Can this monster (in accuracy) be controlled directly with duetwifi?

  • Hi there,

    I'm designing my new 3d printer which will be a coreXY setup in a ultrarigid setup.

    That means the following.

    1. all metal frame
    2. Z bed is 325x328x8mm PEI coated aluminium going up on 3 screws with a lead of 2mm, driven by 3 0.9° steppers which will give effective Z change of 2mm / 400 = 0,005mm per full step.
    3. X and Y axes run on THK linear caged ball guides, SRS12 and HSR12 models
    4. Dual extruder/hotendh (e3d titan aero) for dual direct drive extrude with the shortest possible extrusion path
    5. The main motors for gantry movement are Lin Engineering 5704-03 Stepper motors. These are Nema23, but they are also 0.45° motors, meaning a real 800!!! full steps per rotation for absolute precision. Normal 200step steppers have a permissible inaccuracy of 5% of 1 step length, which is about (360 /200 / 20) = 0.09° , while these motors have an accuracy of 0.017° . (0.09° on a gt2 20T pulley is 0,01mm while 0,017° on that pulley is 0.00188mm

    I do understand that this seems like overkill and maybe it is, but… I already have these parts here + I do intend to not always print with 0.4 nozzles, but to ultimately go printing with the ultrasmall 0.15mm nozzles. Of course that is not meant to print huge parts. 0.15mm nozzles are slow like hell, but the parts I'd like to print in those qualities are maybe max 20x20x20mm . The printer is a lot larger, but thats only for using larger nozzles of course.

    I read (for example at e3d) that with a 0.15mm nozzle you really already see the artifacts of a 1.8° stepper and they advise to use a 0.9° stepper. I have 0.45° so that should ultimately be better in terms of accuracy.

    I am also 100% aware of the fact that these motors will NOT give me the absolute top speeds like 1.8° steppers could give me, because they are larger and therefore will have more inertia, but that is still in quite acceptable ranges.

    A lot of information (just to make clear what I would like to make), but now comes the important question.

    Can I drive such a motor directly from the Duet Wifi @ 24V ? The motors are rated ( at 3.0A , but I read on the website that best practice is to run steppers at about 85% or rated power, so 0.85 x 3.0 = 2.55A

    I see that currently firmware limits the current to 2.4A , so is this OK ?

    Any information would be highly appreciated. I'm spending a large amount on these parts just to be able to deliver the finest possible tolerances I can imagine. It will still be a challenge afterwards to effectively print high quality with it, but form a mechanical standpoint I would like to be sure not a single part can drive me nuts 😉

    Kind regards,


  • administrators

    Hi Bart

    Sounds like a very interesting printer. The motors will work on the Duet Wifi and Duet Ethernet, they are only slightly higher rated than some Nema 23s we used for testing before increasing the limit in firmware to 2.4A. If you are running the motors at 2.4A then you will need a fan blowing over the underside of the board to keep the temperature of the drivers within range.



  • Thank you for the fast answer Tony.

    I have another (unrelated) question for you. I allready posted a couple of things on the duet forum and normally I should get a notification when somebody posts an answer to a thread I follow

    I have setup in my settings on the forum (under contact) that under the item subscriptions both checkboxes are checked, so that means I should get plan text version of new comments + automatically subscribe to every thread I comment in.

    Until now I didn't receive any mails about messages posted on this forum. Even not the notification of your answer. I knew I have to reload the page every 10 or 15 minutes and see if somebody answered.

    Is it possible to have a look what could possibly be wrong ? I would like to receive notifications.

    My email address is and the mail is nowhere (also not in spam or other folders)

    Kind regards,

  • Bart,

    Although as Tony has said, those motors will work, your print speed will be limited. I can't remember what the maximum step frequency is (DC42 may step in here and tell you) but check to make sure that the speed will still be acceptable. I know you are looking for the ultimate in precision but you have indicated that at some time, you'll want to print larger objects with a larger nozzle. Paradoxically, the larger the machine, the higher speed you need in order to keep the print time for (larger objects) to 10s of hours rather than several days (and you do have a largish bed).

