SO3: How do you calibrate X&Y axis to make perfect circles/squares?

Mine seems to be “better than 1/64 across 2 feet” and of course much more accurate over shorter distances, once calibrated.

Just speculating, but the dual y-axis motors/belts seem like a possible source of nonlinearity across the table. In a perfect world they would be calibrated separately.

My machine worked surprisingly well when the set screw for the left stepper motor was not tightened against the flat. It might be worth checking calibration with one side deliberately loose, and then the other side deliberately loose. Unmatched belts/tightness/steppers might (I am totally speculating here, someone who knows better feel free to correct me) result in different y-travel for the left and right sides of the gantry, which could show up as errors in both X and Y.

Right, accuracy needs depend on your application. I have a Nomad and an XXL, and so far I haven’t needed (really wanted, since I am just a hack hobbyist) to make anything larger than 8x8x3 which also required really tight tolerances. Nomad is not a kit: it is very accurate out of the box, and unlike Shapeoko you can be machining within an hour of signing for the UPS guy.

I am sure there are projects which require large work area and .001 accuracy over that large work area, but if I had a project like that I’d probably try to refactor the design first: maybe the precision parts can be made smaller than two feet wide and incorporated into a less finicky design. But that is pretty abstract: it really just depends on the particular need.

1 Like

It may sound like it, but I am not a complete nincompoop as I am a degreed mechanical engineer and worked in aerospace for 2 decades. However, I was a DESIGNER so sat behind a screen 99% of the time. During all that time I also had many hobbies and projects that involved building something so have what I would consider better than average hand skills. I left the aerospace career behind and became a full-time woodworker.

Like I said, I would be happy with +/- 0.010 but I admit I don’t know if that is out of the box accuracy (assuming due diligence setting up as described in the Carbide3D assy instructions) or if that takes more skill and tooling to get there. I don’t know where to start looking in order to rectify this beyond what I have already done.


  1. How square does the machine need to be?
  2. How tight do the belts need to be? If they are too tight, does that degrade accuracy or just wear out stuff sooner?
  3. What leads to inconsistent cutting behavior?
  4. How much do the v-wheels play into any of this (i.e. too tight or too loose)?

If the common answer is, well it depends, that is not helpful. I am looking for something definitive if at all possible.

  1. As square as possible.
  2. Tight, but not too tight. (Trial and error worked best for me here)
  3. Let’s talk about the dirty secret. The rails are extruded aluminum and as a designer, you might be familiar with the tolerance of extruded aluminum. It’s ugly, so sometimes rails are the problem, and still well within factory spec. I know you can’t see it in that pic I posted, but there are 2 pieces of paper under that aluminum plate. In THAT area of my machine the rails dip a little and 3 pieces of paper (~0.012 shim) makes it flat within 0.0005) over 6 inches. Getting to know your machine is a BIG help. I have areas that are very flat/square with no shimming (for my crazy needs) and other areas that I won’t use.
  4. The V-Wheels CAN develop a FLAT if over tightened and YES then can play into this. It’s a FINE line between tight enough and too tight. Again trial and error. I tighten mine, let them sit, and then I roll them checking for a flat spot. PS Dirty wheels, rails and belts also contribute.

Hope this helps Chis.

Edit: typo fix

1 Like
  1. As square as possible
  2. and
  3. loose belts, loose pulley set screws, maladjusted V-wheels, wrong feed/speed, incorrect cutting direction

Definitive is hard to do for a hobby-level machine w/ parts made to an affordable tolerance. It’s a kit:


The Shapeoko is a 3-axis CNC machine kit. It is intended to provide basic functionality while being simple to understand, assemble, and operate — all this at the lowest possible price. The stock configuration forgoes many features that are standard in more expensive machines (but many can be added as upgrades). This imposes limitations that the user has to work around. Despite this, or maybe because of this, many people with no prior experience with CNC, woodworking, or machining have successfully assembled and operated Shapeoko machines. Like most things in life, what you get out of it reflects the effort, preparation and patience you are willing to exert.

