Sizing a stepper motor for Z and three drive shafts.



  • I am building a D-bot core XY printer. As I have a bigger bed than std so I need to use 3 x 1204 ball screws with a belt drive. The problem, I need to size a stepper for this duty and finding it rather hard to do. Has anyone done this and can advise? I am sort of thinking a nema 23 in the 2.8amp range as this seems common and hopefully drive it below 2000mA on the wifiduet. Does this seem silly? I will have a 24volt Vin. Otherwise Nema 17s all seem to peak at 1.8amps….


  • administrators

    A 2.8A Nema 23 sounds suitable to me. I expect to increase the maximum motor current somewhat in a future firmware revision.



  • I have a bed that weighs in at around 7Kg (400mm x 400mm x 10mm thick aluminium + 2020 frame + 6mm glass + rigid insulation + silicone heater). It is driven by a single Nema17 via a continuous belt driving 3 off 8mm x 1mm pitch lead screws. I use these http://uk.stepperonline.com/3d-printer-nema-17-stepper-motor-59ncm84ozin-2a-17hs192004s1-p-18.html. For normal use, I run the motor current at 1800 mA (about 80%) but I drop it to 1200mA for homing and it works fine.

    I don't using any gearing. That is to say that 20 tooth pulleys are fitted to the motor shaft and all three screws. I'm assuming the "4" in 1204 means that your ball screws will be 4mm pitch. Personally I would never use course pitch screws on a Z axis but if you are dead set on it, then I'd advise that you use gearing to lower the torque requirement for the motor and bring the resolution back to the point where you will not be relying on micro steps for positional accuracy. So a small pulley on the motor shaft and big ones on the screws (or just buy fine pitch screws - IMO ball screws are overkill for the Z axis on a 3D printer).



  • my setup has Quad Z screws turned by a single Nema 23 with GT2 Belts on a 2:1 gear ratio
    no issues
    A geared Nema 17 would also work.



  • I use the same setup (1204, Nema 23 2.4 amp) at 4:1. It is running fine at only 800ma but I will ramp it up to 1.8amp when fully commissioned. I think a Nema 17 would have been fine. Pics here http://forums.reprap.org/read.php?397,676997,page=7



  • Interesting first post. An exact verbatim, word for word copy, (and paste?) of something I wrote about 3 posts up.



  • @deckingman:

    8><–--

    I'm assuming the "4" in 1204 means that your ball screws will be 4mm pitch. Personally I would never use course pitch screws on a Z axis but if you are dead set on it, then I'd advise that you use gearing to lower the torque requirement for the motor and bring the resolution back to the point where you will not be relying on micro steps for positional accuracy. 8><---

    Hmm interesting. As a beginner building my second 3D printer it was not something I had specifically considered.

    The ball screw on the Wanhao D6 works very well, and its direct drive. Yes 4mm lead, however 36teeth on each pulley that is on each drive shaft, I can go bigger if need be to get more ratio. Not sure on the pulley on the Nema yet I have 12's and 16's toothed pulleys to hand. My problem right now is finding belts in the length I want to do the job. I seem to be able to find 1524mm and 2270mm GT2 continuous belts and nothing in between, I have 1 of each coming to mess with.



  • Ref pulleys. Try and use ratios that give a whole number of complete steps per mm (or preferably steps per 0.1m). So, with no gearing, assuming a 1.8 degree stepper you get 200 steps per revolution. 1 revolution = 4mm so that's 50 steps per mm or 5 steps per 0.1mm which is fine. With a 36tooth pulley on the screw and 12tooth on the shaft that gives you a ratio of 3:1 so you'll get 150 steps/mm or 15steps per 0.1mm which is also fine. If you use the 16 tooth pulley on the shaft, the ratio is 36:16 or 2.25:1 which becomes 112.5 steps per mm. 0.1mm is therefore 11.25 steps, 0.2mm would be 22.5 steps, 0.3mm would be 33.75 steps etc. i.e, with that ratio you are looking at having to use micro steps for positional accuracy which IMO isn't a god idea.

