Tuesday, March 17, 2009


Skate bearings and bike chain

I've been meaning to document my experiments with bike chain and skate bearings for a while now. I've used bike chain from the beginning for my handmade almost-darwin, but only started using skate bearings because I was unhappy with the wobble and the friction that wood-steel bearings gave. I shall split this post up a little, because there is no reason the two have to be used together. Here goes:

(Various people name their x and y axes differently, on my reprap it is the y-axis which moves the x stage)

Bike chain:

When I started my reprap, I couldn't find the belts and sprockets for less than £60 shipped to the uk. A quick look on some cycling websites showed that I could get the equivalent amount of bike chain and jockey wheels (the smallest sprockets I could find) for about £30. This wasn't just a cost saving measure, I liked the fact that bike chains are unbreakable (with the loads present here), they won't stretch, and I felt I was getting something more significant for the money. Before I go on, here are the drawbacks of bike chain you should be aware of:

-Reduced precision due to larger sprockets. I'm getting 4 steps/mm, which I believe is lower than you'd get with belts. This didn't matter for me personally.
-You need a sturdy frame and good bearings. Since the chains have to be tight, you'll get a lot of friction without good bearings, hence the skate bearings I'm using.

My reprap uses 14 jockey wheels, 8 for the z, 4 for the y, 2 for the x. They cost around £1 each, and seem to be made of some nice HDPE. For the z axis, it was sufficient to trap them with m8 nuts, as the z axis needs very little rotational force to turn. For the z axis, I found it was easiest to use 4 belts instead of a single long one, because it meant many more teeth meshed correctly. The white discs were fashioned out of polymorph to prevent the belt from slipping.
To attach them to 8mm steel rod, I tended to bodge it, with suprisingly good results. I put an m3 bolt through the rod just shy of where I wanted the jockey wheel, and then drilled three small holes through the sprocket. Then, I smothered the whole thing in polymorph. The m3 bolt prevented the polymorph from turning on the rod, and the polymorph seeps through the holes in the sprocket, preventing that turning. It doesn't look great, but there was no movement whatsoever even when gripping both with pliers and turning as hard as I could. To tension the chain, I used a machine screw I had lying around. It would have been a good idea to put a split-ring/sprung washer on as well, but I couldn't find one at the time.

Roller bearings:

My reprap uses 30 roller bearings, but at only 30p each from ebay, I don't think its much more expensive considering there are very few extra parts. Bear in mind I will probably replace a lot of the parts with better wood, and cut things more accurately, but it shows the design works with badly made parts.

As I mentioned, I was getting too much friction, so I decided to make new x and y axes. My favorite aspect of the darwin is that few of the parts must be made to a great deal of accuracy because the threaded rod can be adjusted to suit. I tried to design the axes in the same way, so anyone can build it with simple tools:

Tap and die set - One of the most useful set of tools, you won't realise how great they are until you get some. I got a set of really nice carbon steel ones from aldi for £8.
Drill - I used a cheap pillar drill, but a hand drill would be just as good.
Saw - Whatever's appropriate for the material you use.

The wooden parts could also be reprapped if you had a generous friend. For the y axes, the only parts needed are 8 rectangles with holes in the right places. I chose arbitrarily, so its probably easiest to see how it works from the diagrams. I chose 3cm center-center separation for the x axes. Each pair of bearings is separated by two m8 washers, so they center really well on
the rod. Because the bearings are so good I can tighten everything so the y-axis can even be powered from only one side with no rotation of the x stage. I chose not to do this, since I already had power from both sides. I could measure no backlash whatsoever, and each axes moved with a gentle nudge. The x axes worked in a similar way, I haven't attached the extruder yet, because it obscures the photos, but it will simply bolt on.

Sorry for the long, disjointed post, if anyone wants more info, ask in the comments. Some more photos are here.

UPDATE: pritam has tried the roller bearing setup with success, wish I'd had the space on my reprap to make it so neatly.

Someone else had done a similar bearing setup but he was riding them on pipe. He said his were making a ratcheting sound. He found the offset of the bearings to the side of the pipe was putting torque on the bearings to make the sound. Have you had a similar problem?
Thanks for the post I am definitly going this route. Are the cogs you mention smaller than rear de-railer bike cogs? Also do you plan to oil the chain often:)
Arvin: I haven't had that problem, not sure whether its because the skate bearings are well made or the gap is only about 3mm.

Mike: The jockey wheel are the same used in derailleurs, so a bit big. I'm not sure I'll have to oil the chain too often because they're under a lot less stress and speed than when used for their intended purposes.
Good stuff James! Being that I am following your footsteps in my design it's good to hear what kind of problems you encountered so I can look out for them! Good luck and keep us updated on your progress!
One thing to investigate is the possibility of non-bike chains of the same type -- especially smaller-pitch ones. Industrial equipment uses these things (roller chains) in all shapes and sizes.

You might want to look at SRAM chains if you can get them cheaply -- they come standard with a powerlink that allows you to disassemble and assemble them. Hell, investigate the powerlinks anyway -- they're sold separately and pretty cheap, and you can get rid of those bodges with the screw stuck through a few links.

For larger accuracy, investigate larger sprockets. A single rear sprocket in almost any size can be had relatively cheaply. You could even have several gear ratios, depending on if you wanted speed or accuracy for any given print.
jasper: I'm planning on using better chains for the child of this reprap, I'm hoping I can get good enough accuracy to print working parts without, plus bike chain is really easy to find. The screw stuck through the links is actually to tension them, I didn't make it clear in my post. Larger sprockets will reduce accuracy, because every degree the stepper turns will translate into more movement.
Why not use a derailer-like structure for tensioning? You could replace the traditional spring with a bolt, and that would give you a more aesthetic (and possibly better functioning) result.

The large sprockets will create MORE accuracy on the Z axis if they are only used on the threaded rod and not on the stepper.

To get better accuracy on any other axis, you need to gear down the stepper.

What about this style of drive, only using bike chain?


Should work pretty well, though it means moving the motor on the carriage, which increases inertia...
Annirak: I thought of using a derailleur system, but I don't want to touch it too much seeing as how I'm getting zero backlash at the moment. I see what you mean for the sprockets, but I'm powering one directly with the stepper, so I can't use it on mine. The z axis is also accurate to 0.05mm, so its lowest on my list of things to improve. I don't think the belt drive thing is practical for me, too much work and I'm lazy.
My repstrap goal is 0.1mm accuracy on all axes. The biggest problem is feedback, but I'll have some more info on that in about a month.
Post a Comment

Links to this post:

Create a Link

<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to
Posts [Atom]