CNC Build Ditches Rods For Hardboard

This is a redesigned x-axis for [Peter Jansen’s] selective laser sintering rig. We looked in on his SLS project last month and since then he’s been refining the design. The new component uses a rack and pinion system, relying on some Kapton tape to reduce friction for a nice smooth slide. One stepper motor powers the laser-cut gear box with four gears interfacing the sled to the frame for stable and accurate motion. Now he’s just got to work out the math/physics that go into finding the optimal gear ratios as this prototype is just a rough guess. If you’ve got the skills to work it out please lend [Peter] a hand as we’re quite excited with where this is going.

29 thoughts on “CNC Build Ditches Rods For Hardboard

  1. I can’t help but wonder how accurate that is… A while back there was a post on this very site covering the use of a hinged linkage as a replacement for precision rods (wish I could remember its name).

    I can see problems with skew coming in as well, as the track geometry is rectangular.

    Not that this should stop anyone having a go, I’d hate to sound all-negative, as its still quite a novel use of materials!

  2. There’s a reason why most people go with ground rods, linear rails, and precision ballscrew. It’s about accuracy under load. I realized there’s very little load in this case, but I’m not convinced using laser cut wooden teeth will be accurate or repeatable. Even something as simple as having a humid day could mess with the accuracy.

  3. the issue i’m seeing is using a material that’s not dimensionally stable. MDF and hardboard are stable, that looks like the crappy “plywood” that’s a veneer on soft wood. easy to laser cut, but warps like a wonk’s version of reality.

  4. lexan / plexi cost much more money than hardboard. As I normally do, I will prototype with masonite then change to something like Lexan when the build is all worked out.

    I think Peter is doing the same sort of thing.

  5. hi folks,

    neat to see this up here! :)

    to your questions, there are definitely better ways to build an x/y axis, if you have a set of precision parts available (like ground rods, belts, etc). the goal of this project is to set all that aside and say, i’d like to build a 3D printer that can be *entirely* laser cut (and therefore, also likely printed), with the only vitamins as nuts and bolts — things that would be available at the simplest of hardware stores. the reason i’ve undertaken this design goal is to make the project widely accessible, meaning that as long as you have access to the laser cut (or, potentially, 3d printed) parts, you could create the entire device in an afternoon with nothing more than a box of standard screws and some tape. the *ultimate* goal of keeping it accessible is that 1) the parts would be easy to come by, 2) people of limited technical skill could easily assemble it (say, in an afternoon, as a project with your kids), and 3) it would be extremely inexpensive compared to similar technologies, meaning that even very low income folks could afford it — in this case, the entire axis featured here is $7 total ($1 in material, $1 in screws, and $5 for the stepper). Using acrylic brings this up to about $10, but I’m using the hardboard to prototype the design and work out the bugs.

    ultimately, in terms of its use for the open sls 3d printer, you can think of it in terms of a bootstrap device — it may not be terribly high precision right now (about 0.3mm accuracy, in its current configuration), but that’s easy to change. And, like a repstrap to a reprap, you could inexpensively print out a better, more precise linear axis upgrade design using the linear axis as a component of printer itself.

    i tested a bunch of other designs for a complete 2-axis x/y system using linear slides (even using idler wheels, someone mentioned this earlier), and these are detailed on the reprap builders blog in earlier posts. each had issues with shearing or binding, and this new design is just fantastic in those respects. with the proper spacing between the gantry and the linear rails, things seem to work really well — far far better than you’d expect. there’s definitely going to be some backlash in the system, which i haven’t fully characterized yet (to see if its relatively constant), but that’s mediated by the fact that for rastering (for, say, SLS) you only go in one direction, then back in the other — so you never have to deal with any backlash issues, only monitoring endstops. i can only imagine the motion will get better with a better aligned gearbox — i just eyeballed this one, so its fantastic that it works at all! :) i have also thought of making a version with an etched encoder strip on the linear axis, so that you could sandwich the axis between a cheap opto endstop and have an okay encoder available to you.

    the real creativity and innovation in this project is seeing what you can do when you limit yourself to a given set of build tools and media, with the (possibly lofty) goal of creating an accessible, inexpensive (under $200), and easy to assemble 3D printer for novice makers everywhere to enjoy.

    (if other grad students or folks with access to maker tools with an engineering/computer science background are genuinely captivated by this project and would like to help the design and building efforts, please feel free to send me a message — my phd thesis takes up a great deal of my time these days, and extra helping hands who are genuinely inspired by these ideas would be wonderful!)

    be well

  6. “Even something as simple as having a humid day could mess with the accuracy.”

    What about UHMWPE? It is fairly cheap and is laserable. I should slap some on my laser cutter and see how accurate it is able to be laser cut.

    You could always waterjet cut it but that requires access to a waterjet. Or make the racks out of aluminum or stainless? Or hardened steel that you waterjet cut to avoid changing the temper? A waterjet would give you the accuracy you would need in a material that would be stable and wear resistant.

  7. @peter

    While I do understand that you want to keep the complexity and cost down, I still don’t think that’s an excuse to make a subpar system.

    You are already using exotic parts and tools (lasers/steppers) so I really don’t understand why you aren’t using at least things like roller blade bearings and such.

    In any case – keep up the good work!

  8. @andrew

    You wouldn’t believe the steppers that I’ve gotten out of old lab equipment. Find a large university that has a surplus store, or that auctions off old equipment. I get old lab equipment for $5 that contains massive transformers, various steppers, etc inside of it. I rarely buy new parts, I have so much used stuff laying around from scrapping stuff like that. Best part is, I almost always break even or *make* money off of the equipment as it is generally of aluminum construction and is worth more than $5 by weight. I wait until I have enough to fill my car’s trunk, then take it in to the scrap yard!

