Core XY initial design

[Michael Hackney]

All, I've spent several weeks researching and mocking up ideas for my 300x600x500(ish) CoreXY printer. I decided to use the SpeedDemon SG linear rails and blocks as I've used these on an all-metal Rostock Max build and they are really nice and come in various dimensions to suit the application. Before getting into the details, in addition to building many delta and simple Cartesian printers, I have converted several milling machines to CNC and fabricated 2 large CNC routers. I have a good understand of "rigidity" and motion control.

I'm going with 30x30 extrusions (slotted Misumi) for this build rather than rectangular 20x40. Note that I've left off all the reinforcing and corner attachment details but the frame will be rigid when done. 

On this machine, X is short and Y is long. X will be a SD SG20 rail which is 60mm wide x 22.5mm high. It sits on top of an extrusion that couple the blocks on the Y linear rails. These Y rails are the SG10 and are 28mm wide x 14 mm high. They are mounted UNDER the frame cross bars with the sliding block facing down as shown. The Z rails are also SG20s and mounted as shown. I'm considering milling a slot into the extrusion cross pieces to recess the Z rails by 10-15mm just to clean-up the appearance a little. I'll use 2 rails. The screws are 1204 single start screws commonly used. They are mounted offset from the rails a bit. I understand about torque with offset screws, etc but in this application with these wide rails I think this will be a workable arrangement. My big milling machine has a big screw mounted a few inches off center from the dovetail ways and has to lift a 2 HP motor and auto tool changer and performs well. This is a miniature version of that. I'll use 2 steppers likely mounted up top to drive them with the opposite ends floating.

Not shown is the table base that sits on top of these. The grey sheet is the working X-Y I'm designing for. The support for the bed will likely be full width this L extrusions mounted to the blocks on the Z rails with a 3 point leveling system. I'm going back and forth between melamine or MIC 6 for the table base. I could mill holes in the MIC6 to lighten it and am leaning towards that just because Melamine is more sensitive to humidity changes.

This is a rough drawing of what I'm thinking.

I haven't come across anyone using the SpeedDemon rails for a Core XY yet but I tried to calculate clearances based on other printers with linear rails. 

In addition to feedback on the basic idea and some of the details, does anyone have experience with .9 degree steppers on these printers?

cheers,

Michael

[William Cook]

I really like the look of those SpeedDemon rails. Where are you getting them?

[Michael Hackney]

Automation4Less.com in the US. They list other distributors on their web site (link above). 

[William Cook]

What sort of prices are you seeing for the cars and rails?

You are causing me to completely re-think my CoreXY design, and it's a good thing.

[Michael Hackney]

Pricing is here: http://www.automation4less.com/store/categories.asp?cat=1196

The blocks (3 roller) are around $47 each. Rails are cut to length. So the 692mm long Y rails in my design would be $49 each (all prices US). 

[William Cook]

I like your design...it seems to solve a couple of problems. Are the cars sufficiently rigid to keep the bed support from racking under the weight of the print or other forces?

What do you see as the advantage of the SG system over the OSG system? Are you worried at all about any sort of binding issues sith the two rails opposing each other?

[Michael Hackney]

Yes they are very rigid and you can read the torque specs on their website. 

The SG provides a lower profile and wider rail separation. Also the OSG is a 2 part rail and I didn't want to mess with that. 

As for 2 parallel rails, both of my CNC routers are based on that design and there is no problem aligning the linear rails to be parallel. The PDF tech document on their site even has a section on how to insure the rails are parallel when mounting them.

[William Cook]

OK thank you. I really like those rails a lot better than the Misumi MGN-type rails. Having the separate guide rails would seem to eliminate possible issues with thermal distortion, since it's like a built-in expansion joint. And it looks like you can easily remove the cars without running the risk of having little bitty ball-bearings go all over the place.

You may now return to your regularly scheduled thread, which I sort of hijacked (sorry).

[Michael Hackney]

I used the Misumi MGN rails on a mini kossel and one of my routers. Nice rails for the purpose but I really like these SG rails in the wide format for Z and X and the overall adjustability, quietness, etc. Yes the blocks come separate and there are no balls to loose. Adjusting them is dead simple too, something you can't easily do with bearing based blocks.

Cheers,

Michael

[William Cook]

FYI I checked with LM76 and they are happy to supply STEP files for the blocks and rails.

[Michael Hackney]

Great, I need to grab them. The company is also based near me in Massachusetts and I've talked to their design engineer a few times on my other projects. They are great to work with.

[Michael Hackney]

Here is a design update with refinements to accommodate the mechanics and using the Speed Demon STEP files.

Now that I'm adding in the pulleys and belts I see my primary design challenge - the "crossing of the belts" at the left side in the  drawings. With the linear rail at that position, it interferes with a clean belt path. Options are:

1. use parallel belts and simple run them behind the linear rail under the top extrusion cross piece
2. cross the belts in front of the linear rail
3. others???

I'm leaning towards 1.

The 305x610 plate is the work envelope footprint. The goal is to get a full 12" x 24" foot print. This update now takes the width of the rails/blocks into account for the X and Y max travels. I have some other design elements to think about. The drawings do not show the offset of the X/Y footprint to accommodate the width of the X carriage (60mm) but I did take that into account (plus the extra for the hot end) when I laid this version out. *I can finagle things around once I figure out how I'lll route the belts.

I also need to refine the Z axis dimensions so I can order the rails and screws. I'm not so hung up on having a big Z but since I'm building this from scratch, may as well go for it!

Feedback on the belts and other elements greatly appreciated.

cheers

Michael

[DonaldJ]

Crossing the belts in front of the linear rail should be no problem; they don't require a lot of room.  The build platform won't hit them, unless your hot end (and fans) are flush with the bottom of the X rail.

I had a similar dilemma when building my CoreXY machine a few months back,  but I gave myself lots of extra room around the build plate.  Although only  12" x 18" (X/Y) the Y rails are more than 21" apart.  You can see it here: http://www.thingiverse.com/thing:1014884

[Michael Hackney]

Thanks Donald, that's some machine you built! I don't think I've ever seen a better use for plastic pipe.

[Michael Hackney]

I am considering changing the X axis rail/block from the SG20 to SG15. The SG20 is 60mm wide and the SG15 is 47mm so I would gain some room along Y but more importantly, the overall mass of the X axis would be a bit less. 

[Steve Johnstone]

Wow, looks great Michael!

[Michael Hackney]

Thanks Steve.

Here's iteration 3 with some refinements. This has truncated rails for Z so the belts can pass behind. I still have 19.5" of Z travel. I also routed the belt and I'm happy with the path. Added an E3D V6 hot end just to see how it would look in proportion and to check my Z travel. I have not changed the X rail to the SG15 yet, I'll do that next.

I'm starting to like this design. The inset cross pieces that the top of the Z rails attach to seem like a good idea. All the steppers are up top so enclosing for a heated chamber should be straight forward. 

Next I'll try the SG15 X rail and start to work on the stepper and idler mounts. I'm not one to design to perfection and then build. I like to work out the design and leave a little wiggle room so I can tweak things to fit as I build. 

