Question How to slice a floating object

[PDDXPrinting 0]

Hey everyone.  I'm experimenting with using supports that were made before an item is printed that are fixed and bonded to the bed.

I'm trying to slice a floating object and it gives 0 minutes and 0 material used.

How do I get it to print the floating part and put in the print and travel moves for a non-touching object?

And obviously don't want it traveling through the non-seen support.

 

I've got Don't drop the object to the build plate and Remove empty first layers on.

Thanks

[GregValiant 1,355]

DISCLAIMER:  This is a stone cold guess.

Move it up so it's completely off the build plate.  Call it 5mm for this example.

Turn on Support and make the Minimum Support Area = 1mm².  Turn off Support.

Bring in a support blocker and size it to 1 x 1 x 5mm (the example height).

In Per Model settings change it so it Prints as Support.

Move it to support that lowest point of the model.  (EDIT:  It appears it doesn't matter where on the build plate the blocker goes - it should still keep the model floating)>

In Special Modes turn on Relative Extrusion.

Slice and create the Gcode.

Open the Gcode in a text editor.  The layers that involve the blocker will be first.  Delete all of them up to the point where the model starts to print.

 

Here we have a Benchy near the flag of the 18th hole.  In this example Layers 0 through 199 are all support and so deleting them leaves the Benchy floating in air.  Your initial retraction and prime need to be managed to make sure the filament is where it needs to be in relation to the nozzle when the first extrusion goes down.

You will need an initial Z move up higher before moving over and dropping to the working height.

Being in Relative Extrusion mode makes it easier to start at any point in the file.

 

 

 

I'm kind of curious how you plan to insure that the print head doesn't hit your fixed cradle thing that will be supporting your model.  What about adhesion?

 

As a personal note...it seems like a lot of work to go through to make a golf tee.

 

 

Edited April 26, 2022 by GregValiant

[ahoeben 1,990]

3 hours ago, PDDXPrinting said:

I'm trying to slice a floating object and it gives 0 minutes and 0 material used.

It is always hard to know for sure from just a single screenshot, but the 0 minutes 0 material used thing has nothing to do with that the model "floats" above the buildplate. From the looks of it the model is hollow and has very thin walls, probably thinner than the nozzle diameter. Parts that are thinner than the nozzle diameter will not print in Cura 4.x.

 

Try increasing the Horizontal Expansion, or try slicing with the Cura 5 beta which changes how thin walls are handled.

[PDDXPrinting 0]

Haha thanks Greg. 

I was thinking about deleting supports below in gcode but didn't want to have to create a small object below the tee. The adding exclusion and changing to support works well. And I forgot about the relative extrusion that's great catch.

And I couldn't get it to work without reducing minimum support area and other settings so good instructions. It kept dropping to the plate anyway.

 

Ideally I'd also like to figure out a way that I can do this and have it slice without going in and deleting g code lines every time. 

Because now with this I can do it and it slices the floating without being connected to the support touching the build plate. 

 

If there's a setting that will just allow this without having to have the support there separate touching the build plate, that would be very good.

 

Or making the support height too small to slice or width to small to make a path but Cura still thinks it's meeting the values and slices the floating model.  

So far I've got the support height at .1mm and it won't let me go shorter and object at .3 with layer .2 and it still has the first layer.

Oh shoot I just got it to not print the support by turning off adaptive layers. That's great.

 

But I'd still like to have adaptive layers for more detail and smoothness on the almost horizontal parts, so if anyone still can know how. And would like to not have to add and shrink support every time. But this is great in the mean time.

 

Yeah pretty frivolous for just a golf tee, but I'm trying to experiment so I can get it going for a lot of things to print anything with a round or small surface area bottom to keep it from giving a flat spot or slowing down all the way. Then we could print any sphere shape or rounded object without compromise or splitting.

Have a pre made aluminum base for lots of prints of the same part.

Or print a base support out of plastic that melts at higher temperature, like petg or pc, then print pla part on top to stick but not bond to the support base.

 

Yeah it's hard to tell with just a picture. I was able to slice with it touching the plate and it filled and was solid. I'll put the file below.

G_Tee_T2-2.0_tee float.3mf G_Tee_T2-2.0_tee float w box.3mf

Edited May 4, 2022 by PDDXPrinting
dropping to plate

[PDDXPrinting 0]

I haven't yet tried in Cura 5.0 so we'd have to give it a go.  Hope it still works or other good addition.

 

And it slices with layers biased more to the bottom and going past mesh there, and not all the way to the top of the mesh for layers on top.

I've got slicing tolerance to middle.

