Rolling-Screw Extruder Goes Brushless

In the name of saving weight and pushing plastic, it’s nice to see continuous tweaks on 3D printer extruders from folks in their spare time. And to go where no extruder has gone before, [wayne dalton] has managed to combine the rolling screw thread extruder concept directly onto a brushless pancake motor. The result is a filament pushing mechanism weighing in at just under 90 grams. What’s more, this modification arrives a few weeks weeks after we first saw an open source version of the rolling screw thread extruder land on Thingiverse back in September.

Getting a direct-drive BLDC extruder off the ground comes with a number of geometry challenges. First, filament needs to be fed through a motor with a hollow shaft. Since this motor will be operated under closed loop control via an ODrive, the motor also requires an encoder with a hole down the middle too. [wayne dalton] patiently walks us through all the changes made over seven iterations to produce a setup that will eventually mount onto a machine and start printing. But for that last step we’ll need to check back a bit later. Finally, this progress comes with the promise of a possible open source release of the design files, so we’re pretty eager to stay tuned for future videos.

In the last few years, we’ve seen state-of-the-art extruders move away from NEMA 17 motors in favor of more compact solutions like planetary gears and even lasers to push printers faster. We’re excited to see that folks keep trying to set the bar higher.

Thanks to [ImpC] for the tip!

33 thoughts on “Rolling-Screw Extruder Goes Brushless

  1. He mentions on the video that the rollers had to be custom made. Anyone got ideas where you could get those rollers easily?

    I was googling yesterday and the closest I got was with “knurled edge dye roller”, way too big and knurled wrong(?).

  2. Incredible how much time and effort the 3D printing crowd puts into designing and creating solutions to problems that could fundamentally be solved quite easily by building stiff and strong printer structures and using proper linear motion hardware.

    So much time and effort has been wasted in the pursuit of solutions to problems which could just be solved easily and permanently mechanically just by compromising on paying a little extra – automatic bed leveling is a prime example the obsession with which I have never understood. I installed my print bed to be mechanically permanently orthogonal when assembling the printer and it has never needed adjustment. The pursuit of endlessly light printer hotends is another problem falling into that category that seems to cause no end of grief with the solutions implemented, as much as I admire the quality of work that has gone into this particular solution.

    It’s a problem I see with lots of open source and software-heavy projects – not enough people with actual mechanical design skills, and a total aversion to just paying for adequate hardware for the task at hand. Too many people applying the software hacker ethos of ‘we can just code our own version instead of paying for it’ without realizing that doesn’t really scale conceptually into the mechanical world. So you wind up with horribly convoluted solutions to problems that are usually solvable with off the shelf parts that end up making the projects much less useful – but at least it’s all open source, right?

      1. Thermal expansion is trivial at the structural temperatures for a printer compared to the overall tolerances a printer is expected to achieve. Any concerns over bed warping due to thermal expansion can be nulled using appropriate design to minimally constrain the print platform to the motion base allowing it to float and grow but remain planar. But it’s still such a trivial effect that it shouldn’t matter.

        My bed stays level whether it’s hot or cold within appropriate tolerances. 9/10 of the complaints I’ve seen about bed leveling simply come down to people expecting software to account for their failure to properly calibrate their equipment.

        1. The best solution I’ve seen for printer bed thermal expansion was to have 3 mount points which consisted of a largish (1/4″-1/2″ or so) steel ball, which rested on two parallel steel rods that sat in line with the center of the print bed. A strong magnet underneath the rails gave it some “stickiness” so the mount wouldn’t easily fall off the rails, but did not provide significant resistance to the minor movement thermal expansion would cause

          In this way the plane is always maintained, due to only having three mount points (3 points always produce a perfect plane, as long as the bed started flat it would always remain flat with 3 mount points and no thermal expansion), and the ball on rails allowed the bed to expand in all directions evenly, which prevents the expansion from causing any warping on any axis. So then the only movement of the bed not accounted for would be the microscopic vertical expansion, which is far below any margin of error for calibration.

          It was part of one of those high-speed CoreXY kits, I don’t remember which one.

      1. And that’s a valid point, but at what point do you have a 3D printer for your hobbies, or do you have a 3D printer hobby?

        Regardless, my point is I’m pointing out the persistent blind spot that the 3D printing community has in failing to just build structurally adequate printers to begin with, before opting to come up with Goldbergian software or mechanical solutions to problems that can be solved with more meat or just spending a few bucks.

        To an extent the 3D printing community has never grown out of the RepRap project’s bootstrapping mindset, which is somewhat ridiculous when the range of mechanical components available to consumers/hobbyists now is orders of magnitude what it was 10 years ago and at a fraction of the price. The majority of people live with 3D printers that are structurally and mechanically sub-standard because they carry over the design compromises required to make a 3D printer 10 years ago vestigially (and because the manufacturers exploit those choices to make the printers extraordinarily cheap to make and to ship overseas).

