Non-causal behavior: The cause for concern

The requirement to verify the causality of electromagnetic model response and ensure validity of transient simulations results is gaining momentum for those skilled in the art. In order to capture causal models and verify causal behavior, the engineer must understand which parameters have an impact on model behavior as well as those parameters which inhibit lab verification of said model. Specifically, as dielectric properties continue to trend towards lower-loss, it may become more critical to understand the impact between dissipation factor and model causality. The SI engineer must have a keen understanding of the magnitude by which a model´s behavior can deviate from a causal response before he/she becomes concerned about their analysis. It is common knowledge that demand has and continues to drive computer systems data rates to ever-increasing speeds. As communication interface speeds increase, signal integrity engineers require more accuracy out of the transmission line models (models need to account for more effects). The level of accuracy now required has caused the complexity of transmission line models to increase to a point where concatenation of the models may induce significant error. As such, model developers must guarantee model behavior is causal, not merely frequency dependent. Non-causal models may generate inaccurate time domain simulation results as well as complicate a tool´s convergence process; even if the model´s behavior agrees with measured behavior in the frequency domain. Therefore, model developers must understand the model creation, checking, simulation process, and how to interpret results within chosen tool suites. Equally important, one must realize that model frequency content, step size, length and other parameters that may impact a checking tool´s ability to accurately flag causality violations and how those violations, “real” or “false”, may impact system-level simulation results [8]. This pape- will discuss the interaction between the measured-model acquisition process, the ability to verify causal behavior and the ramifications of non-causal models or inaccurate interpretation of causality-checking results.