New user, with a strange Y axes problem. XXL

I finely got time to work with my XXL. After squaring the frame and Z, I checked my axes travels. The X and Z are fine and repeatable. BUT the Y is anything but.
One would expect to see equal movement + or – from 0. As with the X and Y. Moving it in 0.01 increments there is about a 0.005 difference with a change in direction from 0.
Moving +, zero dial, moving + good. Switch direction we lose 0.005. Works the same when moving - and resetting 0 and continue to move -.
Switched to moving it in 0.1 increments, there is still about a 0.005 change, when changing direction. Both seem to be sum what repeatable. I Checked belt tension and v wheel, both are good.
Could this be a belt problem?

1 Like

Sounds like a little backlash somewhere.
Have you checked your Y motors pulley set screws, no play there ?

EDIT: I remember a great post from @WillAdams about belt tolerances and motor step error (1 step = 1/40 mm = 0.001"), can’t find it right now


marked the pulleys and shafts for a quick check and there is no slippage.

Hello something that catches us all out at the begining is making sure you flash the correct baseline grbl setting under CM make sure you have selected XXL and see if that helps. Though given you description above it can at least be ruled out as not it!:grin:


I thought the same thing and reloaded the the machine settings, with no avail. one thing i also noticed, at times when moving the motors, they will sound like they move but the dial wont move. missing steps? also once in a while they will double the movement instead of moving .01 they will move .02. I have not tested it running in MM. i will try that after work today. i will also try 2 dials on both sides to see if one motor is moving and the other is not. i also need to get some metric dials, i hate to convert from imperial to metric measurements.

Is it possible that the error is not to do with the machine but with the dial and/or how it is mounted? Just a thought…

If I were you, I would use larger distances to draw any kind of conclusion on accuracy (very small movements are tricky to measure correctly). That’s what matters in the end, since you will cutting pieces larger than 0.1" (probably, on a Shapeoko).

If you zero your dial, move Y by half an inch, then back by half an inch, how close is it to zero ?
I’m just trying to say you could drive yourself crazy chasing micro-errors everywhere (don’t ask me how I know), when in the end what matters if the final dimension of your piece, and there’s more to controlling that than “just” the mechanical accuracy. Acceleration, runout, deflection, the material deforming during the cut, all that kind of interesting stuff.


using two indicators, the are reading the same measurements. updated from Carbide Motion version 4.0.417 to Carbide Motion version 4.0.431. same results but now the X travel is two much with my dust boot on. that aside, i still loose 0.004 to 0.005 when changing direction. seams that with my machine this is normal. ??? :confused:

Something has to be moving or loose imo. On a well tuned machine, travel repeatability tests should be within 0.001. But like @Julien said, large movements are easier to measure.

5 thou is quite a bit

1 Like

As noted, the machine can only move based on full steps and micro-steps, and it’s pointless to try to measure/calibrate until one has verified the machine is mechanically sound (pulley set screws, belt tension, V wheel eccentric nut adjustments) and square (trying to measure a lozenge when expecting a square is an exercise in futility).

1 full revolution, divided by 200 steps, with a pulley circumference of 40mm, w/ 2mm pitch (total of 20 teeth), so at full steps one can move along a grid of 1/5th of a mm, and at 8x micro-stepping, 1/40th.

If you calibrate for belt stretch, then those dimensions shift, and your full step and other dimensions become this maddening set of interstitial grid points — the folks who do the best work at the smallest scale will often leave their settings at 40 steps/mm, machine a prototype, measure it, then adjust the CAD file to adjust for the actual size cut — repeat that to verify (sometimes further adjustment will be needed). It might even be that someone has a spreadsheet which notes the actual placement of various machine movements.

All of this suggests a testing methodology:

  • home the machine (in its initial state of 40 steps/mm)
  • move to some point where the actual movement of cutting out a pocket for the Probe will be in full steps (you’ll need to account for the 0.35mm difference of the diameter) and cut a pocket for the Probe — metric tooling would make this sort of thing easier, so long as the diameters/radii match up to grid points
  • place the Probe there and set zero against it — ideally this will be essentially unchanged relative to the previously set origin
  • Probe the other edges of the Probe and based on the distances, one should then be able to calibrate for belt stretch

(and yes, this was part of why I recently ordered a full set of metric collets from Elaire Corp.)


I made a video showing what I’m talking about. movement is in .01 inch. Remember I’m still new to this. I’m I overthinking this?

It tells me the video is private ?

new to youtube, set it to public

has anyone tried to view the video ? i thought i put it to public.

You sure belt tension is exactly the same on each side? It’s weird to have the X and Z fine but off in the Y. I’ve had some funny behavior when a tooth off on one side.

Have you tired cutting anything and measuring the actual results?

1 Like

I have been doing some reading, Julien has a good article called Shapeoko CNC A to Z. I’m going to tare it apart and recheck everything. While I’m at it I’m going to raise it up off the bench a little and add a center support to the bed. also I’m make a one piece bed rather that the 2 piece. I will report back on any changes positive or negative.

That action alone makes it worth the effort, to avoid the very common “sag in the middle of my XXL”.

Also, I think @Vince.Fab may have a lead about checking whether both of your Y-belts are tensioned equally ?

I know you said you checked your belts already. FWIW, here’s how tight my Y belts are (this is on my tiny SO3, I cannot slide a finger under it with the gantry pushed all the way back and lifting it in the middle, and it goes twaaang. it won’t be exactly the same on a XXL, due to the longer belt length)

1 Like

the table is now less than 1/16 of an inch from perfectly square. re tightened all belts, just like the video Julien added. rechecked the y travel. in 0.1 movement. 0 dial after y+, moved it y- by 0.1 and we are short by 0.006, moved it again y- and it is moving 0.10, back y+ short .006 and after moving y+ we are back to moving 0.10. this is not a mechanical problem with the belts pulleys or machine square. what to do next ???

My vote is still for a Cut/accuracy test (with spring passes) and if you aren’t satisfied with that then like Will said, you can adjust steps.

Every cnc router I’ve seen and even Tormachs, don’t advertise better than a ± 0.005 tolerance. Is this 0.006 going to be a deal breaker for your parts?


going to cut some holes for the wast board mounting and get a washboard mounted,then i will try a 123 block cut out. wish me luck.