Cartbot with all three axes controlling the toolhead - why not?

This is a question that has been bugging me for a bit, and it should be simple to answer. Everywhere that I've seen discussion of cartbot options, the three listed all have at least one axis of movement for the build platform.

Why? Is there something intrinsically wrong with moving the toolhead on three axes and leaving the build platform immobile?

Edited 1 time(s). Last edit at 08/05/2008 11:47PM by Joshua Merchant.

No, that would work fine. I have a friend building a RepStrap with that concept.

You'll also notice that there aren't many examples of designs that move the table in all 3 axes. If you split it between the head and table you're slinging around less stuff in general, because no axis is carrying the weight of both of the others (if that makes sense).

Interesting. And yes, that last part does make sense.

The thing about moving the table about in 3 axes is that the table is generally much larger than the toolhead.

Personally, I think it would be best to move only the toolhead as the toolhead should generally be much lighter than the build platform and build object (as far as I know). If we end up making a really heavy toolhead the reverse might apply, but I doubt this.


One interesting idea that occured to me is that if you have the toolhead moving, then the build platform is essentially not a major part of the contraption. You could print on a floor, I suppose.

That led me to thinking what would happen if you just left out the build platform.... and put the toolhead cartbot on wheels.
If you give it a z-axis with an extremely wide range (maybe telescopic or something), and allow the wheels to climb over objects (like the build object), you could build objects of much greater size than the cartbot itself.

It might be a good idea to place paper (or rather, thin films/sheets of stuff that the build material will stick to, instead of using a giant build platform) on certain spots where bottom supports will be placed (you might be able to have spaces between the sheets - if you use overhangs and gap-spanning, you can build a small foot which connects to other feet through the air and holds up the build object). It might be interesting to have the cartbot dispense this film itself (either through a separate dispenser tool or, perhaps, through extruding directly on the floor; for extruding directly on the ground, the contraption would need a closed-loop system with sensors to sense the shape of the extruded platform, so that it could make it flat on top).


Essentially you'd have the toolhead moving on the build platform in two dimensions (the ground). You'd still have the precise cartbot control of all three dimensions, but the fact that the cartbot would be moving on top of the build platform extends the build space much more than just a cartbot would. To have a standard cartbot with the same build area, the cartbot would have to be larger than the build area... so if you wanted a 10'x10' build area... you get the idea.


Anyway, this idea may become more useful after we perfect our current toolheads and add more toolheads to our collection.


[Now that I think about it... maybe a moving cartbot isn't the best idea. If it's going to be a mobile bot like that, perhaps a different style for the precision movements would be better (like a 6-axis robot or something). Anyway, it was a funny enough idea to imagine a malfunctioning Darwin rolling out the door and down a sloped street that I figured I should say something about it.]

Haha, remember also that you'd need some form of highly accurate position/orientation feedback for the cartbot as well relative to an immobile point, which may be difficult to get working.

But I for one welcome any modifications that get us one step closer to ubiquitous grey goo! Nice!

The overhead cartesian design is very scalable in size.
The US navy uses a multi-ton overhead gantry CNC plasma cutter to make ships.
It rolls on wheels, and the weight makes it stable, and despite
it's massive size it is driven down its tracks by a threaded rod.

Does anyone have ideas on specific setup? The best thing I've come up with so far (now, I didn't say it was "good"; "best" is relative!) is a leadscrew design which has the z-axis attached to the toolhead, the y-axis attached to the z-axis, and the x-axis attached to the y-axis (of course x/y are interchangeable; I imagine this would be in firmware or software).

Basically, the toolhead mount can be imagined as a plate with a motor mounted on the top, with the shaft pointing upward. The shaft connects to a leadscrew via a coupling (perhaps like the silicone tube couplings used in the Seedling's McWire). The leadscrew threads through a captive nut (as in the McWire), which is mounted on/in another platform (which is more cube-shaped than sheet-shaped). When the motor is powered, the leadscrew rotates in the nut and moves the plate up and down. Now imagine a square drawn on the plate with the motor and leadscrew in the center. At each corner is a rod, the end of which is connected to the plate. These are guide rods, which go through holes (bearings/bushings) up in the cube-shaped platform with the leadscrew, guide the plate, reduce the x/y strain on the leadscrew. They might also need to go through another plate at the very top to guide the bottom plate (the one with the toolhead) better.

