Microwave device model validity assessment for statistical analysis

Model accuracy is a key consideration when using circuit simulation for microwave designs in general but normally ignored in statistical analysis. A measure of model validity against process variation, MAP, is developed from theoretical considerations. A demonstration of MAP is given for the simple case of a microstrip line and the industry-standard MLIN model. More complex cases are examined for coupled differential microstrip on module for linear design and MMIC HBT diodes for nonlinear design. The impact on MAP of both process variations and model changes is demonstrated. Limitations and extensions to the technique are discussed for other factors impacting the selection of models based on their accuracy.