 
|
  
   |
| About
CE-Mag Free Subscriptions Current Issue Article Archives ESD Help Mr. Static Web Gallery Staff Info Contact us
|
|
|
|
|
|
Distributed-Element Multipole EMI Filters: Material and Design ConsiderationsClyde M. Callewaert, Joseph V. Mantese, Adolph L. Micheli, Norman Schubring, Dominic A. Messuri, William T. Phillips, and Steve A. MusickA novel technology uses selective metallization of ferroelectric-ferrite composites to create a new class of passive components integrating inductive and capacitive elements.As a result of persistent demands for greater functionality, faster operating speed, and reduced size, many electronic devices have evolved into collections of highly integrated systems. Commonly used devices such as digital cellular telephones and wireless local-area networks are complex mixed-signal systems combining high-frequency analog and digital circuitry in a single instrument. At the same time their complexity is growing, these systems are becoming more compact because of the use of VLSI circuitry and surface-mount device (SMD) technologies on highly populated PCBs. Implementation of these advanced systems is hampered, however, by the increased electromagnetic interference (EMI) that accompanies component proximity and high-frequency mixed-signal circuitry.
An automobile's electronics, for example, may be regarded as a collection of subsystems interconnected by distributed wiring. Each subsystem—the engine control module, the antilock braking system—is in effect a highly integrated mixed-signal device. EMI generated by each subsystem can propagate to other subsystems via radiation or by transmission along the vehicle wiring harness, or by a combination of both mechanisms. Therefore, to suppress the reception and transmission of each system's EMI requires some type of shielding or filter arrangement. EMI filter circuits vary in complexity from simple ferrite beads to multipolar filter networks composed of capacitors and inductors. The choice of filter is always a trade-off among considerations of performance, cost, and size. Although shielding a device with a conductive enclosure does provide good protection from EMI, the power and ground wiring to the device still provides a path for interference. EMI filter circuits therefore are necessary to supplement the use of conductive enclosures. As electronic devices become smaller and more integrated, a paradoxical situation arises: devices require ever more EMI filter performance in ever less space. Newer compact filters reduce PCB space consumption by molding discrete capacitors, inductors, and ferrites together into quasidiscrete filter assemblies. Alternatively, some EMI filters have been removed from the PCB and designed entirely into connector bodies such as filtered header connectors employing ferrite blocks and SMD capacitors. Because all passive filters are made up of discrete inductive and capacitive elements (either designed or parasitical), true integration of both properties in one element would require a single material having both inductance and capacitance. Such a concept motivated the development of ferroelectric-ferrite composite materials.1–3 The materials make possible the realization of capacitive and inductive distributed multipole elements whose performance can be controlled by orchestration of metallization patterns and geometry. This approach can, however, require complex patterning or critical geometrical constraints that may be expensive or impractical for mass production. Another method of tailoring filter performance is available, involving adjusting the inductive and capacitive properties of ferroelectric-ferrite composite materials through compositional variation. This alternative promises the possibility of a more manufacturable device, and thus has direct application for existing packages. Although compositional variation as a means of customizing filters is an appealing approach, it requires extensive knowledge of the complex permittivity and permeability characteristics of the composite materials for a range of frequencies and compositions. This article describes the material properties of ferroelectric-ferrite composites and discusses how the materials can be used in the fabrication of EMI/RFI filter elements. The performance of a standard multilayer feedthrough SMD capacitor, formed from a ferroelectric-ferrite composite containing 50% barium titanate by volume, is also considered. Ferroelectric-Ferrite Composite Materials The new composites are two-component systems consisting of a capacitive ferroelectric component (polycrystalline barium titanate) and an inductive ferrite component (polycrystalline copper-nickel-zinc ferrite). The barium titanate and ferrite were variously combined to produce materials with different volumetric proportions of barium titanate, the volumetric compositions, excluding porosity, being accurate to better than 1%. Sintering temperatures were chosen for their ability to produce maximum ceramic density without apparent chemical reaction, as determined by x-ray-diffraction analysis. Optimum sintering temperature for maximum density was that temperature at which no phases other than barium titanate and ferrite were observed in the x-ray analysis.
The electrical properties of the composites depend upon the compositional ferroelectric volume fill fraction, f, and excitation frequency,
Effective media predictions for
Here, Equation 1 thus can be useful for designing the properties of new composite media once the properties of the starting constituents have been determined. The existence of such a design equation obviates traditional cut-and-try methods of producing new materials with enhanced electrical properties. The electrical properties as a function of frequency for a range of barium titanate–ferrite composites are shown in the contour plots of Figure 1. Both the complex permittivity and the complex permeability can be seen to vary smoothly and monitonically with composition (at fixed frequency) from their values for the pure ferrite to their values for pure barium titanate. Thus, a wide range of composites can be formed that concurrently possess relatively large permittivities and permeabilities—interesting materials for a variety of practical devices. Filter Design Considerations Construction of feedthrough SMD packages usually involves alternating active (through) and ground layers, forming multiple parallel capacitors similar to the construction shown in Figure 2. The overall package footprint is that of a standard 1206 package: 120 x 60 mil, with 0.67-mil plate spacings. Unlike a standard two-terminal multilayer SMD capacitor, a feedthrough device has four terminals: two active connections and dual ground terminals. In the 1206 geometry of an experimental SMD EMI filter package, the device had a total capacitance value of approximately 2110 pF when formed from a 50-50 barium titanate–ferrite composite.
