3D Printing

Overview

This week I designed and printed a 3D object that could not be made via subtractive fabrication methods. The CAD modelling process, slicing & printing preparation, and printing steps are detailed below.

3D Printer Design Rules

This weeks group assignment was to test the design rules for our 3D printer(s). The group collectively printed a few test prints, each characterizing a different aspect of the Prusa printers. The three images below are test prints for bridging (teal), overhang (yellow), and clearance (purple). The teal print tells us the Prusa is capable of successfully bridging various lengths with minimal material sag. The yellow shows that the printer cannot print unsupported overhangs greater than 1 mm without sagging material. The purple print suggests that our models should have greater than 0.3 mm of clearance between features and parts for guaranteed separation.

CAD Modeling

Since the goal was to create something that can't be milled, I considered natural and organic shapes as inspiration. Ultimately, I ended up with two objects - an interwoven, spider-web-like structure and a nautilus-shell-like donut. I spent most of my time this week on this step, because of issues I ran into when trying to export a .stl of my model.

I originally created a cube composed of the spiderwebey model, which was a collection of feature rotations, mirrors, and translations. This resulted in several (~200) separate parts that I was not able to union into a single part, without OnShape crashing. When trying to export the parts without union-ing them I noticed the .stl would take incredibly long to export (~30min) and had missing geometry when it finally downloaded. Instead, I tried union-ing parts throughout the design workflow, rather than once at the end. This worked in allowing me to connect most pieces, but I still ended up with 3 separate parts. I was able to export this version, but ran into issues in the next section with this version. In the end, I ended up exporting a subset of the cube, the atomic unit of the cube, as the .stl I used for printing.

Slicing

As mentioned above, I had an issue with uploading my cube of spiderwebeys to Prusa Slicer. I received an error saying my model had incomplete geometry that needs to be modified. In the interest of time, and assuming this is the same issue I ran into previously where I tried exporting multiple separate parts, I uploaded the atomic unit of spiderwebs instead. This posed no issues with uploading. However, it did import tiny in scale. I simply scaled the model up by adjusting the largest of the XYZ values to 30 mm, with the locked ratio selected. I hit the generate button and created the g code file for printing. Before exporting the g code to an SD card, I modified two values – “set bed temp” and “wait for bed temp” from S55 to S72, per Tom’s suggestion.

Printing

The printing process itself was fairly simple and hands-off. After inserting the SD card and selecting my print file on the Prusa, I waited for the bed to heat up, watched the printer to self-set it’s Z offsets, confirmed that the first layer went down cleanly, and left to let it do its thing. About an hour later I returned to find a successful (but extremely hairy) print. I understand this hair is an artifact of the Prusa, when printing at small scales. I tried cleaning these hairs up with a razor blade, but I think a smoothing solvent would be a better alternative approach to get rid of them.

Scanning

Overview

This week I also scanned and printed an "object" using the Artec Leo 3D Scanner and Artec 3D stitching software. The “object” I chose to scan is my lab partner, Ido. In the event I have to work remotely, I’ll have a 3D version of him to keep me company wherever I go. The details of this process are described below.

"Object" Capture

The Arctec Leo scanner is simple to start using. I turned it on via the button in the top left corner, selected a new project, and pulled the trigger to start scanning. The tricky part comes when sweeping over the object, especially when that object is a person. I ended up doing 3 separate scans to get the final model to look as close as possible to the real thing. The key was staying within the “green” mesh, not too close (red) and not too far (blue). Another trick I found was to start directed at Ido’s face, circle around (while sweeping up and down), and stop before scanning over his face again. Multiple passes across the front of his face in the first two scans resulted in some odd texturing, one where the model had 3 eyes.

Post-Processing

The post-processing of the scan was the most time-consuming part of scanning. I selected the scan on the scanner touchscreen, copied it to the SD card, and ejected the SD card to insert into a PC with the Arctec software. In the Artec 3D software, I created a new project, imported then the scan (.pkg) of Ido, and ran the autopilot stitcher. The resulting stitched model created a beanbag-sort-of geometry below Ido’s waist. To remove this, I created a 2D plane and extruded it to erase the model below the shoulders. Finally, I filled the hole at the bottom and re-textured the bottom, creating a model I could export as a solid (.vrml & .stl) with a texture map (.png). I shared these models with Tom for printing in the J-55 printer.

Note – To try and salvage an initial scan of Ido, the one with 3 eyes, Jon suggested selecting all the frames that included the 3rd eye and deleting them. I tried this (third picture from the left, ) which amounted to several hundred frames, then re-ran the autopilot. This resulted in slightly better geometry around the eye but removed detail in the nose. Re-scanning, starting with the face, as described above, resolved this issue.