Introduction: This paper is the product of the research “Generation of a modular electronic platform for the conformation of a WSN” developed in the EAFIT University in 2019.

Problem: Wireless Sensor Networks (WSNs) are used as a technology of information and communication (TIC) to automate processes. Their implementation is considered non-efficient because of their high cost concerning the requirement to replace node batteries.

Objective: The objective of the research is to develop more efficient solar energy harvesting systems that can guarantee an average performance of the wireless sensor node at a low cost.

Methodology: Here, we present the design and the implementation of a solar energy harvesting system that integrates a buck converter, a maximum power point tracking (MPPT) control, and a wireless sensor node. We also measure the voltage at the output of the buck converter and provide an estimate of energy autonomy in the wireless sensor node.

Results: Autonomous operation with the solar energy harvesting system was achieved for the node, sending packages each 20 min.

Conclusion: This article presented a solar energy harvesting system with the implementation of an MPPT control.

Originality: Through this research, parameters of the MPPT control are formulated for the first time for the solar energy harvesting system design, based on the duty cycle limits.

Limitations: The availability of the devices used in the implementation of the proposed design.

Wireless sensor networks (WSNs) are a set of devices connected to collect data remotely from the environment, as required by the user and they are made up of devices called nodes, which collect, process, and wirelessly communicate data through a sensor, a microcontroller, and a radio transceiver, respectively. These WSNs are present in our daily lives [1], agriculture [2], and industry [3]; however, the use of non-rechargeable storage devices as an energy source, i.e., batteries, have limited their implementation. Here, we attempt to solve this concern by designing a solar energy harvesting system (SEHS) to feed energy to wireless sensor nodes. Such an energy harvesting process from the environment is made by a solar panel, which provides the energy stored in rechargeable batteries or supercapacitors to later be used by the node [4]. Furthermore, the advantages of the proposed design are restricted to a lower cost implementation, promoting the use of renewable energy, and avoiding pollution from the non-rechargeable batteries that commercial sensor nodes implement [5], [6].

The connection of the sensor node to an energy harvesting system constitutes an autonomous wireless sensor node. This technology is currently used for the wireless communication of data, mainly because they harvest infinity sources of energy; nonetheless, this kind of energy has a not continuous behavior. For this reason, the design of an energy harvesting system, for example, with solar energy, implies the implementation of specific techniques such as pulse-width modulation (PWM) and MPPT techniques [7]. Besides, these systems require a voltage adequation reached by a buck converter [8]. In the next paragraphs, our proposal will be presented. In Section 2, the literature review of the designs and implementation of a SEHS and a sensor node is developed. Later, in Section 3, we show the proposed methodology to design and implement our autonomous sensor node, and in Section 4, we present the results and subsequent discussion. Finally, conclusions are presented in Section 5.

Existing designs of autonomous sensor nodes are used for animal tracking, air quality monitoring, and disasters [9]. For animal tracking applications, autonomous sensor nodes have carried out wireless communication every 2 hours to track the animal’s position in large and rural areas. These nodes use solar panels connected to rechargeable batteries to ensure the node’s continuous operation for 1 year [10]. Furthermore, in air quality monitoring applications, autonomous sensor nodes have recorded air quality parameters for 180 days, implementing a solar panel of 5W, and 12 V lead-acid batteries [11]. The mentioned applications are relevant to the monitoring and the subsequently study of a subject of interest, nonetheless these sensor nodes operate through solar panels and rechargeable batteries without charge controllers which result in low-efficiency operation because the solar panels themselves have efficiencies under 30% [12],[13].

Current designs of SEHSs present the implementation of charge controllers with techniques like PWM and MPPT, which increase the efficient operation of the SEHS. These controllers can present different architectures that allow for more efficient operations according to the application to which the autonomous sensor node is directed. For instance, the controller presented in Figure 1.a implements an on-off control of the sensors, without affecting its overall functionality and implementing a supercapacitor to provide high energy spikes to the node; mostly required in the transmission process [14]. Moreover, the controller illustrated in Figure 1.b implements an MPPT control, which presents an increment of over 20 % of the system efficiency compared to a PWM control [15]. Among the aforementioned, the MPPT controllers are reported to be an efficient technique to implement solar EH systems and for that reason, in the present article, we propose a design of the solar energy harvesting system with an MPPT control. Also, in the development of the SEHS design, it is necessary to determine the limits of the signal control and to calculate the values of the converter components [16], [17]. Other kinds of controllers in solar energy harvesting systems, i.e., the solar tracking systems, have been implemented and have been proven to provide efficient operation [18]; however, the implementation of this and other kinds of controllers are beyond our scope.

