Hello Guest it is March 09, 2021, 02:31:58 AM

Author Topic: After using the “Steps Per Unit” in “Axis Calibration” my CNC wont run properly  (Read 30914 times)

0 Members and 1 Guest are viewing this topic.

Offline ger21

  • *
  •  6,289 6,289
    • View Profile
    • The CNC Woodworker

Now I have another question now that the "Motor Tuning, Steps Per" is working.

I wrote GCode to draw a 1" square.  Results below.
Why is the 32,000 run way off size and a rectangle.  I was expecting it to be 2"  Any ideas.  I know this is off topic, and I know how to draw a 1" square now, but it doesn't make sense.

At 32,000 it draws a square that is 1.026" X 1.104  Not square
At 16,000 it draws a square that is 0.986"
At 8,000 it draws a square that is 0.489"
At 4,000 it draws a square that is 0.224"


Something is very wrong with the 32,000 setting. However, since it's the wrong setting anyway, I wouldn't worry about it. 16,000 is the correct setting, but it's off a little, so you need to adjust the number a little. If 16000 = .986, then 1" should be about 16,227 steps. Try that and see what you get.

2010 Screenset

JointCAM Dovetail and Box Joint software
You could also run the "Axis Calibration" feature on the Settings page.
(Hello Jim P. ::))
Problem is, his error is not constant, even if we ignore the 32000 setting.  If it were, he should be seeing 0.986, 0.493 and 0.247.  Something is very wrong there.  As he reduces the steps/unit, the error, as a percentage, is increasing! 


Ray L.
« Last Edit: September 03, 2008, 10:32:32 PM by Chaoticone »
Ray L.

Offline RICH

  • *
  •  7,419 7,419
    • View Profile
Change it to the 16227 as Ger 21 suggests

.986/1=16000/X     its a ratio to find how much to increase your steps
X=16000/.986 = 16227  that should very close and if not calculate it again ( probably not  linear)

( Another way is to make a longer travel if you can measure it accurately, but the inch travel should surfice)
« Last Edit: September 03, 2008, 09:58:41 PM by RICH »
I am not too concerned about the actual size of the squares that I ran at 4,000, 8,0000, 16,000 because I was using a ball point pen that was loosely floating inside a piece of copper pipe with a weight on top, and it had quite a bit of wobble which in itself will make the size of the squares inaccurate.  When I actually put a known size bit in and cut wood that will be the proof of the pudding.  And yes I will cut the largest square that will fit on my table.    Then run ¼ table size squares in each corner, this should be a good test.

The 4,000, 8,0000, 16,000 were the same in the X and Y direction, however when I upped the speed to 32,000 it made a rectangle, one of the posters said that to run at 32,000 you needed a PC with a high GHZ rate, I really think that was the problem when I tried to run at 32,000.  Anyway around 16,000 is the ballpark I need to be in not 32,000.

I really want to avoid using the Settings utility where Mach 3 takes the number you actually measure after a fixed move and it sets the step rate.  This utility has caused me many problems, the last time I used it the steppers would not jog more than a couple of revolutions when holding the job button down.   Another member also cautioned me about the utility.  When it is time to set the steps I will do it by trial and error until I find the magic number for my machine.

Today I installed the table hold down track and sacrifice table and made a crude dust shoe, and cut my first piece of wood with a ¼” Zip bit and a “V” shaped router bit.  Things worked as expected.

Tomorrow I start to work on my dust collector, dust, especially fine dust in the 2 3 micron range is a killer.

Thank you all again, I had all my questions answered.

Mr Chips
Mr. Chips
Retired 3M'r
"When it is time to set the steps I will do it by trial and error until I find the magic number for my machine" - That is a horrible idea!  There is one, and only one, correct number, and it is determiend by the hardware, and should be precisely known.  If it isn't, then *something* is wrong, and just finding a number that seems to work will not give you a reliable, properly operating machine.  The proper setting for your machine should be 16,000.  If that doesn't work right, then figure out why!

Ray L.
Ray L.

Offline jimpinder

  •  1,232 1,232
  • Wakefield, West Yorks, UK
    • View Profile

Mr Chips - I fail to see what your problem is, and how you can possibly put all these diffrent numbers into "steps per unit" and hope to get anywhere.

Steps per unit, and we are talking of Config/Motor Tuning - bottom left hand corner, is a finite fixed number. It is not something that can be measured, guessed at, adjusted or anything else. We have covered this many time n this forum - and we still get the same old arguments popping up.

Steps per unit is asking you to say how many pulses the computer has to put out to move the axis by 1 inch (if in inches) or 1mm (if in millimeters). You must calculate this.

