Abstract: This paper proposes a new model for parametric fault diagnosis in analog circuits, which is one of the most challenging problems in circuits and systems. This new model is based on the eigenvalue and phase difference from the time series of the output voltage of the circuit under test (CUT). The phase deviation information of the circuit is obtained via the sampling voltage time series. The sampling voltage time series is reorganized to be a matrix, and dominant eigenvalue of this matrix is obtained accordingly. Finally, by comparing the phase deviation and the dominant eigenvalue of the CUT with those of the fault free circuit in absolute relative error criteria, fault location and parameter identification can be accomplished. Experimental results show that the proposed method performs well in both fault location and parameter identification with very few access points and relatively low computation cost, moreover, fault location and parameter identification can be realized simultaneously, which makes it an effective and efficient method for fault diagnosis of analog circuits.