Discussion: Many heat sinks
have tines that protrude perpendicularly from a base plate such as the
one shown in Figure 1. Today's clock frequencies are so high as to be
able to excite the tines as quarter wave monopoles resulting in
something like an antenna farm on top of the IC, potentially radiating
a clock harmonic or other source resulting in an emissions problem.
One way to measure the resonant frequency of physical structures in general and heat sink tines in particular is shown in the
June 2006 Technical Tidbit, Measuring Structural Resonances
on this site. Basically, one would hold a very small shielded loop at
the base of one of the heat sink tines to couple energy into the tine.
By measuring the reflected signal from the small shielded loop, the
resonant frequency of a tine can be measured as the frequency where the
reflected signal experiences a dip as the tine absorbs energy from the
loop. The lowest frequency this would occur at is the frequency at
which the tine is 1/4 wavelength. The method is described in detail in
the June 2006 article referenced above. Figure 2 shows the method being
applied to a PCB test board mounted over a copper clad board. The
object in that case is to measure the resonant frequency of the LC
tuned circuit composed of the capacitor formed by the PCB and copper
clad board, and the inductance of the short wire between them.
Figure 2. Use of a Small Shielded Magnetic Loop to Measure PCB-Chassis Resonance
I observed a case some years ago where 4.7 GHz was leaking out of every
slot and seam in the enclosure of a product. It turned out that 4.7 GHz
was the third harmonic of the processor clock. When I measured the
resonant frequency of the hea tsink tines, the dip in the reflection
from the loop was right at 4.7 GHz indicating the tines were 1/4
wavelength monopoles, very efficient radiators, at that frequency. If
the tines had been longer or shorter the emissions problem would have
been reduced significantly.