Cartensian Robot Designs (Pros and Cons)

So I'm looking into building my own RepRap. It looks like there are 2 different designs the Darwin and the McWire. Both have 3 DOF, yet the Darwin has the extrusion head moving in the X & Y and the table moving in the Z. The McWire has the extrusion head moving in the Z and the table moving in the X & Y.

Before I start, what are the Pros and Cons of each system?

The moving head system can make an object nearly as big as its own volume.

The moving table system has needs a floor area four times bigger than the object it can make but is mechanically simpler.

---

[www.hydraraptor.blogspot.com]

I think nophead has it pretty much summed up.

A consequence of this to be considered is loading, particularly in stepper based systems which use optimistic positioning.

Your design ideally needs to take into account the pull-in of your steppers (the ability of the stepper motor to move from standstill the number of steps you have instructed it to move without losing one or two along the way) and which axes are affected by a change in loading.

In a moving head (X & Y) system the X & Y axes are unaffected by how much mass (think weight) the workpiece has. The Z axis however is lifting the work piece up and down.

When you are creating a workpiece by deposition as in this case the mass is increasing throughout the run until at the end you have maximum mass for that particular piece. Your design then for the Z-axis should ideally be able to pull-in (ie move without losing steps) when lifting the highest mass you could possibly think it would ever have to, then add a safety factor.

When you are creating a workpiece by erosion ie milling bits of a lump (or Billet) of material the mass will arguably be greatest at the start of the run and if you are hoovering the swarf away least the at the end of the run. Once again the same design considerations for your Z axis should ideally be taken into account.

In a moving Table (X & Y) system the X & Y axes are affected by how much mass the workpiece has. But in a well designed set of tables not nearly as much as the Z axis on the moving head design.

The moving Table, X & Y axes have to cope with the inertia of the work piece and the friction of the bearing surfaces of the work table/s for a given mass of work piece.

The moving head, Z axis has to contend with the work piece dead weight, inertia and the friction of the bearing surfaces of the workpiece for a given mass.

If you are only depositing using light materials (Extrudable Plastics etc) the above considerations should be minimal and you can go with whichever takes your fancy.

If you are intending using the same equipment additionally for CNC milling of heavier duty materials (Possibly metals and heavy lumps of wood) some forethought and planing could save a quantity of frustration later.

In summary these are the factors to help you choose your horses for their respective courses.
There is no Wrong v Right only applicability to what you want to use it for.

Hope this helps.

Cheers aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Thanks for the responses. I was looking for answers more geared towards accuracy. For my senior project, my team is building a RepRap, but we are focusing on accuracy and ultimately resulting in usable parts in robots and other projects.

We will be looking at the extruder and increasing the accuracy of it while switching to a stronger plastic like ABS. We want the Cartesian part as accurate as possible.

Unfortunately neither is any more or less accurate than the other.

It is all Dependant upon your end use matching your specification matching in turn your design and your constructive accuracy/Precision.

Pick which you fancy, make sure you understand the ups and downs (Pun intended) then the do the best you can with accuracy/precision within the budget you have to work with.

Personally speaking I prefer the X & Y moving the tool head purely from the stand point of there is less wastage in terms of having parts that do not get used/worn. ie your bed is the size of your work area.

I terms of circumstances the Table X & Y is cheaper and easier mechanicaly to start with, So I will be starting this way.

cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

For the reasons stated earlier, moving tables machines tend to be small and large machines need to be moving head.

---

[www.hydraraptor.blogspot.com]

You may want to consider using a CNC-ready Taig Mill or equivalent as your RepStrap.

Pluses - It works out of the box, and you'll also be able to mill aluminum and steel.
Minuses - Somewhat expensive, and if you have a machine shop on hand, it unecessarily duplicates existing resources.

More subtly, using a premade apparatus may defeat one learning portion of the senior project, the machine-building aspect, which tests and develops your Mech Eng. skills. However, it will enable you to get to the cutting-edge research part: making better extuders, getting an inkjet-based head working, material research, integrating EMC and the RepRap controller electronics, etc. (Rather like using a stock vehicle as a DARPA challenge robotic navigation platform, rather than building your own car.)

