Accuracy and performance of passive transmission line macromodels based on optimal matrix rational approximations

Elfadel, Ibrahim Abe M.; Dounavis, A.; Huang, H.-M.; Nakhla, M.; Ruehli, A.E.; Achar, R.

doi:10.1109/epep.2002.1057948

Cited by 8 publications

(4 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Such structures can be analyzed with very good efficiency by MRA-based techniques. A preliminary assessment has been attempted in [15] on a limited number of cases. Significant advances on line macromodeling have been achieved in the last few months, and a more exhaustive analysis using the most recent algorithms on a large number of line cases is needed.…”

Section: A Delay Extractionmentioning

confidence: 99%

“…As mentioned above, the conversion into time-domain of (8) and (9) is an easy task if all transfer matrices are rational with (11) pure delay terms. Therefore, we approximate both characteristic admittance and delayless propagation operators as (15) We tested different strategies for the generation of this approximation. One choice was to apply the well-known technique of vector-fitting [13], which is known to produce very accurate estimates for the poles and consequently for the overall approximation.…”

Section: B Rational Approximationmentioning

confidence: 99%

“…External Circuit:This implementation is the only one possible without access to the kernel of the circuit solver. The macromodel is generated off-line for each line under investigation as a netlist of circuit elements corresponding to a realization of (8) and (9) combined with (15). Only standard circuit elements are used for this synthesis, particularly resistors, capacitors, controlled sources and ideal delay lines (the latter to synthesize the modal delays).…”

Section: Timing Analysismentioning

confidence: 99%

“…This implementation is equivalent to the previous one, differing only in the way the macromodel circuit elements are stamped in the MNA matrix of the overall network. Equation-Based: Due to the availability of the API for the definition of the line macromodel, it was possible to skip the equivalent circuit synthesis and to derive an equationbased implementation of (8)- (15). The circuit elements of previous implementations are replaced by properly defined internal states, which are automatically initialized, computed and stored during the transient simulation.…”

Section: Timing Analysismentioning

confidence: 99%

See 3 more Smart Citations

Transient Analysis of Lossy Transmission Lines: An Efficient Approach Based on the Method of Characteristics

Grivet-Talocia

Huang²,

Ruehli

et al. 2004

IEEE Trans. Adv. Packag.

Self Cite

126

View full text Add to dashboard Cite

This paper is devoted to transient analysis of lossy transmission lines characterized by frequency-dependent parameters. A public dataset of parameters for three line examples (a module, a board, and a cable) is used, and a new example of on-chip interconnect is introduced. This dataset provides a well established and realistic benchmark for accuracy and timing analysis of interconnect analysis tools. Particular attention is devoted to the intrinsic consistency and causality of these parameters. Several implementations based on generalizations of the well-known method-of-characteristics are presented. The key feature of such techniques is the extraction of the line modal delays. Therefore, the method is highly optimized for long interconnects characterized by significant propagation delay. Nonetheless, the method is also successfully applied here to a short high/loss on-chip line, for which other approaches based on lumped matrix rational approximations can also be used with high efficiency. This paper shows that the efficiency of delay extraction techniques is strongly dependent on the particular circuit implementation and several practical issues including generation of rational approximations and time step control are discussed in detail.

show abstract

Section: A Delay Extractionmentioning

confidence: 99%

Section: B Rational Approximationmentioning

confidence: 99%