Page 1 of 1

Magnified view of extrusion process

PostPosted: 2013-Aug-Thu-14-Aug
by RetireeJay
I thought it might be interesting for people to see how extrusions fit together - at least theoretically.
This drawing is to scale, with a stock 0.4mm nozzle. The layer height is 0.3mm and the extruded width is 0.5mm. The flat on the nozzle is 2.6mm in diameter.
Red is for hot plastic, blue is for cool plastic.
Extrusion_04nozzle_03Layer_05Width.JPG

Re: Magnified view of extrusion process

PostPosted: 2013-Aug-Thu-14-Aug
by ahaer
I always assumed that the extruded width should be equal to the nozzle diameter...

Why would you choose a extrude width that is wider? ie what effect does it have on finish, strength, etc...

Re: Magnified view of extrusion process

PostPosted: 2013-Aug-Thu-14-Aug
by RetireeJay
Look at the header of your Slic3r generated G-code; it tells you the extrusion width that it is using. If you have told Slic3r you are using a 0.4mm nozzle and are laying down 0.3mm layer height, with width at the default setting of zero, then it will automatically give you an extrusion width of 0.42mm.

If you think of the extrusion as coming out of the nozzle at 0.4mm diameter, and then you take that cylinder, lay it on its side, and squash it down until it's 0.3mm thick, you get something close to 0.5mm width. (My geometry for that calculation is based on two semi-circles 0.3mm in diameter, plus a rectangle in the middle 0.3mm high, and wide enough to make the total area equal to the extruded 0.4mm thread. Not exactly what I drew, but when the extrusions lie side-by-side they must develop sides that are more nearly vertical.)

I think that by using a slightly wider extrusion than the default, I can get better bonding between layers (a wider cross-section and higher pressure as the plastic gets squeezed). On the other hand, maybe I'd get better surface finish if I used the default settings; I haven't explored that possibility.

Re: Magnified view of extrusion process

PostPosted: 2013-Aug-Thu-15-Aug
by ahaer
Thanks for the information - Just when I thought I understood all the parameters...

I wonder what happens if you make the extrusion width a big - say 1.0mm? From your drawing it seems that you'd still get a smooth/flat layer of the desired height - it would just be wider. The only question there is if the plastic would squish out evenly on each side.

It also might be fun to experiment with the difference between a single 1.0mm perimeter vs two 0.5mm perimeters. Since KISSlicer lets you set perimeter and infill extrusion widths separately I can keep the infill width at 0.4 for both so the part should be the same. It seems like that could make it print faster since it would only make one trip around the perimeter.

Re: Magnified view of extrusion process

PostPosted: 2013-Aug-Thu-16-Aug
by KDog
There was a post on another forum where they varied the layer height to see what all the extrusion width settings were. It appeared that Slic3r wouldn't increase the cross sectional area of the extrusion above the cross sectional area of the nozzle. For a 0.5mm nozzle this is 0.2 square mm. Maybe the printer can't control a bigger area than this for extruding?

BTW, I assume that the extrusion cross section is an ellipse. The area of an ellipse is πR1R2. So for our purposes it would be 3.14*(0.5*extrusion height)*(0.5*extrusion width). Not perfect but maybe closer than rectangular?

KDog

Re: Magnified view of extrusion process

PostPosted: 2013-Aug-Thu-20-Aug
by RetireeJay
KDog wrote:It appeared that Slic3r wouldn't increase the cross sectional area of the extrusion above the cross sectional area of the nozzle.


That's if you have it on automatic. I've been using 0.5mm, which creates a thread with slightly more cross-sectional area than the nozzle.
But if you have ever extruded caulk from a caulk gun, you know that when you expel more material than fits easily in the trail behind the gun, the caulk will build up ahead of the nozzle, and in a printer this probably would create some instability in the final width and shape of the trace after the printhead has moved on.

KDog wrote:I assume that the extrusion cross section is an ellipse.


Why would it be an ellipse? when you squeeze a balloon between two flat plates, you get two flat surfaces, plus the rounded shape where there is no constraint. Here is how I calculated the shape of a trace in isolation.

Extrusion shape idea.JPG


However, in actual printing, your thread is very often butted up against a neighbor; that is why I showed the semi-rectangular shape in the drawing. I should edit the drawing to show the un-constrained faces to have a semi-circular shape.

EDIT: Now that I look at it, I'm wondering... the top of the extruded material can't be narrower than the nozzle, can it?
We need someone with access to good facilities to cast an object in resin, then cut it and polish it and take a picture with a high-power microscope!
In the meantime, just consider the illustration at the top of the thread to be a rough conceptual idea. The width of the nozzle, the width of the flat on the nozzle, and the thickness and width of the traces are all to scale, but the detailed shape of the traces is still unknown.