Figure 2. A feedthrough SMD filter package. (top) Top view of the metallization patterns for the active (electrode A) and ground (electrode B) layers. (bottom) Cross section. Scattering parameter measurements of several fabricated devices were taken over a frequency range of 30 KHz to 1 GHz, using an HP8753D vector network analyzer(Hewlett-Packard) and a three-terminal 1206 SMD test fixture designed by Intercontinental Microwave Inc. (Santa Clara, CA). A full two-port calibration was performed prior to measurement; fixture calibration standards were employed to de-embed the fixture. The devices exhibited at least 30 dB of insertion loss between 100 MHz and 1 GHz (see Figure 3).
Figure 3. Measured insertion loss of several SMD filters formed from a 50-50 barium titanate–ferrite composite. Several observations can be made upon review of the insertion loss of the SMD depicted in Figure 2. At low frequencies the composite has very low dielectric and magnetic losses but also possesses high permittivity and permeability. These characteristics would yield an equivalent filter circuit having ideal inductive and capacitive elements that provide classic roll-off characteristics. At higher frequencies the composite becomes very lossy and nonmagnetic, rendering an equivalent circuit dominated by resistive elements. These characteristics give rise to a nearly constant insertion loss. Therefore, the material characteristics of the composite allow the equivalent circuit of the filter to evolve with increasing frequency from ideal elements to a nearly resistive network. Because the filter was fabricated with standard SMD feedthrough packaging as used by a host of manufacturers, it does not fully exploit the novel electrical characteristics of the ferroelectric-ferrite material. The parallel active (through) layers of the device carry equal signal current and thus cause the magnetic fluxes to oppose one another in the space between the layers, as shown in Figure 4.
Figure 4. Vector field representation of magnetic flux density, B, between active layers. Field cancellation between active layers limits excitation of the magnetic-loss properties of the ferroelectric-ferrite material. The reduced field, therefore, cannot take full advantage of the magnetic characteristics of the composite. A simple solution would be to connect as a through only one active layer, thereby allowing the magnetic flux to permeate the entire volume of the device. This modification would increase the series inductance and high-frequency magnetic losses, improving insertion loss. In addition, the effective series resistance (ESR) of these devices usually comprises dielectric and metallization losses. However, displacement currents between the active and ground layers create additional magnetic losses through the composite. Increasing the number of parallel capacitor regions can reduce ESR, further improving the insertion loss of the device. It must also be remembered that a metal plate acts as a shorted turn that sets up opposing eddy currents when a magnetic flux impinges normal to the surface metallization. Thus, optimum magnetic performance is achieved when the electric and magnetic fields are normal. Finally, to achieve maximum EMI/RFI attenuation, the SMD must concentrate the energy density into the entire SMD package, with contributions to the energy density coming from both the electric and magnetic fields. To make sure that happens, the electric and magnetic fields of the SMD must cohabit as much of the same volume of the SMD as is feasible, while yet maintaining manufacturability. A number of overlapping factors must be kept in mind when designing SMDs from these new composite materials.
Figure 5 diagrams the layout of an optimized SMD filter device. The coil that forms the inductive element is shown enclosing the entire package. The capacitive plates are contained within the coil structure; therefore, the electric and magnetic fields are made to occupy the same volume of material, producing maximum energy density within the SMD. In addition, the fields are normal to each other, thereby preventing eddy currents from being established in the metallization that would prevent penetration of the magnetic field within the structure. The multilayer structure of the device minimizes inductive impedances to ground. By design, the EMI filter is a true distributed-element multipolar structure.
Figure 5. Schematic of a 1206 SMD based on the design rules described in the text. Figure 6 shows the modeled performance of a 1206 SMD similar to the device in Figure 5 but tailored to meet manufacturing requirements. It incorporates all the design rules described above. This SMD is expected to yield nearly 40 dB more insertion loss than the simple structure described in Figure 3, in the same package. Although this filter device has not been fabricated, the models have been extensively tested and validated for a variety of structures and provide a high degree of confidence.
Figure 6. Insertion loss as a function of frequency, modeled for a 1206 SMD. Conclusion Optimal use of ferroelectric-ferrite composites, with prudent metallization and packaging design, creates a new class of passive components that are both inductive and capacitive. Because both inductance and capacitance are capabilities of the same material, many other ladder networks are possible, including, to name a few, high-pass, low-pass, band-pass, and band-reject filters. References 1. JV Mantese et al., "Applicability of Effective Medium Theory to Ferroelectric/Ferrite Composite with Composition and Frequency-Dependent Complex Permittivities and Permeabilities," Journal of Applied Physics 79, no. 3 (1996): 1655–1660. 2. GW Milton, DJ Eyre, and JV Mantese, "Finite Frequency Range Kramers Kronig Relations: Bounds on the Dispersion," Physical Review Letters 79 (1997): 3062. 3. T Yamamoto et al., "Evaluation of Ferroelectric/Ferromagnetic Composite by Micro-composite Designing," Ferroelectrics 95 (1989): 175. Clyde M. Callewaert, Joseph V. Mantese, Adolph L. Micheli, and Norman Schubring are with Delphi Research Laboratories (Warren, MI). Dominic A. Messuri, William T. Phillips, and Steve A. Musick are with Delphi Packard Electric Systems (Warren, OH). For further information, contact Mantese at joe.mantese@delphiauto.com. |
Ferroelectric-ferrite composite materials provide both inductance and capacitance.
. A volume fill fraction of zero indicates purely ferrite material, one of 1.0 indicates pure barium titanate, and 0.5 indicates equal volumes of barium titanate and ferrite. For notational convenience, let the complex effective permittivity and permeability of a composite (normalized by their vacuum values) be represented in functional form by 
*eff(f,
*1 or 
