Industry Migration to 3G Multimode Puts Premium on Handset’s Front-End Module Design
By Nick Cheng, Skyworks Solutions, Inc.

During the past decade, cellular handsets have moved from single-band platforms to much more complex multi-band, multi-mode architectures that optimize carrier capacity flexibilities, support numerous geographic air interface standards and enable more multimedia features. More recently, these multi-band, multi-mode handsets have begun supporting new modulation protocols that enable higher data rates, such as EDGE and WCDMA, which is the underlying air interface used for universal mobile telecommunications systems (UMTS), as well as high speed downlink packet access (HSDPA) and high speed uplink packet access (HSUPA). All of these protocols are derived from the UMTS standard as well as enhancements of 3G capabilities, which added channels and applied new modulation and coding techniques while delivering improved error recovery capabilities.

HSDPA/HSUPA has been much heralded as the answer for mobile operators who want to offer users a true mobile broadband experience while competing more effectively with DSL service providers, supporting up to 14.4 Mbps of downlink and 5.76 Mbps of uplink data transfer rates. The main advantage for a 3G operator with a WCDMA network is the ability to move up to HSDPA/HSUPA with relatively insignificant additional infrastructure cost, through a relatively simple software upgrade within the existing WCDMA network. This technology is expected to enable these operators to provide advanced multimedia and content services while delivering on the promise of 3G.

These operators have great hope that it will stimulate ARPU data, since it maximizes the ability of WCDMA to provide broadband services (in the same way that EDGE increases spectral efficiency as compared to GPRS, so does HSDPA/HSUPA). The higher spectral efficiency along with higher speeds not only enable new classes of applications, but also support a greater number of users accessing the network, with HSDPA providing over twice the capacity of WCDMA.

With enriched features and services such as music downloads, movie streaming, video teleconferencing, mobile video games and broadcast television provided to 3G end users, the complexity of handset designs has been escalating tremendously while the size and cost of cellular phones have continued to decline significantly. As a result, there is a clear motivation and path of evolution that highly integrated front end modules (FEM) are needed [1, 2], as opposed to the discrete solutions, in particular for the 3G multi-band, multi-mode handset architectures.

Consequently, new design challenges arise and they greatly impact the development of 3G WCDMA FEM in many ways. Multiple frequency bands have to be supported due to the lack of a single global frequency band for UMTS. For instance, the latest FEM technology addresses WCDMA/HSDPA/HSUPA applications in bands I, II, III, IV, V, VI, VII and VIII and the list is still growing, as shown in Table 1. In addition, spectrally efficient signal waveforms, which often have higher peak-to-average ratio (PAR) to support an enhanced data rate, dictate the power amplifier (PA) to operate at a back-off power level in order to meet the linearity requirement at the expense of reduced power-added efficiency (PAE). This phenomenon is observed in EDGE applications, where PA efficiency degrades since 8-PSK needs to be supported in stead of the GMSK used in the GSM applications for higher data rate. Moreover, a PA’s performance may vary under mismatch or high VSWR conditions when the impedance at antenna port changes, leading to degradation and variation in RF performance at the system level, i.e. reduced total radiated power (TRP), increased current consumption and degraded linearity, even though the level of severity depends on the phase angle and VSWR presented by the antenna load. One thing to keep in mind is that these relatively new challenges are extra additions to the other existing requirements for handset FEM products – small form factor, low bill of material costs, increased functionalities, robust product quality and high production yield. Therefore, FEM designers have to come up with innovative solutions to reduce current consumption, or improve PAE, at various power levels in order to reduce overall talk time while meeting more stringent linearity requirements and minimizing variation and degradation of RF performance under mismatch conditions, in particular for the more demanding HSDPA/HSUPA applications.

