Holes on top surface (pillowing) #771
Comments
|
|
|
I have same problem :-( I have found out that it is done by infill less than 24%, because when the head crosses over to make the squares in the infill, small dots occur in the junction. This dot will then have the head to stop running for a moment when the top layer is laid. It starts a hole and it becomes worse at high temperatures. Because of that I always use 25% or more infill. |
|
I tried a different slicer and it is definitely a problem in Cura. I can go as low as 10 % with other software and it automatically detects approaching top surface and creates a transition zone. In Cura this problem is less pronounced at 25 % infill, but the default 20 % already causes problems. |
|
This is a known problem. From what we've deducted, this has something to do with the new slicer engine using squares as infil, instead of alternating lines (Skeinforges solution). This creates sealed pockets of hot air which cause this effect. We're working on this, but we've not found a completely working solution for all cases as of yet. |
|
Note that changing the infill amount and/or changing the top/bottom thickness helps for the pillowing. (or warts as we call them) |
|
I proposed a 'sealed pocket of hot air' theory on the mailing lists 12-18 months ago for this effect. However I later came to think that it was nothing to do with that - but mostly just a problem with poor bridging. See: http://umforum.ultimaker.com/index.php?/topic/1872-some-calibration-photographs/?p=17300 It is cooling related, but I saw it start to form even on the first layer, when there was no sealed pocket of hot air. The bottom of the filament catches the infill line as it passes over it, but the top keeps moving, causing the broken tail of filament to curl. Better cooling helps the filament to curl less in that circumstance. I always thought it would be worth experimenting with doing the first layers of solid infill over grid infill at 45º to the sparse infill, so that some of the corners of the squares can be bridged first. |
|
My tests also show that it's not fully related to sealed air pockets. I adjusted the code so it does not generate pockets once, didn't yield much change. I did experiment a bit with the angles, but I didn't get much improvements yet. |
|
For a software solution I propose 45deg rotated support pattern with 80 % of the bridge span length. This should make the bridges shorted and easier to connect. |
|
Hello, I had the same problems my solution ist to slow down the print in the regions where the solid infill begins, I'm printing with 80mm/s and slow down to 40mm/s. (20% infill) |
|
This is a known problem of Cura. There are some workarounds:
The simplest code fix would go like this:
|
|
+1 for a dynamic infill density, that is increased towards the top layers! This would solve a lot of problems (and also save filament). |
|
Bug confirmed in Cura 14.12.1 |
|
I'm curious to why this was closed? |
|
@BagelOrb yes, same problem for me. |
|
@BagelOrb Bug confirmed in Cura 15.04. |
|
Yeah, this is still not fixed. Mostly because this is a quite complex problem. Part of this is hardware, part of it is the software. |
|
Which software approaches have been attempted so far? There are several suggestions from several different people in the thread above. |
|
But just because the problem is complex the issue shouldn't be closed. Though why not do the proposed solution as luotinen proposed in his comment: https://github.com/daid/Cura/issues/771#issuecomment-37218393 Or maybe just put up a warning (red triangle icon with on hover text, NOT a popup) in Cura when you choose a small amount on the Bottom/Top thickness combined with lower than 25% infill?! |
|
Material, cooling and print temperatures are the biggest factors in this. |
|
For my part, I have to disagree:
It seems to be a bridging thing like @illuminarti suggested in his message. When I shorten the bridge distance (=increase fill ratio), the problem disappears. The problem with the increased cooling is that it is not very user friendly to the less tech savvy. Also, the fan is not on fully from the bottom layers, so you will still get the warts at low points. You will also get warts wherever the air blow is impeded by the built object itself. |
|
Well, for me printing slower and changing the temperature removed the problem. Odd. |
|
I have no idea why I closed this initally (perhaps a misclick?), but I have just implemented the idea of @luotinen, of which I thought was already implemented. Do mind that you need at least 4 layers top/bottom skin for it to be enabled. Lower density infill regions (and thus higher density closer to roofs) are on the todo-list, but may be removed because it will be quite a hit on performance (computation could take 30% longer) and it is not really a trivial problem. |
|
Sounds nice @BagelOrb. :) |
|
Indeed. For smaller top/bottom thickness is it automatically turned off (in the engine, not the setting in the frontend). |
|
Testing didn't quite lead to the desired results. I am doubtful whether we should keep the option. |
|
Looking at the drawing of @luotinen, it makes sense that it doesnt really help. The holes aren't that much smaller. It could help if we rotate the infill and then shift it by half the width and height |
|
Print thicker layers or wider tracks so that the filament doesn't snap when On 10 June 2015 at 22:35, M. Vondano notifications@github.com wrote:
|
|
@luotinen I don't think the curling up is the problem, but it is a symptom. Fact is that the gaps cannot be bridged. Of course this problem can be solved by making the gaps smaller, or reducing speed, but I am trying to speed up the printing process and so I'm trying to reduce warts without introducing more lines or turning the speed down. |
|
This hasn't always been a problem. It started a while ago. I still believe that it has something to do with the way we create infill patterns. |
|
The troubleshooting guide (http://support.3dverkstan.se/article/23-a-visual-ultimaker-troubleshooting-guide#pillowing) mentions a big jump with the pattern between 24 % and 25 %. Why is there the jump with the infill pattern?
