400-500 mm cube corexy vs deltabot?

780 views
Skip to first unread message

Marc Lee

unread,
Dec 9, 2014, 8:19:10 AM12/9/14
to 3dprintertips...@googlegroups.com
Hi everyone, this is my first post on this forum (having been lured across from the deltabot forum!). Anyway thanks for all the great posts that I've been following - things are certainly developing at a fast pace.

I'm about to build a printer but am stuck at the corexy vs delta decision and would be very grateful for guidance from those who know both please.

I'd like:
a build volume of about 400-500 mm cube (or 400-500 mm diameter & height cylinder)
to print TPU/TPE/flexible filament items like swim fins, shoes etc. both flat and vertical (~200-300g weight)
to print harder plastic (PET+ etc.?) moulds for the above items
a heated enclosure with heat sensitive parts outside

I'm an engineer, competent with hand tools, building stuff, electronics, computers, CAD etc. (but no mills or lathes immediately to hand).

Cheers from sunny & thunderstormy Sydney

Thanks, Marc

Christopher

unread,
Dec 9, 2014, 8:37:22 AM12/9/14
to 3dprintertips...@googlegroups.com
I would say Core XY is the way to go. Just search the group and you can find a lot of instruction and pictures of the bots Jetguy and the other have build.

Ryan Carlyle

unread,
Dec 9, 2014, 11:08:57 AM12/9/14
to 3dprintertips...@googlegroups.com
You're probably not going to have much luck printing long, thin, vertical parts with TPE. It's too flexible and will sway.

That's a very big bot you're talking about. It's getting into the size range where cost escalates significantly and everything gets wobbly unless you over-build the heck out of it. It's possible to make either printer perform well in that size range, but the Delta will be about twice as tall and thus require much stronger structural members than the CoreXY to get equivalent print quality. On the other hand, you're going to have to go with 3 or 4 leadscrew for the CoreXY Z stage, which adds some cost and complexity. The Delta will be a little more brute-force construction and might end up a bit cheaper. 

Do you have a preference on electronics? Around here, most of us seem to prefer Sailfish/Mightyboard which does not support Deltas. 

James Le

unread,
Dec 9, 2014, 2:56:34 PM12/9/14
to 3dprintertips...@googlegroups.com
I would recommend a corexy build over the delta.

You didn't mention speed, which a lower effector weight would help with.

CoreXY is more volume efficient, it doesn't need to be as tall.

You want to print TPU which is quite difficult with a bowden extruder. CoreXY is better designed for direct drive extruders moumted directly on the effector.

Marc Lee

unread,
Dec 9, 2014, 5:03:56 PM12/9/14
to 3dprintertips...@googlegroups.com
Thanks, I've done quite a bit of searching and seen a lot...but is there a complete design/BOM for this size that I am missing? For example is it Jetguy's "midi" that I should be looking at, and if so, is there a BOM/design area for that somewhere?

Jetguy

unread,
Dec 9, 2014, 5:29:25 PM12/9/14
to 3dprintertips...@googlegroups.com
BOM is not complete yet for the midsize.

The reason is there is some heartache on this design.

I used very expensive IKO linear slides for Y and X. I believe that the design's success is dependent on those specific slides. If they are not widely available or affordable, the design must change.
The reason is complex to describe but basically, these are the first linear bearings I used out of the box that are perfect with no extra drag or bumbs as the balls enter and exit the preload bearing surface between the carriage and rail.

I would like to source the entire BOM from Adafruit and use the existing slides that they sell and those are used for Z axis, but I had to wash out the heavy grease that shipped in them and made them insanely stiff. To me, while possible, it just seems a lot of DIY steps required and may give marginal performance.

On the other hand, one could scale up the other design posted earlier that rides on the linear extrusion. Again downsides to this is that user/builder error is introduced, there is room for other problems.

Finally, the Z screws are sourced from Zen toolworks at $45 each. Today, I just ordered on Amazon some impressive 16mm ball screws 670mm long for $59 but there is a HUGE caveat, they require 15mm bearings and even worse, 12mm bore pulleys. The intent now is to turn the shanks down in a lathe to 8mm so 608 skate bearings and 8mm bore GT2 2mm pitch 36 tooth pulleys like I'm using now could be used.

Finally, the entire X gantry was made by hand and probably has 100 precision drilled holes in key locations. It is critical to performance and operation.

