Nathaniel and I met up at CTRL+H last night and hacked on the printer for nearly 5 hours. We successfully "printed" our first "objects" directly from shredded trash flakes, and we learned a lot in the process. One of the most promising things we learned is that coroplast sheet can indeed be used as a build plate. For a printer of this size, making a heated bed like a typical printer has is a daunting and expensive design challenge. But without a heated bed, how do you get your first layer to stick?
I knew that Polypropylene is very self-adhesive, and my hunch/hope was that maybe I could just lay down a sheet of cheap Polypropylene coroplast and that the molten PP would adhere to that, because they're the same material. That way you could just slide in a piece every time you print, and then cut off your object (and then shred up the scrap and print with it, of course). That part worked out quite well- the PP adhered much more strongly to the coroplast than it did to the wood underneath. That means that a heated bed upgrade isn't necessary- putting the cost to replicate this setup at under $500, using almost entirely 3D printable, laser cuttable, or widely available parts.
We decided to try and print out the Hackaday logo [based on this model from Thingiverse], and this the result of our first attempt. It's more of an "ckaday" logo, but it was an encouraging first shot. Nathaniel adjusted the ratio of extrusion to speed, and we tried again. One of the challenges of this kind of printing is that the extrusion is non-linear. It's not moving a spool of solid filament, it's building forward pressure in the extrusion barrel, and so there's a 3-5 second delay between when the auger starts pushing and when the plastic actually starts coming out of the nozzle. Luckily Nathaniel speaks robot quite fluently, and after a bit of arguing, we got this:
Not bad! Not, you know, great either, but pretty damn good for our second try. Next steps are to rebuild the extruder to be little sturdier (it was wobbling back and forth a little when extruding), mount the thermistors more securely (one of them got pinched and wouldn't read) and build an actual wiring harness and material feed. But all in all, I'm really happy with this test, and I am confident that this approach can produce useful (if not-very-detailed) objects.