Clearing the Path to 3G Handset RF Integration

By: Kent Heath, Freescale Semiconductor
( 1 Jul 2007 )

Time has always been a scarce commodity for mobile phone designers, and integration of the phone's RF section has consumed more than its share of precious development resources. As the complexity of cellular handsets has grown, the burden of RF IC integration has continued to increase. However, as wireless handset OEMs work to incorporate EDGE and 3G support into their products, RF component and subsystem vendors can no longer place the burden of this task on the OEM. Rather, the burden is now on RF vendors to not only reduce the component count and cost of their RF subsystems, but to provide the most trouble-free, robust RF front-end subsystems that improve this time to market with yields comparable to last-generation 2G subsystems.

Initial BOM cost and size reduction has traditionally been the principal focus. Typical cellular RF roadmaps reflect this focus as parts shrink and become less expensive over time. For 3G RF subsystems, there has been a migration from multi-chip solutions requiring a number of discrete RF ICs and front end components, typically with two parallel RF paths (one for GSM/GPRS/EDGE signals and a separate path for the 3G UMTS signals) to more recent lineups with significantly higher levels of integration (see Figure 1).

The resulting board-area parts count reductions, vendor consolidation, and supply chain consolidation efficiencies shown in Figure 1 create significant cost reductions for handset OEMs. However, this is only part of the total cost of ownership (TCO) challenge faced by handset manufacturers. They face huge development costs and costs related to manufacturing quality caused by the incredible complexity of state-of-the-art 3G feature phones. Addressing these additional elements of cell phone TCO requires a new approach to RF subsystem design.

The development costs associated with RF subsystem integration include all of the engineering staff costs to insert the RF ICs, power amplifiers, filter elements, low-noise amplifiers, and switches needed to support each mode (GSM/EDGE/UMTS) and frequency band for each phone model. The costs also include creating or modifying the phone's layer one software to drive all of these components, and matching the impedance levels between each component for optimal radio performance to ensure conformance with 3GPP specifications and carrier requirements. This must be done for all bands, modes and power levels of phone operation. It is a highly iterative software/hardware optimization effort, and much of this work is done for the handset designers by the chipset vendors that typically provide reference designs as a starting point for handset development. However, due to the pressure on each handset manufacturer to differentiate its products, customization is inevitable. With traditional RF subsystem architectures, this customization requires many months of effort and tens of highly-skilled RF engineers for the development and testing of the completed phone.

The complexity of multi-mode, multi-band handsets provides many manufacturing challenges to the handset OEMs and their partners that build and test each cell phone before shipment to the consumer. After assembling the cellular phone PC board, and installing all needed components, each phone must be programmed with software that personalizes the phone for operation in the regions of the world where it will be used. It must also embody the unique "look and feel" specific to the marketing channel (network operator, retailer, or virtual network operator) through which the phone will be sold. The phone must then be "calibrated" via software and all bands, modes, and power-levels must be checked sufficiently to ensure that each phone will work properly while meeting the battery life expectations of the consumer, while not generating undesired RF signals that could interfere with other calls or carrier network services. The costs associated with the manufacturing process include the depreciation expenses of assembly time (directly related to the number of components placed on the PC board), software download/calibration time, test time, and the resultant manufacturing yield and debug and rework cost to fix failing handsets (see Figure 2).

With the ever-growing number of bands and modes that are being adopted in mid-level and higher-tier cellular phones, development and manufacturing costs are reaching unsupportable levels with traditional RF approaches. Addressing these components of TCO that are not directly impacted by Moore's law is the new challenge for RF subsystem vendors. In order to meet this challenge, new, smarter RF subsystem architectures are required.

A "SMARTER" APPROACH
Freescale Semiconductor is tackling this challenge with a smart RF approach the company has labeled "Extreme RF or RFX." With the RFX approach, the company is fundamentally changing the economics of RF subsystem integration and manufacturing by embedding RF intelligence into the subsystem. The result is a dramatic reduction in "time to first call," minimizing phone calibration and test times, and increasing yields despite the high level of complexity in new 3G and EDGE phones. An example of the RFX approach is the Freescale RFX275-20 EDGE RF subsystem (see Figure 3).

As shown in the figure, there are two ICs in the RFX subsystem: the MMM6000¡ªa highly integrated GSM/GPRS/EDGE transceiver; and the MMM6029¡ªa quad-band GSM/GPRS/EDGE power amplifier with switch and power control. Beyond an obvious high level of integration, this solution minimizes BOM cost and size. The RFX275-20 subsystem employs a DigRF interface for fast, easy connection to a number of readily available baseband solutions, and two unique elements: a PolarPlus small-signal polar transmitter specifically designed to maximize handset yields (as compared to alternative full polar architectures), while providing long battery life and an RFX control subsystem.

The RFX control system includes a programming method that significantly reduces development cycle time required for the design of new phones. Based on customer feedback, the time to first call is reduced as much as 66 percent, with significantly fewer resources required for layer one programming optimization.

Traditional RF subsystems require engineers to consider critical timing between the transceiver, power amplifiers, switches, low-noise amplifiers, baseband processor and voltage regulators. The RFX controller uses a single-command programming model that reduces calibration steps and practically guarantees system compliance and high cellular phone yield.

In summary, as the complexity of cellular phones increases, the manufacturing and development costs associated with traditional RF subsystem integration rival the actual IC costs paid by the handset manufacturers. Because of this trend, the RF component supplier must take on a greater share of the total cost of ownership currently experienced by its OEM customers.

Creative solutions, like Freescale Semiconductor's RFX, must be applied to ensure that cellular phones remain economical as they support applications enabled by higher bandwidth, including music and video download, mobile TV and Internet-based services.

About the Author
Kent Heath is Director of cellular operations for Freescale Semiconductor's Radio Products Division.
Click here for Illustrations:

Figure 1

Figure 2

Figure 3