Issue link: https://iconnect007.uberflip.com/i/1457913
44 DESIGN007 MAGAZINE I MARCH 2022 center to center, over an FR-4 substrate, 9.5 mils thick, with a modeled Er of 4.4. When a Sonnet EM simulation 3 is run for the structure in Figure 2, and the coupling is mea- sured from port 1 to port 4, as shown in Fig- ure 3, you can see that there is nothing strange going on in the frequency range of 0-4 GHz. is is our baseline coupling that we will com- pare to later on. Adding a copper pour to Figure 2 results in the new structure in Figure 4. e copper pour has a realistic spacing of 10 mils to the aggres- sor and victim traces and therefore has a mini- mal effect on the 50-ohm impedance of these traces. Simulating the coupling of the traces and cop- per pour shown in Figure 4 results in the new coupling plot of Figure 5. e first response to seeing this is sure to be, "It's true. e iso- lation is worse with a copper pour." Perhaps, and perhaps not. It's all in understanding what is going on, because poorly placed copper will ruin any design, at any frequency from DC to light, as we shall see. Antenna Basics To understand what is going on with Figures 4, 5, and 6, we need to understand something about resonances and how antennas work. You may well have heard of a "half-wavelength" antenna. is is an antenna that is one-half of a wavelength long which has good efficiency as a radiator. ere are also quarter-wavelength antennas, simply one-half of a half-wavelength antenna placed perpendicular to a perfectly conductive ground. Anything shorter will not radiate well at all without added matching. If you put a trace or any copper structure Figure 4: Adding a copper ground pour with a 10-mil gap to the aggressor and victim traces of Figure 2 results in this new structure. Figure 3: Simulating port 1 to port 4 coupling of the structure shown in Figure 2 shows nothing strange in the frequency range of 0-4 GHz. Figure 5: The resulting coupling measurement from ports 1 to 4 of the structure of Figure 3. This simulation does indeed show something strange going on at a frequency of around 3.2 GHz.