    Just a thought


  • Absolutely Ian.

    I have to admit 95% of my prints will be less than 50x50x50mm and require the super high resolution. The bigger prints are not that big of a problem.

    Also… These motors are 4 times finer than regular 200steps steppers (at 800 steps) but if I'm not mistaken, then you can set the microstepping over software ? If you normally print at 16 microsteps, then with these motors I can technically speaking work at 4 microsteps and still have benefit of the more precise motion.

    So speed isn't a big issue.

    PS. I still didn't get a mail notification that you posted some answer to this thread. Do you receive notifications Ian ? I have to come manually check in every topic I ever wrote to see if somebody posted a followup message. Easy to miss stuff that way and I don't think this is the intended way of working on the forum.

    Kind regards,


  • I don't subscribe to any threads - I just look at recent activity from time to time so can't help you with that.

  • administrators


    Absolutely Ian.
    Also… These motors are 4 times finer than regular 200steps steppers (at 800 steps) but if I'm not mistaken, then you can set the microstepping over software ? If you normally print at 16 microsteps, then with these motors I can technically speaking work at 4 microsteps and still have benefit of the more precise motion.

    There are (at least) 3 aspects that limit motor speed:

    1. Motor inductance * motor drive frequency. The smaller the step angle of the motor, the higher the frequency you need to drive it at for a given movement speed.

    2. Motor back emf due to motion.

    3. The rate at which the Duet can generate microstep pulses. This can be controlled by using lower microstepping if necessary.

    #1 is likely to be the killer if you use 0.45deg motors. See

  • Whilst I like the ambition of it, I'd would take a slight tangent and say why not build two more conventional machines. One micro sized printer which can be equipped with 0.9 degree motors and tiny nozzles, it's moving mass is very small, and calibrating it is much more straight forward (and accurate) than a larger machine, plus it will use far less power. One larger machine designed for printing larger objects with faster motion (but less accuracy required) enclosure and powerful heaters, I've seen mains heater cartridges being sold, to melt large volumes of plastic.

    This has been my approach, whilst my kossel Xl can print micro sized objects quite well it seems ridiculous to me to be using 500w to heat a bed 100 times bigger than the object I'm printing. So I have a micro delta which is still in the process of being honed to produce micro sized prints, and an enclosed corexy which happily prints with 1mm nozzles in abs.

    I don't want to put you off because if you can do it if will be an awesome achievement, I just worry you'll end up with a machine which is neither accurate enough for tiny things or fast enough for big ones. Think amphibious cars, potentially not great on the road and not great on the water, but if someone can figure out how to do it then I'll be the first to congratulate them.

  • I don't really understand why my setup might be not accurate enough ? Can you elaborate a little more on that ?

    Only thing I can see is that my machine might not be able to achieve super speeds. That is something I do certainly agree with.

    But what would not be accurate on an all metal machine with THK linear slides (these babies cost a couple of $100's a piece) + a 325x325x8mm alu bed leveled on 3 leadscrews with 3 extra lineair guiderails for making it even more rigid + all other parts will be precision milled in Delrin or Aluminium?

    I also did some calculations based on the formula's and this is what I calculated.

    My motor is

    800 steps per revolution or 0.45° per step with accuracy of 1 arc minute (as far as I know only 1 manufacturer in the world makes these: Lin Engineering)
    U = 24V
    R = 1.2Ohm per phase
    I = 3 Amps
    Inductance = 1.2mH/Phase
    Inertia 2.1oz/in

    This looks like very low resistance and inductance if you ask me? These motors are highly specialized stuff costing around $250 a piece. (and yes I bought these but they are currently mounted on my self made laser cutter 😉 so I already have them 🙂 + the laser could work as well with my other regular nema23 motors 😉 and then use these on my 3dprinter)


    Time to fully energize coil = 3 x 1.2 / (24 - 0.5 x 3 x 1.2) = 0.162ms
    Max speed (20steps per mm) -> 1000 / (20 x 0.162) = 308 mm/s

    308 mm/s = 6160steps per second , so in 16x microstepping thats 98560steps or 98,560kHz (perfectly possible on duet)

    Am I missing something?
    Am I calculating something wrong DC42 ? You seem to be very knowledged about these kinds of calculations and I just THINK that I used the formula's correctly.

