The insatiable demand to optimize and engineer new functionalities in electromagnetic devices has risen their geometrical complexity to unprecedented levels. This usually results in computationally multiscale problems with some tiny components of great importance to the overall behavior of these devices. Block meshing is a powerful technique for treating such problems. The usage of variable mesh size is adaptable for adequately representing the geometrical details without exhausting the computational resources with a uniform fine grid in the entire computational domain. In addition, block meshing allows the use of cubic cells only for which time-step in maximum and velocity error is minimal in each subregion. In this article, we present a mathematical formulation for stability and dispersion analysis when using block meshing in TLM method. These relations permit us to compute the maximum mesh-size and timestep that guarantee a tolerable level numerical dispersion, hence, minimizing the computational expenditures. Moreover, we study the case of adopting local time-step and demonstrate the origins of the instability that may appear in this case. Finally, some numerical experiments are presented to show the advantages of the proposed approach when using TLM block meshing. Similar procedure can be used for FDTD or FIT with block meshing.