Hemera And Tool Changer

Discussion in 'Motion System' started by Bob Hepple, Jan 9, 2020.

  1. Bob Hepple

    Bob Hepple Member

    Joined:
    Jun 10, 2017
    Messages:
    6
    Likes Received:
    0
    I put my purchase off in December for the Tool changer I thought I liked the idea of the new Hemera instead of the Titan is the mounts for the Hemera available yet ?? and I also noticed threads on the noise level is this something I should worry about??
     
  2. mhe

    mhe Well-Known Member

    Joined:
    Sep 5, 2015
    Messages:
    279
    Likes Received:
    77
    You can put hemeras on the Toolchanger already: https://www.thingiverse.com/thing:4006801

    The noise level is nowhere near bad, the Duet motor drivers are quiet and if you insulate the bowden tube somehow and prevent them from touching the slightly resonating side panels, this thing is pretty damn tolerable. I can run it in my living room without the missus giving me crap for it.
     
  3. Jai Stanley

    Jai Stanley Well-Known Member

    Joined:
    Aug 28, 2019
    Messages:
    115
    Likes Received:
    60
    The extrusion control from my Hemera is absolutely fantastic: print quality took a substantial step up after installing it. However: it is a very heavy, very wide extruder. This has reduced print speed considerably for the same ringing artifacts.

    There have been hints that a lighter weight Hemera with a smaller motor might be released at some point; and a kit for TC owners. However: no clairty on whether this kit is just a parts-reduced upgrade for those with bowden parts already, or something a bit better with machined parts has been forthcoming.

    I love my Hemera, but I'm also a bit MEH about it in the TC. Others love theirs. YMMV.

    Jai
     
  4. Greg Holloway

    Greg Holloway Administrator
    Staff Member

    Joined:
    Sep 4, 2015
    Messages:
    1,020
    Likes Received:
    631
    I'm going to try to write out my vision of the toolchanger, and hopefully insipire, or motivate you beyond what you can see now. Sorry to highjack the thread, but I feel like there is some narrow focus, hopefully i'll open some eyes, maybe I'll get chased off the thread instead though....

    We've been lauding the motion system / toolchanger (ms/tc) as a research platform, 'a not a printer' and there is a reason for that. We hope everyone won't look at it and see four tools on there and go ' oh look it has four print heads. what print heads can i get, oh they only have bowden, Bowdens aren't very good, but there is Hemera and well it's not quite right either'. Which, in thinking that, you're right. Bowden is bad for retractions, and as stated above the Hemera tool is heavey.

    But that doesn't really matter, the first printers were terrible that didn't stop anyone!

    Personally I hope you'll take the other approach and sit in wonder and think 'Oh look, it can change tools. I wonder what tools I can get? Only two, well I'll get maybe one or two of each. I'll have a go, understand how it works and once I know whats going on maybe i'll make a better print tool and not only that but maybe I'll make something completely unexpected that can insert cables into a print and wire up electronics, maybe other people will want one too!'

    The toolchanger is complicated. It is hard, it has taken a lot of work to get it where it is and there is waaaaaay more work and development to do. Far more than I or E3D have time, or the skills for. We made it open source because we want everyone to explore the capabilities, do the unexpected and run with the idea!

    If you don't like the two tools we have, please go make another, design what you want, other's have already started working on their own versions, some of which are posted on here. You can even get a Slice or Bondtech tool, we welcome anyone and everyone to toolchanging.

    We will be opening the sales to the general public soon as we have worked through the queue. When it's open you'll be able to buy the motion system on it's own, blank tool plates, tool plates with docks, tools and some other stuff too.

    I work 95% of my time on the toolchanger, be that design fixes, design itterations, new tools ;) or jumping on the forum to assist or hijack threads. I do it becuase I can't wait to see what new and exciting stuff is going to be coming out in the future. I hope you all share that excitement and enthusiasm too :)

    Other tools posted here:

    BMG + Mosquito + Tool Changer | https://forum.e3d-online.com/threads/bmg-mosquito-tool-changer.3411/
    Nema 14 BMG direct extruder | https://forum.e3d-online.com/threads/nema-14-bmg-direct-extruder.3461/
    Zesty Nimble | https://forum.e3d-online.com/threads/zesty-nimble.3418/
    Prototype BMG based 'tall' direct drive extruder tool | https://forum.e3d-online.com/threads/prototype-bmg-based-tall-direct-drive-extruder-tool.3479/
    Almost Direct Drive Tool Head | https://forum.e3d-online.com/threads/almost-direct-drive-tool-head.3353/
     
  5. mhe

    mhe Well-Known Member

    Joined:
    Sep 5, 2015
    Messages:
    279
    Likes Received:
    77
    Maybe we can get it to ring less by using beefier motors (maybe even proper servos) on the motion system? I'd love to see what could be done with some Clearpath stuff on this thing.
     
