Printing In Three Dimensions, For Real This Time


3D printers don’t continuously print in three dimensions – they print one layer, then another, then another. This is true for every single 3D printing technology, but now Topolabs has a very interesting technique that changes that. They’re printing in three dimensions by moving in the Z axis while also printing in the X and Y axes.

The basic idea behind Topolabs’ software is to print a support block, then print an object right on top of the support. The support block can be curved and convex, and the finished product follows the contours of the solid support block. Unlike ‘printing with supports’, the printer extrudes along the X, Y, and Z axes, which should make the finished product much, much stronger.

There are a few drawbacks to the technique – a release agent must be applied to the top of the support block. In the video below, Topolabs is using Kapton, but hair spray or glue sticks will also work. There’s also a limit to how steep an incline a printer can print, determined by the size of the extruder nozzle. Lastly, this technique would be much better suited for a delta-style bot, but the team is getting very good results with a normal Cartesian bot.

You can see a few videos of the Topolabs printing technique below.

37 thoughts on “Printing In Three Dimensions, For Real This Time

  1. The concept is nice, but I can already see a lot of failed prints with trying to print too shallow curves, the width of the hot-end could melt some plastic of the printed part. When this however is taken into account in new slicing software in which you can define some basic dimensions of your hot-end, the print quality may indeed -rather vastly- improve, as you can see in the video. So, as always, implementation and attention for details is everything!

  2. On top of being stronger you could save some material if you need more than one copy and can print several of the on the same support.

    Even if a 5-axis machine could do much more curved designs this std 3-axis variant is interesting as it could be a firmware upgrade to existing mechanical designs, just as an extra mode of operation for suitable objects, without any extra (hardware) cost or other negative effects on normal operation.

  3. Three hours ago I posted on a 3d forum asking why 3d printers did not do this and for the last three hours I have been hex writing a print file manually for my printer to do it.

    1. 3D printers, even commercial types, deposit or form on a layer-by-layer process.

      This could be useful, basically the printer would build up the infill first, then lay down the solid layers on top of the infill, and the slicer would have to calculate the angles it can pull off, based on your hot end.

      1. Well like more advanced CNC machines, you could vary the angle of the hot end to get around stuff. That might add to much complexity to current designs, but future machines could have more axis to play with.

        1. Changing the angle of the print head could could be a simple matter of using a servo and mounting the entire hotend to it. As long as you maintain some slack in the filament coming off the reel, you should be able to achieve 45° before the carriage crashes into the bed on the first layers. This is taking into consideration we are printing using a wades, on an i3 style x axis frame. This won’t be as easy to produce with horizontal smooth bars found on other reprap designs.

  4. So, I’m imagining that the limitations are cylindrical parts and features. Does it improve the strength of screw bosses? From the looks of it, this only improves the strength of smooth, shallowly curved surfaces… Which typically aren’t bearing loads or fastening pieces together.

  5. It says “patent pending” on their website. Putting aside the fact that software patents are supposed to be illegal, what is novel enough about this software to warrant a patent? I’ve seen this sort of thing done before with 6 axis machines, so what’s the new novelty that they are trying to patent?

    1. Ceramics industry has used hump molds for centuries. Are they trying to patient the idea of printing on a hump mold? Or claim that certain patterns of printing can only be done using their software?

          1. You know you can look up a patent application, read the claims, and then find prior art that would invalidate the application, right?

            I really don’t know why more people don’t do this. Crowd sourced prior art search, anyone?

          2. Sadly, reporting prior art during the approval process also is not a magic bullet. The patent is likely to get a non-final rejection for it (if the prior-art is significant), but this allows the applicant to change the claims section to avoid it. Often you can simply say ‘but does not cover embodiment xxx’. The patent can then be resubmit (if there is anything innovative left, almost always the case with modern day standards), and this prior-art can never be used to fight the patent anymore…

    2. I don’t think patents are about merit anymore. It is a calculated risk and a lottery ticket rolled into one. If you don’t file your stuff, someone else might do it. You lose huge. Even if you don’t have anything patent worthy there is a decent chance that the patent will be granted anyway. You win massively.

