CAD files for the Antikythera Mechanism

Update June 2015: I’m revisiting this project to finally get it to a version where it can be manufactured at home. Check for future updates. I have also withdrawn the files and re-released them under a similar noncommercial license.

Update June 2016: Haven’t forgotten about this project, but I’m back in grad school so free time is limited. I’ve gotten it to a point where I’ll need a laser cutter to test out design options, so things are pending me finding one I can get.

By request, I’m making the CAD files for the model of the Antikythera Mechanism that I created freely available. (Click that link for an overview of the project.)

Download the Antikythera Mechanism from Google Drive here. You’ll need Solidworks to open it, and 3D printing will probably not work. I’ve gotten asked that a lot.


It isn’t perfect. There are a few issues that would keep it from working if you 3D printed this, and a few more issues that make it not completely authentic to the original Mechanism. Here are the places where the model differs from the Antikythera Mechanism. Italicized issues keep it from working if it was 3D printed. I estimate it would take me 40-50 hours to fix them (it would require a redo of a lot of parts), which I don’t really have time for right now. But… if someone had access to a 3D printer and would print me a copy pro bono, I might be motivated enough to clean it up.

  1. All gears except for the lunar phase geartrain are assumed to have a module of 0.5. This is actually the mean value of the modules for all gears in the Mechanism. A more accurate model would account for variations in the gears. In one instance, this creates a collision between a gear and a support as it travels around. A few gears should also be thicker.
  2. I didn’t use values for the pin placement radii on the epicycle gears. These can be found in Freeth and Jones’ “The Cosmos in the Antikythera Mechanism.”
  3. Two indicators should slide onto the Metonic and Saros pointers, and track in the engraved spiral on the back face. Solidworks had issues with a tangent relationship to a spiral, so these are suppressed in the model (but are still there!).
  4. The Antikythera Mechanism used a combination of adhesives, pins, and friction fits as fasteners. The model uses no fasteners, and would need a few if printed. All the gears that need to have locked rotations are keyed and spaced out with spacers, but some of the keys are really small loose pieces and should be changed.
  5. There are no tolerances built into this model.
  6. The Antikythera Mechanism used triangular teeth. The model uses involute profiles. Solidworks was having issues creating these profiles in a couple cases, and a few gears don’t mesh correctly.
  7. The wooden frame, pointers, and posts are creative interpretations. The large pointers should be removable in the real model to allow the indicators to reset.
  8. The planet positions are not zeroed to any celestial day.

Despite these differences, the form, function, and essence of the Antikythera Mechanism remain intact.


Also, some licensing info!

I am releasing this to the public under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 Licence. This is a mouthful, but it basically means that you are free to use, edit, and redistribute these files so long as you a) credit me for originally making them, b) don’t use them to make money, and c) apply the same license whenever you redistribute it. This means they will be forever free. With the following exceptions!

  1. The files “plate back lower dial v4.png” and “plate back upper dial v4.png” are taken from Freeth et al. “Calendars with Olympiad and Eclipse Prediction on the Antikythera Mechanism: Supplementary Notes”, and slightly edited by me. Copyright belongs to original owners, however because it is a published paper all educational uses are fair use.
  2. The file “plate front v2.png” is taken from this image, originally made by J. Evans and A. Thorndike, and edited by me. Copyright belongs to original owners. (They might mind if you use it educationally. I doubt it though.)

This Post Has 52 Comments

  1. Yanis


    Great job there mate!
    I wonder if you can help me a little.
    I want to 3d print and assemble the gears representing the movement of moon.
    Can you please direct me?

    Thank you

    1. Scott

      Hi Yanis,

      The complete gearing for the moon is large and complicated, but if you just want the part at the top that makes it spin on its axis, you could make it pretty easily. You have to design the parts yourself though – these files aren’t set up for 3D printing.

  2. Denis

    Any chance that you could put a copy of the STL files in the repo you have made?

    There is lots of us out there that don’t use solidworks that could support the development in our own software preference.

  3. Mr. K

    Great work!!!

    Any update on this?

    I’d be willing to help with the physical printing if you’re interested in getting this working 100%

    Would be an amazing option for intellectually minded people for Christmas

    1. Scott


      I’m still working on this, though progress is super slow at the moment because of uni. The plan is to get a version that can be laser cut from sheets of wood, and release the files to the public. But that won’t be done for probably a year – I’m going to start rework over the holidays and pick it up again over the summer. I can definitely let you know when I hit a milestone.

