DUAL-INPUT ENERGY HARVESTING INTERFACE FOR LOW-POWER SENSING SYSTEMS

Transcription

1 DUAL-INPUT ENERGY HARVESTING INTERFACE FOR LOW-POWER SENSING SYSTEMS Eun-Jung Yoon Department of Electronics Engineering, Incheon National University 119 Academy-ro, Yonsu-gu, Incheon, Republic of Korea Chong-Gun Yu Department of Electronics Engineering, Incheon National University 119 Academy-ro, Yonsu-gu, Incheon, Republic of Korea Abstract This paper proposes a dual-input energy harvesting circuit using vibration and thermoelectric energy. Since the maximum power point of a thermoelectric generator (TEG) output and a piezoelectric generator (PEG) output is 1/2 of their open-circuit voltage, an identical MPPT controller can be used for both energy sources. The proposed circuit monitors the outputs of the TEG and PEG, and chooses the energy source generating a higher output voltage using an auto-switching controller, and then harvests the maximum power from the selected device using the MPPT controller. The proposed circuit is designed in a 0.35 μm CMOS process and its functionality has been verified through extensive simulations. The designed chip occupies 330μm 640μm. Keywords Auto-switching control, energy harvesting, MPPT control, thermoelectric energy, vibration energy. I. INTRODUCTION Energy harvesting is essential in developing ultra-low-power and small-size sensing systems such as wireless sensor nodes and implantable bio-devices to achieve self-powering (battery-less) and near-perpetual operation. Among various energy sources, solar, vibration and thermal energy are widely explored for energy harvesting [1]-[9]. There is a maximum power point (MPP) for each energy harvester. To harvest the maximum power, the energy harvesting system should be able to track the MPP such that the harvester always operates at its MPP. The MPP tracking (MPPT) becomes even more important in micro-scale energy harvesting systems because the output power of a size-limited energy harvester is extremely small, often only a few tens of µw. Since energy harvesting, by nature, is sporadic, combining energy from multiple sources is desirable to increase the overall system reliability. Recently, several solutions in multi-input energy harvesting have been proposed such as adding output voltages from individual harvesters by stacking individual storage capacitors [2], selecting the harvester with maximum instantaneous power [3], and sharing single inductor for power conversion [4], [5]. Among these solutions, the last one is promising approach. However it requires an external inductor and more power switches for sharing an inductor, and thus, may not suitable for ultra-small-size sensing systems. Ultra-low-power and very small-size sensing systems are usually utilized for applications operating in active/sleep mode. In this mode the system is active when there is sufficient energy, and it sleeps when the available energy is insufficient. In these applications, simultaneously harvesting energy from multiple sources might not be the most important requirement. The first priority can be rather reducing the number of components required such that the system form factor, cost and power losses are thus reduced. In this paper a dual-input energy harvesting circuit using vibration and thermoelectric energy is proposed and designed in a 0.35 μm CMOS technology. The input source with higher output voltage is selected automatically by proposed auto-switching controller, and an identical MPPT controller using the fractional open-circuit (FOC) method is used for both energy sources. By avoiding the use of individual MPPT controller circuits for multiple energy sources, the number of components required in multi-input energy harvesting system is reduced. 113

