Processing it into gcode with Skeinforge proved to be impossible. That led me into a meandering sidetrack which had me pushing my old Slice and Dice code into working order after several years of neglect after Netfabb ran late with the release date for their STL to gcode processing module. Last weekend, I got Slice and Dice working acceptably.
After work yesterday afternoon, I decided to see if I could make my Slice and Dice alpha code successfully generate gcode for a print of that same light structure. There were a few hiccoughs. Fortunately, Slice and Dice is purposely designed to let me deal with difficult prints. I was able to identify problems by doing partial prints and tweak my settings to get past them.
When I got up this morning I finished processing the light structure STL and set my Rapman printer to work on it. Being a stand-alone 3D printer, it happily worked while I got on with my day job. Roughly 90-100 minutes later, I had a printed module.
I was quite chuffed with my design. Clearing the female connectors on the bottom with a 5/32" bit took a matter of a few seconds and let me mate one of my test prints of the top of the module with the whole printed module. I immediately started printing up a second module. It finished up a few moments ago.
I think it's pretty obvious where this is going. I did a short clip of the print underway so you can get an idea of how fast it proceeds.
I am printing a third module now and plan on printing five over the course of the day. This evening after work, I plan on designing end caps for the assembly so that I can post tension it and then get an idea of how much deflection it exhibits under load for a 500 mm beam.
The lessons I've learned from all this is that if you design light and print vertically, viz, keep the cross-section on the xy print table small, you are going to get what are basically no-warp parts without going through the drama of heated print beds or turkey bags. it's worth thinking about.