Pro's and Cons of various cartesian gantry layouts

I want to have a discussion on various different layouts of cartesian gantry setups found in the various DIY 3d printers that are out there, and understand why each design decision was made.

Traditional Mendel/Prusa: The y-axis is the moving build platform, the x-axis moves the entire extruder assembly, and the z axis moves the entire x-axis/extruder assembly

Rapman (and some others): The x-axis is a dual linear slide with the extruder assembly, the y-axis moves the x-axis/extruder assembly along its rails, and the z-axis is the raising/lowering build platform

Ultimaker: Difficult to describe in words, but esesntially all of the stepper motors are fixed; the x and y are essentiall the same mechanisms, and they are joined together by the hot-end carriage by 2 perpindicular linear bearings. The z-axis is, again, the raising/lowering build platform.

.. then there is the whole discussion of linear bearings vs busings, belts vs threaded rods vs leadscrews, etc....

What is the "best" of these, and why?

Nothing at all on this?

Well, I don't think you can call any of them "best". Each is best for a particular set of design goals.

Each of the designs has its pros and cons; and the design that is best, is the one that works for you.

As for the mechanical parts, there are pros and cons to each of them as well. In any design you have to balance
speed, accuracy, cost, and availability of parts. Linear bearings work better than bushings, but they also usually cost
10x more. Belts are faster than screws, but not as accurate. Threaded rod is cheap, but slow.
You also have to consider the strength of the motors and the drive electronics for each design. Belts have less momentum
than screws so you can use a smaller/cheaper motor and less complicated drive form since you don't have to ramp up and
down your motor speed.
All of the above and more are tradeoffs you make in the design process. People have been talking for years about which one
is best and there is no clear winner.

criswilson10 Wrote:
-------------------------------------------------------

> Linear bearings work better than bushings, but
> they also usually cost
> 10x more.

Are they? It seems to me that brass bushings with tight tolerance would offer less friction and therefore better accuracy. What about the 9 bearing setup that puts the rod between 3 bearings? If adjusted properly it seems like it may offer the least play and friction.

This is all speculation. I currently have linear bearings and 5/16 rod. This certainly not ideal as the linear bearings are 8 mm giving me noticeable play.

What are other's experiences with the different bearing/bushing setups?

Another thought I've had is to create a dual y stepper setup on my prusa. It seems this may allow me to increase speed given the mass of my build platform. Any thoughts?

criswilson10 Wrote:
-------------------------------------------------------
> Linear bearings work better than bushings, but they also usually cost 10x more.

That isn't necessarily true. In my experience, LM8UU linear bearings wear out, and become quite noisy after a short period of time, while IGUS bushings wear quite slowly.

If you were referring to v-groove style linear bearings, then I'd agree with you.

Quote

That isn't necessarily true. In my experience, LM8UU linear bearings wear out, and become quite noisy after a short period of time, while IGUS bushings wear quite slowly.

Weird, I have the opposite experience. LM10UU running for over a year 24/7, no problems. IGUS bushings wore out after a few days.

Edited 1 time(s). Last edit at 12/27/2011 05:41PM by nophead.

---

[www.hydraraptor.blogspot.com]

... igus-bushings are very sensitive to shear forces and dust, so mostly not usable in common DIY-mechanics with inadequate adjustment and open rods.

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

I'll second Viktor's comments about igus bushings. I like them, but for DIYers it is tough to keep the load balanced so that the bushing
doesn't tear up.
Linear bearings like the LM8UU seem to be easier for DIYers to use. Of course you still need to keep the load as balanced as possible.

Either way, you do want the bushing or bearing to match the shaft size. If you have a metric shaft, use a metric bearing. If you have an
imperial shaft, use an imperial bearing.

From IGUS' web page:

- For application in extremely dirty environments
- Compensation of misalignments
- No need of seals or wipers

Hmm.

I'm also looking for the genius solution on this topic. While making rods parallel is doable by adjusting (Mendel) or match-drilling (Mantis), aligning bearings/bushings isn't that simple if you want them to be connected rigidly.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Maybe IGUS's definition of "dirty environment" differs from mine, but I have always found their bushings to lock down with just a little
bit of dust on the shaft. Their "compensation for misalignment" is to break apart in my experience. I do like to use them in final designs,
but I never prototype with them because I find them to be fussy if everything isn't perfect.

The genius solution for parallel shaft alignment is a CNC machine. It's a little out of reach for DIYers though. Cheaper options include using jigs made out of hardwood or measuring and marking very carefully.
Good tools are also a big help. That includes drill bits that are 4 or 6 flutes instead of the standard 2 flutes; and sharpen them often.

Going back to the various cartesian configurations. It is true that the each configuration will have advantages and disadvantages.

I think it is important to consider where future reprap design is going and in particular what direction extruder development will be. For example.

If you think multiple heads are the way, or large "inkjet type" heads may come about, you are probably best with a fixed head (or moving just in Z) and move the part under it. This way if the extruder weighs a lot and becomes large and complex, it doesn't really matter.

