$400 DIY CNC Machine Is Surprisingly Simple!

Once you go CNC you never go back — they’re just too darn convenient! [Drez20001] shows us how he made one for around $400. Who needs expensive roller bearings when you can use drawer slides?

That’s right — the majority of the cost of this CNC machine are the things you can’t really get any cheaper — the servos (or steppers), the belts (or leadscrews), and of course the motor controller plus computer interface. Everything else? Plywood, drawer slides, and a bunch of fasteners can be had for next to nothing!

He had wanted to build a CNC for years but was mostly hesitant in doing so due to the cost and apparent complexity of the build, but when he started to look into it seriously, he found it really wasn’t the case! It’s built on the basic gantry system design where the X-axis drives a bridge containing both the Y and Z-axis. It’s not a heavy duty machine by any means (he just has a small dremel-like tool in it right now), but for his purposes it’s more than enough.

One rather creative way he saved a few dollars is with his motor couplers — he’s actually taken rubber gas line and cinched it onto both shafts, which he says works quite well!

If you’re looking to spend a bit more and want a stronger machine, you might want to take a look at this aluminum variation we shared a few years ago — same basic idea.

[via Hackedgadgets]

33 thoughts on “$400 DIY CNC Machine Is Surprisingly Simple!

  1. Nice one!
    I built a machine for around $300 some time ago. I also used rubber hose couplers and they are very strong. My machine lacked end-stops and I thought the rubber hoses would act like stops (I didn’t hold them in place, just slit them over). Wrong! My machine almost broke.

    In my experieance you can built a very cheap machine but you have to go for accuracy over speed. My lead screws where just threaded rods. Much slower but cheap.

  2. I also have hose couplers on my machine. I also currently lack endstops and by not tightening the hoses, they act as a clutch and will slip before serious damage occurs. Very cheap, simple and effective.

  3. My issue is not with the hardware being difficult or expensive but the SW side. still have a half finished CNC here, as I haven’t been able to start on the SW side. G-code, Cam, 3D drawing etc….. that is the more challenging part for me

    1. If you go Linux you can come out pretty cheap on software, EMC is free and I hear rather good. There are free CAD/CAM softwares out there too. If you are doing 2.5D work then I suggest the free CAD program Inkscape, its free and has Linux and Windows versions. As for CAM software, I looked, I’ve tried but in the Windows world you have to pay a little dough to get something worthwhile. MakerCAM is free but buggy as hell, I dropped $150 to get Cut2D which is a very sweet package, I hear good things about CAMBam too (same price range). I personally have a Inkscape/Cut2D/Mach3 setup. You can go uber cheap using one of the Arduino controller setups w/o a PC too.

  4. I’ve seen designs costing little, and ones that cost a lot. I’ve built one in the ~$3,000 ballpark and still have little slop in my Z axis. I can only imagine what kind of tolerances this thing can hold…. Cool design, but only if it can accurately cut what it’s designed to cut.

    1. What kind of accuracy are you looking for with a wood router? Most woodworking demands about a thirty second of an inch to be considered fairly accurate. Other than the letter spacing I think his experimental sign came out fine. That was likely due to his software, not the machine.

      When I worked in a tool and die machine shop we did not expect incredible accuracy out of our cutting machines. That was why we had a grinding department. Then again we were professionals.

      1. Agreed, the key to having a professional product is making it look good. If you have a certain degree of known slop and can account for it within the process so that the end product comes out to the requested spec, then that known slop doesn’t end up mattering (in most cases). Cases where it would matter will have tighter specs, etc… Not everything needs to be done to 5 zeros… and certainly not woodworking. A 10% change in humidity will smack that precision right out the door…

        1. I tend to think of it all as the Internet weenie factor myself. It all smacks of audiophileism gone mad in a certain way. You know, the same as people chasing after frequencies that they cannot hear, or levels of distortion they cannot discern. Which goes hand in hand with frame rates of video games that they cannot perceive too. All a byproduct of the Dunning-Kreuger effect I’m sure.

          1. The framerate issue isn’t that simple, because video games employ vertical synchronization to prevent visual artifacts. That means the real framerate is always a full division of the screen refresh rate. The point is to make your game always run faster than your screen, because if it ever drops even a bit slower, then your real framerate gets halved from 60 to 30, and the difference is noticeable.

            The other point is that games that are designed to run in a single thread are computationally bound by the framerate, because they’re programmed to wait for the frame to draw. Hence why the faster your framerate, the faster your reaction times because the game goes through the input loop more times per second. If you’re running 30 frames a second and it’s triple buffered, it technically takes 33 milliseconds for the game to register your input and 133 ms before the results appear on the screen, which is almost doubling the average human response time.

            If the game’s running at 300 frames a second, then the input lag is about 16 ms or fast enough that the results show up in the very next frame sent to the monitor, which makes it neglible.

