CoreXY sensitive to belt tension?

[www.youtube.com]

it's claimed here that corexy is sensitive to belt tension compared to other designs.

anyone found this to be the case?

Edited 1 time(s). Last edit at 04/02/2016 07:14AM by bani.

Belt slop causing circles to be ovals is not dependent to corexy but to belt slop. Of course the longer the belt the more important is proper tension. I haven't got much of experience with corexy but this is my experience after working on ultimakers and solidoodles in the makerspace and at friends.

Edited 1 time(s). Last edit at 04/02/2016 12:53PM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

Nope. I tensioned belts to where they are neither too tight nor too loose and did not feel a need to do any further adjustments. That worked twice.

Although I think I know what he is talking about. He has a heavy direct drive extruder and a cheap smooth rod system. It all weights above 1 kg. Mine is even heavier and if I print at higher speed I can see circles looking like a hex. All this mass together with a high speed really shows the direction change on a print.

It is a hype that corexy is a high speed design. That is BS. Most of corexy printers have direct drive extruders and corexy has a whole X-axis moving (extruder, linear system of X -axis, bearings of Y axis). That is a lot of weight to move and it shows in prints although definitely not as bad on a prusa with moving bed.
Now with a linear rail X-axis and a bowden extruder you can minimize moving mass to 500 grams but that is still nowhere near to how little mass delta moves.

And corexy is just a solidoodle design without a motor sitting on X-axis. Nema17 weights 400 grams although I think that a smaller and lighter nema motor should be sufficient to drive extruder's carriage. That would be a reasonable weight penalty in exchange of not having to deal with very long belts and expensive or low quality pulleys (from my experience all bearings sold at reprap shops have play in them even when new and putting them on an undersized bolt does not make a nice pulley).

Edited 1 time(s). Last edit at 04/02/2016 02:28PM by Edvardas.

virtually 100% of the corexy designs i have seen are bowden.

I guess that is because you are planning to use a bowden.
I am a fan of direct drive and more than half of the designs I see have carriage mounted extruder We see what we want to see.
Oh and this comes from looking at designs on thingiverse.

i am looking at designs here and on openbuilds. virtually 100% of them are bowden by design.

it looks like the only ones using direct drive are from people who got frustrated with the original bowden design and converted to direct drive afterwards.

Quote
bani
[www.youtube.com]

it's claimed here that corexy is sensitive to belt tension compared to other designs.

anyone found this to be the case?

what i found is that not having equal tension on both belts resulted in torquing the x-assembly [the flat of the "H"]. not correcting that would result in, obviously, all square shapes ending up as trapeziums! apart from the damage that it would do to the linear bearings to have the x-assembly permanently being twisted.

in theory - in a perfect world with infinite tension on the belts and zero-friction bearings both on the idlers and the motors [and a few other minor conditions which detract from the point to enumerate] - you would need neither the y-rods, x-rods *or* their linear bearings. understanding the physics - that the forces cancel out - is really quite challenging.

so yeah, i am really very happy with the printer i've designed: corexy looked intuitively to me like a really good concept and i'm glad to have chosen it.

Quote
bani
i am looking at designs here and on openbuilds. virtually 100% of them are bowden by design.

it looks like the only ones using direct drive are from people who got frustrated with the original bowden design and converted to direct drive afterwards.

i have a mendel90: i really, really like it, and i even didn't notice how loud it was until i did the first prints on the Sandwich200. a month ago i managed to rack the mendel90 up to 200mm/sec and i've done approximately.... 100 metres of PLA with it, and the prints are *good quality* at 0.15mm (with a 0.4mm nozzle). i had to rack the temperature up to 210C to cope with the speed, which then resulted in jamming (squashing of the filament due to heat travelling up the extruder) - that was solved by running the fan 100% all the time.

now, in the new design (the Sandwich200) i decided to go with a bowden... but i did some asking around and some research, first. two things about bowdens i don't like: first, you waste about 0.8m of PLA on every single reel (basically the length of the bowden tube). secondly, even if you have it set up correctly, there's quite a bit of "lag". gcode software is sort-of solving this by doing relatively-huge, fast retractions of the filament (cura recommends about 4mm). my take on this is that the correct place to deal with this is in the firmware, to add "PID" to the extrusion fwd/reverse. that way you'd actually have the amount of filament coming out that you were actually, genuinely expecting and needing.