  4. Trixie Ray

    Hi Scott, why did you choose 52 teeth for gear k2? Doesn’t the model in the ISAW paper specify k1, k2, e5, and e6 all to be 50 teeth?



    1. Scott

      Hi Trixie, good catch! I tracked it down and it was an error in my gearing spreadsheet – k2 should be 50 teeth. Thanks for finding that, it probably would have gone unnoticed in the remake of the model that I’m doing.

      1. Trixie Ray

        Hi Scott, thanks for the response. Great work on this! How did you choose the size of each gear? Are you modeling to 1:1 scale, using the Supplementary Information section of the Nature 2006 paper, Note 3 (Gears) for the sizing info? Axle K is modeled with a 1.00mm epicenter offset for k1 and k2. Gear k1 is modeled with the distance from the center of the gear to the center of the pin as 8.00mm. How did you determine/where did you source these values?


        1. Scott

          The sizing of the gears was pretty crude – I took the number of teeth for each gear and multiplied it by the mean module for all gears (0.47, from the ISAW paper). This worked, but it loses all the detail in the varying modules between the teeth (from roughly 0.44 to 0.58). In this new version I’m working on I’m basing it off that table in the Nature paper, after manually ensuring a module match between meshed gears. All the gears are a 1:1 scale, but the overall size of the model doesn’t fully line up with the actual mechanism.

          The epicycle axle and pin placements were arbitrary because I was crunched for time at the end of the project to turn it in. But the planetary pin placement can be found in appendices of the ISAW paper. That paper also says that k1 and k2 are offset by “just over a millimetre”, but I’m having difficulty finding a written number to distinguish between 1.0mm and 1.1mm. The heavy corrosion of the mechanism makes me think the difference isn’t all that meaningful.

          Good questions! Are you working on your own model? I didn’t really expect anyone to dig that deep.

          1. Trixie Ray

            Yes, I am working on my own model and build of the mechanism. I suspect mine will be oversized, at least 2:1 scale. I will be prototyping in plywood and intended to do the final version aluminum. Your model will serve as a great reference to cross check my own work. I will keep you posted on my progress.

          2. Trixie Ray

            I noticed that you had a separate fixed gear for venus and the solar anomaly. You should be able to share a single 40 tooth gear for both trains.

  5. Evgeny

    Hi Scott, thank you for the perfect work,

    did you update google drive model ? is something new available for experiments ?


  6. Dave Goodchild

    Hi Scott,
    Wow – only just found this, but I’m much impressed! I too am having a crack at building this amazing contraption but I’ve opted for hand cut wooden gearing as that’s what I’m used to doing (I have been known to build the odd orrery or two). I’m also including the full five planet display as well which I’ve pretty much finished. Unfortunately I know nothing of 3d printing, but having hand cut 60 or so gears Ican see the arattraction….
    The moon gearing (k1 and 2 especially) is probably the trickiest bit to set up, and don’t forget that although they all have 50 teeth, k1 and E5 are slightly smaller than K2 and E6 – but I’m guessing that you’ve already spotted that!
    My machine is about 50% bigger than the original which is partly down to ease of building it, but also down to the fact that the original is really, really small, and I just wanted to do it bigger so that the person winding the handle has a fighting chance of actually seeing what was going on internally. It will be all wood when finished and hopefully quite skeletal in overall design, I want to be able to see the gearing without horrible perspex! I’ve put links to a coupe of photos of how I’m getting on if you want to have a peek.
    Anyhow, best of luck fitting this in with your studies. I’m looking forward to watching your progress.



    1. Scott

      Dave, that’s fantastic! It looks like you’ve made pretty incredible progress towards finishing it.

      It looks very similar in approach to how I am planning on doing it, though I’ll be using a laser cutter to make the gears since I lack the patience to hand cut them (which I’m very impressed with). I’m pretty sure we’ve also found the same source for the telescoping brass tubing.

      If you don’t mind me asking, how did you anchor the ends of each axle, so that the gears can rotate together and still stay firmly on axis? Especially the axles that only have room to be supported at one end, since you can’t constrain those tubes between two supporting plates.

