Modeling and estimation of crosstalk across a channel with multiple, non-parallel coupling and crossings of multiple aggressors in practical PCBs

Increases in the cost of printed circuit boards (PCB) due to increases in the layer count have led to the design of PCB stack-ups with broadside coupled signals. The broadside coupling of signals in the adjacent layers also leads to crosstalk, which can at times be difficult to model and quantify in terms of its impact on the receiver eye opening. The difficulty stems from the fact that in most boards, multiple occurrences of broadside coupling occur between the signal traces at various angles. The challenges involved in rapid modeling include generating models for the broadside coupled section without the overhead of time-consuming Full-Wave simulations. Full wave simulations are time and memory intensive, especially for coupled traces that are at an angle, and real board designs can have hundreds of these angles. The simulation challenges include accurately predicting the impact of crosstalk on the bit error rate (BER). Focus is on alleviating the modeling challenges by breaking the overall geometry into small, unique sections and using either a Full-Wave or fast equivalent per-unit-length (Eq. PUL) resistance, inductance, conductance, capacitance (RLGC) method or a partial element equivalent circuit (PEEC) for the broadside coupled traces that cross at an angle. The simulation challenge is resolved by seamlessly integrating the models into a statistical simulation tool that is able to quantify the eye opening at BERs that would help electrical designers in locating crosstalk sensitive regions in the high speed backplane channel designs. The dielectric media surrounding the traces are not homogeneous and this increases the Far End Crosstalk (FEXT) and throughput (THRU), thereby reducing the signal to noise ratio of the high speed link. FEXT crosstalk impact on eye opening at a specified bit error rate (BER) at different signal speeds for broadside and edge side single-ended and differential coupled traces in inhomogeneous media is investigated and compares the results against homogeneous media models. A set of design guidelines regarding the material, coupled length and stackup parameter selection is formulated for designers based on the signaling speeds.;Additionally, due to the fiber weave effect, designers have also been forced to route signal traces in a zig-zag fashion rather than in straight lines. This type of routing tends to be roughly periodic between the victim and aggressor. The periodic coupled routing creates periodic resonances in the near end crosstalk (NEXT) and nulls in throughput (THRU) transfer functions due to Floquet modes. Due to the periodic resonances, the crosstalk is aggravated, which reduces the signal to crosstalk ratio. The major objective of the present study is to determine quantitatively the effect of crosstalk due to periodic routing. Another objective is to help designers figure out the "do's" and "don'ts" of broadside coupled routing for higher signaling rates. A set of design guidelines based on the angle of routing, length of period and signaling speeds was formulated for designers using statistical bit-error rate (BER) analysis.;A new methodology is proposed to generate BER contours that capture the Tx driver jitter and ISI through the channel accurately using unique waveforms created from truth table bit combinations. It utilizes 2N short N bit patterns as waveforms and jitter correlation from current bit pattern into adjacent bit patterns to get equivalent transient simulation of a very large bit pattern. The method utilizes the residual ISI from any current bit pattern to next successive bit pattern to account for ISI accurately. The statistical eye diagram generated with the above approach includes non-ideal channel characteristics, including inter-symbol interference (ISI), crosstalk (XTK) from nearby aggressor channels, and jitter through the channel.