Cartbot with rotary and linear axes

I couldn't get to sleep, so I came up with an idea for a cartbot which can change the angle between the extruder and the build object to overcome the 45

I made a comment about 4+ axis printing a little while ago: [forums.reprap.org]

To summarise; the problem isn't a mechanical one, its a software one. Calculating toolpaths when you add extra axes is going to be very, very, very hard.

I think just making generous use of scaffolding parts that can be easily trimmed away is the best direction at this point. I've been fiddling with my own toolpath generator a little, following from the docs on the wiki, but they stop dead when the process got to [reprap.org] and I've been to busy to work the whole process out for myself or to reverse engineer the host. I'm sure one day I'll get around to sorting it out, and making scaffolding generators

Ru Wrote:
-------------------------------------------------------
> I made a comment about 4+ axis printing a little
> while ago:
> [forums.reprap.org]
Ah, yes. You're that person I was referring to in the OP.

> To summarise; the problem isn't a mechanical one,
> its a software one. Calculating toolpaths when you
> add extra axes is going to be very, very, very
> hard.
Just as I thought. But then, "hard" doesn't mean "impossible". It just means "expensive" .

> I think just making generous use of scaffolding
> parts that can be easily trimmed away is the best
> direction at this point. I've been fiddling with
> my own toolpath generator a little, following from
> the docs on the wiki, but they stop dead when the
> process got to
> [reprap.org] and
> I've been to busy to work the whole process out
> for myself or to reverse engineer the host. I'm
> sure one day I'll get around to sorting it out,
> and making scaffolding generators
The problem with using scaffolds (nice word choice) on the standard 3-linear-axis cartbot is that you have to directly support the overhang, which means when you cut away a scaffold, you have the imperfection on the overhang. With a 6 axis system (is that the correct terminology?), many of these overhangs won't need supports at all, and for the ones that do, you (or rather, the software) can elect to build them in more optimal positions (to minimize imperfections), because it changes the direction of gravity relative to the build object, and thus changes which part of the object is "overhanging" at any given time.


But yes, the programming will be tough. For the near future, it won't be worth getting in to, but eventually it might be. Basically, it's low priority and high cost, which means it won't be developed until someone with a hefty budget (be that of time or money) and a drive towards perfection finishes with a good majority of the other features.

Quote

when you cut away a scaffold, you have the imperfection on the overhang

A little careful work with a sharp knife should take care of that. Take a look at the sprues and flash you get in injection molded or cast parts, for example. I used to make airfix kits quite a few years back; cleaning up plastic parts it dull, but it can be done quite neatly.

I'm pretty certain that you could extrude the scaffold at a lower temperature too, to reduce adhesion, but I don't know how quickly the extruder will respond to temperature changes and whether this would have an unacceptable speed impact. I'm hoping not, but its an experiment for another day.

Quote

But then, "hard" doesn't mean "impossible". It just means "expensive"

I was thinking that it is a sufficiently difficult problem that the investment of time simply isn't worth it. The fact that commerical rapid prototypers do not use this approach suggests to me that is a bad plan. Support material, once a suitable candidate has been identified, would just magic away all our problems, I'm sure

Moreover, a fairly solid support material would require similar toolpaths and positioning as the scaffold, so scaffold work would have payoffs in the future: support material would have 100% density under an overhang, whereas scaffold would have the minimum required to allow bridges to be built across it.

Quote

With a 6 axis system

I think 5-axes would do pretty much everything you'd really need them to, so piling an extra axis onto a 5-axis one just increases your cost and adds another layer of innacuracy. Hexapod platforms are 6-dof, and more importantly because they use parallel rather than serial kinematics errors in each part don't combine like they do in a 'stack' of axes on a more conventional cartesian-type bot. But hexapods are a different flavour of engineering challenge, and would require newer and cleverer firmware to drive them.

Quote

many of these overhangs won't need supports at all

Indeed. I considered having extra axes before scaffolding, mostly cos multi-axis cnc tools look very cool when they're in action, and I like that sort of complex challenge

It my well be something I look at again in the future (mostly cos I'm interested in continuous 4+ axis movement for engraving various things) but it isn't ever likely to be anything exciting, mainstream and commonplace.

If nothing else, it would increase the complexity of the reprap frame and electronics by 50% or more. The electronics are the most expensive bits right now... even assuming that the components for the frame and motors can be obtained cheaply or reprapped, the L298/297 combo is a pricey one and not easily replaced.

Scaffolds on the other hand just require an extra step in the CAM process, and not a hugely complex one at that. This makes it effectively free and transparent to end users. I've no doubt that someone will be trying them before I get round to doing anything really useful

Another idea would be to have the linear and rotary controls for an axis on the same leadscrew.