      1. Dave Goodchild

        Hi Scott,
        Sorry for the late response – been a bit tied up the last few days.
        Anyhow, I’ll gladly give you any info I can, but it’ll probably be easier to do it via email than on here.
        Send me an email to and I’ll ping back how I put my machines together. It may be different to what you are able to do though as I’m using wooden gears obviously which are quite forgiving and accommodating in all sorts of different ways when it comes to fixing and fitting – plastic gears may need a bit more thought…
        Anyhow, drop me a line when you can.

  7. Trixie Ray

    Hi Dave, nice work! Whose Cosmos gearing did you choose, Freeth, Evans, or Wright?

    1. Dave Goodchild

      Hi Trixie,
      Thank you! The planet gearing is to Colin Freeth’s proposed design, for no other reason really other than it builds up into a single very neat self contained unit with minimal gearing – it really is a beautiful design. Was the original like that? Who knows, but it works for me and runs beautifully!
      Any photos to see of your progress?

      1. Trixie Ray

        Hi Dave, currently I am finishing my research, spreadsheets, and CAD drawings. I have prototyped the k1/k2, e5/e6 gear train and plan to prototype the inferior and superior planets next. I will post photos as I progress.

        1. Dave Goodchild

          Hi Trixie,
          Ahhh, CAD… wish I had the time to learn it properly… all my designs are in note book form, or designed directly onto the wood and then patterns taken from that once I’m happy that everything works. Old fashioned – that’s me…
          I love the way that the machine is still open to a fair bit of interpretation when it comes to the details. I’m hoping that there will start to be several different machines coming out of the wood work in the next year or two that all achieve the same basic functional goal of working like the original would have, but with very different design approaches – should be fun to see what different people come up with.
          Have you thought beyond the gearing as to what your final design will look like?

          1. Trixie Ray

            I great appreciate your craftsmanship to design and cut the pieces via the traditional method. Do you use a depthing tool to layout the arbors? I have a keen interest in horology, clock making, and orreries. I enrolled in the BHI’s distance learning program (but did not take the tests). I have been fascinated by Antikytheria Mechanism for many years and started my research on the subject with the Price and Wright papers in early 2002. Naturally the Freeths articles in 2006, 2008, and 2014 have been extremely enlightening and exciting. Building a replica has been on my project list for a very long time and it finally made it to the top spot. My design will follow the ISAW paper for the gear train, although I may consider using the 145:46 gear ratio for Mercury. For me the beauty of the AM is the mechanical design as much as the results calculates. My design will try to make as much of the mechanic as visible as possible. It will be a skeletal design with engravings on the gear spokes and designs cut into the Date Plate, Superior Planet Plate, and bridges. I am still thinking about how I want the front and back plate to look. I plan to cut most of the device from aluminum or stainless steel, with stainless arbors and roller bearings.

  8. Dave Goodchild

    Hi Trixie,
    I really should check back here more often and keep up – sorry.
    It looks like your final machine should be quite something when done. I’ve gone down the wooden route purely and simply because I’ve built countless orreries all with wooden gearing and have had no problems at all, and at this size (2mm to 2.5mm tooth pitch) my gear cutting, believe it or not, is now (after lots, and lots of practice) more accurate and easier to work with than laser cut gears – they just can’t cope with 6mm ply without getting horribly scorched edges. They work, but aren’t great.
    I too am going for the skeletal look. I can’t see the point of cutting all these gears if people can’t actually see them! I have a design in mind so will just have to see how it pans out – can’t wait to be honest to see it start taking shape.
    I’ve got almost all of the centres for the main gear train cut now, and in isolation all the pairs run as near to perfectly as I can get. I’ll just have to see if this continues as I start cutting the rest of the frames and start actually assembling it with all the gears in place, not just random pairs. Oh, and to answer your question, yes I do use a depthing tool to set up the centres. I built a home made one which works brilliantly. And yes – most of my info cane from Colin Freeth’s Cosmos paper – it’s been incredibly helpful and was really the catalyst I needed to get started with the project.
    My email address is if you want to drop me a line and then I can send you a photo or three if you want. Where are you based by the way? I’m in the uk abs currently enjoying rain!

  9. David Wilkie

    Although the manufacturing techniques and mathematical solutions incorporated into the actual machine are the core interest, I am as interested in the actual results of what the mechanism meant to the designers and users.
    A computer program with the equivalent readout, for this current era would be good. My personal pleasure in this gadget is what it says about the mindset of its creators, and the direct association they had with the Earth and Cosmos, I reckon they are the precursors to Quantum Mechanical thinkers because everything they said and did was timed to, and for the reason, that the stars and planets acted out above them. I don’t think it is any more a religious idea then than it would be now, it is the clockwork that matters, and how much money they could make out of the information it provided the users. It is a business machine.