    Ref belts. Try to bet more than a 90 degree bend around the pulleys. That usually means running the belt in triangle and having a couple of idler pulleys along one side with the motor mounted inside the triangle. The position of the motor can then take up the excess belt length and if you mount the motor on a sliding slotted carriage, it can also serve as the belt tensioner.



  • About belt and pulley contact there is an reference pulley tooth engagement. Which is min 6 for the stated belt specifications.
    There is also min distance between pulleys depending on size, which in 3d printers is almost of no concern.

    Idlers should be avoided as much as possible, if necessary should have bearings and best is to be use double sided belt and toothed idlers. Simple idler could be formed by hardened steel shoulder screw and 1 or a couple of bearings

    I have recently read many incorrect info on 3d forums concerning belts, so these are my 2cents. Especially about plain and toothed idlers. People constantly use plain idlers for tooth contact and this is absolutely ridiculous, even from DIY point of vieew, an i am not an engineer.

    Each machine is as much as accurate as its least accurate part… / Ok , the saying was for strength, but i adapted it 🙂

    There is also excellent belt calculator for simple setups , which i could confirm from experience is very accurate. http://www.bbman.com/belt-length-calculator/



  • In terms of steps there are 200 per revolution? so 4mm /200 = 0.02mm. Typically a print will be in the 0.06 to 0.12? so 3 and 6 whole steps respectively? ie no microstepping?



  • @gzcwnk:

    In terms of steps there are 200 per revolution? so 4mm /200 = 0.02mm. Typically a print will be in the 0.06 to 0.12? so 3 and 6 whole steps respectively? ie no microstepping?

    200 steps per rev for the motor. So with 1:1 gearing it'll be 200 steps per 4mm or 50 steps per mm or 5 steps per 0.1 mm so you won't be relying on micro steps for positional accuracy providing your layer height is divisible by 0.1mm. Also 0.06 layer height would indeed by 3 whole steps (but are you serious?). But 0.05 layer height would be 2.5 full steps so not so good IMO.

    Edit. 1:2 gearing would give you 400 steps per 4mm or 100 steps per mm which would allow any layer height of 1/100th of a mm in full steps.


  • administrators

    You can also get stepper motors with 0.9 degrees per step instead of 1.8 degrees, so 400 steps per revolution.



  • I've decided to go with 4 direct drive Nema 23s on the 4 Z lead screws of my Core XY build. Currently I have 4 N23s I bought from Steppersonline.com but 2 are http://www.omc-stepperonline.com/9deg-nema-23-bipolar-121v-038a-09nm1275ozin-23hm200384s-p-24.html and 2 are http://www.omc-stepperonline.com/nema-23-cnc-stepper-motor-24nm340ozin-18a-23hs411804s-p-126.html. I don't believe I'll need 4 of the more powerful 1804s (1.8°) 1.8a, but I'm concerned that the 0384s (.9°) .38a will have the necessary power to direct drive the XY carriage in Z motion (the bed is fixed). I need to buy 2 more motors….but which one. I have 2 power supplys 24v and a 36v available, and 4 external ST-MT045 microstep drivers. Any recommendation on which route to take? Thanks...


  • administrators

    How heavy is the total XY gantry etc that the Z axis is lifting? That's a lot of motors even for a large printer.

    Also you should consider if you need 0.9 degree motors on the Z axis where you already are likely to have a larger steps/mm (depending on the leadscrew pitch of course).



  • To add to what Tony has said, I have already mentioned that a single hight torque Nema17 will lift a 7kg bed at just over half it's rated current, so 4 off Nema 23s does sound like a lot of overkill. However, it's you build…............

    I'd be more concerned with the fact that you are going with 4 screws. 3 screws will level the bed or keep it level, 4 will twist it. Or in your case, 3 screws will move the XY in a level plane, 4 screws could impart twisting forces (especially when each one has a Nema 23 driving it). That is if the screws are rigidly fixed at one end and the moving nut is rigidly fixed to the XY axes. If they are not rigidly fixed and rely on gravity, you won't get the twisting forces (because one screw/nut won't be in contact).




