In recent years the use of solid state frequency converters is rapidly increasing in the industrial and power plants where electric machines are installed, since it allows variable speed operations for the electric system, thus becoming a key factor which is capable of increasing the overall efficiency, as well as the global plant flexibility. The benefits that can be reached when electric motors are under investigation can be so enticing, in particular from the perspective of the operating machine coupled to the motor, that drawbacks take a back seat and can be overcome. On the contrary, power electronic for variable speed is generally not used in power generation, since the gains brought out by the present technology are not sufficient to attract anyone's interest. There are some exceptions in the wind- and the hydro power plant businesses, where there are some cases of variable speed already introduced. However in the former applications the output power is limited to some MW, while in the latter the converter is sized only to a fraction of the whole generating power, since it feeds the machine's rotor.This dissertation investigates on a possible electronic converter for power generation in the 40 MW range and above, which is also attracting from both the cost's and the efficiency's point of view, as well as from the operation reliability perspective and thus may be proposed as a breakthrough in this field. Among the possible frequency converters, those ones have been selected which perform natural commutations only, i.e. those that employ thyristors as the semiconductor devices, since the efficiency, the robustness and the relatively high performances are actually the key success factors to be addressed from the beginning. Well known matrix converter topologies will be referred to, as well as new matrix converter arrangements will be brought out and analyzed: their behaviours will be targeted by the dissertation, in particular when different strategies are adopted for controlling the operations. Specifically comparisons will be carried out when the same converter topology is used, but driven by either the well-known cycloconverter strategy or the newly brought out "active generator" one. In addition several generator winding arrangements will be proposed and analyzed, with the aim of increasing the actual feasibility of the proposed solution. Finally some test results on a sized down experimental rig will be analyzed and compared to the simulation outcomes.