|
|
Because it goes from both directions each layer to one direction each layer around 25%. |
|
How about a solution with the holes gradually smaller but still with a bit more support from the underlying layers. The second and third image could be printed as the same layer of course. But I separated them in images to make a better visual. Hope you by the pictures can see how think. Regards,
|
|
Now when I look into what I just sketched I realized it's the same idea as @nallath posted! |
|
@hAmpzter Great minds think alike ;) |
|
So what you great minds are saying is that we should alternate the rotation and shift of the infill instead of the skin? |
|
Yeah, but rotate and shift does not do all the work. The pattern when going diagonally must be minimized a bit for the pattern to fill in the holes a bit. It just makes the hole a little bitt smaller (if you watch the image the lines are a bit closer to the yellow x's than the holes between the x's). I'm not sure if just one layer of the red and green needs to be printed or if it should be maybe two of those. |
|
Indeed, the line distance should also be scaled by a factor of sqrt(2). The fully filled top is called 'skin', while the sparse infill inside the object is what I call 'infill'. You seem to be talking of the skin, but the skin is not put down in two directions at the same layer, only the infill is. Doing two directions of fully filled skin at the same layer would be overextrusion by a factor of 2. I don't think your idea enhances the infill. |
|
Just to get things clear, the skin then is the same number of layers calculated by dividing bottom/top thickness with the layer height? So then the deal to do this would be to make maybe the last four layers of the infill (or maybe just two is enough) with this kind of structure to make a better support for the last layers on top to fill nicely when fill density is lower than 25%.. If this is successful maybe the amount of layers of skin don't need to be that many also, which will make an extra bonus! :) |
|
Did a quick test to see if the infill really influences so much. Test methodology:
|
|
@hAmpzter no, as I said, the skin layers are not filled in both directions, but in either direction, alternating between layers. Your ideas are based on the wrong premisses. @luotinen Perhaps the infill structure should default to alternating line infill instead of square infill also for infill percentages below 25% |
|
No, we speak of the same thing. Just that I did not say two alternating directions. But that was what I meant. I do not think the problem is how the layers are laid out, I think it is the fact that the amount of space between the lines when it is to fill over them is to long. The higher the infill the smaller that distance is. So I think we should gradually make that distance smaller with use of the places where it still has as much support as possible. Hence the fact that the red lines are crossing more lines than open areas. If we make the infill alternate at low infill values wouldn't that just make the structure more fragile? I think one of the things that makes the infill "curl up" when it tries to lay those last layers are the fan. You need a good fan, but when it will blow down in the cavities as it is creating the long lines the airflow will also aid the plastic to separate and curl upwards. So if we make the structure gradually stitch the air will still flow through and maybe not have as much impact on those. |
|
You simply have to extrude more plastic so it doesn't snap when spanning a gap. If you can successfully print bridges then those settings will be able to close the gaps in the infill. If not then the problem is the crossing filament in the sparse infill pattern is lifting your nozzle and so leaving a bump that breaks the flow when the next layer passes over it. As this problem seems to be Cura specific that may be the reason as extruding two lines over the same spot is unique to Cura I think. |
|
@hAmpzter you are still confused about terminology. Let's not talk of 'infill', but of 'inside' and not of 'skin', but of 'top'. (The infill doesn't curl up, the skin does!) @nophead Indeed infill (inside) is similar to support and so top skin is similar to bottom bridges. However, there is no code separating the bottom first bridge from the bridge above it. It is therefore not easily implementable to have only the first layer of a bridge, or the first layer of the top skin print with more volume. Implementing such a thing would be a considerable hit on computation time, and it would be very hard to print. The area for the first-top-skin-layer and for other top skin must either be printed separately, which causes the print to have less structural integrity, or in one go, in which case a single extrusion line must switch between higher and lower extrusion rates, corresponding to bridges and non-bridge-skin. |
|
Well you simply can't span gaps when the filament area is too small so the options are use 100% fill so there are no gaps or increase the filament area so it doesn't snap. I think Slice3r treats the first skin layer as a bridge to cope with this. The way I avoid it is to always print with a cross section that can span a gap without snapping. I don't have special settings for bridges, so no need for flow rate changes. I print the whole object with the same flow rate for best quality. |
|