AegisTalons,

What college are you from? I'm doing a build for Oregon Institute of Technology with another student but it isn't my senior project. Just a one term build. We are focusing on aesthetics and durability in our construction. I'm a sophmore so I don't have so much experience yet. Would be interesting to share notes and see how you solve some of your problems.

Demented

A quick note on the McWire version, At our recent RUG one member brought in his mostly complete McWire. He called it a good learning experience. His main complaint was that there was a lot of slop in the table movement, especially under load. While this probably not an issue for a small additive type machine, it will not scale well or support subtractive work. The model could probably be improved with better bearings and so forth, but pretty soon you're designing a whole new machine.

I don't know if it effects accuracy in the real world, but it would seem to me that with a moving head machine structural load and inertia would be constant, whereas with a moving table both of these would be constantly varying. But I think somebody already said that in more exact terms.

I would note that milling machines move the table, not the head and are very ridged. Yes, the McWire is unsuitable for subtractive work excepting of the lightest material but this is a design problem for this specific setup and not an overall problem for the general configuration.

Demented

Looking at the McWire.

A quantity of slop under load could potentially be removed by by replacing the spring tensioning with something somewhat stiffer.

For example mount one bearing using a central fixing and one opposite with an off center fixing (Think Cam) but close enough to rotate the bearing against the slide (removing slop) before tightening the fixing and clamping it in position.

Alternatively again mount one bearing using a central fixing and one with a central fixing opposite with a small amount of space that would be slop. Place a tyre or sleeve around the second bearing to take up the slop and apply positive pressure to the rail. (Positive pressure = preload)

Something else that occurred to me. The steppers are mounted on flat plates without a 90 Deg support to stop the motor from wagging backwards and forwards. Adding this stiffening might help remove some slop under load too.

This could be worth experimenting with if only for elimination purposes.

Some thoughts and observations for what they are worth.

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Hey,

Our concern in choosing between the Darwin and Mcwire has a fair amount to do with alignment (McWire seems to be at a bit of disadvantage here, offsetting its low cost and ease of construction). Given the nature of how both setups are driven, in which the threaded rod provides a 20:1 geardown, and torque requirements are fairly low, the limiting factor is the quality of construction and slop in the system. Assuming we can deal with slop for a light-duty thermoplastic extruding setup, quality becomes the issue. In this vein, we see alignment as potentially most problematic. Because the McWire is setup from threaded pipes, a true 90 degree fitting of the vertical pipe (given the T's rotational degree of freedom) seems non-trivial. So too with the perpendicularity of the x and y stages. Does anyone have suggestions or experience with proper alignment, for either setup? We're looking to get as much precision as the setup can give, and given the 1/20" x (0.9/360) precision theoretically obtained given ideal movement in the x, y, and z, this seems to be most pressing.

Thanks for the help.

I've not worked with either, so take my comments with the appropriate quantity of NaCl.

You could align the head initially, hen drill both the mounting plate and the pipe it's attached to. Then thread the htole for a bolt, or just drop a cotterpin through, (actually, match the cotterpin and drillbit, so that you need a bit of force to get it through, or else use a conical hole, and a conical pin.) Personally, I think more accuracy would be had with a bolt through the plate, than a pin, of any shape.

If you do this, you might also be able to rig an interrupted screw rig. This would make changing heads easier, while still improving alignment. You could, conceivably, grind off just over half the thread of the pipe, doing likewise with the inside of the mounting plate, (in quarters, first and third quarter ground out, leaving second and fourth to hold, alternately in sixths, again alternating between thead and gap.) You could then do the same with the bolt.


Another idea, if you're only concerned about ease of aligning it, and not the permanency of the alignment, grind a groove from plate to pipe so you have a pair of lines to match up.

Just a couple ideas to throw around. Use them, or discard them.

Thank you Sean for your suggestions. My question was more about that initial alignment, though. We had planned on tack welding or, as you said, pinning the pipes at their fitting, but this is useful only once initial alignment has been achieved. We've got a laser angle finder in the lab which is good to 0.1 degrees, For a 12 in segment of pipe, 0.1 degrees loses about .02", or 0.5mm, assuming we can get it that good. Essentially, the part is extruded at a bit of a slant in relation to the .STL specifications, exceeding tolerances the larger the part gets in the z. I basically was curious if people still had good success with only moderate efforts at alignment, such as with a square, table alignment, or other methods which they wish to share. I have seen references to alignment on some forum posts and builder blogs, but never any detailed discussion or quantitative analysis. Also, I was curious if the Cartesian setup for the "Darwin," due to its boxy nature, could be assumed sufficiently aligned so as not to be a constraining factor, and if anyone had any idea about to what degree. Obviously it's going to be dependent on your quality of construction, but assuming the kit is assembled with a moderate degree of workmanship.