      Really it is the way of new things right now. Either game the system or risk others screwing you out of your own work.

        1. Just a thought, should HaD staff read this (maybe I should report it), but you know the report option’s been a little niggle over the years. Is there any way it could change to an “unreport” button when you’ve clicked it, in case you’ve made a mistake?

          You could possibly use a cookie to store the “reported” status. Since the end result goes to a human editor, and it’s only the original reporter who could un-report it, there’s no potential for abuse there. It’d just cut down on reports you have to check on, and the need for apologetic emails like Randy’s.

    3. Software patents aren’t illegal. It’s not literally the code, but a new method of doing something in software can be patented, and has been. To the massive consternation of programmers and anyone who knows what they’re talking about.

      Genetics is possibly an even worse field of patents. They just need to find a gene now, to patent it. Used to be the time you had to invent things for yourself to get a patent, not just notice them occurring in nature.

  6. “Lastly, this technique would be much better suited for a delta-style bot”

    What makes you say that? Delta bots are good for quick movement but limited and has less rigidity than a cartesian setup. Maybe you are thinking Stewart Platform where they can move the head?

    1. Because delta-style printers have zero drawbacks to fast, three-axis moves. Most of the cartesian printers use cheap threaded rod for Z, resulting in fairly high resolution but very slow movement. For example…my Prusa i3 can easily manage 175mm/sec XY moves, but struggles to do a 5mm/sec Z move. The Z axis, in many designs, is only running through a standard nut, not exactly great wear characteristics but it’s fine when gravity helps reduce backlash. However, these problems are all known, and also affect the popular movement to auto-level printbeds by detecting the angle of the bed and compensating in software (resulting in constant Z usage during XY moves). So a lot of work is being done to use better leadscrews for Z, or replace it with a belt mechanism.

        1. A standard delta style printer always keeps the effector platform level. That’s one of the advantages of a delta printer. You would have to have additional axis mounted on the effector platform in order to provide rotation.

  7. And, the last that I knew making the API opensource is not opensource, it’s just making the methods of the API public. I don’t count that as opensource and would appreciate them not using that term.

  8. finally, additive CAM has gotten to the level of subtractive CAM. started realizing how silly something like slicer or skeinforge is compared to the toolpaths that your CNC milling CAM suites have to generate when I started looking into it

  9. color me interested in the software…the machines that do car prototyping do exactly this, but with epoxy paste extrusion. the idea is to build up a uniform layer of material on top of the plug before milling it to tolerance. its millions of dollars, of course.
    If the Gcode generation is solid, this is extremely useful for every day milling cnc. right now, cheap software that can generate decent 3d toolpaths is pretty hard to come by. sketchup and cambam have ok 3d toolpath generators, but forget about compound paths, its waterline, or horizontal. if someone can work out a longer hot end, this can do some amazing stuff just in 3 axis.

    1. There was a really great video I saw years ago that showed the entire automotive prototyping process. It was all done on a huge 5-axis machine. It started with rough foam milling an undersized core. Then a spreader nozzle applied a thick coat of bondo-like epoxy. Once that cured, it was milled down in multiple stages to be an exact full-scale replica of the car.

      Throughout the process, they were using contour milling and printing. This isn’t new at all. Ok, well maybe it’s new in the hobbyist space, but that doesn’t make it patent-able.

      Wish I could find that video. I’ve looked for it a bunch of times. I’ve even found similar ones, but it was the best at showing a completely automated process.

      This one is similar and shows the milling, but the “contour paste” shell is applied by hand rather than by the machine….

  10. Looks like topo labs is kaput. Their URL spits back this error. Application error
    An error occurred in the application and your page could not be served. If you are the application owner, check your logs for details.

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.