Ballscrew Mapping to Having an Effect

So I'm trying to laser calibrate out the error on my ballscrew and I'm not having much luck. One of the quirks of my setup is I'm trying to maintain backward compatibility with the old control so my machine can run the same programs with little to no changes.

The machine's X/Y/Z are homed to the positive limit and all axis move in the negative direction from home. I can't put negative values in the first value of the octet or LinuxCNC gives me errors. So I tried moving the machine from home/limit+ to limit- and vice versa. The change in values after compensation are almost nothing. Like on the last value where the error is .00194" I get a change of maybe .0002". The results are repeatable over multiple runs both comped and not.

I'm trying to at least match the OEM controller that could get the machine within .0002" across 12".

Any help would be greatly appreciated. I'm running 2.8.1 of LinuxCNC

joint 0: compensation values must increase (X)

It created an error for each entry. Which was crazy when I had 118 entries.

Octets were
target out back
0.0 0.000 0.0002
-1.0 -1.003 -1.005
-2.0 etc.....

As soon as I made them positive LinuxCNC didn't complain.

I followed this thread in going down this path at one point:
forum.linuxcnc.org/38-general-linuxcnc-q...adscrew-compensation

The compensation file that resulted worked with linuxcnc. I eventually installed my old 5um scales to create a closed loop system so as to ge better tolerance on resulting parts. But that was more due to other limitations.

Cheers - j.

Well, I feel sheepish...

I'll try that!