Date of Award

1-1-2010

Document Type

Campus Access Thesis

Department

Electrical Engineering

First Advisor

Paul Huray

Abstract

Though the printed circuit board (PCB) industry advances much more slowly compared to silicon density, the signaling rates of PCBs must continue to scale so that platforms can take advantage of these higher performing processors. The nature of the materials used to make these boards is not well understood, especially at higher frequencies. For example, high humidity models of PCB materials do not correlate well to actual measurements and classic models for losses associated with copper surface roughness break down around 5 gigahertz (GHz). Industry often uses the infinite pole Debye model to simulate the behavior of dielectrics as a function of frequency. This model does not predict high frequency resonances, nor does it account for their effects on signal propagation. Resonances not only diminish the amplitude of signals, but can also cause dispersion when the signal is made up of multiple frequency components. If severe enough, these effects can cause serious data errors.

In one particular set of PCBs, high frequency resonances around 35 GHz were measured in the insertion loss of some microstrips. The cause of the resonance was unknown and unaccounted for in previous measurements and simulations of the board's structure. Thus, the original hypothesis of this thesis was that the high frequency resonances were caused by material contaminants and not by the structure. This work characterizes the high frequency measurements and summarizes the trends for this particular set of boards. It also discusses the possible causes of the resonances and the plausibility of each effect, including transmission line discontinuities, trace impedance mismatch/mixed mode propagation, vias, periodic materials, contaminant resonant absorption, and water/moisture.

Share

COinS