EDIT 2: SEE NEXT POST. Slic3r treats the extruded material as having a rectangular cross-section.

Re: Magnified view of extrusion process

PostPosted: 2013-Aug-Fri-17-Aug
by RetireeJay
OK, here's the deal. I created 14 different Slic3r models of a 10mm square in vase mode (no top, bottom, or infill; no raft, no brim, no skirt, one perimeter). All settings affecting trace width were on automatic; the Extrusion Multiplier was 1.0. I was mainly using 0.4mm nozzle and 3.0mm filament. All this was theoretical; no actual plastic was melted in the conduct of this study. ;) This gave me small and simple G-code that I could read and analyze. I found that the upper layers differed from layer 0, so all my calculations are based on layer 1.

Slic3r treats the extruded plastic as having a rectangular shape. The width of the rectangular extrusion is taken into account when Slic3r calculates the path of the centerline of the extruder, so that the outside dimension of the object is precisely equal to the model. Interpreting the data this way is correct to 3 decimal places.

When possible, Slic3r tries to extrude exactly the right amount of material to have a cross-sectional area equal to the area of the nozzle opening. So that means as the layer height is decreased, the width of the trace increases; as the layer height increases, the width of the trace decreases.

EDIT: Apparently they changed the algorithm in Slic3r1.0 Now it appears that the trace width is exactly the nozzle width (at least, in my case of a 0.4mm nozzle and 0.25mm layer height).

However, there are two other rules that come into play:
1) If the layer height is tall, for example equal to the nozzle width, then extra material is extruded to keep the trace width equal to the nozzle diameter plus 4%. I.e. we never extrude a trace that's narrower (or even equal) to the nozzle diameter.
2) If the layer height is short, then the width of the trace is limited to the nozzle diameter plus about 65%. The cross-sectional area of the trace is decreased as necessary to obey this rule.

With a 0.4mm nozzle, the range of layer heights with constant extruded area is about 0.2 to 0.3mm. This corresponds to trace width of 0.628 and 0.419 respectively.

Re: Magnified view of extrusion process

PostPosted: 2013-Aug-Mon-14-Aug
by pomonabill221
RetireeJay wrote:OK, here's the deal. I created 14 different Slic3r models of a 10mm square in vase mode (no top, bottom, or infill; no raft, no brim, no skirt, one perimeter).

Great explanation on how slic3r creates and calculates extrusion/height/width/position! WOW alot goes into it! Impressive!

How did you create the square in "vase mode"?
I am trying to figure out how to create a box with only a bottom and 4 sides, NOT filled and so far, I have only created a cube (filled) using sketchup and freecad (preferred).
I know it MUST be simple, but I am probably "trying too hard" and missing something very basic!
Thanks
Bill

Re: Magnified view of extrusion process

PostPosted: 2013-Aug-Mon-14-Aug
by RetireeJay
Here's another example of why it's best to run Slic3r on its own to create G-code files, and then load the G-code files into Repetier in a separate step.

In the latest version of Slic3r - 9.10 - there is a checkbox on the "Print Settings" tab called "Spiral Vase." But even in older versions of Slic3r, all you need to do is specify Infill "Fill Density" of zero, and Vertical Shells of 1 layer. Specify Horizontal Shells with Top set to 0 and Bottom set to 0 or some other value as you wish. The net effect of all this is to hollow out your solid model and print the exterior of your model with a single thickness.

It's possible to do this "within" Repetier, of course, but you need to save all your settings with a name that is meaningful to you, and then select those modified settings on Repetier's slicer screen before you slice. The allowed name length that is visible to you is pretty short, so it's hard to keep track if you create lots of different configurations for specific purposes. When you run Slic3r on its own at least you can see exactly what options are in effect when you do the slicing.

Re: Magnified view of extrusion process

PostPosted: 2013-Aug-Mon-15-Aug
by KDog
I still have it on my list to take some photos of this on the microscope. Just haven't had the time.

Re: Magnified view of extrusion process

PostPosted: 2013-Aug-Tue-01-Aug
by pomonabill221
RetireeJay wrote:Here's another example of why it's best to run Slic3r on its own to create G-code files, and then load the G-code files into Repetier in a separate step.

In the latest version of Slic3r - 9.10 - there is a checkbox on the "Print Settings" tab called "Spiral Vase." But even in older versions of Slic3r, all you need to do is specify Infill "Fill Density" of zero, and Vertical Shells of 1 layer. Specify Horizontal Shells with Top set to 0 and Bottom set to 0 or some other value as you wish. The net effect of all this is to hollow out your solid model and print the exterior of your model with a single thickness.