 

Does anyone know if there's another way to change this or know how much it is and how to get it to slice mostly in the middle of the mesh outline top and bottom?

I'd like to know as much as possible to plan for squish or it running into the support base or gap in between.

It seems inconsistent and depends on layer height and mesh height when I move it up and down and re slice, so just built in software.

 

 

Edited May 4, 2022 by PDDXPrinting

[ahoeben 1,990]

Have you tried changing the Slicing Tolerance setting?

[GregValiant 1,355]

If I understand what you mean by a "bias" - isn't that the triangle cross section creating that?  It's an orthogonal view of a polygon with an odd number of sides so it's only symmetrical about one axis.  When viewed from the second or third axis it looks "off" unless the view is exactly square to a "point" of the triangle.

Since it is a triangle and since the view is orthogonal then the distance from the point to the top of the rib is X distance and the dimension from the point to the bottom is sin(30) * X distance = 1/2 X.  The bottom "half" of the model will only have half the layers of the top half (in this limited view).

Here is a side view with the bottom model rotated 30° about the X.

 

That will certainly slice differently than the top model because the orientation is different.  In addition, the top rib in your orientation is viewed straight on and so the top "edge" will be a radius.  The other two ribs are at 30° angles to the view and so their edges are ellipses and not radii.  Again, that will produce different slicing.  A round Tee would not have the problem since no matter how you view it from the side, the top edge and the bottom edge would always be equidistant from the point rather than some trig function introduced by the unequal number of sides.

So I don't see any bias, just Cura slicing the part as you presented it.

 

 

Here are those same models but viewed from the top.  They appear to have switched but it's just a question of orientation of the parts when sliced.

 

Either that or once again I've over-thought something and I'm totally wrong.  More coffee might help.

Edited May 4, 2022 by GregValiant

[PDDXPrinting 0]

Yeah I checked and have Slicing Tolerance set to middle.

 

Yeah I saw the asymmetry in the side view from the kind of triangular legs, I was talking about the displayed gcode path going outside and inside the mesh displayed.
I can see the pictures lower resolution when posted so you can't really see it.

 

So I'll post more close up I see the g code paths go out pretty far on the bottom of the mesh outline, and not quite all the way to the side and mesh outline in the top.
So I'm wondering if there is a middle slicing tolerance for this as well or how much it's tolerance is to keep from running into any base part printing on.

 

You can see it from the side, and this is with 0.2mm layer height so not crazy small.

G_Tee_T2-2.0_tee 13m .2l.gcode

[PDDXPrinting 0]

Even when I drop it down to the base

[GregValiant 1,355]

I haven't had a good Sunday Morning Rant in a while.  I don't know if this will qualify as a rant and I certainly can't tell if it's "good" but I'm pretty certain that today is Sunday.  Here I go.

 

A lot goes on in the little space below the nozzle as it extrudes.

Zooming in that close I think the Cura Preview becomes a "cartoon".  That it loses the exactness of a mechanical drawing (if it ever had it) and becomes more of an approximation or an "idea" or simply a "guess".

 

The extrusion can be drawn in a side orthographic view as some sort of rectangle .4 wide by .16 high but you can see in your images that Cura is "representing" them as some pointy sided shape.  That shape likely isn't true.  A radius? Maybe...but absolutely not a point.  Then there is "how should it be represented" vs "what is possible in a representation" given GPU/CPU/video-display hardware and the software constraints.

 

So how should it truly be represented?  In the plan view what sort of resolution should we give it?  In a side view how accurately can we represent it.  When we orbit the view how do we change it?  Heck, what is the actual shape of the plastic as it comes out of a round hole and is squished into some sort of rectangular oblong sort of thing whose "oblongness" is dependent on the speed of the nozzle?

I've used rectangles in this "representation" to create an infinitely thin vertical slice through the middle of the part.

 

 

On the left is the side view of the actual gcode provided by @PDDXPrinting.  When you zoom in to the right - (you have to believe me when I say that) I've drawn the magenta colored spline at the tip of each layer at the points given in the gcode file.  Then I put .4 x .16 rectangles hanging from each of those points.  The cyan spline is drawn through the corners of the rectangles at what approximates the outside surface of the actual print - i.e. how it would look to your eyes when held in your hand.  You can see that I've drawn the cyan spline through the bottom right corner of the rectangles below the tip of the part, and through the upper right corners of the rectangles above the tip of the part.

I figure that somewhere between the cyan spline and the magenta spline reality exists.  Exactly where in that area I don't know.  Do the extrusions of the outer wall actually have square sides?  Probably not.  But I doubt that they have pointy sides as either.  Most likely the unconfined-by-an-adjacent-extrusion outer edge of the outer walls is a radius.  I was too lazy to add radii but a spline through the tangent points of all those radii would end up between the magenta and cyan splines.