        Better built and designed printers could be had at hobbyist price points, but the market is dominated by RepRap’s offspring. If we could give that up we could have substantially better printers in just one generation.

        1. You are probably underestimating the effort needed in mass producing devices with such a high accuracy, and on the other hand over-estimating the efforts spent in overcoming mechanical deficiencies. There is a reason even high-end CNC machines often use software compensation: it is cheap, accurate, works well and can be updated whenever needed.

          But if you really believe there is a gap in the market, go fill it!

          1. I’m a mech eng working in design for manufacture every day, I’m totally across what it takes to manufacture a structure like that at scale, and why most printers are built the way they are.

            High accuracy isn’t required, just proper design is. The geometrical tolerances necessary for an FDM printer are generous to say the least and could easily be achieved with heavy gauge pressings, as some printers have done. Other than approximate axis orthogonality, the only truly important tolerance is setting the bed coplanar with the XY axis, which can be done using a kinematic mount with positional floating for the end user to adjust and lock off.

            High end CNC machines use software compensation to chase positioning accuracy 1-2 orders of magnitude higher than an FDM could ever achieve or would ever need to achieve. At a base mechanical level even your average manual milling machine or lathe has positional accuracy that would match or exceed the needs of an FDM printer.

            Meanwhile there are reams and reams of projects and posts online where people come up with ridiculous solutions to really simple mechanical problems for printers. Which is my point – the kinds of people that generally rely on their printer as the sum total of their fabrication skills are typically not the same people that have the mechanical design skills to recognize how mechanics could be used to solve problems they’re attempting to solve in software or elsewhere.

    1. Why do you feel the need to put down other people’s experimentation and ingenuity?

      Part of what makes life interesting and humanity so adaptable is the urge to try something different… A lot of times it doesn’t work, but sometimes it does!

      I would much rather see someone tinkering away at something than many of life’s other frivolous pursuits.

      1. I specifically said I applaud the ingenuity of this work. The comment is directed more at the 3D printing community in general than this specific body of work.

        My point is that it’s work that is being done to work around a problem that could be solved far more easily by improving other parts of the printer easily and inexpensively in the overwhelming majority of applications. Which would then free the people that waste inordinate amounts of time trying to solve simple mechanical problems with software hacks or unusual designs to work on other things (or god forbid, print things that aren’t more bits for 3D printers).

        As clever as this design is, it’s still optimizing something that can far more easily be optimized by just designing a stronger printer and using more beefy motion control hardware, then going to a far simpler/more robust/cheaper direct drive extruder. Or changing the physical arrangement of the printer to minimize the motion of the extruder.

        1. Except this enables a far lighter design overall which can be extremely useful in something like hardware that will go into space for example. At that point every gram matters.

          1. That’s an extreme fringe case though. In the overwhelming majority of cases the effort would be exponentially better spent on stiffening the printer frame, reducing printhead weight wherever it doesn’t deduct from reliability or durability, siting the printer on a good base to provide vibration damping, and coming up with a repeatable, solid mount for the print bed. All of which would dramatically improve quality of life for the largest number of users.

        2. “As clever as this design is, it’s still optimizing something that can far more easily be optimized by just designing a stronger printer and using more beefy motion control hardware, then going to a far simpler/more robust/cheaper direct drive extruder. Or changing the physical arrangement of the printer to minimize the motion of the extruder.”

          What are the specs of the printer exactly that you would upgrade ? I’m quite interested because you don’t appear to know them or you wouldn’t have made this comment .

    2. Automatic bed (mesh) leveling is a great idea because it vastly improves that critical first layers placement for great adhesion without needing a really over engineered bed that can stay perfectly flat and consistent in position at all the varied temps for every type of filament you try to print… Its a very sensible effective solution to not needing aerospace materials and/or construction precision to get a good first layer for any possible temperature profile that filament wants. It is very possible to engineer and prevent the issue, especially if you only print at x bed temp all the time every time, but its still pretty challenging, when you can save heaps of construction cost by just making a decently stiff mount that copes with the thermal expansion and letting any minor twist and warp be taken out in the software, basically for free – which when you then want to print at a new temp the mesh level will just fix any shifting in the machine for you, or you end up having to really get creative in the bed engineering to keep it perfectly flat and positioned so you can just print in other filaments…

      Light print heads similarly are great, as even if you have a real tough CNC mill frame playing FDM printer the really rapid movements of a properly fast 3d print will make something in it ring and show some artifacts on the print so much more easily when its heavier. Your cutting tool spindle can damn nearly weigh whatever you want – the movement speeds are much more restricted by chip clearance of your rather massive cutting too (compared to even largest 3d print nozzles cutting bits are generally orders of magnitude bigger), part cooling etc so the mass of the tool head isn’t nearly as easily the limiting factor in getting good results faster, where FDM is all about being able to move the head faster and push the plastic through it fast enough to keep up…

      The only way in which lighter FDM print heads don’t make sense is if you are happy for every print to take 20x longer than it could (which seems like it can also have a detrimental effect of part strength and warping)…

      That both these solutions have let the rather cheap end of the 3d printer market get good results with questionable engineering really is testament to how good they are as solutions, and they still have use at the higher end too.

