This paper proposes a discontinuous charging technique for switched-capacitor converters that improves the power conversion efficiency (PCE) at low power levels and extends the input power harvesting range at which high PCE is achievable. Discontinuous charging delivers current to energy storage only during clock non-overlap time. This enables tuning of the output current to minimize converter losses based on the available input power. Based on this fundamental result, an input power-aware, two-dimensional efficiency tracking technique for Wireless Sensor Network (WSN) nodes is presented. In addition to conventional switching frequency control, clock non-overlap time control is introduced to adaptively optimize PCE according to the sensed ambient power levels. The proposed technique is designed and simulated in 90nm CMOS with post-layout extraction. Under the same input and output conditions, the proposed system maintains at least 45% PCE at 4µW input power, as opposed to a conventional continuous system which requires at least 18.7µW to maintain the same PCE. Therefore, the input power harvesting range is extended by 1.5x. The technique is applied to a wearable WSN implementation utilizing the IEEE 802.15.4-compatible GreenNet communications protocol. This allows the node to meet specifications and achieve energy autonomy when deployed in environments where the input power is 49% lower than what is required for conventional operation.