          2. @Dax A quick Google search returns the average human reaction time as 0.25 of a second. Apparently you suffer from Internet Weenieitus yourself. BTW I’ve heard your argument more times than I’d care to admit. If you’re actually good then you don’t need all of the time that you think that you do.

    1. Have a link to a free version of Solidworks or Inventor? because if most of us could afford the licensing for either (especially now that autodesk is a subscription) for hobby use i’m not sure we would be talking about a $400 CNC machine.

  5. Instead of double sided foam tape to hold the foam flat, I would duct the extraction vacuum to a vacuum table (like a reverse air-hockey table) to hold the foam flat. Simple and should work great.

  6. I built one using some scavenged steppers from old Laserjet printers, but splurged a bit on some good Igus slides for the Y and Z but used a pair of drill rods for X to carry the gantry, good lead screws and anti-backlash nuts on all axes. HobbyCNC controller and a scavenged power supply ($30) from an old film developer got me to maybe $500 on the machine not counting the Bosch Colt router. Old Compaq laptop to run Mach3 and some free CAD S/W (Cambam etc.) and it does really well. Birch plywood and MDF for the bed and frame and gantry with some minor fabbed scavenged aluminum bits for brackets and such. So maybe twice as much for a solid, very accurate, fairly powerful set-up. Nowadays some cheaper chinee stuff and electronics would probably enable the build for not much more than this guy’s machine and provide much better performance for the money.

    But props to the guy for jumping in and doing it, hope he enjoys the machine.

      1. I make a conscience effort to avoid Chinese machine tools these days. I didn’t always, but let’s just say that over time my views on the matter have altered for a variety of reasons. I understand that today it is difficult, but worthwhile things in life often are.

        The Bosch router I use as a spindle has a Proudly made in USA sticker inside of it. I know because I saw it myself.

  7. Jesus, would those doubled up cat5 cables really support the high current, even per phase?
    Still I love what he’s done here. I had the idea of using drawer slides myself but failed on the construction and design elsewhere. i’ve thought an awful lot more about it since my failures, and feel that this gantry design is good provided the Y and Z gantry is fixed and X moves underneath it.

    1. Ah, what high current? I’m pretty sure this machine uses one of those imported multi-axis TB6560 based driver boards. If that is the case then most likely the fixed maximum output of that is 1 amp per phase. I know the manufacturers of those things claim differently but that is not what has been observed.

      It isn’t even hard to measure stepper current draw. But few actually do.

  8. Drawer slides have a bad habit of being made out of softer steel alloys, so they wear out and become loose, and eventually the bearings pop out one by one. You might need to replace the slides every so often.

  9. This $400 machine could have been a $300 machine if the controller was the (free) LinuxCNC not the $100 Mach3. Or was the computer and software not included in the estimate?

    1. I’ve heard a lot of those imported driver boards ship with a CD that contains a pirated version of Mach3 on it. Though those in the know would say the user is still getting ripped off. heh

  10. Points here for using drawer slides. Prior to being allowed to purchase a CNC chinese machine, I had pondered building one and began getting slides from work when we didn’t use them for server and equipment installation. Started hoarding steppers too. Never had time to work the design to anything other than “it’d be nice”. Good to see it works. I would caution against CAT5. As it flexes over and over it will break and get intermittent. I suggest swapping it out with Belden 8723 or equivalent.

  11. Thanks for the replies, all very appreciated – just saw the crosslink at Hackaday. I’ve since upgraded the machine to hold a standard sized router. The design is the same – just has a larger gantry. The results have been very good. The registration has been far tighter than I thought it would be. It’s been good enough to do pretty clean lithopanes and a number of signs using mostly open source software. A lot of folks said the machine would have too much slop but it’s been pretty minimal until I got to the far extremes of the slides. An alternative to help increase the rigidity further would be to run the slides in opposing directions. Lots of fun and a good starting point for getting into CNC. The rest of the videos of lots of signs and lithopanes are at my YouTube channel – drez20001. Thanks again for the comments.

  12. I’ve used vinyl hose from the hardware store as a shaft coupler on my drum coffee roaster for years. It easily handles quite a bit of shaft misalignment and transfers torque very well. To quick-connect the drive motor to the drum shaft, I used an off-the-shelf 1/4 inch drive socket. I filed the end of the drum’s shaft to a 1/4 inch square that slips into the drive end of the socket. The other end of the socket is clamped in the vinyl hose.

    Sometimes the cheap solutions are better than the commercial ones.

  13. I’m a poor guy working in a german car-part factory and our house
    Could definitely need some new windows….
    Would a machine like that, made from oakwood (got plenty of it for free) be able to cut windowframes from oakwood?

    Screens, gasgets and hinges are pretty cheap and a Machine like that could save a lot of money. Cutting time doesnt really matter as its just to save heating oil. :-)

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.