how slicing is calculating the speed and feed rate

[almal 0]

hello

I am an ultimaker 1 user and I am working on a new way of 3d printing on a robot arm support. I am looking for some information about the relation between the size of the noze the amount of plastic extrude, the speed , the retraction and the distance between the layer.

how I can define some strategy to slice some model with different plane not just z axe.

My biggest problem is to understand how much plastic I had to extrude to fill a specific amount.

If you had any mathematical formula or test or files I can read to understand this I would be very happy

thanks and sorry for my poor english

 

[rooiejoris 5]

Well since 2 years somebody came up with volumetric extrution. It means that you will know the volume of the output, so you can calculate the volume of the input.

output volume is:

length of the path x width of the path x layerheight [or if you want to be more precise the section is something with 2 half circles and a square x length of the path

input volume is:

input length x filament thickness x something with pi

the only unknown is the input length so that is [pretty easy] to calculate

should be something like:

input lenght = (length of the path x width of the path x layerheight) / (filament thickness x something with pi)

is this what you mean...?!

cheers joris

 

[donmilne 2]

You'll need a time component as well. I.e. the printer will execute the task in a given period of time. Hence the feeder needs to be able to deliver that volume of material in the same time or better.

It's a subject that's been interesting me of late. Problems such as underextrusion are caused by a disparity in this timing (there may be underlying hardware issues that exaggerate the disparity). The normal way to deal with overly tight sync constraints is to have some kind of buffer between the supplying and consuming processes. Perhaps a reservoir of some sort which is kept topped up regardless of the current consumption rate. Of course a larger reservoir would make filament color changes a bigger deal.

 

[almal 0]

Yes this is this type of information i am looking for

I am also looking for the amount of material I need to extrude when I flatten some new material on a layer already extrude for exemple if i want to extrude 1 mm height or 0.5 mm height is this is going to change width of my extrusion.

when you print a layer, the extrusion is flatten on the older layer. I don't really understand how to calculate the shape the extrusion is going to take when you flatten it on a new surface. It seems to be different than filament on top of other filament (the extrusion is not round but more some kind of oval)

 

[almal 0]

You'll need a time component as well. I.e. the printer will execute the task in a given period of time. Hence the feeder needs to be able to deliver that volume of material in the same time or better.

It's a subject that's been interesting me of late. Problems such as underextrusion are caused by a disparity in this timing (there may be underlying hardware issues that exaggerate the disparity). The normal way to deal with overly tight sync constraints is to have some kind of buffer between the supplying and consuming processes. Perhaps a reservoir of some sort which is kept topped up regardless of the current consumption rate. Of course a larger reservoir would make filament color changes a bigger deal.

 

what I understand of the gcode is that he give an extrusion lenght , a speed and two point in the space and the goal of the machine is to extrude the lenght and to move from one point to the other with the speed define. So I think I am going to be able to synchronize the movement with the extrusion ? Am I wrong ?

 

[Dim3nsioneer 557]

The length of a print step, the nozzle size and the layer height give you the volume of the needed filament. With the standard RepRap flavour the E-component of a G1 command has to represent the length of filament used for extruding this volume. So you have to divide the volume by the cross-section of your filament.

If you use the volumetric flavour (implemented in Cura as a alternative RepRap flavour since not so long ago and for a bit longer as UltiGCode for the UM2), the E-component of a G1 command directly represents the volume.

 

[almal 0]

if I am right I find this formula

(Size of the filament X length of filament input) / Size of the nozzle = length of filament output

 

[almal 0]

The length of a print step, the nozzle size and the layer height give you the volume of the needed filament. With the standard RepRap flavour the E-component of a G1 command has to represent the length of filament used for extruding this volume. So you have to divide the volume by the cross-section of your filament.

If you use the volumetric flavour (implemented in Cura as a alternative RepRap flavour since not so long ago and for a bit longer as UltiGCode for the UM2), the E-component of a G1 command directly represents the volume.

 

the problem is I am not working with traditional layers. my layers could be 1 mm at one point and 2 mm at an other with a variation of extrusion between those two point. I searching for a new way of calcul without using some plug in. I am more looking for mathematical calcul the slicing software are using

thanks every one for thoose information

 

[Dim3nsioneer 557]

You may check the procedure yourself: https://github.com/Ultimaker/CuraEngine/blob/master/src/gcodeExport.cpp#L103

 

[almal 0]

thanks I am going to have a closer look to this

 

[gr5 2,192]

how I can define some strategy to slice some model with different plane not just z axe.

 

What? Cool!

Well I usually just tilt the part. Right now Cura only does Z slices. This is more accurate because the Z has no play because it always goes one direction. However it *is* possible to tilt by a little - maybe up to 20 degrees? The nozzle has a flat surface around the hole so you can't slice more than about 20 degrees tilted.