the other thing that people tend to do with bowdens is "direct drive". basically they're relying on the thinner diameter (1.75 instead of 3.0mm) reducing the nozzle pressure, as well as having a low-diameter gear (like the MK8 which is really, really low diameter), and that means that you don't need quite such a high torque generated by the motor.

unnnfortunately for me, i got a MK9 drive gear (which is something like 9mm OD!), and i have over 400m of PLA still around from the mendel90... all of it 3.0mm. soooo... when i made a direct-drive extruder for _that_ combination, it all went horribly pear-shaped very very quickly. even when ramping the current up to 1.38A (which caused all sorts of other problems), and even when going down to 1/4 micro-steps i was *still* getting skipping.... at only 20mm/sec extrusion rate!

that was so pathetic that i decided to create a geared bowden extruder instead. with a 46:13 (3.54:1) gear ratio on that MK9 drive gear, i can now - amazingly - get 200mm/sec flow in static tests, with 1/4 micro-steps, 225C on the PLA, and 1.2A current - these are 1.8 degrees per step high-torque (84oz-in) motors. at 220mm/sec the motor starts to skip. i'll be replacing the MK9 with a MK8 when it arrives in a few days, and i'll do the test again.

honestly, although it has been a pain, when i convert to 1.75mm i am *not* going to be using the 4mm OD tube. the bowden tube for 3mm is about a 6mm OD, and i like the fact that it's much stronger. i have a feeling that many of the problems people have with bowden 1.75mm is due to the thinner tubes, as well as the lack of force that direct-drive extruders can generate. in many extruder designs i saw people putting in "supports" for the tube on the way in and on the way out of the extruder, so that it goes dead-straight both in and out. any variation in the direction has a dramatic detrimental effect: much more force is required as the filament drags against the side-walls both before and after the gears.

so, i hope that you can see, you don't *have* to give up on the bowden concept, but to be honest i don't think people have really thought it through entirely and fully, so there are still developments and experiments being made. bottom line: do your research, find something you're happy with, and make time to experiment.

Bowden would work if the filament were exactly 1.75mm and the bowden tube were 1.76mm and there was no friction and the filament was inelastic.

It was a necessary compromise before flying extruders and flex3drives became available. Perhaps a rare earth magnet geared nema 11 extruder is light enough that we won't need bowden.

Get the best tuned bowden setup in the world and its blown out of the water by a averagely tuned direct extruder.

I've a Kossel Mini which had a long bowden tube and now has a flying extruder and a massive improvement in print quality. Same with corexy was bowden now flex3drive, same improvement.

Edited 1 time(s). Last edit at 04/07/2016 10:57AM by DjDemonD.

so if corexy is not a high speed design, then why would anyone use it over ultimaker type design?

Quote
bani
so if corexy is not a high speed design, then why would anyone use it over ultimaker type design?

I doubt that an ultimaker's gantry could carry an extruder without a lot of deflection on those smooth rods. So wanting a X-axis mounted extruder would be one reason.

And some people just prefer to shave off all the possible weight. For some people losing 400 grams (weight of a motor) off the X-axis is worth having to deal with long belts and a bunch of pulleys. Benefit of course would be a slightly higher acceleration and probably top speed which would result in a slightly faster print times at a same quality.

let's compare apples to apples. for a corexy bowden vs ultimaker (bowden), why would you choose corexy?

I've got a delta, a corexy and a 3 axis cartesian printer (kossel mini, smartrapcore alu, i3)

There is no doubt that the best design is the delta - stationary bed, stationary motors, lightweight print head its got it all going on. But they're a pain to setup, they're more delicate and they're a bit limiting on x-y size (but great in z - not that its all that useful most of the time), with flying extruder you get nearly the whole package (you can enclose it then as well, as there's no bowden tube flapping around.

The corexy is much faster than the cartesian machine, the kinematics are smoother and seem more efficient than with the cartesian, the frame is much more rigid being a cuboid made from aluminium extrusions, it has decent x, y and z build envelope, and stationary motors. You think they're easy to enclose but there still a lot of cabling arching over the top and into the head which is not going to be easy to build something around. If I swapped my 8mm rods for 12mm rods or linear rails I'd happily put a direct extruder on the print head, as it happens a flex3drive does the job beautifully - 85g direct extruder. I might try to make a flying extruder for this machine too, its simpler and cheaper and as effective.

