SIGNAL PROCESSING WITH ANALOG CIRCUIT Chun Lo Analog

SIGNAL PROCESSING WITH ANALOG CIRCUIT Chun Lo

Analog circuit design Main disadvantage: low precision Due to mismatch in analog circuit components such as transistors, resistors and capacitors e. g. Amplifiers: lower input signal dynamic range Comparator: lower signal resolution Transistors: lower accuracy in current mode DAC Advantage: Less power consumption Possibly smaller area

Mismatch correction By trimming device Larger device, layout technique in component mismatch Autozeroing, correlated double sampling in amplifiers Cost of extra area, power and design complexity By in herent programming elements Floating gate MOSFETs

First paper Floating Gates Transistors For Precision Analog circuit Design: An Overview Srinivasan, Graham and Hasler Second paper Analog Floating –Gate On chip Auditory Sensing System Interfaces Hasler, Smith, Graham, Ellis and Anderson

First paper Structure of Floating Gate MOSFET Programming floating gate transistors Programming precision Charge retention Application examples of Floating Gate MOSFET Differential DAC ADC pairs

MOSFET MOS Transistor Floating Gate MOSFET The gate is completely surrounded by Si. O 2 No DC to ground Used in Flash memory, ROM

Programming FG-MOSFET Setting the DC voltage Add or remove charge Mainly two methods Fowler As Hot Nordheim tunneling a global eraser for all floating gates electron injection Program each floating gate to the desired value individually

Programming precision What is the smallest drain current change A single electron change results in an accuracy of or 12 bits

Charge Retention Short term drift Observed immediately after programming Proportional to the amount of change that is programmed into the floating gate Long term charge loss Themionic emission Related to both temperature and time

Applications Differential pair DAC ADC

Second paper Advance in analog VLSI circuits Performance close to DSP More option for designeer Analog Digital Partition Signal Processing Circuits Noise Suppression for speech enhancement Speech Recognition

Analog Digital Partitioning Power and space No overhead for tuning and programming Work in subthreshold save power SN VS Cost

Analog Signal Processing

Noise suppression Main Idea: Attenuate the noise components

Speech Recognition HMMs States, transition probabilities and Hidden states are words, output is audio signal produced by speaker Recognition problem: what is the state of the system