**WORK IN PROGRESS**
Sometimes large cross-sectional area parts fail to print on Ember. One observed failure mode is 'jamming,' where a cured layer appears to adhere both to the accreted-part and the PDMS window. When the stepper motor attempts to rotate the resin tray it is unable to, thus causing a 'jamming' or 'motor-timeout' error.
Another failure mode is severe warping on the edges of prints, which appears to be proportional to the cross-sectional area near the base of the print.
One idea to solve these issues is to use a checkerboard infill pattern, which is inverted, from layer to layer. This will reduce the effective exposed surface area, on average by half. From observation, large surface area seems to induce jamming more readily. From a theoretical perspective (see the paper The Lubrication Approximation), large surface area has a dramatic effect on the force applied to the plate, and thus the resin tray, when it moves downward after the slide action is completed. In addition, large surface area means large shearing forces on a recently cured layer as the resin tray rotates.
1) Using a checkerboard infill will reduce the likelihood of jamming, with the greater number of ever smaller tiles corresponding to a ever reduced likelihood of jamming.
2) Using a checkerboard infill will reduce warping.
** add equations **
1. Reliably induce jamming, and warping
2. Take same set of conditions from 1. and add checkerboard infills, test varying levels of division (4 tiles, 9, 16, 25, etc) to see effect of jamming
Print File Generation
1. Create a processing script which generates the sequence of checkered png's.
2. Generate gray-scale infill prints using emberprinter.com and simple box-meshes in rhino3d.
Step 1: Creating Test Geometry
1. Gray-scale infills are the default on emberprinter.com. Since and stl is required, a simple mesh box was created in rhino3d, with the required base dimensions and a height of 501*.025mm. It was found that this resulted in 500 layers in emberprinter.com's slicer, with 25 micron layer-height.
2. Checkered infills were generated with a custom processing script (attached), which can generate alternating checkered images with set base dimensions, number of divisions, and number of stacks. These files were combined into tarballs. The the printsettings file was copied from the gray-scale print tarballs, downloaded from emberprinter.com, and modified to match the file name. See this instructable on making png stacks into ember prints.
Step 2: Inducing Jamming and Warping on Ember
To test the hypothesis, first it was necessary to reliably induce jamming/warping. Note: Initially the goal was to just focus on jamming, thus the repeated tests trying to get jamming. However, multiple tests showed that another, perhaps more important issue is warping.
Test Geometry: varying cross sectional areas, but always 500 layers tall, with 25 micron layer-height (a nominal height of 12.5mm).
E1_T1. To do this I started with a square-based prism, with an area equal to π*100 mm^2 (in case I wanted to test a cylinder, a radius of 100mm would be necessary). The prism was sliced using emberprinter.com resulting in the gray-scaled infill. I was using the settings under the title of Autodesk Standard Clear 025um(new). This included a First Layer Wait (After Apporach) of 3 seconds, for all types of exposure layers (First, Burn-in and Model). However, this did not induce jamming.
E1_T2. I then printed a 30x30 square base prism, 2.86 times greater in cross-sectional area than the previous print. Same settings, on a different printer, which had been observed to jam or otherwise fail more readily. I was unsure if this would make a large difference. The print jammed, but got very close to completing the build, with 41 layers remaining, out of 500.
E1_T3. Next I tried a checkered pattern dividing the square into four quadrants, with 30x30mm base, This failed, apparently due to jamming, although it is uncertain weather this was also related to the resin tray cracking, and leaking resin onto the resin-tray platform. The part was 5.08 mm tall, suggesting around 200 layers had printed. Significant warping on bottom edges was visible.
E1_T4. The prints were still not jamming as readily as I had hoped, so I increased the square base to 40x40mm, and printed on the original (more reliable) printer. The print was a success.
E1_T5. I increased the base to 60x40mm, almost the maximal size possible. The print succeeded.
E1_T6. At this point, since cross-sectional area was almost maxed-out, I decreased the single variable that I seemed most related to failure: wait time after approach. According to bad-zima, the resin-tray, and thus PDMS window deflects after/while the build pate approaches the window just before exposure. This causes the deflection of the window, and thus the gap between the build-plate and the window, to jump up, and then slowly return to the ideal 25microns. Given this model, is makes sense that allowing a long time, in this case 3 seconds, for the gap to approach the ideal value, reduces jamming. So I arbitrarily decreased the wait time, only for model layers, to 2 seconds, and indeed got jam, at about 4.4mm of height. This corresponds to roughly 176 layers. Warping on bottom edges.
E1_T7. Reprint of E1_T6. No jamming, however small debris left in resin tray, misplaced layers sticking out of side-wall of part. Warping.
Step 3: Next Steps and Remarks
Next Steps (ok I am getting ahead of myself here, oh well): If a checkered infill is desirable for reasons of jamming or warping, or both, the pattern could be applied using a filter, and built into the slicer on emberprinter.com. Other infill patterns designed to reduce cross-sectional area could include
- Random checkered patterns
- Branching hierarchies to encourage flow from the center to the perimeter of a given area
- non-random patterns involving intentional overlap between layers
- Polar coordinate checkered patterns (like a dart board)
- Other tessellations (triangular, hexagonal)
Severe warping, or something that looks like warping, occurs on corners of the large-cross section prints. These warped corners are always on the build-plate side. Given that increasing wait-time after approach seems to increase the reliability of prints, it might make sense to look into this warping issue. Any geometry that is for some engineering, or even aesthetic, purpose is greatly reduced in value by warped corners. However, checkered infill patterns may still be worth investigating, since they may call for shorter wait-times, and thus faster builds.