Long time no see, folks. I’ve got some great news for you. I’ve finally found a method for getting super complicated geometry locked inside of a seamless skin. It’s taken a lot of prototypes to get here, but I think the results are more than worth the effort. There are some wrinkles to iron out (which I’ll get to below) but all in all I think I’m incredibly close to rapid-fire casting working quadrupeds, ready to go in just a few short steps after popping the mold. In other good news, I’ll be dropping some files very soon which should get you your very own working quadruped using any FDM printer. All you need is a Makerbot or similar, a few hours, and some casting materials to have an exact duplicate of my most sophisticated robot to date.

One of the major hurdles for getting the quadruped designs (now called the Glaucus project) perfected has been getting everything to align effectively. Ideally the whole robot would come together in one smooth step – no seams to worry about. This goal introduces a lot of practical limitations, though. If you have a sufficiently complicated wax part forming the core, it’s guaranteed to have stray bits that accidentally collide with the mold or each other if left unsupported. Inevitably you need some kind of supporting structure if you’re going to have complicated pathways of bladders and tubes running through your robot. However, the supporting structures themselves are going to leave cavities in your bot once they’re removed, and they’ll need to be backfilled with silicone.

So, you have this catch 22: either risk a super finicky mold that comes together in one seamless step or use a multi step process and risk introducing a weak seam into your bot. I think I’ve found a way to align the waxes without introducing too many problems by timing the silicone pours very precisely. If you design your molding system with 3 sides (a-la the last rev of the quadruped) and cast your robot timing the first pour to gel and stay solid while you switch mold parts and pour up the final side, you get a seamless silicone skin. The process still needs some proper load testing, but it seems like if you do a second pour within the gel window of the first you’ll get a skin that’s indistinguishable from a single unbroken pour.

I’d figured I’d solved a lot of my cure inhibition problems – detailed on previous posts – by switching the whole process to FDM printed bathtubs. The process goes: you print out a bathtub for every mold part and then you cast a silicone tool from these prints. This was supposed to fix all the troubles I’d had with odd bits of silicone failing to cure. I’d thought it was because the powder printed parts contained elements that might throw off the silicone cure. And yet, my silicone tools were also leaving sticky cast quadrupeds. After some research, I figured out why. It turns out that a silicone with a platinum catalyst will cause one with a tin catalyst to fail to cure and vice versa. All my tools have been made out of Mold-Max (a tin cure silicone) and all my parts are cast out of a blend of EcoFlex and DragonSkin (both platinum cure silicones). Well, I’m switching over tooling materials and that problem should be fixed in short order.

Now, if you’ve fiddled around with molds before you might be wondering why I wasn’t into using a parting material to solve the cure inhibition problems. That would have worked for a single casting but since my process now includes that wax supporting layer any parting materials would end up in between my two silicone pours. Since there’s no simple way to make sure every speck of parting material is cleaned off of a jellied silicone blob, I figured settling on this as a solution cut down on how reliable, distributable, and repeatable the whole method could end up being.

I think this new FDM process is going to be extremely useful. Even though it can’t do the geometries powder printing can, the benefit of the super smooth surface and low warp/shrinkage is huge. Also, it’s much more likely you’re going to find folks with access to FDM printers than working Zcorps. Additionally, the silicone on silicone molds make precision ventilation much more plausible. Now, I’m using a professional quality vacuum chamber to degas my silicone before and after pouring and am getting crystal clear bubble free parts.

Pretty soon, I’ll have a tutorial out on the Adafruit Learning System, a few more vids, and a file release on Thingiverse. In the meantime, go check out Super-Releaser for some more info on what I’m up to as well as Flickr for process photos (down at the bottom).