Raised Edges

[gr5 2,172]

1) I'm pretty sure this has nothing to do with the "pause" effect of retraction because if you look at my first picture it happens on a part 30mm square. You could prove this by turning off retraction - you will still get this affect (and stringing also of course).

2) In the video, with the "lip" already there I can see that it prints in the "bowl" of the lip and then pushes the plastic around like a soup and some of it is pushed up the edge of the lip of the bowl to form more of a lip. BUT how does it get started on this lip in the first place? And why only on the overhang edge?

3) On both your part and my part there were at least 2 passes of the outer edge - I believe it does the inner pass and then the outer pass last, right? Could it be something to do with that inner pass?

4) On my part there were typically 2 MINUTES between when the print head comes back to make the next layer. So the part had long cooled to ambient air temp. This is probably not useful information versus all the above theories.

5) THANKS for looking into this and putting out your theories!

 

[gr5 2,172]

But the priming process would have to be split up between the travel move and the original priming.

 

VERY cool concept - and that would speed up printing. But I don't think it's relevant here.

 

[gr5 2,172]

You get the same thing everywhere if you print small structures too fast (e.g. minimum layer time below 5s).

 

Not sure this is relevant. In my first posting above, in the picture you can see that is only one leg of 4 plus other support structures so layer time is quite a bit over 5s, . But still you may have something as there are 2 or 3 passes of outer rim and the outermost pass is the last pass.

 

[foehnsturm 969]

1) The retraction pause was not intended as an explanation for your issue. I just found it remarkable when watching the video.

2) and 3)

The lip builds up as Dim3nsioneer explains. More or less regardless how long the layer time is:

Nozzle arrives on a tiny area - heats it up - returns after a very short time (too short for enough temperature drop, in the video you can see what a difference 0.5 sec makes) for the second loop - further heating- "stirring" several layers underneath - picking up liquified parts by cohesion (which I can see in my video raw material) .

Regarding the overhang edge, I like Dim3nsioneer's exlpanation:

Why does it happen just on the overhang side? Well it doesn't. The effect is just stronger. You get the same thing everywhere if you print small structures too fast (e.g. minimum layer time below 5s EDIT: I think it's more time between perimeter/infill lines than layer time EDIT ). But on the overhang side the amount of material to be heated up again is limited or smaller compared to the other side. So the material temperature after contact with the nozzle is higher on the overhang side.

 

[gr5 2,172]

>returns after a very short time

Is 5 minutes getting to the next layer the "short time" or is it the 4 second short time from when it does the inner pass to the outer pass? I think you keep thinking of your very tiny supports yet this happens on larger 30mm square supports as well.

>picking up liquified parts by cohesion

Hmm.

So here's my latest theory which I will call "nozzle cohesion". I hypothesize that the PLA sticks a bit to the flat area of the nozzle tip. As the trailing edge of the nozzle goes by it lifts slightly on the PLA. When the other end (the bottom) of this PLA bead is touching solid PLA the cohesion to the surface print is strong enough to hold it down. When the bottom of the PLA is over thin air, the nozzle lifts this bead up at the trailing edge (slightly - maybe 1/3 of a layer height). Resulting in a lifted edge.

I think your video would be more interesting if your cylindrical piece alternated between vertical and tilted segments so we could see the problem repair itself and then start anew on the next layer that is hanging partly over air.

I love the quality of your video by the way- the only problem I have with it is that the problem is well under way. I want to see how it gets started.

If my latest "nozzle cohesion" hypothesis is correct, the problem might be mitigated by thicker layers. Or less sticky PLA (different formulations) or different nozzle tip shapes. Or different XY speeds. Or different temperatures (high temp might make it flow more and gravity might help more here - low temps might make it stick faster to the existing solid pla).

 

[Dim3nsioneer 557]

Hmm...the term I would use for such a relationship between PLA and the brass nozzle is rather adhesion, not cohesion. To my experience, a piece of PLA melts easily on the outside of the nozzle and distributes over the surface, i.e. shows classic adhesion. This is quite nasty if you print one color and still have some other color sticking on the nozzle. Then it begins slowly to crawl down and suddenly you have a spot in your print.

If there is such an effect as you describe it, then it should become better with an underextrusion or just smaller extrusion. I think to remember that this didn't change anything? Something else which should work against such an effect is to raise temperature as the PLA then becomes more fluid and less sticky (it certainly takes less time to get the PLA above glass temperature on nozzle contact). However, to my personal experience the curling effect becomes worse with higher temperatures.

 

[foehnsturm 969]

Looking at your first picture, does it take 4 sec to print the inner pass of these tiny rectangular structures?

And does the Eiffel tower feature the tiny, slicer-generated re-enforcing infill at the lips?

This is where everything begins (watching requires some patience):

Slow motion: 2x

 

 

[illuminarti 18]

Looking at the first minute of last video, the sense I get is that the slope of the left side isn't as steep as the right side (and I presume it should be?). Even in the first few layers, the angle is noticeably less steep.

