Robust ADC testing with very long time records An Adaptive Fourier Analyzer based approach

A common method to test AD converters is the sine fitting. If very large dataset needs to be processed (long records), the iterative four parameter sine fit algorithm is very sensitive to the initial value of the fundamental frequency estimator. Even a very slight error in the initial guess can spoil the fit because of the local minima in the error surface. We propose a robust algorithm to fit the sine wave with a resonator based state observer, together with an Adaptive Fourier Analyzer (AFA). The method has the advantage that amplitude and frequency drift can also be detected and compensated.