Nonlinear noise theory for IMPATT diodes

Previous studies have yielded good qualitative understanding of noise characteristics in IMPATT diodes under large-signal conditions. However, a coherent approach to describing correlation effects in a rigorous manner and demonstrating their relationships to AM and FM noise spectra has been lacking. The present work is intended to fill this need in the framework of a Read diode model with small back-bias. Autocorrelations and noise spectra are periodic functions of time in large-signal operation. Resulting correlation effects are quantified and shown to cause unequal AM and FM noise spectra. The short-circuit spectrum of the avalanche current is calculated and found to have resonant terms at the harmonics of the large-signal frequency. A matrix relationship with the open-circuit avalanche field spectra is derived. Simple analytical expressions are developed for these spectra by considering the interactions between the low-frequency and fundamental-frequency amplitudes. The AM and FM spectra have noise poles at the differential and the large-signal avalanche frequencies, respectively. The open-circuit noise spectra of the diode voltage are used to study noise properties of IMPATT amplifiers and oscillators. Conditions for noise stabilization are discussed. The AM and FM voltage spectra are coupled unless these conditions are met. Numerical results for the noise measures and noise-to-carrier ratios versus signal level are presented. The AM noise is found to be smaller than the FM noise when large bias resistance is used to suppress upconversion of low-frequency noise into the AM spectrum. Strong increases are found in the FM noise measure versus signal level. Previous theories are discussed, and the connections with the present work are established. A new large-signal IMPATT model is introduced to study the effects of the reverse saturation current on the efficiency and the noise measure. The present theory should be useful in designing diodes for optimum power-noise performance owing to the simplicity and the minimum of restrictive assumptions in the derived analytical expressions.