what is an ideal scaleable RepRap build for this application?

Hello all!

I am interested in building a delta type RepRap, with the eventual goal of scaling it up for TIG welder based printing, following kolgery's StrongPrint project [reprap.org] among others. This will be further down the road, but it informs my current decision re: which printer to make. for reasons of budget, etc I'd like to start with a standard form factor before jumping in to a larger scale project.

Though both these guys: [www.youtube.com] and Molten 3D: [molten3d.blogspot.com] used cartesian type printers, Molten 3D is the only diy 3D printer I've seen other than a delta, that has a stationary bed (I could very well be wrong about this). So given the eventual mass of deposited material desired and therefor the necessity for a stationary build plate, it seems to me a delta type printer is the most scaleable design as of yet.:

-no backlash
stationary print area
- of all the stationary bed printers, Delta RepRap has the most documentation.
-there are existing welder based project using a delta rostock type RepRap, like this one headed up by Joshua Pierce [www.appropedia.org] as well as the aforementioned StrongPrint project.


Ideally I would like to start with a larger than average machine, space is not an issue for this project. I am also looking at metal based machines due to eventual uv and heat issues.
Aside from scale and / or scaleability, I am also looking for sturdiness. (I know scale and sturdiness tend to be inversely proportional) A sturdy, small structure is preferable to a large one with inordinate movement, as long as it is scaleable. If I can, I will put the TIG head on a smaller than ideal mechanism and dial down the heat as necessary.

I am currently looking at:

Rostock Max from seemecnc: [seemecnc.com]
Minikossel from makergeeks [www.makergeeks.com]
all metal kossel from makergeeks [www.makergeeks.com] (haven't seen anything on this one, which leads me to--skepticism. gimmick item?)
Griffen: [griffin-3d.myshopify.com].


a little about me:
I have not built a 3D printer, but I have built many other things and am generally extremely capable of design / build projects from the drawing board (3D modeling) to the shop. Having done some large scale mechatronic and automation projects, I have a working knowledge of electronics, various programming languages, and a very good mechanical aptitude. I have an extensive knowledge of the TIG welding process and metalworking in general, which won't help me here ...yet. I have a budget of around $1000 to begin with, but can put more into it in a few months, up to about 5K total within the next 6 months. I already have a lot of the welding equipment necessary, including a 200DX.

But, to begin with, what is the best scaleable delta RepRap? I assume this involves replacement of NEMA17's, drivers, motion control, drive train, and structural components.
What is the sturdiest Delta RepRap?
What is the best large scale delta RepRap design currently out there in the 4' Z axis range?
Based on my stated longer term goals, is delta indeed preferable over cartesian?

If there was a cartesian design that was superior for this project in terms of scaleability / sturdiness / cost I would go cartesian. I especially believe a Z axis moving build plate is tenable but would likely become costly to maintain sufficient tolerances as loads increase. If it is a generally a bad idea to start with a larger than average machine, i'll start smaller.


Any and all input greatly appreciated.

All the best,

-rev

Edited 3 time(s). Last edit at 11/17/2014 12:37AM by rev.

I built a scaled-up Kossel Mini. It was unacceptably wobbly until I added some diagonal supports. It's now as rigid as I could want. I'd highly recommend building one out of 2020 aluminum instead of 1515, but other than that the design is great for rescaling.

This is a bit jumbled, but I hope it helps.

First the good news....
There is also the original Griffin (OS), the Open Source/Reprap version. Both it and the Pro were designed to be scalable.
[griffinprinter.org]

The only other delta truly designed to scale well that I know of is the Wolfstock, which is better for sizes as large or larger than the Griffin XL. The XL though, is pretty much the largest Delta (and cartesian for that matter) you can buy as a kit (or assembled) for anything less than a new or used car and about the biggest for anything even remotely close to your budget. Deltas (and cartesians) can only scale so much before your parts are extremely bloated or extremely fragile.


