Question on carbon delta rods



  • So I'm sourcing all the parts I want for my next Delta/Kossel build, I feel I have a pretty good grasp on how to build a higher precision machine. I'll either go with Robotdigg corners, or I'll CNC mill my own, not sure how they make the effector…ball nose end mill for where the rods connect? Hmm.

    Anyway! Thanks to the Google search of dc42's BOM for his latest Kossel build, I now know of these

    https://www.blueeaglelabs.com/collections/all-products/products/haydns-carbon-fiber-arms-with-cnc-precision-machined-delrin-sockets

    Because I am me….I believe that if it's worth doing, it's worth overdoing. At the Makerspace near me, someone donated a few, quite expensive, rc helicopters. These things are LEGIT, all carbon fiber, cnc milled aluminum, carbon fiber blades. The guy said "they were a couple grand" to build. A Google search of that revealed that two of them exceeded $4,000 to buy just as a kit, one kit was $6,000 if I'm not mistaken.

    The reason this is relevant is the tail/back half was hollow high modulus braided carbon fiber. The Heli is extremely flamboyant. It got me thinking though, if the rods are long enough to ensure the force is exerted in a way to compress the rods then braided directional carbon fiber could be perfect for this, the rods I use are a cheap carbon fiber now. My trunk is carbon fiber and the look of the strands going the length of the weave is eye catching at the very least.

    Why not do the same for a Kossel XL that will be on display? Does anyone know where to get Delta/Kossel arms or RC hollow tubes a specific length and diameter? I may attempt to make my own but I like the idea of those cnc milled delrin rod ends and I want that braided fiber look and as thin and as lightweight as possible.

    I may go overboard with the carbon on my next machine haha how light can I make the moving parts without sacrificing rigidity. I want fast printing! 🙂


  • administrators

    I suggest you contact Haydn directly by posting i this thread https://groups.google.com/forum/#!topic/deltabot/485cfVFrFFU%5B476-500%5D.



  • With a delta with Duetwifi, and linear rails, 24v and the right motors, I'd say the limiting factor in printing speed (not travel speed) is the hotend, Once you are pushing 15mm3/s through a v6 hotend you can't go any faster, and you end up with under extruded parts. Volcano helps but even then you're at about 35mm3/s which still means you're being bottlenecked by the hotend, not by your mechanicals/electronics ability to move or position accurately.

    So use fancy carbon fibre if you think it looks nice, but I wouldn't presume unless you want to turn your hand to designing a better hotend, that you will be turning objects out at 300mm/s.



  • I will, thank you David.

    DjDemonD, that is most likely why instead of staying with this technology I will probably try to reverse engineer Stratasys's PolyJet, inkjet style printer. Much better resolution and, at least in theory, could be faster.

    I have a volcano, I'll use that on the big printer, probably use 2.85 if I do any flexible filament though. I also want a direct drive extruder with a delta that sits on the effector if I can get a high enough geared small stepper. There's only so much I'll do to these three printers before going after the PolyJet style. That is, if dc42 wants to include that in his upcoming releases. Everything I do will be Open-Source unless of course it violates patents, than I could only share what I did on my own time 🙂 what you do in your own home is your business haha



  • I'm keen to see what you come up with, things do need moving forwards, 3D printers have got cheaper, and more accurate, but not really faster. I am with the "faster is desirable" group and disagree that it doesn't matter if things print slowly as long as the quality is high, this only holds water for a hobby machine or a prototyping tool, not for a machine with wider consumer appeal. I printed two bookends on my corexy in PETG, and it took 11 hours. This is silly.



  • @DjDemonD:

    I'm keen to see what you come up with, things do need moving forwards, 3D printers have got cheaper, and more accurate, but not really faster. I am with the "faster is desirable" group and disagree that it doesn't matter if things print slowly as long as the quality is high, this only holds water for a hobby machine or a prototyping tool, not for a machine with wider consumer appeal. I printed two bookends on my corexy in PETG, and it took 11 hours. This is silly.

    Couldn't agree more. Also paradoxically, the bigger the machine the more important print speed becomes but because of the extra masses involved to make it rigid, the more difficult it is to achieve higher speeds. We have a Tassimo coffee machine and I've designed a holder to take the pods - basically 6 cylinders 78mm OD, 74mm ID arranged in a circle roughly 250mm diameter and 330mm tall. I've calculated it's going to take around 60 hours to print. - I might just buy one instead.

    I dread to think how long it would take to print something that would fit the build volume of 400 x 400 x 750 (ish).



  • It's got to be multiple nozzles to lay down the filament. Idex systems with a fine nozzle for detail and a large nozzle for infill might be a good halfway house.

    David's idea for someone's super large machine was pellets (+1 for cheap) fed into a melting chamber which feeds a nozzle and is extruded with compressed air. Take this further with it feeding multiple nozzles, and make a device for opening and closing nozzles and you lay down one neat line or a 5mm wide strip depending on how many nozzles are open. Flow rate could be massive.



  • And that is exactly why I look at PolyJet as the solution to FFF manufacturing (not exactly the same I know, but very similar in a lot of ways). FFF manufacturing is great for low-cost, low-volume production, it can even produce parts pretty accurately. SLS and SLA I find VERY interesting, at this time have no hands-on experience with either, and have read that newer DMLS machines can consistently get within +/-0.003" which when you think of it, is good for a LOT of applications.

    From my understanding, at this point in time (or at least obtainable by hobby/small business…relatively speaking of course) the only way to get tight tolerance parts is either cnc milled FFF parts or SLS which is easier said than done, SLA which is probably worth it for the prototyping accuracy, or PolyJet which is located here: https://www.stratasysdirect.com/solutions/polyjet/ which they say the layer height can be as small as 16 micron which is 0.0006"….dear God...with an X/Y resolution of 0.0017”.

    If my next delta can produce parts accurate enough...(should I just build a corexy/ultimaker 3 dual ballscrew machine? )
    ...then I will begin on an inkjet. I have the engineers and piezoelectric motors to actually do it. Since I do not know how to modify dc42's reprapfirmware just yet, I'll either use marlin to start, or I'll have to quickly learn enough java...C++ might be just a bit easier, but I prefer the duet so I'll probably force myself to learn java better.

    The biggest problem I see right now is that for PolyJet, since it is so "new" the patents aren't extremely helpful and the first prototype will most likely be either A.) COMPLETELY custom which may be easier if I have control over everything, or B.) a hacked inkjet with a Z axis. Hacked inkjet will be my first attempt. We'll see, I have to learn more about inkjet printers first.


  • administrators

    @NoSkillzEngineer:

    Since I do not know how to modify dc42's reprapfirmware just yet, I'll either use marlin to start, or I'll have to quickly learn enough java…C++ might be just a bit easier, but I prefer the duet so I'll probably force myself to learn java better.

    RepRapFirmware is written in C++, so no need to learn Java.



  • @dc42:

    @NoSkillzEngineer:

    Since I do not know how to modify dc42's reprapfirmware just yet, I'll either use marlin to start, or I'll have to quickly learn enough java…C++ might be just a bit easier, but I prefer the duet so I'll probably force myself to learn java better.

    RepRapFirmware is written in C++, so no need to learn Java.

    Is it really?! Why on earth did I think it was Java?! Hmm. I must have gotten it confused with Android. Wow….That's good because I'm much more familiar with C++ than I am with Java, so that's good. I was under the belief that it was in java for most of the last year. Thank you for that correction! 🙂


Locked
 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.