CoreXY explained: Comparison + strengths & weaknesses

CoreXY seems quite popular at the moment, with a range of printers now available from manufacturers. This is on top of great designs such as the Hypercube on Thingiverse. In this video, a comparison between CoreXY, various cartesian and the delta motion system is made, exploring the strengths and weaknesses of each.

A summary for CoreXY:
-Space efficient
-Rigid cube frame (typically)
-Two stepper motors sharing X/Y movement 99% of the time
-Fast and precise when used with a light print head, which is generally inherent with no moving bed for X and Y
-Belt tension is critical
-A rigid frame and pulley system is critical
-Getting these wrong may ruin the output of the printer, getting them right may raise the cost.

Thanks to Seckit3DP and Tronxy for responding to my question during the making of this video.

Take a look around and if you like what you see, please subscribe.

[Trikkie]

A very big advantage of coreXY that is totally underrated is that you can move the steppers outside of the heating chamber rather easily (due to XYZ and E) steppers being fixed in place. This allows for a stepperless volume inside of the frame and this allows for higher temperatures without watercooling.

[sennabullet]

I think I have watched this excellent video 5 times now. Whenever I forget how cartesian, delta and coreXY systems work and need a refresh...there is no better concise source than this video. Thank you again for making and sharing.

[JohnOCFII]

Another wonderfully educational video. I’ve got one Delta printer (a Kossel Mini), one Cartesian i3 style (a Prusa MK3S) and one CoreXY (RailCore 300ZL). I most enjoy watching the Delta, as it just looks so COOL. The Cartesian is a solid, steady workhorse. The CoreXY produces the best prints (beautiful layer stacking, for example). I do think it is easier to get very good prints from the Cartesian than from the others two types of printers, but each has their place.

[DrJeckyl]

We as consumers are the winners in this. Can't wait to get my Sekit.

[timomiller]

The manufacturers left out the most important part about the design which is not having to carry the mass of the x-axis stepper motor around along with the x-axis which reduces moved mass. The mere fact that the two steppers work together to move the printhead is not really an argument for corexy in my opinion as they also do that in regular Cartesian systems when moving diagonally - the same way a corexy uses only on motor when moving diagonally. It becomes more obvious when you imagine rotating the rectangular bed 45 degrees around the z-axis on those printers. It is funny how they seem to have used the design principle without fully understanding the main reason to use it.

[JWH]

That Scara (sp?) setup looked incredible, with a proper mechanical build and mounting I can see it being quiet versatile even beyond 3D printing.

[Jon-to]

Love your videos, they have been extremely helpful. I recently bought a Tronxy x5sa-400 and I'm having a heck of time getting it to print properly. I'm upgrading a number of the components on it so hopefully it will alleviate the sloppy prints I've been getting.

[ShasOAunLa]

CHOPPED!
I like that MCM shirt you wearing ;)

I started with 3d printing by designing my own corexy printer. now, in iteration 5 i tackeled to go to linear rails instead of rods and also getting a open frame design with no belts in the front. so far only good experiences.

[stevehanwright]

Mate I love the way you are so effective with you explanations. I have just bought a Voron 2.4 after using a Prusa for a few years.

[crosswalkguy]

I just bought a Bambu Labs X1-Carbon, and it is my first CoreXY printer. I was trying to figure out the belt system. Thank you for the great video. I love all of your vids!

[BVD]

Hi Michael, thanks for this particularly clear video explaining Core XY printers, and comparing them with the other motion systems. I've been curious about Core XY machines but haven't had a chance to dig into them in a meaningful way, but after seeing this, I'm a lot more intrigued by them. 👍

[matts]

I came home with a donated CoreXY "Ulti'crater" (?) from the free pile at Craigslist the other month and am very new to its configuration. I look to get it in the air for what I can with updated Marlin FW on its Arduino/RAMPS 1.4 electronics; it will be an ongoing project for a while, and we'll see how it goes. Great review - thank you.

[jhon]

I just recently learned there is a new motion system called Croxy. It is kind of a beefed-up version of a cartesian setup, but shares some of the same benefits of CoreXY.

Similarities:

1. CoreXY and Croxy both have a separate Z stepper motor.
2. CoreXY and Croxy both used fixed motors and benefit from lower moving mass on the gantry and the ability to isolate the motors from a heated chamber if necessary.
Differences:
1. Croxy uses four motors, two for X and two for Y, whereas CoreXY uses two. This makes Croxy more expensive, but also more powerful and simpler in some ways than CoreXY.
2. Croxy move calculations are simpler than CoreXY and should be completely identical to standard cartesian, just with two motors running X commands, and two motors running Y commands.
3. Croxy uses a belt routing path that is much simpler and shorter than CoreXY.
4. Croxy uses a crossed gantry design (two bars, one for X and one for Y). This is slightly more moving mass than CoreXY, but each axis having two motors, shouldn't be a problem.
5. CoreXY is weakest doing diagonal moves because, at a 45 degree angle, a single motor is doing all the work. A Croxy is weakest moving in a straight X or straight Y direction, but in either case, it has the power of two motors to perform the move. On a diagonal (longer) move, it has the power of all four motors.

