Stepper calibration problem

Hi, sorry for my bad English in advice :smile:

Been having some issues trying to calibrate the stepper motors, and I´m in need of some advice.

Basicly i can´t get my machine dialed in so it cuts accurate at short and long distances with the same stepper settings. For example i cut a 35mm x 35mm square and measured it with my calipers and adjusted it to the size it should be. Did a new test cut to confirm i calculated it right and its spot on. But when i try to cut a 200mm x 200mm square it cuts it oversize with about 1,8mm. (Can give you the exact messurments and stepper settings if needed, but I´m not in the workshop atm)

I have adjusted the steppers so they barely run on the arduino board, and also about 1/4 of a turn after that to see if it made any difference but its the same issue regardless. So I´m guessing the steppers aint that picky when it comes to power.

I have stiffened both the x and y rails, and I´m using a Dewalt d26200 router with the inventables mount.
Everything is running smooth and the cuts come out rly nice in both wood and acrylic that I´m mainly working with. Its just that i cant seems to get the accuracy I´m looking for between different sized jobs.

Hope someone could shine some light on this problem.

You should determine your steps per mm using the longest distance that you can achieve on your machine. That will give you the least error in position. Once the long distance is dialed in then the short distances should be ok.

PS English is fine, better than many.

I have dialed it in over 800mm aswell using my longest steelruler, but when i do that it still cuts all the small shapes undersized by as much as 2mm. And im trying to be as accurate as possible when setting the steps. I get it down to 0,1 - 0,05mm either on the small or big shapes, but not both at the same time :worried:

Is this issue related to the v-weels and the belt tension? I can move the gantry pretty easy by hand in both x and y axis and its smooth. I will try to loosen them so the barely make contact next i guess.

Was mainly wondering if there is some setting in the arduino i need to adjust other than the steps so get better accuracy.

Loosend all the wheels so they can be turned with two fingers with abit of force, and the belts are nice and tight like a guitar string. But i still cant get it to be accurate at both long and short distance at the same time.

I guess this is the case for all the x-carves, just have to live with it.
Going to have to adjust the steps acording to the job size :confused:

Is the dimensional error the same for all axes?

I don’t have the 1000mm unit so I can’t test long distances. My 500mm is pretty accurate. I ran one test where I used an ink pen to draw a figure 150mm x 150mm. The width of the ink pen ball is 1mm. I ran the job 24 times without turning the machine off on the same piece of paper and all of the lines were right on top of each other. I haven’t measured them yet, but they sure look like they’re 1mm.

This test doesn’t have any side load, so with a spindle in wood it might not perform as well.

Yea the machine is accurate in the sense of replicating the same job several times and it always comes out the same for me, the problem is that if i want to make lets say a box thats 35mm x 35mm in vcarve and one thats 150mm x 150mm i cant do that. I either have to adjust the steps for each pass or i have to account for that error when making the toolpaths, lets say +1mm for a 19mm hole and -1,5mm for a 50mm hole. Might be exadjurating with the actual error but its enough that i cant get a perfect fit when my machine is calibrated to 800mm for example, the error is that big for me. I would like to be able to cut out different sized shapes with around ±0,2mm let say that i could work with.

I´m mainly making dashboard panels and other hard to make parts for a production boat, with holes for different switches and gauges that are different for each customer. It would be nice so have the machine be calibrated so i would only have to draw each part onec with the dimensions listed on the parts that need to be installed, other than having to make several test cuts for each toolpath and adjust acordingly.

If someone got a solution it would be grate, and I´m pretty sure other ppl are having the same issues as i do.

I think the X-carve should be able to do the job you are trying to do.

There are people milling aluminum with good and accurate results. Now, it may be that working in wood you can’t get that accurate. There should be someone on the forum here that can answer that question for you. I haven’t started working it wood yet, so I can’t address that issue.

Well try making lets say a 19mm circle and a 130mm one (diameter) and messure it with calipers and see if they are off (assuming you have calibrated your machine). Its kinda frustrating when you have no idea how big the cuts you are drawing actually end up being even when you have calibrated the machine to its maximum length.

Either its impossible to make thies machines accurate across a wide range, or I´m missing some other setting than the steps.

And i think i forgot to add I´m making the test/calibration cuts with 500 mm/min and 0,5mm depth of cut in plywood so very low stress on the machine after all the stiffening mods on both x and y axis. Im getting no chatter or rough cuts in either wood or acrylic with 2x the speed and depth of cut for both materials that easle suggests for both materials with a stock machine after the mods and spindle upgrade. Cuts like a dream.