  9. i like the idea, if you have a laser cutter it would be very cheep, and a 3d printer unlike a mill or a router he does not need to be repeatable to .001″. he can probably get up to about +-.02″..good enof

  10. What if you combined this idea with the DIY 3d printer idea mentioned a few posts back? Backlash isn’t an issue because you are only going to move in one direction and you only need to move the elevator down. It would be much simpler than a ballscrew system and would be possibly even 3d printable so you could actually reproduce it. That or laser cut it – the accuracy should be fine for 3d printing applications.

  11. @bud

    +/- 0.02″ = +/-0.51mm.

    I don’t think many folks would be happy with 1mm misalignment registration steps all over their parts…

    The main concern between a mill / router vs 3d printer / laser cutter is whether there is a significant load on the head during “processing”. Mills / routers need to be incredibly rigid to retain their accuracy and repeatability under load (i.e. when the head is cutting).

    For me, the repeatability on a 3D printer needs to be at +/-0.1mm = +/-0.004″ to be useable.

  12. You really should consider spring loading the gears on one side of the rails, such that they push the whole head onto the opposite rail. This will help a lot in dealing with lash, binding up, bad alignment, and possibly some of the accuracy issues as well (half and quarter-stepping the motor is pointless if the teeth never engage due to slop). Come up with a system in which two gears are stationary on one rail, and the opposite side gears are individually sprung.

    Also, in terms of accuracy, smaller pitch (tooth size/counts per inch) generally leads to higher accuracy and less lash.

    I like how you slaved all the gears to the stepper (it is a nice simple mechanism that way), but generally, you *don’t* want to do that.
    Driving only one gear on each rail is the best way to go. With “4 wheel drive,” the head is less able to deal with lash, and you have less design options to counter the problem.

    I like the kapton tape. A lot of people will bitch at you for not using skate bearings. Don’t listen to them. Lots of high precision industrial machinery use teflon or plastic slides.
    In fact, if you tear apart some of those 100 dollar plus slide blocks, you will find plastics inside.
    Skate bearings are NOT going to be more accurate. Regardless of design (slides or bearings) they can be only as accurate as the surface they travel on. But with bearings, unless they are 100% perpendicular to the direction of travel, will offer up all sorts of binding and mistracking. The only advantage bearings have over slides is lower friction, and could actually introduce *more* points of slop.

  13. oh.. and this might sound like a *crazy* idea, but trust me its not:
    since your envelope is rather small, you could spring the whole head to one end of the rack. you would effectively have zero backlash, and the modification would be two holes and a spring.
    You just need to find one that has a pull force just a bit more than the weight of the head and stepper. a letter weight scale (perhaps 2) might about right.

    This has been done on a few machines I have worked on using pneumatic cylinders at low pressures rather than springs.

  14. the steppers that i’m using are KP35FM2-035, which i get at sayal electronics ( ), which has a bunch of stores in and around toronto, ontario, canada.

    mre, those are some really interesting ideas! keeping with the reprap philosophy and ideal of making 100% of the parts rapidly prototypable, the current design (save for some tape, and possibly glue, as well as the stepper) is nearly entirely 3d printable, were one to replace the screws with printed plugs or shafts. the whole idea of this project is that there are a LOT of ways to build a successful axis if you have access to any parts, but with some creativity you might be able to come up with comparable methods using entirely laser cut or 3d printable parts. so i’m not sure if i would be willing to add springs yet, for the following reason:

    the springs are intended to keep the axis tight against the linear gear, to prevent any slop — but i can design the linear gear to slide in then be snugged up with some bolts (and, infact, did this the other day :) ), dramatically reducing the slop.

    trying to drive with two instead of four gears (and having two “idler gears” is an interesting thought — i wonder how well it would work? it would certainly make the captive gear box far less complicated…

  15. I get lots of stepper motors by dismantling old printers and scanners. Friends of mine know that if they are getting rid of an old printer or scanner that I’ll tak eit off their hands.

    These are also a good source for ground rods. I’ve got a selection of rods from 6mm to 11mm in diameter from various devices. The trick is finding two identical – easiest if you can find two very similar printers as each maker uses a different size.

  16. OK so instead of springs what about something like surgical tubing? I have no experience with 3d printing but I do have a little with molding and have successfully made “stretchy” tubing in a pinch using a straw and wire for a mold and RTV as the fab material.It didn’t last long but it was off the shelf material.If you are going to be3d printing you are going to be resourceful in materials acquisition anyhow so why not get started right away !

  17. Many lightweight industrial designs use two stacked spur gears with a torsion spring between them. This pushes the design “out” of the rack, but against the far rail instead, or against a precision glide on the back of the pinion. This reduces the need for absolute accuracy of the gearing itself and allow for automatic wear compensation. The rule of thumb for most industrial designs is W<2*D. The width of the separation of the rails (or drive gears if uses as supports) should not be more than twice the distance between the front and rear support areas/wheels/gears. This prevents binding in a passive manner.

    As long as you don't push harder than the spring can push back, you'll get zero backlash operation. This type of spring could easily be printed out of a more flexible plastic (similar construction to the Mendel chassis springs IIRC).

    Guess it's time for me to catch up with the reprappers again.

  18. just continuing the ideas set forth on these replies. driving a single gear from the stepper will give you the biggest bang for the buck. those other gears can just be tensioners, and in all R&P tables I’ve seen, any extra “wheels” on the rack are independently adjustable. every iterative design I’ve seen progress has settled on that solution, and for good reason, too. driving the pinion independently of the other gears does two things, you can design different ratios for your motor without drastic alterations, and it makes tramming the assembly easy enough for a 10 year old.

    wonderful work, btw. I love seeing these things on HaD.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.