[Jetguy]

The Z axis violates a number of engineering practices. http://www.igus.com/wpck/8107/5_linear_bearing_mistakes_you_do_not_know_you_are_making?C=US&L=en
First one being rise over run. Your vertical bearing distance in the rail compared to the width of the Z stage (yes, even the narrow edge, not the long side) is off.
Part 2 is if you are going to put the Z leadscrews like that, with improper bearing spacing, then you preload the Z stage in a rocked state. Remember, the distance amplifies the angular error.
You might only get minimal deflection at the bearing (well with the expected tolerance) only to amplify it at the outer corners of the bed.

1. The bearings are not spaced far enough apart relative to the cantilevered load or drive force.

Symptom: The bearing sets chatter as they slide in one or both directions of travel (when carrying a cantilevered load) or, the bearings do not move at all (when moved by a highly offset drive force).

1.2 It is important that all acting forces follow the 2:1 rule.

I see far to many designs (Gigabot is the worst offender) selling an overpriced machine with this type of engineering.
Huge bed with no bearing spacing on a centered setup not unlike yours.

Your XY is fine, just fix the Z. It's a crying shame to have a good XY and slop out on Z.

Again, pick any corner of the bed, set a 1/4 pound weight (something you could easily print in that corner volume) and note the deflection of both that corner and the opposite side of the bed.
Back to why a 3-4 leadscrew system can handle the large mass with minimal deflection of realistic print loads- let alone dynamic vibration loads during printing which can be magnitudes greater than static.

[Jetguy]

Also, a point of Core XY is that the belts eliminate this rule for the XY.
The belts act as a huge brace or bearing preventing the gantry from exerting an angular load to the linear bearings.

H-bot has to follow this rise over run rule because it imposes forces across the gantry from side to side (nobody does it though- especially not MakerBot 5th gen).
Ultimately, what would happen is a huge frame and a tiny build axis with a properly engineered H-bot.
Hence why the Mojo is a large printer with a tiny 5 inch XYZ build axis because it is H-bot, but with proper rise over run calcs to prevent the gantry position error- well, that and slow speed.

[Michael Hackney]

Thank you very much Jetguy. What would you recommend as the best option for Z for this machine? I thought a 2 screw/2 rail design would minimize issues with synchronizing the screws. After reading the igus page, I'm still don't understand the relationship between the screw/s and linear guide/s. In a situation like this, is the goal to spread the guides apart to the corners and then use a single screw between them? Or vice versa?

cheers,

Michael

On Tuesday, March 8, 2016 at 7:47:58 PM UTC-5, Jetguy wrote:

[Jetguy]

Sorry, I did a poor job of explaining. What you need is either one really long carriage per side, or use 2 shorter ones spaced vertically, then basically triangulate out a plate to stiffen the ends.

For example, you could use a 3 bearing block for the ends of the gantry of the X Y. This is because there is no angular load.


For your current method of Z axis, minimally you want the 5 bearing option but even better is 2 4 bearing units spaced vertically in the rail.



This is an end view

[Michael Hackney]

Very helpful, thank you. So I need to follow the 2:1 rule. I can do some calculations with the various rails widths and one 5 roller or two 4 roller carriages to achieve that. The tradeoff is the increased height of the machine. I can live with that and even reduce my working Z to keep the overall machine height reasonable. Are there alternatives for Z that are more compact vertically?

[Michael Hackney]

I did not see the drawings you made Jetguy, they were collapsed and didn't show up at first. So in parallel I drew up some options using the various carriage sizes. My target printable width is 305mm. I added 25mm for a little space for hold downs to get a width of 330mm. Here's the layout I came up with.

The rise of 82.5mm to 165mm gives me the target 1:2 ratio.The red lines are the 4 and 5 roller SG10 blocks, purple are the 4 and 5 roller SG15 and the blue are 4 and 5 roller SG20. For comparison, the green is the 3 roller SG20 I used in my original design. Of the options, it seems like I could either downsize the real to SG15 and use a 5 roller carriage for a single carriage solution or use two 4 roller carriages and scale down to the SG10 rails. I think the single 5 roller SG15 (the taller purple rectangle) should give a little extra support. Does it matter if the angled brace is as shown or upside down? It seems that shouldn't make a difference.

[Michael Hackney]

I forgot to mention, the design had a much wider sub-bed than is actually needed so I decreased the width to 330mm as a more realistic width.

[William Cook]

Jetguy, what is your preferred Z system? I seem to recall you prefer a 3-leadscrew solution with two linear rails, and from what pictures I can find that appears to be what's on the Big Boy.

Are any of your printer builds documented somewhere? Or at least pictured in detail somewhere?

[Michael Hackney]

William, a quick search turned up this: http://thegreatfredini.com/2015/05/05/first-pri-t-on-jetguys-big-boy-corexy-3d-printer/

Photo of a VERY IMPRESSIVE 1st layer on the full bed.

[Michael Hackney]

And this: http://thegreatfredini.com/2015/05/02/the-maestro-of-3d-printer-building-jetguy/

Ok, I'm completely reconsidering my Z axis now. I had started down the path of a 3 screw design and then abandoned it because my Yankee gene kicked in urging me to reduce the size and overall complexity of the machine. But now that I see these photos and the "form follows function" of this design, I'm intrigued. My major question (and this is why I started rethinking myself out of the 3 screw design) is I have a 2-1 rectangular shaped bed, not a square. So the 3 points support of the screws is not an isosceles triangle. I was concerned that the corners opposite the rails (on the corners) and 2 screws would be unsupported. Any recommendations on this Jetguy? Would it be better to put the rails on the short edge or long edge in this configuration?

[William Cook]

Yes, I was just looking at that. From the pictures it appears that he has two MGN-type rails for the Z axis mounted on the back face/interior slot of the 2040 rail at the front of the machine, each coupled with a lead screw. A third lead screw is at the back center of the machine, and is not paired with a linear rail. So the leadscrew at the back supports the Z entirely, and noe of the leadscrews are fixed at the top end.

Presumably this system would be quite free of any binding or racking issues with the bearings and, if the leadscrews are locked in sync with a belt it would be very stable.

It also seems to be quite efficient in terms of build area/printer area.

So one could certainly use the SpeedDemon rails in a similar fashion

[Michael Hackney]

V2 iteration 1

So I went back to the drawing board with the Z. This version uses SG10 rails everywhere and uses 3 screws. I have lots more detail to work out. Here are the drawings as they are now.

One stepper in the base with belt drive to the 3 screws. Basically the same CoreXY as in the previous versions with some tweaks so I left most of that off to focus on the unknowns. As drawn, the X-Y footprint is 320mm x 610mm so I'll likely add a bit for margin. Now the questions I need to sort out:

1. with the bottom driven screws and to conserve Z, what's the best way to turn the screw ends for the bearing and coupler?
2. it looks like you use smaller diameter screws Jetguy, what type of screw and ballnut are you using? Your ballnuts appear to be mounted inside of the aluminum crosspiece extrusion, is that correct?
3. How should the single screw in the mid rear be connected to the front cross bar. I looked at as many photos of your machines as I could find and there were none from the back. Is it simply a T from the support bar in front? Or something else?

