Open Source 5-axis CNC Router

This 5-axis CNC router could soon be an open source tool. [Mike Calvino] built it for the School of Architecture at the University of Arkansas. It can be used as a router or as a plasma cutter/welder. Now he’s trying to raise some money that will underwrite his time and effort to develop and release instructions, design files, and specifications to make it an open source hardware project.

It is extremely large, and in addition to the X, Y, and Z axes that you’d expect to find on CNC machinery, it can tilt and rotate the cutting tool. This is not something you’re likely to build at home. But the availability of plans would be a huge contribution toward making machine tools accessible at a relatively small price tag. It’s not hard to image universities building this as a class project. We also think it would be a perfect group project for you and your buddies over at the local Hackerspace to undertake. Check out some milling action in the clip after the break.


44 thoughts on “Open Source 5-axis CNC Router

  1. I came to also say holy crap. Awesome.

    I don’t want to put the brakes on this because it is awesome but building this, as the summary suggests, is a huge undertaking.

    Wonder what the costs are in parts alone? I would have to guess easily $20k but considering a machine like this would run 50 – 100k easily- that’s a bargain. But time and labor counts for something, even if you have the design already done (and open sourced)!

  2. Protip: if you want to show off your machine, don’t show it cutting for 5 minutes straight. 5 minutes of cutting = ZZZZZZZZZZZZZZZZZZZZZZ.

    Show it progress. Time lapse it. Use video editing techniques!

  3. Based on how often the side of the tool was crashing into previously carved areas, and the parts where it was rotating on the same axis as the cutter, I think it wasn’t carving anything in particular.

  4. As macona noted, the hardware side of this is trivial. Guys at cnczone have been building machines of this caliber for <2000$ for years.

    Its the software that will drain five figures from your wallet. Computing toolpaths for 5 axes is mind bendingly complicated. I can't even imagine the complexity involved in programming in the parameters for your machine so that it doesn't run into itself.

  5. did a quick skim of the article, but didn’t see anything about accuracy… what kind of accuracy are we looking at here? i ask because it looks like the whole carriage has to move for the z-axis… that’s a whole lot more work than just plunging the tool head ala a haas machine.

    then again, i only have experience with 3- and 4-axis machining. does the extra axis change that much about the machine?

  6. Not sure why 5 axis is seen as so complex and expensive software-wise, our limited and low budget workshop has a full 5 axis machine and the (inverse)kinematics for 5 axis are not too complex at all?

  7. Open source 5-axis cnc router and he’s proposing to use software like 3D Max ($3500) or Rhino ($1000) which are closed and Windows only.
    There’s no point of using either of those, because Blender 3D 2.5 exists.

    1. Deskproto now supports 5th axis machining.
      The commerical license is less than $2k.
      Hobbyist license is something like $300-400.
      Go to their website.
      The software is not that expensive anymore.

      I’m sure the more expensive software has different functionality, but you can get good and cheap 5th axis software.

  8. @ryan – I dont have access to that right now but I can pretty much guarantee our workshop wouldn’t be paying that for it, they get upset spending a K on tooling. As I say, the maths isn’t THAT complex, no reason why someone with a bit of spare time and knowledge couldn’t knock something useful together for it.

  9. “no reason why someone with a bit of spare time and knowledge couldn’t knock something useful together for it.”

    Knocking something together and making something feature rich and useful are VASTLY different things.

    +/- 0.125″? If that’s true that’s…. terrible. We hold 0.003″ easily all day long.

  10. Cheap, Accurate, Practical- pick any two. Or alter the metrics of those parameters?

    Some of those are inherent in the technologies we currently are using. There’s exceptions like the Makerbot and RepRap projects. This 5 axis Hack counts as a attempt to lower the entry bar by creative applied engineering. Call this an example of “good enough” at work.

    Subtractive Fabrication does have a place alongside additive methods. The difference between Addiive and Subtractive tech often allows useful work at far larger tolerances for Subtractive.

    If you’re making a bolted together from 2X4’s object Vs a 4-6 mm per face D20 your tolerance stacks can be indeed that far apart. You just do not need Jo Block tolerances for bolted wood picnic tables in the real world. And a 5 axis can do art/furniture you would have to work harder at to make any other way.

