Despite the superabundance of DIY tools that can be found in books and on the internet, very few designs approach dome building from a non-shelter perspective.
Rightly so, the common thought is to design a connector that’s as rigid as possible, to prevent any mechanical wiggling that could occur. While rigid connectors are great for building a load-bearing spaceframe, they’re not so well suited for the type of dome building I was interested in.
What follows is the story of how it took two degrees in design and six years of prototyping for me to create the final form: a rubber hexagon with holes in it. It’s been a slow, arduous process. I’ve made a lot of hilarious and expensive mistakes along the way. Let me tell you.
I started by trying to 3d print dome connectors on my Makerbot (an early batch 5 Cupcake, #96). Working from an open-source design I found on Thingiverse, I invested the weeks-worth of time it took to modify the design and successfully print the 26 connectors I needed to make a 2V icosahedron dome.
I had my mind set on making something bigger. While I could rapidly design and prototype with my 3d printer, going into production mode meant I had to stop my research & design process and tend to a finicky robot that needed constant attention.
So, rather than trying to print things myself, I experimented with crowdsourcing the printing of the connectors I needed for my next dome. The following was sent to all the 3d printing mailing lists I could find:
Within two weeks I had all the parts I needed, printed by makers from all over the world. I invited over my friends Ezra and Andrew and we put it together in the backyard. It was awesome! Big enough to get inside, so naturally we did.
This design had holes printed that would hold the strut at a precise angle. In order to make the dome self support, all the struts had to be the exact length and the structure built on a surface that’s perfectly flat (that’s why you see the wood planks in the picture above).
Wondering what would happen if the connectors could be flexible rather than rigid, I found and modified a 3d-printable ball joint connector on thingiverse to make the 5 and 6-way connectors required to make an icosahedron dome. This required a great deal of mechanical precision, and after a lot of calibration I was able to get my shaky Makerbot to produce a usable ball joint.
Despite each connector taking a good 6 hours to print, the design elegantly allowed each strut to self-adjust to the right position. There was no requirement to build it on a flat surface, and the flexible nature of the connector was easier for groups of people to build the structure.
With the feeling I was onto something, I launched and successfully funded a Kickstarter project that delivered 3d-printed domes. With the help of my friend Mark Cohen from Brooklyn, NY, we printed over 1000 individual components in a couple of weeks.
It was kind of a ridiculous thing to do, but we learned from this experience how boring production-level work with a desktop 3d printer is. The same object, over and over until the machine eventually breaks and hours worth of printing time is wasted. An obvious way to scale is to just injection mold it, but this design wasn’t ready for that. 3d printing is cheap, but slow. Maybe there was another way?
For the Kickstarter project, I offered a cheaper alternative to the 3d printed design out of the prediction that folks would be more likely to buy a cheaper dome kit than a more expensive 3d printed dome kit. This prediction proved correct: we sold only 3 of the expensive 3d-printed kits compared to 27 of the cheaper design.
This alternative design was cheap and dirty, based off of something I saw in Domebook 1. Get a tube, cut it into lengths, put a bolt through the middle to hold it together, then stuff a dowel in the end. Crude but effective, it worked fairly well with flexible PVC tubing. Further iteration revealed an even lazier alternative: why drill holes when a zip tie will do?
Despite its supreme simplicity, this design had its problems. I remember once trying to orchestrate a dome raising on a beach during a family vacation. The sand was so hot that the tubing went limp and lost its grip on the dowels, and the whole thing fell apart. This lead into a search for a better material — I tried silicone, natural rubber, neoprene, all sorts of stuff, but getting the right blend of friction was difficult. Too much friction and it’s impossible to disassemble, too little and it won’t self-support.
Not to mention that these were no fun to make. While the 3d-printed design was boring because a majority of the work was automated, the production of these things required a lot of different parts and had to be assembled by hand. Days upon days of cutting tubes into equal lengths, drilling a hole in each, threading a bolt, over and over.
While I only had to fulfill 27 kits, that was over 2000 connectors and more than 10,000 individual parts. My hands hurt just thinking about that experience. There had to be a better way.
Up to this point, all of my designs were composed of multiple parts that together expressed a flexible behavior. More parts typically means more complexity and greater expense, so I put a whole Summer into figuring out how to think about the problem in a different way. Why couldn’t it be a single piece, expressing the same sort of simplicity of my first 3d printed (non-flexible) design?
This prompted a material hunt, looking for as many flexible materials as I could find. I found flip flops and beer koozies had the flexible properties I was looking for, so I built a set of connectors to test the idea.
While at first the foam had a decent grip, over time it deformed due to the mechanical pressure of the dowels and never returned to its original shape. I realized I needed a more resilient material and realized rubber was an obvious alternative. Within minutes, I purchased a couple sheets of EPDM rubber from McMaster-Carr and they arrived the next day. After cutting out a couple shapes, I realized that I was onto something.
I iterated on this design, trying different sizes and hole patterns. From the perspective of cost savings, I explored thinner sheets of rubber, ending up around 1/8" as the absolute thinnest you could go with just a single hole for the dowel to pass through. After watching my wife weaving on a loom, I realized that you could actually make the rubber thinner if you could weave the dowel into it by having multiple holes for the dowel to pass through.
I felt satisfied in how this design solved the problem of getting enough friction by interweaving the two materials together. I tested it on a couple classes, A\B-ing it with the older designs. Students preferred the thinner weave design to the thicker, single-hole rubber connector because it held the dowel more firmly during assembly.
Testing also revealed that a slit worked better than a hole for maximizing friction for the dowel, so I opted for a design that was effectively a hexagon with 12 slits, which turns out can be water-jet cut for a reasonable price and a super-high degree of precision. After six years of working on this problem, I felt like the design had reached a level of simplicity and was comfortable enough to commit to mass production.
Working with Precision Cutting Service out of Savannah, GA, we cut enough pieces for a small batch of kits. Together with my wife, we screen-printed and sewed canvas bags for carrying all the parts. We are so happy to finally have these available for sale on our website, http://domekit.cc.
Architect/Philosopher Christopher Alexander talks about how the best designs are unfolded, meaning that they grow through continued adaptation from repeated use. I feel that this design has done something similar, but it’s nowhere near done. Now that I’ve committed to duplicating this design thousands of times, I’ve realized all the mistakes I’ve made.
Turns out, a slit really isn’t the best shape for the hole because if you don’t have the right kind of rubber, it can rip too easily. A circle is better because there’s equal pressure around the hole and it’s less likely to rip. I wish I had considered this sooner, but that’s what it takes to learn and that’s why I’m sharing this here.
Through this process, I’ve learned to be patient and let things unfold, because slowly everything will find its right place after being used for long enough. To make this work, you can never stop prototyping. You can always make another prototype.