The large grey plate represents the X-Y extents. The orange plate is just for perspective to see how things line up on the cross bar and rear screw mount. I like the short legs at the bottom, should be easier to level and provides rom underneath for electronics, etc. I understand how to route the Z screw drive belt to maximize belt wrap as per a few of Jetguy's photos. That all makes sense.

I show 2 purple hotends simply so I could measure the travels.

I have a bunch of 30x30 Mitsumi extrusions in route for the frame. I'll order the rails and screws once I work out more of the details. I have lots of aluminum plate and extrusions in house to make the other parts (braces, motor/screw mounts, etc) and a CNC mill with a 19"X*10"Y*16"Z work envelope to fabricate the parts. Anything bigger than that can be cut on my 2'x3' CNC router if needed.

[Michael Hackney]

What about a 4 screw system - 1 at each corner - for this machine. That would address my concerns about the unsupported corners. Is 1 or 2 additional linear rails needed with 4 screws? What about using 2 rails diagonally opposite each other?

[Michael Hackney]

Here's a basic design question I have - is one of these two options better than the other for supporting the build plate? The plate will have a 2-1 ratio of side length.

[DonaldJ]

Common sense tells me that the short edge support is better.  Although the supported areas are equal in area, the unsupported distance is less with the short edge support.

[William Cook]

DonaldJ, I would have reasoned exactly the opposite. In terms of plate deflection, I am more concerned about weight in the center of the plate, where it will be most of the time, than I am with it at the corners.

I agree that given identical point loads, the corner in the long edge support case would deflect more than the center in the short edge support case, but the weight will be in the center more of the time and I suspect that is the governing concern.

[Michael Hackney]

William, I understand your concern about possible deflection at the center. My plan is to use a MIC6 bed with PEI on top as the print surface. This will be very stiff so I am not that concerned about sag or deflection. I could even reinforce it longitudinally with aluminum angles if necessary. 

--
You received this message because you are subscribed to a topic in the Google Groups "H-bot and CoreXY 3d printers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/h-bot-and-corexy-3d-printers/sVyTyfYoOwQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to h-bot-and-corexy-3d-...@googlegroups.com.
Visit this group at https://groups.google.com/group/h-bot-and-corexy-3d-printers.

[William Cook]

I understand that, but even so the weight of the plate and prints would seem to me to indicate that the long edge support is more likely to produce a flat bed.

[William Cook]

What size ballscrew/leadscrew were you planning to use?

[Michael Hackney]

William, just to be clear I am investigating/asking questions about several parallel paths...

The recent question about the 3 points of support was actually related to the print bed itself - which will sit on the Z stage whatever that turns out to be. 

The more general question is how best to design my Z stage. After Jetguy's input, I do think a multiscrew system should produce excellent results. The options here are 3 screws as per my updated V2 iteration drawings or a 4 screw version. I'm leaning towards a 4 screw version given the 2-1 aspect ratio of the Z stage. I plane to se 1204 screws (12mm OD) similar to these from RobotDigg: 1204 trapezoidal screws.

I have found a few more photos of Jetguy's BIG CoreXY printer and the Fredini printer. The BIG one uses 4 screws with the rails mounted on diagonal corners. I think this arrangement works best for a stage with a 2-1 aspect like mine. Here's an update 4 screw, 2 diagonal rail rendering:

The MIC6/PEI bed is depicted by the light grey sheet. It would be mounted at 3 points - 2 on the short end and one in the middle of the opposite long end.The orange sheet is just for my visualization of the design. I'd likely just use additional cross braces at the appropriate places for the bed mount points to keep the mass down.

[William Cook]

I like a whole lot of what you've done. I'm running along a parallel path for a machine with a 12"x12" bed and a volume that can scale vertically from 12" up to 36" (I make rocketry parts - they tend to be tall).

I agree that the 4-screw solution is a good one for your setup. I'm probably going to do the 3-screw setup for my smaller bed. When I get something together in my CAD tool I'll post it.

I'd be interested in the links you found.

What is the advantage to running the x-axis track under the extrusion instead of on top of it?

[Michael Hackney]

The all-inclusive list of photos of Jetguy's CoreXY machines (that I've found!):

1. http://thegreatfredini.com/2015/05/02/the-maestro-of-3d-printer-building-jetguy/
   
2. http://thegreatfredini.com/2015/05/05/first-pri-t-on-jetguys-big-boy-corexy-3d-printer/
   
3. https://blog.adafruit.com/2014/09/25/jetguys-ulti-replicator-2/
   
4. https://www.flickr.com/photos/90025904@N04/albums/72157649795398404/with/16636890324/
   
5. https://www.flickr.com/photos/90025904@N04/albums/72157659638963789

I hope I don't come across as a stalker, just a serious researcher!

[William Cook]

I hope I don't come across as a stalker, just a serious researcher!

Ha! 

[William Cook]

There's another machine with a similar Z arrangement here:

http://www.thingiverse.com/thing:612857

[Michael Hackney]

Normally, the ballscrew is inserted into two opposing bearings. Here's the Y axis for one of my CNC mill conversions:

The bearings in this case sit in pockets on the left and right side of the large aluminum block. The end of the screw is threaded and a pair of nuts secure the screw and eliminate backlash. What I've seen on some of the CoreXY Z screws is the use of flanged bearings in a similar way. But like the second link you posted, they seem to only use a single flanged bearing on top and secure the pulley to the shaft below to remove any slop. Here is the description on the thingiverse page:

The turned down ends of the leadscrews sit in the flange bearings. Gravity will keep them well seated. They will not be able to completely pull out as each will have a timing pulley acting as a collar below the build plate.

I'll use two opposing flanged bearings on mine. 

[Michael Hackney]

Hopefully Jetguy will take a look at this, I'd really appreciate his feedback. I've gone back to the drawing board after his analysis of my Z stage. I've seen the  light for a multi screw Z stage that is not over-constrained. I've read every post I could find and scrutinized photos of Jetguy's machines. I did find several posts by Jetguy that list design principles. 1 post was clearly for smaller machines with 3 screws and the other for much larger machines with 4 screws. By combining the key elements I believe this is correct for a 4 screw design:

#1 use supported rails, never round rods.
#2 position the 2 rails as far apart as possible (I pulled this from one of Jetguy's posts and on his big printer it appears that the rails are on the diagonals)
#3 Position the 4 leadscrews close to the corners (paraphrased for the 4 leadscrew design)
#4 Never, ever constrain the top of the leadscrew.

So here is my latest design that I believe addresses all of these principles:

1. I'm using OG10 Speed Demon rails screwed to 30 x 30 aluminum extrusion (frame) - they are fully supported
2. My Z rails are on opposite (diagonal) corners so they are as far apart as possible. This seems to make sense for a rectangular printer like this
3. all lead screws are near the corners and two are as close to the rails as possible
4. the tops of the leadscrews float so as not to transmit leadscrew "wobble"

I'm very near the final design I think and only have a few questions:

1. is principle #2 above correct? It makes sense. The other option is two Z rails on the long edge.
2. how best to connect the stage edges (the dark green parts - which are aluminum angles). In Jetguy's large printer, it looks like two rails on the long edge and then two shorter rails on the short edge. This makes sense but I want to validate it. 