  6. Greg Holloway

    Greg Holloway Administrator
    Staff Member

    Joined:
    Sep 4, 2015
    Messages:
    1,020
    Likes Received:
    631
    Ecoders would be so cool, and it's something I have discussed with the guys at Duet, I want to go over step (lol) further though and have linear encoders on the axes themselves. That's where the real precision is :)
     
    Jason Chodakowski and Ntesla66 like this.
  7. mhe

    mhe Well-Known Member

    Joined:
    Sep 5, 2015
    Messages:
    279
    Likes Received:
    77
    You don't need to do anything with encoders on the Duet, if you look at Teknic's Clearpath SD series, it can be connected to a Duet merely with a step/direction interface, the servo is self-contained and does everything by itself. You basically just give it power, step/dir signal and you connect it to a PC once and it auto-configures optimally to your system. Downside is that they are only available in Nema23 upwards, so there will be some modifications necessary to fit it to the TC, but these things start at around 250 bucks a piece but damn are they worth their money.
     
  8. Greg Holloway

    Greg Holloway Administrator
    Staff Member

    Joined:
    Sep 4, 2015
    Messages:
    1,020
    Likes Received:
    631
    yup, kinda like mechanduinos and smart steppers, they are not proper closed loop as there is no feedback to the duet and positional data only comes from the rotation of the motor. for true closed loop data needs to go back to the duet, and measurement should be on the moving axes.

    the encoders are cheap too, it's just a matter of time :)
     
  9. Jai Stanley

    Jai Stanley Well-Known Member

    Joined:
    Aug 28, 2019
    Messages:
    115
    Likes Received:
    60
    Good lord Greg: I hope I didn't offend you. I just like to point out the pro's and cons of every approach. There's a reason the Hemera is the go-to head on my TC: it prints the best. Hopefully I've got the mesage right though and am trying to explore other options to see if they're better/suitable for something else. Hemera is still my go to atm though.

    On my list of to-do's are: the BMG style extruder that uses mostly the parts from the bowden setup, an interchangeable tool system (quick release) so I can drop in tools/sets of tools for different purposes and a toolhead that has a filament cutter for using something similar to the Prusa MMU without the issues of retracted filament whisps (powered on or off the toolhead, TBC. Aside from that I'd love to see tools like inspection cameras, pick and place heads, threaded insert heat setters etc etc.

    I think the platform is fantastic: I puchased mine because I was developing something similar and it made no sense to create something in isolation when E3d (yourself of course) and thus a community of others would be singing from the same hymn book. I apologise if I seem to be a ney-sayer! I'm sure I come across less blunt in real life. Possibly.

    Back on topic: linear encoders are ridiculously cheap these days. Inkjet printers use them to incredible effect for pennies; it's the development that scares me. Also; active, DSP controlled motive feedback to combat resonance is entirely feasible; just way out of my wheelhouse I'm afraid (just a former manufacturing engineer and tech sales unfortunately). Getting the minds at Duet onto a plug and play solution would be amazing. Keep hassling them.... It'll do wonders for precision of high end machines and drive the price of low end machines way down in the long run (there's a reason your inkjet printer is so cheap: naff all motors, but with precision feedback).

    I have been discussing how to improve precision on the extruder with another forum member recently also. getting fine control over the deposition of material and feedback on the motion of the extruder would be a good start. There are some nice options to add quasi closed loop control of a standard Nema17 (makerbase for example http://www.aliexpress.com/item/32917408111.html )

    I'll update my extruder thread in a bit: made some progress (ie they're useful after about the forth iteration!).
     
  10. mhe

    mhe Well-Known Member

    Joined:
    Sep 5, 2015
    Messages:
    279
    Likes Received:
    77
    Well, as long as the motors do EXACTLY what they are told to, what does it matter that there is no feedback given to the Duet? Sure, from the view of the Duet, it behaves like an open loop system, but instead of being one overall closed loop controlled by a central Duet, every motor is its own closed loop system, not requiring feedback to the central unit at all?
     
  11. Jai Stanley

    Jai Stanley Well-Known Member

    Joined:
    Aug 28, 2019
    Messages:
    115
    Likes Received:
    60
    Yea: I think you're right as a first step... That allows you to use slicer or firmware to predict behaviours of the motion system and pre-program countermeasures into the g-code/motion, like applying correction/eq to your music in advance (like a record RIAA). A better solution, or perhaps the next step, is to apply feedback on the actual performance of the system and correct on the fly: like noise cancelling headphones (to stretch the analogy).
     
