I’ve finally got tests working on the drawbot, here are some pictures of calibration efforts.
View calibration gallery here.
2013-11-26: updated link to github project!
Among the various features are…
The green is one half of my first attempt. The two on the right are attempts 2 and 3. The assembled joint in the bottom left has an M3x20 screw and a 1/8 ID, 3/8 OD bearing. In every single case the round part that holds the bearing cracked when I inserted the bearing. In each case I increased the size of the hole. In the last version I tried to file the inside edge and even smooth it with a drill bit. While 3D printing is great for proof of concept I don’t think the makerbot is up to the levels of accuracy I’m going to need from these joints.
I wonder how it would do with shaft couplers and shaft collars. Hmm…
Sometimes I have to think outside the box. I designed this to be a kid’s game at the Vancouver Mini Maker Faire. Three people will have to move the red pieces in a team effort to move the blue piece to accomplish some goal (draw a picture, stack blocks, etc). Notice that this version has a complete frame that rests on a table, while the earlier versions don’t.
It would be straightforward to add three motors that turn screws to move the red pieces up and down. In fact, here’s a video of a Festo iFab 3D printer that does exactly that.
Tonight I will try printing parts again. Wish me luck!
Let’s review: 
So far…
…I’ve defined success
…I’ve used Math ™ to calculate the size of the humerus (red), ulna (yellow), wrist (light blue) and base (dark blue).
…I’ve changed the joints (white) so that they more closely mimic the math (see image). In the previous version the lengths of the ulnas was much harder to keep track of. Curious note: I’m told that Solidworks reported a lot of “rebuild errors” with the previous type of joint and those are now fixed. Yay?
…I’ve written an open source computer simulation of the math, and included Arduino code to run a delta robot with three hobby servos.
…I’ve tried to print the previous joints and found the hole for each bearing was too small.
I have yet to…
…3D print both halves of a joint and assemble a working model. This is my current challenge.
…3D print the wrist. Nothing is stopping me, so this will be my next challenge.
…3D print the brackets to hold the steppers (green). I want to find a way to add a limit switch so that I can calibrate the robot the instant it turns on. Even better would be a rotary encoder so I could track the arm movement in real time. I suspect such additions will change the shape of the brackets.
…make the base plate.
…test all the wiring. Make sure that the steppers move as instructed and that the encoders/switches work.
…make the humerus bars and assemble the machine.
…test axies: moving to zero, then 1 cm on each axis (to make sure +x,+y, and +z are not flipped in my software)
…test range of motion: moving from 0,0,0 to the eight corners of a box at each corner of the envelope.
…test repeatability: moving from 0,0,0 to some fixed point with a digital touch probe. Repeat 500 times and find the max of the error.
…test drawing straight lines in the air: video the tool moving in front of a checkered background.
… test drawing arcs in the air: video the tool moving in front of a checkered background.
… test speeds: ? ? ? At this point my knowledge completely runs out.
More news Tuesday night when I try to reprint these joints.
