1964
Under extruding will exacerbate the problem.

Slowing down will also help. If you jump to 18:00 in his video, you can see his actual print speed, which seems quite slow to me. Maybe 40 mm/s? Ah, you can see his print settings in the video, 50 mm/s. Good guess, me.

You can test the results of any settings without having to reprint the whole object by sinking the model down into the bed until the surface you're interested in is just over the bed. Makes it less wasteful when testing.

[Bratag]

Any reason you aren't using the version of the part with the built in support?

[CCVirginia]

Any reason you aren't using the version of the part with the built in support?

The print pictured was done with the built-in support, I have tried both.

I think I see what is going on. The mostly horizontal surface is at an angle and, perhaps, a slight bow - so the layers would be "terraced" in some way. S3D's response to this is to print this "wiggle" pattern but for some reason is not doing it well - you will note that there is space between the filament lines on the same layer - the layers are not fully filled. On a test I flattened the horizontal bottom and it then made a square pattern. that was filled.

So problem one is that S3D is not slicing this well and is producing a sparse layer. Then compound this with the "wiggle" over a support and it pulls the filament out of wiggle room. Perhaps if it was very slow this may work, but I think problem 1 would still be there. Does this sound reasonable?

So I may try fooling S3D in various ways or try Sli3r or Cura - does anyone have current (V4) profiles for these, I could not find them? I can also try posting to the S3D forum.

Has anyone printed these models with success?

[sthone]

100_4098.JPG (73.71 KiB) Viewed 14008 times

I printed one last night with similar results. (as seen above) I've never done anything with bridging so I didn't know what to expect. Over all I don't think it's too big a deal as they get covered with foam and Velcro anyway but who knows what the separation is between the built in support and the actual model. You might try downloading the non supported version and running S3D supports and see what happens.

-Steve

[jsc]

Yes, it is "terraced". No, I don't think that matters. If you are using dense infill layers, it will be printing over those anyway. The amount of terracing is dependent on the layer height and the angle of the surface. Other slicers aren't going to magically do anything different from S3D, although their default settings may be different so you might get different results. But looking at that video guy's settings, other than the print speed his settings seem pretty normal: two filament shell, .40 width, .20 height.

I have printed curves over dense support before, many times. Other than an occasional problem with small holes, I haven't had any issues. I think you should run an extrusion calibration, it can't hurt and it may help.

Oh, hey, I just sliced the one with the included support. You definitely don't want to use that. I was wondering why you guys kept referring to "bridging" when it shouldn't be bridging at all. Well, with the included support model, it doesn't know that it's support and is in fact trying to bridge on the bottom layers. Plus, the "support" is actually a mostly hollow box with beams (scrub the End slider at the bottom to see what the interior looks like) and it will be trying to actually run curves over thin air between the beams; with only three top layers the quality of the top of the support structure will be abysmal. Use S3D supports with 3 dense 80% layers and one top layer separation, and it should do a much better job. Although, it kind of sounds like you already tried that?

[jsc]

Here's my test of just the area in question.

IMG_3057.JPG (147.77 KiB) Viewed 13998 times

And after some half-assed sanding:

IMG_3058.JPG (132.09 KiB) Viewed 13996 times

Also, that screenshot doesn't show the same part. It's not the gauntlet, it's "left palm supported".

If I were printing the whole thing, I would flip the whole thing upside down, set supports to 60 degree overhang and 2mm grid, and let it rip. Print time reduced by about half.