A comparison between HBt small-signal model optimization using a genetic algorithm and direct parametric extraction

This work for the first time shows that physically meaningful, wideband, multi-bias small-signal modeling of HBTs can be efficiently and accurately achieved using a genetic Algorithm (GA). The physical significance of the equivalent circuit parameters extracted by the GA was checked using a Direct Extraction Technique (DET). The two procedures were applied to HBT S-parameters measured at different bias points. The simulated S-parameters match very well with the measured ones over the whole frequency range investigated. For each point we obtained quite a good agreement between the parameters extracted by the DET and by the GA, which demonstartes the ability of the GA to efficiently extract a physically significant HBT small-signal model.