Bluetooth designs
04/01/2001
How low-temperature co-fired ceramic substrate can maximize performance
BY DEAN MARSHALL
The advent of the Bluetooth wireless specification has enabled a new set of advanced wireless applications in computing, communications and consumer electronics.1 Initial implementations are now in production, built using low-temperature, co-fired ceramic (LTCC) technology that includes a variety of co-fired tape layers that create a multilayer module. This article reviews the high levels of integration possible with LTCC technology and why it is ideal for Bluetooth designs.
Bluetooth Specification Requirements
The Bluetooth wireless specification is one of the most innovative and exciting technologies now available for a new generation of small, high-performance mobile electronic products. Originally conceived and developed by Sweden's Ericsson Co. Ltd., Bluetooth is a global specification for short-range (approximately 10-100 meters) wireless voice and data transmission, developed to replace cables and enable wireless links between handheld devices and also between or among stationary electronic equipment. An open specification based on radio transmission in the 2.4 to 2.5 GHz industrial-scientific-medical (ISM) band, the Bluetooth standard is now gaining acceptance as suppliers introduce Bluetooth-enabled products into the marketplace.
The Bluetooth-based radio is being designed as a highly integrated subsystem module that uses a frequency hopping scheme to minimize signal interference in the cluttered ISM band. Its potential for consumer, computer, industrial and many other wireless applications is significant. Bluetooth is expected to revolutionize personal communications by allowing cellular phones, personal computers, personal digital assistants and other devices to exchange information with no need for any hardwired connections. Implementing Bluetooth technology is not trivial, however. A Bluetooth system must include a microwave radio with antenna, a modem, radio controller hardware, data processing capability and working memory. For the widest acceptance, Bluetooth solutions must not only be very low in cost, but also must integrate as many of the needed functions as possible into the fewest number of semiconductor chips. Integrated sections can include an RF synthesizer, low noise amplifier, 0-dBm power amplifier, voltage controlled oscillator (VCO) and limiter, as well as the data processing and radio controller hardware. However, depending on the semiconductor technology used, there remain some functions, which are best left "off chip," either because it would be impractical to implement or for electrical performance reasons.
External Functions
Functions that are sometimes difficult to integrate in semiconductors may include the antenna, various filters,impedance matching, VCO tuning, RF switching and, nearly always, power supply decoupling. These functions are usually implemented with various active and passive discrete components, such as resistors, capacitors, inductors and diodes connected to the ICs by trace routing, either on the main printed circuit board (PCB) or on the substrate at a module level. Filters are needed at several places in the radio design and some are particularly troublesome to integrate on semiconductors. Many designs use intermediate frequency (IF) filtering in the receiver, and because most use a relatively low IF, this filter is easily incorporated into the semiconductor chip. However, at the antenna, radio frequency (RF) filtering is needed to keep strong out-of-band signals from overloading the receiver input and at 2.4 GHz, this filter is not usually practical to integrate on semiconductors. For performance reasons, these RF filters are normally implemented externally. Very low frequency filters are also generally used as part of the frequency synthesizer. Unless these low frequency filters can be implemented with digital signal processing (DSP), they, too, are not easily integrated because of the relatively large element values needed.
 Figure 1. Cross section of a three-dimensional LTCC substrate. |
The large number of passives needed for filtering and impedance matching elements can quickly add up to a significant amount of board space. However, careful design and material selection may allow several components or whole functions to be incorporated into the underlying circuit board or substrate material. Realizing as many of them as possible in the substrate as lumped inductors and capacitors or as distributed elements using controlled impedance transmission lines can realize cost and size advantages.
Another benefit is that embedded elements can be located directly beneath sensitive surface mounted circuitry, yielding the shortest possible connection length for maximum performance. This can be especially useful for power supply bypass capacitors buried in the substrate beneath supply connections to surface mounted semiconductors. Depending on the chosen substrate technology, designers can decide which components are best embedded in the substrate material to optimize performance, size, cost and power consumption.
An Optimal Technology
An optimal technology for manufacturing transceiver modules that must embed key passive structures is LTCC substrates. Several features make it ideal for emerging high-frequency applications, such as Bluetooth. The LTCC substrate provides the ability to combine RF design structures in three dimensions with high-density interconnects and advanced assembly technology. LTCC materials are readily available, and the technology is mature, shown to provide high density and small size required for Bluetooth components. The process is similar to thick film hybrid construction, but less expensive because layers are processed in parallel rather than one at a time, and, once collated, all of the dielectric layers and conductors are fired simultaneously, rather than sequentially.
 Figure 2. Bluetooth RF modules and LTCC substrate arrays. |
There are alternatives to LTCC substrates, as listed in Table 1. For example, high-temperature co-fired ceramic (HTCC) materials are multilayer structures using dielectrics, such as alumina. HTCC substrates have been shown to be reliable in a broad range of aerospace and other applications, but the high firing temperature of around 1,500°C required for HTCC materials requires the use of refractory metals, such as tungsten or molybdenum-manganese, that are relatively poor conductors. In contrast, the LTCC firing temperature of 800 to 900°C allows the use of highly conductive silver and gold metallization. Also, shrinkage that occurs during firing is not as well controlled in HTCC as in LTCC technology, which translates to poorer dimensional accuracy and, thus, looser available component tolerances for embedded circuit elements.
