In the thesis, my consultant’s, Dr. János Ladvánszky’s of noise-analysis in time domain will be presented. This analysis uses quadrature demodulation to determine the noise parameters (amplitude noise and phase noise). With the help of the presented offline method the relative amplitude noise and phase noise of a previously fabricated negative resistance oscillator have been determined. In addition to that, the auto- and cross correlation functions of the noise parameters have been calculated as well. Using these functions, the autocorrelation function and the power spectrum of the output signal of the oscillator have been determined. This method is shown in the Appendix I. of . This power spectrum and spectra determined by other methods are compared. According to the results of the comparison we state, that the equation used is adequately close at the fundamental frequency of the oscillator.
In the second half of the thesis, I present a two-step spectral cleaning algorithm made in Matlab Simulink system, and its effects in case of a 1.8 GHz oscillator. After the succesful results of the two-step spectral cleaning algorithm, the first steps of a realtime one-step spectral cleaning algorithm are implemented on an FPGA. The complete realisation of the one-step and the two-step spectral cleaning algortihms will be done during my PhD project.