Practical tuning of an OTA-C bandpass biquad via recurrent geometric programming

The geometric programming which can be globally solved special cases of nonlinear problems is operated recurrently with calibrated object and constraints that are based only on ideal behaviors to overcome the unmodeled nonidealities or nuisance factors of a practical current-mode OTA-C bandpass biquad, and successfully tune its responses. By the end of each GP-based optimization, error related to desired goal is computed by measuring practical responses obtained through HSPICE simulations. If the calculated error is significant, next optimization is introduced with calibrated goal based on computed error. A second-order bandpass requirement is successfully tuned within a few iterations. Based on the fact that the HSPICE simulation is the most time-consuming process, this method is very efficient as others heuristic and evolutionary algorithms fundamentally need at least several tens responses measuring.