**2. Proposed Topology and Operational Principle**

The proposed storage unit is comprised of two supercapacitors, a small (SCsmall), a larger one (SCbig) and a backup battery (Figure 2). The small supercapacitor is mandatory since it is the main storage element that provides power to the control unit. The large supercapacitor and the battery elements are considered optional and their integration on the unit depends on the needs of each specific application.

**Figure 2.** The topology of the proposed energy storage unit.

Specifically, SCsmall provides a regulated supply voltage with a 50 mV voltage ripple for the internal control circuits of the energy harvesting system and power output for connection of external loads (e.g., sensors, processing units, low-range RF modules). The 50 mV voltage ripple is selected to minimize the switching frequency — hence the power consumption — and at the same time, it is considered safe for most load types. SCbig offers an extra unregulated output to the system, for more power demanding loads (e.g., wide-range RF modules, GSM modules), since many off-the-shelf components operate in a wide supply voltage range. The voltage window of this output can be adjusted by the user regarding the specifications of the connected module, with a minimum window of 200 mV. To ensure the extended survival of systems that their continuous operation is critical a backup battery can be connected to the unit. Finally, a bleeder resistor is used to protect the system from excess input energy, which is activated whenever the supercapacitor(s) are fully charged.

The flowchart depicted in Figure 3 summarizes the operational principle of the unit which is described as follows:

**Self-Startup:** PMOS switches are used to control the charging of the supercapacitors (i.e., switches S1, S2 in Figure 2). Initially, all control signals are at zero potential since the control unit is inactive. Thus, the PMOS switches are ON, and energy is provided to both supercapacitors. As soon as the small supercapacitor voltage (VSCsmall) reaches a sufficient level, the control unit is activated, monitoring the charging of the supercapacitors.

**Charging:** Initially, the SCsmall supercapacitor starts to charge through switch S1. Meanwhile, SCbig remains disconnected as the main objective is the power-up of the system. The charging process continues until VSCsmall reaches a maximum threshold (Vmax1). At this state, SCbig begins to charge. The charging of SCbig continues until its voltage level reaches a high threshold value (Vmax2), but only if VSCsmall remains within a small voltage window (i.e., <sup>∆</sup>V<sup>1</sup> = Vmax1 − Vmin1 = 50 mV). This way, SCsmall has always charging priority and supply voltage regulation is achieved for the system internal control unit.

**Figure 3.** A flowchart of the proposed system operational logic.

**Loads Connection:** Two power outputs are available for load connection. A light load can be connected to SCsmall through the switch S1\_load. Granted that VSCsmall is above Vmin1, the load supply can be enabled. Power demanding modules can be connected to SCbig. If VSCbig is higher than a lower threshold (Vmin2), the load can be supplied through switch S2\_load.

**Preservation:** In the worst-case scenario, where input energy is not available, a backup battery can provide energy to SCsmall in order to sustain the operation of the system. If needed, the unit can be configured to provide energy to SCbig as well. The battery support is triggered if the supercapacitor's voltage level drops below the predefined thresholds, Vlow1 and Vlow2, respectively. These thresholds are set lower than Vmin1 and Vmin2, to avoid unnecessary battery activation and the external loads enabling during energy starvation periods. The charging priority feature is also applied here.

**Protection:** When the supercapacitors are fully charged, a bleeder resistor is connected to the input in order to dump any excess input energy and protect the system from overvoltage stresses.

**External control:** The charging and discharging thresholds for the supercapacitors are externally selected using resistor dividers. Large value resistor networks should be utilized, for ultra-low power consumption. Alternatively, digital-to-analog converters can be used along with a microcontroller unit (MCU), to dynamically change the thresholds, or even deactivate unnecessary modes by monitoring the available energy and voltage level at the input. Finally, the load outputs can be enabled or disabled by the MCU, through the en1 and en2 pins.

The selection of the supercapacitors values should be based on the needs of each specific application and be decided according to the available input energy and the load's demands. For instance, if long starvation periods are expected, large supercapacitors should be used, which may increase the required start-up time but will secure the extended survival of the system. Generally, demanding loads that can draw large instant currents during activation (e.g., wide-range RF transmitters) should be connected to SCbig. For SCsmall, a relatively low capacitance is recommended, as it provides fast startup to the system. Finally, for ultra-low-power systems the leakage current of the selected supercapacitors should be considered.

In the proposed design the maximum input current is set to 500 mA and the maximum output current is set to 100 mA. The working voltage thresholds of both supercapacitors can be set anywhere between 1.2 V and 3.6 V.