[stephenwollenberg]

From an engineering prospective I found that one of the biggest weaknesses of the CoreXY is the that long belt length and tension distribution throughout the X & Y axis's. I built one of these and was disappointed with the result. I use my 3d prints for engineering projects, and found that the CoreXY struggles more to print with precision. If it is going to bother you that your print is going to come out with slightly oval holes instead of the circular holes in your design, I would recommend the Cartesian style over the CoreXY. If you know you are going to be printing mostly pieces of art that won't rely on a definite level of precision, you may benefit more from off loading one of the motors in the mechanics with the CoreXY. Honestly though, I feel that the drawbacks from carrying the extra motor with the Cartesian style printer is worth the sacrifice. You could always use a larger motor to compensate for the extra mass that needs to be carried. At the end of the day it's more important to me to have the quality that the Cartesian style provides, even if it takes slightly longer to pull the print off the bed. Never tried the Delta style due to all the problems I predict that I would deal with getting an accurate print. Firmware is getting better however to address some of these issues hopefully... eventually.., but that still would add additional calibration settings to deal with, and like mentioned these add additional calculations that will need to be made, but perhaps not an issue to concern you on a 32 bit board.

[BDP]

I enjoy my Core XY X5SA, I have modified it a bit to make it print stable at 100mm/s. Love the speed and that it prints the same quality as a 30mm/s print.

[vizionthing]

Error @ 12:30 "Kinematics are harder to crunch for the firmware"
coreXY moves on basic G1 commands like any other printer all handled fine on an 8 bit controller, the only time you need 32bit controllers is when you need to print at much faster rates or use the exotic G2 and G3 arc commands and has nothing to do with the coordinate system of the printer.

When converting G0 or G1 commands into x and y step calculations on coreXY the math functions are addition and subtraction along with division by two and that division can be handled with a simple binary shift, compare that with the calculations needed for a cartesian which will need to use sine and cosine functions and it would seem that your statement is fundamentally wrong.

[LoneTech]

A CoreXY printer doesn't actually need to do any more calculations; it's still doing cartesian movement, just in a 45 degree rotated reference frame. If you rotate the bed 45 degrees in your slicer, only the homing procedure would need to know anything different from a printer where the axis are aligned with the bed edges. And the math is ridiculously simple; A=X+Y, B=X-Y. Delta, Scara or polar movements are far more complex.

The part about both motors participating in a move is a bit of a double edged sword; not only is it only for moves aligned with X or Y axis, it means the motors are moving a different weight depending on whether it's moving mostly in Y or mostly in X. And they do so by varying tension along the Y axis differentially, so it's critical the X axis bar is rigidly perpendicular to the Y axis even as it moves along it.

The Ender 5 design instead uses a stronger Y motor to cope with the larger weight it moves and a common axle drives both Y belts to hold the X axis at a constant angle, even if it somehow isn't quite perpendicular to Y.

[fritzhusselmann]

Subtle correction on a super-awesome video: the Prusa i3-style mechanics didn't start with the Prusa Mendel, it started with the RepRap Mendel (of Bowyer, Oliver, et al. fame), which the Prusa Mendel is based on and ultimately kickstarted the home/DIY 3D printing game.

[CapinWinky]

I was a controls engineer that worked with industrial machinery and I specialized in complex motion, including robotics and unique kinematic systems. I've moved into management and to keep scratching that engineering itch, I'm looking to get into 3D printing. I just so happens that I work very closely with H-Bot, I-Bot, and Deltas and the design criteria for when you would use one over the other.

With an eye to cost optimization and increased reliability at a consumer level, I think the coreXY is an excellent choice of kinematics. You remove the cost of an axis for the delta and have a simple/robust mechanical system to hold a heavier head without needing larger motors to control it. At the scale of a printer, H-Bot has a disadvantage in the form of a moment placed on the beam. CoreXY exists specifically to cancel out this moment through the use of two belts. You could tackle that problem with more robust bearing/slide design on an H-Bot and get the same accuracy with a more reliable belt situation, but that may increase the size of the machine as you must place the bearings further from cross bar ends to fight the moment, limiting the crossbar travel in the perpendicular direction.

At some point, the physical properties of the print material are the speed limit; your prints would sag as new layers come too quickly to cool. With high quality belts, CoreXY could get there, but it would be easier to get a Delta there. Scara is nice for access, but it would need to be very ridged and mounted very rigidly to go at speed.

[prxZen]

What really stands out to me, is the required footprint for a heated enclosure. Might even be able to print pei and peek with this, if you find belts that can play ball with the heat at such lengths. Since the steppers are static, these could be mounted outside of the enclosure, greatly easing the cooling system design.