Rack and Pinion on a McWire RepStrap

Ever since reading the "Rack and Pinion" thread, and namely goinginreverse's statements about the mcwire being too slow for it's extruder (and he's not the only one I've heard this from), I've been contemplating an alternative McWire design that would replace the leadscrew with a rack and pinion, probably one from McMaster.

The Rack and Pinion discussion seems to be all the rage right now for Mendel. So I'm interested in ideas for placement, and solving problems like alignment.

My initial thought is to mount the rack on the underside of each stage in the middle, with the pinion gear on a rod, supported on either side and attached to the motor. If we move the X motor to a position on the back side of the base, and put the Y motor on a mount perpendicular to it's current position, there shouldn't be much else to change.

I should note that I write this being about 25% done with my own McWire bot, so if my naivety in terms of building/printing are apparent, that's probably why. Like many, I want to get on the fast path to printing parts so I can get to a genuine RepRap, but if we can side step speed issues by changing the linear motion method, I'm all on board.

You may want to also consider trying a belt drive; put a 10 x .2" tooth pitch gear on the stepper, mount it orthogonal to the direction of travel off one side, mount an idle pully to the other side, then attach securely the belt to the underside of the platform. This is effectively what the current version of Mendel uses, and the timing belts appear to work quite well there.

Rack and pinions have great potential to be printed, but that advantage is lost if you don't have a printer yet.

I am currently experimenting with this rack and pinion from mcmaster (nylon, cheap):

57655K63
57655K53

There are a couple minor issues to work out. I mounted the rack using small holes and set screws through the face. This is not as secure as I would like, the rack has a natural tendency to pull away from its mounts and bend into a U because of the way the nylon is molded. I will be using epoxy and drilling top putting screws top to bottom through the rack (the set screws were through the face).

I have tried both motor on the plane/rack on the frame and the reverse. I think rack on the plane is better but it works best if you can centrally mount the motor (in the middle of the plane traversal) otherwise you lose most of your travel length. There isn't really a good place to put the motor on the mcwire frame and there are some clearance issues to work around.

The other problem is finding a way to shim the motor shaft out to the 1/4inch bore on the gear. This is the biggest problem. Right now I am using a drilled our aluminum insert but any slight offsetting causes some issues in the quality of the meshing.

Otherwise I don't see a reason this won't work, it can definitely move fast enough and accurately enough. I am using the higher torque nema 17 steppers from interinar though (I don't think the makerbot one would cut it and in fact my advice is to just get the most powerful nema 17's you can afford for everything).

Belts are a bit of a headache because you need to properly size the pulleys, belt, bearings for the pulleys to rest in and then deal with tension issues. Just look at all the extra stuff on the mendel design to handle that. The mcwire doesn't really have any good place to put/mount all of that.

Have you dealt with any movement of the gear to one side or the other? I assumed that in spite of the rails, there might be a tendency for the gear to move off path. I was figuring I'd mount a "guide" piece of PTFE to either side of the rack.

Do you have pictures of your build/design?

goinreverse Wrote:
> Belts are a bit of a headache because you need to
> properly size the pulleys, belt, bearings for the
> pulleys to rest in and then deal with tension
> issues. Just look at all the extra stuff on the
> mendel design to handle that. The mcwire doesn't
> really have any good place to put/mount all of
> that.

Fair enough. I guess I rate belt somewhat 'easier' because of the bias I had using them in the RepolaRap design: 1 pulley, one motor, one round platform, one belt. Per axis. Done. I've not yet had a problem with tensioning, but may run into that, in which case, I'll need one skate bearing and a rubber band/spring per axis to create a tensioner.

The mcwire is all "mushy" to me, tolerances are huge and the bearing arms are not really rock solid. I looked at trying to do a belt driven but it ended up being possibly another $70 in belts, pulleys, plates, thrust bearings, additional mountings and it just didn't seem like there was a good clean way to mount it. There are some of those problems with the rack and pinion but they just seemed more manageable, plus I wanted to experiment with the rack and pinion for my 80/20 device. I probably could have hacked something to together to be cheaper but the rack and gear were $11 combined I think so I don't think I could have beat that.

With the rack on the mcwire I had to double up the rubber bands I use to keey the bearing arms against the guide rail to increase the resistance holding the plane in line. The only movement problems I had are because of either too much friction (I wasn't that careful with the tops of the u-channel when I cut them and were they rub against the PTFE could be better) causing skipped steps with the motor (a higher torque stepper would be better) or the mounting of the rack, set screws through face, flexing in some cases. I should be clear that I am talking about the X and Y axis, I don't think the rack and pinion would be any use on the Z (yet) but that doesn't matter because you don't need speed in that direction.

Conveniently for the 80/20 device the 1/4 in face width rack slides right into the t-channel of the extrusions with rubber panel liner insert in place. The gear then also fits slightly in the recess of the t-channel to assist with alignment of driving against the rack. Really nice combination.

