Thursday, July 15, 2010


changing the game, and using kapton tape as a slide

Hi folks,

I've been working on redesigning the design for an entirely laser-cuttable x/y stage, and have some neat progress.

This design also uses a rack and pinion system, with a bunch of changes:
1) the linear gear is mounted horizontally, rather than vertically -- the weight of the entire axis is no longer suppored by the gear itself, but rather by a structural outcrop.

2) the axis is driven at four points (two from each side), rather than only a single point. this removes a lot of the shearing/binding issues. this is accomplished using a captive system of gears sandwiched inside the axis itself to transfer power from the stepper to four points along the edges of the axis.

3) a long piece of kapton tape is used as something like a linear slide, significantly reducing the friction between the axis and the linear gears that it rests upon. the axis slides *really* well with the kapton tape.

The result is a working linear axis that can be entirely constructed with a laser cutter, using only a handfull of screws, a $5 stepper, and a few pieces of kapton tape as vitamins. That's fantastic! :)


So far it works pretty okay! There are a couple refinements to make:

4) Right now the gear on the stepper is just press fit -- I really need to key the gears and motor shaft, as the press fit wears quickly with the hardboard.

5) The gearbox is mostly just eyeballed, and there's probably an ideal solution when solving simultaneously for (a) a rack, with (b) two gears of radius R1, being driven by (c) another larger gear of radius R2; all for a given tooth size. Myself and a bunch of other physicist/math nerds at school have each had our hands at the problem, but haven't come up with a fantastic method of solving it yet (start with C, then find all the angles that B can attach to C, then determine if the distance between the centres of the two gears B modulo the gear pitch is an integer for each angle?).

Other thoughts:

6) This design seems great for the cross-axis. I have a similar design for the supporting axis with a MUCH larger captive gear system in it -- I'm not sure if the friction would be too much in that situation, or if it'd work out okay. (I stick a tiny washer under the gear, to reduce the friction a great deal -- maybe a nylon washer, or some kapton tape might also work out well?).

The alternative might be using a combination of two of these tiny "captive gear" axes, one at each end, for a supporting axis. This adds an extra stepper to the design (which is a negitive, although these ones are only $5!), but it also keeps the whole system fairly low, which would be kind of nice!

If anyone is curious to tinker, I've put the design files up on Thingiverse ( ). Happy tinkering, and thanks for reading! :)

[part 1] [part 2] [part 3] [part 4] [part 5] [part 6] [part 6 video] [part 7] []

This comment has been removed by the author.
That's really intersting stuff! Have you measured the backlash in the assembly?

Industrial CNC routers often use rack and pinion when they are large, because the cost and inertia of a lead screw is too high. However, backlash is then an issue.

One of the 'accepted' industrial solution is to use a helix angle on the gears, which in your case isnt feasible to keep it cuttable in 2.5 d.

But, you have the advantage that you have low loads, so perhaps you could use a coil spring in the system to automatically remove backlash?
This comment has been removed by the author.
What you should do is realign the gears so that each is 1/4 step more advanced than the previous- That will help with gearlash while all gears are the same size.
(edited for typo)
Or software backlash compensation? If you can measure it then its not difficult to work around, assuming the backlash is even a problem. Cool stuff, thanks for sharing it with us.
You might be able to overcome all the issues of backlash, etc with a linear encoder. You might be able to make a cheap one from an optical mouse sensor. If you had a stable material for the sensor to look at you could compensate for all the problems.
as far as measuring the backlash, why not just do it optically? Just run a set of stripes and a reflection sensor... or do it mechanically....
If you are looking for a something more effective than the kapton tape you are currently using for the slide, may I recommend 3M's PTFE Film Tape 5490. It has a very low coefficient of friction and has a high heat tolerance, for long jobs that may create some heat due to some amount of friction.
jake: thanks! if that kapton (which is very inexpensive) doesn't end up working all that well over time, we might have to switch to something a little better like the PTFE tape!

backlash: i don't see the backlash as a huge issue right now -- i had planned on rastering the layers, which means the travel on a given axis is only from one end to the other, then back again -- so you can just use endstops. it would definitely be worth characterizing the backlash once a little better revision of the gearbox is made, to see if it's something constant (that software could easily work with), or potentially etching an encoder strip into one of the linear gears to get some inexpensive position information!
This is a great project! Regarding item (5), I have done some work on this before. One solution is posted at

The optical linear encoder is a good idea. You could maybe use the output of one of the cheap digital calipers (their weakness is slow update rates). I personally have an Epilog laser and have been playing with various designs for wood/plastic cnc units and US Digital has some sweet prices on encoders - (not affiliated, yada yada)
Great work! I am currently using these same stepper motors for a different project and cannot find any information on them anywhere. Especially the wiring diagram for the coils. I have no idea what the voltage and current ratings are either. I have tried to contact Japan Servo co. without luck. Any help of how you wired them, and what speed you are running them would be great. I am using an Arduino microprocessor with an Easydriver v.4.2 to drive them. If you could please email me at that would be very much appreciated! Thank you!
Nice work.
Going back to the idea of pre-heating the powder to reduce the amount energy needed by the laser.
While thinking about this I reckoned back to my days as a photocopier tech. The final stage in a photocopier and also a laser printer is to melt the toner into the paper. This is done by a roller with a quartz halogen light inside to act as a heater.
I was thinking that instead of heating the entire powder container, if a recoverd roller from a laser printer was used as the leveling drum it would rapidly heat the small amount of powder as it was transfered to the main chamber, It would take some fine tuning but I think it should help.
I don't have a work shop to play in at the moment, so maybe someone can try the idea out for me.


I sent peter and e-mail to see if he'd made any headway on this or has been consumed by life.

I am very interested in this project, but unfortunately my specialization is in finance, not mechanical engineering. I am really hoping that this affordable SLS technology will not turn out to be nylon powder in the wind.

Troy Douglas
This is great! Is there an X/Y version? I would like to create a dual axis slider.
Post a Comment

Links to this post:

Create a Link

<< Home

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

Subscribe to
Posts [Atom]