So that's the Z-axis. The Y axis is similar, except that the captive nut and motor positioning are reversed: the captive nut is on the moved platform, and the motor is on the stationary. This is more like the axes on the McWire, and it is preferable because the motor does not have to carry itself as part of the load.

The downside about it is that the thread has to go all the way through the load, meaning it pokes out the other end and extends all the way across the full range of movement; if the Z-axis was like this, the screw would reach all the way down to the build platform, and collide with the build object, etc. I'm not sure how to avoid this (other than displacing the z screw to the side of the build platform and using guide rails), and that's why I put the z motor on the z plate (it seemed like the simplest design).
[I took a look at Darwin's cartbot, and I was a bit surprised. Am I correct in thinking that it just uses a leadscrew system for the Z-axis which moves the platform? It seems to me that the motor is connected to one leadscrew, and the belt connects the one leadscrew to the other three... is that right? Interesting design.

Anyway, so the Y-axis leadscrew's captive nut is in the same cube-shaped platform as the Z's captive nut. It should be noted that this cube-shaped platform is at the top of the cartbot (though the Z-axis leadscrew and stabilizing rods extend through and over it). Anyway, the Y leadscrew threads through this captive nut, and is attached to the Y motor, which is attached to another platform. When the motor is powered, the motor remains stationary, and the leadscrew moves the captive nut (and the platform it is attached to, which is attached to the z-axis and the toolhead) back and forth along the Y axis.

The platform the Y motor is attached to has bearing/bushing, through which is placed a rod, which is a nonmoving stabilizer attached to the top corners of the cartbot (sorta like the Darwin's corners). (I later refer to this rod as the Y-rod.)

The X-axis captive nut is also in the cube shaped platform with the Y and Z captive nuts (they're all having a lil party in there). The X leadscrew threads through it, and connects to the X motor mounted on a platform. When the X-motor is powered, the motor remains stationary and the leadscrew moves the captive nut (and the central platform it is attached to) back and forth along the X-axis. It is essentially the Y-axis rotated 90 degrees.

When the x-axis is powered, it moves the X-captive nut. This transfers the X-motion to the central platform, which transfers the X-motion to the Z-axis, which transfers the X-motion to the toolhead plate, thus moving the toolhead along the X-axis. Not only that, but when the central platform moves along the X-axis as described, the X-motion is transferred through the Y-leadscrew to the Y-motor and the Y-motor's platform, which then slides parallel to the X-axis along the aforementioned Y-rod.

Similarly, the X motor is attached to a platform, which has a bearing/bushing in it, through which is a rod attached to the corners of the top of the cartbot. This X-rod is just like the Y rod, except rotated 90 degrees for the X-axis.
So when the Y-motor is powered, it moves the X-captive nut, which transfers the Y-motion all the way down to the toolhead (as I described for X-motion) and also through the X-leadscrew to the X-motor and the X-motor's platform, which slides parallel to the Y-axis along the X-rod.

It would also probably be a good idea to have the X and Y leadscrews stabilized by platforms on the opposite ends of the motors, where they would otherwise simply be sticking out. These extra platforms would be similar to the motor platforms, with the rods to glide along when the perpendicular axis moves, except without the motors. To attach the end of the leadscrew to these platforms, they could either be directly inserted into a bearing/bushing, or coupled to a smooth shaft which would then be inserted into the bearing/bushing (I'm not sure if the threaded-ness of the screws would lead to less friction due to less surface area contacting the bushing, or if it would create more friction due to some other problem).


Now, so far, I've only painted a picture of an XYZ cartbot which is hovering in the air. To obey the rules of gravity (I'm such a boy scout), those "top corners of the cartbot" (to which the X and Y rods are attached) would be large enough to be attached to a support structure. I suppose the most basic structure would be making it a cube (or rectangular prism) by connecting vertical rods to each top corner, and attaching the bottom of the rods to similar coners, which have horizontal rods forming a square at the bottom. For more stability, I think we could use two diagonal rods on each of the prisms' square faces. Or perhaps cut the middle of the rods and stick them into a piece in the middle to connect the diagonals together (though I suppose you could just wrap them with tape or something). If more structural strength is necessary, I think we could use those triangular shapes they used to use in skyscrapers.... or something. I don't know much about structural engineering.

Of course, there's always the option of using a flat plate to cover and support the entire side, instead of just spanning it with diagonal rods.

