能源自适应MPPT

An Energy-Adaptive MPPT Power Management Unit for Micro-Power Vibration Energy HarvestingJun Yi, Feng Su, Yat-Hei Lam, Wing-Hung Ki and Chi-Ying Tsui Integrated Power Electronics Laboratory Department of Electronic and Computer Engineering The Hong Kong University of Science and Technology Clear Water Bay, Hong Kong SAR, China Fax: (852) 2358-1485; Email: {eeyi, sufeng, hylas, eeki, eetsui}@ece.ust.hkAbstract— A batteryless power management unit (PMU) that manages harvested low-level vibration energy from a piezoelectric device for a wireless sensor node is presented. An energy-adaptive maximum power point tracking (EA-MPPT) scheme is proposed that allows the PMU to activate different operation modes according to the available power level. The harvested energy is processed by an ac-dc voltage doubler followed by on-chip charge pumps with variable up/down conversion ratios for higher efficiency. Interleaving technique is employed for the high-power output to reduce both current and voltage ripples. The PMU is designed using a 0.35 μ mCMOS process, and simulation results are presented to demonstrate its functions.I. I NTRODUCTIONEnergy harvested from the environment could replace or extend the lifetime of bulky batteries in a wireless sensor network, and vibration is a promising source of energy. F or piezoelectric materials at a given vibration status (magnitude and frequency), there exists an optimum output voltage at which maximum output power can be extracted [1]. An ultra-low-power maximum power point tracking (MPPT) scheme in a time-varying environment was discussed in [2] that tracked and held the maximum power point periodically using a very small duty cycle, and consumes only a fraction of the power required by prior MPPT schemes. All functional blocks are activated or shut down together. However, a sensor node consists of functional blocks with very different levels of power consumption. For example, signal processing may consume 20 μ W while RF power amplification and transmission may consume 1mW. The power management unit (PMU) should activate different blocks as the input power level changes. For example, at very low input power, the whole system may be duty-cycled; at medium input power the critical blocks work continuously while the rest are duty-cycled; and at high input power, all blocks are activated. To achieve energy-adaptive MPPT control, information on the absolute or relative amount of available power is needed. The problem is how to obtain and utilize this information without using power-demanding computational methods such as quantizing voltages and/or currents. Prior research works focused on applications that consume power in the order of 100 μ W to 10mW or even higher [3] [4]. Switch-mode power converters (switching converters) are often used due to its high efficiency. However, an off-chip bulky inductor is needed. Research in recent years has reduced the total power consumption of a sensor node to less than 1mW, and fully integrated small size solution becomes increasingly realistic and desirable in some cases. In addition, [5] shows that low-level vibration with available power of less than 100 μ W and fundamental frequency of less than 200Hz can be widely found in many situations. One challenge is to harvest energy efficiently from low-level vibration with low source voltage. Another challenge is to provide high power conversion efficiency over a wide range of source voltage without using switching converters. Furthermore, a fast power transfer rate is desirable such that the system could make the most use of short bursts of high-energy instants. In this research, we propose an energy-adaptive MPPT power management unit for harvesting energy from very low level vibrations. Available power is measured by a simple “load perturbation“ method that is basically a trial-and-error process. With this information the PMU could instruct the system to operate in different modes. A high-efficiency ac-dc doubler provides an adequate output voltage without resorting to additional step-up dc-dc converters for low vibration levels. Fully-integrated charge pumps with variable conversion ratios are used to replace switching converters. Fast energy accumulation and small input and output voltage ripples are achieved by time-interleaving techniques. II. SYSTEM OPERATIONFig. 