So the truth is, while much seems easy and is sourced right out of the Adafruit CNC section, there is also a fair amount of very precise hand made parts that no drawings exist of. They are one of a kind until I can document them. And with work keeping me away, that process is very slow.

Marc Lee

unread,
Dec 9, 2014, 5:52:14 PM12/9/14
to 3dprintertips...@googlegroups.com
The long thin parts would include fins made from reasonably stiff 95 Shore A TPU (e.g. http://filaments.ca/collections/flexible-filaments/products/flexible-tpu-polyurethane-filament-blue-1kg-spool ) and vertical would be the best for those parts if it can be done. Buttresses may be required to minimise wobble. If I can't print the fins direct then I would like to print the vertical moulds for them out of stiffer plastic. I would probably print both mould sides together, and temporarily linked together, so that if they warp a bit then they warp together, and still fit as a mould after I cut them apart.

I wonder if there is a difference between (well-built) CoreXYs and Deltas, relating to sway/wobble causing issues for tall, thin prints? I had previously thought Deltas might be better in this regard because of their fixed build plate. But is that the general experience?

I don't have a preference on electronics (should I?!). I was just planning to go with the best recommended by the groups for either CoreXY or Delta.

As for size, 400 mm is the absolute minimum for the parts to fit, and I would prefer 500 mm as a buffer in case I need it. Is there a particularly big design difference in moving up to 500 mm from 400 mm?

Chris P

unread,
Dec 9, 2014, 7:02:31 PM12/9/14
to 3dprintertips...@googlegroups.com
 The intent now is to turn the shanks down in a lathe to 8mm so 608 skate bearings and 8mm bore GT2 2mm pitch 36 tooth pulleys like I'm using now could be used.

Jetguy may already know this, but for the wider audience reading this thread...

Word of warning, my local machine shop and I have spent countless hours and probably a few grand developing a repeatable process to turn (nearly) perfectly concentric shoulders on leadscrews.  I of course don't know how you plan to set up a lathe to turn down the existing shoulders on the ballscrews, but it's a tough nut to crack. If the shoulders aren't concentric you will introduce Z artifacts.  If you get it wrong you basically have to throw the ballscrew out and start over.

Ryan Carlyle

unread,
Dec 9, 2014, 7:24:16 PM12/9/14
to 3dprintertips...@googlegroups.com
If you're going CoreXY and only scaling one stage, you can go as long as you want, really. Just use good supported rails on that axis (Y or Z) and the length becomes fairly unimportant. (Length is bad on unsupported rods so avoid long X axes.) There are limits to how long you can get cost-effective lead screws, but 500mm usable length shouldn't be too bad.

Deltas do scale vertically quite easily. You pretty much just make the columns longer and you're done. There are some firmware limits though -- specifically with Repetier using integer step math. Again, shouldn't be a problem at 500mm.

On the TPE print wobble -- you definitely wouldn't want to use a RepRap style moving bed, but with any printer you have nozzle drag. Whenever the nozzle is moving across the print (whether extruding or traveling), it's pushing on the print. With tall, flexible prints, that's going to cause some XY registration issues.

Jetguy

unread,
Dec 9, 2014, 7:24:35 PM12/9/14
to 3dprintertips...@googlegroups.com
Which might be cause for me to just retain the existing 15mm shaft/shoulder and just turn down the 12mm tip to 8mm for the pulley only.
The reason behind going 8mm is the timing pulleys.
#1 I already have a ton of GT2 2mm pitch 36t pulleys (Cupcake days) VS, the only 12mm bore pulleys from SDP-SI are higher pitch belts (XL, T5, etc, thus forcing massive belt change and even more cost).
#2 the pulley portion is less sensitive to any deviation.
#3 the only 12mm bore puleys are larger diameters resulting that the ballscrew positions are moved back the frame rails for the pulleys to clear.

Marc Lee

unread,
Dec 9, 2014, 7:34:37 PM12/9/14
to
Thanks for the update Jetguy. Yes finding the path of least heartache for everyone is key.

Are you suggesting scaling up this CoreXY: http://www.thingiverse.com/thing:393155 instead? How do (well-built) carriages running on extrusion, compare with the linear slides on X & Y? I'm guessing you had reasons for moving to linear slides....X&Y precision perhaps?

On Wednesday, 10 December 2014 09:29:25 UTC+11, Jetguy wrote:
BOM is not complete yet for the midsize.

