An ANFIS based fuzzy controller for corner compensation

Abstract

As threshold voltage of CMOS transistors is the main parameter that takes effect from process variations, in this paper a novel method for corner detection is presented which senses the variations of fabrication process through threshold voltage of the devices. A new general purpose 2-input, 2-output, 25 rules, ANFIS based fuzzy controller is proposed to compensate the variations subsequently. In this controller novel structures are presented for each block including membership function generator, Min-Max selector and defuzzifier. As an application, bias points of comparators of a typical flash ADC are controlled through introduced system in order to compensate the process variation effects and minimizing total power consumption consequently. Due to differential structures used in the architecture of the blocks, major part of the power supply noise is rejected. The Hspice (level 49) simulation results are given using a generic 0.35 lm standard CMOS technology parameters and power supply of 3.3 V with total power consumption of 15.6 mW for 7.4 MFLIPS. Because of simple and symmetrical circuitry, layout of the proposed controller is very compact, about 410 lm 9 210 lm.

An ultra high-speed high-resolution low-offset low-power voltage comparator with a reliable offset cancellation method for high-performance applications in 0.18 µm CMOS technology

“…Process variations, during the integrated circuit fabrication, degrade the performance of the designed chips in terms of the speed, accuracy and so on through engendering some non-idealities like introducing offsets to the systems [8].…”