Which Stepper Motor? Motion system?

Hi All,
Looking to build a reprap type device down here in Australia but I'm finding that I can't make a decision on what type of stepper motor to use. What are the current reprap contributors using? I'm worried about spendin money and not having a powerful enough one. I'm thinking NEMA 17 but might be too powerful?

I'm also kinda uncertain about using the threaded rods to move along the axis - wouldn't a pulley system (as is used in current printers) be faster & easier? I've also been considering a cog and track type system since cogs are one of the first things to have been demonstrated as being printable (threaded rods would be difficult to print)

Note that even if my project falls flat on its face I hope to be able to produce a list of suppliers down under that can help future reprappers to produce the main reprap device.

thats a good idea nathan. infact, i think reprap 'darwin' which is v1.0 actually uses a belt/pulley system to move the xy assembly.

i really like the idea of being able to print the drive/positioning system. that would REALLY make the thing more self-replicating.

The rack and pinion approach you're suggesting has reared its head multiple times during the development of the system. It is really appealing just because it can be more easily fabricated by the machine itself, which is obviously one of the ultimate goals of the system. They are also very conceptually simple and seem like they would be easy to make.

One of the reasons for not using this approach now is that small production errors in the rack lead to generational errors that can build up and get worse over time.

Also the accuracy of version 1 (Darwin) isn't really up to producing highly accurate gears on its own, so the errors would creep in very quickly.

So for Darwin the possibly slight safer approach of timing belts is being used. After that, when we have a good platform for further experimentation I'll go slightly out on a limb and say I think there is fairly universal agreement that either the belts will go, or there will be a subsystem that will produce the belts automatically. I think only good experimentation will provide the right path.

***Also the accuracy of version 1 (Darwin) isn't really up to producing highly accurate gears on its own, so the errors would creep in very quickly. ***

We were using threaded drive rods with Zaphod, but Adrian dumped them for belts with Darwin. Belts are faster, but it would appear that they would have coarser resolution. Does anybody know what kind of XY resolution that Darwin is on-track to have?

Oh, regarding the pulley system, yes I think that's a great idea too. Just about all head based printers use that approach and it works pretty well for them. It is certainly very fast for positioning compared to threaded rod.

As I understand it, the new Darwin is doing just that for one of the axes. Although that might change before the final version.

With common 1.8 degree/step motors, to get approach our worst-case required accuracy levels of <0.1mm you'd need a wheel diameter of 12mm (assuming half-stepping), which is about what you find in some printers. It does limit accuracy quite a bit though.

Remember, printers also have the advantage that they just fly their head in one direction. They don't need accurate positioning, they just need to accurately know where the head is. It's a much more dynamic action. Fly the head along, squirt out ink as you fly by the right point. In fabrication, you need to dwell on a point longer to extrude material. That means you not only need to know where you are, you need to be able to stop at a particular point. Your typical inkjet head can't do that. You also need to change direction and move at different speeds. Older line printers and some dot matrix printers used a system that did allow for better positioning, but they weren't highly accurate either.

Threaded rod is used in several of the other axes. And for most of the earlier experimental platforms people went with threaded rod because we know it works well and it effectively gives a great deal of gearing without needing to actually make our own gears. You can buy threaded rod globally. Although it's slow, we have more accuracy than we need rather than being borderline. So it's a good safe bet.

The goal here is to get some initial base platform out there so other people can start experimenting too. We are not suggesting the current methods are the best or only methods, they're just very simple so we can get something out there sooner rather than later.

Hi Nathan,

I've been taking to storing all the supplier information i've gathered on the wiki. It's a bit sparse, but this page has all the info on stepper motors: [reprap.org]

I dont know exactly what size motors you need. I got some nema 17's and a 23 for testing purposes. If you can swing it, i'd suggest getting one of both for testing and then once you know what size you need, then get more of it to complete the machine.

Once you find a supplier in Australia, please let me know so I can add that stuff to the wiki. Alternatively, we can go about getting you access to it if you feel like doing it yourself =)

With the Darwin prototype we get a resolution and repeatability of 0.1 mm with the timing belts. But they're almost certainly better than that - that's just the finest step on the steppers. Remember that inkjet printers use them to position the head to withing the diameter of an ink-jet spot.

One of the main reasons for going with timing belts is their ubiquitous use in printers and scanners. We suspect that Hewlet Packard may have spent maybe as much as a couple of hundred dollars researching all the alternatives...

It's occurred to me that as far as I can tell, microstepping those steppers could be done with just a software change to have the pic generate a PWM signal through each coil of the stepper...

Am I missing something, or has that software just not been written/needed?

-Jonathan

Half-stepping is currently implemented. Adding microstepping has been discussed, however since most of the experimental designs have used threaded rod, which gives a very high gear ratio, it is far more accurate than is needed already.

Even for the direct belt drive, it would still provide the initial goal of 0.1mm without microstepping. So the extra complexity just hasn't been justified yet.

In the longer term we may move away from steppers entirely.

Further on again, my guess is we might move towards brushless DC motors, in which case something similar to microstepping might be useful again. Nothing very certain just yet though...

this is exactly why its so critical for us to get v1.0 done and out the door to people who have ideas like yours. i've seen tons of ideas on this board that are fantastic, but unfortunately until you have a machine in front of you its pretty much impossible to test it.

anyway, good idea, and i wish we could send you a machine to test it on =)