Address: P. O. Box 1457, Los Gatos,
Figure 1. A path crossing a break in a ground plane
EMC and signal integrity engineers know that a signal that crosses over
a break in a ground plane often causes reflections as well as immunity and
EMC problems. An additional problem that often results is excessive crosstalk
between paths crossing the break. Figure 1 shows part of a circuit
board with a single path (of approximately 50 Ohms impedance) that crosses
a 5 cm break in the ground plane. The path is about 10 cm in length. The
board's solid ground plane extends downward from the bottom of the figure
for about another 10 cm. A 300 mV signal with a risetime of about 300 ps
was launched on the signal path. The waveform across the 47 Ohm resistor,
nodes 1 and 2, is shown in Figure 2 using 100 mV/div and 5 ns/div scale factors.
The voltage across the ground plane break at nodes 3 and 4 is shown in Figure 3. Both waveforms were taken with a Fischer Custom Communications BCP-1 differential probe. Notice
the that the peak voltage in Figure 3 is almost 100 mV. That voltage represents
the voltage drop caused by the signal's return current flowing in the ground
plane through the impedance (read that "inductance") of the path around the
end of the break in the ground plane. As bad as this is for radiated emissions
imagine what would happen if a second path crossed the break 1 cm above the
signal path, about halfway between the existing signal path and the top
edge of the board above nodes 3 and 4.
|Figure 2. Signal Voltage (V1-V2)
|Figure 3. Slit Voltage (V3-V4)
Normally, there would not be significant crosstalk between two paths spaced 1 cm apart over
a ground plane. However, in this case, both paths must
share the impedance around the end of the ground plane break in their return
paths and thus the voltage measured across the break in Figure 3 will be
coupled into the second, victim, path. For the common case of a low impedance
gate output driving a high impedance gate input in the victim path, the ~80
mV peak voltage of Figure 3 will show up as crosstalk between the signal
path of Figure 1 and our hypothetical victim path. In this case, the peak
crosstalk is nearly 1/3 of the original signal amplitude, not a value that
can be ignored.
As a general rule, any paths crossing a break in a ground plane (actually in all the ground and power planes of a board) will likely be strongly coupled resulting in significant crosstalk. The plane breaks I have seen in some equipment are much longer that the example of Figure 1. For those cases, even signals with much slower risetime could be strongly coupled. For cases of a broken ground plane over a solid power plane, or vice versa, there may or may not be a problem depending on several factors including plane spacing. Such a case is beyond the scope of this short Technical Tidbit.
The waveforms in this article were taken with an Agilent
Infinium 54845a oscilloscope.
Top of page