Robot Challenge Marathon

3. Materials required: 8 plastic parts

Eight plastic parts

Whew!  It’s been a wild few weeks!

Just a few weeks ago I found out about WyoLum 2013 Innovation Grant and discovered that the AFRON 2013 Design Challenge had been extended.  I then began a foolish feverish attempt to enter both challenges before their respective deadlines (12/31/2013 for the WyoLum Innovation Grant and 1/15/2014 for the AFRON Ultra Affordable Educational Robot challenge).  I found it difficult to get the robot’s plastic part design and software to an acceptable point for submission for the WyoLum Grant and probably more difficult to get all the documentation in for the AFRON challenge since their challenge required so many parts – pictures, videos, putting on a robot building workshop.

If you’re interested in reading a LOT about the Tiny 3-axis CNC drawing robot, my entry into the AFRON 2013 Design Challenge is here.  Between that page and the software user guide, I’ve written more than 11, 000 words and added 115 pictures and 5 videos.

If there was just one page on this entire site to teach you more than you ever wanted to know about how to build a tiny drawing robot (including what parts you could scavenge, where to find them, what parts you can substitute, how to wire up the robot, program it, and get it drawing) look no further.

Designing with Injection Molding in Mind

Injection molded parts

Injection molded parts 

I had always assumed injection molding was a pretty straightforward process.1 You send your digital files to the injection molder, you pay a bunch of money, and plastic parts show up.  While looking into the process of injection molding, I discovered there are all kinds of design requirements.

  • Uniform Thickness.  Apparently having a non-uniform thickness to plastic injection molded parts causes lots of problem.  The plastic can flow into the mold unevenly and cause bubbles or voids.  The thinner parts would cool quicker and the thicker parts would stay warm longer, causing the part to warp as it cools.
  • Draft.  Apparently all parts that are injection molded require some amount of “draft.”  This means that a part should be tapered outward slightly – so that it can slide easier out of the mold and incur less friction as the mold parts slide together and apart.  The various resources I’ve found suggest a minimum draft angle of 0.5 degrees to as much as 5 degrees for parts with lots of surface texture elements.
  • Part Radiusing.  Since the plastic shot into a mold is basically a viscous liquid, it flows better around curved corners and has a difficult time flowing around sharp angles.  The guides online suggest that internal curves should have an internal radius of 0.5 times the wall thickness and an external radius of 1.5 times the wall thickness.  Plus, proper radiusing means consistent wall thickness, even around part corners.
  • Coring Out.  The process of removing excess material, leaving the bare minimum uniform wall thickness in walls and ribs for strength.  This allows the finished part to be of uniform thickness to prevent uneven shrinking and internal part stress.
  • Radiused Corners.  As a part’s geometry is carved by a CNC mill out of the metal mold the CNC can only carve with a minimum diameter equal to the CNC’s bit.  This means corners won’t ever be true corners, but rather small curved internal corners.

Interestingly, these design requirements also explain why so many plastic parts are basically shells.  I had always assumed this was done to reduce plastic and cost.

  1. Photo courtesy of Creative Tools []

Competing Design Ideals in a Drawing Robot

New possible design direction for the Tiny 3-Axis CNC

New possible design direction for the Tiny 3-Axis CNC

I find myself at a design crossroads, as it were, with the Tiny 3-Axis CNC.  There are certain improvements that I think are necessary to make the overall robot more functional and reliable.  However, to adjust these designs to accomplish these improvements would require a compromise of some part of the design ideals I’ve been employing so far.

One unintended consequence of having a low number of interlocking parts is that when I make a design change to one part of the robot, the design implications ripple throughout the rest of the robot.  Thankfully, this is made slightly easier by using OpenSCAD which automatically adjusts parts depending upon changes to variables.  These changes also have the side effect of making each version of the robot unique enough that almost no parts are compatible with other versions.  As a result, I’ve got a pile of parts from intermediate non-functional versions which don’t really work with any robot version.

In order to overcome some problems with the last design1 , I basically now need to choose what I value most:

  • Elegance.  Design elegance is a very murky and personal topic.  I think of design elegance as incorporating the fewest possible number of parts, simplicity, and (when possible) symmetry or the reuse of parts.  Even “simplicity” is a convoluted and subjective ideal.  I think of this as the least amount of plastic and least amount of complex features.
  • Low Part Count.  I love the idea of a super low part count.  If I were to print parts with support structures or had a very complex design for injection molding, I could probably reduce the part count to the absolute bare minimum possible number of parts – 7.2 However, if I try to make the parts easily printable without overhangs or support structures I have to increase the number of parts.
  • Easy to Print.  Even if a 3D printer is capable of printing with mild overhangs and support structures, a design is more easily printed if it doesn’t include such features.  It is very important to me that the parts are easy for people to replicate on their own.  For me, part of being “easy” to print is having as little plastic as is required.  The most recent stable version of this design takes about 2 hours of printing on my Replicator.  I think I can do better.
  • Easy to Assemble.  Generally, I’ve found the more complex and numerous the parts, the more difficult and less intuitive a thing is to assemble.  I would really like this robot to be able to be assembled by my 6 year old with minimal adult intervention.  She’s pretty good at building Lego designs from the graphical instructions and I’d like to have something similar here.  Fortunately, OpenSCAD make it really easy to create Ikea/Lego style graphical assembly instructions.

After discussing these issues with some friends, I think I’m going to sacrifice the super low part count as I push the design forward.  I don’t anticipate this will cause the design to have a higher plastic content – just a few more pieces.  Overall, if I had to sacrifice one particular design ideal in order to adhere more closely to the others, I would have to choose the super low part count.  After all, I could always publish an additional version that combines two or more parts into a version that could be printed with support structures.  🙂

  1. Principally the XYZ carriage tipping out of the X rack []
  2. I think this would require: the X rack/base, the X pinion, X motor mount, Y pinion, Y rack, Z pinion, and Z rack. []

Pen Lift Achievement Unlocked!

Drawing vectors with the PlotterBot

Drawing vectors with the PlotterBot

I recently developed a simple printable pen holder for my PlotterBot.  Although it worked wonderfully as a pen holder for single-line drawings and although I had designed it to work with a servo for pen lifts, I had literally never tried actually using it with pen lifts until yesterday.

Above is my first attempt to draw vectors with pen lifts and, frankly, it came out beautifully.  The design of the pen lift system could not be any simpler.  The pen holder has a mount for the micro servo which holds it as close as possible to the wall-facing surface of the pen holder and a rectangular hole for allowing the servo’s arm to sweep through.  I was concerned that the sweeping action of the servo arm would cause a slight stray mark on the paper.

Despite my concerns, the pen lift test was successful.  It really came down to a matter of balance.  Once I had swapped in a different pen lift arm and connected the pen lift servo cable, I was able to easily adjust the balance by changing the location of the filament attachment.  With the proper balance achieved, the pen lift essentially worked flawlessly.

Today I tried my longest drawing ever, an 8-hour process, with voluminous pen lifts.  The result was… amazing.

Stay tuned for a picture.