Stepper calibration problem

This would be a good case for using a guitar tuner to check the tension of belts and be able to compare them to other folks belt tension.
If you measured with a 500mm distance between bridges and got say a key of C and the frequency for that key and I checked my machine at the same distance and got a higher or lower number I could adjust accordingly.

I am starting to think that Mikael has his belt way to tight.


Guitar tuners can be installed on your phone as an app. Check App Store if you have iPhone. Search for “guitar tuner”.


I´m using a 3mm engraving Vbit and measuring from 0 to 135mm on my calipers or 0 to 800 on my steel ruler. OR the second method I´m using is making profile cut squares with my 3mm flat endmill, and then dividing the 2 measurments and multiplying with the current steps i have the machine set to. Both methods give me perfect squares at the calibrated distance every time. You would think that the machine would be calibrated at that point but i guess thats not the case somehow.

I will go out to the workshop in an hour and start adjusting the belts from super loose to tight and calibrate the machine each time and see how it effects the accuracy on different distances.

Thanks for all the answers :grinning: lets get to the bottom of this, and help a few ppl with the same problem in the process.

As I noted above, you’re starting too early on cutting out shapes.

Drill holes first, then measure the drilled holes from center-to-center (or same edge to same edge). Use those measurements to calibrate steps/mm,

Then cut shapes or slots to calibrate for endmill diameter/runout only.

If you must cut out shapes to calibrate for steps, then you’ll need to use higher math to determine what the effective endmill diameter/runout is at the same time as you determine your actual movement.

Please see:

Mill multiple sizes and graph the results as actual/expected. Linear regression will reveal that there is a constant error (cut size not exactly equal to nominal cutter diameter) and a proportional error…

I have now managed to get the y axis working as it should, i get all distances down to 0,1mm and thats good enough for me. I Loosened the belts alot and it seemd to help.

But the x axis is still the problem, no matter what tightness on the belts and v wheels i cant get it to be accurate at both 35mm and 135mm (size of my calipers). And if i calibrate it to 700mm the error on short distances are even bigger.

So basicly if i calibrate the x axis by drilling holes 135mm apart and adjusting the steps acordingly so that at 135mm of movement i get exactly 135mm betweene the holes (same edge to same edge) the 35mm holes are closer to each other by 0,5mm and 150mm holes are 0,5mm further apart then they should be after calibration. So i can only get the x axis accurate at the distance i calibrate it for :confused: it seems like its not even linear if thats the word for it.

I´m just confused how i can get the y axis to be dead on, but not even close on the x axis.
I´m starting to think either the x axis stepper is broken in some way, or that i have to tweak some other setting than the steps for the x axis.

And also changing to power for the x axis motor on the arduino doesnt make any difference. The next step might be to swap the z and x motors around, but i would rather not since its kinda alot of work if there is a easyer solution I´m missing.

I’m inclined to agree w/ DavidSohlstrom that you’ve got a bad belt. Replace it.

My suggestion would be to go ahead and upgrade to 9mm wide belts — they seem to be a much greater improvement than the 150% increase in (already overspec) tensile strength would indicate. They look better too.

(oops dubble post)

And also, the x axis is way smoother than the y axis thats accurate when you push it by hand, so its not binding up either. Its rly smooth now after i loosend up the belt and the v wheels, you can push it with one finger from side to side.

Usually the Y-axis binding is caused by uneven belt tension.

Also be sure your set screws in your pulleys are tight — mark them and the shafts so as to be able to determine if they spin on the shaft?

They do not spin because its accurate at the distance i calibrate it for every time in both directions. It just seems like its not linear in some way when you try to move any other distance than the calibrated one. I will tighten the set screw again just to be sure tough.

And i find it hard to believe its a faulty belt, then it would not be accurate at all when you move it to one of the ends and do the same test. But it still the same when you drill holes that are the distance you calibrated it for. Its dead accurate on the calibrated distance, its just not linear when going any other distance.

I i know im repeating myself alot, but i just dont know why its doing this, when the y axis is perfect with the same calibration method.

Hmm i think the random behavior on my x axis vent away now. Not exactly sure why but.

The stepper motor started making a grinding noise mid testing/calibration. So i took off the belt, didnt see anything wrong. The grinding noise was still there, so i turned down the stepper power all the way and ran it a few times and it struggled to move as expected. Turned back up the power so it barely ran and then 2 more notches on the dial and it seems to have done miracles, and no more grinding noise.

I have no idea why its not all over the place anymore, but i could not be happier :smiley:
Fingers crossed its working now, but i will have to do more testing later and run a few jobs and see if its accurate. But the calibration tests seem fine now :smile:

Ended up with:
$100=40.194 (x, step/mm)
$101=40.135 (y, step/mm)

Most likely the current limit was too high or too low. By moving it you seem to have landed in the sweet spot. If your machine starts acting up again try adjusting the current limit again.

Just and update on the calibration, everything is accurate now but with very low tension on the belts so im getting some shatter when cutting acrylic with higher feeds and depth of cut (1mm 1000mm/min. So i guess if i want to get back to using my previous settings when cutting i have to start over again with the calibration and increase the power to the steppers to get higher belt tensions.

Also the best way for me to get a good calibration has been cutting a 35x35mm and 135x135mm square (the sizes im able to accuratly measure with calipers) with decent load on the machine and adjusting according to the 2 different cuts and either giving the steppers more power or loosening the belts/v-wheels while adjusting the steps until both squares are spot on. Just drilling holes and calibrating based on them has not worked out at all, since its not accurate when cutting actual jobs. Either the 135mm square is oversize or the 35mm square under sized based on the drilling calibration, witch would be way less time consuming if it was accurate.

If anyone got any ideas on how to make the calibration process less time consuming feel free to comment, since my method takes for ever and uses up quite a bit of acrylic in the process. But like i said the only method that i have come up with that has given any good results.

Edit: And having sharp endmills when calibrating and cutting jobs makes a huge difference on the accuracy with the x-carve when going for thous last 0,1mm:ers.

I know this is an old issue, but I recently upgraded the Z axis and I’m having this same problem with cutting 2 different dimension squares. But I’m using a 1/8 bit (the one form inventables). So… I tell the machine to cut an outside cut 76x125mm and then in the same carve an outside cut 50x50mm and I get a pretty close result on the bigger square, but in the small one I have about a 0.20mm short…

What would you suggest that I should do? I’m kind of frustated cause I wanted to make a small ring box and it’s messing with the tight fit “lid/box”.

Any help would be appreciated about your first step on this.


The tolerance stated for a “tuned” Xcarve is around 0.075-0.13mm

I´d simply fudge the bit diameter value to counter the difference.

1 Like