"You should use larger incerement, at the integer number of mesh size." this is my suggestion. Please get the mesh size. for example, you can use specified override mesh, which you know the mesh size since you set it. or you can use the scrript below:
e1 is dataset: Introduction to Lumerical datasets
similarly for y and z if the FDE is in the other cross section. The sweep should use n*delta_x where "n" is an integer number.