Z Bed setup with auto calibration, three leadscrew design



  • Hi Guys and Gals,

    just finished my own custom build printer. It works well however there are a number of things I'd like to change for my next design, one being the Z assembly which moves the bed.

    I have an idea but not sure if it'll work as well as I think it will. A bit of background, the printers are for my dad who has ALS and thus my mom will have to do the physical things like changing filament and stuff like that. She's as nontechnical as they come so the goal is to have a printer ANYONE can use!

    My current Z assembly has two motors on one driver. It's ok. However does not fill the needs. Especially since the printer has a kraken 4x hotend. I often need to reset the z axes. Now I've made some work around which makes it easier, but nothing I want my mom to have to do for a long time.

    Now for the new printer I'm going to have a larger print surface (400x400, which will compound on the issues I currently have) as well as a different Z approach.

    The Z will move the complete Z assembly. The Z will consist of three independent motors each with their own driver. Each motor will drive a 700mm long SFU1605 ball screw. Each SFU1605 will be securely mounted.

    The idea is to have only the SFU1605's and no other Z guidance, so no rods nor wheels running on the alu profile. I assume since the SFU1605 is a 16mm "rod" this will work well.

    A BL-Touch will handle bed sensing and the duetwifi will make sure the bed is level using the 3 motors.

    What do you guys think? Will this indeed work well. Without any tinkering needed after the implementation (assuming I implement it correctly 😛 ).

    Edits: Some corrections and clarifications.



  • @nemesis said in Z Bed setup with auto calibration, advise requested:

    The idea is to have only the SFU1605's and no other Z guidance, so no rods nor wheels running on the alu profile. I assume since the SFU1605 is a 16mm "rod" this will work well.

    That seems kind of dicey. Even 16mm rods will be prone to wobble when the bed is lifted. Try holding up a chopstick by pinching it between your thumb and forefinger. Now wave your hand around, does the top of the chopstick sway at all? I'll bet it does.

    What's the reason for not wanting to use linear guides?



  • @phaedrux said in Z Bed setup with auto calibration, advise requested:

    What's the reason for not wanting to use linear guides?

    Because I thought I wouldn't need em if I did it this way. It would be a simpler and cleaner design. So basically no other reason then it looks cool 😛



  • @phaedrux said in Z Bed setup with auto calibration, advise requested:

    Even 16mm rods will be prone to wobble when the bed is lifted. Try holding up a chopstick by pinching it between your thumb and forefinger.

    Hmm good point I realize I haven't provided enough context. The rods will be secured on top and bottom with bearings. The printer will have a extruded 2020/2040 alu frame.

    Some images of the current design
    0_1545420802676_94d1e16b-3bdb-422a-b872-b831ab069bf2-image.png
    0_1545421606731_2dd60d78-2dad-4330-be98-689010a8ca04-image.png
    0_1545421668828_537ef0f7-e2d8-404d-80d2-5fde3fd0e357-image.png

    The design is not final yet so bear with me 😉

    • In the bottom the SFU's (orange) would rest in a 12mm bearing mounted to the frame.
    • The SFU would of course be mounted to the motor (yellow) and the motor in turn to frame. (not sure yet what motor couplings I'll use not happy with my current ones)
    • In the top the SFU would be kept in place by a 10mm bearing.
    • For the Carraige (red) I'm thinking of using CoreXY principle. it'll run on the 2040(green) with vslot wheels. The 2040 (green) you see in the top will use vslot wheels to run over the 2040 (purple)

    Now what I could of course do is add a 2020 in the back and use vslot wheels on that and the existing 2020 in the corners.



  • This setup could work, BUT it is over constrained. If there is any misalignment between top and bottom on any of the rods you'll get binding. It's not best practice in other words, and will likely be frustrating to implement because of the need for constant perfect alignment.

    Since you're already using V-slot for the XY motion, you might as well look at what the DBot does for using vslot in the Z axis. I'm not saying what you have planned can't be made to work, just saying that it will be very tricky to get and keep working.