Now some bad news...
Considering your plans, I would forget the Mini Kossel right from the start, not because it's a bad printer or that it doesn't scale well, but considering the amount of weight you will be moving around on the effector, a Mini Kossel will probably be too flexible, especially if you try and scale it. While the Ultibots version uses 20x20, it still uses a lot of the Kossel upper frame which isn't meant to take that much moving mass and dragging a feeder tube around. An issue you may face with the Rostock Max is the amount of wood involved, it uses A LOT, which could catch fire. That isn't to say a Griffin won't either, like the Mini Kossel, it still uses cast (Pro)/printed corners (OS) that are not metal. While the Makergeeks All Metal solves the fire problem, being based on the Rostock and relying on linear rails for the frame, it will not be a very stout machine (I had an original Rostock). Realistically, just like the Kossel Mini, the Griffin XL is probably too flexible for what you want to do with it but the Mini and Std would be good. The mini is one of the most stout printers around.

This is just the tip of the iceberg in terms of issues you will face.,
Due to the heat coming off, and flying debris, don't count on belts lasting long, in fact belts may not be a good idea at all (they too could burn). There is also a question of torque and mass, considering the amount of weight you want to move, standard Traxxas rod ends are probably not going to work, besides wear and tear they could melt.



Now the dream killer...
It's not even going to be remotely close to being within your budget. Large format is NOT cheap to do right, it can be done cheap, just not right, and this alone pretty much destroys your budget. If you just wanted to make a small welding printer above, you could probably do it for a few thousand. Throw size into the mix, you just doubled your cost. Add in the fact that you haven't built a printer before, double it again. While Nema 23's aren't badly priced, the stepper drivers needed to drive them properly are, they alone will suck up a large portion of the budget you set.

Using a delta may also be a bad idea... While it can offset the need for larger motors, it can only do it for so long and as you go larger, keep in mind that deltas become less and less space efficient while cartesians become more space efficient. Then there is calibrating... A cartesian is linear, you lift this end x amount, it goes up that amount. Deltas work in a bowl shape, the further out you go horizontally, the more accurate your calibration needs to be. For example, most consider a bed accuracy of around .01mm across the build surface good, in order to get that on a Kossel Mini sized delta, software calibration will probably reach it using a diagonal measurement in the .001mm range. To get that same bed accuracy Griffin XL sized delta, your diagonals need to be calibrated own to the .0001 range, if not more. It can take 10x longer to calibrate an XL than a mini, if not more. Trying to get a 4foot build plate accurate down to even .1mm may not even be within the firmware's ability.



I'm not saying what you want to do is impossible, however, you need to be realistic. Large format is still relatively experimental (and expensive), metal printing with a welder is highly experimental (and needs lots of custom metal parts I.E. expensive), and you have zero experience even with a normal printer/knowing what will or won't work. And you want to do it with a delta, which complicates things even more... I'm not saying your experience won't help, but it won't keep you from making mistakes and large format simply costs much more.

My advice, find a printer you like, buy or build it, enjoy it, learn it, then start planning how to build a larger printer or metal printer, and then finally combine the two. While this sounds time consuming and expensive, it will actually save you money because you will have a more knowledge. It's cheaper to make mistakes on a smaller printer and what you save on the large format will easily pay for the smaller printer. Not to mention you will get a printer you can use now to help build/prototype your next ones, and believe it or not, you will end up at your goal sooner rather than later.

All that said, don't expect it to be a couple weeks or or even a few months, what you want is cutting edge, and you have to learn the basics first. Before you even head down this path (metal printer), you should understand that what those guys were doing, while technically metal 3d printing and impressive, it's not what most would consider very useful yet. It's more about experimentation and pushing limits than it is about getting something out of it that you can actually use. You could use it with the expectation of putting it in a milling machine after I guess but that's something you should keep in the back of your head, after all this time, effort and money, you may not get anything physical out of it.

Hi. This is an interesting project. Keep in mind the precision of a Mig based printer will be extremely relative. In many cases existing CNC machines will do a pretty clean job faster.