This printer meets my objectives with a 320m x 620mm (conservative) X-Y footprint with 405 mm of Z travel. I might increase the Z travel a bit. There is enough room below the lower crosspieces to mount the Z stepper, power supply, controller, etc with plenty of clearance. 

Cheers,

Michael

[Michael Hackney]

Correction: SG10 Speed Demon rails.

[Jetguy]

It's up to you on that rail placement. We could have a debate about it and some engineer might pop in with some aspect I've not covered.

On a relatively small design (again, relative terms compared to some of my builds) 2 on the long front edge might (probably) would look fine and be more pleasing in some folks minds.

The thing is that you have a very high aspect ratio bed in this design.

The other thing to keep in mind is one side is the guide, the other is "anti-rotation" around the Z plane.

You probably aren't follow the latter to the "T" and neither have I on my designs.

By that, I mean the one side should connected to the Z stage in such a way that it only restricts motion in one specific direction and "floats" in the others.

Technically, by rigid mounting both bearings to the Z stage, we over constrain it. The reality is, I've yet to see this be a problem in our application if you follow an specific assembly pattern.

That is, tighten one rail and ensure it's dead vertical (perpendicular to both X and Y planes). Leave the other rail slightly loose to the frame. Now, lower the Z and tighten the nearest rail to frame mounting screw. Then raise the Z stage to max and tighten the nearest rail mounting screw. Now we know that both rails are dead parallel and not binding.

Also, since I make my machine frame from aluminum and the Z stage is aluminum, then the frame and Z stage should have similar rates of thermal expansion and not bind.

If you go cross corner to cross corner, there is a lot more frame involved, a lot more room for the rails to not be parallel in all direction, more chance that you get a bind from top to bottom as the Z stage moves.

[Michael Hackney]

Thanks Jetguy, greatly appreciated input. As Albert Einstein said "In theory, theory and practice are the same. In practice, they are not." You have demonstrated practice and that trumps theory in my view.

I have what I need to finalize the design. I am going to go with the parallel rails on the long front edge, not diagonally but I'll leave the dimensions of the frame as they are so I can retrofit/experiment with diagonal placement later if needed. On the "rotational constraint" rail, it might be interesting to use a slot in the Z stage cross member (parallel to the long front rail) and a pin through the slot. For now, I'll use KISS and over constrain and carefully assemble the rails as you describe.

This is perfect timing, my 30x30s arrive today. I can finalize the dimensions in my CAD drawing to determine my rail and screw lengths so I can order them while I assemble the frame.

cheers,

Michael

[Michael Hackney]

Jetguy, do you have any recommendations on ballscrew type/size for this machine? And how did you do the bearings and pulley attachment on the lower end. I've scrutinized your photos and can't really find anything with enough detail. I have read that some only use a single flanged bearing and rely on the weight of the stage and the pulley attachment to keep the screw in place. Do you use opposing bearings? 

[Ryan Carlyle]

I don't think "opposite corners" or "same long side" placement of the rails will matter at all to stability. The length of unsupported plate cantilevered out past the rail is nearly identical.

Now, that said, good general practice is to put rails as close to screws as possible. This ensures the XYABC constraint is nearly in the same place as the Z constraint. That minimizes possible flexure modes. (Imagine a rail is five feet from its screw -- it's carriage will be able to move up and down quite a bit with the screw stationary.) So the position of the screws should arguably control the position of the rails.

[Michael Hackney]

Ok, more details clarified:

re: bearings - Ryan provided some info on why a single bearing on top is a good option. Most of the load is downward within reasonable limits. Also, the screw is freeze to move around a bit to compensate for any bending so this is not transmitted to the Z stage. The lower end is turned down for the pulleys, no threads. The pulley acts as a lock color underneath.

Now I have to figure out which screws to use. I am not familiar with trapezoidal screws but I've used ball screws in a number of CNC machine and printer builds. Here are some particulars:

1. I estimate the bed to be somewhere in the range of 4.5 to 5.5 KG (10-12 pounds)
2. I plan to have 520mm of Z travel (~20") so the screws will be in the neighborhood of 600mm OAL
3. The Z stage has four ball screws - one at each corner
4. I'd like to get at least 50 mm/s of Z max speed
5. I plan to use a Duet - which has 16 micro steps and can do 100kHz/step
6. I plan to use NEMA 17 .9° steppers that have a resolution of 400 steps/rev
7. I plan to run the steppers at 24V

I'm leaning towards 1004 ball screws (10mm D, 4 start) but that's just a SWAG.

[Steve Johnstone]

This is such a great post Michael, I'm learning so much.

Thank you Jetguy and Ryan. Your explanations on not over constraining the Z-stage really make sense now.

On the Clone R1 z-stage upgrade I've 4 slides & 2 screws. One of my biggest worries was getting the 4 slides aligned and ended assembling it pretty much as jetguy explained.

I don't think I've got it quite perfect but should do for now. I just need to fit the stepper motor / screws and will definitely not be constraining the tops of the lead screws.

Here's some video of where I'm at - LINK

Also I've just modeled the Duet v0.8.5 with Think 3d's touch screen bundle. Let me know if you would like the models and I'll check with the guys over there if its ok to share.

 

[William Cook]

I would very much like to have the Duet models.

[Michael Hackney]

Now that I have the basics of the frame, CoreXY stage and "general" Z  stage  (i.e. 2 rails on long edge, 4 screws) design locked down I'm trying to apply some discipline in determining what Z screws and steppers to use. The X-Y stage/s are pretty straight forward but Z stage is challenging due to its mass and trying to find workable lead/ball screws.

I am using Ryan Carlyle's stepper motor simulator and configuring for a Duet controller with Allegro A4982 stepper drivers. It looks like with TR8x8-p2 (trapezoidal 8mm, 2 mm pitch, 4 starts, 8mm/turn) screws that seem to be common on smaller (300x300mm) machines driven in 1:1 by the stepper, my max Z will be in the neighborhood of 40mm/s with a 1.8° stepper and half that with a .9° stepper. I use a fast (100mm/s) Z lift to manage PLA stringing on my deltas but have little experience with Cartesian printers. The Bowden on a CoreXY is a lot shorter than a typical delta Bowden and I have the option of direct drive so perhaps I can manage stringing in other ways. I'd appreciate any feedback/insight on this.

