Abstract: | A neuro-space mapping (Neuro-SM) for modeling heterojunction bipolar transistor (HBT) is presented, which can automatically modify the input signals of the given model by neural network. The novel Neuro-SM formulations for DC and small-signal simulation are proposed to obtain the mapping network. Simulation results show that the errors between Neuro-SM models and the accurate data are less than 1%, demonstrating that the accurcy of the proposed method is higher than those of the existing models. |