1 shows the PMU of the sensor node with two types of load. (1) Baseband analog and digital circuits such as sensor front-end, A/D converters, computation logic and memory circuits consume about 10-20 μ W and should remain active if possible. (2) RF circuits consume about 1mW and should be activated if enough energy is accumulated for transmission. Due to size limitation the piezoelectric device may only provide a maximum of 100 μ W. Because the total power consumption exceeds the maximum power provided by the source, the transmitter has to work in a duty-cycled fashion. The energy for transmission is accumulated on storage capacitors C SRF1 and C SRF2. The RF blocks are supplied by This research is in part supported by Research Grant Council CERG HKUST 614506 and 620305. 978-1-4244-1684-4/08/$25.00 ?2008 IEEE 2570Energy-Adaptive MPPT ControlAC-DC Doubler Charge PumpsPiezo filmV SC SRF2Energy-Accm. QP3C SRF1RF Circuits BRF Circuits AAnalog Temperature: -20 o ~ 100 o). A 5-stage ring oscillator could generate all driving signals with a frequency that ranges from 1.6MHz to 2.5MHz. The total current consumption for this block is around 400nA with V ref = 758mV and f osc = 2MHz at V CSIN = 2V and T = 25 o. The start-up circuit is not shown. IV. SIMULATION RESULTSFig. 7 shows the simulated voltage waveform of V SIN for three cases: (1) I p=15μ A, f=30Hz and P avg(opt)=8.54μ W. (2) Ip=30μ A, f=60Hz and P avg(opt)=17.08μ W. (3) I p=50μ A,f=90Hz and P avg(opt)=31.63μ W. Both the load power P LBB and PLES are equal to 10 μ W. The first two cases have the same V s(opt) of 3.578V. It can be seen that, for case 1, V SINfluctuates between 3.652V and 3.431V, with an average of 3.541V. For case 2, V SIN is between 3.725V and 3.507V, and the average is 3.616V. For both cases, the average voltage is very close to the optimum value. The input power for case 3 is larger than the sum of P LBB and P LES so V SIN stays at 4.234V, which is above the optimum value of 3.976V. V. C ONCLUSIONSIn this paper, an energy-adaptive MPPT power management unit for harvesting energy from very low level vibration is proposed. By a simple control method, it sets the system in different operation modes depending on the available power level. F ully-integrated time-interleaved charge pumps with variable conversion ratios are used to replace the switching converter, eliminating the need of an external bulky inductor. A very low power voltage reference is used to generate the reference voltages for start-up. The functionality of the MPPT scheme is verified by simulations. VI. REFERENCES[1] G. K. Ottman, A. C. Bhatt, H. Hofmann and G. A. Lesieutre, “Adaptive piezoelectric energy harvesting circuit for wireless remote power supply,“ IEEE Trans . Power Electronics , pp. 669-676, Sep. 2002. [2] C. Lu, C. Y. Tsui, and W. H. Ki, “Vibration energy scavenging and management for ultra low power applications,“ IEEE Int l Symp. on Low Power Elec. Devices , pp. 316-321, Aug. 2007. [3] N. Shenck and J. A. Paradiso, “Energy scavenging with shoe-mounted piezoelectrics,“ IEEE Micro , pp.30 – 42, May-Jun. 2001. [4] G. K. Ottman, H. F . Hofmann, and G. A. Lesieutre, “Optimized piezoelectric energy harvesting circuit using step-down converter in discontinuous conduction mode,“ IEEE Trans. Power Electronics ,pp.696 – 703, Mar. 2003. [5] S. Roundy, P. K. Wright, J. Rabaey, “A study of low level vibrations as a power source for wireless sensor nodes,“ Computer Communications , pp.1131-1144, 2003. [6] Y. H. Lam, W. H. Ki and C. Y. Tsui, “Integrated low-loss CMOS active rectifier for wirelessly powered devices,“ IEEE Trans. on Ckts. & Sys., Part II , pp.1378-1382, Dec. 2006. [7] F. Su and W. H. Ki, “Design strategy for step-up charge pumps with variable integer conversion ratios,“ IEEE Trans. on Ckts. & Sys., Part II , pp.417-421, May 2007. C 1V ddV ss|GNDC2C LV OVddC 3V ddC4V ddS1S2S 3S 4S 51S 52S 6 S 9 S 12S 7S 81S 82S 10S 111S 112S 113S 114SO1SO2SO3SO4123456 87Fig. 3. Schematic of charge pump QP 2 for RF-A. S1S2S3S4S51S52S6S113S114S12SO1SO2SO3SO4r1|φ 1(r2)φ 2(r 54|r 32|r74 |r2)r 12|r 34|r1 |φ 1r1|φ 1(r12 |r32|r2)φ 2(r54|r 32|r2)φ 1(r12|r 54|r32 )|φ 2(r 34|r74 )r 1|φ 1(r 34|r74 |r2)| φ 2(r 12)φ 1(r34 |r74)φ 1(r34 |r74)r1 |φ 1(r 2)| φ 2(r 12|r 34)r1 |φ 2(r 12|r 54|r 32|r 2)r1 |φ 2(r 12|r 54|r 32|r 2)r1 |φ 2r1 |φ 2S111S7S81S9S82S10S112r 1|φ 1(r 2)φ 2(r54 |r32|r2 )φ 1(r 54)|φ 2(r 34|r74 )r 12|r 1|φ 1 (r54)r 1|φ 1φ 2(r54 |r32|r74 |r2)φ 1(r 2)Fig. 4. Control logic of power switches in Fig. 3. RingOsillatorfromRef&CLKBranchV CP3C SRF2φ 4Energy-Accum.QP3QP 31QP 32QP33QP34φ 2φ 1φ 3RingOssilatorφ 1φ 2φ 3φ 4φ 5V refM3M 2M1 M4M7M 6M 5 M812 : 1 : 1 : 312 : 1 : 1 : 241.2R RM 9: 3CROVCSINFig. 5. Interleaved QP 3. Fig. 6. Reference and clock generator.Fig. 7. Simulated voltage waveforms of V SIN for different power levels.2573