Crazy flying extruder idea #92

  • Ok Mr. 42, lets try one more absurd idea.

    On my delta (duet wifi) would it be possible to create a 4th tower that simply follows true Z as determined by the other towers? Ie Z5 = 5mm off the build plate equals tower at 5mm? Essentially I'd like create a 4th carriage to ride on the outside of my extrusion and hold my extruder, then the bowden need only travel x/y diameter and it'll be on it's own drive/belt/motor so it won't affect the mass of the carriages carrying the effector. Basically a 4th tower that just follows the height of the effector? It'd probably need its own endstop too?

  • Here's a previous thread about something similar:

  • Interesting, good to know I'm not the only interested in this idea, look forward to hearing if it's been implemented, or how I could get something comparable happening!

  • administrators

    If you or someone else wants to implement the mechanics, I'll do the firmware changes. It will have to wait until next month because the next 2 weeks will be taken up by the TCT show and preparation for it. You will need to provide a homing switch for the additional tower.

  • And it's not a crazy idea either IMO (although I know nothing about Deltas). I did similar on my CoreXY for which I needed 3 extruders (now 5). Started with a sort of suspended, counterbalanced arrangement, then turned it into a second XY gantry but passively dragged around by the main hot end. Then changed that again to a second, driven XY which carries the extruders and sits above the main hot end XY. So frame size is 600mm x 600mm but Bowden tube lengths are only around 200mm (and could be shorter if I didn't also want to be able to swap hot ends easily). Best of luck with it.

  • Amazing DC42! I look foward to implementing this on my machine. No rush, if you can give me firmware that supports a 4th endstop and motor with its own steps/mm and can set it to track actual Z I can come up with a hundred ways of making it work mechanically.

    Ideally I'd avoid a second belt setup all together. I'm gonna try to design a printable part that is essentially a rigid belt flattened out that fits in the back cutout of typical extrusion, then a wheeled carriage with matching gear wheels to ride up and down on it with both steppers (the one moving the carriage and the extruder). It'll be tricky but it'd be easier to universally implement an add-on thats totally printable and doesn't entail running additional belts because people use different corner pieces etc. I'm pretty sure I can design a self driving carriage that'll fit the backside of any 2040 based printer. Because its just tracking Z it's not moving all that much or that fast, the most stressful move it'd make is homing. If a self driving carriage doesn't work out I can come up with a mount that will attach a 4th motor, belt, and idler to all extrusion based printers like so:

    It doesn't actually need to ride on a rail because precision isn't that important. I'll design a mount that fits on the top and bottom horizontals and holds a motor and gear (and endstop) on top and an idler on the bottom to mount in between towers. I'll design a "carriage" with a belt tensioner and extruder mount. It'll essentially dangle in the air and be reeled up and down by the motor at the top. This is easier to implement than a carriage that rides on rails and I'll start by testing this out when firmware supports it.

    But yeah gimme firmware for a 4th endstop, and motor, with individually defined steps/mm and its own endstop, with the capability to track Z and I will give you back an easy mechanical way of having a much shorter bowden tube without a heavy stepper flopping around on elastic! Please get in touch if you need any help with the firmware side of it (ie understanding the project or the needs, I can't actually help code it lol)

    Seriously thank you and your fine company for being so responsive to the community. I don't take for granted how amazing it is that I can come up with an idea that is completely outside of the realm of ordinary delta stuff and be told by the programmer that he'll change the whole firmware to make it possible. Amazing stuff.

  • administrators

    What I'll actually implement is support for a 4th tower with the same kinematics as the other three but separately configured diagonal rod length and XY position. As I said in another thread, if you specify a very large diagonal rod length, it would effectively track Z movement. But you will also have the option to configure a shorter rod length, so that when the effector is farther from the extruder drive, the extruder drive is lowered to compensate.

  • Is it not possible to have a 4th tower that operates on Cartesian principles? I don't know how the firmware treats delta mode and if it'd be impossible to except a 4th motor from delta treatment so it just treats Z=5 as be 5mm from the buildplate. If not, would it be possible to program the 4th motor to move at the average of the 3 main towers? Then it'd stay hanging around the same height as the effector still. You obviously understand delta kinematics and programming better than I do, so I'll trust your implementation, but I'm not sure how I'd calculate theoretical rod length for a carriage without rods to trick the system into keeping the extruder carriage at the same Z height. You're the best in the business though, so however you can pull it off I'd be so grateful and I'd find the simplest and most adoptable mechanical solution to get the extruder carriage where we need it to be.

  • Requesting an update on this idea 🙂

  • I'd consider trying this also, it would seem, although forgive my lack of deep understanding of the firmware, that this axis would need a max endstop, and would home with the delta towers, if it supported a pivoting-extruder then it would follow XY movements passively, only moving in Z to track the effector (although conceivably it would need to have an offset, so that after homing it would not move downwards until say 20cm offset had opened between the effector and the 4th axis, presuming you'd want the bowden tube to follow a broad curve downwards from the extruder to the effector which would minimise bend in the bowden tube.

