Incorrect wall thickness?

[brousant 0]

Hi,

I have been using an Ultimaker 2 since a week or two. The primary use is to print special parts for electronics test jigs. The parts do not need to be particularly beautiful but they must be mechanically strong. So I print at normal speed but with increased wall and bottom/top layer thickness settings.

Now the problem is that in some areas the printed object is not closed at the top and bottom.

Here is a screenshot of Cura with a part that shows this problem (this part was straightforwardly designed in OpenSCAD)

In the below image you can see that there is a space between the walls that is not filled in with a bottom and top layer.

But maybe the real problem is not the absence of a bottom and top layer, but that the wall thickness is much smaller than it should be. If the wall thickness were 1.5mm as specified, that hole would be closed to begin with, and there would be no need for a separate bottom/top layer! However the measured thickness of the wall is only about 0.8mm.

Another problem with this print is that one wall did not fuse together (not only is the wall not fused to the top/bottom layers, but the two layers that make up the wall are not fused to each other).

I am printing with the first roll of PLA that came with the Ultimaker 2 and apart from the changes shown in the CURA screenshot I use all standard settings.

What is the problem here and how can I fix it or work around it?

Thanks in advance

Simon

 

[gr5 2,026]

Shell thicknes should ALWAYS be a multiple of nozzle - at least until you know exactly how cura works. So setting the wall thickness to 1.5 is basically the problem. There is a complicated set of instructions but in this exact case you end up with 0.5mm traces - which means 3 outer shell passes (except only 2 in your thin section). And they are spaced .5mm apart and the printer overextrudes (or tries to) by 25% to get .5mm of plastic through a .4mm nozzle (not easy to do unless you print slow and hot!).

Now if you did this nice and hot and slow it would come out fine but if you are printing at 50mm/sec, .15mm layers and anything cooler than 240C then you will likely get underextrusion - it's just hard to force that massive amount of plastic through that tiny nozzle.

So what is happening is you are underextruding and getting about .4mm wide traces (spaced .5mm apart) which means your shells aren't touching.

If you simply change the "shell width" to 1.2 (or 1.6) things should be much better off.

Also another tip - make sure you look at the part in layer view (and wait for it to completely load - there is progress at bottom sometimes hidden behind other stuff) so you can see what's going to happen at your thin walls.

 

[brousant 0]

Hi,

Thanks! I should have noticed the significance of the box being yellow... I also was not aware of the layer view.

Anyway the parts I printed so far were acceptably strong for the purpose, with the correct shell thickness it can only get better

Best regards,

Simon

 

[brousant 0]

Today, I have been happily printing parts with excellent mechanical strength.

However, here is a part where I encountered a similar issue. Despite using correct shell thickness setting now, as in the printed part, in the layer view you can see that the space between the shells (indicated by the black arrow) is left open. The fill is missing in this area as well as the top layer. In the normal view you can see that this area should have been closed.

The only changes I made to the "fast" quickprint setting were the shell thickness (which was 1mm as copied from the quickprint settings, doesn't that mean the quickprint settings are incorrect?) and the fill density.

If this is again a case of beginner's mistake, I wonder what it is!

 

[gr5 2,026]

I think you got me on this one. The gap that you point to is probably a little smaller than .4mm and the yellow diagonal infill is supposed to be .4mm traces so maybe that is part of the reason (if you make the wall 2.1mm instead of the current approx 2mm it might fill it in - you could just scale the part up as a quick test by 1.1X). Another experiment - try .4mm walls just to see what happens in this area.

But the other area nearby is skinnier and *much smaller* and it seems to have yellow infill (it's so incredibly tiny that I'm not certain it's filled in but I think it is).

 

[brousant 0]

Hi,

The problem area is 2mm wide, so 1.2mm shell walls would have overlapped. Apparently the shell thickness is locally reduced to 0.8mm. This leaves a space of 0.4 mm which I think could be filled without a problem. However it isn't.

