Accurate spectral testing with non-coherent sampling for large distortion to noise ratios

Spectral testing is widely used in characterizing the dynamic performance of analog signals. Achieving accurate test results is an expensive and challenging task. This paper focuses on eliminating the challenging requirement of non-coherent sampling in the presence of an impure signal with large distortion to noise ratios. A new method was proposed that iteratively estimated noncoherent fundamental and harmonics. Comparisons are made with several widely used methods: windowing, 4 parameter sine wave fitting, and FIRE method, to address the need for this new method. They are quantitatively examined for robustness at different input signal purity and non-coherency level. The advantages & limitations of these methods are discussed. Extensive simulation results show the effectiveness and robustness of the proposed method, which can tolerate wide range of signal purity and any level of non-coherency. It can be implemented for on-chip accurate spectral test, and can also be suitable for accurate ADC spectrum test when the pure source is not available.