In all honesty if I built a decent large enclosed delta (which is certainly harder to do) then I'd probably not need the other 2 machines, they would offer nothing extra.

cost????
could use duet hardware with control of the printer via a browser

---

appjaws - Core XYUV Duet Ethernet Duex5
firmware 3.1.1 Web Interface 3.1.1
Ormerod 1-converted to laser engraver, Duet wifi
OpenSCAD version 2020.07
slic3r-1.3.0, Simplify3D 4.1.2, Cura-4.4.1

Going back to the topic I printed a 20x20mm cube and it was 20.16 x 19.76. So I adjusted the belt on one side. I reprinted it this time is was 20.00 x 19.50 so it made no difference. I have not yet tried changing the x and y steps per mm to see the effect.

I have made a CoreXY that is direct drive.

Quote
bani
so if corexy is not a high speed design, then why would anyone use it over ultimaker type design?

compared to a reprap: much-reduced x-y dimensions. i.e. to be specific: greatly reduced extra surrounding space around the print bed.

let's look at the original reprap. the bed is moving, yes? so you now have an actual space required that's double the length of the printbed, plus a bit more. then, it's a dual-Z-screw design, yes? so you have to have room for two z-motors plus mountpoints plus support, that's 120mm left and right (!). i know some people have managed less than that, but it's still a heck of a lot.


so this explains why the mendel90 for example is 500mm x 450mm for a printbed size of only 214x214mm!

by contrast, i'm currently working out a design in my head where the top area will be about...270mm width x 290mm depth (most of that depth being taken up by the motors being at the back), on a printbed of the same size. there's no penalty in terms of an increase in height because the space used up by the hotend is now surrounded by belts and rods etc. that are now level with the hotend, whereas normally in a corexy design they would be approximately 20-30mm lower than the top of the hotend.

compared to an ultimaker - i've seen one, once - i don't believe the extra space around the printbed was anything like a 30-40mm gap, it was months ago but i seem to recall it was more like 50-60mm. from their website: Dimensions: 357 x 342 x 388 mm

the deltabots - they're circular! i hear that as long as you keep out of the way of the triangular bracing, i hear you can actually build parts that are outside of the dimensions of the build plate and the printer itself, but you'd need to first build rafts to do it. ... mehh

Quote
DjDemonD

It was a necessary compromise before flying extruders and flex3drives became available. Perhaps a rare earth magnet geared nema 11 extruder is light enough that we won't need bowden.

flying extruders are just.... going back to the idea of direct-drive, but instead of on rails it's... on the delta head. big deal. what's old is new. now, flex3drive, on the other hand, that's an *awesome* idea, i absolutely love it: use a flexible but strong tube to transfer the power of a stationary motor over to the printhead. cooool. so thank you for mentioning that, i will look at getting one as soon as i can

Well I'd suggest you try the flying extruder first, I bought a flex3drive for my kossel and didn't get round to fitting it before I tried the flying extruder. It's not on the effector, it's above the effector so it's mass and inertia only effects the movements on pure z moves, on x and y moves the effect of the added mass is very little as the extruder swivels. A z move uses all three motors, working together, so they are not as badly effected as you might imagine.

It's cheap and simple and it considerably improves the packaging of the delta printer.

The flex3drive is now on my corexy and its wonderful but its a more complicated and expensive piece of equipment. There are other limitations such as the flex drive itself cannot be cut to length or it loses its strength.

Edited 1 time(s). Last edit at 04/14/2016 06:22PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
DjDemonD

There is no doubt that the best design is the delta - stationary bed, stationary motors, lightweight print head its got it all going on. But they're a pain to setup, they're more delicate and they're a bit limiting on x-y size (but great in z - not that its all that useful most of the time), with flying extruder you get nearly the whole package (you can enclose it then as well, as there's no bowden tube flapping around.

Deltas have some very nice characteristics, but I think it's a bit of a stretch to claim they are the best. As I understand deltas, resolution varies depending on XY coordinates - the resolution is higher near the center of the bed and decreases as you move away from the center. How are deltas for printing exactly sized and shaped parts, particularly rectangular shaped objects? As you have pointed out, alignment and troubleshooting a delta is a little tricky. Cartesian machines of all flavors are a little easier to set up, and debugging print problems is usually pretty easy because errors in the prints tend to show you where to look for trouble. Ease of fixing problems is a big part of printer design.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

I base my opinion on owning and operating all three current, popular printer designs. Once the delta machine is properly configured and printing well the print quality and accuracy is very good, it will also print faster for the same quality than the i3 or to a lesser extent the smartrapcore alu machine. I do not print great numbers of objects right to the edges of the bed, but then neither do most people with a cartesian machine and a square bed. If I needed to regularly print objects close in size to my kossel mini's 20cm circular bed I would consider building a kossel XL with a 30cm bed and printing in the centre. I am thinking about this right now.