The reason is there is some heartache on this design.

I used very expensive IKO linear slides for Y and X. I believe that the design's success is dependent on those specific slides. If they are not widely available or affordable, the design must change.
The reason is complex to describe but basically, these are the first linear bearings I used out of the box that are perfect with no extra drag or bumbs as the balls enter and exit the preload bearing surface between the carriage and rail.

I would like to source the entire BOM from Adafruit and use the existing slides that they sell and those are used for Z axis, but I had to wash out the heavy grease that shipped in them and made them insanely stiff. To me, while possible, it just seems a lot of DIY steps required and may give marginal performance.

On the other hand, one could scale up the other design posted earlier that rides on the linear extrusion. Again downsides to this is that user/builder error is introduced, there is room for other problems.

Finally, the Z screws are sourced from Zen toolworks at $45 each. Today, I just ordered on Amazon some impressive 16mm ball screws 670mm long for $59 but there is a HUGE caveat, they require 15mm bearings and even worse, 12mm bore pulleys. The intent now is to turn the shanks down in a lathe to 8mm so 608 skate bearings and 8mm bore GT2 2mm pitch 36 tooth pulleys like I'm using now could be used.

Finally, the entire X gantry was made by hand and probably has 100 precision drilled holes in key locations. It is critical to performance and operation.

So the truth is, while much seems easy and is sourced right out of the Adafruit CNC section, there is also a fair amount of very precise hand made parts that no drawings exist of. They are one of a kind until I can document them. And with work keeping me away, that process is very slow.


On Tuesday, December 9, 2014 5:03:56 PM UTC-5, Marc Lee wrote:
Thanks, I've done quite a bit of searching and seen a lot...but is there a complete design/BOM for this size that I am missing? For example is it Jetguy's "midi" that I should be looking at, and if so, is there a BOM/design area for that somewhere?

On Wednesday, 10 December 2014 00:37:22 UTC+11, Christopher wrote:
I would say Core XY is the way to go. Just search the group and you can find a lot of instruction and pictures of the bots Jetguy and the other have build.

Am Dienstag, 9. Dezember 2014 14:19:10 UTC+1 schrieb Marc Lee:

Ryan Carlyle

unread,
Dec 9, 2014, 7:51:35 PM12/9/14
to 3dprintertips...@googlegroups.com
Chris P with Fusion gets very good results at 300mm using extrusion as the X rail. Lots of Deltas put the carriages directly on V-slot extrusion columns. It's a perfectly viable option.


On Tuesday, December 9, 2014 6:34:37 PM UTC-6, Marc Lee wrote:
Thanks for the update Jetguy. Yes finding the path of least heartache for everyone is key.

Are you suggesting scaling up this CoreXY: http://www.thingiverse.com/thing:393155 instead? How do (well-built) carriages running on extrusion, compare with the linear slides on X & Y? I'm guessing you had reasons for moving to linear slides....X&X precision perhaps?

On Wednesday, 10 December 2014 09:29:25 UTC+11, Jetguy wrote:

Jetguy

unread,
Dec 9, 2014, 8:16:10 PM12/9/14
to 3dprintertips...@googlegroups.com
I agree, please don't take my comment as a huge detractor to using rollers on the rails. Deltas use them, Chris uses them, they are successful all over the place. It's something I contemplated using myself.

I had the linear ways on hand sitting for a project

Now, I will say there is an advantage to building it exactly the I did is lower footprint.
The rails attach to the main frame extrusion but allow for fully enclosed and bracing side panels (something I hadn't implemented yet, but could be done at any time.
When you ride on the frame, then you now have something that has to clear everything from a motion perspective.

But if one were to reproduce this, it gets expensive. Further, I have 2 carriages on every slide making it more expensive. For example, this keeps the entire X gantry from rocking and that would in turn cause the nozzle to rise/fall since it hangs over the edge of the X. Same with the X, I wanted the extruder with maximum bearing underneath. 

FYI, the Type A machine 2014 uses single carriages. I could honestly tell the difference in the machine during a print by sound. There was a distinct rattle and I have no doubt after some careful strain testing with the motors locked there was car more flex from a single carriage per rail than my doubles.

Is it necessary? No, I just had them in hand and so they got used.

Chris P

unread,
Dec 9, 2014, 9:29:02 PM12/9/14
to 3dprintertips...@googlegroups.com
#2 the pulley portion is less sensitive to any deviation.

