Pillowing and Warping on Extra Fine Settings

[Kathryn 0]

Recently I've begun test printing a part for production on my UM3 that has a number of dimension-sensitive holes in the X and Z directions. As a test I went ahead and printed the part with both Normal (.15) and Extra Fine (.06) settings using Ultimaker PLA with PVA supports. The Normal print turned out fairly well, but the Extra Fine part ended up with some rather severe pillowing on the top layer, as well as significant warping in the back right corner despite using a thin layer of glue as directed.

 

 

I used the default settings for the Extra Fine, so there should have been 17 top layers which should have reduced the pillowing somewhat, but clearly it didn't do much. What's also odd is that the finish of the Extra Fine print is actually much more matte than the Normal part. 

 

Does anyone have experience printing larger surface areas with the Extra Fine setting that might reduce this pillowing?

Also are there any methods other than a glue stick that are useful for reducing warping? For the time being there isn't an enclosure we can use for the full printer, but if that is a common solution it'd be good to know.

 

Any other general tips and tricks for getting a part to print with consistent dimensional accuracy would be welcome, since as of now I've had to go into the files for every part I've printed to make adjustments for the printer.

[gr5 2,173]

There's 4 issues here.  Let's take them one at a time.

1) Going down to .06mm layer is only going to improve dimensional accuracy in the Z axis.  So if those cylinders and holes are the wrong height/depth this may improve that.  For example if one of those holes is supposed to be 0.13mm deep it will actually be printed .1mm deep if layer height is .1 and .12mm deep if layer height is .06mm.  I'm guessing the Z accuracy isn't your issue so I recommend going back to 0.1mm layer height.

 

2) Pillowing.  Has to do with temperature and infill.  Yes 17 top layers should have been enough!  I guess it wasn't.  Make sure the side fans are at 100% as this helps a lot with pillowing but since the 0.1mm layer height worked fine maybe you can ignore this advice.  But if you really need 0.06mm layer height you could increase the infill to 90% and I'd set gradual infill on and set it to 0.3 step size (5 layers per increase of infill percent) and 5 levels of graduation.

 

3) dimensional accuracy.  I assume you mean the holes and cylinders are too small.  That's a common problem with FFF printing and PLA in particular.  The best solution is to adjust the diameters in CAD.  Sorry.  You would have to do this for injection molding as well (well someone in china would have to do it) as injection molding you also have to do weird adjustments to the model to get it to come out to the correct final size - for example if you want a 90 degree angle in injection molding you usually have to go for 88 degrees in the mold.  Something like that.  After a while you become a pro at adjusting the diameters.  I usually add 0.4mm to every vertical hole diameter (horizontal holes don't need adjusting).

 

4) Corners lifting.  Personally I don't use auto leveling - if you are doing manual leveling then just turn the 3 screws CCW 1/4 turn (looking from below) to move the bed up a little which will cause the filament to squish a little better.  If you are doing auto leveling you can turn the 3 screws after AL finishes but eventually you will run out of screw/spring space.  Alternatively you can set "initial layer flow" to 150% so that the first layer squishs into the glass really well.  In addition to squish it's good to have a very very thin layer of PVA glue on the glass.  There are 3 easy ways to do this.  One of the ways is to put a thin layer of glue stick only on half of the glass (put down stripes of glue stick) then spread it with a wet tissue such that you remove 90% of it and spread it extra thin.  Heat the bed and wait for it to dry before printing - usually dry by the time the bed reaches 60C.  Much more info on getting your part to never come loose here:

 

[yellowshark 153]

Personally I am not convinced that 17 layers is enough, especially if your infill is sparse. If you increase your infill to 90% as per @gr5 then you may find it is OK - only one way to find out! Thinking about it, which I never have really, I would say from experience that the thinner your layer height then the thicker you need the total of the solid top layers to be to get the job done - this is though significantly affected by infill % - with 90% the first few layers of 0.06 will probably lay down quite well.

I have never used a slicer that provides variable infill density so have always increased the upper layers

Edited May 31, 2018 by yellowshark

[Kathryn 0]

Thanks for the feedback! Since the z dimension isn't actually that crucial I will probably just go back to using the .15 layer height I've been testing with.

 

@gr5 it's interesting that you mention the horizontal holes shouldn't require adjustment, since I've actually had to tweak both vertical and horizontal holes by anywhere from .01 to .02in at a time, at least for this part in particular. In your experience do these issues become more pronounced as you scale up? 

You also mention this is a particular issue with PLA- are other materials like Nylon better suited to dimensional accuracy?

 

As for the warping it is possible that that may be due to me tweaking some of the first layer settings to help reduce the elephants foot on the part- I have been using the glue stick + wet paper towel method but I'll be sure to check out your other suggestions as well.

 

Thanks!