If I move the "hole for top clamp" (see OpenSCAD code below) 0.1 mm, 0.2 mm and 0.3mm to the right, increasing the problem area width to 2.1; 2.2; and 2.3 mm, then the space between the shells does become filled, but with green line, no yellow.

If I move the "hole for top clamp" 0.4mm to the right, increasing the problem area width to 2.4 mm, then a bit of yellow fill-in/top appears.

If I move the "hole for top clamp" 0.1 or 0.2 mm to the left, reducing the problem area width to 1.9 and 1.8 mm, there is again no fill between the shell walls and no top layer.

The width of the skinny area you saw is 1 mm and there the shell wall is further reduced to 0.4 mm, leaving a space of only 0.2 mm. But in this tiny space fill-in and a top layer are present! With a width of the problem area 1.8 mm we have a space of the same 0.2 mm, but in that case no fill-in and top layer...

It looks like local reduction of shell thickness and filling/top layer are inconsistently handled and I wonder if this can be qualified as a bug.

Below is the OpenSCAD code I used to make this part.

difference(){

union (){

// base

translate([0,-12,0]) cube ([17,32,3]);

 

// support

translate([0,-12,0]) cube ([17,5,5.6]);

 

// pole

translate([0,0,0]) difference() {

translate([0,0,0]) cube ([13,8,7.2]);

 

//hole for top clamp

translate([2,1.5,2.9]) cube ([8,5,8]);

}

 

// side guide

translate([12,-12,0]) cube ([5,5,10]);

translate([12,-12,0]) cube ([5,18,7.2]);

 

// top clamp

translate([-20,-2,0]) cube ([12,10,3]);

translate([-17.9,1,0]) cube ([7.8,4.8,7.2]);

}

 

// recess for pcb break bridges

translate([5,-3.5,3]) cube ([4,4,10]);

translate([8.5,-8,5.6]) cube ([4,5,10]);

 

// screw hole 5mm

translate([10,16.5,-1]) cylinder (h=11.5,r=2.5,$fn=100);

}

 

Best regards,

Simon

 

[brousant 0]

By the way, if I set the shell thickness to 0.8 mm, in the problem area the actual shell thickness remains the same, but now fill-in and top layer appears.

But this is not an acceptable solution, it should be possible to choose a thicker shell for overall increased mechanical strength.

Best regards

Simon

 

[lars86 42]

I'm having trouble with wall thickness management as well.

I tried to print a calibration part which is a 0.4mm thick shell, that is 1" x 1" on the exterior.

The STL seems clean, but Cura refuses to generate a path. Even lying to it and entering a 0.3mm nozzle width. I get the same result.

Thoughts?

 

[gr5 2,026]

You have 2 options:

1) The worse option is to lie and tell the printer you have slightly under .2mm nozzle (say .19mm or .195mm). This should work because CURA needs to make two passes - one for the outer wall edge and one for the inner. That's just how it works - sorry. It's open source if you want to edit it but it would be massive work.

2) There is a special case for cookie cutters and cups and vases and stretchlets. You can model this part as a solid part in Cad (give it top and bottom) then in CURA you can have it print the shell only and only one pass (set shell to .4 and nozzle to .4). Do this several ways - either with "spiralize" or with "outer mesh only" right next to spiralize feature. Also disable top and bottom.

Actually I think you can also do #2 by simply removing top and bottom and turning off infill (0%).

 

[lars86 42]

I appreciate the help, but I wan't able to get it working. I only wanted a single pass ( preferably not spiralized). Even after enabling "only follow mesh surface" (which also kills your brim) and/or "spiralize", it wants to make 2 perimeters, which badly violates the model.

I got what I was after with Slic3r, but wish I could rely on Cura.

 

[visu-al 2]

for a single pass: model the stl to one solid piece (no shell) and in cura remove top/bottom and set shell to 0.4.

thats it

 

[gr5 2,026]

Lars when you tried my suggestion #2 you skipped a step which VISU-AL mentions again. It works incredibly well - it's a very special case scenario and works for this cookie-cutter like part. You don't really need spiralize or mesh.