Application of design of experiments methodology to model the effect of multiple parameters on simultaneous switching noise

A new methodology is developed for efficiently analyzing simultaneous switching noise (SSN) using design of experiments (DOE) to model the effect of multiple adjustable parameters on SSN responses, such as ground bounce and local supply droop. Each SSN response is fitted to a 2nd order polynomial equation allowing quick calculation of SSN as a function of load capacitance, number of drivers switching, power/ground inductance, power/ground resistance, on-chip de-coupling capacitance, and rise/fall time of the driver input pulse. This method is applied to two separate driver types to demonstrate the approach, and show that certain parameters are more or less significant for one driver, but not the other. Statistical analysis shows that only two of the five responses are unique to SSN. Drivers are characterized using this method, and the analytical expressions provide immediate feedback for making engineering decisions regarding future IC pad-rings and package designs. Results are presented to demonstrate the validity of the proposed methodology and its application for optimizing SSN designs.