An enhanced operational amplifier macro model based on artificial neural network

Abstract

AbstractAn enhanced operational amplifier (OPAMP) macro model based on artificial neural network (ANN) is developed for analog circuit simulation. The model uses ANN to capture the relation between the inputs and outputs (currents and voltages) of the circuit module. Both direct current (DC) and alternative current (AC) signals are considered in the model. By adopting the adaptive sampling algorithm, the amount of data required for model training can be significantly reduced and the accuracy of model fitting can be apparently improved. The model is also validated by the simulated data from OPAMP, Bandgap, and LDO circuits. Compared with SPICE model, enhanced ANN OPAMP macro model has nearly 8 times faster simulation speed without apparently degrading the model accuracy. The predictions made by the neural network are also compared to the experimental measurement results of LDO circuit fabricated in 0.18‐μm process. Both simulation and experimental results show the feasibility and accuracy of the proposed enhanced ANN OPAMP macro model.