A Novel De-embedding Method Based on Artificial Neural Network

De-embedding methods are used to extract the S-parameters of the device under test (DUT) from the combination of fixture + DUT. The rapid iteration of electronic products raises higher demands for a fast de-embedding method. A novel de-embedding method without the need of any calibration standard is proposed in this article. Firstly, the time domain reflectometry (TDR) is used to extract the reflected wave from the DUT, and then the S-parameters of the DUT are extracted from the reflected wave of the DUT by using two previously trained artificial neural networks (ANNs). The numerical and measurement examples demonstrate that the proposed method achieves both higher efficiency and accuracy than the traditional de-embedding method. The proposed method and the same ANNs can be applied for different fixtures, which greatly reduce the measurement time.