Feature request: 2nd Layer Horizontal Expansion and independent X/Y Horizontal Expansion

[rcfocus 0]

The "elephant foot" issue is improved on my Ultimaker 3 by setting "Initial Layer Horizontal Expansion". But it is not fully resolved. I have tested several expansion values from -0.2mm to -0.4mm. By inspecting the prints in microscope, the 1st layer is fixed but the 2nd layer is still wider than expected. So, I think it would solve the problem if I can also adjust the 2nd layer expansion.

 

I also find another issue. The printed hole is always smaller than expected. So, I print a cube 10x10x10mm to verify the issue. The measured dimension is 10.2x10.12x0.97mm (excluding the elephant area). Obviously the X-Y directions are wider than expected. The error 0.2mm is too large. So, I set "Horizontal Expansion" to -0.07mm. It works well. But for this, I have two questions.

 

(1) When setting both "Horizontal Expansion" and "Initial Layer Horizontal Expansion", does Cura calculate the final 1st layer based on the adjusted "Horizontal Expansion"? Or they are independent? If they are independent then I have to compensate it myself. For example, if I want "Initial Layer Horizontal Expansion = -0.25mm" then I have to set it to "-0.25mm - 0.07mm = -0.32mm". This would be very inconvenient.

 

(2) The error of X and Y axes are very different. I have printed the cube several times and measure the average dimension. The error of one side (X or Y) is more than 0.2mm while the other side (Y or X) is around 0.1mm. So, I think it would be better if I can adjust X and Y Horizontal Expansion independently.

 

Btw, I use auto-leveling and "Fine" print setup for the above experiments.

 

Hope the above two features (options) can be added to Cura. Then Ultimaker 3 would be perfect for my applications.

 

[gr5 2,229]

All the other "initial layer" parameters override settings for the first layer and don't sum.  So you probably want -.32 in this cae.

 

When measuring the error of the part make sure the micrometer doesn't touch any corners!  Corners have their own error caused because the print head is slowing down.  In fact if you want high precision then make sure you turn off:

1) acceleration control

2) jerk control

And make sure all the printing speeds are the same (because when it switches from (for example) infill to shell if it speeds up it will underextrude for a few seconds or if it slows down it will overextrude for a few seconds.

 

vertical cylinders are indeed going to be small because the filament is like a liquid rubber band and is pulled inward.  I usually add 0.4 or 0.5mm to all my vertical cylinder holes in cad (but not to horizontal cylinders.

 

You can add a chamfer at the base of your part to compensate for all of the horizontal expansion/elephants foot.  I know it's not as easy as using initial horizontal expansion.  Sorry - that's all I have for you.

[rcfocus 0]

Thanks gr5 for comment. I will try to turn off acceleration control and jerk control.

 

I have used the "chamfer trick" for many years (since UM2). But it is really very very inconvenient for my case. I design 3D models for customers and use UM3 to verify the design. Finally customers will make the mold according to my design. I hope the design is consistent. If I add chamfer then there will be a risk that I forget to remove it before sending the model to customer.

 

In fact, chamfer is also not perfect solution. I have tried chamfer from 0.2mm to 0.8mm, 45 degree. The result is not good enough. Theoretically fillet should give better compensation but it is not suitable for printing.

 

I think an option to tune the 2nd layer expansion would be the ultimate solution.

[rcfocus 0]

I just print the cube with Acceleration Control and Jerk Control disabled, Horizontal Expansion = 0mm and Initial Layer Horizontal Expansion = -0.25mm. Unfortunately the measured X & Y error is almost the same as before. The measured dimension is 10.19x10.10x9.90mm. It looks like Acceleration Control and Jerk Control don't matter in this cube test.

 

 

Then I enable Acceleration Control and Jerk Control again, set Horizontal Expansion to -0.07mm & Initial Layer Horizontal Expansion = -0.32mm. This time I get 10.00x9.98x9.93mm (excluding the elephant foot). The result is pretty good since the error of all axes are smaller than 0.1mm.

