How to stop convex overhangs curling up?

I have problems printing any shapes that involve convex overhanging perimeters -- the perimeter thread curls up, and in areas of persistent overhang the amount of curl accumulates each layer. Eventually it curls so much that a subsequent layer will go completely outside of the folded up section and things potentially start over. One model I experienced this with was the pink panther woman, where she mostly printed fine, but had problems up her back due to the prolonged convex overhangs.

I am currently trying to print some fairly coarse screw threads and am again encountering this problem. In this case the curl results in the thread profile turning out sawtooth rather than zigzag. I have been trying various combinations of layer thickness, thread width, temperature +/- 5 deg, speed, but nothing seems to have an effect. Here is a picture showing several test prints:

Effect of perimeter curl

In this picture, layer thicknesses vary from 0.35mm (left), a couple at 0.15mm (middle), 0.1mm (bottom right), and 0.05mm (top right, my first ever attempt at this resolution). Typically I've been setting w/t to be around 0.35mm (my nozzle diameter), but have also tried smaller and larger.

Does anyone else have these problems, and any inklings how to address it?

Other Info: "Sells" Mendel, 0.35mm MG nozzle, PLA from diamondage, Makerbot Gen3+PSMD+3G5D electronics/FW.

Thinner layer heights and/or slower layers (give them more time to cool off and become solid) will help immensely.

---

www.Fablicator.com

I'm very interested in this too. I've had this problem to varying degrees on my cupcake for the past year. Tried printing cooler, hotter, thinner layers, thicker layers, changing loop/perimeter/fill order..

Hopefully when I get my prusa running I will be able to tell whether it is specific to the way I've configured my machine or to the way I choose slicing settings in general.

It probably depends on how much you are stretching the filament. If you measure the diameter extruded into mid air and compare the cross sectional area of that to the cross sectional area of your target layer height and filament width you get a measure of how much it is being stretched. It has to be stretched a little but I think too much will make it curl when there is not much underneath.

---

[www.hydraraptor.blogspot.com]

Trying to reduce the stretching was my motivation for printing the one with 0.35mm layer height 0.35mm layer width since that should be a pretty close match to the no-stretch scenario for my 0.35mm nozzle. I haven't done the calculations though, and will have another go after applying some math.

I was surprised at how little difference there was in terms of the curl effect over the very wide range of layer heights that I tried in my experiments, and I was starting to wonder whether the volumetric 5d skeining was ensuring that the curling was consistent whatever the layer height...

Yes the height doesn't affect the stretching if you keep the width constant because lower heights get proportionally lower flow. It is the width that counts. SF is better than SFACT and I think Slice3r because it relates the width to the height. If W/T is constant then thinner layers does stretch it more.

0.35 width is very thin for ABS and a 0.35 nozzle. If you measure the plastic extruded into free air it will be substantially bigger diameter than the nozzle, maybe 0.45. That is before you squash it flat by having the layer height lower than the width.

Typically I would use 0.30 layers with a 0.35 nozzle with W/T of 1.5 giving a width of 0.45. Or 0.35 layers giving 0.525 for even less stretch. The most stretch I would use would be 0.25 layers with a width of 0.375, still more than yours. For lower layer heights I would increase W/T, so W would never be less than 0.375.

---

[www.hydraraptor.blogspot.com]

I'm struggling with this problem too. I tried using the SF 'cool' module, but plastic wants to ooze out after the motor stops, and likewise, after it orbits and waits and starts the next layer it takes a sec for the plastic to resume again.

I thought about having it build two at a time instead. That way one is cooling while the other is being built, and the extruder doesn't need to stop pumping for extended periods, but I figure this will cause the ooze problem to put strings between my two objects :[

Nophead, I read on your blog that getting the extruder to reverse every time it stops could mitigate the ooze problem, but you did it with python software on your printer? Do you know if SF can be coaxed into doing something similar with my g-code firmware?

As for slowing down the print to give the layer a chance to cool, doesn't that just give the hot-end more of an opportunity to transfer its heat into the plastic? I tried asking the 'cool' module to just make it print slower, but there didn't seem to be much of an effect.

I haven't had any ooze or strings since I started reversing 1-2mm, (depending on the extruder) at about 10mm/s. That was more than two years ago [hydraraptor.blogspot.com]. I don't extrude particularly fast compared to some people though although I think I printed PLA at 100mm/s once and I don't remember any ooze. The object did sag though.