    Why not just essentially slave this axis to Z coordinates of the effector with a definable offset and its own movement limits to ensure it does not crash the end of the travel. I suppose on power up there would have to be a presumption that this axis was with a range of offset positions from the effector to limit the possibility of pulling the bowden tube out.

    I'd definitely try it, I have a movement channel spare (E1) and all it would need is a rail/carriage, a belt and a motor, and a printed/metal bracket to hang a titan extruder on that allows it to pivot, I have a similar arrangement on my microdelta but do not need z tracking as it only has 120mm of z travel, my kossel XL has 460.

    Having said all this, as cool as it would be, and unusual, it might be considered quite a lot of complexity compared to a flying extruder of a more conventional design, unless you intend to mount multiple extruders which would not be suitable (I've tried it) for a flying extruder bracket due to their mass.

  • administrators

    This feature is something that I mooted, but I haven't implemented it yet. The basic idea was to allow you to define the X and Y coordinates of the 4th tower (real or imaginary) and the "diagonal rod" length for the 4th carriage as you like. This would allow the greatest flexibility.

    I hadn't given a lot of thought to homing. One option would be to turn the 4th motor off when homing the other 3 carriages, so that the Bowden tube pushes it to the correct place. This might work for some configurations. Another option would be to give it its own homing switch.

  • Its interesting but in all fairness I cannot see it being better than RDD extruder or flying extruder, its more a novelty IMO.

  • I disagree Dj. A flying extruder is just not an elegant design, bouncing the whole assembly all over the place with rubber bands on a precision machine. I am getting absolute terrible results from my RDD extruder and will almost certainly be returning it. Moving the motor between towers and positioning it close to the effector while decoupling it entirely from the motion components seems like the best, most elegant design possible (excepting RDD, which I'm starting to think isn't there yet)

  • Sure but conversely, and the reason why I took that position, I've had great results with both flying extruder (I'm talking about at up to 90mm/s not at 180+!) and with RRD (both flavours). My fyling extruder didnt bounce around much apart from short fast infill on ceratin models, as it was pulled down towards the effector and always under tension) Is it possible that there is something else going on? What issues have you had?

    I would try this driven extruder platform idea though as it would be novel, interesting, unusual and potentially very successful but mainly just for hell of it. I wonder if we can do it already without any firmware changes by just slaving (using firmware) one of the tower motors and running the extruder platform on its 4th rail to run parallel to the tower of your choice. No endstop and with the offset just physically built into it. What is the min/max difference between say the Y tower carriage and the distance the bowden tube would have to move to ensure a short enough tube to get really good results? I suspect compared to straight bowden it would be considerably shorter.

    I might try this approach without any firmware changes, if it works - great if it needs firmware help, then I've got all the kit in place already it just needs the update to get it working.

  • administrators

    Here's another idea, inspired by deckingman's CoreXY machine.

    Phase 1: put 2 carriages on each belt, one a fixed distance above the other. The lower one connects to the effector as normal. The upper one connects to a second effector, with the same diameter and the same rod lengths as the first. This second effector carries the extruder drive(s), connected by a short Bowden tube or tubes to the hot end.

    Phase 2: That decouples the heavy extruder drives from the main effector and the rods that position it. However, the mass of the second effector may still cause ringing because of coupling via the belts. So the next phase is to put 2 pulleys on each stepper motor shaft, to drive 2 separate belts - one for each carriage. That eliminates coupling caused by elasticity of the belts.

    Phase 3: There may still be coupling due to the 'elasticity' of the stepper motors, i.e. the variation of torque with slip angle. So the third phase is to use 3 more stepper motors to drive the second set of belts independently from the first.

    This is almost certainly overkill for a single-extruder delta printer, because there are other solutions such as RDD extruders or lightweight highly-geared extruders (as used on the Ultibots DVS300). But for a delta using a 3- or 5-input Diamond hot end, it might be a good solution.

  • Yes - phase 3 - go for it! Then I won't be the only one with an insane printer.

  • Ok, so lets move forward on this. Lets say I want to set up an extruder carriage on a fourth rail to just mirror one of the x y z towers until firmware supports the new axis thing DC was talking about. How might I go about doing this. Anything I'm missing homing wise? Do I even need to home it, will the computer treat it like a new axis, or will it be treated like a second motor on one axis (like a dual z setup with one endstop), if I could just tell it to mirror X without setting up homing stuff that'd be ideal, as I'd just mount it a little lower than that carriage and it'd stay in sync.

  • Why not just use the dual Z motor output to run a second "Z" motor on your extruder platform axis. Get the driver current right, given you are driving two motors not one, position the extruder carriage where you want it and you're all good as long as your rail is long enough. Sure you can't necessarily have it higher than the effector but at the same height wouldn't be a big problem.