Joris did a TRUE 3D print using a cura plugin he wrote:

http://www.thingiverse.com/thing:75735

You can see in the above cup his max tilt was about 20 degrees. If your part is simple you could combine a plugin with tilting the part and get true angled slices.

 

[Dim3nsioneer 557]

I guess the idea is not only to tilt the part once and slice it but to slice different parts of the same object with different z orientations. One could call this 6D-printing as three rotational degree of freedom are added (just wanted to mention the name before a Brooklyn based company issues a trade mark and/or patent on it). It might be the solution to any overhang issues as actually there are no overhangs left.

 

[almal 0]

Yes

[gr5 2,192]

It's - brilliant - if I understand correctly. But realize this only improves the angle by about 20 degrees. So you can make a 45 degree overhang, printed at a 20 degree angle print as good quality as a 25 degree (from vertical) over hang. Wow - that's actually an amazing idea.

and you could print an absolutely straight overhang shelf at 90 degrees from vertical by slicing it at 20 degree angle and it will print with the quality of a 70 degree (from vertical) overhang. Wow- this might actually work. You could even have the left half of the model sliced at one angle and the right half at another angle.

I don't know if this really helps print quality because the nozzle tip is still flat. But it should help as the new layer should be pretty well connected to the previous layer. This is very intriguing. It would help if the nozzle itself could tilt up to 90 degrees from vertical! Then it would be a 5 axis print head! (X,Y,Z, roll, altitude (no yaw needed)).

I'll put this on the end of my somewhat long list of tests I want to do.

 

[Dim3nsioneer 557]

gr5, you still think too limited...

It's actually a 6-axis-printer. With the right robot arm you can print in a arbitrary layer direction at a arbitrary position within a volume.

Imagine, the initial 'print bed' could be something like a sphere holded by a very thin strut or similar. Then the robot arm with the nozzle at its end could print onto that sphere and create a print of e.g. the shape of the spherical head of a dandelion, i.e. something which is almost impossible today. Of course simplier prints could use simplier forms of 'print bed'. Maybe we will call it rather 'print seed' than 'print bed'.

The really ugly part of such a development would certainly be the slicing algorithm for such forms. For the above mentioned print, slicing would happen in spherical shells instead of layers. From a mathematical point of view it doesn't matter. It's just a different object. I hope, Daid will survive if he reads the last two sentences...

@almal: Sorry for hijacking your idea, but it is just too cool not to think about where this might lead... B)

 

[Daid 306]

The really ugly part of such a development would certainly be the slicing algorithm for such forms. For the above mentioned print, slicing would happen in spherical shells instead of layers. From a mathematical point of view it doesn't matter. It's just a different object. I hope, Daid will survive if he reads the last two sentences...

 

No problem. I live.

Math and programming are 2 different fields for a reason. Some stuff is hard in math and easy in programming. Some stuff is easy in math and hard in programming.

 

[gr5 2,192]

If the nozzle hole is round (symmetrical) then I don't see a need for a 3rd rotational axis. Two should be plenty so I still stand by 5 axis. However if the nozzle hole is rectangular - then you could rotate the nozzle hole for different orientations when printing different features. The wide length for low res fast printing, the short way for higher resolution slower printing. Like a calligraphy pen. But I don't think that is reasonable. I think it's better to use a round nozzle.

 

[Dim3nsioneer 557]

If the nozzle hole is round (symmetrical) then I don't see a need for a 3rd rotational axis. Two should be plenty so I still stand by 5 axis. However if the nozzle hole is rectangular - then you could rotate the nozzle hole for different orientations when printing different features. The wide length for low res fast printing, the short way for higher resolution slower printing. Like a calligraphy pen. But I don't think that is reasonable. I think it's better to use a round nozzle.

 

Well, of course you're right. 5 axis should be enough for a round nozzle. However, I was thinking of a revolving toolhead or something similar (but it's not a full axis...).

 

[almal 0]

gr5, you still think too limited...

It's actually a 6-axis-printer. With the right robot arm you can print in a arbitrary layer direction at a arbitrary position within a volume.

Imagine, the initial 'print bed' could be something like a sphere holded by a very thin strut or similar. Then the robot arm with the nozzle at its end could print onto that sphere and create a print of e.g. the shape of the spherical head of a dandelion, i.e. something which is almost impossible today. Of course simplier prints could use simplier forms of 'print bed'. Maybe we will call it rather 'print seed' than 'print bed'.

The really ugly part of such a development would certainly be the slicing algorithm for such forms. For the above mentioned print, slicing would happen in spherical shells instead of layers. From a mathematical point of view it doesn't matter. It's just a different object. I hope, Daid will survive if he reads the last two sentences...

@almal: Sorry for hijacking your idea, but it is just too cool not to think about where this might lead... B)

 

No problem you have exactly the same idea as me

now I have to devlope the slicing not the easy part