Temporarily mount a rod alongside the extruder in perpendicular, make sure the robot is plumb, then make sure the rod is.

Do the same thing with a slide projector. Mount the rod so it hangs down in front of the arm, then project light on it, and check to be sure the rod shadows the arm down its length.

How about using a magnetic degree gage. Real cheap from Ace Hardware. And you can tell if it is off by even 1/10 of 1 degree visually.



The frame is setting on a table saw. Table saw is horizontally flat by gage. The frame is aligned to the flat table surface and the horizontal pipe is aligned for zero pitch.

That should be straight enough to fix it so that it can be tack welded to shape.

Edited 1 time(s). Last edit at 01/31/2008 10:25PM by Robert Teeter.

---

Bob Teeter
"What Box?"

Bet you could get even better than 0.1 of a degree if you added a vernier scale. Grin

Anyone seen the vernier clock someone made on t'internet. Cool toys.

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Another difference between moving head and moving table:

With moving table, if the table is not exactly perpendicular to the z-axis the object will have slanted verticals but be the same height all over. With moving head it will be taller at one side than it is at another. I.e with MT the distance to the table is always constant but with MH it varies with XY position (if it is out of alignment).

Of course you can always map inaccuracies and compensate in software as long as they are repeatable and you have suitable measuring equipment.

At the moment the tolerance of the filament and the extrusion process and shrinkage far out way any error in axis alignment achieved with a giant try square.

---

[www.hydraraptor.blogspot.com]

I have some suggestions for alignment issues with the McWire.

For my setup, the alignment is truly trivial. The machine can be quickly and easily put into shape with a machinists square of appropriate size and a height gage.

Start by lineing up your x/y axis. The only things that must be lined up here is your two rails which the tables clamp to via your bearings. These two rails must be perpendicular and flat in relation to each other. Flatness is easily fiddled into shape with a height gage or other such device and a bit of tinfoil or other thin slices that can be placed between a couple washers on each bolt. Further simplifying this step is the fact that we only need the rails to be "flat," i.e. the same height, in relation to each other. So simply take height measurements of each and adjust them with the tinfoil so that they have the same difference at all points tested. Perpendicularity is obtained by clamping a flat plate to the x-axis rail and using a machinists square to extend a perpendicular out from the y-axis rail. Adjust the y-axis rail until you cannot see any light between the plate and the machinists square. They are now perpendicular.

The reason that it doesn't matter if they are flat in relation to anything else is that they move uniformly in a given plane to which we can square the rest of the machine. Obviously it is desirable to have them "eyeball" correctly so that things don't look screwy even if they print straight.

Now take the z-axis plate and true it up with the y-axis table. This can be done by pushing the vertical pipe backward and forward and by rotating your flange. Place your machinists square on edge against the x axis table with one end pointing up and turn it so that the upright is against the z-axis plate. Since we can assume the y/z axis tables have an acceptable degree of flatness, lining the two up should give acceptable tolerances. Adjust the vertical pipe and the flange so until you cannot see any light between the table and the machinist square. On a milling machine this is used to get--as my shop teacher says--to within .0005" of true. That's pretty good. Do this along the face of the z-axis plate and along the z-axis rail for the remaining 2 degrees of freedom.

That's it, you're done.

Hope that made sense. It's hard to explain without pictures or a mill and square to show you.

Demented

Guys your observations on alignment have been pretty timely.

I have been agonizing over the McWire frame for a short while but couldn't work out what was bothering me.

You guys have pretty mush summed it up and that is maintaining alignment whilst under load and being subject to a lot of vibration if milling. (Yes I want to be able to mill things as well)

Particularly after seeing the footage of a CNC Taig milling machine being used as a lathe for small parts. Cool or what.. (Thanks Viktor)

I have opted to go for a frame made using Galvanised Key Clamps and Galvanised Malleable Iron Tube.