The addition of the flying extruder mod really brings moves the delta forwards, as their main limiting factor was previously the bowden extruder (especially with a long bowden tube). I refute any claims that a well tuned bowden set up can compete with a direct extruder also well tuned (or flying extruder or flex3drive system) in terms of filament control. Bowden could compete if it could have feedback at the nozzle to tune extrusion but it seems there is no means anyone has demonstrated, by which we can acquire that information at the present time. Or if the filament was a precise diameter, extremely inelastic, the bowden tube perfect diameter and friction was eliminated

I accept the point that building a delta with the level of precision required is harder than building a cartesian machine. A very small inaccuracy in the build results in major problems during calibration rather than the smaller and easier to detect problems that a misalignment might cause in the cartesian machine.

But imagine this is a case of producing consumer machines, not enthusiast machines, the building issues are not the consumer's problem, assuming the factory could use a jig to precisely align the delta's towers before shipping, and assuming you could ship it without it going out of alignment, then the other features of the design, the speed, the stationary motors, the fixed build plate, the ability to auto calibrate the kinematics (if not the extruder - as yet) would make it a good choice of design. Perhaps a machine with a one piece/unibody frame and build plate would solve this problem.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
DjDemonD
[direct extruder recommendations on a delta printer] It's cheap and simple and it considerably improves the packaging of the delta printer.

i won't ever get a delta printer. just to name a few factors: the accuracy required of the rods, the extra space (height-wise) and the round printbed are all "No" for me. i am printing parts mostly that are 230mm long, 12mm wide and 30mm high (libre laptop parts). to get what i need, i'd have to have at least a 280mm circular printbed, which goes immediately outside of the "low-cost" range. most large-diameter delta printers are also higher: i don't need height but they come with extra height anyway... so i'd be looking at a custom design. i *really* do not want to custom-design a delta printer, or even to be researching one. thought about it, and got completely turned-off by the combination of many different factors, all of which add up to "NO".

Quote

The flex3drive is now on my corexy and its wonderful but its a more complicated and expensive piece of equipment. There are other limitations such as the flex drive itself cannot be cut to length or it loses its strength.

cool. good to hear that you really like it - i'm getting one and it's going on a corexy (portable, folding) design. GBP 100 i wouldn't consider to be expensive, especially given what it is. i love the fact that it's a compact 40:1 gear ratio with specialised plastic gears that have had lubricant pre-added to the material. that just has "really good engineering" written all over it. i'd never consider cutting the flex drive: it's a precision designed part, and the length as-is is perfect.

Edited 1 time(s). Last edit at 04/15/2016 06:40PM by lkcl.

---

-- sandwich200: compact portable folding corexy printer [reprap.org]

Quote
the_digital_dentist
As I understand deltas, resolution varies depending on XY coordinates - the resolution is higher near the center of the bed and decreases as you move away from the center. .

... and i'm printing 230mm-long parts where the aesthetics are important, and there are critical-accuracy ends that have to fit into other parts. some of the parts are printed upright, and others are at right-angles. even *assessing* whether the 230mm long parts would print out well on a delta printer and fit together well is a headache! so... Big No

Quote
lkcl

Quote
the_digital_dentist
As I understand deltas, resolution varies depending on XY coordinates - the resolution is higher near the center of the bed and decreases as you move away from the center. .

... and i'm printing 230mm-long parts where the aesthetics are important, and there are critical-accuracy ends that have to fit into other parts. some of the parts are printed upright, and others are at right-angles. even *assessing* whether the 230mm long parts would print out well on a delta printer and fit together well is a headache! so... Big No

On a decently built, "square" delta, shouldn't be a problem. Last print I did on my delta was supposed to be 90.13 mm in length, and it actually came out to 90.17 mm. I can live with a 0.04% error, and I doubt you'll find much better on a cartesian.

While the constant rectangular <=> polar conversion makes my brain hurt, it doesn't bother the electronics on the printer that much (especially if you're running 32bit).

Most of the accuracy / detail complaints about delta printers are from poorly built printers that are overly reliant on on software calibration (my personal opinion).

Having said all that, I'm printing parts for my CoreXY on the delta, because while the delta is pretty, I want something a bit more industrial in terms of speed / volume.