Yes I do it in my python code. Skienforge does the same thing with g-code I think. There is also an extra restart distance but that should always be zero IMHO.

I used to get ooze during warm-up, but now I park the nozzle just above the bed so that it blocks itself. That means I can tell the machine to build something and there is no manual intervention until it finishes. Then I remove the object and the skirt and also the blob where it has warmed up. Than it is ready to go again.

I slow down to make the layer time not less than 20s and that seems to work for most things. If not I print two.

---

[www.hydraraptor.blogspot.com]

I've got a sells mendel with a Budaschnozzle1.0 and ive had issues with smaller parts curling at the edges. i learned about stretch and other culprits.
Then i realized the heatblock was causing these issues in areas just under it. I added a little pillow underneath out of ceramic fabric and kapton.
It's never an issue for larger parts but the small extruder gear is unprintable even with the pillow. I can get it done by printing 4 at once and keeping the travel feedrate painfully low. seemed worth mentioning here.

I have been having the exact same issue and it's been extremely frustrating. this problem just kinda popped up one day out of nowhere, i had never experienced it (to a noticeable degree) before. i used to be able to print the pink panther woman with no ill effects. now, when i do so, even the very slightest overhang (the bottom of her right leg, for instance) causes nasty curling such that subsequent layers and "mushing down" the perimeter as it goes around. eventually it'll curl to bad that the thing is extruding into thin air where it expects a surface to be. the only change i've made since is switching from printing ABS to PLA, though you'd think PLA would exhibit this less since it contracts less.

i've tried lowering the extrusion temp, bed temp (all the way to 0 on blue painters tape), layer heights in .050 increments from 0.1-0.3, cool feature (30 sec minimum layer time... very slow), slower overall feed rates (down to 15 mm/s), nothing seems to work. i don't think it's the hotend heat as nothing has changed in that area between now and before.

any ideas? is there some magic bullet software setting im missing? i'm using slic3r.

try turning on your fan all the time during the build. I build with ABS and this has worked wonders for me. I also have a duct on my fan so it's blowing only on the nozzle and not cooling down the entire bed.

I'm having the same problem, with the same models. I can print the pink panther woman all day long in ABS with varying speeds, temps and layer heights. When I print with PLA, however, I get nasty curling with even very slight overhangs.

I turn two 40mm fans on after the first layer is laid down. I had the exact same problem. I couldn't build the heart gears until I added cooling. The print bed stays hot enough to keep the part down even when cooling. Commercial printers have ridiculous amounts of cooling inside. Its their little secret.

Not that we needed more proof of Nophead's genius, but this immediately solved every bit of my upwards curling on overhangs. Thanks once again Nophead! ; )

Quote
nophead
Yes the height doesn't affect the stretching if you keep the width constant because lower heights get proportionally lower flow. It is the width that counts. SF is better than SFACT and I think Slice3r because it relates the width to the height. If W/T is constant then thinner layers does stretch it more.

0.35 width is very thin for ABS and a 0.35 nozzle. If you measure the plastic extruded into free air it will be substantially bigger diameter than the nozzle, maybe 0.45. That is before you squash it flat by having the layer height lower than the width.

Typically I would use 0.30 layers with a 0.35 nozzle with W/T of 1.5 giving a width of 0.45. Or 0.35 layers giving 0.525 for even less stretch. The most stretch I would use would be 0.25 layers with a width of 0.375, still more than yours. For lower layer heights I would increase W/T, so W would never be less than 0.375.

That would be great, if that did it!

i have this problem on almost any print I try with even slight overhangs.
On the calibration Cube it does appear, but to a degree, I can handle/the printer can.
But thos curled up edges are really killing me - having to sit next to the printer for hours on some prints and pull down on the just layed layer with some spatula to keep the PLA down...
By the Way, I am using 3 x 50mm fans....so cooling should not be an issue, though no ducting.

Will try this tomorrow,
if it helps, i´ll post it here.

i currently have set Layer height to either 0.2mm or 0.248mm on most prints and Perimeter Layer width to 0.35 and 0.4 wwith infill.

so how much bigger width-wise should i go?

Alex