Vibration energy harvesting for wireless sensor networks: Assessments and perspectives
Thanks to the reduction of circuit sizes and progress in microelectronics, basic electronic functions are consuming less and less power, allowing us to use a new ecological and durable supply source for wireless sensor networks (WSN): ambient energy, including sun, temperature and vibration.
Wireless sensor networks: goals and needs
Today, one of the goals of researchers and R&D engineers is to develop sensor networks able to collect data from their surrounding environment. WSN (Figure 1a) are made of several sensor nodes (Figure 1b); each node is able to get information from its environment (temperature, vibrations, light, etc.), to turn it into numeric data and to send it to a base station. Many fields, such as transportation, industry and aeronautics, have a strong interest in developing and using WSN to increase their productivity (real-time monitoring), reduce their costs or limit machine downtimes (predictive maintenance).
Batteries can power those devices for a limited time. Another solution consists in using energy harvesting (EH), aimed at converting ambient energy into electricity. This green technology also gives a theoretically unlimited lifetime to sensor nodes, in contrast with batteries.
Unfortunately, the power output of micro energy harvesters (Eh) is generally limited to some tens or hundreds of microwatts and the power consumption of RF-emitters or microcontrollers can reach some tens of milliwatts, banning a continuous running mode and implying intermittent measuring and data sending. Therefore, in EH and autonomous WSN, it is more appropriate to look at energy consumption for one measure instead of power consumption.
Also, it should be noted that the value 500µJ is a key number for WSN. This value corresponds to the needed energy to get a piece of information from the environment (temperature, humidity, etc.), to convert it into numeric data with an analog-to-digital converter (ADC) and to send it using standard protocols such as Bluetooth Low Energy or Zigbee. This energy could be reduced to some tens of µJ in the near future.
Therefore, functioning mode of EH-powered WSN can be summed up as follows (Figure 2): the energy harvesting device harvests power from its environment and stores it in a buffer (capacitor, battery) (1); µC, sensor and emitter are in standby and consume about 5µW. Measurement (2) and emission (3) are performed when enough energy is stored in the buffer. Buffer is emptied; system returns to standby, waiting for a new measurement cycle (4).
Figure 2: WSN measurement cycle
Figure 3: EH-based supply source
Vibration Energy Harvesting (VEH)
The VEH principle is quite simple and mainly relies on resonance phenomena. Vibration energy harvesters’ (VEh) basic architecture is a mass-spring system, damped by mechanical friction forces (fmec), that resonates when subjected to ambient vibrations (Figure 4a); this structure makes it possible to amplify low-amplitude vibrations and VEh output power (Figure 4b). Indeed, it is important to note that ambient vibrations (natural and man-made environments) are generally characterized by low frequencies (<100Hz) and low amplitudes (<50µm) that do not allow harvesting much power without using resonance effects.
The goal of EH researchers is to develop converters able to turn part of mobile mass kinetic power into electric power. The effects of this converter on the system are modeled as an electrical force (felec) that slows down mobile mass displacements when mechanical power is extracted and turned into electricity.
Figure 4: a) VEh general model and b) power amplification at resonance (e.g. f0=50Hz)
Piezoelectric converters (Figure 5a) use piezoelectric materials that generate charges under stress or strain. Electromagnetic converters (Figure 5b) are based on Lenz’s law: the movement of a magnet in a coil generates a current. Finally, electret-based electrostatic converters (Figure 5c) use electrets to induce charges on electrodes; a relative displacement of an electrode compared to an electret generates a variation of electret charges’ influence on the electrode and charge circulation.
Nevertheless, whatever the converter, VEh output power is limited by physics and will not exceed Pth (except in certain circumstances; e.g. non-linear behaviors). VEh output power is therefore proportional to mobile mass and acceleration squared and inversely proportional to the harvester’s frequency bandwidth.
