A deviation

Hey all...

Just so you know: I don't know a bloody thing about Rapid prototypers. I've been reading the blog for a long time, but I've never actually seen a rapid prototyper in person. In many ways, this is kind of a neat position to be in, because I'm pretty much a clean slate. (That said, sometimes I ask stupid questions simply because I don't know better)

I was thinking about the designs I've seen so far, and they've all been cartesian-like designs. XYZ axis. The problem is, making something reliably square can be a pain. I was trying to think of methods of simplifying the whole process, and I think I came up with something at least mildly interesting.

The idea is fairly simple: work over a rotating turntable, and have the extruder head only move on one axis. That way the extruder head wouldn't be a whole lot different than an injet print head. You could rotate the turntable with a high amount of precision... and so long as the extuder head and the top of the turntable were level, things would print out right. Also, finding the center of the turntable would be pretty easy.

This would be exceptional for round parts (gears anyone?), but I think that even though you'd be working in radial pixels you could do a pretty decent job of creating square objects (so long as the movement of the head is very fine).

The other idea (actually the first) was an inverted pyramid (extruder head held by three rods over the working area). On further thought though, I think it'd just be more complex.

I'd love any feedback on my little thought experiment.

~D

We've talked about doing a turntable design quite a lot. To the best of my knowledge, however, Hans Wargers is the only guy who actually built one.

[staff.bath.ac.uk]

I don't think he ever carried through with his design, however.

Yup, and here is one earlier thread you can read on the subject:

[forums.reprap.org]

personally, i think the turntable is a fantastic design idea. for various other reasons, we've decided to go with an XYZ cartesian bot for v1.0. however, i really hope someone else tries to reprap a rotational bed once we get things launched.

Yeah, I certainly intend to build a copy of my "briar patch" design eventually.

[builders.reprap.org]

Mine has four axes, xyz and a rotational capability around the z axis. A briar-patch Tommelise would have a much smaller footprint for the same working volume than the Reiyuki/CNC design I'm using now. As well, as we all know, it could do a much better job at doing solids of rotation.

Yes, I did quit with it because you need a drop on demand printhead for it, you can not use an extruder head for it and
it seems to me (i did try) that it is almost impossible to make such a head simply eand easy.
That is the reason why i now think that the XYZ cartesian bot is the best choice.

I think it's necessary to be able to move the axis of the turntable, otherwise you can only make concentric circles, but no ovalshapes nor two circles or anything else.

So..... It's quite entertaining to think about a turntablesystem, but in fact it's mor difficult to handle than an ordinary 2.5d system.

'sid

"otherwise you can only make concentric circles, but no ovalshapes nor two circles or anything else. "

That's like saying a cartesian system is only good for making straight lines. While in some ways true, due to the scale of what we're working on, round objects are possible. The reverse is true as well.

You can make just about any shape with a turntable system... You just have to keep in mind that you use radial pixels/addressing. If you can arbitrarily reach any part of a plane, then you can create any 2d shape on that plane.

~D

"You can make just about any shape with a turntable system... You just have to keep in mind that you use radial pixels/addressing. If you can arbitrarily reach any part of a plane, then you can create any 2d shape on that plane."

If I understand your argument correctly, it also applies (or equally does not apply) to a cartesian system, which invalidates the whole need for a turntable to begin with.

Unless, of course, the turntable system ends up being simpler/cheaper to construct, easier to calibrate, able to achieve higher accuracy, speed, durability, etc. Moving from linear to rotational motion can have more implications than just shape of output.

I dare say, a rotational system could be much simpler hardware-wise than the current cartesian approach. For example, you could do it while only moving the extruder head along a single axis. There's something to be said for that.

Just because I'm looking at an alternate approach to the same problem does it mean that either approach is invalid. The design will simply have alternate constraints and implications.

~D

Yes, sorry draco -- I never meant to say that a turntable approach was unthinkable -- just that the previous arguments I've seen would no longer hold water if the penultimate argument here (which I previously replied to) was true.

dracolytch,
well of course you can make any shape, as you can on any ordinary 2.5d system...
but if you want to make any circle that is ofcentered the turntables axis, you need to lock that axis! (in other words.. it's useless therefor)

cheaper to construct? an additional axis? I don't think so.

'sid

My design was not the one with an additional axis, but to replace XYZ with AngleDistanceZ. By doing so, you can eliminate a large amount of mechanical complexity, making it cheaper to construct.

~D

There is at least one good reason to not do a Theta YZ layout instead of the current XYZ design given "Darwin 1.0" electronics.

Taking any axis to a rotary axis will require accurate parts. Backlash becomes a more significant concern (especially if you build something that's distant from the center) as does gearing torque to get you the oomph to accelerate that interial mass of the turntable. This also applies to adding any rotary axis that reverses.

The cheapest way to get around this requires a closed loop controller that uses a small idler based position sensor that friction rides the edge of a circular base. In the end, this should be more complex mechanically than the XYZ layout, but not significantly so. You'll get better resolution towards the build table center, and possibly worse torwards the edge. Mechanical slop will probably cause some different but probably no worse accuracies.