LTCC materials also have advantages when compared with organic materials. The dielectric loss tangent of LTCC materials can be an order of magnitude better than traditional PCB materials, such as FR4. Although the relatively expensive, high-performance polytetrafluoroethylene (PTFE) materials can match the high frequency loss characteristics, the low dielectric constant means that embedded capacitors and transmission lines must be much larger. LTCC technology also provides layer thickness control better than most organic substrate materials for closer tolerance on transmission line impedance and embedded element values. Neither organic material can match the combined high frequency performance, size and cost of LTCC substrates.
 Table 1. Substrate comparison. |
A challenge sometimes faced is the increased power dissipation levels in high-frequency components. The transmitter in a Class 1 Bluetooth device must generate up to 100 milliwatts of output power, which can lead to a significant temperature rise in tiny modules. LTCC substrates can provide a thermal benefit. For example, the thermal conductivity of a LTCC is 2 to 4 W/m/°K, which compares with 0.3 W/m/°K for organics. By using thermal vias in LTCC (vias placed solely for heat conduction, not electrical interconnect), the effective thermal conductivity can reach 25 W/m/°K.
Another thermal characteristic of LTCC substrates is the relatively low Tf, which is the temperature coefficient of resonant frequency - a combination of the thermal coefficient of capacitance (TCC) and coefficient of thermal expansion (CTE). It is a measure of the thermal stability of frequency determining components. For LTCC materials, the Tf can be as low as 10 ppm/°C, which compares with values many times higher for most organic materials. Ceramic packaging in general offers benefits for a wide variety of environments. Its low CTE of about 6 to 7 ppm/°C provides a closer match to most surface-mounted components, especially flip-chip semiconductor devices, than other substrate materials.
LTCC Construction
An LTCC substrate is fabricated by printing metal conductive patterns on various pre-cast dielectric tape layers. Interconnect between layers is accomplished with vias that are punched and metal filled before stacking and lamination. Tape layers consist of a cast glass-ceramic composite between 50 and 250 microns thick while the conductive layer is typically a precious metal, such as gold or silver, although some LTCC systems can use copper.
In addition to providing for interconnects, the metallization patterns printed on the tape layers can be used to create capacitors, inductors, controlled impedance transmission lines and various other RF structures within the ceramic. It is also possible to embed resistors into the substrate by screen printing thick-film resistor formulations on tape layers before lamination.
Individual tape layers are then collated or stacked, laminated under high pressure, and fired to form a single homogeneous multilayer ceramic structure. The number of tape layers that can be assembled using this process is nearly limitless. For applications such as radio transceiver modules, active components can be attached to the outer layer of the module by soldering as pre-packaged devices, as flip chip die or with wirebonds. Additional passive components can be solder-attached, as well; Figure 1 illustrates a cross section of a three-dimensional LTCC substrate. LTCC modules have shown to be cost-effective and provide space savings in VCOs and high-frequency filter designs. The LTCC process enables a high level of integration by offering the ability to embed a wide range of capacitors, inductors and other RF structures for high-frequency filtering, impedance matching and power supply bypass. LTCC modules are typically fabricated in array format, and various array sizes are available. The designer can fabricate multiple design permutations on one prototype array. The final design can then be converted to a large production array. Substrate processing, as well as subsequent module assembly operations, benefit from multi-piece handling when high-volume manufacturing is needed.
An Example Module
The Bluetooth radio module shown in Figure 2 measures approximately 14 x 10 x 1.6 mm and uses an LTCC substrate with nine tape layers, some less than 50 microns in thickness. The module uses a backside land grid array with a 1.27 mm pitch for connection to the next level assembly. Interconnect routing is predominately 0.13 mm traces and spacing. Layer-to-layer vias are 100 microns throughout, and there are multiple internal ground planes used for shielding and to form controlled impedance striplines. Embedded within the LTCC substrate are four inductors, 16 capacitors and 10 transmission lines. Approximately 40 additional active and passive components are solder attached to the top surface, including a flip chip BiCMOS IC.
Conclusion
While other options are available, the use of low-temperature co-fired ceramic is measurably a good choice for Bluetooth radio modules. The complete radio can be integrated within the module, thus simplifying a designer's task of embedding a spread spectrum radio into an OEM product and reducing time-to-market. Using direct attach of bare ICs and embedding passives in the substrate achieves a significant level of integration and a reduction in size and cost compared with other technologies.
AP
Reference
- www.bluetooth.org
DEAN MARSHALL, senior project engineer, can be contacted at CTS Corp., 1201 Cumberland Avenue, W. Lafayette, IN 47906; 765-463-2565; Fax: 765-497-5399; E-mail: [email protected].