Cross-talk noise in repeater networks

Abstract

To avoid unjustified optimism in noise prediction in the absence of accurate analysis of switching activity, commercial tools have to make the most conservative assumption about aggressor switching which leads to an overly pessimistic result in situations of strongly correlated signals connected by repeaters. Such pessimism in the noise analysis hides out the effect of interleaved repeaters used to minimize noise in long, coupled interconnects. In the proposed approach we accurately analyze coupled networks with repeaters by considering timing and logical correlations, and multiple signal switching. We demonstrate an 81% reduction in the number of false noise violations in a design where interleaved repeaters are used for crosstalk critical global interconnects.