My experience indicates otherwise.  The system appears more or less equally sensitive to runout in a) how the shaft is machined, b) if the running surface of the pulley is not concentric with the axis of rotation of the screw, c) if the pulley is not mounted concentrically. 

If the pulley running surface is nonconcentric, your leadscrews effectively move at different rotational speeds as they turn.  This means the platform is not moving the same vertical distance at all points, and therefore you get Z artifacts.  Granted, they are extremely minor and unless you're running translucent filament you won't even see them.  But they are there :-)

Chris P with Fusion gets very good results at 300mm using extrusion as the X rail. Lots of Deltas put the carriages directly on V-slot extrusion columns. It's a perfectly viable option.

It works very well as long as you understand the limitations of the system and design with those in mine.  It is not nearly as stiff in resistance to cantilevered and bending loads as other linear motion systems.  However you can achieve very high stiffness in other directions.  And if you're designing precision linear motion systems, you should be avoiding cantilevered loads anyway.

Chris

Marc Lee

unread,
Dec 10, 2014, 7:19:47 AM12/10/14
to 3dprintertips...@googlegroups.com
Thanks for the further explanation to put it in perspective.

What about using wheels connected to the bed and running on the vertical extrusions for the Z (still with 4 lead screws, one in each corner)? Is that a viable option too?

Marc Lee

unread,
Dec 10, 2014, 7:29:27 AM12/10/14
to 3dprintertips...@googlegroups.com
Yes, for flexible filaments, it seems to be commonly said that bowden extruders are not really suitable, but I don't know if this is true?

So maybe I would have to find a light extruder option for the effector (especially on a Delta), like the Ultibots Micro Extruder (not yet tested for flexible filaments to my knowledge) or the Flex3Drive (not sure how well that works either). I don't know of any other options - maybe other people do?

Jetguy

unread,
Dec 10, 2014, 9:10:22 AM12/10/14
to 3dprintertips...@googlegroups.com
I would not do it that way.
It violates the rules and causes over constraint.
 
You only use 2 guide posts, not 4.
 
YMMV and sure, try it if you want.

Marc Lee

unread,
Dec 11, 2014, 9:14:58 PM12/11/14
to 3dprintertips...@googlegroups.com
Let's break this down into two parts.

Yes, Z is fully constrained by 3 lead screws. But I understand that the consensus is that over-constraining Z by adding a 4th lead screw (so that there can be one in each corner for symmetry) works better for larger beds as we are discussing here (to cope better with bed flex etc.). Is this correct?

Regarding, XY constraints, I understand that you are fully constraining X and Y via two guide rails on adjacent vertical corner columns. Is that correct? If so, I am wondering how wheels connected to the bed and running in the vertical corner columns could best achieve the best XY constraint and how this would compare with your two guide rail solution for precision. I didn't mean to imply necessarily over-constraining X and Y with too many wheels/carriages running up the 4 columns. In fact, I think only two wheels running up the facing sides of adjacent verticals might be all that is logically needed to constrain X and Y. Although adding an extra pair of wheels to the other two columns might give a better result (to cope better with wheel/bearing flex etc.) even though it logically over-constrains X and Y. What is the consensus on this?

A key aspect in our quest for precision is that aluminium extrusions are only rated as accurate to about 0.5 mm per m. That does not sound good enough precision for larger printers, so presumably they usually come out more accurate as people do seem to get good results with them. But guide rails are generally rate far more accurate (unless you bend them by bolting them onto bent aluminium extrusions). So, to guide rail or run in extrusions? - that is the question.

Ryan Carlyle

unread,
Dec 12, 2014, 10:45:22 AM12/12/14
to 3dprintertips...@googlegroups.com
I'm sure it depends on the quality of extrusion you get. Keep in mind that for reputable companies, rated dimensional specs are "do not exceed" numbers, and actual tolerances will usually be much better than that figure. Any part that exceeds the figure is scrap -- and statistical quality control techniques usually place part tolerances on a bell curve type shape so the only way to have a low scrap rate is to have the "average" deviation be a very small fraction of the rated spec.

For example, I would say the Misumi 2020 I've been using to build frames is "perfectly straight" to the best of my ability to measure it. This is fairly easy to tell by holding two pieces up to each other and flipping the faces around in different combinations. Then the only real tolerance issue is adequately squaring up the frame during construction. This is difficult if you cut the extrusion yourself, but fairly easy if you order it precision cut to length.