Regarding belt tension, I don’t see a recommendation for the newer 9mm belts. The video must be assuming a small SO3, not the XXL which weighs factors more (using the “lift one side of the machine with the belt” approach).

My primary goal here, after seeing significant inconsistency, is to FIRST achieve CONSISTENCY, FOLLOWED by ACCURACY. If I can get to a consistent +/- 0.01 then I will be pleased. If I can cut that in half, I will be happy. If I can cut that by 2/3, then I will be ecstatic.

You should not need to change the values of the steps per rotation, I think this is not going to help.

1 Like

Get a spring / luggage / fish scale?

Alternately, rig a weight and a pulley? (or flip machine upside down?)

or, do it the old-fashioned way — get as taut as possible by hand, then tighten using the new belt clamps — adjust as needed for even / adequate tension (be sure not to bend the motor shafts).

Once you’re up and running see

Here is how I set up my machine.


  1. How square does the machine need to be?

This is important because everything begins on the rails. If one of the rails is crooked the machine will run crooked. Get a machinist square and loosen all the Y rails in the 4 corners and bring them back into square. You may have to actually physically twist left to right on the machine to get it square depending on how out of square it was. Tighten the bolts up slowly in each corner while checking to make sure you stay in square. When you have them all tight check it all again. Then work on the X rail. Make sure it is square in the Y direction and the X direction. Adjust the screws accordingly. You wont have a great deal of movement in the mounting holes but you can move them all a small amount.

  1. How tight do the belts need to be? If they are too tight, does that degrade accuracy or just wear out stuff sooner?

My belts are tight. When you move the machine to the end of its travel, you should be able to raid the middle of the belt up about 1/2 inch. There are some videos on youtube showing how to calibrate them with a fishing scale by hooking the belt on the scale and pulling it up an inch or so and measuring the weight. I dont think that is necessary but if you want a good measurement you can do that.

  1. What leads to inconsistent cutting behavior?

Anything that is loose, not calibrated, not square etc. Check every single screw and make sure they are all tight. A pulley that is not tightened down will make for very erratic operations. Make sure the belt pulleys are aligned with the belt. Some have reported the X axis drive pulley has been found backwards. The belts should be straight and aligned with all pulleys. Check the V-wheels for cracks or flat spots. Make sure there is no debris on the rails. Ask yourself if the machine was a car and you were riding on it, what would be the result of that component rolling as the car drives down the road.

  1. How much do the v-wheels play into any of this (i.e. too tight or too loose)?

Can be huge. If too loose, they wont grip and you will slip steps, If too tight, the motors wont be able to move them on the rail correctly and you will lose steps. How I adjust mine is like this. One v-wheel is fixed on a screw and there is not adjustment. On v-wheel is on an eccentric which allows for adjustment. Usually the eccentric is on the bottom of the rail. With your machine off, the spindle should be able to move on all rails easily. After verifying this, take one of your fingers and stroke the eccentric v-wheel and see if you can move the spindle down the rail. If not, the eccentric is too loose. You want the eccentric v-wheel to barely touch the rail while allowing it to move the spindle down the rail. Too tight and you will bind the v-wheel. Play with this a little bit. You want it to be only as tight as your finger can easily move it.

And the last thing. It is completely likely that your spoil board is not level with your spindle. There are two issues here. First is the support in the middle of the table probably allows the spoilboard to sag. Once you get everything square, calibrated and adjusted you will want to so a spoilboard flattening to make the spindle perfectly aligned with the spoilboard.

After you do this, you may notice little ridges running either left to right or back to front. If so, this indicates you spindle is not square to the spoilboard. You may have to shim your router to correct this and then relevel you spoilboard.

I know this all sounds like a crazy amount of work. But if you will be patient with tuning the machine, and get it close, your accuracy will improve greatly.

Last but not least, dont get frustrated. We have all gone through some subset of this process and are here to help. Apollo frequents the boards also and the Carbide 3D team is awesome in customer service. Just ask and someone will give you some idea of what you need to do.

