Opinion on Core XY belt layout



  • So, I have been working on designing a new machine for a while now. I had already modeled the frame/ enclosures/ wire paths/motion and etc. BUT then e3d released that sneak peak at their tool change design! Now although the tool change is fascinating I found myself liking their idea with the belt routing and hiwin attachment (except the cantileved pulleys of course) I took a look at my frame design based around "L" misumi extrusions and realized I could very easily use twin aluminum plates to make solid pulley mounts similar to e3d.

    The reason for this venture is that all my parts are machined aluminum for this design, My original concept used no fewer than a dozen cnc'd brackets that would need careful tramming. I can do this easily enough but it is a lot of machine setups on the cnc to address all the needs. However, using two plates with pulleys/spacers sandwiched between allows everything to be done in 2 machine setups.

    I started laying things out and came across one problem with the coreXY mechanics and wanted an opinion on if the belts on the carriage can be aligned the way I have them. It is a stacked belt design using 9mm gt2 belt. The issue/concerning area is that the front smooth idlers on the Gantry bar are 13.75mm closer to the center of the machine in the "x" direction compared to the toothed idlers on the backside. All other pulleys line up parallel as well as perpendicular. The axis are driven by motors indirectly (for higher belt tension since this machine should be about 400mmx400mmx650mm build area. )

    Another note, all the pulleys are below all the hiwin rails!
    I rough drew the belt path freehand in paint real quick



  • It's a bit complicated but it might work. You don't need the belts marked in Red. It's acceptable for the belts not to be parallel between the motors in your plan, as long as they are parallel to the Y axis down the sides and the X carriage. It's better to have the belts on the centre line of the X carriage. The way you have them in the diagram, either side if the rail, will impart a twisting force to the carriage. A lot of people do it that way but it's not correct IMO. So ideally, the idlers either side of the x carriage rail should be moved inwards toward the rail.

    Edit. So you probably need to move all the pulley above the rails, so that you can attached to belt ends to carriage (preferably the centre line of the carriage).



  • This look a lite bit like the E3D Tool Changer, I suggest checking it out.



  • @dragonn Well a lot of the design was based around the e3d tool change design. The main difference being that the pulleys are sandwiched between two plates rather than being cantilevered. Toother and smooth idlers. And the stepper motors are separated from main drivetrain.

    @deckingman Awesome to see you chime in here! With the way my carriage is attached to the X-rail the belts below the hiwin rail are actually center to the carriage. (Y-rails are upside down while the x rail is right side up) The belts marked in red are going from a secondary pulley to the motor so that the motor is separated from the main belt runs, allowing higher tension without loading the stepper motor radially(probably overkill but wanted to be sure I could achieve proper tensions with the large build area and 9mm belts)
    I haven't redrawn the carriage but it is 1/8" aluminum plate on either side with 30mm spacing in between. So, in my current layout the belts attach to the outside of the carriage on either side. Hmmm, I wonder how much effort it would take to re-align the pulleys to be center of the rail? I've just always used the pulleys to the front and back since my first machines were all C-BOT variants. I am just concerned with how the belts would attached to the carriage plates like that.

    To mount it, I would have to increase the offset between the pulleys in the "x" direction so as to allow room for the aluminum spacers between the pulleys and aluminum plates.

    Oh, also not shown is that the two pulleys in the very front are mounted in a slide track with a forward tension bolt that allows them to be precisely tuned for tension by threading the bolt in/out (rather than the C-BOT method of grabbing the belt on the carriage and pulling it tight with pliers/etc)



  • @zakfarias.

    Thanks for the explanation. I understand better now.

    Ref the red belts and "remote" motor - as you say, probably overkill but the point you make about side loading the stepper bearings is valid so yes, why not? Just a word of caution about belt tension though - it really doesn't need to be too tight. As long as the belts don't "slap" when printing. Too high a tension just increases friction on all the idlers and bearings making more work for the steppers, and generally increases stress on everything. Also the wider the belt, the more tension needs to be applied. I use 6mm belts and throw my 4Kg mass around just fine so 9mm may be overkill but there is no harm in it.

    Absolutely agree about having idler axels constrained both top and bottom.

    Ref getting the belts on the centre, take a look at my printer here https://somei3deas.wordpress.com/my-corexy-printer-build/. The 6th picture down. The belts are only offset 3mm apart, 1.5mm either side of the centre line and they are clamped either side of a 3mm wide pillar on the X carriage which keeps everything nice and parallel. You can just about make out the clamping arrangement on the pictures further down the page. I printed my own spacers which are a lot thinner than the aluminium ones. TBH, once the belts are tensioned, the idlers find their own position along the axel so spacers aren't really needed at all, but they do make it easier to fit the belts in the first instance.