I'm also having a bit of difficulty figuring out what to use/what's needed for the Z screws. As previously mentioned, my Z stage is about 4Kg. I have 4 screws (1 at each corner) and the effective screw length is about 500mm. I've found a few online "whip" calculators and with the relatively low speeds according to the stepper simulator, the Tr8 leadscrews should be fine. But I wanted to investigate other options. That's when things get tricky. There are Tr10 leadscrews but they are very difficult to source and the few I found were very expensive. 1002 and 1004 ballscrews (10mm D and either 2 or 4mm lead, single start) are available but they are 1) "slow" meaning that my Z max speed would be 1/4 to 1/2 what I calculated above and 2) also very expensive. The next size up, 1210 ball screws (12mm D, 10 mm lead, single start), are readily available, affordable and result in a bit of extra speed (50mm/s as calculated with Ryan's simulator) but result in 320 microsteps/mm of travel and will probably require a beefier stepper. I could reclaim resolution by moving to a .9° stepper (giving 640 ms/mm of travel) but decreasing max speed by 1/2 in the process.

It looks like the 2 reasonable options are:

1. Tr8x8 with a 1.8° stepper
2. 1210 with a 1.8° stepper

With the Tr8x8 leadscrews, is it standard/best practice to turn down the ends to, say 5mm, to seat on a bearing and provide a smooth shaft to fix the pulley? With the 1210s turned down to 8mm for the bearing and pulley seems reasonable.

Tradeoffs, tradeoffs, tradeoffs. I'm dangerously close to analysis paralysis and need to push forward soon.

[William Cook]

I've been looking at this also, and have to show some ignorance about stepper motors and drivers here. As I understand it, you've sim'd with 8 microsteps/full step. Is there a technical reason why you cant go lower? It would be noisier for sure, but do you need resolution beyond 0.025mm/step?

[Michael Hackney]

The Duet 0.8.5 is fixed at 16 microsteps. 

[Ryan Carlyle]

I think the max speeds you're interpreting from the simulator are probably the "guaranteed safe" speeds (no significant inductor lag) and you have a fair margin of safety to go faster than that. But 50mm/s is not an unusual "safe speed limit" for a screw-based Z stage.

I don't think 320 steps/mm is low Z resolution. In fact, I bet it's higher Z resolution than your deltas.

If you don't use auto-tilt-correction or Z hop, a Cartesian bot stays at perfectly constant Z height through the entire layer. Whereas in a delta (even WITH dc42 non-segmented movement) the nozzle is constantly moving up/down +/- 0.5 microstep relative to the desired XY plane as the kinematic transform rounds each carriage height to the nearest integer step.

320 steps/mm will put you at nice full step increments for all the typical 0.05x layer heights. That means consistent distance between layers, even if the motor/driver system has some microstepping angle error. It's kind of ideal tbh. 10mm lead is a great choice if you want metric layer heights.

[William Cook]

Ah, that is a very good reason. Hmmm. That might be enough reason for me to stay in the RADDS ecosystem.

[Ryan Carlyle]

The 4988s at 1/16 are a good, reliable driver. I'll take a tried-and-true 4988 at 1/16 stepping over a microstep-rippling 8825 at 1/32 stepping ANY day. 

Only real downsides to the 4988s are that you can't get the noise reduction from finer microstepping, and the current doesn't really go high enough for NEMA 23s if you're building something big. 

[Michael Hackney]

Thanks Ryan. As much as I love my deltas (and I've put a lot of effort into them) I am looking forward to the advantages you describe - that's the primary reason for building this CoreXY machine. I'm hoping I can lay down a .10 or .15mm thick first layer 1cm wide by 600cm long!

[William Cook]

The 4988s at 1/16 are a good, reliable driver. I'll take a tried-and-true 4988 at 1/16 stepping over a microstep-rippling 8825 at 1/32 stepping ANY day. 

Only real downsides to the 4988s are that you can't get the noise reduction from finer microstepping, and the current doesn't really go high enough for NEMA 23s if you're building something big. 

I was thinking more in terms of going to 1/8 or so on the Z axis.

And the noise reduction of 1/32 is a significant plus in my environment.

It's certainly not a question of accuracy or resolution. 

[Michael Hackney]

Sourcing lead/ball screws has been a PIA - a major one. My big lathe is down for a motor/speed control upgrade right now so I don't want to turn the ends myself. I hoped to find sources that would machine the ends but cannot. ZenToolWorks has stock Tr8x8 screws that I could probably use but they do not list the specs for the length of the turned down end. They have no phone # and they have not replied to my email. I've used aliexpress and alibaba to source parts for a lot of things in the past and I keep coming up dry trying to source these.Basically, I need a screw that's 600mm overall length with a turned down end that's 50mm long. The end should be 5mm D for 8mm screws or 8mm D for a 12 mm screw. I've attached a basic drawing of what I'd like but I'd settle for Tr8x8 turned down to 5mm (50mm long). It's ridiculously easy to source larger ball screws (16mm or >).

[William Cook]

I've been having the same problem, and was considering going to 1610 ball screws just because you can actually get them.

[Michael Hackney]

Overnight I have found a source for 1204 ballscrews and nuts for about $29 each for 4 from a reputable Chinese company (aliexpress) that I can have in-hand in about 8 days. I also have a quote for HIWIN 1210 ballscrews and nuts for $96 each. Both of these are machined to my specs with a simple 8mm D 50mm long end (on one end) and 550mm of screw for an over all length of 600mm. These are part #2R12-10A1-FSW-650-650-0;05

There is an old saying in car racing "speed costs, how much do you want to spend?"! That seems to apply to ballscrews too. 

I'm tempted to bite the bullet on the HIWINs. I am thinking of using angular contact bearings. The mount will likely be 1/2" aluminum for rigidity with a pocket for the bearing on top. Considering using 32 tooth pulleys so the total length of the turned down end only needs to be 31mm. A 50mm turned end is probably too long - although I can cut it to final length. I think I'll change the specification to 35mm turned down ends.

[Michael Hackney]

A couple of quick things:

1) I realize that aliexpress is not a company, it is a market place. The company with the 1204's is a reputable company based on their aliexpress feedback

2) I've decided to bite the bullet and get the 1210s. In a stroke of divine intervention I received an order for 5 3D printed reels from a Japanese customer that almost exactly covers the cost of these ballscrews. 

3) Now that I have the components sourced I can do the final design tweaks. I'll post it as soon as it's updated.

[William Cook]

Where are you getting the 1210 ball screws?

[Michael Hackney]

Can you PM or email me and I'll give you some details.

[Michael Hackney]

Caution, potentially harebrained idea ahead...

Ok, I've got my ballscrews & nuts, pulleys and bearings ordered for both the CoreXY mechanism and the 4 screw Z stage. I've been thinking about the belt to drive the 4 ballscrews. I know that the 3 and 4 screw Z stages use a continuous belt but I was thinking of a way to use a cut belt. Basically the idea is this:

My ballscrews have a 10mm lead and a working length of 615mm. That means it takes roughly 61 turns for the ball nut to travel the full length of the screw. With the T32 GT2 pulleys I got, they are 25mm OD. For simplicity, I'll use this as the pitch diameter. That means the belt travels:

belt travel/turn = 25mm * Pi = 79mm/turn (rounded up)

That means the belt travels 4791mm for full Z travel. Now back calculating the diameter of a circle with a circumference of 4791mm we get:

diameter = 4791mm/Pi = 1525mm or 152.5cm

Of course, that's if the belt had a full wrap around the pulley but as you can see in the drawing, the belt needs to "come off" this big idler pulley. From my rough calculations along with positioning the smaller idler pulleys it looks like a pulley 180cm in diameter could be used as shown here. The ends of the belt are fit into slots at the perimeter as shown. This allows a cut belt to be used without requiring a joint to produce a continuous loop. This large idler pulley could easily be 3D printed and made very light and even have a built in belt tensioning mechanism. 

