Luis Arturo Martínez Alvarado

Director:

Dr. Jordi Madrenas Boadas

Reading day:

06 th July 2011 

Abstract:

This thesis presents a development methodology and optimization of reconfigurable mixed-signal integrated architectures. This includes the architecture analysis, modeling, mapping, experimental characterization of applications and finally the optimized architectures proposed. To carry out the analysis, a FPAA developed by the UPC AHA Research Group is used as a test bench. Chapter two presents a state of the art study on FPAAs, the translinear concept and high-level modeling. In chapter three the reconfigurable translinear cell (RTC), the basic building block of the FPAA is discussed in detail. Chapter four presents an acceleration methodology to optimize the development time of complex reconfigurable mixed integrated systems by means of the simulation with high level models and the simplification of configuration. Each one of the RTC modules was modeled in high level, quantifying the accuracy of the models and their incidence in simulation time reduction. In chapter five were performed mapping tests for several signal analog processing functions on the FPAA, including static and dynamic circuits. The results of the mapping, the experimental measures and the performance obtained for each application were presented. Based on the application mapping resultsand the experimental measurements, in chapter six the optimization and the reconfigurable translinear cell synthesis to circuit level were performed. Several circuits enhancements are proposed to improve the translinear element performance. Regarding architecture, we propose a basic translinear cell including two elements as a differential pair, minimizing non-ideal effects and allowing the reuse of auxiliary modules. These improvements provide an enhanced functionality with an increased silicon area of 29.1 % respect to the original cell. Finally, chapter seven presents the conclusions and future work of this thesis.