And make sure you get a machinist square, a dial indicator, a caliper and a set of 3-2-1 blocks. These will help you get it all dialed in.

Hope this helps and good luck. Remember this is no stupid question. If you have it, someone else does too. And we have likely all asked it previously.


Okay, so first thing I have done after all this is to reset the X and Y steps back to their default of 40.0. I then used a nice metal 24" ruler with fine graduations. I ended up changing the X step to 40.052 and the Y step to 39.948. I spot checked a few random dimensions to see if there is consistency in the result, seems to check out for now (but that is not in the thousandths of precision). I have not yet made a test cut to confirm anything.

Now what I am experiencing is a little odd. When I use CM to hit the various rapid position points, the LH side points all run up to the rail and hit without triggering the hard or soft limits. I have also changed the X max travel ($130=850.0). I have tried altering that value with no effect. I have turned off the machine, performed several homing cycles and after each one, the carriage slightly hits the LH support.

I went ahead and ordered up a few measuring tools I don’t have. In the meantime, I have checked all fasteners, readjusted all v-wheels and gave a real good overall look to see if there is something glaring that I’ve overlooked. Nothing stands out. It is square within about .015 as best as I can measure. Y-rails are parallel as I can measure. Y-rail heights are as equal as I can measure today (1/32" graduations, but I think I can eyeball 1/64). Gantry height to wasteboard is easily within 1/64" from side to side. I keep expecting to grab something and find it to be loose, but no.

I ran a few more test cuts and they are worse than when I had messed with the x and y factors more (remember I essentially took them back to default, but altered them slightly). I cut a square that should be 50 mm…it measures 47.47 in X and 49.32 in Y. Pathetic. Something has got to be way off in order to see this sort of inaccuracy.

I am at a loss here. Really don’t know what else to do until I get better measurement tools but even then I don’t expect to find much based on my review of everything a few minutes ago. I simply cannot believe it is this finicky just to achieve half-ass accuracy or there would be a ton of people saying the same thing. Something’s off but I’ll be damned if I can find it.


You are 100% sure that ALL the pulleys have set screws located on the motor (Shaft) flats?

I have verified both the Y motors and the Z motors, but of course I cannot see the X motor pulley. I thought I had checked that one too, but perhaps not, so that is the next thing, but looks like a few things have to come off to verify it.

Yup, all good. Some of the set screws might have been ever so slightly looser than I’d like, but I will try a test cut again just to see.

There has been a hell of a lot of messages on this thread. Let me just say that I am COMPLETELY satisfied with my system. With the homing sensors and new plate, I just go on my merry way. If you dig into it in great detail, you can find the limitations. But working with wood, no issues. You’re not making parts for SpaceX, so try not to compare to a machine that is top of the line.

Loose set screws on the motor shafts is a good thing to check regularly, I did have an issue with that once.

I never intended to sound like I was unhappy. I think it’s one of the best tools I ever bought and continues to get better.

1 Like

The X axis is your biggest error too. I am a little concerned that we have spent this much time trying to help you, and you didn’t check one of the very first thing that’s on the list? Ok put the fork in me, I’m done…

1 Like

Well, that must have been it. I will affirm that nothing was what I would consider loose. There’s certainly a difference between “loose” and “could be tighter”…it was the latter is all I can say. The X and Y values did still need to be tweaked a bit, but are much more in line with reason. Thanks for putting up with me, hopefully it was at least entertaining.

1 Like

So if I understand you right, it is working better? If so, can you post some pictures. Remember if you adjust anything you will want to recalibrate the steps/mm calculation with the ruler across you span.

I’ve found the set screws to be a significant gremlin. It would be better if the pulleys themselves had a flat spot to match the motor shaft flat: then the set screws would just be there to hold the pulley on, they wouldn’t be critical to machine alignment and performance. It might even be worth making a little insert or finding pulleys with a matching flat, if anyone makes such a thing.

I’m glad I’m not the only one who has to hear something several times before it sinks in, haha.

Not aware of pulleys w/ a matching flat. They’d be expensive / time r consuming to make.

Alternately a key way would work.

1 Like