The thing most people seem to have difficulty with is the pitch of the leadscrew. I converted my old manual machine and here the pitch was 1/10th inch, i.e. 10 turns per inch - easy.  You must find out accurately what the pitch of your leadscrew is. If you don't have a leadscrew - say a belt drive, it still comes down to the same thing - how many turns of the cog wheel produces 1 inch movement. If your leadscrew is in millimeters, then I would set up the whole machine in millimeters, it is easier.

Most motors are 1.8 degree per step these days, so therefore 200 pulses are required to turn them one rev. All drivers do microsteps , and 8 microsteps are the norm - so settle for 8, it gives a good combination of smoothness and power, and is well within the capabilities of the standard P.C. Multiply 8 by 200 = 1,600 steps per 1 rev of the motor.

If you have any gearing between the motor shaft and the leadscrew shaft, then this is added now - mine is 3 to 1 (steel cutting lathe) - so the equation is now 200 x 8 x 3 = 4,800. The only thing left is the leadscrew - I said mine is 10 turns per inch - therefore my result was 200 x 8 x 3 x 10 = 48,000 steps per inch.

This is a fixed number - it cannot be altered - particularly not to adjust your supposed shortfalls - the fault will lie elsewhere.

This is entered in Config/Motor tuning/(select the axis) - bottom left hand corner. Make sure after you have entered it you press the "Save Axis Setting" before you leave that axis, otherwise the figures will revert to the previous entries. Whilst on this page I would also set the speed at a modest 4 inches per minute and acceleration at 0.5. You want accuracy at the moment - not speed.

The way to check this for accuracy, if you must, is not to muck about with squares. Simply move you axis using the MDI line.
Make sure backlash compensation is OFF
With the table central on the machine, zero the axis DRO. Type in G0 X1 and the axis will move 1 inch. Do not try and measure this - it was only to get rid of backlash (and do not believe anybody who says they don't have backlash). Set up your measuring equipment and zero it (I use digital calipers) and then type in G0 X2. The axis will move another inch - and this you should measure and it should be 1 inch. (Give or take a couple of thous either way - because accurate measuring is impossible unless you have really specialist equipment.

If the answer is not 1 inch, then you have a problem - and you will have to come back to us.

While you have the measuring set up, the type in G0 X1, the axis will return 1 inch - only it won't - and if you check your measurement then the shortfall is backlash - make a note of it.

Repeat with the other axis.

You can then enter the backlash into Config/Backlash and switch on Backlash compensation. Set speed at say 50%. Backlash compensation sounds very disconcerting when you first use it, but it is accurate - and adds that little bit extra on each change of direction to take up any movement in the gears etc as the get their shoulders down to pushing the other way.

You could now try your 1 inch square.

You can now try motor tuning !

Gradually increase the speed on each axis. (Dont forget to save axis settings each time) You will reach a point where the motors stall as they start up - back off from this point to your last safe setting. You can try the same with acceleration. Don't try and get to the moon - I need accuracy - not super fast speed.

My motors, at first would not get past 4 inches per minute - 220Ncm motors, wired in series at 24 volts. I wired them in parallel and changed to Gecko drivers - and I can get up to 40 inches per minute  (although I have reduced this for accuracy - they keep stalling on a couple of tight spots on my lathe bed) and my latest improvement is using 36 volts for the drivers.

Setting up the "step per" and speed and acceleration and backlash is an exact science - not something you can find by trial and error.

Sorry about the big post - but I had to get that diatribe off my chest (again)

Not me driving the engine - I'm better looking.

Thanks for taking the time to explain accurately and in detail.  It is always best not to assume that I know anything, in this case you are correct.  He He

I am using Acme 1 start ½” 10 TPI, direct drive.  Stepper motors are 200 oz/in with a 24V PS.

My configuration is a fixed gantry so I don’t have to push a big heavy gantry around. 

Thanks for the detailed step by step instructions for eliminating backlash and getting accurate results on distance traveled. 

When you say “Gradually increase the speed on each axis.” Do you mean air cutting or cutting the typical material that I will be using?


Hager (AKA Mr.Chips)
Mr. Chips
Retired 3M'r
Here are the test procedure and results from measuring the X and Y axis.  As you can see all the results from moving 1” came out to be 0.996” on all three tests that were ran on each axis.  I am using ½” 10 TPI 1 start Acme rod.  All three axis were cut from one piece of rod.   This is not a high precision piece of iron.
After setup I did not have to touch the indicator, it came back to zero every time.

I will try and attach the spreadsheet with test details.

What do you think of the test results? 
What would be my next step?


Mr. Chips
Retired 3M'r
Hey I'm ready to cut but want to drill a hole through my work table.

When you use the Wizards in Mach 3 such as circle cut and hole drill, do you position the Z axis bit down and just kiss the surface to be cut then press the Zero axis button and then Press Start??

Thanks ???
Mr. Chips
Retired 3M'r