Simulation and modelling techniques for noise in radio frequency integrated circuits

In high speed communications and signal processing applications, random electrical noise that emanates from devices has a direct impact on critical high level specifications, for instance, system bit error rate or signal to noise ratio. Hence, predicting noise in RF systems at the design stage is extremely important. Additionally, with the growing complexity of modern RF systems, a flat transistor-level noise analysis for the entire system is becoming increasingly difficult. Hence accurate modelling at the component level and behavioural level simulation techniques are also becoming increasingly important. In this work, we concentrate on developing noise simulation techniques and mathematically accurate noise models at the component level. These models will also enable behavioural level noise analysis of large RF systems.; As a first step, we develop a new quantitative description of the dynamics of stable nonlinear oscillators in presence of deterministic perturbations. Unlike previous such attempts, this description is not limited to two-dimensional system of equations and does not make any assumptions about the type of nonlinearity. We show that the oscillator output is the sum of two stochastic processes: a large signal stochastic process with Brownian motion phase deviation and a small “amplitude” noise process. We further show that the response of the noisy oscillator is asymptotically wide-sense stationary with Lorentzian power spectral density. We also show that the second order statistics of this output are characterized by a single scalar constant. We present efficient numerical techniques both in time domain and in frequency domain for computing this constant.; We also extend this analysis to circuits that are driven by more than one large periodic signal that are corrupted by phase noise. We show that, similar to the one tone case, the output of a nonautonomous circuit driven by two or more large tones is also asymptotically stationary. We show that phase noise of each input signal contributes one additional white noise source that is modulated by derivatives of the steady state response.; These models for autonomous and nonautonomous components of RF circuits will enable one to perform nonlinear noise simulation at the behavioural level for large RF systems. These models can also be used for behavioural level performance optimization and constraint generation for performance optimization and constraint generation for the RF components. (Abstract shortened by UMI.)...