The big advantage to using separate rails/rods is that the motion stage can "self square" by running it through the range of motion before tightening all the bolts. That ensures parallelism anyway.

Marc Lee

unread,
Dec 13, 2014, 6:45:06 AM12/13/14
to 3dprintertips...@googlegroups.com
Thanks Ryan. Do you or other people think OpenBuilds V-Slot (2020, 2040, 2060 and 2080 mm) is similarly "perfectly straight"? (Misumi extrusions do not seem to be available in Australia but OpenBuilds are.)

I assume your "self square" comment is directed at CoreXY but I can't see why this technique wouldn't work even with wheels/carriages running on the extrusions (rather than separate rails/rods).

And, for comparison, is there a similar technique for squaring up a delta by running through ranges of motions? I can't think of one. Deltas seem much harder to square up.

Ryan Carlyle

unread,
Dec 13, 2014, 9:52:29 AM12/13/14
to 3dprintertips...@googlegroups.com
You would have to build a jig to do it with a delta. You're really reliant on the vertices being printed/machines properly to get a delta square.

Marc Lee

unread,
Dec 16, 2014, 10:43:33 PM12/16/14
to 3dprintertips...@googlegroups.com

I found a CoreXY build with the kind of XY constraints that I was discussing earlier in this thread. So I thought I'd share it to see what you CoreXY people think of it.

It uses 4 wheels running in vertical extrusions (rather than using vertical rails for Z axis). See http://imkovb.blogspot.nl/2014/05/okmi-3d-printer-v02.html#gpluscommentsfor some good pics and also http://www.openbuilds.com/builds/okmi-3d-printer.1316/. Are these wheels a good solution?

The design includes some other neat ideas too. (I can already hear Jetguy asking for more wrap on the 4 Z-screw pulleys!) But what about the XY axis/carriage design and the XY belt twist-cross? And do you think it would scale well to a 500mm cubed build volume?

Ryan Carlyle

unread,
Dec 16, 2014, 10:56:30 PM12/16/14
to 3dprintertips...@googlegroups.com
I don't understand why people put so much work into weird belt crosses. It doesn't hurt anything to have them rub a little at the cross. Or you can build a staggered arrangement where each belt is planar and not have a cross at all. (That's how my current bot is built.)

Marc Lee

unread,
Dec 16, 2014, 11:11:42 PM12/16/14
to 3dprintertips...@googlegroups.com
Thanks Ryan. Do you have a pic of your preferred XY belts please? And any views on the other questions I asked? Thanks.

Ryan Carlyle

unread,
Dec 16, 2014, 11:29:58 PM12/16/14
to 3dprintertips...@googlegroups.com
Have you seen Dan's Flickr?

Joseph Chiu

unread,
Dec 17, 2014, 4:19:58 AM12/17/14
to 3dprintertips...@googlegroups.com

That looks like the standard belt cross to me?

On Dec 16, 2014 7:56 PM, "Ryan Carlyle" <temp...@gmail.com> wrote:
I don't understand why people put so much work into weird belt crosses. It doesn't hurt anything to have them rub a little at the cross. Or you can build a staggered arrangement where each belt is planar and not have a cross at all. (That's how my current bot is built.)

--
You received this message because you are subscribed to the Google Groups "3D Printer Tips, Tricks and Reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3dprintertipstricks...@googlegroups.com.
To post to this group, send email to 3dprintertips...@googlegroups.com.

Ryan Carlyle

unread,
Dec 17, 2014, 11:25:59 AM12/17/14
to 3dprintertips...@googlegroups.com
Dan's belt cross is "standard" to me. They just cross and harmlessly rub a little, no attempts at tricks. The OpenBuilds bot linked has a belt cross that's all complicated -- they put an unnecessary half twist in both belts to arrange them flat at the cross so less off-plane deflection is required to prevent touching. That's fine, it doesn't hurt anything, but it's just unnecessary effort for no particular benefit.

Here's my current belt arrangement... not sure if I'm going to keep it. The big downside is, the loads on my carriage belt clips are asymmetrical so they're sitting crooked. Doesn't affect print quality but I don't like how it looks.

Print quality (0.1 layers, 100mm/s fill, 60mm/s perimeter):




On Wednesday, December 17, 2014 3:19:58 AM UTC-6, Joseph Chiu wrote:

That looks like the standard belt cross to me?