    The sliding tracks for the front idlers are fine - that's how I tension the belts except that my motors are where your front idlers are. The first picture on this page shows the general layout https://somei3deas.wordpress.com/my-corexy-printer-build/. The upper and lower CoreXY gantries are identical except that the upper one is wider to facilitate the wider extruder carriage.



  • Just had an idea while eating breakfast at the diner. I wonder how bad acceleration, jerk, and overall speed would be affected going to two "cars" on the y axis and two hiwin rails on the "x" I mean in reality I don't need high print speed since I primarily print PETG, Polycarb, and ABS; I just like maintaining the capability.

    If I did that I could actually clamp belts on center very similiar to your machine (which is top 3 favorite builds just saying) Currently My setup has the "X" rail bolted to a 1/4" thick 6061 plate that is bolted to the "y" rails. I work in a tool room as a mold designer/CNC programmer so the plan is to cut 2 parallel 1/16" deep groove on the backside of the aluminum along the entire X-axis and PUK weld in 3/16" rods to act as gusset ribs. With my current setup it would be a classic single rail with the hotend hanging off the front (e3d v6 driven by a nimble with a quick swap option to volcano) With the gussets added I figured that would make the gantry stiff enough but now I am wondering about dual rails instead with the hotend mounted on a plate in between like yours (except linear guides instead of v-slot.) I would still have the 1/4" plate with gussets underneath, just going to two rails instead of 1 as well as two cars on the "Y" to spread the load on the aluminum "x" gantry. Side thought this would also allow me to swap back and forth between the bondtech bmg and nimble as well.

    End goal of this machine is to replace a highly frankensteined and worn out D-bot that acts as my main large format machine while all my other printerss are smaller scale for small productions. (small hobbyish business) That being said, high surface finish and accuracy is the ultimate goal. Speed is an afterthought since my speed is limited by filament anyways.



  • I'd always go with dual X rails. I tried a single one once with the hot end "hanging over the side". Hated it. No matter how I did it, it was possible to "flex" the hot end by applying finger pressure to the nozzle. It's just so much more rigid being slung between 2 rails. I can still get up to 300mm/sec print speed by employing multiple melt chambers even when I'm throwing the 5 extruders around as well as a massively heavy Diamond hot end (4kg in Y direction) https://www.youtube.com/watch?v=-HhxSiv5ajs&t=35s and https://www.youtube.com/watch?v=NAFd3Hj9Wmc

    As a general rule, it's always good engineering practice to reduce moving mass but personally, I think people are too fixated about it when it comes to 3D printing and the speeds and accelerations at which we can reasonably melt and lay down filament. An added bonus to more mass is that you don't get "ringing". It just doesn't happen.

    Edit. I envy you your access to metal machining facilities. I don't have that, hence the open build V slot and all printed parts.


  • administrators

    @deckingman:

    I'd always go with dual X rails…

    Or perhaps one wide one, like E3D is planning for the Z axis of their motion system?



  • @dc42:

    @deckingman:

    I'd always go with dual X rails…

    Or perhaps one wide one, like E3D is planning for the Z axis of their motion system?

    Not ideal IMO as the hot end would have to "hang over the side" which leads to a less rigid mount and shifts the centre of gravity. Tried it myself and hated it so went back to duel rails. Having said that, I hate the way that E3D hot ends are mounted in any case. Fixed at the very to the top of the heat sink with the rest hanging down and able to flex. Is the screw mount better than that God awful groove mount?



  • Well I redrew everything as was discussed above, and I am starting to fall in love with the layout

    In order to fit the pulleys into the design with a gantry plate bolted to the top of two hiwin rails with 4 cars on "X", I did have to switch back to 6mm belts. Oh well.

    For fun I am going to try the e3d threaded heatsink mounted to a small triangular aluminum plate which is then bolted to the gantry plate with spacers. This should allow quick swaps between the v6 and volcano as well as moderate adjustments on heatsink height by swapping spacers.

    However, a new problem arises as I lay everything out back into my frame (roughly). The heatsink is kind of split, 2/3 of it below the gantry plate (3/16 thick 6061) and the other third is above the plate. It looks pretty but cooling is looking to be a bit of a pain since the machine is enclosed(hot) with there being a chance of active heating (currently planning for passive heated chamber). I could make a light duty enclosure box to mount to top of gantry plate (pretty and allows a fan to force air over everything) but still only contacting a third of the heatsink. Pretty sure that won't be enough

    Instead I could also drop the heatsink lower and have the cooling set up under the gantry plate and between the linear guides. However, I dont like the swaying down there or the risk of cooling the nozzle. Water cooling would be great, but out of all of our machines at work our lathe (due to age and wear) is more of a lathe like object and can't be trusted with tight tolerances.