Hopefully I described this clearly enough. Is this completely harebrained or potentially good way of using a cut belt to drive the Z stage?

[Michael Hackney]

Note: the small idler to the left of the large idler is not the stepper. The square drawn around it was an artifact (I forgot to disable a layer). Most likely, the stepper would replace one of the two guide idlers to the right. Of course a more optimal belt path that provides more wrap around the pulleys will be possible but I wanted to focus not he core of the concept to see if it makes sense.

[Ryan Carlyle]

Did you draw a 180mm pulley instead of 180cm? It should be bigger than your entire printer.

[Ryan Carlyle]

You could probably imagine wrapping the belt around a multi-wrap helix to get the diameter down... If it doesn't get too tall...

[Michael Hackney]

"off by 10". Yes, I did. I thought of this driving home from NY last night (6hr drive) and then wrote it up before sleeping on it. 

4.8 meters of belt travel is probably a non-starter for this idea.

Back to our regularly scheduled build thread...

[Nick Foley]

Having built a similarly proportioned & designed CoreXY to the one that started this thread (~250X x ~400Y x ~400Z, a somewhat accurate 3D model: https://cad.onshape.com/documents/6fe81cf35e2740b3bbde9aa6 ), I would say that (at least with wheels on rails) the practical simplicity of the two-screw Z-stage outweighed the theoretical cantilever problems. I was worried that the Z-stage would be pre-loaded strangely, or have too much flex as a result of the two-screw / center rail system, but in practice, the amount of Z-planar flex introduced by Z-stage motion system in an asymmetrically-loaded-bed-scenario is probably 1/10th the amount of flex that is enabled by my 3-screw sprung bed leveling (which is nonetheless relatively stiff & precise).

I built my machine with enough Z-height that I could add a taller bearing block for the Z-stage if it became a problem, but based on initial performance, it seems that will be an unlikely change. While the printer is still new, it has been churning out large, immaculate prints for a few hundred hours at this point, and everything in the motion system still feels exactly like it did when I started. 

On Tuesday, March 8, 2016 at 7:47:58 PM UTC-5, Jetguy wrote:

The Z axis violates a number of engineering practices. http://www.igus.com/wpck/8107/5_linear_bearing_mistakes_you_do_not_know_you_are_making?C=US&L=en
First one being rise over run. Your vertical bearing distance in the rail compared to the width of the Z stage (yes, even the narrow edge, not the long side) is off.
Part 2 is if you are going to put the Z leadscrews like that, with improper bearing spacing, then you preload the Z stage in a rocked state. Remember, the distance amplifies the angular error.
You might only get minimal deflection at the bearing (well with the expected tolerance) only to amplify it at the outer corners of the bed.

1. The bearings are not spaced far enough apart relative to the cantilevered load or drive force.

Symptom: The bearing sets chatter as they slide in one or both directions of travel (when carrying a cantilevered load) or, the bearings do not move at all (when moved by a highly offset drive force).

1.2 It is important that all acting forces follow the 2:1 rule.

I see far to many designs (Gigabot is the worst offender) selling an overpriced machine with this type of engineering.
Huge bed with no bearing spacing on a centered setup not unlike yours.

Your XY is fine, just fix the Z. It's a crying shame to have a good XY and slop out on Z.

Again, pick any corner of the bed, set a 1/4 pound weight (something you could easily print in that corner volume) and note the deflection of both that corner and the opposite side of the bed.
Back to why a 3-4 leadscrew system can handle the large mass with minimal deflection of realistic print loads- let alone dynamic vibration loads during printing which can be magnitudes greater than static.

On Tuesday, March 8, 2016 at 6:33:56 PM UTC-5, Michael Hackney wrote:

Thanks Steve.

Here's iteration 3 with some refinements. This has truncated rails for Z so the belts can pass behind. I still have 19.5" of Z travel. I also routed the belt and I'm happy with the path. Added an E3D V6 hot end just to see how it would look in proportion and to check my Z travel. I have not changed the X rail to the SG15 yet, I'll do that next.

I'm starting to like this design. The inset cross pieces that the top of the Z rails attach to seem like a good idea. All the steppers are up top so enclosing for a heated chamber should be straight forward. 

Next I'll try the SG15 X rail and start to work on the stepper and idler mounts. I'm not one to design to perfection and then build. I like to work out the design and leave a little wiggle room so I can tweak things to fit as I build. 

[Michael Hackney]

Thanks Nick. A couple of questions if I may...

How much does your bed/Z stage platter weigh?

How tall are your bearing blocks and what is the construction of them? (i.e. shoulder on screw, bearing, block, thrust washer/washer/bearing?, etc)

I am mostly concerned with the first layer. I need to be able to print a 24" long x .5" first layer .15mm to .20 mm thick. flawlessly and constantly (85%+ success). 

[Nick Foley]

Not totally sure about the weight - key items are two 500mm lengths of 20x20 V-slot & a 10" x 16" x 3/16" aluminum plate (from Online Metals) for the heated bed. I had been planning on adding glass on top of that, but the aluminum has proven to be flat enough even when heated to reliably take the first layer.

Bearing blocks are these: http://openbuildspartstore.com/mini-v-gantry-set/

--
You received this message because you are subscribed to a topic in the Google Groups "H-bot and CoreXY 3d printers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/h-bot-and-corexy-3d-printers/sVyTyfYoOwQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to h-bot-and-corexy-3d-...@googlegroups.com.
Visit this group at https://groups.google.com/group/h-bot-and-corexy-3d-printers.

--

Nick

www.nhfoley.com

 

[Michael Hackney]

Thanks, that's enough to compare with mine. I'm using 1/4" MIC6 aluminum quite a bit larger, no glass and two 12"x12" silicone heaters. What are you using for heaters? I'll also has PEI on top of the aluminum.

[Nick Foley]

Heater:

http://www.ultibots.com/24v-kapton-heater-240mm-x-310mm/

Also using PEI.

--
You received this message because you are subscribed to a topic in the Google Groups "H-bot and CoreXY 3d printers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/h-bot-and-corexy-3d-printers/sVyTyfYoOwQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to h-bot-and-corexy-3d-...@googlegroups.com.
Visit this group at https://groups.google.com/group/h-bot-and-corexy-3d-printers.

--

Nick

www.nhfoley.com

 

[Michael Hackney]

I presume you are using 2 of these?

On Sunday, April 3, 2016 at 5:44:25 PM UTC-4, Nick Foley wrote:

Heater:

http://www.ultibots.com/24v-kapton-heater-240mm-x-310mm/

Also using PEI.