The energy consumption problem presented in the design and the implementation of a solar energy harvesting system is a current research topic. In fact, recent designs have addressed this issue through the selection of components based on energy requirement measurements. In this scenario, the selection of components was reported according to the mathematical equation of the adjustable resistance, battery performance evaluation, and current consumption [19]. Moreover, cyclical calculations of components like the inductor and the input and output capacitor of the converter were also registered [20]. Along with the latter, efficient designs of SEHS have been reported, with calculations of the current consumption of the nodes, in agriculture [21] and precision irrigation applications [22]. However, these reported designs are implemented in sensor nodes that require low energy consumption values, so they do not apply to nodes with average energy consumption.

Altogether, different SEHS design methodologies have been implemented. These methodologies are based on low energy consumption components selection [23], small components selection [14], and low energy consumption [21]. The continued importance of study in the development of new, more efficient SEHS design methodologies is evident, because of the extended presence of WSNs in many kinds of contexts, like animal tracking [10], air quality monitoring [11], and removable energies [24]. Their implementation is generally oriented to achieve the necessary power to feed the node [10] and guarantee a continued energy delivery to the node [16], [25]. Therefore, here we report the development and implementation of an SEHS design to guarantee the operation of an autonomous wireless sensor node. The proposed methodology to design such a SEHS was based on the mathematical model of a buck converter coupled with the mathematical model of a solar panel.

The design of the SEHS for the autonomous operation of a wireless sensor node was proposed. This design integrated a solar panel of 4.5 W, a buck converter with MPPT control, and two li-ion batteries, as suggested by recent developments of autonomous wireless sensor nodes [15]. The aforementioned three units supply energy to a wireless sensor node, designed with an Atmega328P microcontroller [26], which wirelessly transmits data every 20 minutes and has an energy consumption close to the average energy consumption of commercial sensor nodes [27], [14], [21]. Each of the solar energy harvesting components was selected based on the implementation of simulation and analytical models concerning the solar panel performance, the batteries, and the converter. The purpose of our proposed methodology was to develop an efficient design to guarantee the autonomous operation of our sensor node.

Concerning the solar panel simulation, it required experimental measurements of the open voltage, short circuit current and intermediate points of voltage and current. The purpose of the simulation was to obtain the curve characteristics of the selected solar panel. The simulation was performed following the methodology in previous works [28], the solar panel mathematical model presented in Equation 1 and using the Matlab/Simulink software. In Equation 1, the current I is shown as a function of the voltage, V. Also, the resistances in series RS and parallel RP are associated with the loss produced by the leakage currents and ohmic contacts of the cell materials. These last resistances were calculated using the Newton Raphson method, as presented in [29]. Other parameters required to perform the simulation were the irradiance G, the reference irradiance Gref the reference photocurrent Iph,ref , the temperature coefficient μSC , the difference of temperature ΔT, and the diode current I0 . Figures 2a, 2b and 2c present the block diagram implemented for the solar panel simulation, the characteristics curves of the solar panel and the measuring instruments for the Solar panel measurements.

In addition, the implemented converter was a buck converter because the node voltage was lower than the solar panel voltage. Its design was based on an MPPT control to improve the efficiency of the energy harvesting system [30]. The definition of the values of its components like the inductor, and output capacitor, were calculated with Equation 2 and 3[31] and assuming continuous driving conditions for a 100 kHz switching frequency and a ripple voltage ΔV of 0.5% in the output capacitor. In theseequations, f is the switching frequency, and D is the duty cycle, which corresponds to the ratio between the converter output and input voltage.

In the developed design, the converter was connected to a pair of Li-ion batteries, that provide a continuous source of energy to the SEHS [32]. These batteries were chosen based on the comparison between different kinds of rechargeable batteries and their gravimetric densities, volumetric densities, charge and discharge efficiencies, nominal voltages, and self-discharge rates [33]. For this comparison, supercapacitors were not considered because, compared with the rechargeable batteries, the latter provide a more continuous operation instead of supplying high values of the current for short periods [33]. Figure 3a presents the basic blocks of the SEHS design, connected to the solar panel and the sensor node. In this figure, the buck converter performs the reduction of voltage required between the Li-ion cells, the node, and the solar panel. This reduction was requested for the implementation of the IR2104 integrated circuit to minimize the spikes of current in the inductor caused by the high frequency used to produce the switching process and, therefore, this circuit corresponds to a highly efficient technology [34]. The control signal and the implemented circuit are presented in Figure 3b and 3c.

Along with the design and implementation of the sensor node and the SEHS, the autonomous performance of our system was proved, through measurements of voltage and current consumption in the sensor node and the rechargeable batteries. These measurements were performed with 115 fluke multimeters and one TBS 1102B-EDU Tektronix oscilloscope. In the proposed set up, the SEHS was connected to the sensor node to supply energy and to a B&K power supply to simulate the source of energy, which allowed for changes of input voltage from 10.26 V to 7.2 V. The role that the charge resistance plays in the definition of the limits of the duty cycle, in our design, was done using a 15 Ω load. In this scenario, the converter control signal was changed between 50% and 99%. Furthermore, the proposed design was simulated, connecting the mentioned solar panel simulation and the buck converter circuit simulation in the Matlab/Simulink software. The data obtained with this simulation validated the proposed design [28], defined the required power to supply to the solar panel and estimated the maximum power points of the solar panel PV curves.

