An empirical methodology for foundry specific submicron CMOS analog circuit design

Analog CMOS amplifiers are the building blocks for many analog circuit applications such as Operational Amplifiers, Comparators, Analog to Digital converters and others. This dissertation presents empirical design methodologies that are both intuitive and easy to follow on how to design these basic building blocks. The design method involves two main phases. In the first phase NMOS and PMOS transistor design kits, provided by a semiconductor foundry, are fully characterized using a set of simulation experiments. In the second phase the user is capable of modifying all the relevant circuit design parameters while directly observing the tradeoffs in the circuit performance specifications. The final design is a circuit that very closely meets a set of desired design specifications for the design parameters selected. That second phase of the proposed design methodology utilizes a graphical user interface in which the designer moves a series of sliders allowing assessment of various design tradeoffs. The theoretical basis for this design methodology involves the transconductance efficiency and inversion coefficient parameters. In this dissertation there are no restrictive assumptions about the MOS transistor models. The design methodology can be used with any submicron model supported by the foundry process and in this sense the methods included within are general and non-dependent on any specific MOSFET model (e.g. EKV or BSIM3). As part of the design tradeoffs assessment process variations are included during the design process rather than as part of some post-nominal-design analysis.;One of the central design parameters of each transistor in the circuit is the MOSFET inversion coefficient. The calculation of the inversion coefficient necessitates the determination of an important process parameter known as the Technology Current. In this dissertation a new method to determine the technology current is developed. Y-parameters are used to characterize the CMOS process and this also helps in improving the technology current determination method. A study of the properties of the technology current proves that indeed a single long channel saturated MOS transistor can be used to determine a fixed technology current value that is used in subsequent submicron CMOS design. Process corners and the variability of the technology current are also studied and the universality of the transconductance efficiency versus inversion coefficient response is shown to be true even in the presence of process variability.