    One thing I really do like about this design (dual rails with belts on center) and how it is wedged into the frame is that I should be able to easily run drag chains now (light duty). This allows me to enclose the top of the machine far easier than I was anticipating otherwise.

    @DC42

    Quick side note: The Z stage is currently 3 12mm rods for guidance with 3 four start 2mm leadscrews (openbuilds) Since it seems the true auto leveling has evolved tremendously, I will run all three screws independently with 40t on the motors (.9 degree) and 20t on the leads. This should give me the proper gearing for 0.01mm resolution, I believe? I would run straight 4:1 ratio with 1.8 degree motors but I can't get a clamping style pulley(which is also GT3) for the leadscrews smaller than 20t and an equivalent GT3 style motor pulley maxes at 60. Say worst case scenario, the bed is 10-12kg and the motors are setup as so…is it enough? Motors are 17HM19-2004S.



  • Ref the hot end - it sounds to me like a case for liquid cooling https://e3d-online.com/titan-aqua-hotend-and-extruder

    Ref lead screws. Multi start screws for Z axis are one of my personal pet hates but let's not go into that, except to say that if you used a finer lead, then you could do away with gearing.

    Anyway…

    You've got the gearing the wrong way round. You want the smaller (20T) pulley on the motor and the larger (40T) pulley on the screw.

    So 2mm pitch 4 start is 8mm lead. That is to say, 1 full turn is 8mm. You should aim for multiples of full steps for layer height. Now with 2:1 gearing (20T on the motor and 40T on the screw), one full turn of the motor will move the screw half a turn so give 4mm lead. Using 0.9 degree motors, you get 400 full steps per rev so 400 full steps = 4mm therefore 0.1mm would be 10 full steps. Therefore you can use any later height that is a multiple of 0.01mm.

    If you used 1.8 degree motors, you'd have 5 full steps per 0.1mm which is fine unless you wanted to use layer heights that aren't multiples of 0.02mm (1 full step). So 0.14mm would be fine as it's 7 full steps, but 0.15 is not so good as it would be 7.5 full steps and you'd be relying on micro stepping for positional accuracy. But all the "normal" layer heights (i.e 0.1, 0.2, 0.3 0.4 etc) would all be fine with 1.8 degree motors and 2:1 gearing.

    Edit. Ref torque requirement. If it's any help I use a single Nema 17 driving three 1mm lead screws via a continuous belt which lifts a 400mm x 400mm x 10mm thick aluminium tool plate, mounted on a 2020 frame and topped with 6mm glass. Can't remember how much that weight though.

    HTH



  • I've looked at the aqua. If I had a plate spaced below the top gantry plate then I could use it,(so that nozzle sits below gantry rails.) However, mounting it has me a bit concerned.

    On the lead screws, I find it hard to source reliable screws which is why I have always settle with using openbuilds. ( that and I have several of their machines) However, their screws steps/5mm are always something like 1 percent off on all builds. Which would mean if need to calculate that into gearing. Hmm, may be better off trying to source some 2 start 1mm leads.

    On gearing. Haha duh blonde moment here on me!

    Planning weight on gantry (including x rails and braces) is stretching into 3-4kg territory.



  • @zakfarias:

    On the lead screws, I find it hard to source reliable screws which is why I have always settle with using openbuilds.

    Yes, I'm a huge fan of Open Builds - but not their screws. Amazon is you friend. That's where I got my 1mm lead (1mm pitch single start) trapezoidal screws from https://www.amazon.co.uk/s/ref=nb_sb_ss_i_4_14?url=search-alias%3Daps&field-keywords=8mm+lead+screw+1mm+pitch&sprefix=8mm+lead+screw%2Caps%2C144&crid=24TH7CG3NH1CY&rh=i%3Aaps%2Ck%3A8mm+lead+screw+1mm+pitch



  • In US I see CNC INC on Amazon has some 800mm long 1mm lead, assuming single start. I already have the 0.9 degree motors. Those would put me at 6400 steps/mm though. Probably should plan to buy more 1.8 degree steppers instead.

    Maybe use 0.9 degree for XY on one of the lighter C-BOT variants.

    Still musing over how to mount an Aqua if I go that route. It doesn't look user friendly for someone mounting to aluminum plates/my current revision of the gantry. I'll have to post the newest version here later tonight for musing.