2 II. PROPOSED DUAL-INPUT ENERGY HARVESTING INTERFACE CIRCUIT Fig. 1 shows the block diagram of the proposed dual-input energy harvesting system using vibration and thermoelectric energy. Vibrational and thermoelectric energy are transduced into electrical energy using a piezoelectric generator (PEG) and a thermoelectric generator (TEG), respectively. The auto-switching control block outputs control signals (VMC, TMC) to select the harvester with higher output voltage. The power switching circuits provide power switches (SW1 VB, SW1 TEG ) with on/off signals (VPSW, TPSW) based on the output signals from the auto-switching block. The power switching circuit [10] compares the harvester s output voltage (V VB or V TEG ) and the voltage on the storage capacitor (V Sto ), and then provides higher one with the power switch, such that the PMOS switch can be properly turned off. Moreover, dynamic bulk regulation technique [6] is adapted to avoid leakage currents and latch-up. The MPPT control block makes the selected harvester operate at the MPP to harvest maximum available power. The proposed system operates in active/sleep modes. The application load switches between active mode and sleep mode depending on the harvested energy. Fig. 1 Proposed dual-input energy harvesting system A. Energy Harvesters The targeted PEG is the Quick Pack QP20W [1] that can generate 125 µw with an acceleration of 7 m/s 2 at 80 Hz. The PEG is usually modeled as a sinusoidal current source I PEG in parallel with an internal capacitor C PEG as shown in Fig. 2(a) [4]. The amplitude and frequency of the current source are set to 3 V and 80 Hz, respectively. The internal parallel capacitor is 0.2 µf. The TEG can be modeled as a voltage source V TEG,OC and a series internal resistor R TEG as shown in Fig. 2(b) [4]. For the targeted TEG [7], the V TEG.OC and R TEG are 3 V and 20 kω, respectively. Fig. 3 shows general I-V and P-V curves of a PEG including an AC-DC converter and a TEG. Since the V MPP of a TEG and a PEG is 1/2 of their open-circuit voltage V OC, an identical MPPT controller can be used for both energy harvesters. I PEG + C PEG V PEG - (a) V TEG,OC + - R TEG + V TEG - (b) Fig. 2 Equivalent circuit of energy harvester (a) PEG (b) TEG. Fig. 3 I-V, P-V characteristics of PEG with ADC and TEG. B. AC-DC Converter Since the output of a PEG is similar to an ac voltage, an AC-DC converter (ADC) is required to generate a dc voltage. The ADC must have high conversion efficiency in order to transfer the maximum amount of power from the harvester to the load. Conventional passive ADCs using four diode-connected MOSFETs or gate cross-coupled ADCs using two MOS diodes and two MOS switches suffer from diode voltage drop that reduces the available output voltage and thus system efficiency. In this paper an active ADC [8] as shown in Fig. 4 is designed for full-wave rectification without diode voltage drop. The ADC consists of two active diodes and two MOSFET switches. The active diode consists of a PMOS switch and a comparator. The simulated peak efficiency of the designed ADC is 95.6% at the load resistance of 20 kω. C. MPPT Control Block Fig. 5 shows the architecture of the MPPT control block that is used to regulate V Sto at the MPP of the selected energy harvester and transfer the energy harvested to load. It consists of a pulse generator, a sampler, and an enable generator. The pulse generator is used to generate a clock pulse (MC) for MPPT control. As shown in Fig. 6 the pulse MC is one cycle long for every 128 cycles of the clock signal (CLK) that is generated using a ring-type oscillator. During this cycle (MC = 1 ), the switch SW1 is open, and the V MPP of a selected harvester is sampled by the sampler. In practice, V OC /8 is sampled instead of V OC /2 (= V MPP ) for proper operation of internal circuits. 114

