Axes not moving correct distance

Hey all,
What factors outside of Linuxcnc can cause the axes to move the wrong distance?

I say it this way because I just built a new Linuxcnc install with the new 2.8 image, clean configuration using pncconf, on a totally different PC. I don’t see how it could possibly be the computer or software at this point.

I have been chasing this problem for a long time and have not put my new machine into service yet because of it,

The X axis moves .150 inch more than the commanded distance
The Y axis moves .120 inch more

The interesting thing is that the pncconf built-in axis testing feature moves the axes even more than the final working config does. It was moving about 2.5 inches for a commanded 1 inch move.

I am using made up numbers that work for these values, changing them makes no difference in the moves.
Dirsetup 10000
Dirhold 10000
Steplen 5000
Stepspace 5000

Acceleration and maxvel changes also don’t affect the move distance.

This is a gantry style plasma table, XYYZ design
Mesa 7i96 board
Keiling 5056e drives
Rack and pinion drives with 3:1 reduction
Cheap Amazon multi-voltage power supplies
Good wiring in a nice solid enclosure
Shielded wiring



Any help is appreciated.
Thanks,
Jamie

Is the motion consistent?
That is, after multiple motions, does a return accurately to a specific location?

It is consistent.. I use a 2” dial indicator for testing and the indicator returns to zero every time.

Thanks!
Jamie

So the extra distance is the same regardless of the commanded distance or is it larger for longer distances?

If the latter is the case you would need to adjust your STEP_SCALE in the INI file.

That is a great question, I don’t remember so I will re-test in the morning.

Is it normal to have to “tune” the step_scale number? I used math to determine the correct step_scale, checked it with the gear reduction manufacturer, and both axes have identical motors, drives, and racks.

Thanks!
Jamie

The missing value we need to help is details of your pinion. What module? How many teeth? how far does it move for one revolution?

On one of my axes, I had to resort to trial and error.

I usually divide the commanded distance by the the actually measured distance travelled and multiply that with the STEP_SCALE used.

Thank you all for the help.
@rodw I am using the rack and pinion drives below. They have a 20 tooth stepper pulley and a 60 tooth pulley driving the rack gear.
www.finelineautomation.com/product/rack-and-pinion-drive/

With 8x micro stepping step_scale should be:
(8 micro steps per step * 200 steps per rev * 3 gear reduction) / (1in pitch diameter * pi) = 1,527.88745 steps per inch

@Aciera and PCW, after lots of testing I believe this is a step_scale problem. The longer moves have more error and it goes past 0 on the way back. See table below.

Is it common to have to tune step_scale even when the math should be absolute?

Thanks again,
Jamie

Is it common to have to tune step_scale even when the math should be absolute?

Certainly not unheard of.

As I mentioned earlier you can calculate your step scale quite easily now for each axis:

STEP_SCALE = STEP_SCALE x "Expected Table Position" / "Actual Table Position"

Great suggestion , I will give it a shot.