  12. mhe

    mhe Well-Known Member

    Joined:
    Sep 5, 2015
    Messages:
    279
    Likes Received:
    77
    I'd say that is overthinking it already - these servos are way more capable than what I'd consider reasonable for a 3D printing application, they are just so affordable that we could afford to try messing about with them for printing.So all the would-be corrections I'd consider moot, since the motion system can be assumed perfect given the power and precision of these servos.

    But sure, for price reduction it would make a lot of sense to enable the Duet to read encoders instead of relying on electronics in each servo separately. If your servos are not as overkill, you could read and visualize overshoots etc and auto-tune accelerations directly in the Duet, now THAT would be freaking awesome.
     
    Nibbels likes this.
  13. Greg Holloway

    Greg Holloway Administrator
    Staff Member

    Joined:
    Sep 4, 2015
    Messages:
    1,020
    Likes Received:
    631
    No offence taken, and don't worry about doing so anyway. I can take it :p
     
  14. Greg Holloway

    Greg Holloway Administrator
    Staff Member

    Joined:
    Sep 4, 2015
    Messages:
    1,020
    Likes Received:
    631
    That is exactly what I'm thinking, feed the duet quadrature from the encoders and let the duet manage it all. Linear encoders are about £40 an axis. And also, the guy that designed the motion system may have been to visit the manufacturers of said encoders during the initial design investigation and made space for the magnetic stripes to be added at a later date......
     
  15. mhe

    mhe Well-Known Member

    Joined:
    Sep 5, 2015
    Messages:
    279
    Likes Received:
    77
    I do get the feeling that I might have to hold said inventor to this at some point. Until then, I'll be busy salivating and pestering @dc42 for encoder support in the Duet3, thanks! :p
     
    Greg Holloway likes this.
  16. Jai Stanley

    Jai Stanley Well-Known Member

    Joined:
    Aug 28, 2019
    Messages:
    115
    Likes Received:
    60
    Hells yea. This is VERY high on my want list.... for precision more than print quality, if those things aren't the same thing. I guess with a toolchanger this is even more important.
     
  17. dc42

    dc42 Well-Known Member

    Joined:
    Aug 16, 2016
    Messages:
    484
    Likes Received:
    216
    FWIW we have a prototype single driver CAN-connected prototype board under development, and we are taking the opportunity to include an encoder input on it for experimentation. This may in time lead to a CAN-connected board with TMC2160 + encoder input, or perhaps to a Duet 3 main board with encoder inputs on it. It's not trivial to do because apart from the firmware considerations, reading quadrature encoders at the high speeds used in 3D printers can't be done reliably by the main processor without hardware assistance. But I've worked out a solution to that.
     
  18. yngndrw

    yngndrw Well-Known Member

    Joined:
    Dec 26, 2019
    Messages:
    55
    Likes Received:
    19
    What about a tool board with a TMC4671 driver chip (That looks to support field oriented control of stepper motors) designed to bolt onto the back of a Nema 17 stepper motor with a magnetic encoder IC mounted just over where the shaft is?

    That way you don't need any special tricks with the inputs of your main microcontroller and you don't have to worry about implementing field oriented control. (And more importantly you don't need to implement it with the performance to suit such a high pole-count motor) No need to close the look right back to the trajectory planning as it doesn't make sense with a 3D printer's pre-calculated extrusion, but you should be able to attain closed loop positioning with this scheme with relative simplicity.


    As Jai has pointed out in other threads a stepper motor has an effective spring rate so when used in an open-loop system on an extruder the filament is actually displaced from the expected position and that displacement depends upon the load. At maximum load, an open loop stepper motor may be displaced by upto 90 electrical degrees in either direction. If my calculations are correct, for a Hemera with its 1.8 degree stepper motor that displacement can be upto 0.352mm of filament. (And this is in either direction, where the reverse direction is applicable to retractions etc) With closed-loop control of the extruder stepper motor, the integral component of the PID controller will remove this error regardless of the load. (At least, until the maximum torque is exceeded)
     
    #18 yngndrw, Jan 12, 2020
    Last edited: Jan 12, 2020
  19. Cam_Motion

    Cam_Motion Active Member

    Joined:
    Jan 6, 2019
    Messages:
    30
    Likes Received:
    4
    @Greg do you have a link to the linear encoders you mention? Have you heard of the ODrive, it uses hobby brushless motors and encoders to use as servos and can be used with linear encoders. This could allow the Duet to control the motors via the ODrive in stepper mode.
     
  20. Greg Holloway

    Greg Holloway Administrator
    Staff Member

    Joined:
    Sep 4, 2015
    Messages:
    1,020
    Likes Received:
    631
    Hi,

    The encoder are here | https://www.rls.si/en/products/linear-magnetic-encoders/incremental-encoders

    Yes, I know Odrive. We have an early control board although we've never used it, there was a reason why we didn't but I cant remember what it was!
     

Share This Page