Micro Power Systems Overview Dan Steingart PhD Student UC Berkeley Thanks to Shad Roundy Luc Frechette Jan Rabaey and Paul Wright UC Berkeley 2002 2004 If you fall asleep Ambient computing desperately needs new power sources A number of promising small scale power sources energy storage devices are under active development Variety of needs applications indicates that there will not be a AA Battery solution to the problem Over the wall design from application to power source not possible UC Berkeley 2002 2004 Topics Driving forces for micro power systems Energy scavenging collecting systems Energy distribution mechanisms Energy reservoir power generation systems Design considerations from theory and practice UC Berkeley 2002 2004 Why Micro Power Now Simple example At an average power consumption of 100 mW you need slightly more than 1 cm3 of lithium battery volume for 1 year of operation assuming you can use 100 of the charge in the battery which you can t Energy density of rechargeable batteries is less than half that of primary batteries So someone needs to either replace batteries in every node every 9 months or recharge every battery every 3 to 4 months In most cases this is not acceptable UC Berkeley 2002 2004 Comparison of Sources UC Berkeley 2002 2004 Power Lags Behind UC Berkeley 2002 2004 Two Paradigms In Sensor Nets Modular Off the shelf tech fabricated together on one small PCB Allows for sensor design flexibility Monolithic Eliminate layers between radio and sensor Goal to design hardware quickly around only desired functionality lower energy needs Different design paradigms create different power needs UC Berkeley 2002 2004 Energy Scavenging Areas Solar Ambient Light Temperature Gradients Human Power Air Flow Pressure Gradients Vibrations UC Berkeley 2002 2004 Solar and Ambient Light Sources Noon on a sunny day 100 mW cm2 Office Lights 7 2 mW cm2 Collectors SC Silicon 15 30 efficient 6 V open potential needs series stacks Poly Silicon 10 15 efficient Photoelectric Dyes 5 to 10 efficient BWRC BMI Solar Powered PicoRadio Node UC Berkeley 2002 2004 Temperature Gradients Exploit gradients due to waste heat ambient temp Maximum power Carnot efficiency 10 C differential 308K 298K 308 3 2 Through silicon this can be up to 110 mW cm2 Methods Thermoelectric Seebeck effect 40 W cm2 10 C Piezo thermo engine WSU 100 s W mm2 theoretical Bahr et al WSU Piezo thermo engine UC Berkeley 2002 2004 Human Power Burning 10 5 MJ a day Average power dissipation of 121 W Areas of Exploitation Foot Using energy absorbed by shoe when stepping 330 W cm2 obtained through MIT study Skin Temperature gradients up to 15 C Blood Panasonic Japan demonstrated electrochemically converting glucose UC Berkeley 2002 2004 Air Flow Power output efficiencies vary with velocity and motors Applications exist where average air flow may be on the order of 5 m s At 100 efficiency 1 mW cm MEMS turbines may be viable UC Berkeley 2002 2004 Pressure Gradients Using ambient pressure variations On a given day for a change of 2 inches Hg density on the order of nW cm3 Manipulating temperature Using 1 cm3 of helium assuming 10 C and ideal gas behavior W cm3 No active research on pressure gradient manipulation UC Berkeley 2002 2004 Vibrations Sources HVAC Engines Motors Three Rules for Design P M P a2 P 1 f Existing Designs Roundy UC Berkeley Piezo Bender Roundy 800 W cm3 at 5 m s2 similar to clothes dryer Future Plans MEMS piezo MEMS capacitance UC Berkeley 2002 2004 Energy Distribution RF Radiation 1 r2 fall off through walls actually 1 r4 For a radio station For an 50 000 W FM station at 20 km in open air 7 W Wires self defeating Acoustic Power Attenuation is very high in air better in water Sound level of 100 dB 1 ft from a lawn mower corresponds to 0 96 mW cm2 Light Guided Light through walls UC Berkeley 2002 2004 Energy Reservoirs Power Generation Capacitors Batteries Fuel Cells Heat Engines Radioactive Sources UC Berkeley 2002 2004 Capacitors Useful for on chip power conversion Possible secondary storage for frequent but non periodic energy sources Energy density too low to be a general secondary storage component UC Berkeley 2002 2004 Ultra Capacitors Differ from capacitors in surface area take advantage of highly porous electrodes Good potential for secondary storage Issues ELNA Co http www elna co jp en ct c dynac1 htm Size Leakage Distribution of Pores UC Berkeley 2002 2004 Battery Basics Closed system with respect to reactants except zinc air Area of electrodes determines power Volume of electrodes determines capacity Chemistry of battery affects potential limiting current density and cycle life Discharge Examples UC Berkeley 2002 2004 Micro Batteries Ni NaOH Zn Harb BYU Shows excellent cyclability Sealing Problems Low potentials Thin Film Lithium Bates ORNL 1 D microscale 2 D macro Uses lithium metal Shows promise for high power densities 3 D Lithium Ion Dunn UCLA Very High Power Densities In initial stages UC Berkeley 2002 2004 Micro Batteries Microfabrication Friendly Li Poly Ion Photoresist based anode graphite Sol Gel process cathode V2O5 Spin on Electrolyte PEO UC Berkeley 2002 2004 Micro Batteries 1 5 W on SiO2 2 Expose and Process Photoresist 6 Add Photoresist Add and Dry V2O5 Composite 7 3 Pyrolyze and Process Photoresist 4 Remove Photoresist 8 Add Photoresist Add Electrolyte UC Berkeley 2002 2004 Fuel Cell Basics Load H Oxidant Cathode Electrolyte Anode Fuel ions O2 4H 4e 2H2O 6 X lithium 15 X rechargeable lithium e 2H2 4H 4e Like a battery current is directly dependant on available electrode area Unlike a battery the reactant is stored elsewhere Methanol energy density is 17 6 kJ cm3 O UC Berkeley 2002 2004 MEMS Fuel Cell Current Generation Toshiba 1 cm3 hydrogen reactor 15 efficient Produces 1 watt Transients may be too slow for low duty cycles Fraunhofer Next Generation Planar Arrays Fraunhofer 100 mW cm2 Stanford 40 mW cm2 research in progress S J Lee et al Stanford University UC Berkeley 2002 2004 Micro Heat Engines MEMS scale parts for meso scale engine 1 cm3 volume 13 9 W Poor transient properties Micro size heat engine ICE s thermoelectrics thermoionics thermo photo voltaics via controlled combustion Meant for microscale applications with high power needs UC Berkeley 2002 2004 Bencharmarking Radioactivity High theoretical energy density Power density inversely proportional to half life Demonstrated power on the order of nanowatts reactor less Obvious Environmental concerns UC Berkeley 2002