  • Okay so back on the topic of the belt system,

    I have hit a minor/major problem. Minor problem with a major fix. The machine has gotten a little ambitious with build volume for certain parts on the agenda. As in,I am going to run out of filament long before I run out of build volume on some prints, especially if I were to run volcano at times. That and when running any kind of need for support material I would be forced to run the same material as support, creating a lot of cleanup and waste.
    So…
    1.)convert to cartesian IDEX
    2.)Try to butcher together a CoreXYU IDEX and probably fail miserably
    3.)Dual extruder on the main hotend (blahhhhhhhh)
    4.)diamond...unfamiliar to me and would require setup like @deckingman or bowden (and I reallllllyyyyy suck at tuning a bowden since my previous bowden builds have all required 600mm+ bowden tubes)
    5.) Run the titan aqua as the "main hotend on the plate and come up with some sort of system to raise and lower a secondary bowden style or nimble based hotend that runs support material only. Mayhaps a servo driven e3d-v6 mounted on the same gantry plate as the AQUA?

    I really would like to run IDEX but the motion system starts to become a headache for me if wanting to maintain coreXY



  • @zakfarias:

    In US I see CNC INC on Amazon has some 800mm long 1mm lead, assuming single start. I already have the 0.9 degree motors. Those would put me at 6400 steps/mm though. Probably should plan to buy more 1.8 degree steppers instead.

    Maybe use 0.9 degree for XY on one of the lighter C-BOT variants.

    Still musing over how to mount an Aqua if I go that route. It doesn't look user friendly for someone mounting to aluminum plates/my current revision of the gantry. I'll have to post the newest version here later tonight for musing.

    On my Dbot I went with 3 belted 1mm lead single start screws and decided for speed and torque and to limit steps/mm I went with 1.8 deg motors and I'm glad I did. Even so I have to run fairly high current on the Z motor to prevent binding at the very low end close to the belt. Mind you it's a very heavy bed.



  • @zakfarias:

    I have hit a minor/major problem. Minor problem with a major fix. The machine has gotten a little ambitious with build volume for certain parts on the agenda. As in,I am going to run out of filament long before I run out of build volume on some prints, especially if I were to run volcano at times.

    The options as I see it, in order of complexity - least to most.

    1. Buy filament in 2.3kg spools. If that's not enough then…

    2. Change filament "on the fly". This is something I've done on numerous occasions, the first time had me in a cold sweat but now I'm quite relaxed about it. What I mean is this. With a Titan extruder, there is a short piece of PTFE tube sticking out of the top. When the old filament is just about to run out, cut it just above the tube (so that you don't get the kinked end coming off the spool). Take the new filament and when the old filament has disappeared inside the tube by about 5mm, push the new filament in. Maintain light downward pressure so that the two ends are always in contact. Keep the pressure up until you are sure that the hobbed bolt has got hold of it and that any retraction won't pop it out.

    3. Use pause and resume. Not something that I've tried personally but it should work. Pause the print, chnage the filament, then resume.

    4. Use "Y splitter". As above but more automated.

    5. IDEX

    6. A multi input hot end such as a Diamond or multiple hot ends like the Cyclops or Kraken. Actually this might be less complex than IDEX if you don't mind long Bowden tubes. Personally I hate them which is why I have a second CoreXY stacked above the first taking the extruders. I'd say dual CoreXY is probably a bit more complex than IDEX but it gives you more print area. Anyway, a 3 input Diamond would give you 3 x 1kg or 3 x 2.3kg and with a 5 input you could get up to 5 x 2.3 = 11.5Kg of loaded filament (ought to be enough). You'd set the mixing ratio to use equal amounts of each filament.

    Multiple hot ends would allow you to print support material separately to the main material. This is also possible with the Diamond and easy if you choose materials with similar temperatures (e.g E3D edge + Scaffold). If you have filaments with wildly different temperatures, then IDEX might be better.

    If you are likely to use large nozzles, I would give serious thought to using a Diamond. The multiple melt chambers allow you much higher melts rates an dthus high(ish) print speeds even with large nozzles. The down side is that you can't just change nozzles - you have to chnage the main brass part and for that to be practical it means having a number of complete hot end assemblies which gets a bit expensive.

    7. Tool changing - i.e swapping between "parked extruders". Definitely the most complex.

    Those are just my thoughts…..........



  • I have used pause resume without problems when I have caught it. I am considering some future load cell based filament run out detection and warning, maybe combined with a Y splitter, ambitions are nothing without the time, willpower and resources to see it through though 😉


  • administrators

    We're doing a small production run of the Duet3D laser filament monitor.



  • Count me in for one or ten David!


Locked
 

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