The build platform would be a stationary plate either attached to or resting on the bottom of the frame. I think the best option would be to have a plastic platform across the entire bottom, which has clamps at the edges that allow quick fastening and unfastening of the build plate. This would allow for quick setup, quick removal of the build object, and quick change between build platform materials (for testing the differences between build platform materials, as well as changing between extruded materials which may require different build platform materials).


What's really interesting about this is that it is very scalable. If you, instead of putting rods into the upper corners, attach them to one of those string positioning systems, you might be able to get a cool room-sized RepRap. Alternately (and this is something I might actually try), instead of attaching the upper corners to the string system, attach them to a larger overhead cartesian bot, like the giant US Navy one mentioned by Tim (though probably not so big, but you get the point).

If the larger machine had less accuracy in positioning, then you could move the larger cartbot into one position, lock it in place, then use rangefinder-type devices (either light or sound based) to measure the actual position, then begin printing with the smaller cartbot. When you're done printing in one volume, the larger machine is unlocked and moved in a 3D gridlike fashion to each cube, locked, ranged, printed, unlocked, etc, until the entire build is complete. It's rather simple really, and the rangefinding equipment can be made for under $100 (I think, but I suppose it does depend on the accuracy and range required).


It should be noted that any part of a RepRap cartbot that is too large to be printed directly can be split into smaller pieces which can then be put together (with various types of permanent and non-permanent joints) to form the larger part. This may seem obvious, but it means that the larger machine could be created by the smaller. (I wonder if larger pieces can be made by putting separate smaller pieces in contact and creating a mold from that which could be used to cast the larger piece?)



Anyway, any comments/improvements/alternatives? I know I rambled a bit, and there's no illustration (I'm not sure what exactly I should have used to create a visual aid for that; what do the other guys use for those 3D models? Just AoI and stuff?) but maybe you survived .

my upcoming cartbot
still without leadscrews (I guess you'll figure where they'll go
The two leadscews for the x-Axis are going to be connected by a timing belt.

Oh, it'll have a baseplate that I didn't include in the picture, just a plate, nothing special


'sid

sid Wrote:
-------------------------------------------------------
> my upcoming cartbot
Nice.
> still without leadscrews (I guess you'll figure
> where they'll go

> The two leadscews for the x-Axis are going to be
> connected by a timing belt.
Yeah, I figured I would have to do something like that if I didn't do it the way I did it (where the x axis moves the y axis and the y axis also moves the x axis - that is, they affect each other).
I think it's a tradeoff:
In my design, the Y motor uses more energy as it has to move the mass of the X-axis (in your design it doesn't), but it only uses one leadscrew without a belt, and the motor shaft is in the middle of its axis (which I think might be a slight structural advantage).
In your design, the Y motor uses less energy than the X motor (in my design they use the same amount of energy), but requires two leadscrews and a belt.
Something about the offset leadscrew unsettles me, but it should be fine (for practical purposes, the main reason I wouldn't go with that design would probably be the need for the belt, which makes the whole thing a bit more difficult to design and replicate).
> Oh, it'll have a baseplate that I didn't include
> in the picture, just a plate, nothing special

>
> 'sid
How do you do your z-axis? From the picture it looks like you mount the motor on the top platform, put the captive nut in the middle platform, and then connect the end of the leadscrew to the bottom (toolhead-mount) platform in such a way that it can rotate freely in the connection but still allow the leadscrew to push and pull the platform back and forth (if that makes sense).

If this is the case, then it's basically the reverse of my z-axis, which has the motor on the bottom side. In yours, it looks like the powered motor will move the bottom and top plates together up and down relative to the middle platform.

The advantage of having the motor on the bottom is that you don't have to have the top platform and you don't have to have the complicated connection to the bottom plate (as it's reversed, you might think you'd have the complicated connection at the top, but I removed the top platform altogether so the connection is no longer necessary - you would just have the leadscrew sticking up into the air, or perhaps through a stationary bushing for stabilization).

Unless, of course, I'm mistaken about your Z-axis operation.

Joshua Merchant Wrote:
-------------------------------------------------------
> sid Wrote:
> --------------------------------------------------
> -----
> > my upcoming cartbot
> Nice.
>

thank you
that's my... let me count.. third try to do it this wy, originally to build a cnc milling machine, but that thing just isn't stiff enough (in the desired size (300x400x100 movement) to do serious milling, but it should be sufficient for RepRapping and some pcb milling I guess (I'll see)
Oh btw: it's made of LDPE, so it should be reprapable too...

