I led the end effector team for the watercolor machine this week. After brainstorming a few times, we decided to make a purely mechanical system for the brush (no electronics except the x, y and z movement from the rest of the machine). We elected to make a gimball mechanism to allow the brush more "natural" actuation and a simple cup holder for the paint.
It was difficult to coordinate meeting times with all the members of the end effector team, so we ended up parallelizing the tasks. Kate took on the ink cup holder, Joel took the brush holders for three classes of brushes (very large, medium, and tiny), and I took on the gimball mechanism for the brush.
I modeled the gimball in Rhino to fit on the z-axis of the machine. The rotation in the gimball is created with very thin flexures connecting the rings.
Kate and I did a test for the gimball design on the lasercutter with some acrylic. The flexures in this test were too thin and stiff. The pieces kept breaking after lasercutting.
We also did a test 3d print test for the gimball design. The thinness of the flexures was fine, but the structure was too stiff.
Per kate's suggestion, I trimmed the flexures to half the height on the test print. This allowed the rings to twist.
The test brush holder (also 3d printed) barely fit inside the gimball. I chose to modify the press fit design with a compression clamp design in the next iteration.
I made a new iteration of the gimball design. I added a shelf so the gimball could easily attach to the z-axis. All holes in the design accomodate 4-40 screws.
I 3D printed the modified design with the compression clamp and shorter flexures. I printed extra gimballs just in case the flexures broke.
It went together really well!
I joined the assembly team to get the machine put together. As expected, we encountered a few problems. First, the pressfit joints were missing dogbones so we couldn't fit the pieces together. We fixed all the corners of the joints on the band saw and a few by hand drilling dogbones.
We also needed to countersink all of our drill holes so that our screws would lie flat in the material. Drilling holes perpendicular to the thin dimension of the material was quite challenging as well. But we got some fun plastic spirals out of the ordeal.
We attached the tinyG to the bottom of the robot box. And fortunately, we were able to force our gantry into the pressfit holes.
But then we found out that the motors in our digital model were different from our real motors. So we had to cut new mounting plates. We didn't have any HDPE left, so we scavenged around for some 1/4" material.
We also needed to find M3 to fit in the motor mounting holes. Luckily Soma was able to find some at the CBA shop.
We lost one of the milled pinions and had to 3d print a new one. Jen printed 2 for us on the ABS printer, but for some reason the inner diameter in from the 3D file was way too large. So we modified the model, added a hole for the set screw and printed on a sindoh.
We had to redrill holes into the HDPE so the pinion would interlock with the "teeth" on the x and z axis. We were extremely precise and used a marker.
Because we used 1/4" plywood for our mounting plates, we also had to use spacers (a stack of washers) to attach our motors to the HDPE.
We had to shave a bit off the z axis to make sure it slid freely. We used the bandsaw, and it took slightly too much off, but we had to make do. I drilled some holes in the z-axis to attach the end effector. Sadly, I only had 1.5" 4-40 screws left, so they stuck out quite a bit.
And finally added the cupholder.
How to Make (Almost) Everything | Fall 2017