# Core XY Explained

If you are building a CNC machine, a 3D printer, or even a plotter, you have a need for motion in both the X and Y directions. There are many ways to accomplish this, for example, some printers move the tool in the X direction and the bed in the Y direction while others move the entire X carriage in the Y direction and yet more use a delta mechanism. However, one of the oldest means of doing this is the Core XY method. It is interesting because both motors remain stationary and the business end moves entirely on belts or cords. This is similar to the H-Bot technique, but with some differences. [Michael Laws] has a video (see below) that explains how two stationary motors can move a tool anywhere in an XY region.

The idea behind Core XY goes back to at least old drafting tables. You can think of it as an object held by two ends of the same belt. As one end of the belt gets shorter the other end gets longer. The belts are arranged so that motion of one motor causes the tool to move at a 45 degree angle. That means you have to move both motors to go in a straight line.

There’s some distinct advantages to doing things this way, but — of course — also some trade offs. There’s no perfect world and every choice you make will change the design of your machine in some way.

The video also talks about some other mechanisms, including delta printers. Controlling these printers takes a little extra math, although the math of a core XY mechanism isn’t all that tricky.

We have looked at a few machines that use core XY before. We’ve also compared the mechanism to the similar H-Bot design.

## 28 thoughts on “Core XY Explained”

1. Saabman says:

I did think initially that was how an “etch a sketch” mechanism worked as the two “motors” remained stationary but that’s not the case as one control moves left/right the moves up/down. I need to pull one apart

1. CityZen says:

The Etch-a-Sketch mechanism more closely resembles that of the Ultimaker, which uses cross beams to move the printhead, which always stays at the intersection of the beams. The beams are moved by circular belts at both sides. The Etch-a-Sketch uses wires instead of belts, and they have a somewhat more complex path to follow.

2. Neolker says:

I am really looking forward to Prusa’s Printer XL based on CoreXY

3. Ginger says:

Why is the article picture clearly NOT CoreXY?!

1. mrehorst says:

I was wondering the same thing…

4. curious_one says:

So to get all the benefits and none of the drawbacks of a core xy on my Monoprice Maker Ultimate (carriage driven by independent x and y steppers, z moves bed vertically), all I have to do is rotate my prints 45° before exporting from Slic3r. Benefits being x and y are both used to move the nozzle (this happens anyway, and you can’t really claim it’s unique to core xy) and I don’t get the drawbacks associated with extremely long stretchy belts. Core XY seems extremely over hyped.

1. Ah, shoot. Sorry, I clicked on report instead of reply… too early in the morning!

Exactly my thoughts as well! The following comment was posted under a youtube video by me:

Core XY printers are not different to cartesian printers like the Ultimaker or the Wanhao Duplicator! The motion system of Core XY printers is rotated by 45° in respect to a “normal” cartesian printer. However, this does not make them superior to cartesians. The work load is not better distributed to the two motors than on other cartesians like the Ultimaker which is therefore not the reason why they can print faster etc.
Remember, that you have two degrees of freedom for x- and y-motion and therefore you need two independent motors. You can’t magically combine two independent motors which control two independent perpendicular axis to move the print head in one direction faster than with only one motor. Think about it, this is simple vector algebra.

2. Stephan says:

In my opinion the actual benefit is that both motors are stationary. Thus the overall weight of the carriage is lower, thus you can build more lightweight / live with flimsier motors / accelerate faster with same motors resulting in better corners etc. On the cons – next to the long belts – the cross connections of the carriage need to be very stiff to avoid canting which leads to reduced accuracy. And to really benefit of the weight loss you should use a bowden extruder which has its own drawbacks. Personally I don’t think these advantages overcome the drawbacks.

1. True, this is the only valid argument. However, the Ultimaker, Wanhao Duplicator or the Zortrax M1 does have a belt system where the motors stay stationary, too. So their mechanics is also very lightweight and is not relying on a Core XY system. In my opinion, a Core XY system is not superior to a cartesian system of a Ultimaker-like. Additionally, I suppose that the belts of a Ultimaker-like system are much shorter than of a Core XY system, making belt tensioning etc. easier.

