Thesis Defense: Energy Efficient Control Power Management Circuits...


Event Speaker: 

Saurav Bandyopadhyay

Event Location: 

34-401A (Grier A)

Event Date/Time: 

Friday, April 26, 2013 - 9:00am

COMPLETE Doctoral Thesis Title:
Energy Efficient Controlfor Power Management Circuits
operating from nano-watts to watts
Energy efficiency and form factor are the key driving forces in today's power
electronics. All power delivery circuits, irrespective of the magnitude of
power, basically consists of power trains, gate drivers and control circuits.
Although the control circuits are primarily required for regulation, these
circuits can play a crucial role in achieving high efficiency and/or minimizing
overall system form-factor. In this thesis, power converter circuits, spanning a
wide operating range- from nano-watts to watts, are presented while highlighting
techniques for using digital control circuits not just for regulation but also
to achieve high system efficiency and smaller system form-factor.

The first part of the thesis presents a power management unit of an autonomous
wireless sensor that sustains itself by harvesting energy from the
endo-cochlear potential (EP), the 70-100mV electrochemical potential inside the
mammalian inner ear. Due to the anatomical constraints, the total extractable
power from the EP is limited to 1.1-6.3nW. A low switching frequency boost
converter is employed to increase the input voltage to a higher voltage usable
by CMOS circuits in the sensor. Ultra-low power digital control circuits with
timers help keep the quiescent power of the power management unit down to
544pW. Further, a charge-pump is used to implement leakage reduction techniques
in the sensor. This work demonstrates how digital low power control circuit
design can help improve converter efficiency and ensure system sustainability.
All circuits have been implemented on a 0.18um CMOS process.

The second part of the thesis discusses an energy harvesting architecture that
combines energy from multiple energy harvesting sources- photovoltaic,
thermoelectric and piezoelectric sources. Digital control circuits that
configure the power trains to new efficient system architectures with maximum
power point tracking are presented, while using a single inductor to combine
energy from the aforementioned energy sources all at the same time. A dual-path
architecture for energy harvesting systems is proposed. This provides a peak
efficiency improvement of 11-13% over the traditional two stage approach. The
system can handle input voltages from 20mV to 5V and is also capable of
extracting maximum power from individual harvesters all at the same time
utilizing a single inductor. A proposed completely digital time-based power
monitor is used for achieving maximum power point tracking for the photovoltaic
harvester. This has a peak tracking efficiency of 96%. The peak efficiencies
achieved with inductor sharing are 83%, 58% and 79% for photovoltaic boost,
thermoelectric boost and piezoelectric buck-boost converters respectively. The
switch matrix and the control circuits are implemented on a 0.35um CMOS
process. This part of the thesis highlights how digital control circuits can
help reconfigure power converter architectures for improving efficiency and
reducing form-factors.

The last part of the thesis deals with a power management system for an offline
22W LED driver. In order to reduce the system form factor, Gallium Nitride
(GaN) transistors capable of high frequency switching will be utilized with a
Quasi-Resonant Inverted Buck topology. A burst mode digital controller and
driver have been used to perform dimming control and power factor correction
(PFC) for the LED driver. A custom controller and driver IC was implemented in
a 0.35um CMOS process. The LED driver achieves a peak efficiency of 90.6% and a
0.96 power factor. The digital nature of the controller helps remove the
passives that would be normally required in analog controllers.
Presenter's Affiliation: MTL
Thesis Supervisor(s): Prof. Anantha P. Chandrakasan