Z height compensation problem (with GcodeRipper)

Hi all.
I'm having some problem with my z height compensation procedure.
I use Scorch's GcodeRipper to fix my gcode files.
I'm doing some work test on some PMMA
Situation: 3 gcode in sequence:
- 1st: a rounded rectangle dept 0.2 mm with end mill 1 mm
- 2nd: an engraved text dept 0,15 mm with 0.1 10° bit
- 3rd: some holes and the final cut with a 3.175 mm single flute.
My z probe is direct into the chuck and z zeroed (this way the probe offset is 0 for all the axis)
This is what I do for each file:
- open in GcodeRipper,set the total probe points and export the probe file.
- launch the probe file in linuxcnc and generate the points file
- load the points file in GcodeRipper, fix the original gcode and save the fixed one.
I don't zero the probe each time (i think is useless,isn't?)
Well,at this point, i mount the first bit (the end mill 1 mm) in the spindle, z zero the bit, load the 1st fixed file in linuxcnc and start the job.
The bit doesn't "carve" anything.....the first pass stays at about 0.6 mm from the surface. If i stop the spindle (and the job) and go to Z0,the bit touch normally the surface.
I tried twice to remake the "ripping" procedure,but no changes...
At the end i tried to "cheat": after z zeroing,i went to Z0 and manually re-zeroed Z, giving an offset of 0.5 and launched the job again.
I expected the bit to work propeperly (or at least to touch the surface),but nothing to do: the job goes exactly as the first time and at the same height (0.6 mm)
Honestly,i don't understand why this behaviour....
Any suggestion?

Andrea

Maybe found the issue.
Z speed to high.
'cause I'm using a microswitch based probe, I'm pretty sure the problem is related to bouncing / ghost trigger problems.
At low (very low....) speed (0.05 mm/s), the probe is almost reliable, but just at 0.5 mm/s it seems to be problematic...
Need to look for a solution....