This work proposes a methodology for designing power electronic converters called “Automatic Design for Manufacturing” (ADFM). This methodology proposes creating Power Converter Arrays (PCAs) using standardized converter cells. The approach is greatly inspired by the microelectronics integrated circuit design flow, power electronics building blocks, and multicell converters. To achieve the desired voltage/current specifications, the PCA conversion stage is made from the assembly of several Conversion-Standard Cells (CSCs) in series and/or parallel. The ADFM uses data-based models to simulate the behavior of a PCA with very little computational effort. These models require a special characterization approach to maximize the amount of knowledge while minimizing the amount of data. This approach consists of establishing an experiment plan to select the relevant measurements that contain the most information about the PCA technology, building an experimental setup that is capable of acquiring data automatically and using statistical learning to train models that can yield precise predictions. This work performed over 210 h of tests in nine different PCAs in order to gather data to the statistical models. The models predict the efficiency and converter temperature of several PCAs, and the accuracy is compared with real measurements. Finally, the models are employed to compare the performance of PCAs in a specific battery charging application.