The simulation of the solar panel of 4.5 W provided the average power for each of the curves at different irradiance, i.e., for an irradiance over 500, the simulation presents a maximum power of 1.7 W. Figure 4 presents the other MPPs of the selected panel. The simulation data and the measurements of voltage and current in the solar panel presented an error of 5%. The converter implemented the MPPT technique “perturb and observe” algorithm. The calculated values of its inductor and output capacitor were 19 of 133, respectively. The storage device used in implementing the autonomous sensor node presented an equivalent resistance that ranges from 30 to 80 Ω. The control sign turned off the storage device when its voltage was under 5.5 V and turned on over 8.5 V.

The final schematic of the energy harvesting system and the designed sensor node are presented in Figure 5 and Figure 6. Along with the circuit of the converter, the design of the energy harvesting system includes a sensing circuit to measure the input and output power, a switching circuit to protect the MOSFETS of the converter from burn as a consequence of the high frequency of the MPPT signal control, and a protection circuit to isolate the solar panel and the sensor node from any reverse current or high current spikes. The former circuit described is presented in Figure 3.a Also, the design of the sensor requires the connection of additional circuits. Figure 3.b presents the microcontroller connection, the sensor and the transceiver to regulators that provide outputs of 5.0 V and 3.0, a dual bidirectional voltage-level translator for improving the wireless communication the connection and bypass capacitors. Figures 3.b and 3.c present the PCB design of the node and solar energy harvester system, which was developed in the free version of the program EAGLE. The presented PCB boards were chosen to have width traces of 0.025 to guarantee current transportation of 2A without any damages of the designed board.

The efficiency achieved with the proposed design oscillated between 70% and 85%. The variable efficiency reached by the implemented design depended on the input power, as is presented in Table 1. For a limit cycle that ranged from 0% to 99%, the inductor voltage presented spikes that ranged from 5% to 30% of the converter output voltage. If these spikes remain for long periods in the converter, it may damage some parts. For that reason, the limit of the duty cycle was redefined to change from 70% to 80 %. Figure 7 presents the discharge curve of the storage device and the discharge time of this device connected to the sensor node transmitting data packets of temperature ranging from 25 to 27 °C, wirelessly, each 5 sec at 5V. This discharge time was 2.6 h. Subsequently, the sensor node autonomous operation was verified for this node, transmitting data packages every 20 minutes.

With the implemented design, the autonomy of the sensor node was demonstrated, based on the solar energy harvesting system proposed. It was shown that efficiency depends on the duty cycle. Figures 8.a and 8.b present the inductor voltage signal for a converter connected to a 15 Ω resistance and programmed to change its duty cycle to values up to 80%. The spikes presented in Figure 3a not only depend on the programmed duty cycle but also of the resistance connected to the converter, which in the present case corresponds to the sensor node and the storage device. Thus, the efficiency of the solar energy harvesting system depends not only on the duty cycle of the converter but also on the connected resistance at its output. The study of this relationship will be addressed in future research. From the commercial implementation of buck converters, it is possible to establish that these spikes are avoided by implanting complex circuits that make the process complex and exclusive [35], [36]. However, the duty cycle limits change if the connected resistance changes. Thus, future research will focus on developing an optimized mathematical model for the solar energy harvesting system design that allows us to understand the relationship between the efficiency and the resistance connected to the converter.

This article presented a solar energy harvesting system with the implementation of MPPT control. The system has proved to be useful in determining the real consumption of a wireless sensor node and a pack of two cells of Li-ion batteries. The control parameters and estimated energy autonomy were defined for the proposed design. In the proposed design, the duty cycle was a topic of discussion because understanding its limits provides the required understanding of the system to propose a more efficient design of it. For this reason, future works will be directed towards improving the converter’s efficiency through the optimization of the solar panel’s resistance. This optimization can be achieved by developing a mathematical model of the solar EH system that allows us to describe the limit of the converter’s duty cycle as a function of the design parameters. Also, we introduce the topic of the spikes in the signal of the inductor. These spikes present values close to 30% of the output voltage of the converter. This data was necessary to compare this value with the design’s reported output signal and define a limit value for the spikes on the converted inductor signal [37], [38]. Based on this comparison, it was established that this value might not overcome the 22% of the output voltage to guarantee a converter’s continuous transition performance. The reduction of these spikes and its relationship with the identified duty cycle limits are future challenges to investigate.