Skip to Main Content
An improved fundamental identification and replacement technique is presented in this paper. The proposed method achieves fast and accurate spectral testing results for approximate sinusoidal signals without requiring coherent sampling. The new method uses data record lengths comprising only prime factors 2 or 3, which improves the computational efficiency. Furthermore, a new algorithm for counting the integer cycles in the data record is developed, which guarantees the robustness even when the signal frequency is close to Nyquist frequency and/or when noise is high. The new method provides an alternative option for testing engineers to achieve a good result in the case of noncoherent sampling without carefully choosing the windows. Theoretical analysis, simulation, and experimental results collaborate to prove the validity of the proposed method. Comparative studies demonstrate that the proposed algorithm achieves spectral testing accuracies similar to those obtained using perfect coherent sampling method. The proposed method is well qualified for high precision spectral testing.