TABLE OF CONTENTS This chapter summarizes the electrical signaling environment of PCI and mechanical issues related to add-in cards. The objective is to highlight the electrical features of PCI without getting bogged down in details that are primarily of interest to integrated circuit designers. To dig deeper, refer to the current revision of the PCI specification.
Many aspects of PCI's electrical specification are explicitly intended to reduce power consumption. Not only is this environmentally correct, it is essential for mobile and portable devices. PCI is based on CMOS, which means that steady state DC currents are minimal and in fact, most DC drive current goes to pull-up resistors. The bus protocol assures that bus receivers are not allowed to float such that they might oscillate and consume unnecessary power. Finally, the most interesting aspect of low-power consumption is that PCI is based on "reflected wave" switching, rather than the more traditional "incident wave" switching.
Traditional bus architectures, such as the Unibus and Qbus, have stressed the need for proper termination of all bus lines to prevent unwanted reflections. Every signal on a backplane bus is really a transmission line with a characteristic impedance of about 120 ohms. If the ends are not terminated, a pulse travelling down the line will be reflected back from the end possibly causing unwanted interference.
The solution is to terminate both ends of the bus in the characteristic impedance. Figure 5-1 shows a typical termination arrangement. The "Thevenin equivalent" impedance...
