Steps/mm Calibration

I think those numbers are posted on here somewhere, but if you run the tests in the video you should be able to tell. If you run the Easel setup for your machine, you should get to the default number meant for your machine.

I have to wonder if he is really over-rating his steps/mm because of backlash. I believe it is a fine balancing act to dial in steps/mm while compensating for the backlash that happens when you go back in the opposite direction. It would be wise to move in both directions to see if you have enough backlash compensation first before you start trying to finalize the steps/mm too finely. I am not at this point yet due to scheduling conflicts (work gets in the way of fun too often), so I would love to hear from someone who has already addressed this, and maybe see another video on the subject. Hey @Zach_Kaplan, wouldn’t an adjustment/setup/fine tuning video be a great Tip Jar Bounty?

1 Like

@makermark i’m gonna share this internally and discuss.

I bubbled up my results on my SO2 in another thread but will touch on them briefly here.

When running a calibration pattern, the square and circle were not exactly as they should have been. I didn’t keep my notes, so I’m going from memory here. My long axis is X, gantry is Y. I was able to test my X axis over a range of 40" and came up a tiny bit short. After some experimentation, a setting of $100=40.04 corrected the error. Before I checked the Y axis, I set $101=40.04 and got an exact measurement. For my Z axis, I had updated to a NEMA23 belt drive with ACME thread lead screw. The initial setting of 188.947 was to be reduced in half because of the 2/1 pulley arrangement, yielding 94.4735. I was getting more travel than commanded on Z so I adjusted the setting to $102=94.5 to get the correct travel.


@Zach_Kaplan - Here’s another thread to discuss:


I’m working on a calibration video for my X-Carve (Creating Calibration Guide). If your smart fellers have any specific numbers that shouldn’t change, please let me know and I’ll make sure to say “don’t change this, ever” in the video. I wouldn’t mind a Tip Jar Bounty though… :smile:


1 Like

Back when I was running the Shapeoko 2 and the TinyG I noticed I was not getting ROUND circles and everything else was squared and all that fun stuff on the mill checked out fine… I ended up using a .001" dial indicator mounted to the slotted bed to measure the SPU on the mill. I found the stock values were off on every axis. Once I got these dialed in for each axis, the circle came out far better. But as the TinyG did not allow for any sort of backlash to be allowed for, they were never 100% perfect. But calibrating each axis was a big improvement over leaving them stock. By the math, the values should be 40… But the real world measurements did not bear that out on the Shapeoko 2 or on the X Carve once I got it either. Now that I have switched to a Planet CNC MK3 controller, I am able to measure and allow for backlash as well as calibrating each axis since this is a feature on the controller. While the values are only SLIGHTLY different from the stock value of 40, and the backlash is only around .0550mm on the X and Y axis for the X-Carve… Getting those values calibrated over the stock setting has allowed me to greatly increase the repeatably of the mill from one project to the next. Just in doing my write up on milling of PCBs I took a break on night and decided to measure and factor in the backlash on the mill, then I reran the PCB mill job. The difference in the boards was easy to see with and without the backlash settings.

I can not offer a true explanation as to why there is a subtle difference in the real world measurements VS the “stock” numbers they should be at. But I know the real world difference the mill performs at since I have calibrated each axis with the dial indicator

I look forward to seeing where the debate on this lands and also very much what the guys at Inventables have to say and think about stock Vs measured SPU values.


I have to believe that the belts that come with the machine are only 2mm pitch when there is no stretch to them. Having only fiberglass tension members I have to believe that when they are under tension that the pitch increases just enough to cause the difference from the stock value of 40.


1 Like

That was my working theory also. Since there is not a true set way to make sure all the belts have the same tension from one mill to another would seem to account for the minor differences we all see from one mill to another.

Does anyone know if there is a belt type out there that does not have the stretch to it that would alleviate the belt issue?

If you can find Kevlar or steel reinforced GT2 2mm pitch 6mm wide belting let the rest of us know.

1 Like

I changed my belts to 9mm a few months ago and haven’t had a slip yet. I had to make my own belt clips but all other parts were available online.

1 Like

BillArnold - Interesting concept to use the 9mm. I may look into that. Not a bad price on amazon for the change over (~$26). What did you use for the belt clips? I assume you had to change the grbl settings also?

@AngusMcleod - Wider to fit the 9mm belt

@JeremyHill - I had some Lumicite (acrylic with fiberglass) in stock. I modeled my update after the original SO2 clip. No change in grbl since the configuration is the same as the 6mm belt.

Just a quick comment to the video’s author. The calibrations were correct on the X and Y, but the way you calibrated your Z was wrong. This is because after calling G92 (which isn’t actually needed. More on why not in a minute) you moved down to your work surface which was a pre-calibration move, which would then have thrown off all future measurements from zero because you don’t actually know you’re at true zero yet and definitely won’t be if your ratios are off.

I’ll over-exaggerate the numbers to better visualize the problem with how you did it. Say your touch-plate was 5mm thick but your $102 was off by a factor of 2 (was 80 when it should have been 40.) You touch off the top of the plate, then set G92 Z5 (because you know you should be exactly at Z5), then say ‘Go to zero’ but your tip plows into the surface of your machine thinking it had only moved 2.5mm so far. That’s because the ratio was off while you were moving to zero.

Say the reverse was true and your ratio was off by a factor of 0.5 (you have it at 20 and it should be at 40) and you said 'Go to zero. This time you’d stop at a point hovering 2.5mm above your work surface although the machine would be reporting you’re at zero even though clearly you’re not.

Both of those would be obvious that something was ‘off’ and you couldn’t just ‘measure from zero’ but say you’re only off by 0.1mm. You may not notice that and are measuring from a zero that’s actually 0.1 off as part of your calculations.

But as I said above, you don’t need to set zero to calculate your ratios anyway. You only need to be able to move a relative, but exact distance from a known point, and you have that known point–the top of your touch plate–thanks to your probe!

As such the proper way to calibrate Z is to probe down to the surface of your plate then simply issue a relative Z move (set G91 for relative coords, then issue M0 Z20 or whatever the actual thickness of the extrusion is, which you can easily measure with your tool) and measure where you are relative to the top of the extrusion as the extrusion sits on top of that plate, not the surface of your machine! Again, at this point, the surface of the plate is the only absolute, which is why neither the plate’s thickness, nor where zero actually is actually matters yet.

When you measure everything from the top of the touch plate, this guarantees all Z-axis movement is considered in the adjustment calculation, which is what your initial jog down to Z-zero missed and threw off.

Once you’ve set all your ratios, then go and re-probe, and set your G92 to identify zero because at that point, all ratios have been properly adjusted so you will be at true zero then.

Hope this helps!

(Note: I’ve copied and pasted this into the video page’s comments on YouTube so he’ll see it there as I’m not sure he even knows his video is linked here.)

Where and how do I access these commands to adjust my z?
My stock layers appear to be way too deep.
FYI I just got my unit up and running about a week ago and I am working to calibrate it

Hello…I am rizalyn. We have 30 inch cnc longmill at home and I’m having trouble with my machine because the machine locked up. And I dont know how to fix it. I have no idea what to. I hope someone can help me. Thank you.

Locked up in what way?

Perhaps the controller throws an error or is awaiting a homing unlock sequence, or status reset.
Try to send a $X or $H command through the controller.

It locked up when I install a gcode sender because I want to carve an image but it didn’t work. So when I go back to sign in my easel the machine didn’t work any was only having a sounds but not moving any more… so we can I find that controller? Thank you for your reply.

I would suggest making your own thread for clarity, where you in detail describe your hardware and workflow + where it is snagging.