    I've had my own adventures with overconstraint. My lead screws are oly fixed at the bottom, but they are mounted in double bearing blocks that don't let them wiggle very much, and combined with being practically constrained at the bed mounts it makes for an overconstrained system and is prone to binding if they are not very carefully aligned to be perfectly vertical. So do as I say, not as I do. 😉



  • To start off just a disclaimer: I'm a software engineer/architect. So on a structural level I do not really know what I'm talking about 😉

    @phaedrux "It's not best practice"
    I find that really weird for many reasons. Yeah I agree it's much more difficult to get everything aligned for a constrained system. However I think it would have many advantages: higher repeatability, far less vibrations.

    That having wiggling parts of any kind is the defacto standard is strange to me and just feels wrong (Yes yes I do often over engineer 😛 ).

    From what I have read and deduced up to this point is that you want to keep vibrations to a minimum. So having rods free balling sounds counter intuitive to me. On the other hand I do get your point of not having the alignment headache.

    @phaedrux Curious why you'd say it woud be difficult to keep working. I would actually assume that since it's contrained. Repeatability would be higher and if aligned properly would be much less prone to need tinkering.

    On the other hand since I'm going the CoreXY route I'll have the alignment nightmare anyway. Since the principle requires near perfect alignment (from what I read).

    My first design looks as follows and works fairly well, I will of course change many of the principles. I think most would also call this over constrained.

    1_1545472984220_1f4c77c6-bf42-436f-b5d2-34de7f349dd7.jpg 0_1545472984214_f255aeae-d379-480e-82aa-5876a05798aa.jpg



  • @phaedrux said in Z Bed setup with auto calibration, advise requested:

    I've had my own adventures with overconstraint. My lead screws are oly fixed at the bottom, but they are mounted in double bearing blocks that don't let them wiggle very much, and combined with being practically constrained at the bed mounts it makes for an overconstrained system and is prone to binding if they are not very carefully aligned to be perfectly vertical. So do as I say, not as I do. 😉

    Just wanted to reply to this, with some findings of my own. I've been looking at how industrial applications solve these issues and discussed this with a friend of mine (who has a Phd in structural engineering).

    So what is normally done (this is for stuff that needs to be able to handle heavy loads) in these application is creating a fully constrained enclosure which actually looks like my previous design (that where I actually stole it 😛 ). That would then be mounted where needed removing any off the alignement issues (since the part itself takes care of aligment). Alignment to motor will be handled by a proper motor coupling.

    Something like this:
    0_1545473919473_666aa319-8797-4cc1-8288-da186b48d70f-image.png

    Even though cost is not really an issue for me (up to a certian point that is) this is a bit of over kill. I realise this one is for horizontal motion only 😉



  • That ball screw probably costs as much as an Ender3. 😉

    I'm not a mechanical engineer. Far from it. I'm only relating my received wisdom as it relates to building a DIY desktop 3D printer. The general advice is to not over constrain the leadscrew since it should be your linear guides that handle the XY and rotational constraint, leaving your Z axis lead screw to only provide force in the Z direction. Having the end of the rod free to wiggle helps account for rods that aren't perfectly straight, which was very common in the early days of reprap, especially when the designs used threaded rod more often than actual machine lead screws. Even cheap chinese lead screws seem to be pretty decent now days, and almost no one uses threaded rod.

    Now as I said, it's not that an overconstrained system can't be made to work, it just requires alignment, and that alignment can't drift, or it's prone to binding. My own printer is over constrained by definition, but when properly aligned it works perfectly. It did require some hair pulling however. The industrial designs you mention take care of that alignment and eliminate the hair pulling.

    But now that you're aware of the prevailing advice please feel free to ignore it and build whatever your creativity and ingenuity can come up with. It would be rather boring if we all built the same printer, and it's far too early in 3D printing to say the perfect FDM design has been conceived.

    These links may prove useful to you in your design phase.

    https://somei3deas.wordpress.com/blog/

    https://drmrehorst.blogspot.com/2017/07/ultra-megamax-dominator-3d-printer.html

    https://reprap.org/forum/list.php?397



  • Hi,

    I am working on a D-Bot CoreXY machine using a set of Chinese parts. The printed parts were done by Shapeways using their "Versatile Plastic".

    Despite the simplicity of the design it seems to be working very well. The movement in all axes is smoother than I expected given the use of the "wheels-on-extrusion" approach.

    In regards to your idea I would not try to use lead screws in the manner you described. Let the lead screws do the lifting but let something else do the guiding.