It's possible to do this "within" Repetier, of course, but you need to save all your settings with a name that is meaningful to you, and then select those modified settings on Repetier's slicer screen before you slice. The allowed name length that is visible to you is pretty short, so it's hard to keep track if you create lots of different configurations for specific purposes. When you run Slic3r on its own at least you can see exactly what options are in effect when you do the slicing.

Will have to give this a try!
I thought that I could create the part in freecad/sketchup by creating a "square" and extruding the edges for the walls, but when I tried that, the "box" didn't have a bottom and slicer didn't like it.
Think I did something wrong so will have to try again...
Thanks!
Bill

Re: Magnified view of extrusion process

PostPosted: 2013-Sep-Wed-15-Sep
by rho-wan
I tested Slic3r with extrusion 1 or 0.8 and the output text always gives me 0.43 mm as extrusion width fpr 0.29xx mm layer thickness.
If I play with the multiplier I will probably have exact inner and outer size for my printed object (as RetireeJay wrote elsewhere), but I wonder whether adhesion between adjacent threads on the printing plane will be good, or they will not adhere on their sides well because they are now thinner even if the expected extrusion width is expected to be the original one.

Have you made tests on this regard too?

Re: Magnified view of extrusion process

PostPosted: 2013-Sep-Wed-17-Sep
by RetireeJay
Well, yes if you turn down the extrusion factor too far, you get vertical walls that are not adhered laterally. Happened to me two days ago. :(

Re: Magnified view of extrusion process

PostPosted: 2014-Apr-Mon-17-Apr
by RetireeJay
Apparently they changed the algorithm in Slic3r1.0.0RC3. Now I'm seeing a trace width exactly equal to my nozzle diameter (0.4mm, with layer height 0.25mm). Again, this is based on assuming that the trace itself has an exactly rectangular cross-section, 0.4mm X 0.25mm.

Re: Magnified view of extrusion process

PostPosted: 2014-Jun-Fri-19-Jun
by rho-wan
Good, in other words, when there is no neighbouring thread, the actual thread will be wider due to the shape shown in viewtopic.php?p=29560#p29560
This means that calibrating printer extrusion using single-walled walls is not reliable and, if done that way, the adhesion between threads would be in real prints worse than it should be.
Am I correct?

Re: Magnified view of extrusion process

PostPosted: 2014-Jun-Fri-19-Jun
by thawkins
rho-wan wrote:Good, in other words, when there is no neighbouring thread, the actual thread will be wider due to the shape shown in viewtopic.php?p=29560#p29560
This means that calibrating printer extrusion using single-walled walls is not reliable and, if done that way, the adhesion between threads would be in real prints worse than it should be.
Am I correct?


also in the discussions above, i saw no mention of the "die swell" effect, if you push hot plastic under pressure down a tube, what comes out the end is bigger than the tube diameter as the plastic expands as the pressure is relived.

http://en.m.wikipedia.org/wiki/Die_swell

Re: Magnified view of extrusion process

PostPosted: 2014-Jun-Fri-20-Jun
by RetireeJay
My former employer was a tire company, and they routinely extrude rubber through dies to prepare different parts of the tire assembly. So I learned something about die swell.

The wikipedia article is mostly correct, but there are some subtleties that they did not make clear. Polymer molecules aren't little spheres; it's more like a tangled mat of fibers.

For practical purposes, we can assume that the volume of material entering the die is exactly equal to the volume of material exiting the die. And the volume exiting the die is exactly equal to the volume several seconds later after it has undergone the "die swell;" the volume did not change.

The die squeezes the plastic down to a certain cross-sectional area when it exits the die. For the most part, the shape change going through the die is permanent, but not completely. There's a little bit of "spring tension" left in the polymer molecules when they exit the die. What we call the "die swell" is caused when the polymer molecules relax and relieve that spring tension. The length of the extruded thread gets shorter while the width gets wider.

This all applies to the un-constrained case where the extruded material is not touching anything (extruding into air). When the extruded material touches and bonds to something, then that something (a print bed, or a previous layer) constrains the thread not to shrink (as much) in length and therefore reduces the "die swell."

BTW, this shortening of the thread after exiting the die explains why circular holes tend to print undersized.

The trace width that Slic3r uses for calculation - even of the outermost layer - is based on a perfectly rectangular geometry. This could be true for inner layers, but we know it's not true for the outside layer; the ridges for each layer are clearly visible. So the thickness of a single-trace wall will certainly measure more than what Slic3r is claiming.

Re: Magnified view of extrusion process

PostPosted: 2014-Jun-Sat-04-Jun
by rho-wan
Perfect explanation, thanks.