I do think that smarter firmware with all the acceleration issues worked out and perhaps optimized for this case is needed to fully realize the potential of rack and pinion.

Maybe I haven't been listening closely enough, but I've yet to hear of anybody printing anything significant with a McWire repstrap.

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

Forrest Higgs Wrote:
-------------------------------------------------------
> Maybe I haven't been listening closely enough, but
> I've yet to hear of anybody printing anything
> significant with a McWire repstrap.


If that's the case, it represents a significant problem within the project. I think many (including myself) have been given the impression that if we can't get ahold of parts for a Mendel, we can build a McWire repstrap and print out parts for a Mendel. The further I've gotten into my reading, the more I've arrived at the conclusion that the McWire is not as plug-and-play as it seems at first glance. I'm gonna guess this is why The Makerbot CupCake exists in the first place. The McWire is a great idea, and a pretty good execution, but I'd love a hand-count of how many people have successfully bootstrapped a RepRap off of one...

Never-the-less, I intend to complete mine and print out the parts for a Mendel...

Quote
sircastor
The McWire is a great idea, and a pretty good execution, but I'd love a hand-count of how many people have successfully bootstrapped a RepRap off of one...

Me too. It seems to me Mcwire is bordering on being time-prohibitive for use to bootstrap Mendel but as I have little other choice I will be building one as soon as my parts come of back order.

I have heard many people with ideas on howto speed up Mcwire and am looking forward to seeing some designs.

I would also like some opinions on this guys idea (see the video about half way down) about adding mass to the threaded rod to help overcome any binding. He claims to get almost 80mm/sec but admits that the firmware needs to ramp up to that speed. If this works this looks be one of easiest mods to Mcwire to add speed.

I am pretty sure nobody has booted off a McWire. I have see a lot of people start building them but not many complete them and start printing.

I don't remember anybody else bootstrapping from a RepStrap to a RepRap, other than Makerbots and Rapmans for that matter. There might be one or two.

I am sure with acceleration the McWire would go at least twice as fast. With slow acceleration it could be even faster but it would need the 5D code and a stepper driven extruder to keep it in synch.

---

[www.hydraraptor.blogspot.com]

I have in fact printed parts of the mendel, specifically the frame vertex and the extruder carriage, plus unsuccessful attempts at several others.

As for the fellow with the higher speed threaded drive, that is a fine idea but he using completely different motors and drivers than are recommend in the mcwire instructions and in a mendel frame setup (rather than mcwire which has a lot more mass in each plane I believe). Also that design requires printed parts... which defeats the purpose of a bootstrap.

Printing a single frame vertex on the mcwire took me pretty much an entire weekend day of fits and starts with tons of skeinforge tweaking. The extruder carriage is very messy but could be cleaned up with some drill press time and filing. With 40+ unique parts and in some cases several copies of certain parts I calculated print time at 200-400 hours. In other words completely not worth it.

I would be interested in hearing anyone's experience printing with the makerbot cupcake, looking at the size of some of these parts I think there would be some significant problems and also given the STL's a lot of manual skeinforge tweaking as well. I believe it has been confirmed that it can be done but I would have to guess it also takes a lot of extra tweaking time. I have made my position on printed parts clear and others disagree, requiring printed parts without standardized interchangeable parts as options make the idea a complete gimmick and not practical at all. The evidence is overwhelming, the 'official' bootstrap method presented by the project does not actually work.

wikipedia 'gimmick': In marketing language, a gimmick is a quirky feature that distinguishes a product or service without adding any obvious function or value. Thus, a gimmick sells solely on the basis of distinctiveness and may not appeal to the more savvy or shrewd customer.

Reprap isn't selling anything so I don't see how the definition applies.

The idea of using a McWire to bootstrap was Zach's idea, but he never built much more than a mini-mug and then moved on. I agree it should not be an offcial recommended idea.

The printed parts are practical though. I have printed two Darwins and most of Mendel and hence can make those machines for the price shown on the wiki, less in fact as I will be using much cheaper electronics.

So they achieve the objective and being able to make machines very cheaply. I use a much simpler version of skeinforge so I don't have any tweaking. I just slice the parts and print them. My current extruder has proved reliable enough to print a whole machine without so much as a hiccup, assuming it does not fall over on the last few parts.

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
>
> I don't remember anybody else bootstrapping from a
> RepStrap to a RepRap, other than Makerbots and
> Rapmans for that matter. There might be one or
> two.
>
Well, you're one of the few that have.

I think that the problem with being able to build a repstrap good enough to print Reprap parts is that once you have a repstrap that good there isn't all that much incentive to actually build a Reprap that isn't going to be much better than your Repstrap.

I think that nophead and Frank Davies fall into this category. Nophead still uses his repstrap as far as I know, even though he's printed several sets of Reprap parts and, to the best of my knowledge, built up a Darwin.