On Sun, Apr 3, 2016 at 5:41 PM, Michael Hackney <mhac...@eclecticangler.com> wrote:

Thanks, that's enough to compare with mine. I'm using 1/4" MIC6 aluminum quite a bit larger, no glass and two 12"x12" silicone heaters. What are you using for heaters? I'll also has PEI on top of the aluminum.

On Sunday, April 3, 2016 at 5:38:44 PM UTC-4, Nick Foley wrote:

Not totally sure about the weight - key items are two 500mm lengths of 20x20 V-slot & a 10" x 16" x 3/16" aluminum plate (from Online Metals) for the heated bed. I had been planning on adding glass on top of that, but the aluminum has proven to be flat enough even when heated to reliably take the first layer.

Bearing blocks are these: http://openbuildspartstore.com/mini-v-gantry-set/

--
You received this message because you are subscribed to a topic in the Google Groups "H-bot and CoreXY 3d printers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/h-bot-and-corexy-3d-printers/sVyTyfYoOwQ/unsubscribe.

To unsubscribe from this group and all its topics, send an email to h-bot-and-corexy-3d-printers+unsub...@googlegroups.com.

Visit this group at https://groups.google.com/group/h-bot-and-corexy-3d-printers.

--

Nick

www.nhfoley.com

 

[Nick Foley]

Actually just using one. Given your larger bed size, it might not be big enough, but for 10"x16", it heats the whole thing quickly to 60°C, with what seems to be acceptable evenness.

On Sun, Apr 3, 2016 at 6:05 PM, Michael Hackney <mhac...@eclecticangler.com> wrote:

I presume you are using 2 of these?

On Sunday, April 3, 2016 at 5:44:25 PM UTC-4, Nick Foley wrote:

Heater:

http://www.ultibots.com/24v-kapton-heater-240mm-x-310mm/

Also using PEI.

On Sun, Apr 3, 2016 at 5:41 PM, Michael Hackney <mhac...@eclecticangler.com> wrote:

Thanks, that's enough to compare with mine. I'm using 1/4" MIC6 aluminum quite a bit larger, no glass and two 12"x12" silicone heaters. What are you using for heaters? I'll also has PEI on top of the aluminum.

On Sunday, April 3, 2016 at 5:38:44 PM UTC-4, Nick Foley wrote:

Not totally sure about the weight - key items are two 500mm lengths of 20x20 V-slot & a 10" x 16" x 3/16" aluminum plate (from Online Metals) for the heated bed. I had been planning on adding glass on top of that, but the aluminum has proven to be flat enough even when heated to reliably take the first layer.

Bearing blocks are these: http://openbuildspartstore.com/mini-v-gantry-set/

--
You received this message because you are subscribed to a topic in the Google Groups "H-bot and CoreXY 3d printers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/h-bot-and-corexy-3d-printers/sVyTyfYoOwQ/unsubscribe.

To unsubscribe from this group and all its topics, send an email to h-bot-and-corexy-3d-...@googlegroups.com.

Visit this group at https://groups.google.com/group/h-bot-and-corexy-3d-printers.

--

Nick

www.nhfoley.com

 

--

You received this message because you are subscribed to a topic in the Google Groups "H-bot and CoreXY 3d printers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/h-bot-and-corexy-3d-printers/sVyTyfYoOwQ/unsubscribe.

To unsubscribe from this group and all its topics, send an email to h-bot-and-corexy-3d-...@googlegroups.com.

Visit this group at https://groups.google.com/group/h-bot-and-corexy-3d-printers.

--

Nick

www.nhfoley.com

 

[Ben Holmes]

That's an interesting thought.  I've been trying to figure out how to get a heater at the right size for my 10"x10" buildplate and have only been finding smaller or larger heaters.  With the proper heat spreader does a smaller one work well enough?

Ben

You received this message because you are subscribed to the Google Groups "H-bot and CoreXY 3d printers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to h-bot-and-corexy-3d-...@googlegroups.com.

[Ryan Carlyle]

With a thick enough heat spreader, it barely matters how big your heater is. Problem is, the thicker your spreader, the more the bed weighs and the longer it takes to heat up and cool down. For example, I recently did an 8x8" heater on a 10"x12"x1/4" heat spreader... the surface temps are even enough, but it takes a LOOOONG time to hit the target. 

The thing I'd be worried about with two heaters is different power outputs. The resistances will probably be a little bit different. That means one half of the bed will likely end up hotter than the other. So you really ought to use separate zone heating for each independent heating element, rather than a single temp control. That's pretty easy to do with an external PID controller unit, or you could treat one of the beds as an additional extruder. 

[Ben Holmes]

How about using an 8"x8" heater and only printing objects up to that size when you need to use it?  I'm very comfortable printing PLA on Gluestick Glass for most things but want to start printing ABS and Nylon.

Ben

On Mon, Apr 4, 2016 at 2:27 PM, Ryan Carlyle <temp...@gmail.com> wrote:

With a thick enough heat spreader, it barely matters how big your heater is. Problem is, the thicker your spreader, the more the bed weighs and the longer it takes to heat up and cool down. For example, I recently did an 8x8" heater on a 10"x12"x1/4" heat spreader... the surface temps are even enough, but it takes a LOOOONG time to hit the target. 

The thing I'd be worried about with two heaters is different power outputs. The resistances will probably be a little bit different. That means one half of the bed will likely end up hotter than the other. So you really ought to use separate zone heating for each independent heating element, rather than a single temp control. That's pretty easy to do with an external PID controller unit, or you could treat one of the beds as an additional extruder. 

--

[Nick Foley]

Yeah, these were the reasons that I stuck with a single, almost-large-enough heater. I thought about using a copper heat spreader instead of (or in addition to) aluminum to keep the mass down while maxing thermal homogeneity, but decided to KISS with 3/16" aluminun and everything seems fine.

On Mon, Apr 4, 2016 at 5:27 PM, Ryan Carlyle <temp...@gmail.com> wrote:

With a thick enough heat spreader, it barely matters how big your heater is. Problem is, the thicker your spreader, the more the bed weighs and the longer it takes to heat up and cool down. For example, I recently did an 8x8" heater on a 10"x12"x1/4" heat spreader... the surface temps are even enough, but it takes a LOOOONG time to hit the target. 

The thing I'd be worried about with two heaters is different power outputs. The resistances will probably be a little bit different. That means one half of the bed will likely end up hotter than the other. So you really ought to use separate zone heating for each independent heating element, rather than a single temp control. That's pretty easy to do with an external PID controller unit, or you could treat one of the beds as an additional extruder. 

--

You received this message because you are subscribed to a topic in the Google Groups "H-bot and CoreXY 3d printers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/h-bot-and-corexy-3d-printers/sVyTyfYoOwQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to h-bot-and-corexy-3d-...@googlegroups.com.
Visit this group at https://groups.google.com/group/h-bot-and-corexy-3d-printers.

--

Nick

www.nhfoley.com

 

[Michael Hackney]

I'm considering having a custom heater made but need to think it through a bit. I'll be using MIC6 as my print bed with PEI on top so isolating 2 separate heaters would be tricky.

