Core XY iteration - asking for feedback

[vinvin]

Hi guys 

I wanted to make my own printer. This is it!

http://www.thingiverse.com/thing:588277

I went for a CoreXY (http://corexy.com) configuration with a bowden setup.
I decided to use SFU1204 ballscrew for the Z axis, synchromesh cables and guitar tensioning mechanisms. 10mm rod for the X axis, and 12mm for the Y axis. 

I will be more than happy to read your opinions/critics/thougths about it!

ALSO : I haven't choose my electronic board yet, and doesn't know with one to choose. I wanted to go with 1/32 microstepping drivers since as i understood, Marlin implementation of corexy system divide the resolution by factor 2, and also for quietness of operation. Should I choose a 32bit capable board like the Azteeg X5 mini or a regular atmel 2560 will do?

[vinvin]

[Jetguy]

Constraining the tops of the ballscrews is going to introduce Z wobble artifacts.

Part 2, because ballscrews are so free moving, syncing 2 Z motors is asking for trouble and you'll likely have unrepeatable leveling as when the motors are off, the Z stage can become 1-2 full steps out of alignment, then when you enrgize the motors, they jump to the nearest pole alignment and nothing ensures that is consistent. Preferred method is use a single motor and slave the shafts to each other via timing belt.

Everything else is fine, but again, there is a better way to build a Z stage, it's well documented and proven.

[Jetguy]

Here's the reason why and already in this group and floating around the web you guys are sharing the same bad element over and over.

No Z screw that long is straight. Between manufacturing process and shipping/handling, at some point, it is HIGHLY likely there is some runout. Either non-concentric or just plain old bent screw.

When you constrain both ends, EVEN with some form of isolation at the nut, there is still some force from the eccentric rotation of the shaft being pushed into  X Y motion of the Z stage.

Guys have come up with all kinds of methods for mounting that nut, but in the end, they still transmit some force in X Y.

If you let one end float this effectively decouples the leadscrew from being able to transmit small but perceptible wobble into the z stage.

Finally, some pics of my methods

Double V arrangement for maximum belt wrap on every pulley and still only uses 2 idlers.

608 skate bearing idlers.

Motor and overall arrangement.

Moving stage plate detail from the bottom side.

[vinvin]

Hello Jetguy,

First i'd like to thank you for your ideas, and the way you explained them. just to be sure : you are suggesting to take the BF10 end support out of the design, and find a way to drive the ballscrews with a single motor via belts/synchromesh cable?

The flexible couplings won't be needed anymore? Would i need to change my ballscrews/shafts arrangement for a 3 screws design like yours, or just 2 would be sufficient?

Thanks a lot

[Jetguy]

No, drive with GT2 timing belt and matching pulleys from SDP-SI or other supplier.

How can you link a continuous loop synchromesh cable and have it run over pulleys in a loop? I don't think you can and that's why you use a closed loop timing belt.

Yes, remove the UPPER BF 10 ends as all they do is hold the ballscrew rigid. All ballscrews, threaded rods, etc have some amount of runout and almost always are not straight. It's nearly impossible to machine and maintain an perfectly straight rod of any threaded type. By the time i's shipped to you, eventually, there is some level of runout.

Removing the upper bearing lets the rod angle ever so slightly as required so that the runout is no forced into your Z stage as wobble. Again, we are talking fractions of a fraction, of an angle. It's nothing at all and fully within the compliance of the bearings involved.

3 is better and after a certain size (around a 12 By 12 in bed and larger) 4 screws are required.

Look, I know a LOT of printers like the MakerBot Z 18 are using an arrangement like yours and it does generally work (FYI, even the makerBot Z18 uses a belt and a single motor to sync the 2 screws.

The problem is, it's not optimal.

[vinvin]

Ok ok, thanks again for your advices. 3 ball screw are going to cost me an eye...How did you manage belt tension with your setup?

 I'll make the modifications as soon as possible and post the results here

[Jetguy]

The idler holes are slots so the 2 bearings in the middle space are how you adjust final belt tension

You cannot see the slots because I used big washers to spread the load so the idlers don't bend from 90 degrees to the wooden floor.