  11. Cheap, Accurate, Practical.

    Why not “fairly inexpensive, reasonable accuracy without needing lasers and CMM, good open sourced software with an eye for usability, ease of use and polish”.

    Why can’t we have all of these? Linux does all of these things and gets better every single day. Commercial units can do this but cost solid money. Why can’t we get a good enough unit that is open sourced, easy to repair and yet also has a solid software base? Surely these things grow and build on each other and get better over time?

    1. Maybe you can be the one do do it first CutThroughStuffGuy, and then you’ll probably know why no one else has done it. We CAN get a good enough unit that is open sourced, easy to repair and has a solid firmware base, but SOMEONE has to be that guy that actually gives it a go. Maybe it will be you?

  12. Nice try. He’s asking 16k for documentation and release as open-source while he could just dump all info he has and let the community sort it out. But he fails to mention the basics so I guess the machine cost him too much for what it’s capable of so he’s trying to sell

  13. Wow and I thought I was hard to impress… I don’t know what you guys are expecting for an OPEN SOURCE project. I personally think this is an extraordinarily good thing for the community to have access to.

    I tell you what though if you think you can make it more accurate for less do it. Until then keep your mouth shut if you have nothing of value to offer.

  14. Im impressed I wondered what was involved in a project like this.

    As to accuracy. Im sure most of the open source 3 axis machines started like this. Once more people are involved and more man hours are spent on refining it then the accuracy will improve with time.

    The software will follow as well.

    Well done to Mike Calvino for getting the ball rolling and donating his time and expertise.


  15. Serious question:

    Why not reverse engineer an existing high end machine? Make a BOM, parts lists, assembly instructions, troubleshooting info, etc? The software might be the holdup if it had to be rewritten from scratch but this is what the Chinese do and look at what it is doing for their technological prowess. Right or wrong, they are moving forward by decades in only a few years.

  16. @CutThroughStuffGuy

    There’s no secret in the hardware. A high-end 5-axis machine is expensive because it uses high-quality materials made to tight tolerances. You can reproduce one, but you’re not going to save much (if any) money. This project is more a study in what you can accomplish on a lower budget with inferior parts.

  17. @Neckbeard: Please don’t promote the common misconception that ‘open source’ means ‘low quality’; there are a lot of high quality open source projects available. It is understandable how that widespread viewpoint about OS projects came about, though, as, unlike commercial ones, they don’t hide away until all the bugs are worked out, but will usually ‘share the adventure’ as soon as their devs have anything to show. Hell, many are in constant use even before they are out of beta, or even alpha.

    @naysayers: Seriously, guys? I doubt that even one of you has even attempted to build anything more complicated than a sandwich.

    This is (AFAIK) only the second CNC machine that he has built, and the first 5 axis. He’s having to learn as he goes, and doesn’t have a huge R&D budget. Give him a bloddy break.
    Check out some of the other large-scale, high-precision CNC and 3D fabrication projects out there; none of them started out matching the output of commercial rigs, either.
    As far as tolerances go, he repeatedly says “should be assumed to be +/- 0.125 inches.” In other words, he probably hasn’t measured how much it is out, but knows that his output isn’t quite dead-on.

  18. I agree-and more, with DeadlyDad.

    I’ve both built and serviced stuff from drop dead simple “driveway bells” for a campground, on up to retrofitting a shut off for a brake lathe. So much of our world has to have a multiverse of “serial # 1” projects before we get to the levels of performance we take for granted.

    That’s a rather different concept by the way, from blessing or slagging any one Hack/Project for ANY reason.

    For every thing in our lives that’s become what it is, there’s an often startling length of tool chain from Iron, Oil, Beach Sand etc to the finished product.

    Building a 5 axis on your own is both an education of both learning “how to” and teaching others who “audit the course” watching it on HaD and in other places. There’s a few of us who may be somehow “better” in our minds than Mike Calvino. Not me. He’s become the better of many more HaD posters as Mike went and made something that works. Armchair Critics Vs those who make stuff work? Looking it from that POV? That says a lot.

  19. @DeadlyDad wind your neck in. Seriously. The man is a 1 man band doing a project it is bound to be inaccurate, it’s only once a device develops a userbase does it improve. Don’t be so naive, besides I’m more than happy with 0.125 for general purpose stuff.


