Issue link: https://iconnect007.uberflip.com/i/1391285
48 DESIGN007 MAGAZINE I JULY 2021 ance. Sometimes we remove the area under a high-speed pad to reduce the impedance at the same place it increases due to the component package above the pad. Cherie Litson: Only two instances come to mind: in a flexible circuit where you need to bend the board, and when weight is an issue, such as with items going into space. When the ground ref plane is not solid, the grid should be such that it will still support a return path for the signals above and below it. Also, it should not create in issue for fabrication. It can cause acid traps on the outer layers. With inner lay- ers, this is not as much of an issue. Rick Hartley: ere are some unusual filtering techniques available (i.e., defected ground structures) that remove small areas of ground plane under some traces in RF circuits. But these techniques should only be used if com- pletely understood. In general, splitting or removing ground is considered a really bad idea. If you consider where the energy is located in a transmission line—in the E&H fields, in the space between the trace and plane—removing any portion of the ground under or near the trace will simply cause the fields to spread out from their nor- mal position in the PCB dielectric. When fields spread, it almost always leads to trouble. e engineers who write app notes and data Lee Ritchey: ere is no defendable reason to ever cut a ground plane. Sure, there are data- sheets that say to do this. ey are simply wrong. Eric Bogatin: e only case for this practice would be if you are dealing with uV analog sig- nals on a board that is also carrying 10A tran- sient signals. Otherwise, no. Heidi Barnes: Typically, one can get in a lot more trouble removing ground planes. Just consider what the field lines are doing when the ground reference is removed. e field lines will spread out and have a bigger chance of interfering with something else or radiating. Removing ground planes can reduce capaci- tance and in some rare cases provide isolation, but it is always best to use an EM simulator to make sure the performance benefit is under- stood. Carl Schattke: Sometimes we purposely want a large area for a return path. We may want to do this on the area below an antenna, for instance, because we want that energy to propagate outward and not have a direct return path. Another reason: if we want to change the ref- erence layer (with voided-out copper) to make a wider RF trace with a corresponding lower tolerance to manufacturing variations, giving us lower overall variation in the trace imped- Is there any instance in which a ground ref plane should not be solid? Certain datasheets used to recommend removing the ground ref plane in some situations. Q