Reprap Design Strategies - Ultimaker

I am new to 3d printing and was investigating different reprap designs and trying to understand what current challenges repraps were facing. I was wondering a few things:

1. With regards to speed of fabrication, what is the limiting factors on the speed of printing?

2. I found that the Ultimaker is one of the best in terms of speed for a given precision; so what exactly about its design is so good compared to other printers ? (If say the other printer also has a light Bowden extruder)

3. Also, what type of problems does one run into with this design in comparison to other designs?

4. I know the cost is also pretty high for the Ultimaker; which parts therefore are the most expensive? And why is it cheaper to make, for example, Mendel style designs in comparison?

5. I was also wondering the benefits and drawbacks of direct belt systems versus double belt systems or rack pinion setups?

Quite a few questions I guess but just wanted to get an opinion on why different designs are better. Cheers for reading.

The mass moved by the motors limits the accelleration of the axis, this becomes significant at higher speeds.
The Ultimaker has very low moving mass, so it can run aggressive accelerations, which results in higher print speeds.
On a conventional Mendel style printer the Yaxis would become the limiting factor if you used a Bowden extruder, on most simple gantry designs, you end up having to move at least the mass of a one motor along an axis. The exceptions are HBots an the variants there of (coreXY etc) and the Etchasketch (ultimaker like) designs.

---

___________________________________________________________________________

My blog [3dprinterhell.blogspot.com]

I can attempt to speak to rack & pinnion as I am incorporating one onto the Y (long) axis of my printer which is going to yeild about 27" + or-. The benefit is after a certian length, the tension needed to prevent a belt from sagging will cause other problems including wear and may possibly overcome rod stiffness depending on the layout. If you are doing rack and pinnion, it needs to be adjustable to perfectly match the linear travel in parallel to the drive gears, and the drive gear must be perfectly parallel to the rack. I'm still early in construction, as things progress I will see just how this design plays out. I am only doing R&P on the one axis however, the extruder carriage, bowden equipped will be belt driven.
If you are going to use 23s on anything other than a fixed application, speed goes out the window. If you want to go big, in my opinion, Z for the build bed is the best way to limit dynamic weight, and the issues it could cause. Also, balancing weight and loads is important to overall performance.
Most parts are available in a number of options and then those options can be had with various features and properties. Some parts can be gotten from China a bit, or a lot cheaper than domestically. If it is exactly the same part, you save money, if it is a quality part, you make out. It really pays to do a lot of research to see what you want to save money on, and what you are better off sourcing based strictly on quality. Example, you may save money buying a cheap hotend, only to find out that the few dollars you save buys frustration compared to buying a better quality part for a few more dollars. Many of the linear bearings we use however can be gotten cheaply from China if you are willing to wait a week or so, and most originate from the same place. Hope this is of help.

First of all, thank you both for replying.

I looked at the CoreXY and the HBot designs and those were quite interesting. Is there some specific reason why they aren't used more than say the Ultimaker style pulley and idler system? I can understand that the mechanics of the system is more complicated in the HBot design, especially with both the motors having to move simultanenously in opposite directions to provide a diagonal motion, but I guess that would be more of a control and programming challenge, perhaps. I found a difference of opinion while looking at blogs on its though which was quite confusing. Some saying that it could achieve high accelerations and speeds and others not. Don't quite get the reason for the difference in opinion. Also, what about precision, would it be more difficult considering both motors would need to work together precisely?

With regards to the rack and pinion, I guess it gets rid of a lot of problems with belts as the build volume gets larger. I would also figure it limits build volume for smaller designs depending on whether the rack is stationary or moving with the gantry.

Until recently there was no real support for CoreXY or HBots in the firmware, though the change is basically trivial.
I had issues with my HBot prototypes I could run very fast with acceleration similar to those used ultimaker's, however there are racking forces introduced by the belt path and I could easily see the effect of those in motion across the H. Others seem to have managed to get them running without this issue, so perhaps it was something in my design exacerbating the issue.
CoreXY removes the racking forces, but I've never built anything with the dynamics to test it.
Both geometries require long belt runs, which are expensive and makes tensioning difficult, and both couple X and Y motion together, this last part seems like the real weak point to me, not because it's inherently bad, but because other mechanisms achieve basically the same thing without having to do it.

I have some parts for a 4th HBot prototype, which I'll get to when I've finished with my current build, though my intent this time is to use the CoreXY mechanics instead.

---

___________________________________________________________________________

My blog [3dprinterhell.blogspot.com]

I think that a rack works best ( on a rod system anyway) anchored to a non-moving frame member. In my case that is the Y axis, and the bed is the Z axis, so in this case it actually expands build area potentially, depending on your carriage arrangement. Bear in mind, that my frame is rectangular and encloses the entire print area.

ok new idea, it uses the hbot configuration for the structure, but there are two belts for the x axis, then you have a square rod running down the side of each rail on the x axis, these square rods are connected to motors and on the x axis bar the square rods turn pulleys on the x axis bar, which drive a belt connected to the x carriage. [garyhodgson.com]

I will probably try make a corexy machine, that seems a very good solution for me, the raking equilibrium lessen the needed precision, especially with a small bowden tube and 2 or 3 heads. Quite cheap and good precision achievable with only DIY tools.

For the belts, one thing to be aware of is that the same belt (say T2.5 or GT2) exist in at least 2 versions with either plastic (high flexibility) or metallic (high resistance to tensioning) reinforcing wires. T2.5 exist also with kelvar wires. We should use, i think, only the metallic versions. Il dont like the idea of fishing lines, the potential to slip is too problematic. Metallic wire cables (bike brakes) could be used too. No tensionning problem in that case.

You'll find you can't find GT2 belts in the commonly used sizes with anything other than neoprene with fiberglass reinforcement. There is nothing wrong with that it should be fine for any purpose on a reprap. With metal reinforced belts there is an associated minimum bend radius and it can be quite large so I'd check it's actually larget than your planned pulleys and idlers.

In general timing belt will be better (and certainly pose less pitfalls) than cables (fishing line or otherwise), but a working cable driven version is an interesting exercise, because it removes an expensive component.

---

___________________________________________________________________________

My blog [3dprinterhell.blogspot.com]

dont know about GT2, it is an US standard, but T2.5 is available in steel :

[products.beltcorp.com]

I used it in kevlar versions too. Anyhow, it is just a matter of adjusting the width of belt, we can use fiberglass too.

Minimal inner radius for metal T2.5 is R6 which is fine with what we have now. outer radius (cambering on the spine) is much bigger, but that is not a problem.

A for expensive, T2.5w6 is 5€/meter on ebay, a 300x200 corexy will need about 3m, that is a max of 15€.
You need also 8 pulleys but i think that you can print those if you do it in nylon and still get a precise item.