  • I constructed the chassis out of C-beam V-slot extrusions. I'm using Openbuild 8mm 4 start lead screws at each corner. The bottom of the screws are rigid with the motors and the lead nuts are fixed to the corner plates of the XY carriage. The XY carriage itself is C-beam along the Y axis and 2040 V-slot on the X axis. Currently the X, Y, and Z axis are mounted with V-slot mini wheels, but I'm testing another type of linear guide system for Z and Y. I plan to leave the X on mini-wheels for now. The XY carriage will weigh a little less than 15 lbs loaded up with extruders, parts and pieces.

    I have spare N17s I could use for each Z corner but everything is already built for N23s so adding 2 more is just a matter of bolting them on versus having to change all the mounts and flex connectors for N17 motors. I'd rather buy 2 more 23s and be done with it. At 1:1 direct drive, the calculators say 1.8° motors produce 2560 steps/mm vs. 5120 steps/mm for the .09° motors. .09° may be overkill for Z but its an option since I already have two 23s of that type. The question is which of the two 23 types would be the best choice given the current state of the build. I agree that either type is overkill from a power standpoint but better to have too much than not enough IMO… Does anyone think .38 amp wiring on the shorter 23s will be an issue? I have no experience with steppers or electronics to know how to make that decision.

    Your point about twisting the XY carriage using 4 screws is understood Deckingman, but 4 screws fit logically a 4 post design opposed to just 3. Originally I planned to use just 2 screws and motors at cross corners of the carriage, but during assembly I recognized there was enough play in the large carriage structure to know that over time the non-screw corners would sag. So I decided to add 2 extra screws and support all 4 corners and belt drive all the screws together with 2 motors.

    dc42 informed me about upcoming firmware changes to Duet WiFi that would eliminate multiple motor sync problems on the Z axis. So now its practical and logical to add extra motors direct driving the 4 screws. This also eliminates another problem for me which was a leveling mechanism for the large, heavy, heated bed. The chassis is fabricated around a braced stainless cart where everything is squared and plum to the cart counter top. So by fix mounting the bed squared with the chassis, I can now probe the bed and use the Z motors to level the carriage itself to the bed rather than the other way around. There is sufficient flex in the carriage corners to allow for a small degree of adjustment necessary to get a virtually perfect level every power cycle.

    Using this 4 screw design, I think one could reasonably build a range of motion into each XY carriage corner connection where no rigidity of the carriage itself existed. The combination of C-beam Z and Y rails are sufficiently rigid to ensure true motion along those axis even if the XY carriage end rails shown in the drawing were eliminated. My design requires the end rails because that's where my XY motors and cable bearings are located. Easy enough to change that but I don't see the need for this build; although it would make it a lot easier to string the synchromesh drive cables without the end rails.

    Thanks for the feedback...TP.



  • I'm a big fan of Open Builds V slot - that's what I used and have no regrets. I have heard some negative things about the mini wheels though, so on my build I used the "standard" larger wheels (the Delrin version). After a few hundred hours of printing, there are no discernable signs of wear or "slop".

    I'm a bit concerned about your calculations of steps per mm though. Assuming you are talking about micro steps rather than full steps then your 2560 becomes 160 full steps per mm. A 1.8 degree stepper will do 200 steps per rev so 160 steps per mm would mean a pitch of 1.25 mm. BUT, with a 4 start screw 1.25 would be the lead. So, the pitch becomes 1.25/4 = 0.31mm which I have never ever seen (don't think it's possible).

    A quick gander at the spec for Open Builds Acme Metric screws show the pitch as being 2mm with the lead (that you should be using to calculate steps per mm) as being 8mm. This means that one revolution of 200 steps equates to 8mm of linear movement or 25 steps per mm which equates to 400 micro-steps per mm at 16 X micro stepping. This means that any layer height that is not an exact multiple of 2.5 will require the use of micro stepping for positional accuracy. i.e. 0.1mm will be 2.5 full steps (not good), 0.2 mm will be 5 full steps (good), 0.3mm will be 7.5 full steps (again, not good).