> > The two leadscews for the x-Axis are going to
> be
> > connected by a timing belt.
> Yeah, I figured I would have to do something like
> that if I didn't do it the way I did it (where the
> x axis moves the y axis and the y axis also moves
> the x axis - that is, they affect each other).

That's a wonderful design, there's a video of s.o. who actually did that using servos, a bit slow.. I can't find it anymore...

> I think it's a tradeoff:
> In my design, the Y motor uses more energy as it
> has to move the mass of the X-axis (in your design
> it doesn't), but it only uses one leadscrew
> without a belt, and the motor shaft is in the
> middle of its axis (which I think might be a
> slight structural advantage).

I'm unsure about that advantage, because of the need of larger parts to be reprapped
with only one leadscrew for X-Axis, I think i'd go for a portal mill and only move the table not the axis.

> In your design, the Y motor uses less energy than
> the X motor (in my design they use the same amount
> of energy), but requires two leadscrews and a
> belt.

True unfortunately, that's why my original intention was using two synchronized Motors for that axis, Maybe I'll try both someday...

> the main reason I wouldn't go with that design would
> probably be the need for the belt, which makes the
> whole thing a bit more difficult to design and
> replicate).

True, but darvin has lots of belts and pulleys, I'll have one belt and two pulleys, so I think that's still easier than the original darvin for now

> How do you do your z-axis? From the picture it
> looks like you mount the motor on the top
> platform, put the captive nut in the middle
> platform, and then connect the end of the
> leadscrew to the bottom (toolhead-mount) platform

True Motor on Top, Nut in the Middle and Toolhead on the bottom platform.

> in such a way that it can rotate freely in the
> connection but still allow the leadscrew to push
> and pull the platform back and forth (if that
> makes sense).

It does, it surely does

> If this is the case, then it's basically the
> reverse of my z-axis, which has the motor on the
> bottom side. In yours, it looks like the powered
> motor will move the bottom and top plates together
> up and down relative to the middle platform.

That's what it was made for, yes.
In fact I'm a bit unhappy about having the Motor lifted too, but In a previous version with a different z-axis I figured that it makes more sense than to move the center (you'll need a long plate where the toolhead is mounted to have enogh space for the just printed material)

> The advantage of having the motor on the bottom is
> that you don't have to have the top platform and
> you don't have to have the complicated connection
> to the bottom plate (as it's reversed, you might
> think you'd have the complicated connection at the
> top, but I removed the top platform altogether so
> the connection is no longer necessary - you would
> just have the leadscrew sticking up into the air,
> or perhaps through a stationary bushing for
> stabilization).

Well yes and no.
No, the connection isn't complicated at all, the rods are grubscrewed, and the Leadsrew runs in a skatebearing (locked by a nut). That's it, no magic.
unsupported ends of leadscrews and rods... I'm not a fan but it'll work I think.

Only thing that I'm wondering is:
How do you attach the motor? turned at 180

Edited 1 time(s). Last edit at 08/27/2008 12:34PM by sid.

Joshua Merchant Wrote:
-------------------------------------------------------
[snip]
> It's rather simple really
[snip]

LOL.

Sorry, as you were.

sid Wrote:
-------------------------------------------------------
> I'm unsure about that advantage, because of the
> need of larger parts to be reprapped
What larger parts? (Anyway, I don't personally see larger parts as a problem, it just takes a bit more printing time.)
> with only one leadscrew for X-Axis, I think i'd go
> for a portal mill and only move the table not the
> axis.
There are a few reasons I choose not to move the table, which is actually the subject of this thread (see the title). The main two reasons though, are that I figure the toolheads should be lighter than the largest build object and that overhead cartbots are more expandable and versatile, even if they are perhaps not rigid enough for milling (though I think with proper structural reinforcement it could be possible [it should be noted that I don't intend to spend time making it stable for milling]).

> True unfortunately, that's why my original
> intention was using two synchronized Motors for
> that axis, Maybe I'll try both someday...
I would think that two synchronized motors would be even more complicated than using a belt. In any case, though, it would probably be more expensive.

> True, but darvin has lots of belts and pulleys,
> I'll have one belt and two pulleys, so I think
> that's still easier than the original darvin for
> now
One of the reasons I'm looking to design my own cartbot is that the belt/pulley systems the Darwin uses seem a bit overcomplicated. That aside, adding a belt and pulleys where they are not necessary is... not necessary?