FYI the 3 screws are a perfect triangle pattern

The oddball is the motor below the 3 screw in the middle on the left hand side. Keep in mind why I used that path.

#1 maximize number of teeth in contact on each pulley (belt wrap angle)

#2 Use an existing long length belt I had on hand.

#3 minimum amount of idlers VS other methods and belt paths to achieve maximum wrap.

[vinvin]

Hi Jetguy

I have made some modifications regarding the Z axis. I went with your setup. The printer is quiet shorter thank to this!

Tell me what you think!

http://www.thingiverse.com/thing:588277

[Jetguy]

Very, very nice!

You get the Jetguy stamp of approval.

 I am Jetguy and I approved this message.

[adam paul]

Only because I am starting a similar project, how do you plan on adjusting belt tension?

I like your utilization of the 20x40 to offset the idlers, I was planning on using a plate, I may adopt your design. For my build I can rotate the bar the idler is on 90 degrees making it adjustable.

[vinvin]

Thx, glad to be ISO-Jetguyed

[vinvin]

Belt tensioning is done by moving the pulley that control the angle of the motor's belt warp. When you put that pulley to the right limit (as close as possible to the other belt path), i'm getting 1840mm for the belt lenght, I guess that will be enough for a 1830mm belt.
20x40? Are you talking about the extrusion? Those are 40x40mm aluminium extrusion. I placed this extrusion in that way so that the slot holding the pulley isn't parrallel to the belt tension

[adam paul]

Thanks for the explanation, I has wrongly assumed bet tension on the idler. Makes sense to put the cross member perpendicular to the belt pull.

[Jetguy]

One very minor note.

You could save both money and prevent over constraint by not using the particular Z axis linear bearings you chose.

In this new design, It's counterintuitive but the Z axis linear bearings ONLY need to brace motion in the X Y plane. By using tall and rigid linear bearings the way you are, you set yourself up for friction and binding if all 3 leadscrews are not initially timed to a natural state when the linear bearings are mounted and all is rigid. 

A cheaper/better way is to mill holes in the aluminum Z stage plate for common single linear bearings to exactly fit tightly in. Since Linear bearings often have grooves for C clips, simply use them + spacer to center the bearing in the plate. This gives you some minor angular freedom for the plate to be leveled on the 3 screws and NOT bind the linear bearings. In fact, to prevent real over constraint, one side is a tight press fit round bore and the other could be a tight slot.

The one side is the true linear guide, the other is the anti-rotation of the plane. 

Caution, absurdly crude drawing.

[vinvin]

You're right, all of this makes sense, even thought this approach "looks" weak! I think that i'll try with only 2 lmf12uu bearing, which are 30mm tall, and see what the results look like. Have you tried the C-clips+spacer configuration? Don't you think that the "non-perfectly linear ballscrews" will be more prone to make the bed support move with only one true static constraint? The reason I went with 3 large bearings in the first place is because i'd like to print ceramics, and i thought that I would need a sturdy Z axis to support the weight of the printed object....

what do you mean by "timed to a natural state"?

Thanks again for all that input

[Jetguy]

Again, the problem here is you used flanged tall linear bearings. They want to maintain the plane at a perfect 90 to the rod.

the 3 leadscrews together provide a plane of lift.

If both of those planes are not lined up- we get binding.

So by timing, I mean all 3 leadscrews pulleys and timing belt would be locked down to the leadscrews shafts in -situ.

If you arbitrarily lock the pulleys to the leadscrews BEFORE installing everything, your leadscrew Z axis plane may not match the natural plane of the bearing when all is mounted.

Just as much care needs to go into key steps during assembly as there is during the design phase of a printer.

You can have a great design and screw it up during assembly.

[Ryan Carlyle]

I agree with everything Jetguy is saying about Z stages.