    IMO, multi start threads are only good for what they were designed for. That is, to give a large linear move for a small angular rotation where linear speed is required. Fine for linear actuators but not what we need on the Z axis. Also, an 8mm pitch (lead) screw will require 8 times the torque to drive it than a 1mm pitch screw.

    I'm not trying to criticise but as you are the design/build stage, I'd strongly urge you to look again at the screws and source a finer "lead".

    HTH
    Ian



  • Thanks for bringing those calculations to my attention Ian. I used the RepRap calculator to produce the numbers cited above but that calculator doesn't allow you to change lead distance so my numbers are wrong.

    As you say the openbuilds screw has an 8mm lead so I'm looking now at the Pursa calculator that does allow input for an 8mm lead which confirms your 25 steps per 1mm @ 1:1 steps and 1:1 gearing. 25 x 0.1 = 2.5. Moving to .9° motors changes whole steps to 50/1mm which is whole number divisible at each 0.1 layer. Seems then that using 0.9° motors solves that problem. (Sorry to be so repetitive here but I'm learning as I go).

    If I understand correctly, its best to use a combination of steps and lead that are whole number divisible using 0.1 as the layer height example? In terms of torque requirements for a 4 start screw, my carriage weight is approximately 15 lbs, that gives 3.75 lbs per corner. Using the .xls stepper motor calculator I found on Openbuilds, it looks like the screw force for 8mm lead screws using the 0.38a /.09° Steppersonline motors I linked above is 95.38 lbs. ea. So looks like torque requirements are covered. Speed is not very fast @ 37.3 IPM but that's less than a minute Z travel in each direction for my 800mm height.

    The OB .xls calculator gives screw steps/inch for this combination @ 1270 for whole steps (.02 mm layer height) and 20,320 1/16 micro step which is less than .00005 layer height. Is that right? Given the 800mm Z travel of my design, I think the longer 8mm lead is preferable given the available torque at all 4 corners.

    Thanks for helping me to understand how this all applies Ian.
    Terry



  • Hi Terry and glad to be of help.

    Yes 0.9 degree steppers would solve the issue or 1:2 gearing (but I guess that would mean a major re-design). A you say, you have 4 motors so it looks like you won't have an issue with torque. IMO yes, having a combination of FULL steps and lead that will give a whole number divisible by 0.1 is a good approach as you can use any layer height in 0.1mm increments.

    It's a long time since I worked in inches but my quick calcs would be as follows. 0.9 degree stepper gives 400 full steps per rev. 1 rev=8mm (the lead of your screws) so 400 steps is 8mm or 50 steps per mm so we agree that 1 full step = 0.02 mm and therefore 0.1mm would be 5 whole steps.

    For info, my Z travel is 750 mm, so similar to yours but I use 1mm pitch single start screws. Z travel is slow but I can get away with a single Nema 17 - As the saying goes "you pays your money and take your choice".

    I hope you don't mind me pointing this out but there is something else I'd be a bit concerned about with your build. It may not be a problem but I see that you have the motors under the screws and it looks like the motors will be taking the full weight of the bed? Difficult to tell from the pics. If so, then my concern is that steppers don't generally have thrust bearings and aren't usually designed with the spindles to be loaded in that manner. If you have some sort of thrust bearing under the screws or if the motors you plan on using have thrust bearings (which is unusual) then it's not an issue.

    HTH

    Ian


  • administrators

    @deckingman:

    I hope you don't mind me pointing this out but there is something else I'd be a bit concerned about with your build. It may not be a problem but I see that you have the motors under the screws and it looks like the motors will be taking the full weight of the bed? Difficult to tell from the pics. If so, then my concern is that steppers don't generally have thrust bearings and aren't usually designed with the spindles to be loaded in that manner. If you have some sort of thrust bearing under the screws or if the motors you plan on using have thrust bearings (which is unusual) then it's not an issue.

    I would be concerned about that too. You could mount the leadscrews on thrust bearings, then it would be convenient to use 1:2 gearing to drive them from the motors. Or revert to using belt drive. Or perhaps it is possible to find a thrust bearing/leadscrew combo that lets the leadscrew shaft protrude below the bearing, then you could use a flexible coupler to connect it to the stepper motor.


Log in to reply