On Dec 16, 2014 7:56 PM, "Ryan Carlyle" <temp...@gmail.com> wrote:
I don't understand why people put so much work into weird belt crosses. It doesn't hurt anything to have them rub a little at the cross. Or you can build a staggered arrangement where each belt is planar and not have a cross at all. (That's how my current bot is built.)

--
You received this message because you are subscribed to the Google Groups "3D Printer Tips, Tricks and Reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3dprintertipstricksreviews+unsub...@googlegroups.com.

Joseph Chiu

unread,
Dec 17, 2014, 11:36:58 AM12/17/14
to 3dprintertips...@googlegroups.com

Ah, ok. I just recently started working on a CoreXZ for a non-printing application, and with the limited X distance in the proof-of-concept demo for the client, I needed to put in a half twist to make the belt crossing possible... It didn't strike me as being any extra effort, so I just figured it was "standard"...

To unsubscribe from this group and stop receiving emails from it, send an email to 3dprintertipstricks...@googlegroups.com.
20141215_162316.jpg

Bonekollector

unread,
Dec 17, 2014, 2:27:59 PM12/17/14
to 3dprintertips...@googlegroups.com

I agree with Ryan - mine cross & there are no issues (500+ hrs of print time in).


Dan Newman

unread,
Dec 17, 2014, 5:00:52 PM12/17/14
to 3dprintertips...@googlegroups.com
On 17/12/2014, 11:27 AM, Bonekollector wrote:
>
>
> I agree with Ryan - mine cross & there are no issues (500+ hrs of print
> time in).

My guess is that the pulleys themselves present more wear/tear to the belts
than the belt crossing does.

Dan

Marc Lee

unread,
Dec 18, 2014, 5:51:20 AM12/18/14
to 3dprintertips...@googlegroups.com
Thanks - I had a good look through that. I think I understand most of the XY belt options now.

Still trying to figure out if a Z-axis running wheels is as good as (or better) than the more usual vertical rails. I found a CoreXY build with the kind of XY constraints that I was discussing earlier in this thread. So I thought I'd share it to see what you CoreXY people think of it.

It uses 4 wheels running in vertical extrusions (rather than using vertical rails for Z axis). See http://imkovb.blogspot.nl/2014/05/okmi-3d-printer-v02.html  for some good pics and also http://www.openbuilds.com/builds/okmi-3d-printer.1316/. Are these wheels a good solution?

(PS I tried to post this last question as a new post but it was blocked for some reason)

On Wednesday, 17 December 2014 15:29:58 UTC+11, Ryan Carlyle wrote:
Have you seen Dan's Flickr?http://imkovb.blogspot.nl/2014/05/okmi-3d-printer-v02.html

James Armstrong

unread,
Dec 18, 2014, 8:08:21 AM12/18/14
to 3dprintertips...@googlegroups.com
Yes. Mine were wearing bad from the sharp edge on the bearing flanges. I just printed some parts to cover them and extend the flange some.

John Gelnaw

unread,
Dec 18, 2014, 1:07:31 PM12/18/14
to 3dprintertips...@googlegroups.com

On Wednesday, December 17, 2014 2:27:59 PM UTC-5, Bonekollector wrote:

I agree with Ryan - mine cross & there are no issues (500+ hrs of print time in)

Been meaning to ask you... looking at that picture, it almost looks like you could uncross the belts by switching the the two belts on the two pulleys closest to the camera... Is that an illusion caused by perspective, or am I missing some other geometry problem if you reverse them?

I admit, my only issue with crossed belts is purely aesthetic. 

Scott Wells

unread,
Dec 18, 2014, 1:45:46 PM12/18/14
to 3dprintertips...@googlegroups.com
I suppose I could. I followed the plan that Jetguy, Dan, Ryan etc...  had laid forth without questioning it much. I had briefly considered 2 offset idlers in the middle of the top x span to separate the belts slightly at the "rub" point, but there were earlier discussions here along these lines & I was convinced it didn't matter.

I would guess that the longer the span, the less that doing what you suggest would matter in terms of twisting force... or perhaps adjusting height placement of idlers/motors asymmetrically would support it. I just feel like it may complicate a proven design.

--
You received this message because you are subscribed to a topic in the Google Groups "3D Printer Tips, Tricks and Reviews" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/3dprintertipstricksreviews/bafawM8Vnwo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to 3dprintertipstricks...@googlegroups.com.