> True Motor on Top, Nut in the Middle and Toolhead
> on the bottom platform.
>...
> It does, it surely does
>...
> That's what it was made for, yes.
> In fact I'm a bit unhappy about having the Motor
> lifted too, but In a previous version with a
> different z-axis I figured that it makes more
> sense than to move the center (you'll need a long
> plate where the toolhead is mounted to have enogh
> space for the just printed material)
Actually, you could have the Z motor stationary relative to the Z axis by having it rotate the captive nut in the middle platform instead of the leadscrew. I opted out of this because it increases complexity and I don't expect the Z motor to be very heavy compared to the toolhead assembly.

>...
> Well yes and no.
> No, the connection isn't complicated at all, the
> rods are grubscrewed, and the Leadsrew runs in a
> skatebearing (locked by a nut). That's it, no
> magic.
Yeah, (and there are actually simpler ways to do it), but you still have a connection; why have a connection (which takes money/material and time resources) when you don't have to?
> unsupported ends of leadscrews and rods... I'm not
> a fan but it'll work I think.
They don't actually have to be unsupported; it's just that they would be supported by supports which are stationary relative to Z-motion.
> Only thing that I'm wondering is:
> How do you attach the motor? turned at 180

starting from the end

Quote

Well, the Z motor is mounted on the top of the bottom platform, with the shaft pointing up.

Oh.. okay
Then there will be a gap between the lowest point of the leadscrew and the carriage for the toolhead.
thinking of the dimensions of my motors and my leadscrews that'll be about 15cm
(nema 23, ~8cm long... 4cm shaft 3.5 cm coupling) that's what you'll loose on the traverse path.
My rods and the leadscrew for the z-axis need to be 14cm longer than the desired traverse path (3cm motormount, 8cm bushing and nut housing, 3cm tool head mounting) adding another 15cm for the motor and coupling I would end up with rods of 40cm to have 10 movement.
(for comaprision: y-axis 44cm to gain 30cm movement)
Well of course you can use much smaller motors than I have, reducing the loss, still you'll have to cope with some serious gaps making sure the motor not touching anything of the structure.
I'd love to see a picture of how you did it.
I'd love to see pictures of the whole cartesian to be honest

Quote

They don't actually have to be unsupported; it's just that they would be supported by supports which are stationary relative to Z-motion.

Then they ARE unsupprted, the bushings and the captive nut, do not support.
You need something on both ends of the rods and screw, to prevent distortion;
or something really big on one end with small length of rods. It'll work without for printing at least I guess so. But still: I'm not a fan

Quote

Yeah, (and there are actually simpler ways to do it), but you still have a connection; why have a connection (which takes money/material and time resources) when you don't have to?

Well, You'll need to have a connection much more complicated than that because of the need to mount the motor to the baseplate.
If you need less than three holes, two screws and one nut let me know

Quote

One of the reasons I'm looking to design my own cartbot is that the belt/pulley systems the Darwin uses seem a bit overcomplicated. That aside, adding a belt and pulleys where they are not necessary is... not necessary?

Yes, pulleys are hard to fabricate and expensive to buy.
And reduce them where they are unnecessary is the best thing you can do reducing the costs and hassle. But for my cartbot I need to have either a fourth motor or one belt, no matter what I do.

Quote

I would think that two synchronized motors would be even more complicated than using a belt. In any case, though, it would probably be more expensive.

That's what I thought too, that's why I'm with the belt

Quote

There are a few reasons I choose not to move the table, which is actually the subject of this thread (see the title). The main two reasons though, are that I figure the toolheads should be lighter than the largest build object...

Because of the threadtitel I posted my cartbot
The toolhead is indeed lighter than the largest printed object,
but, is the y-Axis plus the z-Axis plus the toolhead ?
I do not think so! So if you want to move less mass with the motors, I think the best way is to have an x-y moving table and an extra z-Axis... just like nopheads hydraraptor (because most printed parts are much lighter than the z-axis itself)
So this is definately not a pro for this type of cartbot.

Quote

What larger parts? (Anyway, I don't personally see larger parts as a problem, it just takes a bit more printing time.)

Well, both ends of the y-Axis and the captive nut housing should be connected, so that's a part of 300mm length at least, that's much larger than anything else yet. Not to talk about precision here, that's not easy I'd guess.
In fact that's one of the reasons I split the x-Axis and use two leadscrews

'sid