  • administrators

    The inductance alone means that torque will start dropping off at around 200mm/sec., but that may be enough speed for you. See the link I posted earlier

  • The frame is very important and it's not an easy task to make a rigid and straight frame of that size with holes in correct places.
    This is completely out of reprap thinking and more into CNC weld>age>machine world. Large build size and super precision also don't pair well with H-Bot / Core XY simply because of belt drive. If you are going to print mainly small objects and use high end components think about using screws for X and Y movement also.
    Core XY is about less mass at effector and thus more speed. You won't get speed with 0.45 deg. motors anyway so use more precise and easier to debug cartesian gantry on screws. Otherwise the precision goes out of the window just because of belts which are elastic and support backlash feature very well and pulleys which are always off center.

    I'm not sure about power supply but probably not impulse one (transformer based) would be the best. There are special supplies for stepper higher class motors.

    I think building such a machine is a dream of every one of us 🙂
    And a 2mm long hole in the nozzle just to smooth the turbulence.

  • Well I'm certainly not doing "reprap" thinking, thats a fact. I don't even think e3d would advise anyone to put a 0,15mm nozzle on a machine based on printed parts.

    I allready made an all metal machine, which was a 1600x1200x800 cnc 80W co2 laser cutter. It uses the same principle and achieves quite good speeds and precision. It is simple cartesian and not corexy, so there is a lot more mass on the X gantry.

    I'll try it anyway with the nema23 0,45° steppers . If it doesn't give nice results, the swapping to nema17 is maybe 15 minutes of work, because in a coreXY setup motors are easily switched.

    I do understand what all of you above are saying, so it is not that I want to pretend I know better. I don't know better for sure, but I have already experience with these motors and will give them a try. If they don't give me enough speed, then they are easily swappable for another motor.

  • Bart,

    Go for it. As you say, if you already have the motors why not try them? Also, as you rightly point out, it only takes about 15 minutes to change motors. So if you did need to print something big and fast, it wouldn't be the end of the world if you had to swap motors to do it. It sounds like you already know how to make a rigid frame which is a long way towards getting good quality prints. Let us know how you get on with motors though.


  • Thanks Ian.

    By the way, I am drooooooling when looking at your video's with the diamond hotend.
    I do have an E3D Kraken here, but I'm affraid that a bowden setup is certainly not going to let me use 0,25 or 0,15mm nozzles without troubles.

    So in first instance I'll design for single and dual titan aero direct drive, but since my carriages will be cnc milled out or alu and delrin, I will provide easy mounting options to exchange the printing head without having to retighten the belt system.

    I'm currently busy designing the important pieces in fusion 360. I don't know if I'll model everything in fusion because I think that might be a loss of time and a lot of mechanical thinkwork is easily solvable in 2D too (like seeing how far a carriage has travel and where nozzles will be when carriages are at their end.

    I'll certainly keep posting and the intend is to go for a duet as driving board. Most possibly NOT a duet wifi as my main alu frame is 45x45mm alu and from experience with my lasercutter (which has built in computer too) inside those faraday cages you don't get a lot of signal 🙂 . So I might go for the ethernet version

  • Bart,

    There are a few downsides to the diamond, one of which is that you can't change nozzles so you are kind of stuck with 0.4mm and as it's brass, it may not be suitable for abrasive materials. I do have a 0.9mm diamond which was a special order and which I intend to use with Taulman T glass but I doubt if you'll get a 0.25 or 0.15 (although it's always worth asking the question).

    Can't help with fusion - I just use OpenScad as it's the only thing I've been able to get to grips with. Cad design isn't my strong point.

    Best of luck with the build.

    Edit. I now have the Ethernet version of Duet and personally much prefer it to the WiFi version - that's just my personal preference of course.

  • We might be able to buy 0,25 or 0,15mm custom diamond, but what I actually meant was that pushing filament through a 0,15mm nozzle through bowden…. is probably not going to work at all.

Log in to reply