Ground Bounce

This term has appeared in the literature to describe voltage drops on the ground plane on printed circuit boards.  The problem is most severe when the logic has very short rise and fall times.  To appreciate the problem it is necessary to review some simple principles.

    Voltage is defined as the work required to move a unit charge between two points.  The points can be in space or between points on a circuit.  The forces involved in doing work are between a unit charge and the electromagnetic fields present.  A voltage measurement is basically a field measurement.  In the case of a circuit trace carrying a signal the fields involved are confined to the space between the ground plane and the trace..  It is essentially impossible to get into the tight space under a trace to make a useful voltage measurement. 

    For measurements made between ground points at the ends of a trace run, the probe will couple to fields not related to the trace run.  To call this voltage a ground bounce would be misleading.

    The story really takes place under the trace.  The current path in the ground plane is confined to the area directly under the trace.  This is where the fields are confined.  Because of skin effect this current does not penetrate very far into the ground conductor.  Electrically the path is a narrow thin conductor imbedded in the ground plane.  The resistance of this path causes a voltage drop and this is the ground bounce. 

    The voltage drop in the ground plane along the trace reduces the logic voltage received at the IC.  This tapered voltage represents a field that is reflected and transmitted at all the connections to the IC.  The affect on the logic is complex.  The approach taken by most designers is to allow part of the total error budget to this voltage drop.  It should be noticed that for any wave action there is a corresponding voltage drop in the trace.  They are about equal.

    This discussion brings up a very basic philosophical issue.  Ground bounce is really a field problem.  The analysis we use is related to circuit theory.  How do we combine these two disciplines to get an understanding of how to proceed?