Joseph Chiu

unread,
Dec 18, 2014, 1:55:25 PM12/18/14
to 3dprintertips...@googlegroups.com
I've been playing Portal with my older son recently.  Sure would be nice if such things could be mounted on a flanged ring and mounted! :)

--
You received this message because you are subscribed to the Google Groups "3D Printer Tips, Tricks and Reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3dprintertipstricks...@googlegroups.com.

Ryan Carlyle

unread,
Dec 18, 2014, 5:00:46 PM12/18/14
to 3dprintertips...@googlegroups.com
So, things to consider:
- Belts wear dramatically faster when they exit flanged bearings and drive pulleys at an angle, so keeping the belt paths close to a plane is advantageous
- The traveling belt runs attached to the X-ends need to be as close to parallel to the Y rails as possible, so you don't get any weird lengthening/shortening skew effects as the gantry bridge moves front to back. But belt runs between fixed points can be safely tilted, twisted, etc without causing any skew
- The longer the belt run, the less force is needed to twist/angle it
- Putting the two belts on different planes adds some twisting moments to the gantry bridge
- The "typical" CoreXY X-ends we've been using have non-flanged idlers, so there's nothing to keep the belts centered except clean departure angles

Basically, the objectives of all the belt layout discussions are to minimize skew, minimize twisting/racking forces, minimize belt wear, and minimize parts count. There is no single "optimal" way to do that. Some people (like me) put the belts on two separate planes. Some people put them on the same plane through the traveling gantry runs and move off-plane for the fixed crossover. Diff'rent strokes.

Bradley Balach

unread,
Mar 19, 2015, 10:26:07 AM3/19/15
to 3dprintertips...@googlegroups.com


So what about a design like this where blue belts are high and red belts are low?

Viesturs Lācis

unread,
Mar 19, 2015, 10:38:35 AM3/19/15
to 3dprintertips...@googlegroups.com


On Thursday, March 19, 2015 at 4:26:07 PM UTC+2, Bradley Balach wrote:


So what about a design like this where blue belts are high and red belts are low?

No offence, but thanks for a good laugh, made my day :))
With tightened belts there will be somewhere around ZERO travel for the gantry, the only motion you will get is the carriage on the gantry. Because to move the gantry you will need to change the lengths of the belts (I thought that my colorblindess fooled me, but the placement of belts in your sketch was confirmed by another person as well).

Jetguy

unread,
Mar 19, 2015, 10:40:13 AM3/19/15
to 3dprintertips...@googlegroups.com
Entirely possible, just now takes up more vertical space for design of the frame. If done right- no big deal, it's just potentially more complicated to solve for all pulley mounting locations, mounting, etc.
In a drawing it's easy. I assure you in real life, this gets a lot more complicated to figure out mounting. In particular, have 2 belt planes in the X axis ends might get tricky. Also, don't forget attachment points at the X carriage- now over/under means that you might have interference keeping the nozzle as close to the plane as well. You solve one problem (belt crossing) and create 10 more new problems.

Bradley Balach

unread,
Mar 19, 2015, 11:03:19 AM3/19/15
to 3dprintertips...@googlegroups.com
Sorry if it wasn't clear the idlers are mounted on the gantry in this case.

Viesturs Lācis

unread,
Mar 19, 2015, 11:16:02 AM3/19/15
to 3dprintertips...@googlegroups.com
Well, I understood it that way. Now you confused me and now I fear that I have made a fool out of myself here. Do I understand correctly, that for gantry to move down (looking at the sketch), it will try to extend blue belt and when trying to move up, it will try to extend red belt thus both belts will actually keep the gantry from moving anywhere? What are those 2 lines on square perimeter - just a machine's frame?

Ryan Carlyle

unread,
Mar 19, 2015, 11:45:22 AM3/19/15
to 3dprintertips...@googlegroups.com
I'm with Viesturs, that belt arrangement as shown can only move in X, it's immobile on the Y axis.

Incidentally, I DO run my CoreXY with parallel non-crossing belt planes at the moment. Jetguy is right, it's more complex for no obvious benefit. I've tried it both ways, there's simply no visible performance difference. 

Bradley Balach

unread,
Mar 19, 2015, 11:45:30 AM3/19/15
to 3dprintertips...@googlegroups.com
Yes the frame
Reply all
Reply to author
Forward
0 new messages