Analog Processing based on Quasi-Infinite Resistors
Abstract
This work proposes a technique for design of CMOS analog integrated circuits such as offset compensated amplifiers, low-frequency filters, programmable current mirrors and membership function generators, based on high-value (quasi-infinite) resistors. The proposed technique incorporates transistors operating in weak-inversion mode in order to reduce the area requirements and minimize the DC-offset. In addition, improvement on both, noise performance and linearity, are achieved along with an enhanced speed-accuracy-power tradeoff. Those features make easier the processing of low-frequency signals and allow the design of systems with multi-decade tunability of gain and frequency. The presented circuits are attractive for implementation of high-accuracy processors for signal conditioning as well as architectures usually reserved to digital approaches, for instance neural networks, adaptive filters, and neuro-fuzzy systems, to mention a few. Characterization through computer simulations, statistical analysis and experimental measurements of prototypes in a double-poly, three metal layers, 0.5μm CMOS technology are reported. The attained results follow the course anticipated in the design of the circuits.
Keywords
CMOS, amplifiers, filters, fuzzy logic, quasi-infinite resistors.