I think that what happens is that when you print the overhanging pass into free air, it doesn't just sit there obediently in free space, but pulls in against the existing print - effectively trying to minimise its exposed surface area that isn't touching other plastic. So you have the right amount of plastic in less space that it's supposed to occupy, so that it thickens and grows upwards faster than it should.

Maybe what I'm seeing is effect rather than cause, or maybe I'm just restating the problem in different terms. I don't know. But it definitely seems like the lowest layers aren't ending up where they should, even before you've built up a stock of excess plastic. Once you have of course, bad things are going to happen ever faster, because now the head has to push that out of the way just to have something to print on at all.

 

[gr5 2,172]

I definitely see quite a few lower layers compress as it prints upper layers. On the left side only. But I'm not convinced this is the *cause*. It might be the effect. I like Illuminarti's theory also. I'll call it the "surface tension mound" theory where surface tension on the plastic pulls it back inward and makes a pregnant ridge belly stick upwards.

 

[foehnsturm 969]

+1 for illuminartis explanation.

It looks like this ridge formation is a tendency of FDM process per se. Caused by cohesion, adhesion, surface tension or a mixture of all.

It only develops to a problem if the underlying structure isn't solid enough to hold back the extruded filament. And this the case when there are overhangs and / or the underlying structures are significantly reheated and softened, which is dependent on heat input = printing speed.

 

[gr5 2,172]

But maybe we can reduce the problem somehow - maybe if we raise or lower the temp or modify the nozzle or if we do one layer of skin only, or...

I will try these things soon but I still have to publish another test I did today...

 

[mastory 44]

Just thinking about the melt puddle adhering to the nozzle and pulling up behind the nozzle. If there is something to it, maybe coating the nozzle tip with non-stick PTFE or ceramic coating like used in cooking pans would help.

According to Dupont the limits on their cooking surfaces;

"Heating nonstick cookware above 500°F (260°C) can discolor the surface of the cookware or cause it to lose some of its nonstick properties. ... ...If an empty nonstick cookware pan is accidentally heated above 660°F (348°C), a temperature that far exceeds what food preparation calls for, the nonstick coating may begin to deteriorate."

260C seems to be the limit which would be just OK for ABS.

These coatings are commonly applied to engine components on special order basis. There is a group of specialty companies that do one-off jobs for hobby engine builders and race teams. Most common is coating piston skirts. The cost may be a little high, but not exorbitant.

Just an idea. Not sure of its merit

 

[gr5 2,172]

I may have solved this issue by putting both UM2 fans at 100%. I have to do more tests but the fan made a huge huge difference. I will post pictures when I'm done with all the testing.

 

[foehnsturm 969]

That's been my assumption as well. ("I looks like it is all about print speed and - indeed - cooling. Not about cooling down the part before the next layers starts, but immediate cooling in the moment of extrusion").

I think it makes sense to distinguish between intra-layer (e.g between first, second pass and infill) the and inter-layer cooling. The first causes problems with small islands, even on bigger parts while the latter is more important for small parts and for avoiding shrinkage and warping. Depending on the part geometry, one of the two issues will limit your printing speed.

 

[IRobertI 520]

I guess I should've posted about the cooling earlier but I thought it might just be a fluke for my prints and didn't want to send people looking in the wrong direction. On my UM1 adding a second fan to the head made a huge difference on prints like the hollow pyramid which is a text book example of this issue. My theory on it, which probably is wrong, is that when you cool the "skin" of the extruded filament as quickly as possible it helps resist the effects of shrinkage. Since the extruded filament is so thin I figured the rigidity of the surface might be just enough to hold the shape of the filament until the core has cooled down moments later.

 

[gr5 2,172]

Oh - well I didn't explain what causes the problem. So far it is basically the Adhesion problem but I want to test other shapes. I tested a part with square layers. A cube turned into a parallelogram at 45 degrees. The problem occurred almost exclusively at the 2 overhang corners and the problem made it's way very slowly towards the center of the overhang edge. I watched with 10X reading glasses as it happened:

When the nozzle goes around the corner it squirts the plastic where it should be and then when it turns the corner it pulls the plastic along with it a bit.

With the fans the plastic cools so instantly fast that the plastic stays where it was placed. This helps with bridging also - instant cooling of the plastic.

A .4mm bead of plastic hanging in air has SO MUCH more surface area versus volume than things I deal with in real life normally. So it cools SO MUCH faster.

I'm not convinced this is the only thing that causes the raised edges - I want to test other shapes next.

 

[eldrick 65]

Thinner layers should increase the problem as there is less room for the extruded filament and less clearance for the print head.

My experience is that thinner layers seem to help alleviate the problem on overhangs. IMHO this is simply because with thinner layers the overhang is less for each layer, so it has less material hanging out to curl, and being thinner, the material can cool faster after extrusion and is "frozen" into shape before it has time to curl.

 

[gr5 2,172]

I'd love to see you do a test and report back. I have been meaning to do a test for a month now. I guess the next time I have to print a propeller I will try this experiment first.

