文档介绍：518 Chapter 18
of strongly coupled design variables. It allows simultaneous optimization of
multiple variables to minimize phase noise subject to certain design constraints.
Especially, a method of selecting on-chip spiral inductors is devised in the
course of the oscillator optimization.
This chapter is a review of the work using GNP in design of integrated
LC VCOs and anized in the following manner. Section illustrates
the methods of GNP with an example. In Section , a specific oscillator
topology is chosen as a design example, design constraints are imposed on
the oscillator and the expressions describing phase noise of the oscillator are
provided. Section explains the details of our design and optimization
process using GNP. Section further discusses the main result of integrated
LC VCO optimization from fundamental physical points of view. Elaborate
simulation results of the optimized VCO are presented in Section with
accurate prediction of phase noise. Section presents the experimental
results pares the performance of our VCO to that of other reported LC
oscillators to prove the adequacy of our design methodology.
. Graphical Nonlinear Programming
In this chapter, we will use a generalization of the well-known Linear pro-
gramming (LP) [27] to optimize a nonlinear objective function subject to
multiple nonlinear constraints by graphically visualizing the objective func-
tion and constraints. We will refer to this generalized approach as GNP. In this
section, the method of GNP is illustrated through an example containing both
linear and nonlinear constraints:
Use GNP to minimize an objective function subject to
The constraints given in () are visualized as the shaded area in the xy-
plane of Figure . The objective function can be visualized using curves
described by g(x, y) = k for different values of k. The minimum of g(x, y)
can be found by changing the parameter k in g(x, y) = k and mov