    You folks are doing something interesting, and have probably convinced me a 3D plastic model will not be a working machine.

  10. Dave Goodchild

    Hi folks,
    Just thought that I’d share a couple of photos of how my machine is coming along. It’s almost at the stage where it runs – I should be at that happy stage within the next month or so. I’m starting to get excited now that the whole thing might actually work as planned! The photos just show it mocked up to trial fit everything to check spacings etc. Next job is to rip it all apart again and actually link it all together. I’m keeping everything crossed…
    So how are the rest of you getting on – any news?



    1. Scott

      That’s fantastic progress! I’m afraid my update is still in the middle of the drafting phase – a move and new job have slowed things down a bit, though I did manage to grab the website I wanted for when I eventually do release the final version and am sitting on that.

      If you don’t mind me asking, what are you using to anchor the axles with only one end mounted, while still letting them spin? (Like here: I haven’t figured out how best to do that yet.

      1. Dave Goodchild

        Hi Scott,

        Ok – in effect I cheat.

        Right from the beginning when I was building my first wooden orreries I realised that for most of their lives they were going to sit looking pretty and gathering dust. After a flurry of enthusiasm from the new owner when the machine first arrives the handle will only be wound for 10 minutes a week, and probably not even that. In other words the one thing that really wasn’t an issue was wear and tear – I reckon that my machines will easily out last me (I’m 46). My earliest orrery is now coming up for its 10th birthday and working just as well as the day I finished it. In a nutshell I use techniques that work for me in the low stress orrery type situation, but they wouldn’t be so great if the machine was motorised and running all day every day.

        So, on my machines the gears are interference fit onto the brass tubes with a little super glue, after which they are going nowhere. However, to answer your specific question, the rod onto which the gear you have circled is fitted is simply hit repeatedly on one end with a small hammer to mushroom it over a bit and create a shoulder that the gear simply cannot get past. The gear itself has a brass sleave fitted at it’s centre so that the tolerances are quite close, any distortion of the rod is then enough to keep the gear firmly where it lives, so my hammered shoulder works a treat. The rod obviously has to be push fitted into the frame and through the gear once the shoulder has been created.

        The other way to do it is to fit the gear onto the rod and have 3 or 4mm of the rod protruding above the gear and then simply solder a 2mm sleave onto this protrusion to stop the gear from escaping again.

        The second option is probably better engineering but takes up more space (not really a problem in the situation that we’re looking at though).

        I’m not sure that this is going to help much with 3D printing, but it works great with the materials I build from. I guess though that it wouldn’t take much to simply increase the diameter of the printed rod beyond the gear to simulate my soldered sleave.

        I hope that this helps at least a little, but our materials are very different so detail bits like this will tend to differ.

        I can’t remember if I shared this, but below is a YouTube link to a film of the whole planet mechanism working.

        Best of luck, and I hope that things have calmed down enough for you to have a play with this again.


        1. Scott

          No judgement at all for fudging it a bit! I watched your video of it working, and it looks incredible! Runs smoothly and the motion is very apparent. Well done! I’m a long way off from that point, I feel, though things have calmed down and I’m gearing back up to give it another push.

          Thanks for the info on how you mount the axles. I was hoping to do wooden axles that turned in place, but I agree that sleeves around a central rod make a lot more sense. It will just be more expensive to make. I’m planning to have this next version be laser cut from sheets of wood rather than 3D printed, so the processes from your model are actually very applicable! Thanks for the help and comments, it’s nice to have someone to bounce these ideas off of.

          1. Dave Goodchild

            You’re more than welcome. If you need any more complex info use my email address – – easier than using the blog me thinks!
            Are you in the uk by the way? Just curious, that’s all.

  11. Dave Goodchild

    Hi Scott,
    Just thought you’d like to hear that I’ve sort of got it running! I had a good play with it this evening and was amazed at how close I was to actually getting it going, so a couple of hours of fiddling and this is the result:

    A bit clunky due to a couple of binding gears – easily fixed – and a few gears are still missing from the bottom of the machine, but the complicated bit is done.
    It’s only taken two years… but it does feel good.