[www.fisheralvin.com]

No threading needed, but costs a little more. Should be somewhat more robust (And I can reuse the clamps for something else later if I change my mind)

For the Z axis support (I might make it X as well, not sure yet) I am going to make a bridge arangment (Think of a set of Soccer Goal posts with an extra cross bar. SO it is supported at both sides not just one.

This would be very difficult to do with threaded tube. The robustness for the clamps should also remove most of the alignment issues except longitudinal twist which I should be able to get pretty much ironed out using a flat table (The kitchen one).

I am putting my order in tomorrow and will post some piccys when the frame is put together.

My plans (All being well) are to make removable tool heads for milling (Dremmel and Small Router) and Extruding. I would also like to use this to strap my self a Reprap.

The original strap can then remain a milling/lathe machine and the extrusion tool head will move onto the dedicated RepRap.

I guess I will be able to experiment then with the servo drive too. Yippeee.

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Andy - here is a book that I have that might save you some time in building your new machine.

[www.amazon.com]

Homemade CNC machine.

You can also use the gen 2 electronics with this system and a parallel adapter board.

Edited 1 time(s). Last edit at 02/04/2008 06:14PM by Robert Teeter.

---

Bob Teeter
"What Box?"

For those of you that worry about how to make your RepRap/RepStrap create a precise parts. You may want to take a look at this information. "Volumetric Compensation" was brought to my attention because of another product that I use for Wood Working. The "eCabinet System" uses a Thermwood CNC router system for making precision wood parts. They have used this method to correct for non-solid frame construction in relation to the moving tip of the router or in our case the tip of a moving extruder. This method should work with our systems I would think.

[www.sheffieldmeasurement.com]

---

Bob Teeter
"What Box?"

Hey Guys,

First off great response from the community. I've been busy with alot of projects. To answer Demented Chihuahua, I go to Carnegie Mellon University in Pittsburgh. Senior Capstone currently is one semester class.

Update: My partner and I chose the McWire setup because cheaper and less machining (Cartesian has a multitude of joints that would suck up time machining). We have the electronics finished except for 2 or so resistors that we cannot find in our Mouser order. As for the Mechanics, we worked on the McWire setup last night. We chose galvanized nipple pipe from Home Depot. We have a digital angle finder in the FSAE (Formula SAE) garage. We used the mill's table for a very flat and level surface. Used the angle finder to get to .1 degree accuracy. To keep the pipe from moving, I tack weld the pipes.

As a side note to tack welding galvanized pipe, it sucks. Even when you scrub your surface clean of the galvanization and acetone it; the galvanization still smokes up badly, and you get a lot of splatter even on a low amperage.It still welds its just not pretty at all. Not to mention doing a tack weld from the threads to the angle pieces is difficult and takes a few practice runs to get it "right".

As for the flange, we wanted to machine it down to a perpendicular surface relative to the base of the McWire. When we tried it, we ghetto mounted the McWire into the vise and got it level using the digital angel finder. Under a light cut the flange was rotating and then one of the tack welds broke. I'm going to look into mounting it to a machined cube, or just not worry about it at all.

For stabilizing your joints if welding is proving to be tiresome you might want to consider aligning it, clamping it and then drilling a hole through where you can hit both bits of pipe (Union and threaded pipe end). Then put a self tapping screw through to hold it.

It doesn't need to go all the way though both sides. Just one side but both bits of pipe.

The advantages (although it is a chew to drill) is that you can take it apart for transport and when you want to put it back together it should self align....

cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Aegis,
I don't know if you know this already, but I have heard that fumes from welding things galvanized with zinc can make you ill.

[en.wikipedia.org]

---

[www.hydraraptor.blogspot.com]

Fumes from welding just about anything can make you seriously ill. Whereas I am certain that zinc fumes would be very bad for you, just the fine dust that welding creates can really wreck your lungs. I know that I wear a really high quality dust mask on the rare occasions where I have to weld something.

Probably even less well known is the fact that brazing can also cause respiratory problems as well. Using copper/phosphorous brazing rods is particularly problematic. You always need to do that kind of work in a very well-ventilated environment.

i was thinking about using threadlocker... that or some jbweld =)

I haven't found a need to lock it yet. Let's me swivel the head this way and that to work on it. Very hand so far. Just needs to be tight enough not to wobble around.

Demented