6.1 Introduction

Early circuit boards were made of bakelite, where components were soldered on to mounting pins and vacuum tubes were plugged into sockets. Insulated wires were used to interconnect sockets, pins, and components. Building a board was labor intensive. Component leads were cut and bent to provide strain relief, interconnecting leads had to be stripped and tinned, and all the parts had to be assembled and soldered by hand.

A great step forward was made when copper/epoxy laminates were introduced. Traces replaced insulated leads. Components were mounted in holes that were plated and tinned. If shielding was needed, then critical circuits were surrounded by metal boxes. The logic was slow enough that circuit traces were routed point-to-point, and there were few cross talk or overshoot problems. Wire-wrap technology was introduced when the number of interconnections could not be handled by traces on two layers. Progress in IC design then increased clock rates, which meant that rise and fall times were shorter. Attempts to operate at higher clock rates were often met with failure, as there were just too many problems in signal integrity. It was soon recognized that wire-wrap technology was a blind alley, and it was necessary to go to multilayer boards with added ground planes to provide circuit performance. The trend to smaller trace widths and multilevel boards soon followed. We often take for granted the amazing technology that we now enjoy. It makes our computers, servers, and cell phones possible.

Two-layer circuit boards are in common use even today. These boards are copper-clad laminates composed of a glass epoxy core with copper plate on both surfaces. The expression “two-layer board” means that there are two conducting surfaces. The board manufacturer drills holes for vias and mounting components and etches the copper to form traces and component pads. The manufacturer then plates the holes with copper and solder plates holes and pads so that components can be mounted and soldered to the board. The traces on this board must connect the power supply and common to every active component, as well as interconnect all of the signals.

Today, integrated circuits perform many tasks. This requires a large number of pinouts on individual ICs. A large pin count means that many interconnecting traces are required. For many commercial products, four-layer boards are the preferred solution (four conducting surfaces). Even larger trace counts have made designers turn to smaller trace widths and closer spacings. In this approach, conducting planes and traces are intermixed on each layer. Any unused areas on a layer are “flooded” with copper connected to ground or power. Flooding is a good idea, as it provides a more robust board, reduces board warpage due to heating, and helps in the radiation of heat.

The number of transistors on one die can now exceed 1.8 million. To achieve this component density, the individual transistors have been made smaller. This has resulted in devices that switch in a shorter time. The present 90 nm technology will soon be even smaller making the rise times even shorter.

The issues of cross talk and radiation are closely related to rise and fall times and not clock rates. This one fact has a significant impact on circuit performance. To illustrate the problem, assume that a driver transitions in 100 ps. If energy to drive the connected logic traces is not immediately available, the voltage to the device will sag. This can have two effects. The logic can malfunction, and there can be increased radiation from the board. The problem increases when a number of signal transitions occur at the same clock time.

There are techniques that can be applied even to a two-layer circuit board that allows it to handle fast logic. A practical solution to this problem is discussed in Section 6.7.