    3. Hint, you are reading and replying to an article on HACK-a-day…

      Anyway I think you are overlooking the design requirements which may or may not align with your design ideals. Maybe lightness was the main criteria, maybe cost, maybe purely to explore an alternative approach. Generalising to the 3d printer community overall, one of the original idealisms was to have machines that could print their own parts, which lead to geared extruders with plastic gears. Maybe a planetary gearbox, cyclyodial gearbox, or belt driven would be a better solution.

    4. i build heavy! 20 size lineair rail 80 size frame 4x 400watt gantry (corexy) and everything else strong and sturdy!!

      having counterweghts on the x-y for resonance canceling weiging 3kg.. costs more to get it to move but printing at 2m/s +!!! and no ringing etc.. no jerk no.software compensation and 5000mm/sec acceleration..
      costs a lot of power but in my opinion worrh it..
      having a nozzle with a adjustable diameter 0.1 to 6mm makes it printing even faster.. beqause much more than 2m/s gives extrusion problems at a 0.4mm nozzle..
      i hear nobody chasing variable diameters???

  3. Mesh leveling is a crutch for not properly dialing in the printer in the first place and not having a frame stiff enough to stay that way, and is mainly popular because people generally seem to exceedingly struggle with doing a proper bed leveling job to the point that it’s considered a chore – usually because they don’t have the right tools and because the structural design of the printers works against them.

    Somehow I never have problems with bed warping due to thermal expansion and my bed stays level within an acceptable tolerance cold or hot – probably because it’s mounted properly and it was level when I installed it, and my printer frame is sufficiently stiff not to come out of square, and I have realistic expectations for the dimensional tolerances of my prints. A good bed heater, a good thermal connection between the bed heater and the bed, and supporting the bed such that it is minimally constrained to remain planar pretty much eliminates warping as an issue. The engineering difficulty of this is massively overplayed, but it’s a hard problem to solve when the average printer is mechanically a wet noodle.

    Same with print heads – design better structures and the issue becomes less important. Better structures usually means more mass which means more shipping costs which is exactly why consumer printer manufacturers don’t do it. Adding 1kg of mass to stiffen a small printer frame is a better optimization than decreasing the print head mass, because it also improves all the other mechanical properties of the printer. If that avoids complex extruder designs, or unreliable bowden-type extruders, fantastic, but it’s where time should be spent first and almost never is.

    The problem is the 3D printing community still lives in the shadow of the RepRap project and maintains all kinds of strange vestigial organs that it probably should shed to improve printers that come from that. Good printers shouldn’t be built as if their origin story is derived from the ideals of RepRap, but from basic mechanical design as any other machine would be. To an extent I think carrying on with RepRap derived designs holds the whole market back – the big problem there is that the proprietary alternatives that could be better are usually awful or completely unsupportable after the first year or two.

    1. I’ve seen many problems, especially in RepRap style printers, that were down to the mechanism being too flexible. Very often, the table it is set on is not flat. Or there is detritus, or even a wire, that is under a corner. In one case, someone set the printer down so one corner was off the table and over time, all the 3D printed corner pieces warped so it was not flat even on a flat table.

      I built several torsion boxes for the printers in the makerspace and the printers, once fastened down properly, no longer had so many issues.

      1. I would chalk half the problems I see reported with 3D printers as being a direct result of poor structural stiffness either directly or indirectly. Setting the printer on a stable, immobile base with sufficient weight to provide good damping is the number one thing I see people failing to do that will amplify that problem.

        Problems derived from using Bowden extruders to reduce printhead weight are right up there as well, which would be far less necessary with better built, stiffer printer frames.

    2. Making the gantry heavier to make it stiffer will require bigger motors, and the bigger motors have more rotor mass so it takes even bigger motors again to compensate for both the increased gantry mass, the increased rotor mass and the heavier extruder you want to put on it and maintain a decent acceleration. So putting big servomotors on a 3D printer to maintain the same printer speed is going to cost a lot more than the shipping weight cost you are talking about. And hobbyists are not going to buy printers costing many thousands of dollars. All in all you don’t appear to know what you are talking about.

  4. man Marcus Aurelius is a grumpy person…. If you’re so much better at designing and building 3d printers, go make some? the entire community is doing what its doing because that what it thinks is best for the scenario at hand… if you don’t agree with the masses, maybe assume that you’re the odd one out, and let everyone live their life?

    This extruder is awesome, and the machine its on is as well and i hope it keeps going and does the thing he wants it to do. And if not, i bet he learned a lot more about a lot of things than old mate Marcus could have learned had he decided to enter this post with an open mind and not an elitist one

    1. Thanks :) Im hoping to release it open source this week … just had a couple of events that are sucking up my time like crazy right now . Maybe github is the best way to publish this .

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.