Accurate Sensitivity Analysis of the Dynamic Characteristics of a Power Converter

Proper design of a switching power supply regulator loop requires proper assessment of the impact of component value variations upon the loop frequency response characteristics. Worst-case phase margin and worst-case gain margin are pieces of information a designer can use to assess whether or not the converter will be properly regulated for given component value variations. To find this type of information reliably using sensitivity analysis requires first reliably determining the loop frequency response. Models can be used to determine the converter loop's frequency response but they do not accurately predict the converter's dynamic behavior at higher frequencies whereas brute force simulation, though more accurate than modeling, can be time-consuming. A novel approach is proposed which uses the steady-state analysis capabilities of the software tool SUPRA coupled with Fast Fourier Analysis in order to obtain frequency response characteristics more reliably than through the use of models and more quickly than through the use of brute force simulation. A two-stage approach is used to reduce the complexity of determining the loop frequency response. The search of the combination of component values providing the worst case frequency response characteristics is performed using nonlinear optimization.