fritzw
commented
22 hours ago This is a nice analysis, and I think it would make a good research paper on 3D printing techniques, or maybe an article on hackaday.com or a similar outlet. You should definitely submit it somewhere where it may get more attention, because those are some nice results. (Also, your English is not really that bad.)
However, as a feature request, I think it would require more detailed specifications on what exactly should be implemented, since this is no small undertaking. That also requires more testing beforehand, with more complex models than a simple cube.
If you want to do more experiments yourself, using more complex models than cubes, you could try one of the following:
- Mesh Mixer can create a solid shell from a mesh, as far as I know.
- Blender has the "solidify" modifier, which also makes a shell from a given STL. You can then export as STL with "apply modifiers" checked. This might not work on all models, specially ones with broken geomerty.
You could then load the generated shell STL into the slicer and try your infill-only method on that. However, rectilinear infill is probably not a good fit because it will still cause long straight lines in some orientations. Maybe "Hilbert Cuve" infill would be more suitable.
Is there an existing issue for this feature request?
Is your feature request related to a problem?
I have the simplest open-case model of the Rep-rap printer Voxelab Aquila C2 (analogue of Creality Ender 3 Pro) and carried out all the experiments on it. I was specifically targeting such model, knowing that I would modify it, and for me no need more for my home needs. Well, ABS plastic has attracted me since childhood, I gave it preference in my crafts as it is visually beautiful and easy to process. The story began when I was looking parameters for ideal printing for ABS plastic. At first I encountered the problem of layers delamination during printing, the model being torn off the bed, and unstable sintering strength of layers. I dealt with the first two problems by slicer settings for small models, but the visual quality always led to a deterioration in strength.
Here is an example of my printer's print:
The "ideal" cube with 2 cm at each wall...
I thought about how to solve this problem in such a situation by hardening (tempering, cementing…) the entire model. Since there was no special equipment, I decided to proceed from the existing one, namely, heat the printed model directly on the bed, covering it with a specially made foil cap. The result was something similar to a baking chamber. All samples were kept under these conditions for 1 hour.
Fig. test models under a foil cover on the hot bed.
Initially, it was thought that heating to a temperature of 140°C (the maximum for my printer) would soften the plastic, and due to its adhesive properties, it would bind the layers more strongly. Yes, this particular process took place, but another negative one was revealed. It turns out that the model began to shrink along the X and Y axes, and, conversely, stretch along the Z axis. This curiosity greatly puzzled me, since if there were temperature shrinkage, then the model would shrink (or rather, expand) evenly as the temperature increases!
Sample after hardening and its dimensions
As you can see, the compression along the XY axes was 15%, and the expansion along the Z axis was 12%. It must be said that almost all tested samples at the end of printing (before tempering) had the same ideal geometric dimensions. They received deformation when heated on the table, when the plastic became soft like TPU, and internal stresses distorted the shape. Having thought that completely different principes were involved here than temperature deformation, I began to figure out which ones. To do this, I began to print models with changes in basic parameters, such as flow, wall and layer thickness, airflow, nature and volume of filling, bed and extruder temperature… Having printed a dozen models, I noticed that the most effective parameter was “layer thickness”. Moreover, if, when printing with a 0.4 nozzle and having the same 0.4 layer, the model shrinks 70% less than with the standard “popular” height of 0.2. The second factor in reducing this effect is the replacement of linear printing of wall with diffuse one. I applied the "Fuzzy skin" slicer setting to all the walls (the far-right cube of the first picture).
Tempered samples compared to the "ideal" cube. The further to the right, the selected parameters gave the best result.
Reflecting on the results obtained, I came to the conclusion that the stress that leads to delamination is not inherent in the plastic itself and its physical properties, but is formed during the printing process! Since we are using a layer smaller than the cross-section of the nozzle, it turns out that we first compress the plastic vertically and then stretch it horizontally. And rapid cooling fixes this internal tension. In fact, throughout the entire volume of the model, an ordered integral spring is obtained, which will ultimately spoil the model when printing and weaken it during use!
This conditionally shows the physical process of accumulation of internal tension in plastic. The volume of plastic can be represented as a «set of springs» consisting of polymer chains. Let us consider only those that form vertical and horizontal tensions.
When the flow is compressed, the vertical “springs” are also compressed, but the horizontal ones are stretched. Moreover, they are “fixed” from below on the cooled layer, not in the same horizontal position, but at a slight upward angle in the direction of the nozzle movement. This force determines the fact that delamination occurs with the ends of the model lifting upward.
Then it was only necessary to figure out how best to make the printing process so that there was no hidden deformation forces left in it. I developed a model in a slicer, and after several experiments I modified its thin-walled shell, print by infill only without external walls. The left model contains a layer height of 0.2, the right one - 0.4, wall thickness – 1.2 (of 3 nozzle diameters), filling – as indicated in the figure. The filling factor is 90, so as not to cause expansion of the walls, and the experiment would be successful. The infill turned out to be loops, and along the outer surface it resembled a brick wall.
Model for testing «brick walls»
Here is the result:
The left sample is printed with 0.2-layer thickness, and the right of 0.4.
The sample with a layer of 0.4 turned out to be the most successful.
As can be seen, its dimensions along the XY axes decreased by only 0.5%, and along the Z axis increased by 1.5%, which can be interpreted as a statistical measurement error associated with the general deformation of the model during its transition to the elastic state. The model now does not contain internal stress and hidden defects, which means that they will not affect the quality of your prints in any way!
Conclusions According to my ideas, the physics of the flow will now be like this:
To print walls of shrinkable materials like ABS, not necessary to create squeezing and stretching forces generated by the extruder nozzle. And when the extruded thread comes out freely, no any deformation occurs. The walls must be printed in such a volume that it fits into the required place under its own weight. In other words, the cross-section of one thread of the layer must be equal to the flow area of the nozzle. It is necessary to interrupt the flow in one direction in every possible way so that all internal deformation forces do not add up, but diffuse. The strength (as well as tightness) of the walls must now be ensured by the spatial interweaving of threads of hot plastic.
To avoid additional contracting forces in the solid filling, it is recommended to choose such patterns “Gilbert Curve”, “Spiral Octagram”, and “Archimedes Chords” …
For filling, any one that does not form straight lines from wall to wall is suitable, for example, “Gyroid filling”.
Which printers will be beneficial to this feature?
Marlin
Describe the solution you'd like
I have only given preliminary conclusions on how the printer can be configured to overcome this age-old problem of 3D printing - shrinkage and delamination. But we need a software implementation of this idea, which I can no longer implement on my own. Using this principle, you can come up with different algorithms for quickly printing such walls. You can, for example, make some kind of “3D piles” that will prevent the layers from coming off, or try to implement the “trembling flow” algorithm, which will not cause excessive tension on the wall lines. Or, indeed, print “3D bricks”, or maybe loops that include several vertical layers... Can try a combination with regular outside walls, so as not to degrade the visual quality of the print, or, conversely, make a textured outside wall of the model... Here a great horizon opens up for the implementation of this idea!
Describe alternatives you've considered
No response
Additional context
I'm sorry, English is not my native language, and if you have any questions, you can ask them. I will try to answer them in more detail. If you need a model or g-code, I can also provide it.