3 by the comparators are used by the latch to generate the EN signal determining the on/off states of the PMOS switch, SW2, which defines the charge and discharge phases. The storage capacitor (C Sto ) is repeatedly charged and discharged around V MPP, and thus, the selected harvester always operates near the MPP. The power charged on C Sto is delivered to the load during the discharge phases. Fig. 4 AC-DC converter (Full-wave rectifier) Fig. 5 Block diagram of MPPT control block Fig. 6 Timing diagram of MPPT clock MC For the sampler a two-stage sample and hold (S/H) circuit [9] as shown in Fig. 7 is designed to reduce leakage and to keep a sampled voltage during the long hold times. The first sampling capacitor discharges linearly to the input, but the second capacitor discharges based on the difference between the two capacitor voltages, forming a quadratic voltage profile. Therefore this S/H topology is more space-efficient than a single-stage circuit using a larger capacitor for a given sample rate. Two reference voltages, V MPP,max /4 and V MPP,min /4, are generated by using an amplifier in non-inverting configuration. Fig. 7 Sampler schematic Fig. 8 Enable generator schematic and its timing diagram D. Auto-Switching Block The PEG at resonance, along with the ADC, can also be modeled as a voltage source V VB,OC and a series resistor R VB as similar with the model of TEG in Fig. 2(b). The equivalent resistance R VB is approximately constant over the harvester s operating range [4]. In this design the targeted harvesters have the approximately same internal resistance, R VB R VB = 20kΩ. In this case, the harvester with higher V OC can be selected for energy harvesting because the maximum available power depends on V OC only as shown in equation (1). 2 2 VTEG, OC VVB, OC PTEG, max, PVB,max (1) 4R 4R TEG The auto-switching control block as shown in Fig. 9 is used to select the harvester with higher output voltage. While MC is high, the open-circuit voltages (V TEG,OC, V VB,OC ) are compared by the comparator, and the result is stored in the D-latch. To reduce current consumption the comparator is disabled while MC is low. As shown in the timing diagram of Fig. 10, if V VB,OC is greater than V TEG,OC, the signal VMC follows MC for energy harvesting from the PEG while the signal TMC maintains high level for disconnecting the TEG by turning SW1 TEG off. If V TEG,OC is greater than V VB,OC, the signal TMC follows MC while VMC maintains high level. VB The enable generator consists of two comparators and a latch as shown in Fig. 8. The upper comparator detects whether the voltage on the storage capacitor (in practice V Sto /4) reaches the predetermined maximum MPP level, V MPP,max /4, generated from the sampler, while the lower one detects whether it reaches the predetermined minimum MPP level, V MPP,min /4. The signals generated 115

4 Fig. 9 Block diagram of auto-switching control block Fig. 12 shows simulation results when V VB,OC = 0 V and V TEG,OC = 3 V. It can be seen that the TEG output voltage V TEG reaches the open-circuit voltage V TEG,OC periodically while MC = 1, and, except this time period, V TEG (and thus V Sto ) is maintained around the V MPP (V TEG,OC /2). The V Sto during the discharge phases is delivered to the load voltage V Load. Fig. 10 Signal waveforms of auto-switching control block (t < t2 : V VB,OC > V TEG,OC, t > t2 : V VB,OC < V TEG,OC ) III. SIMULATION RESULTS The designed circuit is simulated using a 0.35-µm CMOS process parameters. The selected values of the capacitors C VB, C TEG, C Sto are 300 nf, 100 nf, 47 μf, respectively. The load resistance is 5 kω. It can be seen in Fig. 11 that the auto-switching and power-switching functions are properly conducted. When V VB,OC = 2.8 V and V TEG,OC = 0 V, the signal VPSW (the power-switched version of VMC) follows MC. The situation is reversed at 2 s, and the signal TPSW follows MC. The power-switching operations can be observed through the arrows with broken lines on the figure. It can also be seen that the voltage on the storage capacitor V Sto is maintained around the MPP of the selected harvester. Fig. 11 Simulation result of auto-switching and power switching (0s < t < 2s : V VB,OC =2.8V & V TEG,OC =0V, t > 2s : V VB,OC =2.8V & V TEG,OC =3.3V) Fig. 12 Waveforms of V TEG, V Sto and V Load (V VB,OC = 0V, V TEG,OC = 3V) Fig. 13 shows power efficiencies of the designed circuit at different load resistances. The peak efficiencies for vibration energy harvesting and thermoelectric energy harvesting are 91.5 % (@ 10 kω) and 95.9 % (@ 15 kω), respectively. Efficiencies for thermoelectric energy harvesting are greater than those for vibration energy harvesting in the measured range of load resistance, 200 Ω ~ 30 kω. This is because the loss in the ADC is reflected in the efficiency reduction for vibration energy harvesting. Table I shows a comparison of the proposed dual-input energy harvesting system with previously published multi-input energy harvesting designs. The proposed system is for applications operating in active/sleep mode, while the other systems are for applications in continuous operating mode. The efficiencies of the designed system are relatively high thanks to a simple MPPT controller with FOC technique. The proposed energy harvesting system can be implemented in a very small form-factor such as miniature sensor nodes because it does not require external inductor and an identical simple MPPT controller is used for both energy sources. The layout of the designed circuit is shown in Figure 14. It occupies the area of 330μm 640μm. 116

