MakerBot Industries

Broken gears are a sign of progress

I originally called this “Project RoboSpider,” but I’ve decided “Clockwork Spider” is significantly cooler.  Once I had printed the original “RoboSpider” parts, I noticed a few problems.  I’m hoping that someone else might be able to make use of the lessons I learn in designing and printing this multi-part mechanism.

1. Don’t make parts too thin. The parts I printed tended to be designed too thin.  If you are designing a multi-part mechanism, don’t skimp on plastic and make the parts unnecessarily thin or small.  My original gears and cogs were 2mm thick.  The problem was that it was easy enough for the teeth in one gear to simply miss the other thin gear.  My new design uses gears that are 5mm thick and they never miss one another.
2. Don’t use vertically printed snap parts. I printed all the parts using a vertical resolution of 0.25.  Dave Durant was right, this is basically a sweet spot for printing.  It’s quick enough that I’m not waiting forever for parts, the resolution is high enough that it looks pretty amazing, and layers close together enough delamination just isn’t an issue. ¹  All of the gears were designed with prongs so that they could be snap-fit to the chassis.  My first attempt at these failed because the prongs were so incredibly thin they couldn’t be printed.  My second attempt failed because the prongs just snapped off as they were flexed to go through the hole in the chassis.
3. Don’t reinvent the wheel… or gear.  The gears in my original designs were very very home brew, and it showed.  I just created a flat cylinder and extended a bunch of small cubes off the edge.  You can build a gear such as this easily in OpenSCAD, but using the “module” command makes it SOOoooooOOOOOooo much easier. ²   However, there was a lot of trial and error for me in getting the length, width, and space between the gear teeth properly proportioned.  For my second revision, I designed the gears over from scratch using MCAD, Greg Frost’s Involute Spur Gear Script, and cbiffle’s Spur Gear Fitter Script to create gears that meshed well.  Yes, it is more work to learn how to incorporate others’ works in your own, but you also get to benefit from their knowledge, experience, and expertise.  You’ll also save time by not having to print a bunch of crappy gears that don’t work. 
4. Ask for help.Thingiverse citizen Dna responded to my call for a rubber band powered motor with his Rubber Band Ratchet Engine.  My second revision benefited from Dna’s comments and input, as well as from getting to modify my designs to work with an alpha version of his rubber band engine.  And, last but not least, for creating this super sweet video of a variation on the clockwork spider in action:

What lessons do you have for others creating multi-part mechanisms?

1. Dave Durant: “You can go down to tweaking it by 0.25 if you want but any more than that is overkill, IMO. Other variables (ambient temperature, filament inconsistencies, how well X/Y rods are oiled, etc) can effect the print more than tweaking that much. If you can get it to 0.25, you’re going to be really happy with the prints.” [↩]
2. What’s that?  I haven’t covered modules in OpenSCAD?  Don’t worry – they’re easy and actually a lot of fun. [↩]

Or, transmuting a digital object into a real 3D object embedded with a 2D representation of its “DNA.”

Digital DNA

A little over a month ago I had pondered how cool it would be to have things on Thingiverse embedded with QR codes – so you could show the bottom of the object to a friend, they could grab the digital file, and print off a duplicate.  Well, l0b0 has done just that with his parametric box!

Green into red? Madness!!!

I’m a big fan of bar codes. ¹  I really like the idea that you can have something physical instantly transformed into something digital. ² In a way, a bar code is the opposite side of a MakerBot coin.  I realize that print resolutions, thing shapes, and QR code size requirements would preclude this, but it would be so cool to have each object printed from Thingiverse to have a little QR code built into its side.  If you wanted to give a copy to your friend – just flash the QR code on the bottom at their phone/webcam/MakerBot and they can have one too.³

An interesting factoid about QR codes is that the size of the QR code box is related to the amount of information being encode into it.  The longer the URL, the larger the QR code needs to be.  By using a URL shortening service⁴ on a Thingiverse URL⁵ along with a QR code generator you can essentially compress the data required to reach a Thing on Thingiverse into a smaller QR code.

Using a 3D scanner to duplicate an object will basically guarantee the digital version of that object will be rougher than the original.  However, using a built-in QR code, you could have duplication without generational degradation.  It would be like stamping everything on Thingiverse with the DNA necessary to build a duplicate.

Semi-random thoughts:

1. Can all of the DNA in a human be expressed as a long string of text?
2. If so, it would be very interesting to me to try to encode that long string of text as a QR code.  I wonder how large it would have to be?
3. One way out is to link to the Human Genome Project, run their link through a URL shortener, and then create a QR code from that.  Now you can print people!

The Human Genome Project

1. Photo courtesy of Mario Caruso [↩]
2. Zach’s recent Tweet got me thinking about this post. [↩]
3. Then again, at that point it might be easier to use image recognition software to match the printed thing with the Thingiverse catalog. [↩]
4. I like YOURLS, but that’s because I like open source stuff, rolling my own versions of things, and the idea of having my own URL shortening service. [↩]
5. Or link to someone else’s STL. [↩]