@BagelOrb Right now I'm more disturbed that you did not at all comment on what I said than the text it self. And I can't see what was wrong with my comment. When I was talking about "the higher infill" and "low infill values" I was referring to the percentage of fill. And that is the "inside" you are talking about in your comment. When you have a higher "percentage of fill" the distance between the lines in the infill (inside) is smaller, hence lesser amount of distance needed to be bridged. Therefor I think that my theory to make the distances smaller (with making the distance gradually smaller between them) may work. I still think my thought on why the plastic in the skin (top) is curling up because of the returning air-flow is a valid one. Is that anything that has been tried? And as I said, if we make the infill (inside) alternate between the layers on the low percentages. Wouldn't that make the part much more fragile? PS. In this comment I have put you new definitions in parentheses, I hope we now can engage in discussing the issue instead of what part in the print we are talking about. DS. |
|
@hAmpzter Yes I finally understand what you are saying. Damn the gods of confusion! I do indeed think it might have something to do with airflow, but currently I do not have the time test the hypothesis. Gradually changing the infill percentage is a hard thing to do, because on every layer top skin might be present and so at a single layer multiple areas of different percentage gradually less overlap can occur. I've been thinking of implementing a two stage infill, where the lower infill is such that it covers half the higher infill lines. This is a feature which is on the roadmap, but currently not top priority. @nophead, Yes the safest and easiest option is to just turn op the infill percentage and turn down the speed. |
|
Did a quick test to see if the speed really influences so much. Test methodology:
|
|
Suggested remedy: mark first layer of skin (top) as bridge. Limit the speed for succesful bridging. Speed can probably be recovered back to user setting after the first layer. |
|
You can already set the top/bottom speed separately from the rest of the speeds. Separating the first top layer results in bad prints when the top of the object is not perfectly horizontal. |
|
Is it possible to only set the Top (skin) speed? The bottom speed is OK and does not contribute to the problem. Then 100 mm/s could be achieved by setting 100 mm/s to global speed and 50 mm/s to Top speed. Then I have double speed and same amount of surface warting as with default settings.
|
|
Problem: Cura with higher than standard speed (e.g. 100 mm/s) will cause bulging or pillowing or warting of the top surface (poor surface quality). Conclusions: The problem is alleviated with 25 % infill and solved with 30 % infill. However, the infill is not the problem. This is a bridging problem and can be solved by dropping the top (skin) surface speed. This does not increase material consumption and only slightly slows down the printing speed. Remedy: Two things are suggested: first add a separate top speed control (currently it is a shared top/bottom setting). This will help to target the actions and maintain highest possible speed. Secondly, add a control box to advanced setting with [ ] Pillowing prevention. If set, it overrules the top speed and sets 50 mm/s (or printing speed if lower than 50 mm/s).
|
|
The top speed cannot be set separately from the bottom speed, since the areas for top and bottom are simply classified as top/bottom. Classifying them separately means that they have to be printed separately, which can result in disconnected areas. Simply setting the speeds lower is a good fix, but not what I am looking for. |
|
I nearly always print in 50mm/s but I still get that problem. It's probably because I have the heated bed platform on and maybe because I normally have a warm area where the printer is. |
|
Well it has been identified as a bridging problem and can be remedied with the typical bridging remedy (reducing speed). Slic3r solves it by classifying the first top (skin) layer as bridge. We could look into alternative bridging techniques? Or somehow scam Cura into believing this is an area to be bridged? |
|
Bridging is currently implemented quite naively. Currently bridging only determines the angle of the top/bottom skin. Indeed bridge areas should be identified and handled separately from the top/bottom skin. I agree that this could quite well solve the pillowing problem. |
On default settings Cura makes pretty sparse infill. On topmost layers the bridging of these large gaps is difficult. Observing the process the following appears to happen: part of the material sticks to the edges of the holes and forms cantilevers. Due to fan cooling the top and heat radiating, conducting and convecting from below part, the cantilevers bend upwards. After a few passes they act as wipers and result in gaping holes in the model or weakened and thinned centers of the holes.
In other software, the bottom and top infill can be set separately and the problem may be avoided in the slicer. The hardware solution is a stronger fan. For Cura I have some suggestions:
Pillowing in top column (worse) and bottom flat regions (uneven quality) using default settings and layer height of 100 um.
The text was updated successfully, but these errors were encountered: