Low-power system-on-chip (LPSoC) processors provide an interesting alternative as building blocks for future HPC systems due to their high energy efficiency. However, understanding their performance-energy trade-offs and minimizing the energy-to-solution for an application running across the heterogeneous devices of an LPSoC remains a challenge. In this paper, we describe our methodology for developing an energy model which may be used to predict the energy usage of application code executing on an LPSoC system under different frequency settings. For this paper, we focus only on the CPU. Performance and energy measurements are presented for different types of workloads on the NVIDIA Tegra TK1 and Tegra TX1 systems at varying frequencies. From these results, we provide insights on how to develop a model to predict energy usage at different frequencies for general workloads.