Resonance conditions in the system can compromise an otherwise good design and can
result in very efficient radiators. All the features of the entire final structure are candidates for
resonance, and the resonance conditions are not typically predictable. This is one of the
reasons that the possibility of changes late in the design cycle is always a threat.
Unanticipated resonators could be heat sinks, chassis structures, resonant apertures, cables,
traces, patches, or a wide variety of other structures. As a general rule, to avoid resonance,
structures in the system should not be left floating. For instance, heat sinks should be
grounded. The author knows of one incident where a system clock chip was located
underneath a heat sink for a CPU. In this case the clock chip was coupling energy upward to
the heat sink and causing a resonant condition. Increasing the distance between the heat
sink and clock chip solved this problem. Although resonances are generally not predictable
(i.e., the resonance of the heat sink did not seem to correlate to any particular dimension of
the heat sink), it is very helpful to keep in mind that small structure changes can make big
differences. A knack for applying trial and error is very helpful in this regard.
One predictable resonance is that of a resonant slot aperture. At a slot length of /2, the slot
is resonant and can be considered a perfectly tuned dipole. This antenna will exhibit no
shielding properties. This, along with the reasons discussed previously, is a good motive for
keeping slot sizes small.
Chassis resonances often dominate high frequency emissions. If resonances dominate
emissions, the importance of making the resonances lossy, which will decrease the effective
Q, can be seen. Often, this can be accomplished by placing lossy material inside the
enclosure. When experimenting with such materials, it should be realized that a populated
PCB board is itself a lossy component. Experiments, for example, of chassis resonances in
the absence of a populated board will yield results that may not be as significant when a
populated board is present in the system. Subsequently, when experimenting with lossy
materials in the chassis, make sure that the fully populated board is included.
Resonance conditions in the system can compromise an otherwise good design and can result in very efficient radiators. Although resonances are generally not predictable (i.e., the resonance of the heat sink did not seem to correlate to any particular dimension of the heat sink), it is very helpful to keep in mind that small structure changes can make big differences.