 

So I think the best Cura can do as a representation is an approximation.  In CAD software I would expect better.  In Cura the generated Gcode path is what it is.  We know where the theoretical point that is at the intersection of the centerline of the nozzle with the plane that is the end of the nozzle is going to be at any given point in time.  But the shape of the plastic under the nozzle is dependent on a lot of different things and the more you zoom in the more guessing is involved.  Educated guesses to be sure, but guesses none the less.

 

It's also possible that all of that is BS and that I've had too much coffee for my own good.  I think it's time for more.

[GregValiant 1,355]

By the way...turning off "Remove empty first layers" allows you to slice a floating object.

 

 

 

 

[PDDXPrinting 0]

I tried with turning off "Remove empty first layers" but it just brought the object down to the plate

 

I got it to work only when Remove empty first layers was off, no support, and Use adaptive layers was off. Otherwise it would still drop it to the plate.

 

I still can't get it to work without turning Adaptive Layers off. Without support and deleting in gcode.

 

 

With the gcode volume path visualizer it makes it a vertically mirrored pointed shape, from an orthographic view, but the path still goes past the mesh outline below and not to it above.  Even though it follows the shape and a rectangular cross section view of an extrusion would also represent it, it's still different on each side.

 

I found turning on Adaptive Layers shows that it slices outside and not to the stl mesh part skin, as shown in these.

But turning off Adaptive Layers makes it go right up to the stl mesh skin. No gaps to the perimeter. Shown here.

G_Tee_T2-2.0_tee float 13m Use Adaptive Layers.gcode G_Tee_T2-2.0_tee float tpu 10m 5w 30p 0.2L 220 30-15-22ow e3.gcode

[GregValiant 1,355]

That's a nice analysis.  In the end though, it looks like we're talking about a difference of what...a few microns?  It's fine to look at a theoretically perfect representation (is it really?) but at some point the real world is going to intrude.

If instead of pointy ends the representation of those extrusions were drawn as ovals (which is a lot more likely to be true) would you draw the same conclusions?

Starting with the extruder - will it in all cases be perfectly adjusted to a 1:1 "volume-in : volume-out" ratio?

The actual size of the nozzle will vary with the size of the drills that made the hole in the nozzle and the machining tolerance of the lathe, automatic screw machine, or other gizmo that made the hole.

The actual diameter of the filament.

The moisture content of the filament.

The temperature of the material as it extrudes.

The shrink factor of the material.

The efficiency of the layer cooling.

The rigidity of the printer in holding it's path and the ability of the mainboard to explain that path to the steppers.

The elasticity of the material as it is drawn around the corner.

How close the visualization is to reality.  Did the software designers nail it(?) or more likely, is there some possible error involved between the interpretation of the image in the software and the actual display?

Can we assume the blue line that indicates the edge of the part is perfect?

Can we assume that the pointy edges of the extrusions (which will definitely not be pointy edges) are also perfect and are placed exactly where the unconstrained-after-leaving-the-nozzle hot plastic is going to end up?

 

I think it's more likely that those extrusion ends should have some sort of tolerance drawn in to indicate a likely "minimum" and the likely "maximum"?  I think there will be some fuzziness in there and that it isn't being shown.  Again, what I'm seeing looks like a few microns.  My printer is pretty dialed in and I know it wouldn't be that consistent from print to print.

By increasing the flow by maybe 1% would the ends of those extrusions move outward?  In theory they should.  I can also move them using the Horizontal Expansion setting, Outer Wall Inset, and other settings in Cura.  Those are things a user can control.  I think there are things a user cannot control and mostly they are at the production end.

 

In the world of sheet metal - it's why tools have shims for all three axes.  All the software along the way may have been perfect, but the fender that comes out of the press will not be.  This is why we always referred to our technical drawings as "cartoons".  What we designed and what we ended up with were rarely exactly the same.  That is how I view that representation - as a cartoon.

 

Speaking of tolerances and as an aside - software has no concept of scale.  If you tell Cura (or AutoCad, SolidWorks, Catia, etc.) that your object model is 200 x 200 x 200 the software doesn't know if you mean nanometers or parsecs.  It's going to display whatever it's displaying at it's internal maximum resolution as it plays with the numbers.  In AutoCad that resolution was to 12 decimal places but the display was to 8 decimal places.  The display was good, but there was a tolerance if you zoomed in close enough.

 

In closing...I may place the ball perfectly on that "perfect" tee but there is an excellent chance that the ball is still going to end up in the lake and the tee will be in two pieces.