Frank, otoh, built his repstrap and then just got on with designing and printing cool stuff.

[www.thingiverse.com]

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

And that's why there are very few RP RepRaps.

Sort of like the reason you never see them doing real work in the holodeck on STNG, most usefull room in the whole ship, and what's it used for? Play.

---

repraplogphase.blogspot.com

Has anyone tried printing thicker layers on a McWire to speed things up? It seems like switching from 0.5 mm to 1 mm layers you could print stuff in a quarter of the time. Sure, you lose accuracy and the texture of the parts is a lot more noticeable, but with some finish drilling the parts should be serviceable. You might have to slow things down for some specific parts, like the pulleys.

I've come to the conclusion that print speed (volume/time) has everything to do with the extruder. The cartesian bot, and to a lesser degree the nozzle, sets the accuracy that you can attain at a given speed. On my Darwin I can leave the flow rate the same, speed up the feed rate and reduce the layer height and get more accurate parts in the same amount of time. The only issue I run into is that PLA can't stretch around small details well. A smaller nozzle would fix that.

On the Repstrap census, my Repstrap is very much on the Darwin plan, but it uses no printed parts, so I think it counts as a Repstrap. I've printed a full set of Mendel parts and given it to someone else a few weeks ago. I hope to get pictures of that and do a blog post this weekend.

---

---

Darwin clone, Gen 2 electronics, Arduino Duemilanove w/ AtMega328, 5D Firmware, Pinchwheel extruder
[www.codeerrors.com]

There is a limit to how big you can go for a given nozzle size. If you use a filament size that is larger than the diameter that it swells to after leaving the nozzle, then it tends to extrude a zig-zag rather than a straight line. Also you can't span gaps without some tension.

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
> There is a limit to how big you can go for a given
> nozzle size. If you use a filament size that is
> larger than the diameter that it swells to after
> leaving the nozzle, then it tends to extrude a
> zig-zag rather than a straight line. Also you
> can't span gaps without some tension.

Hi Nophead,

Do you know if there are issues going the other way? I.E, printing smaller filament with a larger nozzle? Any idea on some formula that relates to the range of viable filament sizes for a particular nozzle, I.E, from 150% to 70%, so for a 1mm nozzle, you could print from 1.5mm to .7mm, and for .5mm nozzle, you could reliably print .75mm to .35mm?

I suspect some of it may be material related too... Getting info like this in table form on a wiki page for general plastic extrusion information would be nice.

When I attempted to print a driven-pulley with 0.3 mm layers and a 0.5 mm nozzle it did not work. The material didn't stick well enough on the perimeter to be stretched around the fine details. I ended up with a cylinder, not a pulley. When I printed with 0.45 mm layers (and 0.54 mm thread width) it worked much better. I tried this in both PLA and ABS.

---

---

Darwin clone, Gen 2 electronics, Arduino Duemilanove w/ AtMega328, 5D Firmware, Pinchwheel extruder
[www.codeerrors.com]

The limit going smaller is when the filament is stretched so much it snaps. There is probably some loss of accuracy as well when the filament is smaller than the aperture and too much stretch causes corner cutting.

It depends on the plastic, particularly how much it swells, which is also a function of the flow rate and the aperture size.

With a 0.5mm nozzle I extruded ABS between 0.3 and 0.5 successfully
With a 0.3mm nozzle I could do 0.25mm to 0.4mm. 0.5mm did not work very well as not enough stretch.
With a 0.4mm nozzle I have done 0.3, 0.4 and 0.5. PLA is not very good as 0.5 though as it has very little die swell.

With HDPE through a 0.4mm nozzle it swells so much that I struggle to extrude less than 0.5mm as too much stretch causes it to cut corners.

For mass production of parts I am going to try 0.5mm filament from a 0.6mm nozzle to see if it is good enough. Certainly 0.5mm filament from a 0.5mm nozzle and 0.4mm layers were good enough to make a Darwin. There are some fiddly bits in the x-axis bearing plates of Mendel that might be better at 0.4 though, and a ridiculously small x cable clip.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
I am pretty sure nobody has booted off a McWire. I have see a lot of people start building them but not many complete them and start printing.

I know this thread is ancient, but it came up in a search and I didn't see it before when I was running my McWire. I completed a McWire way back in 2009, and immediately converted it to be belt driven. From there, I printed two full set of parts for Prusa Mendels and assembled those machines. One got sold, and the other I kept for myself to replace my McWire. See my site here: [www.cameronmaclachlan.com] (I just noticed some of the images are broken. I was hot-linking them off of Thingiverse. I will update it when I have time.)

Quote
nophead
I don't remember anybody else bootstrapping from a RepStrap to a RepRap, other than Makerbots and Rapmans for that matter. There might be one or two.

Maybe I am the only one who has done this?

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!