To unsubscribe from this group and stop receiving emails from it, send an email to h-bot-and-corexy-3d-printers+unsub...@googlegroups.com.

[Ryan Carlyle]

You don't really need to isolate them, do you? If each thermistor is near the center of its heater, they shouldn't interact or anything. You'd just want to use a third temp sensor to calibrate both setpoints to match so you don't get one heater thinking it should be on when the other thinks it should be off. 

[William Cook]

Just an idea, but I think that EcoCast aluminum plate comes in 5mm thicknesses and similar specs to Mic-6. So you'd have a 21% reduction in thickness and mass and the same flatness.Can order from here, the quote comes to £31.01

[Michael Hackney]

5mm echoCast would be great. It could be braced with small angles underneath for longitudinal rigidity if needed. I think shipping to the US might be more than the material but I am going to check it out. I'll also see if I can find a US source. Thanks!

[Michael Hackney]

Apparently AluminumWarehouse is not set up for US orders or shipping. A quick search did not turn up a US supplier.

[Nick Foley]

6061 plate from online metals is astonishingly flat.

--
You received this message because you are subscribed to a topic in the Google Groups "H-bot and CoreXY 3d printers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/h-bot-and-corexy-3d-printers/sVyTyfYoOwQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to h-bot-and-corexy-3d-...@googlegroups.com.

[Michael Hackney]

Several years ago I manufactured 12.125" diameter disks of 6061 from OM as heat dissipators for the Rostock Max delta printer. I lost a lot of money as I refunded most of the sales because the material warped in transit after machining. MIC6 and EcoCast are purpose made for these sorts of applications and I'd prefer not to risk even a little warp. 

To unsubscribe from this group and all its topics, send an email to h-bot-and-corexy-3d-printers+unsub...@googlegroups.com.

[William Cook]

Dang, that upsets my plans a bit. Gleich also makes 5mm cast aluminum tooling plate but seems only set up to ship in Germany.

I wonder why these guys don't want to ship.

[William Cook]

I got an email from Gleich confirming that the will ship 5mm cast tooling plate to the US, but you need to send them a message on their contact page because the web site is not set up to calculate shipping and so forth for international orders. It looks like they are marginally more expensive than the aluminum warehouse guys, and I don't yet know what shipping is like, but it's a substantial weight savings and might be worth it.

[Michael Hackney]

Making progress once again on my big Core X-Y. I decided to not get bogged down in analysis paralysis and over designing. Originally I was going to machine all of the stepper mounts, ball screw bearing mounts, etc. But because the materials and time is $$ (my time) I decided to print these parts in carbon fiber PETG. This stuff is great, it prints beautifully, holds dimension really well and is very rigid. It allows me to prototype and print a couple of iterations in cheap PLA and then do the final print in PETG-CF - all in my office. I've got the bearing mounts finished for the 4 ballscrews. I just made a quick update post with photos: https://www.blogger.com/blogger.g?blogID=1024848558336119865#editor/target=post;postID=6820451113682415610;onPublishedMenu=allposts;onClosedMenu=allposts;postNum=0;src=postname

Now I have a question... I'm using 32 tooth pulleys for the Z stage drive. I have 20 tooth pulleys I intended to use for the Core X-Y stage. Are these sufficient or should I use larger pulleys?

On Monday, March 7, 2016 at 1:20:07 PM UTC-5, Michael Hackney wrote:

All, I've spent several weeks researching and mocking up ideas for my 300x600x500(ish) CoreXY printer. I decided to use the SpeedDemon SG linear rails and blocks as I've used these on an all-metal Rostock Max build and they are really nice and come in various dimensions to suit the application. Before getting into the details, in addition to building many delta and simple Cartesian printers, I have converted several milling machines to CNC and fabricated 2 large CNC routers. I have a good understand of "rigidity" and motion control.

I'm going with 30x30 extrusions (slotted Misumi) for this build rather than rectangular 20x40. Note that I've left off all the reinforcing and corner attachment details but the frame will be rigid when done. 

On this machine, X is short and Y is long. X will be a SD SG20 rail which is 60mm wide x 22.5mm high. It sits on top of an extrusion that couple the blocks on the Y linear rails. These Y rails are the SG10 and are 28mm wide x 14 mm high. They are mounted UNDER the frame cross bars with the sliding block facing down as shown. The Z rails are also SG20s and mounted as shown. I'm considering milling a slot into the extrusion cross pieces to recess the Z rails by 10-15mm just to clean-up the appearance a little. I'll use 2 rails. The screws are 1204 single start screws commonly used. They are mounted offset from the rails a bit. I understand about torque with offset screws, etc but in this application with these wide rails I think this will be a workable arrangement. My big milling machine has a big screw mounted a few inches off center from the dovetail ways and has to lift a 2 HP motor and auto tool changer and performs well. This is a miniature version of that. I'll use 2 steppers likely mounted up top to drive them with the opposite ends floating.

Not shown is the table base that sits on top of these. The grey sheet is the working X-Y I'm designing for. The support for the bed will likely be full width this L extrusions mounted to the blocks on the Z rails with a 3 point leveling system. I'm going back and forth between melamine or MIC 6 for the table base. I could mill holes in the MIC6 to lighten it and am leaning towards that just because Melamine is more sensitive to humidity changes.

This is a rough drawing of what I'm thinking.

I haven't come across anyone using the SpeedDemon rails for a Core XY yet but I tried to calculate clearances based on other printers with linear rails. 

In addition to feedback on the basic idea and some of the details, does anyone have experience with .9 degree steppers on these printers?

cheers,

Michael

[Michael Hackney]

Eee gads, sorry to propagate that horrible design in the rendering in my previous post! I just clicked reply in this thread and didn't realize it was going to reply to the first post. The machine design is significantly different/better thanks to feedback here. Details on my blog.

[Michael Anton]

If you search google for "minimum belt wrap", you will find that GT2 belts are rated for a minimum of 6 tooth engagement, without a reduction of rated torque.  For each tooth under 6, you lose 20% of the torque rating.  So, if the belt is wrapped around half of the pulley, then you can use a pulley with a minimum of 12 teeth.  This is a good reference: http://ie.rs-online.com/webdocs/002d/0900766b8002d3b7.pdf.  Note that if using small pulleys, then it would not be a good idea to use belts with a steel core, as they would fail very quickly (they have a minimum bend radius).  The normal fibreglass belts that are used for 3D printers should be fine.

[Michael Hackney]

Michael, thank you - exactly the info I was looking for. So at the corners where the belt is making a 90° turn, 20/4=5 teeth engagement is suboptimal. This is a big 300mmx600mm printer and although the Core X-Y mechanism is not terribly massive it's probably best to increase the tooth count.

[Michael Anton]

But aren't those 90 degree corners just idlers?  You don't need any tooth engagement for those, as they don't have any force being applied to them.  Now, if the 90 degree corners are where your motors are located, then you have a problem.

[Michael Hackney]

Ah yes, they are idlers.