Each of these converters presents both pros and cons that are summed up in Table 1:
Table 1: Pros and cons of the different converters
Figure 6: Electrostatic VEh a) scheme and b) prototype
Actually, the resonant effect is probably both the main advantage and the main drawback of VEh as they can harvest much power when ambient vibration frequency fits their resonant frequency but have a tight frequency bandwidth that does not exceed some Hz in the best case.
Non-linear effects and tuning of frequency to increase VEh frequency response
In fact, two main ways exist in the state of the art to widen VEh frequency bandwidth: use of non-linear effects (passive) and tuning of frequencies (active). We present here, as examples, some solutions currently being studied in our labs.
Non-linear effects are present in all mechanical structures but have only begun to be exploited to increase VEh output. Non-linear effects do not have to be activated by a control circuit – they are passive phenomena. They are added to the VEh structure during manufacturing and appear as soon as springs leave the linear domain (for high amplitude displacements). We have already proven that thanks to non-linear behaviors, VEh output power can be increased by 50 percent compared to standard linear behaviors in some cases (car engines, motors).
Even though non-linear effects have proven attractive to increase VEh output power and reliability, ”tuning of frequencies” is the most promising way to increase VEh frequency bandwidth. Its objective is to change VEh natural frequency by modifying spring stiffness. These changes are controlled by an active circuit aimed at searching optimal parameters to maximize output power. Two main ways are currently under study in our labs on piezoelectric VEh.
The first one consists in using three-layered beams made of two piezoelectric layers and a silicon beam (Figure 7a). Piezo 1 (Figure 7a) is linked to a control circuit that applies a voltage able to modify piezo 1 stiffness and therefore beam resonant frequency. Energy is harvested on piezo 2 (Figure 7a).
The second method is based on electrical load adjustment. When piezoelectric layers have strong coupling coefficients, it is possible to modify layer stiffness (and therefore resonant frequency) by adapting the load (Figure 7b).
Figure 7: Piezoelectric VEh applying tuning of frequency a) by applying an electric field and b) by changing the load
As presented in Figure 3, the final step (which is essential) to develop an EH module is the electrical-electrical converter inserted between VEh and storage because VEh output power is characterized by high AC voltage and low current that cannot be used as is to power electronic components.
This step can be achieved by different DC/DC converters: buck, buck-boost or flyback. These converters allow a DC/DC conversion that can reach more than 80 percent of efficiency and need quite simple control circuits that generally consume less than 5µW. An example of a flyback converter and its application to VEH is presented in Figure 8.
Figure 8: VEH, DC/DC converter and storage/application
Roadmap – Perspectives
As VEh considered as complete systems rely on technologies derived from many research fields (electronic, wireless communications, materials, power management etc.), predicting future developments is quite difficult. As a consequence, we will only present in Table 2 our vision of VEH Today, Tomorrow and After Tomorrow with expected sizes and output powers and their applications to WSN.
Table 2: VEH today, tomorrow and after tomorrow
Thanks to consumption reduction of basic functions, it is today possible to replace batteries by VEH for WSN. Measurement frequencies are still small but the concept has been validated.
The future of VEH and more generally of EH seems pretty bright as it is consistent with present topics, such as power budget reduction, green power and durable supply sources. Nevertheless, technologies are still emerging and they have not yet been adopted by industry.
Our main objectives today are focused on three points: (i) increase frequency bandwidth, which is probably the most important point to improve in VEh; (ii) increase VEh output power and (iii) decrease electronic circuit power consumption. Improving these different points should help a more widespread use of VEh to power WSN.
Sebastien Boisseau and Ghislain Despesse are researchers at CEA-Leti, a French institute focused on micro- and nanotechnologies and their applications. CEA-Leti is part of CEA, French Atomic Energy and Alternative Energies Commission.
The authors would like to thank their VEH coworkers, B. Ahmed Seddik, J.J. Chaillout, A.B. Duret, P. Gasnier, P.D. Berger, S. Riché and S. Dauvé for their contributions to this article.