If it is cutting out the correct parts in the wrong location, that would typically indicate a position loss in at least one axis during the rapid travel movements to and from cuts. This setting is defined by the Maximum Feedrate setting under the Control-Machine-Feedrate/Ramping page. The default value is 300 i/pm. You can try changing the speed down in increments of 25 and test the table by zeroing out the program coordinates and then running a point move test in which ever axis was loosing position. Run the table out the length of the axis in a rapid travel speed, and then do a point move back to program zero at a rapid travel speed. The machine should return dead on to where it started, if it does not, or if it binds up during any of these movements, the speed may be too high.
I am going to assume that you have already checked the gear rack for dust/debris as well as checked all the set screws on both axis to make sure nothing is loose.
The problem typically occurs at the higher speeds due to the stepper motor's torque curve, having the least amount of power at it's top speeds, allowing any kind of resistance in the table to cause the motor to bind up. This is either caused by the gear meshing with the gear rack too tightly, or tighter in one area then another. (Gear rack not level with the rail, bowed in the center, etc...) The Cam followers should be supporting the weight of the gantry, not the spur gears. The gears should just be sitting lightly in the rack. Also if the gears on the driveshaft are positioned so that when one is meshed the other is not, and you force the other gear into the rack, it can cause enough resistance to make the motors bind. Or possibly if the Cam followers are setup to hold the driveshaft at an angle to the gear rack, this would also cause enough resistance to make them bind up.
Mike @ Torchmate.com | www.Torchmate.com
Toll Free : (866) 571-1066 M-F 7:30am-4pm PST