If some filament printers mecanical principles can be transposed to metal deposition, it's a new world and you will have to invent a lot. IMO, you can't go on your project in one step. First I would experiment and build a small scale working prototype, then I will enlarge the prototype to the working size I need. Any plastic filament design cannot be transposable to your project because you wil not deal with the same heat, weight and technique. A 20x20 aluminium profile is far to stand the moves of a metal part, definitely. You will need a stiffer structure, plan on at last 15 times the weight of your part. The mecanical parts should be strong enough to stand that, and electronics too, forger Nema17 design. Then let's talk about temperature, a Mig works around 43,000 °F (24,000 °C) which is another world. The overall machine must stand that heat proximity and on the long run prevent the overheating.

Even if the inverted delta supporing the building plate may have some advantages, I'm not sure it could be realy interesting. First, Mig technology does not need a lot of speed which is the principal advantage of the delta. Second, the delta only prints a dome cylinder in it's centre, the footprint of the machine is larger than a cartesian one for the same printing area. If you want a big printer, thik about that. If I had to design a Mig printer, I would take the best ideas from plasma cutters and DLP printers. Strange mix uh ? Well plasma cutters are very close to the Mig technology with THC management. They use a pool to absorb excess heat. The DLP uses also an immerged building plate. This could be an interessting way to cool down the work. My think is a catesian XY fixed over a pool, and a plunging Z building plate. If the part only move along on axis, the building construction should be lighter.

---

Collective intelligence emerges when a group of people work together effectively. Prusa i3 Folger (A lot of the parts are wrong, boring !)

Thank you guys for the input.

Sheepdog43 and Zavashier, your points about the limitations of delta printer scaleability are duly noted. And I fully agreed re: getting acquainted with the existing tech and scale before attempting anything as far afield as a TIG based machine.

Sheepdog43 You have raised some key points about delta vs cartesian scaleability, space efficiency, and accuracy. I will do some more research into moving Z axis cartesian machines before making the plunge. Any recs? I have been looking at the flash forge creator. I think lowering a potentially heavy load along the Z axis throughout the print is the only viable cartesian printer scenario for this application. So it comes down to which Z axis motion is the most stable / scaleable. It seems that fishing line (cable) based drives will scale better than belt. This might seem incredibly naive, but perhaps this Z axis motion could be achieved passively via a ratcheting mechanism and counterweights. Or weight could be maintained constant by storing filament to be used on build platform. But this is a long way off. I am not building a large machine any time soon. Just one that has the greatest potential for this application. Scale is important only in that it will keep the machine further away from the work initially due to heat issues primarily.

Zavashier, interesting ideas re: the MIG process. Heat management is certainly a huge issue. But it will involve dissipating it throughout the work to avoid distortion rather than trying to draw it out. BTW I am thinking exclusively about the TIG (Tungsten Inert Gas) process, not MIG. MIG is much dirtier and less precise. TIG welding is capable of much more precision than most people may realize, and does not generate sparks or smoke. You can use a 1/16" electrode sharpened to a needle point and achieve some truly minuscule welds. Add in pulsing techniques and high frequency arc focusing and the machine is capable of much more than the human hand can coax out of it. Whereas the temp of some exotic plasma processes may reach 20,000 F at the hottest point of the arc, it is no where near that 1 inch away. The TIG process reaches perhaps 6000 F at the hottest point, then dissipates rapidly to a few hundred (the operators hands are a few inches away). The speed of the delta RepRap could be advantageous in dissipating this heat by jumping from one side of the part to the other, thereby reducing warpage. But you are right, heat is a major concern with this process. As for tolerances, I agree, it would not approach laser sintering or FDM by orders of magnitude, and would necessarily require post processing for critical parts, but even this would be an incredible advancement. Whereas I am certainly thinking of the process as a 'roughing in' of parts to be finished, there are also applications for printed metal geometries with non critical tolerances.

Thank you guys again for your input. Please let me know if you have any recs for a cartesian machine, or any other thoughts.

-rev