A Lorentz gas consists of a set of non-interacting point particles that move along a certain type of trajectory (for example, straight or circular trajectories) and a set of fixed obstacles in space with which particles collide specularly. The obstacles can be distributed in different ways. Probably the simplest nontrivial case is where the trajectories are straight and the obstacles are 2D disks distributed periodically. This system is equivalent to the Sinai billiard, which was proved to be chaotic. If we consider circular trajectories instead of straight ones on this periodic Lorentz gas, some trajectories become non-chaotic. A simple way to find if there are non-chaotic trajectories is with a Poincaré map. A chaotic trajectory will fill most of the phase space, except maybe for some "islands" where drifting trajectories appear, which move effectively in only one direction with a constant velocity. Because the islands have a positive measure, the diffusion behavior is ballistic, i.e. <x^2(t)> ~ t^2, where <x^2(t)> is the mean square displacement, and t the time. If the islands disappear, then the diffusion becomes normal, i.e. <x^2(t)> ~ t. When the obstacles have a Poisson distribution, the diffusion is always normal if there is no localization of particles. If the density of obstacles is high enough, then the maximum diffusion coefficient is for a radius of trajectories different from infinity. Quasiperiodic arrays of obstacles have an angular symmetry as the periodic arrays, but there is no longer a translational symmetry as with random distributions. In this talk, we will first summarize some of results for the periodic and random arrays of obstacles with circular trajectories, and then we will show numerical computations of the diffusion coefficient for a Lorentz gas with quasiperiodic array of obstacles and circular trajectories. The obtained results are unexpected for high densities of obstacles, where we find more than one local maximum in the diffusion coefficient. We also studied a Poincaré map, finding in some cases islands similar to the periodic case. However, those islands do not correspond to drifting trajectories. Contrary to the periodic case, all trajectories produce normal diffusion.