Analysis And Design Of PI And EMI In PCB-Level For High-speed Circuits

As the developing period of the products is becoming shorter and shorter, the signal transmission problems are also becoming more and more complicated. So the analysis and design of the high-speed circuits are the main issue home and abroad. In the next decade, the PI and EMI problems will be more important as the work frequency is becoming higher and higher. The signal via, grounding via and the continuity of the reference plane can all introduce the power distribution network noise. These noises can not only suppress the PI property of the circuits, but also lower the noise margin, and become the source of the electromagnetic interference. If we don’t pay enough attention in the initial design period of the product, then a lot of testing work and cost will be spent in the later period of the product. So taking the PI and EMI problems into consideration in the PCB-level of the high-speed circuits is the best choice for us to saving time and cost.So in this paper, based on the Maxwell theory, we derive the resonant equation of the arbitrary shape PCB from the mode method and the Green’s function method, and extend the existed mode method from the single PMC boundary to the PEC when there are ground vias around the boarder of the PCB or even the PMC-PEC hybrid boundary which stands for the case that the ground vias are added only in part of the boarder, and then we conclude the self-impedance and the mutual-impedance under the different boundary condition. To suppress the noise between the power and ground planes, we study the capacitive device and structure, such as:decoupling capacitor, stitching capacitor and embedded filter. To decrease the noise caused by the ground via, we conclude the placement of the ground via when the noise source is located in the different place in the PCB. Then the absorbing material is introduced as the optimal solution of the PI and EMI problems. At the last, but not the least we show the relationship between the PI and EMI and realize the PI-EMI co-analysis.