I'm glad you were inspired by my humble efforts. Do you have some pictures of your bots? I took the code from this tutorial: http://forums.trossenrobotics....

Although, nowadays, I would derive it from my quaternions library.

turns out that McMaster-Carr cancelled my order they didnt want to send to europe :) so another source of those parts would be much appreciated

Don't take that as a stress picture. They're just pretty colors. It was the first time I had done the Bezier curves, so I was just looking for something to apply them to.

Putting a fillet where the axle holder attaches to the effector would be a good idea, as would removing those sharp corners.

There's also these more 'traditional' effectors:
http://www.thingiverse.com/thi...
http://www.thingiverse.com/thi...

The second one requiring a different connectio
n.

After some thought, I think I get your point.

I have considered what this joint would look like if I made it from custom milled parts. The axle itself could have a hole in it, and the female clevis attached to the arm could be attached to that hole directly with a pin. that would make it coplanar as well.

For these plastic parts, you're suggesting essentially the same thing. The mail clevis would have to come off the top, rather than the side as well I think.

I'm not quite getting the 'smaller vertices' comment. But, there is a 'hang up' if the angle is too small. I'll print a pair reversed to visualize the solution you're suggesting.

I forgot to answer the stepper question.

I've played with steppers. The problem is slippage. You'll never know when they're in the right place. Of course, you could use steppers with magnetic encoders, and then you'll know for sure what position they are in.

In the original moc up of this thing, I used steppers. It's an expensive way to go with the encoders and all, but yes, that's a viable option, and possibly more interesting if you put a gear on it and a giant gear wheel on the arms.

Thanks for the Kudos. I like those snack bars.

The volume can be anything from 'small' to 'pretty large'. Meaning, given these simple arm designs, the things that govern the space are the lengths of the arms, and the strength of the servos. So, the real answer is, build one of the size you want, then make arms to fit that volume.

Although y
our arms might be physically capable of reaching the far corners of the build space, your real usable volume is probably something like a hemisphere, or cylinder.

and the shorter anser, you'll roughly get something between a TOM and an Ultimaker, if you were making a 3D printer out of this design.

Oh yah, that icosa utilizes your tetrahedron vertex. 36" on each side.

I guess I'm a fan of yours. Please do more stuff!

Well, you know, oddly enough, when you look at my code, you will find some similarities. In fact, I looked at your code. Originally from Trossen Robotics: http://forums.trossenrobotics....

It's very easy code to play with.

The precision is governed by the servos. With these inexpensive hobby servos, about .35 degrees per step. They are very twitchy.

My next version will use the Dynamixels, which are a bit more accurate. I've toyed with using a gear, for greater accuracy, but th
at can be for another revision.

This one is good enough to do some pick/place, and scanning work. Not accurate enough for .25mm extruded printing.

Video - next on my list

Firmware - Right now, the only thing I've coded up is the demo with the wiimote. I figure it might not be too hard to use a gcode parser for the Arduino.

The machine itself won't generate gcode. Rather, it will read gcode, and provide movements. There are actually two parts to the software (at least that I'm working on).

One part is the raw movement (moveby, moveto, speed, etc). The second part is tied to whatever tool you so happen to attach t
o the effector. So, in the case of a 3D printer, you'd have the appropriate stuff to slice a .stl, and run an extruder.

If you switch out the tool, and put a laser scanner on the effector, for instance, you'd load up a laser scanner bit of code (and the raw code running the servos would remain the
same. If that makes sense.

Way more accurate, and very repeatable, although the demo software in the pack is a bit hacky as I was messing around.

This is good enough of a platform to begin worrying about the software instead of the raw hardware. Very easy to crank these out in minutes and hours, depending on how much raw materials you have.