Essentially, you have the end of the screw rigidly attached to the stage (I'm using McWire terminology, as I haven't taken a good look at the Darwin cartbot in a while), and at the end of the screw you have a motor which can rotate the screw. Between the stage and the motor, you have a nut threaded on the screw. The nut can be rotated by a second motor (perhaps by having gear teeth on the outside of the nut).

When the nut motor rotates and the bolt motor is kept still (might have to implement some sort of holding/braking system to make sure rotors don't move when they shouldn't), the screw moves linearly, as does the stage.

When the bolt motor rotates, and the nut motor rotates at a speed with a certain ratio to the bolt motor's speed, the screw and the stage rotate without linear movement (if the nut motor was kept still, then there would be a slight linear movement).

Seems simple enough to me. There would be some structural complications (the frame would not be as simple as the first idea or a standard 3-axis cartbot), but they can all be overcome, as far as I know.

Anyway, that's just an idea to simplify rotation and translation along a single axis (as you don't have to split the two up).

That sounds a little... contrived. You'd need a braking nut on the leadscrew, and you'd need some sort of rotating and braking system on the supporting rails for the stage... they'd have to be mounted on a bearing to allow them rotate with the stage, and then locked in position when linear motion was required again. Two new and complicated mechanisms, for the AX axes, which in turn need to support the new and complicated BY axis.

What makes that simpler than having a toolhead fixed in the XY plane, and a gimballed platform with the AB axes attached to the current Z axis platform?

That seems like a logical way to build it to me, its fairly straighforward and you can always detach it from the normal XYZ cartesian bot when trying to handle multi-axis fabrication becomes just too much

Just listing options as they come to mind. Sometimes options that seem more complex can actually be more effective and sometimes simpler in practice (usually due to the effectiveness). For example, given proper geometry, you may not need parallel supporting rails on the stages. It also may be easier than one might initially think to have braking/locking mechanisms.

But yes, at the moment, the idea is not relevant to RepRap.

Quote

you may not need parallel supporting rails on the stages.

So I apply the brakas, and turn the drive shaft 45 degrees to tilt the stage. I then release the brakes. What keeps the stage at 45 degrees?

Ru Wrote:
-------------------------------------------------------
> you may not need parallel supporting rails on the
> stages.
>
> So I apply the brakas, and turn the drive shaft 45
> degrees to tilt the stage. I then release the
> brakes. What keeps the stage at 45 degrees?

The brakes on the bolt motor. Obviously, this would put stress on the connection between the bolt and the stage, and the connection between the brake and the bolt. Both connections would have to be strong enough for the max load on the stage.
There are essentially two obvious ways to solve this: strengthen the connections, or reduce the load. For experimental testing, I would probably start with a small version so I wouldn't have to worry about it as much.


Perhaps a better alternative to a rotating platform in the first place is to combine additive and subtractive rapid prototyping in one unit (or I suppose you could have them separate but used on the same object). For portions of the build object that would be difficult or low quality using additive, you could extrude it solid and then cut it out. It may require that the subtractive toolhead be allowed to rotate, however.

Joshua Merchant Wrote:
-------------------------------------------------------
> Ru Wrote:
> --------------------------------------------------
> -----
> > you may not need parallel supporting rails on
> the
> > stages.
> >
> > So I apply the brakas, and turn the drive shaft
> 45
> > degrees to tilt the stage. I then release the
> > brakes. What keeps the stage at 45 degrees?
>
> The brakes on the bolt motor. Obviously, this
> would put stress on the connection between the
> bolt and the stage, and the connection between the
> brake and the bolt. Both connections would have to
> be strong enough for the max load on the stage.
> There are essentially two obvious ways to solve
> this: strengthen the connections, or reduce the
> load. For experimental testing, I would probably
> start with a small version so I wouldn't have to
> worry about it as much.

Maybe I don't understand what you're saying, but I think what Ru was asking was what happens when you want to move the stage along the axis while it is at an angle? If you rotate the drive axis, the stage just spins instead of translating. If you release the bolt lock so you can translate, how do you maintain your angle?

There are ways to get around this, like having a disk-brake-type system that attaches to the bottom of the stage and helps it maintain its angle, but now you're adding back in more complexity than you eliminated in the first place.

> Perhaps a better alternative to a rotating
> platform in the first place is to combine additive
> and subtractive rapid prototyping in one unit (or
> I suppose you could have them separate but used on
> the same object). For portions of the build object
> that would be difficult or low quality using
> additive, you could extrude it solid and then cut
> it out. It may require that the subtractive
> toolhead be allowed to rotate, however.

This has been discussed. IMHO adding parallel additive/subtractive capabilities increases the complexity significantly while providing little benefit over additive-only.