5 Process (μm) Input source Architec -ture Operatio n mode [2] [3] [4] [5] 0.25 Solar & Vib. Stackin g storage capaci -tors 0.35 & 0.18 RF & Selectin g the harveste r with maximu m power Fig. 13 Power efficiency of the designed energy harvesting circuit Fig. 14 Layout of the designed energy harvesting circuit TABLE I This work Solar & Vib. & Single inductor sharing Vib. & Cont. Cont. Cont. Cont. MPPT Maximu m power efficienc y (%) External device Hill-climbi ng for solar Solar 83 Vib Vib. & Single induct Auto or -switchin sharin g g Active/ Sleep FOC 74.9 Vib. w/ 91.5 AD C Ther L & C L & C L & C L & C C Year COMPARISON OF MULTI-INPUT ENERGY HARVESTING CIRCUITS IV. CONCLUSION This paper presents a dual-input energy harvesting system using vibration and thermoelectric energy for ultra-low-power sensing systems. This system is capable of auto-switching to the input source with higher output power. An identical and simple MPPT controller with the FOC method is used for both energy sources. The proposed circuit is designed in a 0.35um CMOS process and its functionality has been verified through extensive simulations. The peak efficiencies for vibration and thermoelectric energy harvesting are 91.5 % and 95.9 %, respectively. The designed chip occupies 330μm 640μm. The proposed systems have applications operating in active/sleep mode such as environmental sensing systems that allow a relatively low duty cycle. V. REFERENCES [1] J. Colomer-Farrarons, P. Miribel-Catala, A. Saiz-Vela, M. Puig-Vidal, and J. Samitier, "Power-Conditioning Circuitry for a Self-Powered System Based on Micro PZT Generators in a 0.13μm Low-Voltage Low-Power Technology," IEEE Trans. on Industrial Electronics, vol. 55, no. 9, pp , September [2] N. J. Guilar, R. Amirtharajah, P. J. Hurst, and S. H. Lewis, An energy aware multiple-input power supply with charge recovery for energy harvesting applications, Dig. Tech. Papers in IEEE ISSCC, pp , [3] H. Lhermet, C. Condemine, M. Plissonnier, R. Salot, P. Audebert, and M. Rosset, Efficient power management circuit: From thermal energy harvesting to above-ic microbattery energy storage, IEEE JSSC, vol. 43, no. 1, pp , Jan [4] S. Bandyopadhyay and A. P. Chandrakasan, Platform Architecture for Solar, Thermal, and Vibration Energy Combining With MPPT and Single Inductor, IEEE JSSC, pp , [5] Y. S. Yuk, et al., An energy pile-up resonance circuit extracting maximum 422% energy from piezoelectric material in a dual-source energy-harvesting interface, Dig. Tech. Papers in IEEE ISSCC, pp , [6] M. Ghovanloo and K. Najafi, Fully integrated wideband high-current rectifiers for inductively powered devices, IEEE JSSC, vol. 39, 2004, pp [7] V. Leonov, P. Fiorini, S. Sedky, T. Torfs, and C. Van Hoof, Thermoelectric MEMS generators as a power supply for a body area network, in Proc. 117

6 Int. Conf. Solid-State Sensors, Actuators and Microsystems, pp , [8] D. H. Lee, J. H. Jeon, J. T. Park and C. G. Yu, "Design of a High-Efficiency Rectifier for Vibration Energy Harvesting," 2010 SoC Conference, pp , [9] M. D. Seeman, S. R. Sanders, and J. M. Rabaey, An ultra-low-power power management IC for energy-scavenged Wireless Sensor Nodes, in Proc PESC, pp , [10] T. Y. Man, P. K. T. Mok, and M. J. Chan, A 0.9V Input Discontinuous Conduction Mode Boost Converter With CMOS Control Rectifier, IEEE JSSC, vol. 43, pp , Sep Acknowledgements: This research was supported by the Basic Science Research Program through the National Research Foundation of Korea ( ) and was partially supported by IDEC. 118