Civil Engineering Reference
In-Depth Information
4.5
CAPACITIVE COUPLING EFFECTS
4.5.1
Introduction
As feature sizes have shrunk with advancing technologies‚ wires have been
brought closer to each other. While the width of a wire has correspondingly
shrunk‚ wire heights have not scaled proportionally‚ as illustrated in Figure 4.11‚
since such scaling would increase wire resistances tremendously. The net effect
of this has been to increase the capacitive coupling between wires‚ and this
leads to two effects in terms of circuit performance:
Crosstalk noise
corresponds to noise bumps that are injected from an switch-
ing wire to an adjacent‚ nominally silent‚ wire. Several fast crosstalk noise
metrics have been proposed in [Dev97‚ ‚KS01‚CaPVS01‚ DBM03]‚
and a detailed description of these is beyond the scope of this topic.
Delay changes
occur‚ as compared to the uncoupled case‚ because of the
additional coupling capacitances in the system.
The terminology that is frequently used in the context of crosstalk analysis is
to label the wire being analyzed as the
victim‚
and consider any wires that
capacitively couple to it as
aggressors.
In a full-chip analysis scenario that consists of a prohibitive number of ag-
gressor/victim scenarios‚ it is vital to reduce the cases to be considered to a
manageable number. The effect of an aggressor on a victim depends on a num-
ber of factors‚ and not every aggressor will inject an appreciable amount of
noise into a victim. Pruning filters that identify unimportant coupling cases
have been proposed in
To illustrate the mechanisms that affect crosstalk‚ consider an example of
two coupled lines of equal length. While this is a very simple case‚ the concepts
illustrated here can be extended to understand crosstalk in the multiple line
case. The equivalent segmented RC line with a Thevenin driver model is shown
