Non-Foster Matching Circuit Synthesis Using Artificial Neural Networks
Qianyi Li, Ting‐Yen Shih · 2021
In this work, we study the use of artificial neural networks for non-Foster circuit synthesis. Non-Foster matching circuits are often used in military and civilian communication systems to provide wide instantaneous bandwidth. The design of active non-Foster circuits is very challenging and only a few practical designs have been reported to date due to the complexity, the non-linearity, and the inherent instability of these circuits. We propose using an artificial neural network algorithm to allow for automated non-Foster circuit synthesis. Within the examined input impedances, where the equivalent negative inductance of the circuit ranges from -91.6 nH to -51.9 nH, and the positive resistance ranges from 6.5 Ω to 15.4 Ω, the mean synthesis error of the inductance and the resistance are 0.061% and 0.078%, respectively. The synthesized circuits are stable.