Today, low-temperature co-fired ceramic (LTCC), multilayer ceramic (MLC) and ceramic mono-block technologies are the dominant choices for the implementation of surface mount components for RF passive filters, diplexers, baluns and front end modules. LTCC is the most popular ceramic technology since it uses miniature lumped components, which can be optimized for operation over a wide band of frequencies, whereas mono-block and MLC components use different materials for different frequencies and limit the integration of devices for multiband applications. LTCC, with its ability to integrate in excess of 20 layers, has become a platform for the integration of modules in multiband applications that combine several lumped element filters, baluns and diplexers for cellular and WLAN applications.
It is typical for such modules, as well as the components, to consist of 15 or more metal layers with microvias connecting layers. The need for many layers typically translates into additional design time, higher tooling costs and inadequate model to hardware correlations. Moreover, LTCC has lower performance due to process tolerances (15 percent component tolerance) and higher dielectric losses (Dissipation Factor, Df =0.005-0.007 at 1 GHz) compared to MLC and their mono-block counterparts (2 to 5 percent component tolerance and Df = 0.0005).
With the advent of a new class of thin, low loss, organic dielectrics, large area fabrication techniques and unique lumped element design topologies, high Q, low loss RF components have been realized. These new materials can be made as thin as 5um, with a dielectric constant (Dk) of 7.6 and a Df of 0.002 at 10 GHz. These materials are stable over frequency and have very little moisture uptake, typically < 0.04 percent. In the past, organics were typically shunned for such applications due to their variability with changes in temperature and humidity (typical FR4/5 materials). However, this is no longer the case, as these materials can attain Moisture Sensitivity Levels to MSL 1 at 260oC.
An MLO component consists of one or more RF dielectric layers embedded between layers of other laminates to provide routing, shielding and bonding pads for SMT placement. The MLO technology can also be used as an RF or mixed signal substrate, in which the layers support placement of both RF and digital ICs. Figure 1 illustrates a typical MLO cross-section. Variations of this stack-up may be used if, for example, thinner components are required. The dielectric layer must have low loss at the common wireless frequency ranges and, at the same time, have a high Dk to provide high capacitance density. Unfortunately, these are counteracting properties for most materials. High Dk is usually obtained by filling polymers with dielectric materials, which can increase loss.
For this reason, very thin dielectric layers are necessary. Recently, a number of low-loss formulations of polytetrafluoroethylene polymers (PTFE) and liquid crystalline polymer (LCP) have become available in copper clad films as thin as 8 microns. For PTFE, advanced filler materials are used to increase Dk without significantly increasing loss. LCP has low Dk and has been shown to have very favorable characteristics for a wide variety of RF and high-speed applications. By carefully selecting laminates and bond ply materials, it is possible to utilize standard lamination processes to produce high performance RF components.
Once the design is completed, metal and dielectric layers are etched and laminated. A typical six-layer construction is shown in Figure 2. These customized stack-ups are patent protected and allow for the highest passive component density while achieving increased functionality. To achieve the desired component densities, line widths and spaces as small as 15 microns can be obtained using semi-additive process techniques. The introduction of laser direct imaging (LDI) technology is paramount, as it allows for the realization of fine line geometries and tight tolerance structures. These structures typically form the resonant structures of the embedded components and must be reproducible with high precision over a large area. A distinct advantage of MLO over ceramic and Si based processes is cost. By leveraging the economies of scale, over 0.2 million 0402 devices can be constructed on a single 18” x 24” panel.
Diplexers are an essential component in today’s multiband systems and perform a multitude of functions. They isolate, transmit, and receive separate bands at different carrier frequencies.
An example is a product used in the IF stages of satellite TV systems. The performance required was as follows: channel 1 pass band of 900 MHz to 14450 MHz with insertion loss <3dB and stop band rejection > 40 dB from 1650 MHz to 2150 MHz; channel 2 pass band of 1650 MHz to 2100 MHz respectively with insertion loss <3dB and stop band rejection >40 dB from 900 MHz to 1450 MHz. This type of diplexer is typically implemented using a ceramic monoblock device or discrete inductors and capacitors. However, in order to reduce cost, it has been designed in MLO using lumped elements.
The resultant size for the finished MLO component was approximately 20 mm x 5 mm x 2 mm versus 35 mm x 12 mm x 5 mm for a ceramic monoblock. Up to 26 components were integrated into the MLO device, which resulted in a 15 percent reduction in board area. All electrical specs were met as shown in Figure 3.
A second example is the 0603 WLAN diplexer, designed for use in mobility and computing applications. When the number of bands increases in these devices, so too will the requirement for higher Q and lower loss RF passive components. With phones that support multiple wireless standards, such as Bluetooth, GSM, WLAN, GPS and UMTS, multiple antennas are required. Antennas add cost and provide interoperability challenges, which can impact performance. To minimize the number of antennas, designers will typically incorporate very low loss RF diplexers into their designs. Low loss diplexers provide high isolation between bands and reduce the number of antennas, enhancing performance while decreasing size and cost.
Diplexers are initially synthesized using standard RF design tools and are then optimized to exceed specification using an in-house custom design methodology. The physical layout is virtually constructed using EM simulation software. This basic design can span case sizes and frequency bands by simply linearly scaling the internal components and/or reusing existing element values. Figure 4 shows measured results from a MLO dual band WLAN diplexer where all specifications were achieved in a 0603 case size.
In sum, MLO components offer many advantages over ceramic and Si processes, including lower cost to manufacture, higher product yields, matched CTE to FR4 and enhanced performance. Low variability provides repeatable results and faster time to market. Additionally, process and design methodologies are scalable and can be used to implement SMD components, as well as to realize custom RF and mixed signal modules.