I emailed Jim Cullins about this issue also. The last thing Jim had me do is add a grounding rod near the table and ground the plasma chassis and table with a 1 inch copper strap.
The problem is consistent. The THC300 sends up and down signals to Mach3 and Mach3 thinks it's moving the z axis, according to the DROs and It will quickly get to the maximum correction of +/- 1.000 inches, but when I physically measure the z on the table it is unchanged or at least no where near 1 inch of change.
The software and table function fine as a 3 axis mill including the z axis and I can plasma cut parts all day long without the THC as long as my material is flat.
Below is the simple GCode I'm using for testing.
N0030 G20 (Units: Inches)
N0040 G53 G90 G40
N150 M3
N200 G1 Y20 F90
N300 M5
N310 M30
Thx for any help.