Nozzle Diameter correction

[viennamik 0]

Hi,

What do you do for right diameter in the parts?

I see that cura have no correction of the nozzle diameter, and so diameters are to small, and other lines are to big.

Do you change the modell?

Thanks

Mike

 

[illuminarti 18]

Cura does allow you to specify the nozzle diameter that you are using, and takes that diameter into account when planning the toolpaths that it uses.

Nonetheless, overall parts diameters are going to end up slightly off, due to both mechanical imperfections and variability, and the fact that plastic contracts as it cools.

The effects depend on the particular model geometry, so typically you have to adjust the model on a case-by-case basis.

 

[gr5 2,005]

In general, everything is too small. I don't know if it's due to shrinkage or something else but holes are too small and a simple part like a cube will be smaller than desired. It's not steps/mm as that is very accurate. I think it's a combination of many factors including ringing, play/backlash, belts that can stretch, and shrinkage due to cooling. In addition I beleive that sometimes there is a pulling affect when laying down a "thread". Especially seen on first layer when there is bad adhesion to the bed that makes circles and other shapes deformed into a smaller shape then desired.

Like illuminarti says, nozzle diameter is taken into account. You can see this if you slice a 10mm cube with nozzle set to .4mm. If you look at the gcode it's very simple to see the movements are 9.6mm which takes into account the .2mm radius of the nozzle on each side of the cube.

 

[gr5 2,005]

Oh! There is one more affect. When creating a model with cad software, circles/cylinders tend to have a limited amount of sides - for example maybe 10 edges around a circle or a decagon. Each segment in the decagon is cutting through the circle a bit and this also makes holes smaller than desired and makes cylindrical parts also smaller than desired.

Also note that z dimension is much more accurate then x and y because none of the affects in my previous post affect z height (unless you print in a heated chamber).

 

[viennamik 0]

Thanks for answers :-P

I will do some tests with different diameter and outher dimensions, to get the best results.

Changes in CAD is not a big problem, i use a parametric programm.

Output via STL and 0.001mm hight of triangle is no problem for cura.

 

[Daid 303]

For "checking the dimensions of the GCode", the older Cura releases used to have a "mixed" 3D view, where you could see the GCode inside the transparent 3D model. This showed that the path was properly put down inside the model where it should be.

However, there are a few physical effects in play that are hard to account for. The first and most obvious one is shrinkage. This effects all prints in X/Y, and also effects cubes the same as cylinders.

Next there is the dragging effect, PLA wants to stick to the nozzle a bit, and thus the nozzle drags it inwards to the circle.

Finally there is an effect which I have no idea how it is called. But molten PLA wants to "blob together" this is a different effect from shrinkage as it effects small circles more then large circles.

Some say "the software should account for this!", but that's currently out of my league. There are so many different effects, so much math, and event different colors of PLA that are acting differently.

It's also not odd that you slightly have to adjust your model for the fabrication process. In injection-molding you also need to tweak the final mold till the parts are up to specifications.

My personal design rules are:

* Keep 0.2mm clearance between any moving parts or push-fit parts

* 3mm holes need to be designed 3.5mm for a hole where a M3 screw can fall trough for moving parts.

* 3mm holes need to be designed 3.2mm for fixing parts with M3 screws.

* 3mm holes need to be designed 2.9mm if you want to tab threads in it and keep the part fixed (like my fanduct)

* No overhangs larger then 60deg (where 0deg is straight up)

* 90deg overhangs can be done for very small sections (0.5mm) without any visible problems. Or up to 1mm with light visible drooping

* Bridges are king. With clever design you can print a chair upright without much issues. Check the resulting GCode if your bridges are detected properly.

 

[owen 19]

How far can it bridge generally Daid?

 

[viennamik 0]

Thank you Daid for your helpful hints.

I have test it and get some very good results.

What i also found was the shrinkage of the material:

ABS 0,4-1,0 %, and for PLA 0,6-1.2%

Shrinkage depend on the cooling too. I test it with different fan parameter, and get different shrinkage.

A very interest topic !

 

[Daid 303]

How far can it bridge generally Daid?

 

Depends on how much "slack" you can have and how well you have dialed in temperature and speed for bridging.

My record is a 120mm bridge at 0.2mm layers. (Thinner layers make bridges a bit harder)

 

[owen 19]

OK thanks for that. Didn't realize that about layer thickness making a difference. 120mm is a fair way.

 

[illuminarti 18]

I haven't done any testing, but my gut says that bridging is another area where the volume flow rate is the critical variable. Printing thicker layers and/or faster probably make for better bridges as it increases the ability to get the plastic to leave the nozzle cleanly. There's probably capillary and electrostatic/adhesive forces at the tip of the nozzle resisting the movement of plastic; at low flow rates those are able to resist the smooth flow of plastic needed for bridging, when there isn't a solid object below with its counterbalancing forces.

The downside of achieving the necessary flow rate through faster movement is the likelihood of tearing the filament bead before it hardens - leaving thicker layers as generally the best option.

[illuminarti 18]

I think this is also why sparse infill can sometimes be a messy- it's basically all bridging, if its a linear pattern on alternating layers. I get particularly good results in kisslicer when I do infill slower but fatter and for multiple layers at once. (Having a wider nozzle helps hugely with this).

In fact... If you are using linear infill that does lines roughly at right angles from one layer to the next... Shouldn't they be printed twice as fat, to allow for the fact that except at crossover points they're really needing to cover two layers of space? Does The new Cura do that?

I have to say I much prefer hex style infills that lay down the same pattern on each layer, but from different directions.

[Daid 303]

The new Cura beta does the line infill different, instead of only 1 diagonal line, it does 2 at 90 deg angles. So it does a full grid every layer. This gave better results then doubling the extrusion, which caused problems when moving back from infill to perimeter, as doubling the extrusion amount causes quite a bit of pressure buildup.

 

[illuminarti 18]

Ok yes, makes sense.

 

[Daid 303]

Oh, and on bridging, the 120mm bridge at 0.2mm layers is just what I needed for a print at some point. It was by no means an test to see where the limit was. I think you could span the whole machine with a 200mm bridge at 0.1mm layers if you get everything dialed in properly.