  • administrators


    Ok, so lets move forward on this. Lets say I want to set up an extruder carriage on a fourth rail to just mirror one of the x y z towers until firmware supports the new axis thing DC was talking about. How might I go about doing this. Anything I'm missing homing wise? Do I even need to home it, will the computer treat it like a new axis, or will it be treated like a second motor on one axis (like a dual z setup with one endstop), if I could just tell it to mirror X without setting up homing stuff that'd be ideal, as I'd just mount it a little lower than that carriage and it'd stay in sync.

    Yes you can configure it as a second motor for one of the existing towers, and then you don't need to home it.

  • Great. Is there any way to set it up to track (either via separate rod theory as you planned, or just tracking Z height of the effector) Z height yet, or is my best plan to just clone a tower for now.

  • administrators

    For now you will have to clone a tower.

  • Understood. Thanks as always, you guys are the best in the business!

  • Hi all, just thought I'd toss this message in here for two reasons:

    firstly, I too have at one time desired this functionality and so it's nice to see some movement in the topic.

    Secondly, because ask the necessary people are already involved in this thread that I would have direct messaged otherwise with the fact that currently I'm in the process of outfitting my second MetalMax delta frame with a newly built Diamond Hotend and am running into serious issues that I'm hoping one of the veterans in here can address:

    • I'm using three Nimble extruders and had ordered from Shapeways a properly printed effector mount designed by the Zesty guys for this purpose. Initially, I mounted all the hardware and attempted to deploy it, only to find that the positioning of the extruders severely impeded the range of motion needed by the diagonal arms in order to reach positions that I require (such as dead center – homed). I posted this in the Zesty forums and it turns out the consensus is that I must be the first to have tried this since I built everything correctly yet still have this debilitating problem with how to achieve motion given the crowded airspace immediately above the platform which causes problems.

    I continued my determination to get things working and was able to fashion a way to mount the entire effector from near the nozzle with the hopes that this mounting method, paired with trimming the Shapeways effector to remove the protruding sections which originally were there to host the magnetic ball/diagonal arm interface, would free up the effector to allow movement.

    The result is that the effector IS able now to move, but is still severely limited in its range of motion. Depending on the orientation (of which there are three) I connect the effector at, I am limited thereafter to a triangle (roughly) which reaches to only about a quarter of maybe a sixth of the 300mm build platform and offset toward the side of the effector with the biggest space between adjacent extruders happens to be.

    For this very reason, Stefan's original idea presented above would be competely desirable solution since, though I'm not certain if when using the regular bowden tube compliant mount the effector is limited from reaching the bed extents or not, but assuming it has full range, it would be a simple solution with which to run the Diamond as I so desire to do while removing the complications evidently introduced by attempting to utilize three RDD extruders to operate the diamond hotend with.

    Thanks for taking the time to read this message. This makes me think that the delta may not really be a suitable platform for utilizing the Diamond Hotend, and brings to the forefront of my mind once more the conversation between Deckingman and myself where I had at one time expressed my desire to build a duplicate rig, based on his, and so on the absence of any workable suggestions which to try at the place I've been stuck at (as detailed above), I respectfully re-request from Deckingman the permission (and associated plans and bill of goods I would need to obtain) in order to follow the master down a tried and true path for Diamond success!

  • @jrlederer:

    …............................... I respectfully re-request from Deckingman the permission (and associated plans and bill of goods I would need to obtain) in order to follow the master down a tried and true path for Diamond success!

    Hi Jonathan,

    I happened to stumble across this post. I don't normally pay much attention to Delta related topics as I have almost no knowledge of these weird beasts. The limited knowledge that I do have leads me to think that Deltas are best suited to light and nimble print heads, and that they are susceptible to effector tilt which can be bought about by external forces. For those reasons, I agree that Deltas may not be the best platform for a mixing hot end, which is necessarily heavy and has to be fed with multiple filaments.

    Of course, you can copy my design. However, as you know it has "evolved" into what it is now and it's fair to say that, if I was staring from scratch, I would likely do things differently. For one thing, I would likely use 5 remote drive extruders such as Zesty Nimbles which would negate the need for the second CoreXY gantry to carry the E3D titans. Having invested a fair amount of cash on 5 off E3D Titans, I can't bring myself to abandon them now and buy 5 off Zesty Nimbles. (I'm rapidly approaching the time when I only have my pension income to live off).

    As for plans and BOM, my printer was never intended to be made and sold as a kit, so it would take me a lot of time and effort to document every nut, bolt and washer. I could put the OpenScad files in a shared directory. In fact, I have done this in the past but that was a few design iterations ago so even that process would take a fair bit of time.

    As we are hijacking the OPs thread, it would probably be best if you email me and we can continue this discussion.


Log in to reply