Some thoughts on a cheap, fast and accurate SCARA reprap

I had a few ideas on how a selfreplicating scara robot could be done.

The first issue that popped up is rigidy. Normally extruded PLA isn't rigid enough, especially where forces are perpendicular to the extruded strands. But this is only half the issue.
The way slicing currently works it doesn't generate the optimal path for internal support structures. Paths are normally just straight lines, filling the interiour with the hope that the squashing of the strand between the nozzle and the bottom layer provides enough contact to be sturdy. While this works and is fast it is not the optimal solution in any respect.
My solution would be 3D voronoi cells. They form a honeycomb like structure which should be lighter, stronger and more rigid than any of the current techniques. In a way they are also a mathematically optimal solution.
Constraints on the angle of their walls has of course to be met. They are computationally intensive constructs, but that shouldn't be a serious isse.

Next is the Z Axis. Commercially available SCARAs normally use a special made screw for it and I see no reason why we shouldn't be able to print one. Yes we can't print it straight up and horizontially the support structures would mess up the surface and interiour.
But we could print it at a reasonable diagonal angle. Now that wouldn't be of the quality of their metal cousins but should suffice if we overwork it manually.
And if that fails we can always can back to threaded rods...

more later.

Interesting idea for support. Have you read Gary Hodgson's thoughts?

The SCARA robot is going to be a challenge. You will need very high angular positioning quality to achieve prints comparable to the Cartesian bots. The second challenge will be using a printed screw for Z axis. Since Z position and layer thinckness are critical to print quality, you might want to start with commercial screws for early versions, just to make things easier.

... read through this document: [citeseerx.ist.psu.edu]

It's describing the benefits of a parallel scara structure in comparison to a normal 'single-arm' system ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

The rigidity of the parts is one problem, but I would be much more concerned about the positioning mechanics. The positioning repeatability, speed and acceleration have to be something comparable to the common cartesian design (0.1mm, 100mm/s and 1000mm/s*s).

The paper talks about technologies such as zero-backlash micro-gears and servomotors with integrated position encoders. These are most likely not feasible for hobbyist systems (because of cost). The positioning mechanics for cartesian printers cost something like 100€ total in most configurations. That's about the price point the SCARA reprap must meet, for the positioning mechanics, to compete with the cartesian designs.

What kind of rotary positioner is capable of that kind of repeatability at the required speeds and accelerations... and costs no more than 50? I don't think anything like that exists. Achieving the linear repeatability at the end of the arm translates into very strict angular repeatability requirements at the moving joints. (Something like 0.03 degrees with a quick calculation.)

edit: The Z axis would be the easy part. Just mount the whole arm onto a screw-driven Z carriage. (It would be crazy to put the Z movement at the end of the arm, it's not necessary in a RepRap machine.)

Edited 1 time(s). Last edit at 04/17/2012 07:56AM by ttsalo.

Heres my attempt at a SCARA robot. There is a video half way down of it moving.

Helium Frog SCARA Robot

The Z axis lead screw is very slow, so I am currently developing a belt driven Z axis. I think this will then make an effective pick and place robot.

Resolution of my robot is less than a Reprap, but with 400 step / rev motors and the gear ratio you see on my robot, the resolution does approach that of the Reprap, so this is not too great an issue if you are considering self replication.

Rigidity of the tool head is a problem. As Viktor has pointed out, a double arm or parallel arm setup would probably be the way to go if you want self replication. There are also designs with two fixed motors which would work well.

One major complexity with a SCARA is the elbow hinges as these need quite a few components in them (Two bearings and a long bolt). A living hinge or a good simple pinned hinge design would simplify the build a great deal.

With the experience I have gained with the above project, I think a rising table under a dual arm SCARA with fixed motors would compete with a cartesian robot. The rising Z axis table could be a Sarrus Linkage so as to avoid the requirement for printing linear bearings. This would also be handy as it means that the project could be divided into two separate projects should a team wish to work on development.

"Something like 0.03 degrees with a quick calculation"
360/16384 = 0.02197
Thats roughly the resolution of a 14 bit encoder. (like the AS5048 for ~7$ each)