Monolithic finite gain amplifiers employing active voltage attenuators in the feedback and a charge conserving macromodel for MOSFETs
Date
Authors
Major Professor
Advisor
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Altmetrics
Abstract
Two separate topics, both focused on analog and mixed signal monolithic circuit design, are presented. The first concentrates on the design of monolithic finite gain amplifiers employing active voltage attenuators in the feedback loop. Three MOS active voltage attenuators suitable for realizing finite gain amplifiers along with an operational amplifier are investigated. Monolithic single input amplifiers, two or more input summing/subtracting amplifiers, and differential amplifiers are readily attainable with this approach. The attenuators and amplifiers are theoretically analyzed, simulated on SPICE, and experimentally characterized in the laboratory. The performance of the attenuators and amplifiers is characterized with experiments by measuring the accuracy of the gain, the range of the linear operating region, and the degree of linearity. It is verified experimentally that amplifiers can be built with gains that are accurate to 0.5-2% of the design value. A differential amplifier designed in a single 5V process was experimentally characterized and exhibited a maximum signal output to total non-signal output ratio of 66.2dB at an output amplitude of 482mV (0-P);The second topic concentrates on developing a charge conserving macromodel for MOSFETs. The mechanisms which make MOSFET switches inject charge is reviewed. Based on the review and the nature of the parasitic capacitances in the MOSFET, a charge conserving macromodel is developed. This macromodel provides a convenient and accurate means for simulating the non-ideal effects of charge injection of MOSFET switches. A simple sample-and-hold circuit for characterizing the accuracy of the macromodel was designed, fabricated, and tested. Experimental results for this test circuit showed a maximum difference between the simulated and measured errors of 1.31 mV over a wide range of input voltages.