Comparison and Consistency of On-PCB Microstrip Line Modeling by 3D Fullwave and 2D Quasi-Static Approaches for High Speed Packaging System Signal Integrity Simulations

The S-parameter model and TDR impedance of the uniform microstrip lines by 3D fullwave approach and 2D quasi-static approach are compared. In general, the model consistency is good on most S-parameter components with exceptions in return loss and simulated TDR. The main root cause of this inconsistency has been found to be the loose S-parameter convergence error setting in the 3D electromagnetic simulations. We can use tighter error setting, however this will bring a denser mesh and then longer electromagnetic simulation time, larger computer memory requirement, and even some simulation stability issues. The ideal transformer model, in either the S-parameter or the SPICE controlled source formats, is applied to transform the impedance of the structure and then correct the small error in return loss or simulated TDR without too tight error setting in the 3D fullwave electromagnetic simulations. The expected impedance target is either by 2D approach which easily uses very dense mesh or by expected impedance convergence point from the 3D approach using denser and denser meshes. The corrected model is based on calibration of the uniform transmission lines and can be applied to either uniform transmission line or 3D transmission line of similar wire dimension. Although this correction almost does not affect the simulated eye diagram in most applications, it does provide a more accurate model for TDR simulation which is helpful for obtaining an accurate impedance value in package system design. An interesting phenomenon of single-ended transmission zero in the coupled microstrip lines is also studied in this paper. It can be used as an easy-to-measure signature character of the coupled microstrip structure. It is helpful to verify the model-to-hardware correlation.