  12. Gustavo Keimel

    About the Gearing Module:
    The general actual module of gearing is near 0,5 mm which is (surprisingly) a big coincidence, since there is not possible that the ancient Greeks did know the Module System (metric standards based, created circa 1900 AD).
    But the entering gears (a1 and b1 as the standard denomination marks) had 0,6 Module (which is 0,6 mm x pi diameter pitch). Many theories there are to explain it. My own, is that probably the builder had some gears taken from other mechanism, or an older version of THE mechanism, Actually, the main gear had 223 or 224 teeth (the remains are incomplete) but the only function of it is to provide the main round movement, not to generate relations between other outputs. But since there are a relation of 223 lunar months involved in the mechanism values, it is possible that, in a early version of it, the builder made a 223 teeth gear to power it (By the way, believe me that a gear of 223 or 224 teeth is a big challenge to a builder that don’t had a nonius or CNC gearing machine to build it).
    In my own version of the Antikythera mechanism ( prior I assume that the lunar anomaly gearing had two different modules: one to the e5-k1 gearing (0,5 Module) and other to the k2-e6 gearing (near to ,55 Module). But, in this version of mechanism, I eccentred the k1-k2 axles in radial direction. Later, I realize that the eccentric displacement could be made tangentially with the same effect and so the module of gearing could be the same to all four gears.
    The gearing shape that I choose to my version is the 20º evolve (modern gears) one, opposite the original mechanism shape, which is equilateral triangle. My own reason was that that I think that, if the Antikythera Mechanism builder knew the modern gearing theory, surely he did choose it to his work. Recently I saw a video (, in Spanish) that supports this thought. In it tells that the original mechanism includes mechanical solutions to avoid errors of 2 degrees, but the triangle teeth gave errors of almost 15 degrees. So, it’s impossible that the original mechanism was used to calculate real cosmic events. It is more probably that it was used to provide a mechanical model or to a didactic use. The model relations are perfect, the gearing is not.

  13. Andrew Friesen

    I’ve been looking at this model in SolidWorks and MasterCAM and am hoping to build it. I will probably be cutting all of the gears with a wire EDM and CNC machining what I can’t do in the EDM.

  14. Bert

    Hi Scott,

    You’ve did a great job! Wondering why you used a scale factor for calculating axle distances. The modern gears (evolvente) have already a small tolerance. Calculating axle distances = Ad=0,5M(Z1+Z2)

    Greetings Bert.

  15. Ben Burch

    Greetings! Has anybody made any additional development of this model? I want to make one and would like to see what others have done first. Thanks so much!

    1. Scott

      I’ve done quite a bit of work on a new version that will actually be manufacturable rather than just being a computer model. Still have work to go, but I’ve gotten as far as making some test parts on a laser cutter. Trying to get a version 1.0 done by the end of summer! If you plop your email down at that will get you on the list of people I’ll notify when I finish up.

      1. Glenn P

        just found this thread. great work. registered on your site.

  16. Johan Täljemark

    Great Scott!
    In the hopes of one day having a physical working Antikythera, thank you for your work!
    All the best!

  17. GarciaBurton

    Hello, i also live in and work at night in Denver. I just happened across your page by chance tonight. do you have laser cutting files available for this?

  18. Robert

    Thanks for sharing the file, after trying to print them for a few day’s I found out that they not for 3D printing.
    I always learn the hard way (Age related).
    Can i use a router to carve the gears.
    Do I need to convert them.
    Can’t wait to build the model.
    I am going to build an Orrery next.

    1. Scott

      Yeah, this model isn’t meant to be built, it was only meant for the computer. A manufacturable model is still on my back burner, but it will be some time before I can devote serious time to it.

  19. Seth

    I don’t seem able to download your cad file, is it still available? Thanks!

    1. Scott

      Thanks for the ping! Fixed the link.

  20. Pano K.

    Hi Scott. I plan to provide Greek language lessons in an elearning format to students which will include small cultural lessons as well. I will be paid for the language lessons but not for any media that will be shown (images, videos, etc.). Will including the video of the animated Antikythera device shown at as part of a mini-lesson on Greek inventions be okay as long as I credit you?

    1. Scott

      Sure thing, go for it!

      1. Pano K.

        Thank you!

  21. Peng

    This thing is really good! But it was too late when I saw this webpage, and the link for downloading CAD files on Google Drive seemed to be invalid. Can you re-upload a link or tell me another way to download it?

    1. Scott

      Thanks for the ping, link should be fixed now!

Leave a Reply