I believe I’ve done something similar, but I can’t find it right now. I’ll do another test later in the day, maybe tonight and post the results.

I’m going to take a shot here. Back you said that your belts are guitar string tight. Now here may be the problem. The belts have a fiberglass tension member in them so they will stretch when they are real tight. The belt pitch is 2mm with no stretch. As the tension on a belt increases the pitch of the belt will increase. This stretch may not be linear. So the pitch at the fixed end of the belt may be different than the pitch of the belt at the adjustable end of the belt.

Try this. move the spindle to the center of the machine and using the most accurate method you have check distances moved in X and Y for say 200mm adjust the step per mm until you get a move of 200mm in the center of travel. See how this differs from the standard 40 steps per mm.
Now move an axis to one end of travel and check distance moved. Is it different. Now move to the other side of the travel and check the distance. Is it the same as the other end or different.
Do the same for the other axis. Compare all of your results. what do you see.

Now we are going to check one more thing.

Make up 4 triangular bridge pieces from any hard wood say 10mm on a side and 15mm long. You will also need the use of a guitar tuner. If you own one great if not check with your friends that play guitar and see if you can borrow one. Last resort is buy one they are not real expensive.
Now that you have your equipment. Move the gantry to the fixed end of the belts. Put a bridge under the belts say 50mm from the smooth idler on each side and put another one say 50mm from the adjustment end if the belts. Check to see that the distance between the bridges is exactly the same for both sides.

Now with the tuner on, pluck a belt and note the key and or frequency of that belt. Then plunk the other side. Are they the same. If not lower the higher frequency to match.
If you are getting a fairly high key or frequency your belts may be to tight and you may want to ease off on the tension. This will reduce the amount of stretch and amount of pitch change in the belt.

Just use the tuner to get the belts the same tension. If you can put the bridges in the X axis and get the same distance between them then you can tension that belt to match the Y axis.

If you lower the belt tension then you would want to run the distance checks again and see if they have changed. If they have then we know for sure that belt tension has an effect on steps per mm.

My machine is still in pieces so I can not test this my self but when I get to the point of setting belt tension this is how I plan to do it.

Hope this helps


1 Like

You got some grate points, i will test more tomorrow and report back. I did not change the tension on the belts today when i adjusted all the v wheels as good as i could, spent about an hour making them all as close as i could.

I do not own a guitar tuner or know anyone with one, but i will spend alot of time making them as close as i can by ear tomorrow. The question is, how tight should they be? I got them fairly tight atm and they got a high pitch to them but they are for sure not the same on all axis. But i only tested and calibrated the x axis today to start of with so it didnt rly mather if they wear different.

And as you said i happend to calibrate the steps in the middle of the worktable, i will test if its off by the same amount at the ends after calibration on shorter movements. And also test many different tensions and report back.

Thank you for the response, i will dedicate alot of time on solving this and i am willing to work with you guys taking advice and reporting back.

If you want any information that is missing now or later just ask and i will provide it in text or picture :grinning:


where are you located in the X-Carve world?


I´m located in Europe, Finland to exact :sunglasses:

OK if you go to Ebay UK

You can pick up a tuner for very little money. That will take the guess work out of belt tension.


If it comes down to the belt tension i will for sure order one, thanks for the link.

We will see what kind of a differense tension makes after my testing tomorrow.

Just to check. You are measuring in such a way as to not be influenced by endmill diameter / runout?

Drill 3 holes in a right triangle as large as you can measure accurately, measure from edge-to edge on the holes (consistently, leftmost edge and leftmost edge, frontmost edge and frontmost edge).

Use that to set your steps/mm

Now carve out a circle-diamond-square test, measure it to determine your runout/effective endmill diameter — use that diameter in your CAM program for cuts made w/ that endmill.

Ok, I tried this. Using the default values for steps/mm from Inventables. The G code was generated by Vcarve Pro.

Both circles were correct +/- 0.1mm


Just wondering what your steps per mm are for X and Y during this test cut.


Actually, it’s not a cut. I still have my X-carve in the house and I’m using a 3D printed ink cartridge mount for testing.

Steps/mm are the defaults that Inventables sets up 40 steps/mm both X and Y. It works so well in my test setup that I haven’t bothered to do a calibration. That may change when I go to acrylic and wood.