    While the idea of auto-leveling is grand, based on my limited experience it is not needed in these types of printers.

    I only have a manual leveling system. I do this once a month of so. It only takes a few minutes and the printers work just fine.



  • @fcwilt said in Z Bed setup with auto calibration, advise requested:

    I only have a manual leveling system. I do this once a month of so. It only takes a few minutes and the printers work just fine.

    Thnx for the reply. For myself I would agree. However this machine is for my dad who has ALS. My mom isn't going to calibrate the bed for him, nor do I want her to get near the printer with anything other then a spatula to get the print off 😉

    I do not dispute the D-Bot would work great. I'm sure it does.

    Would you mind giving me the reasoning behind why you would not use 16mm diameter leadscrews to do this?



  • @phaedrux Thnx for your insights I really appreciate it and as you've surmised I'm not one to conform to standards 😛

    The SFU1605 700mm costs 30 Euro a piece, which isn't that bad! Also DuetWifi with Duex5 controlling the printer is far more expensive then the Endor3 😛

    The hair pulling isn't really an issue, I have little to non left anyway 😛 And what is a hobby project without a bit of frustration?

    I'll go through the links you posted and probably fine tune the design and build it with 3 SFU1605's just to prove it can be done 😛

    Also I have a feeling that current 3D printer DYI designs are based on costs and my main goals are repeat-ability, ease of use and long time till failure.



  • @nemesis take a look at the Railcore II and the Voron 2 as examples of modern reprap designs that emphasize quality over cost.



  • @phaedrux Will do 🙂



  • @nemesis said in Z Bed setup with auto calibration, advise requested:

    Would you mind giving me the reasoning behind why you would not use 16mm diameter leadscrews to do this?

    Lead screws are not intended to serve the same purpose as linear guides and the like.

    The diameter is not an issue. I simply wouldn't try to use them as guides.

    Perhaps they can be purchased but I have yet to see a lead screw that was perfectly straight.



  • @fcwilt said in Z Bed setup with auto calibration, advise requested:

    Perhaps they can be purchased but I have yet to see a lead screw that was perfectly straight.

    I've ordered one and will let you know. I have a 200mm SFU1205 lying around (from china) and that thing is definitely straight. I do agree that I have never seen a straight trapezoidal leadscrew (only seen 8mm ones tbh).

    As for a leadscrew not being intended for it I do definitely agree. I might add a guide to the design.

    However a leadscrew is definitely not designed to be mounted on only one end and many printer design rely on that.



  • @nemesis said in Z Bed setup with auto calibration, three leadscrew design:

    However a leadscrew is definitely not designed to be mounted on only one end and many printer design rely on that.

    Two points define a line. So it works fine to have a lead screw setup that way when the bed is mounted using linear guides or the like.

    Clearly if you are going to rely on the lead screws as guides then they will have to be mounted at both ends.

    Frederick



  • i believe the m3d micro uses 4 lead screws in its corners as linear guides. im not sure if they are in any way used for bed leveling tho.



  • @fcwilt It would depend on the rigidity of the bed construction. However it is something that has crossed my mind.

    Thnx for the insights. I won't be able to give a definet answer if it works or not until I get all the parts and that will be a couple of weeks at least. I will post back with the results.

    From my tests with the short SFU1205 I have concluded it has little to no play so it should work fine. However I am working on a lot of assumptions.



  • @antlestxp 4 lead screws would be a option. However the issue with 4 points instead of 3 is that it can be a hassle to get it right and can even lead to torsion in the bed.

    Also not sure if the duet wifi can do 4 point calibration.



  • I would not try to build this without any other support rods at how large a Z chassis you want. The Z-chassis in my large printer is 34x22", it requires 3x 12mm lead screws and 5x 16mm smooth linear rods to keep the chassis perfectly level/rigid, anything less then this gave issues. You will absolutely run into vibration issues at the top end of your design without more support (I realize my example is much larger then what you're making, but will balance on the same principle). 2/3x 12mm smooth rods and 3x 12mm lead screws would do the job great, larger lead screws are just a waste of money unless you're building something giant. It can be difficult to find drive couplers for 12mm lead screws, I'm actually using nema 23's to drive them rather then standard 17's.

    0_1545572625618_IMG_0403.jpg


Log in to reply