[gr5 2,173]

On 5/31/2018 at 9:28 AM, Kathryn said:

I've actually had to tweak both vertical and horizontal holes by anywhere from .01 to .02in at a time, at least for this part in particular. In your experience do these issues become more pronounced as you scale up? 

I tend to simplify.  We all do.  Otherwise I'd have to write a 300 page book to answer every question.

 

The vertical holes are going to need more adjustment than horizontal holes.  It helps to understand the problem.

 

1) STL files only include triangles.  No curves.  So your cad program which has a cylinder of diameter e.g. 10mm is converted into a polygon with e.g. 12 sides (this is a setting in your CAD when you export to STL).  The points on the polygon are ALWAYS on the circle so they lines of the polygon will be INSIDE the circle so this makes holes smaller.  This is a tiny, tiny affect.  Increasing the points helps a bit but if you have too many points then the printer will get upset (it doesn't like it if your polygons triangles start to get much smaller than 0.2mm).  Again this is not a big deal - this is a tiny effect.

2) The main problem is this - PLA sticks to itself when liquid.  Like mucus.  Like snot.  Not like water.  This makes PLA a wonderful material for this kind of 3d printing (FFF).  PLA is also cooling extremely rapidly at first - in milliseconds it's already cooler and contracting - although this may not be important for our issue here.  As you are printing the circle on a cylindrical hole that stretchiness is like a liquid rubber band and tends to pull inward.  The PLA wants to take a short cut and not go all the way around the cylinder.  It pulls inward.  THIS is the main reason cylindrical holes are smaller than desired.

 

The interesting thing is that the effect isn't a percentage of the diameter but more of a constant as far as I can tell.  So it tends to be around a half mm - maybe 0.4mm - not sure as I don't usually worry about the last .2mm (but sometimes I do!).  I suppose it would make sense that really large cylindrical holes don't need any compensation so maybe the compensation actually decreases as you get bigger?  I'm not sure.

 

yes, other materials won't have this problem but this exact property is what makes PLA so great (ABS is very similar).  I've printed materials like PVA that don't stick to themselves so well and, well, it makes for ugly results.  filled filaments such as brassfill, have this issue - some of them are 90% brass and they don't stick to themselves so well and it's a pain in the neck filament to print with.

 

The solution is always to adjust in cad.  If you need consistent tolerances better than 0.1mm regularly you should consider a different technology such as a resin printer like the form2 or a powder printer like what they use at shapeways.  To get even 0.1mm consistent results you need to never change printers, temperature, speed or layer height or nozzle or filament *color* (yes different filament colors print differently - slightly).  You probably even have to worry about air currents so use a front cover so that people walking by won't make your parts come out differently.

 

Another common solution for some people is to drill out holes after you print the part.  It seems a bit crazy but it's so easy to do since the hole is mostly there to guide you.

 

Note that this issue of holes being smaller also affects every surface on your print that isn't a straight line.  So any sharp corner (interior or exterior) the printer is going to try to "cut corners" a bit.  But also the printer slows down on corners so over extrudes so exterior corners are more likely to sick out than in.  Interior corners definitely will not be as sharp is one would want either.  Again this can be compensated in cad or improved by slowing things down and keeping acceleration high.

[gr5 2,173]

On 5/31/2018 at 9:28 AM, Kathryn said:

As for the warping it is possible that that may be due to me tweaking some of the first layer settings to help reduce the elephants foot on the part

If you mean only the bottom layer micro brim (if that's your definition of elephants foot) then the easiest solution for me is to set "initial horizontal expansion" to roughly negative half the diameter of the nozzle although this varies depending on your leveling.  I recommend manual leveling if you are going to want to get this dialed in perfectly.  leveling will not change on this printer for many months so no need to relevel all the time - once you get it dialed in you can leave it alone.

 

I like to level low such that the bottom layer squishes really well but you get that "elephants foot" which can be fixed with "initial horiz expansion" set to around -0.2.  But this feature can hurt really tiny features (features under 1mm across) on the bottom layer.  Like the teeth on a tiny gear.

[KickahaOta 14]

By the way, in some scenarios, another way of reducing the problem of hole shrinkage (especially the problem of inconsistent hole shrinkage on different printers with different materials) is to use "polyholes" -- rather than designing the part with a circular hole, design it with a polygonal hole, with a relatively small number of sides, that matches the inside dimensions you need.The "rubber-banding" effect that gr5 described so well is much less pronounced on these sorts of polygons than it is on standard many-sided-polygon "circular" holes. So results tend to be both closer to the original design and more consistent.

 

Of course, if your design specs call for circular holes, this generally isn't practical.

 

You may find https://www.thingiverse.com/thing:2918438 helpful in exploring the way holes work, since it lets you easily generate holes in a range of sizes, using both traditional circular holes and the most common implementation of polyholes.

[Kathryn 0]

Thanks for the in-depth response! We'll certainly be considering some of these methods going forwards. The initial horizontal expansion setting is not one I had noticed before, so that on it's own is a great find.