 

So, the current Horizontal Expansion option should be enough for me. The only problem is the elephant foot. The 2nd Layer Horizontal Expansion option should completely solve the issue.

 

[gr5 2,229]

Well if the bottom two layers are so important to you then I would definitely turn off autolevel (if you have UM3 or S5) as manual leveling is more consistent.  It will not change for months once you get it dialed in.

 

Also I recommend trying a 0.1mm bottom layer.  I'm not confident this will be any better but I think it's worth experimenting on a small cube.

 

Cura defaults to a thick bottom layer (0.3mm) because the reasoning is if the glass is not flat then this is thick enough that you will get filament everywhere.  But in reality the pressure from the nozzle is usually enough to push the bed down just enough such that this compensates a little bit for non-flat glass.  Doing a bottom layer of 0.1mm works surprisingly well.

 

I know for sure that the bottom layer will come out well but I'm not sure if the second layer will be better or worse regarding "elephants foot" also called "micro brim".

[rcfocus 0]

I will try manual leveling.

 

On Cura 3.4.1, the bottom layer thickness seems calculated by "Bottom Thickness / Bottom Layers". The Fine profile already sets Bottom Thickness = 1mm and Bottom Layers = 10. So the layer thickness should be 0.1mm. Is this correct?

 

[gr5 2,229]

Ignore that.  You want "initial layer height".  Typically it defaults to 0.27mm (used to be 0.3mm).

[Dim3nsioneer 558]

@rcfocus, what material are you printing? The behavior you describe is different for different materials.

Edited July 26, 2018 by Dim3nsioneer

[rcfocus 0]

Thanks you gr5. I will try different Initial Layer Height values.

 

Hi Dim3nsioneer, the material is PLA.

[heqe95 2]

When will this be fixed? It is a common 3D-printing related issue, that should be solvable in the slicing part, not as a chamfer in the model. (3D-printer should be able to print the model out as it is modeled)
Isn't it just a copy-paste thing from the first layer HE behavior?

Please add even 2nd, 3rd etc layer HE, or, if there could be an automatic chamfer calculated from the first layer HE, even better!  I'm using values of -0.35 or more negative when the first layer is otherwise perfect and there's still a lot of elephants foot.

If someone fancies to make a working mod, I'll paypal 40USD.

[Dim3nsioneer 558]

An elephant foot is usually the result of an overextrusion or a too high temperature (and therefore a too low viscosity) on the first layer. Those should be the parameters to adjust for getting rid of it. From second layer on, no overextrusion should happen if e.g. the flow on the first layer is reduced a bit. Therefore I don't see a need for a second layer horizontal expansion adjustment for a well adjusted print profile.

[heqe95 2]

On 12/12/2021 at 8:10 PM, Dim3nsioneer said:

An elephant foot is usually the result of an overextrusion or a too high temperature (and therefore a too low viscosity) on the first layer. Those should be the parameters to adjust for getting rid of it. From second layer on, no overextrusion should happen if e.g. the flow on the first layer is reduced a bit. Therefore I don't see a need for a second layer horizontal expansion adjustment for a well adjusted print profile.

Many times first layer is visibly perfect, but still there is slight overextrusion/elephants foot in couple of the latter layers. And to get perfectly flat underside without line marks, this slight overextrusion is even needed on the first layer (I'm talking about maybe 5%) so reducing 1st layer flow doesnt really make it better overall (2nd and 3rd layer flow could maybe be reduced though to fix it)

One could maybe play with temperature settings layer by layer if maybe the heating effect of the buildplate plays a role or something, but a negative 2nd and 3rd layer HE would make everything easier and have more room for error.

Another idea, there could be an experimental "chamfer" option that applies to models facing the build plate, like there is "make overhangs printable". Aside from maybe adhesion, chamfer is always more useful and prettier than an elephants foot.

Edited December 20, 2021 by heqe95