On the XY gantry, two thoughts:

1) Marlin is generally not fast enough to run a 1/32 stepping printer. Plus, 1/32 stepping doesn't improve resolution at all -- no microstepping over about 1/8th adds any resolution, because the marginal torque per microstep is too small to move anything. The only genuine advantage of 1/32 stepping over 1/16 is that it's a bit quieter. get stepper dampers first, and if you still think it's too loud, switch to 1/32 stepping drivers later.

2) You're going to want bigger idler bearings if you plan on running the printer hard. The smallest genuine synchromesh size is a bit big for 624v bearings. It works, but it's not a great sizing choice. You end up violating the minimum bend radius, which will dramatically shorten the fatigue life. Switching to 625v now shouldn't be too hard, I don't think. Another option to consider is ordering more expensive U-groove bearings (eg from Misumi) which will support the synchromesh cable with less vibration/stress and be dramatically higher quality than all the crappy Chinese 624v bearings (that I've found, anyway).

[vinvin]

Thanks for this.

I was planning on using the 3.81mm pitch/2.2mm outside diameter synchromesh cable from SDP/SI. But i can't find any minimum bend radius recommendation.... 

Do you think that these would be better ?

http://www.aliexpress.com/item/V625-120-V-grooved-straightener-guide-wheel-bearings-V625ZZ-625VV-V-groove-pulley-roller-bearing/2040005078.html

But with these i will need to increase the X axis rods length and the space between them

[Ryan Carlyle]

I forget where I found the MBR, it was one of SDP-SI's pages though. Maybe a cad file download or an info page, I forget. (Or maybe I just took the MBR for the wire core...? Hmm.)

Those 625 bearings don't have much groove depth -- you might get it to run ok and stay in the groove once everything gets tensioned up, but running the cable through the gantry may be hellish.

[Viesturs Lācis]

On Thursday, December 11, 2014 3:58:12 PM UTC+2, vinvin wrote:

Hi guys 

I wanted to make my own printer. This is it!

http://www.thingiverse.com/thing:588277

I went for a CoreXY (http://corexy.com) configuration with a bowden setup.
I decided to use SFU1204 ballscrew for the Z axis, synchromesh cables and guitar tensioning mechanisms. 10mm rod for the X axis, and 12mm for the Y axis. 

I will be more than happy to read your opinions/critics/thougths about it!

Most probably I am too late, but still just out of curiousity - why did you put so many extruded aluminium profiles in the frame (which obviously would make the frame itself extremely stiff and rigid) and still use round rails that are supported only at their ends (and need relatively low load to bend by 0,1 mm or so), which makes it useless to have such a frame in first place. 

IMHO half of those profiles should be removed and for the saved cost a precision profiled linear guide could be used, thus providing better actual stiffness for the endeffector. And less profiles would allow you to see better, what is going on inside, which is very useful during first prints, when there are so many things not working as they should (speaking for myself here).

[adam paul]

IIRC this printer will be used to print ceramics.  The extra rigidity would be to support those weights

[Ryan Carlyle]

All the print weight is carried by the lead screws, which are mounted to the bottom of the frame. Everything above that just supports the gantry. It's definitely overkill as far as extrusions. 

[vinvin]

Did you notice that i used 12mm rods for this axis? The moving mass will be around 800g.... I "thought" these would be enough....

I thought about using SBR rails like these :

http://i00.i.aliimg.com/wsphoto/v0/744932396/SBR16-Length-500-mm-sbr-linear-bearing-supported-rails-Rail.jpg

But : - I couldn't find a way on how to mount the X axis rods

[vinvin]

Sorry.... And i didn't think that you could use the carriages associated vertically

I also thought about using precision linear rails... but i could't find any affordable one cut to specified length 

At last, I tried these : 

http://cdn1.bigcommerce.com/server2300/itwgldve/product_images/uploaded_images/v-slot-rail.jpg?t=1398725710

But same thing, i couldn't find a way on how to mount the x axis rods.....and the idler bearing and everything else

[vinvin]

The thing is that ceramics will be for experimentation.... i will use this printer mainly for abs.....

Le mardi 6 janvier 2015 16:31:12 UTC+1, adam paul a écrit :