 

[DidierKlein 729]

So i started a subject but dimensionneer redirected me here

 

 

Hi all,

This is an issue i often have but it can be clearly seen on this object:

 

It always happens for me in these typical angles.

 

When printing it looks like the layers are curling up, on the following layers the head presses them down and the results are these ugly layers.

 

I've read on the "Show your latest print" topic an advice by colorfabb to increase the minimum layer time from 5s to 8s is this the way to go?

 

I'm wondering how this affects very small areas like the antennas of this model if it spends 8 seconds printing every layer won't it finish melting it?

 

It happens only in layers that are layed outwards of the model not inward as you can see the forehead is very nice

 

Model printed at 0.08mm print speed of 40mm/s and temp of 200°c bed at 60°c in Ultimaker blue PLA

 

Thanks all

 

Didier

 

So i've read through the whole subject and find that there is no clear answer yet

For my case the fans were running 100%, the print is with 0 infill also i do think that this has an impact on the behaviour in my particular case. The shell was 1.2mm i think so this probably also increases the behaviour.

Just wondering about the minimum layer time question i mention above, would it be usefull? And what about the rest of the object like the antenna?

 

[gr5 2,172]

Well the only clear answer is: "more fan". However, I believe that thicker layers might help but I *still* haven't tried it so I really don't know. One fix for sure it to print at 5mm/sec (for those overhang layers - about 1/3 of the model) if you read stuff above I believe this is discussed.

Personally I don't have much trouble with 45 degrees - I get the raised edges and possible parts knocked off but I don't get these bumps so much - my surface is usually almost perfect. It's at overhangs even worse than 45 that I get the bumps. So I'm not sure what the deal is. If it were me and I needed to print the above object I would start doing tests at different filaments, temperatures, speeds and so on. I got through my issues with more fan but at some point I want to revisit this.

So my advice is simple: do experiments - lower this part so that you start printing at the chin with the "lower part" quality setting. First thing to try is 20mm/sec. Next try .2mm layer. I know maybe you think .2mm layer looks like crap but just do the experiment - we might all learn something.

Try .4 shell, .8 shell, 1.2 shell. Try reducing polygons by 90% using meshlab (I'm wondering if the randomness of your surface is amplified by cura and the curling problem):

Meshlab:

http://www.shapeways.com/blog/archives/226-polygon-count-reduction-with-meshlab.html

Alternative to meshlab:

directly in mudbox, the latest version has an optimization tool that is less destructive.

http://www.shapeways.com/tutorials/polygon_reduction_with_meshlab

 

[braddock 1]

things like "increase layer thickness" and "reduce polys by 90%" are all detrimental to the quality of the print anyway, so you're exchanging one undesirable result for another. Ieyo is not insanly high poly anyway, he was reduced a great deal with decimation master in ZBRush which preserves detail.

These overhangs improve with overall scale of the object, my original Ieyo bust was 180mm tall I think, and while there's still some overhang issues under this chin, it's much better.

 

[DidierKlein 729]

I will do tests because it's not the only object on which i get this behaviour. Maybe i print to fast.

Yesterday i tried the same object again scaled down to 70% with 10% infill 2shells and 0.16mm layers and i still had the behaviour but it's less messy, braddock has a good point on bigger object it probably won't show.

I can't put more fans because at that height they are already at 100% however maybe i should print the fan duct i saw on youmagine a couple of weeks ago to make the right fan more efficient?

I will try some tests, i'm thinking of creating a little M letter in 3D and printing it vertically to see how it goes

 

[takei-naodar 0]

I am currently on my Project for improving the Fan Cooling.

Here is the improved nozzleshield to prevent the Nozzle from cooling to much

 

[gr5 2,172]

however maybe i should print the fan duct i saw on youmagine a couple of weeks ago to make the right fan more efficient?

 

Usually those hurt more than help.

I don't recommend changing 3 variables at once unless you plan to go back and print the other combinations. Now we don't know if it was scaling, speed, or layer height that made the difference.

In theory scale should make no difference. A 45 degree slope is a 45 degree slope. However this part is "rough". From one layer to the next it might be the equivalent of 25 degree slope, then 65 degree on the next layer and continue on alternate layers. That's why I mentioned reducing polygons. That's the only reason I can think that scale would help because the surface isn't smooth enough. A Z brush expert should be able to (maybe?) smooth that chin area better so that from one layer to the next the slope never gets steeper than a certain limit.

Also if you are doing tests, The letter M sounds like a good idea. And again - change one variable at a time and label each printed part with a marker because after a while you will forget which was which. And keep good notes somewhere. I keep a notebook that has every print. Also if I print the same print with different settings I give it different names so years later I can reload it back into Cura and check the settings I had used in case I didn't write something down.

 

[Dim3nsioneer 557]

A few days ago I made an interesting observation. Maybe some of you know this for a long time (in this case please just let me enjoy my delight... ).

If you blow with a fan from the underside, the overhang doesn't curl up as it is cooled down from bottom to top. So the solution might be an additional fan blowing from slightly below the present printing level upwards. What do you think?