Extraction of small signal equivalent circuit model parameters for statistical modeling of HBT using artificial neural

We found different performances for the same device due to the variations in the process from die to the other on the same wafer or on another one. Yield analysis becomes one of the important tools into commercial computer aided design (CAD) programs. Statistical issues are crucial in yield analysis for microwave circuits. Yield analysis needs accurate statistical properties between the parameters of devices´ models to reflect correctly the physical variations. Normally, on the level of the device modeling, the statistical properties between the model parameters like means and standard deviations are noisy by using the known techniques (optimization-based and direct) for extracting the small signal equivalent circuit model parameters of active microwave devices. We introduce how is artificial neural network (ANN) accurate and efficient statistical extraction method for small signal model parameters of hetero junction bipolar transistor (HBT). Utilizing this methodology provides a robust statistical model for our device.