1. CityZen says:

The Ultimaker setup is nice, but it is more complex with more moving parts. As a result, it also doesn’t scale quite as easily/cheaply. It requires 3 precise rails per axis (2 fixed, one moving), whereas core xy only requires 3 in all (2 for Y, one for X).

2. CityZen says:

Your point about the cross connections of the carriage needing to be stiff is not correct. It only applies to H-bot, not core XY. That is the whole reason why core XY was invented to replace H-bot. It balances the forces acting on the carriage such that the canting you mention is avoided.

5. curious_one says:

My theory on the alleged superiority of core xy really has little to do with the drive belt configuration and more to do with having to build a rigid frame to support it, which in turn is good for any 3d printer, regardless of configuration.

6. RetiredHobgoblin says:

Starts actually talking about CoreXY at 6 minutes.
I will never get why youtube presenters think they have to yak about unrelated shit all the time.

1. watchm4ker says:

In all seriousness, to increase watch time. One of the goals of YouTube’s automated ranking system is “How long does watching this video make people watch YouTube videos?” Same reason why they ask people to both like the video and post a comment, as one of the early goals was to maximize ‘engagement’ with YouTube.

7. If the problem is things like wobble, why is closed loop control not standard? The ones that do use it(If the kickstarters are honest….) seem to have pretty good results. Mechanics is hard to get right, and when you do, it gets heavy and complicated.

1. fonz says:

closed loop does nothing to fix flimsy wobbly mechanics

1. People are able to do precise artwork by hand with all kinds of imperfectly-ridgid tools, so it clearly is possible to compensate for such things given enough information. Sensors on the motor might not help, but knowing the absolute XYZ position of the head probably will.

2. Stephan says:

Closed loop does usually not help with wobble. Wobble occurs because of the frame being not rigid enough, bent leadscrews, pulley bores not centered / to big for the axle etc. Closed loop control of the motor position does not help at all (it only can avoid loosing steps). Closed loop control of the absolute position of the axis (like the big heavy milling machines do) can help with some wobble but i have not yet seen that on a 3D printer.
Closed loop is important for the big machines where you would like to use DC motors instead of steppers (as their power to weight ratio is much better than steppes). Closed loop on a stepper does save the job in that it can correct an error (i.e. loosing steps) but the error still happens and still can cause inaccuracy of the parts etc. +
In the end, for me a stepper loosing steps is simply to flimsy for the application. If a one looses steps on my machines, there was a different problem somewhere else before (extruder cooled / broken, part coming releasing from the bed etc.).

1. It doesn’t seem like there’s any particular reason you can’t put an accelerometer on the printhead and the bed and compensate for deviations from what you expect. Real linear absolute position sensing might be too expensive for consumer stuff, but you probably don’t actually need it.

1. ThisGuy says:

MEMS accelerometers are FAR too noisy and inaccurate to use for that application. The movements and accelerations are pretty small. Piezo accelerometers might get you closer but the accurate ones are way too expensive for any of the possible gains. A few hundred euros/dollars can get you a lot of added frame and drivetrain stiffness that will be far more effective at solving the problem

8. An additional difference… increased belt wear as the pulleys are orientated vertically, making the belts want to migrate down, relying on puley end flanges to keep them in place.

1. CityZen says:

I think this is a negligible factor, given the lightness and sturdiness of the belts in question.

9. Dhawal says:

Why dont yjey rotate the xy system by 45 deg during g code processing. Then core xy could be used as any normal xy system

1. Dhawal says:

*they

2. CityZen says:

I’d expect this is exactly what is done.

10. MT says:

fun fact.. when you take aerospace classes, and as much math as you have to for said classes, you learn that core xy, delta, and the popular bed slingers are all cartesian because they all use the cartesian plan. Possible exception is the delta which can at times use polar plan. Kinematics is defined as the branch of mechanics concerned with the motion of objects without reference to the forces which cause the motion. So the saying “core xy printers use different kinematics” is actually false. It uses the same physics as the others, just a different physical setup. On this planet, motion is the same, it is defined as displacement over time.

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.