If you think lightweight heads are the way forward, laser sintering etc. a fixed part and a movable extruder head are probably best.

Currently we have a machine (mendel) that is somewhere in between, probably capable of all of the above to some extent, but not ideally suited to either.

One further thought, cartesian robots do not lend themselves to self replication as an accurate linear slide is currently difficult to reproduce in plastic and is likely to remain difficult for a long time (The Sarrus linkage seems the most promising so far). Polar robots may offer an alternative way.

I don't think that people realize that polished shafts used with linear ball bushings are actually TOO smooth for Igus bearings! Igus recommends a rougher shaft and it makes a difference.

Also the best polymer bearings have a coefficient of friction of not better than 0.12, and most are higher. So for 1 lb of force against the bearing surface, from load, misalignment, or tight bearings you will need 0.12 lbs of force supplied by the motor. Recirculating ball bushings can be much lower, near 0.01, and hard steel wheels on hard steel shafts even lower. So ball bearings make a real difference with small motors.

garyhlucas Wrote:
-------------------------------------------------------
> I don't think that people realize that polished
> shafts used with linear ball bushings are actually
> TOO smooth for Igus bearings! Igus recommends a
> rougher shaft and it makes a difference.

Standard stretched steel rods, not chromed are good enough.

> Also the best polymer bearings have a coefficient
> of friction of not better than 0.12, and most are
> higher. So for 1 lb of force against the bearing
> surface, from load, misalignment, or tight
> bearings you will need 0.12 lbs of force supplied
> by the motor. Recirculating ball bushings can be
> much lower, near 0.01, and hard steel wheels on
> hard steel shafts even lower. So ball bearings
> make a real difference with small motors.

Actually coefficient of friction of igus W300 or R materials are 0.08 and there is specialty items from other furnishers with coeficients as low than 0.035 (those are polymer + Mo2S). W300 is even one of the low cost materials of igus.
Now, it is true that recirculating balls or 3 rollers combos will have lower friction, but at the expense of needing a lot less play in the parts. For a DIY maker, polymer bushings are probalbly easier to handle.

Considering the low loads on printers, even at great speeds, the PV factor is good enough that PTFE coated steel bushings would be better than polymers, i think. However this kind of bushing need very good alignement because the PTFE is very thin and easily scratched

Why is a rough shaft better for Igus. Doesn't rough imply more friction and more abrasion?

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
> Why is a rough shaft better for Igus. Doesn't
> rough imply more friction and more abrasion?

For friction, it is a matter of coefficient of surface contact.
On a rougher rod, there is more microscopic peak and valleys (on stretched rods, more valleys than peaks). A fair amount of valleys is important as it decrease the surface of contact but also allow air to circulate reducing succion cup effect as well as creating oil reservoirs.

Abrasion is a different beast and is not correlated to the roughness (Ra, average value of peaks and valleys) but to the size of the biggest accidents (Rz) so you can have a relatively rough rod surface which is not abrasive.

- nophead:
It actually is quite interesting how different materials interact with each other. For example, POM (acetal, Delrin) and PA (Nylon) prefer rougher surfaces while e.g. UHMWPE and PET will have lowest wear with smoother ones.
From the Igus datasheets, I also got to know that their different polymer formulas exhibit quite the differences in running performance against various metals. Cf53 and aluminium seem to be the best choice for most of their products.

Also, the initially rougher surface of the shaft will abrade tiny bits of the bearing, which then act as dry lubricant, so that the valleys get filled and smoothed. I didn't expect it to be that noticeable, but I could feel a difference in smoothness on V2A rods where my UHMWPE bearings ran 10km and where the shaft was left in its original state.

Yes metal against plastic is totally alien to my experience and intuition. Aluminium is an awful bearing surface against metals because it is soft and wears and it gawls and binds. Seems with plastic it is totally different.

My experience with Igus bushings and "precision ground" steel rods is they only last hours before having massive slop and my experience with RJMP08 is they come out of spec and have massive slop out of the box and are really expensive.

I would love a low friction alternative to LM8UU that had no play and would last for years in an FFF machine. LM8UU only cost about £0.55 and you don't get much else for that price.

---

[www.hydraraptor.blogspot.com]

I have heard that PBC bearings are supposed to be good and longer lasting than Igus ones. They also come in the standard bearing formats like the Igus ones. No idea about the exact price, but it is higher than some Igus bearings.

Or you could try DIY bearings with appropriate gliding materials like the ones I am currently working on:
µGen bearings
The initial design is less than ideal. I want to make Generation 2 easier to adjust and completely change the geometry of the glide surfaces.
Theoretically, you could use different materials for different shafts, ranging from UHMWPE (smooth shafts) to PA and POM (rougher shafts). The total cost is extremely low - about the same or less than LM8UUs. Also, they would be far easier to self-manufacture.