The tale of soluble support and our new printhead Jun 02 2014
Our redesign is done, and soluble support, at long last, is here! After months of design work, redesign work, extruder drama, supplier drama, and several garbage bags full of failed prints, we've finally gotten it to work. We're saying it's in beta for now, given that the adhesion between PVA and PLA is not quite as good as we'd like it to be, but we're working on an entirely new support material, designed from scratch to solve this problem. More on that later. It's been quite a journey getting to where we are now, and we'd like to fill you in on it.
First, some background. Soluble support, or the lack thereof, has long been the single biggest factor holding back consumer 3D printers. Present in one form or another in enterprise-level FDM and polyjet machines, soluble support enables 3D printing to be freed from the constraint of requiring one layer to be built atop another beneath it, meaning you can actually print anything, not just objects with flat bottoms and no overhangs or unsupported features. It takes us a long way toward fulfilling the promise of additive manufacturing of designing without constraint. It's something we've been excited about for a long time, and we're thrilled to finally bring it to market.
There are a number of reasons soluble support has taken so long to happen, not the least of which has been a lack of will. The pack of consumer 3D printer companies, led by the esteemed competition in Brooklyn, has taught the world that 3D printers are toys, useful for making plastic trinkets and tchotchkes, an application that excuses the glaring lack of ability to create unsupported geometries. The technical hurdles were also formidable. Making soluble support work requires shrinking the extruder, increasing its reliability, solving the nozzle leveling problem while circumnavigating the Stratasys rocking extruder patent, and coming up with a suitable support material that adheres consistently to PLA. Thus far we've conquered all of these problems with the partial exception of the last.
The first issue is the design of the extruder itself. Although our initial idea involved a modification of the old Budaschnozzle to accommodate two separate channels, it became clear early on that this would simply be too hard to machine. We needed an extruder with a clean sheet design. It had to be more reliable (two extruders means twice the chance of failure), it had to be reasonably lightweight and compact, and it needed to be able to sit hot for extended periods of time without jamming. It also had to be minutely adjustable so as to be able to get the two nozzles absolutely level, so that one nozzle would not be dragged through the material laid down by the other.
We got really excited when we saw QU-BD's then new X-truder, a design featuring a small stepper motor geared down by 3:1, and geared into two separate feed wheels, but we were disappointed when we ordered one and found that it didn't actually work. It jammed left, right, and center. Its plates suffered from alignment problems, which ate up power from its already under powered motor. Its barrel, under siege from the notorious PLA swell problem, jammed consistently. And its poorly designed feed wheels wasted way too much of the motor's torque biting into the plastic. Still, the concept was sound, and we used it as a taking off point.
We spent October to January going through numerous iterations of our extruder, ending up with something that bears only superficial resemblance to the X-truder. We increased the size of the motor to a NEMA 17, adding 100 grams but tripling the power; we redesigned the feed wheels to have smaller but sharper teeth, such that they wasted less energy biting into the filament; we added spacers to fix the alignment issues; we lined the barrel with PTFE and tweaked its heat sinking to solve the jamming problem; we changed the mounting to accommodate our vertical X carriage; and we added a captive nut mechanism to permit precision adjustment of nozzle height. We've gotta say, the result is fantastic. It's the best thing on the market, bar none. Our new printhead never jams. The twin feed wheels hold the filament so tightly that it never slips or strips. True, nozzles still occasionally clog, but they're inexpensive and easily replaceable. We're releasing our design on under the Creative Commons Attribution ShareAlike 4.0 International License; you can view all of the source files at downloads.isis3d.net.
The next problem was choosing a support material and making it work. There were two options already on the market: polyvinyl alcohol (PVA) and high impact polystyrene (HIPS). Neither was ideal; PVA was expensive, and HIPS looked like a nonstarter, dissolving in limonene instead of water. Lacking the timeframe to develop something new from scratch, however, we tried both. Neither blew us away at first. More specifically, neither worked at all. Nothing adhered well enough to PLA.
After a month of fighting with settings and nozzle geometry, and a lot of time wasted barking up the tree of my brilliant (not) "blunt nozzle" idea, we got PVA working reasonably well. It's still not perfect. Most prints work, but in some cases the object will delaminate from its support, causing the print to fail. Right now, our success rate for soluble prints is around 70%, compared to 95+% for single extruder prints. Certainly we're not done yet, but when you consider that the industry average success rate for single extruder prints hovers around 70%, I'd say we're doing pretty well.
The next step, required to get soluble support out of the land of beta and into the realm of things that truly, reliably work, is to create a new support material. It needs to stick really well to PLA, it needs to be water soluble and nontoxic, and it needs to be cheaper than PVA. We're aiming at a retail price of $40/kg, the same as PLA. We've got some ideas, and we've begun work. Hopefully, we'll have something to market in the next couple months. Every one of our fabulous and patient pre-order customers will get a free kilo.
That's all for now. Check back next week for more, and as always, feel free to email us questions you'd like us to address here.