In order to verify the superior level of effectiveness, radiation measurements were conducted on a test vehicle using the RF3178 TxM (Figure 1). The measurements clearly show a substantial gap in the performance of the two shielding technologies, with MicroShield clearly outperforming the embed- ded shield technology. On aver- age, the MicroShield integrated RF shielding technology is 15 dB better in attenuating radiation than the embedded technology. However, achieving these results as a TxM designer does not come without issues to resolve. From a TxM design perspective, adding a shield poses several problems to the designer. Primarily, the close proximity between the shield and electromagnetically emanating circuits alters the frequency response compared to an open-faced (unmolded), perfectly tuned, TxM—changing the performance. These effects are observed especially at higher frequencies. As a result, modeling and EM simulations are paramount to ensure good results when adding a shield. is greater than “b,” the dominant mode is TE10. Hence, equation 1 is rewritten to: c f c 1 2 0 = 2⋅a ⋅ ⋅ r r Where “c” is the speed of light, “ r ” represents the relative per- mittivity, “r” the relative perme- ability and “a” is the opening. Equation 2 shows us that the cut-off frequency, as expected, in- creases with a decreasing opening dimension, a. When several holes are present in the shield, the formula becomes even more complex, fur- ther emphasizing the importance of not having any openings at all. Figure 3. A 3D simulated E-field distribition. Implementing the shield There continues to be the use of metal shield cans applied external to TxMs and RF sections of the handset, however, recently there has been a pronounced trend toward embedded shielding within the TxM. As such, several methods have been developed to implement shield- ing for TxMs. One such example uses an embedded shield made of a simple metal can, however, this method requires that multiple holes be present in the can to allow the mold compound to ow easily throughout the module; a necessity in module assembly. But recall that according to the waveguide theory presented earlier, the shield effectiveness diminishes not only with the size but also by the number of holes in the shield. Since 3-D EM simulations can be time-consuming, depending on the complexity of the circuit and the number of tetrahedron elements needed to provide adequate accuracy, it is worthwhile to start with a less complex circuit and identify its key areas of importance. For ex- ample, from eld theory it is noticeable that closer proximities between eld-carrying signal lines lead to greater coupling. Those signal lines carrying time-varying charges, which are already embedded in the substrate and encapsulated by metal such as a ground plane, exhibit virtually no additional disturbance in the eld lines when an external shield is applied. Only signal lines, components or wire bonds, face signicant changes in their respective eld lines because these elements are subject to air or overmold as boundary condition. Alternatively, RFMD has developed a patent-pending MicroShield integrated RF shielding technology. This integrated shield is produced by applying a thin metal coating to the outside of a packaged semicon- ductor’s mold compound as a nal step in the assembly process. The result is a shield with negligible impact to the height of the module and repeatable in production in the reduction of EMI and RFI emissions.
Figure 2 is an illustration of the impact on the output match of the power amplier portion of a TxM with overmold but without a shield and then with the shield added on top of the overmold. The two-port simulation was done using HFSS, a 3-D EM software tool from Ansoft.
Although, the output match only represents a small portion of the passive circuits in a full TxM, the usefulness in determining coupling
Shield evaluation board
Six layer board. Edge plated. No copper areas are patched
VBAT, VDD and reset components under separate shield can
VBAT SMA connector
GSM850/900 RF input SMA connector
Input baluns under separate shield can
Antenna SMA connector
Figure 4. The Smith chart shows the input impedance S11 represented at the collector of the GaAs die. At the fundamental frequency there is virtu- ally no change in the impedance but at higher frequencies, such as 8 GHz or higher, the resonance changes character and the harmonic content will be different whether the shield is present or not.
GSM1800/1900 RF input SMA connector
SDI and Vramp SMA connector
All SMA connectors are surface mounted and fully enclosed
RX 50 Ohm termination, AM and Vramp components under separate can
Figure 5. Test board designed for radiated measurements.