Crosstalk pulsewidth calculation
Section snippets
Introduction and background
Advancements in the field of VLSI have lead to a decrease in device geometries, high device densities, high clock rates and small signal transition times. Thus, interconnection lines that were once considered to be electrically isolated can now interfere with each others and have an important impact on system performances. One such interaction caused by parasitic couplings between wires is known as crosstalk. A number of simple crosstalk noise models have been proposed in literature. Refs. [1],
Maximum crosstalk evaluation
The inductive contribution on crosstalk voltage is always smaller than the capacitive one in VLSI circuits due to the high interconnection resistance. In this paper, we neglect the inductive coupling and focus exclusively on capacitive crosstalk voltage VC. We have shown in [7] that it is possible to liken a distributed coupled-line RC system crosstalk waveform with a lumped RC system by using a correlation method (similar to a model order reduction method). Comparisons between simulated and
Determination of VCmax/2 times t1 and t2
Let NpC be the capacitive crosstalk voltage pulsewidth taken at VCmax/2. It can be calculated according to the following relation: NpC=t2−t1. The capacitive crosstalk voltage VC reaches VCmax/2 at t1 and t2. Since we correlate the distributed crosstalk waveform with the analytical one, calculate NpCeq consists in determining in the time domain the solutions of the following equation:
However, it is not possible to
Spice simulation comparisons
From the analytical expressions (16), (19), we verify if the new times t1 and t2 calculated values are more or less accurate than the former ones leading to a better knowledge of the pulsewidth/critical timing interval when crosstalk voltage must be taken into account to assure the signal transmitted integrity and propagation along the lines. A lot of validations have been realized to verify the two expressions viabilities. Some of them are presented in this paper.
Comparisons Tables (Table 2,
Conclusion
In this paper, we propose analytical expressions to calculate very accurately capacitive noise voltages characteristics such as the maximum voltage amplitude and the noise pulsewidth taken at maximum amplitude half height which corresponds to the critical time domain when the coupling effects must be taken into account. All these expressions have been deduced from the correlation method/model order reduction method we presented in previous works; this method allowing to assimilate a distributed
Reference (8)
- K. Shepard, et al., Globalharmony: Coupled Noise Analysis for Full-Chip RC Interconnect Networks, in: Proceedings of...
- P. Heydari, M. Pedram, Capacitive Coupling Noise in High-Speed VLSI Circuits, in: IEEE Transactions On CAD of...
- J. Cong, D. Zhigang Pan, P.V. Srinivas, Imporved Crosstalk Modeling for Noise Constrained Interconnect Optimization,...
- R. Suaya, Interconnect Modeling for High Speed Digital Circuits-the Role of RLC Coupling, in: Proceedings of MSM 2002,...
Cited by (2)
Analytical model of single event crosstalk in near space
2012, Wuli Xuebao/Acta Physica SinicaAccurate analytical model for single event (SE) crosstalk
2012, IEEE Transactions on Nuclear Science
- 1
Tel.: +33 467 41 86 42; fax: +33 467 41 85 00.