    Good post mate.

  20. I think this is realy good endaver we nead all the defrant kinds of O.S. Tools that we cane gate awer hands on, kind of like what the printing prase did for books.
    Like the rep rap and the makerbot it will improve over time espasly if we mix in some blander revarse kinomatic with it.
    This is a realy good place for the comunay to staret gating some experance with this stoff
    IF we remamber blander histery to OS you might recole the comunay paing a big price to make it open sorce becuse he nead to save somting or ather and now thay are doing inter HD movies with it. ther is nothing rong with paing for someting to be open sorced espsaly if the farst one on the harzen.

  21. Man! This is a awesome plan of attack. If I ever come up with something great, I’m going to sit on the info, and hold it ransom.

    In fact, I should just design things that LOOK cool, but don’t really work. Throw them together, snap a bad picture, get it on HAD, and then wait for the money to start rolling in.

  22. I hadn’t seen/heard where he mentioned the tolerances shouldn’t be assumed better than +/-0.125. That is pretty awful, but not surprising as in some of his videos you can visibly see the Z column shaking as the rotary axes turn and stop.

    I have my own 5-axis build currently in works. The design is about 98% finished and I’m in the process of gathering parts/quotes. I hope to be cutting this spring. This is my first CNC build, and yes it is definitely a learning process.

    If anybody is interested in my build, the link my flickr link is here:

    To the guy who said people have been building similar projects for less than $2000… you couldn’t be more wrong.

  23. I’d love to see open source 5d cam and someone using 5 axes on EMC.

    I’m not discounting the difficulty in designing and building a 5 axis machine (I built a MechMate) but the software is what’s missing.

  24. If the tolerances are too loose for you why not help the man out? Tell him where he has gone wrong or even better get a kit and start developing for it!

    The tolerances of the machine can easily be solved and improved take for example the jitter mentioned above. Sheesh…

  25. Making A 12 Axis CNC
    For the Hack A Day People I whold go with.
    6 degrees of freedom one on the botome as a holder and the ather on top as a carver.
    Then I will mount two 6Dof SixAsix PS3 on ether side.
    then I will mount AR tool kit on the botem of the tool hade and the holder with a corasponding camra.
    and a free head tracker.
    That way all three divices will be spiting out 6Dof data that can self calibrate each ather.
    Then I will grabe macing machean that is smaler and liter for hand carving with a dremel and a block of wood on the stand and start carving, all the data will be stord on the computer in blander and the big 12axis CNC will be abeal to replacate my movment for aver, and bay the way taligances is a mater of gear racheaos and strong asambly so I will make a 20 gear seat kit.

  26. @Neckbeard: Sorry. Reading your post again, I can see that I might have taken it the wrong way. I guess that I’ve gotten too sensitive about the persistent notion that a project is only open sourced if it isn’t good enough to sell. (This summer I almost convinced a school to save the money that they were going to spend on upgrading their computers and software, by going to a Linux setup (LTSP) that ran OpenOffice/etc. They were all up for it until they wanted to talk cost, and I said that the software was all free because it was open source. They promptly changed their minds because they ‘didn’t want to have to deal with second rate software’. :sigh:)

  27. @joker EMC is simply the controller software. It is not CAM software. For CNC you need both. CAM creates the toolpaths and generates the g-code, the controller software interprets the g-code and controls the motors on the machine.

    There is no open source simultaneous 5 axis CAM software. Even CNCtoolkit requires 3DSMax.

  28. TReid : Actually, 3DSMAX has a FREE version called GMAX. you can use CNCTOOLKIT with GMAX. GMAX is exactly 3DSMAX version 5 minus the saving and exporting capabilities, which is not needed if all you want to do is save GCODE.

  29. In regards to the 6 Dof CNC post hear is a Hexapod this is what I have bean toking bout but with two of them one for the router and the ether for the holder 2 time 6Dof will be nice. with two input device like. Novint’s Falcon 3D Touch Controller or two Microscribe one for the holder and one for the dremal touter for houmane input path programing.
    So this is how I wold do it Carving a Head out of Wax.

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.