Issue link: https://iconnect007.uberflip.com/i/1014812
AUGUST 2018 I DESIGN007 MAGAZINE 61 substrate or dielectric between the PCB trace and the adjacent reference plane reduces sig- nal crosstalk. The crosstalk analysis is continued with simulation topology depicted in Figure 3 for each transmission line model (i.e., differential mode) listed in Table 2; the transmitting end of the aggressor's positive and negative lines are assigned as port P1 and P2, respectively, while receiving end of the victim's positive and negative lines are assigned as port P3 and P4, respectively. The receiving end of the aggres- sor pair and transmitting end of victim pair are terminated. The coupled spacing and length between the aggressor and victim are set triple the trace width and 3 inches, respectively. These four transmission line models are laminated with medium-loss substrate mate- rial (i.e., 3.6 Dk and 0.01 Df). Each model has a trace thickness of .0012". Model 3A is set as microstrip, with a substrate thickness of .003", trace width of .006", and intra-pair spacing of .008" to achieve characteristic impedance of 92.1 ohms. Meanwhile, model 3B has the same parameter setting as 3A, except the substrate thickness is increased to .004" to achieve char- acteristic impedance 105.8 ohms. The substrate thickness of model 3B is increased by not more than .001" versus model 3A to limit the trace impedance of both models within ±10% toler- ance of the nominal 100 ohms. On the other hand, model 4A in Table 2 is set as a symmetrically centered stripline, with Table 2: 2D transmission lines in differential mode for varying dielectric thickness. Figure 2: Simulated plots of far-end crosstalk for transmission line models listed in Table 1. Figure 3: Crosstalk simulation topology for transmission line models listed in Table 2.