The PA is the engine and key building block of a handset FEM, and gallium arsenide heterojunction bipolar transistors (GaAs HBTs) are currently the technology of choice for handset PAs due to their excellent performance characteristics and established high-volume manufacturability [3]. Each FEM design may consist of one or more PA blocks or paths, depending on how many frequency bands need to be supported. Typically, the bias circuitry and control logic associated with each PA often reside on the same HBT die and may include internally generated reference voltage (Vref), which is a desired feature in many applications for cell phone designers [4]. In general, the efficiency and linearity performance of FEMs is largely determined by the load line presented to the final transistor stage of the PA and there is a trade-off between efficiency and linearity. Assuming that supply voltage is fixed and current saturation is avoided, a load line with lower impedance corresponds to better linearity margin due to less collector voltage clipping and distortion. However, current consumption increases and efficiency drops due to higher loss associated with the matching networks resulting from a higher impedance transformation ratio. In the same fashion, a load line with higher impedance corresponds to lower current consumption and better efficiency but degraded linearity. Higher PAR of the HSDPA/HSUPA signal waveforms, along with the current trend of moving towards lower battery voltage, imposes even more constraints on the PA load line design. Since the load line is established with reactive components and set at the center of the frequency band of interest, the load line changes with frequency and has a bandwidth limitation. The FEM performance tends to degrade or exhibit linearity/ efficiency trade-offs more as the PA operates at a frequency further away from the center. It is noteworthy to mention that the definition of operating bandwidth for a PA and FEM has less to do with the conventional “3dB” bandwidth defined by using the insertion loss of the passive output matching network but more to do with the requirements of efficiency and linearity.

The configuration of a WCDMA FEM can be quite different depending on the application. It may consist of one PA block/path, an inter-stage filter and a duplexer along with an integrated coupler/detector, as shown in Figure 1, supporting only one specific frequency band, i.e. band II from 1850 to 1910MHz [4, 5]. This configuration greatly simplifies the development tasks of handset designers since all of the PA filtering elements are included and the FEM is the only component placed among the RFIC (TX), the LNA (RX) and the antenna. In addition, the loss of the PA output matching network is minimized since direct impedance transformation from the duplexer input to the transistors of the PA output stage can be achieved; therefore, the PA can operate as close to compression as possible and better efficiency can be attained, assuming that the linearity requirement and peak output power at antenna are both met. The challenge for the FEM is to minimize coupling and leakage. For example, one must maximize the isolation between the TX and RX ports. This is achieved with proper filtering, grounding, shielding and layout within the module. Since different designs are optimized for different frequency bands of interest, a different FEM is needed for each different frequency application. In this particular example shown in Figure 1, a coupler and a detector are also integrated inside the FEM for open-loop power control, along with an internally generated reference voltage with a form factor that measures at 4mm × 7mm [4].

The WCDMA FEM configuration can take on a completely different direction by incorporating more than one PA block/path and coupler/detector but excluding the inter-stage filter and the duplexer, as shown in Figure 2. This 6 × 6 mm FEM consists of two PA paths; one supports the 900MHz band, or low band (LB), i.e. from 880 to 915MHz for band V and VIII, and the other supports the 1900MHz band, or high band (HB), i.e. from 1850 to 1980MHz for band I and II. Accordingly, the bandwidth requirement increases by more than two-fold, from 3 percent (band V) and 3.9 percent (band VIII) to 10.5 percent (band V & VIII) at low band and from 3.1 percent (band I) and 3.2 percent (band II) to 6.8 percent (band I & II) at high band, escalating the design challenge of PA load line. However, with this configuration, different combinations of frequency bands can be easily mixed and matched by the phone designers with the proper selection of external inter-stage filter and duplexer based on only one FEM design, suitable for multi-band phones, i.e. band II & V for the U.S. market or band I & VIII for the European market. This approach reduces the number of different devices required to be purchased and improves the supply chain, enabling the reduction of inventory cost since different FEM products covering different applications will not be needed. Both of the configurations discussed are based on Skyworks’ proprietary Load Insensitive Power Amplifier (LIPA) topology, which significantly minimizes performance degradation and variation under a mismatch condition [4, 6].

For the development of handset WCDMA FEM products, a system-in-package (SIP) approach based on multi-chip modules (MCM) renders great flexibility and cost/performance trade-off [7] since the best process technologies can be chosen for each building block with the overall considerations of cost, performance and manufacturability, i.e. GaAs HBT for PAs, PHEMT for RF switches, BAW or SAW for filtering elements and GaAs chip or laminate for couplers. This approach also enables shorter time-to-market and design re-use when it comes to developing a family of derivative products for a variety of frequency bands since relatively less development time is needed to modify an existing FEM design, i.e. from band I to band II, by switching the filter and duplexer and re-centering the on- and off-chip matching elements, provided similar maximum output power is required for the FEM. The top view of a typical Skyworks’ FEM developed for WCDMA/HSDPA/HSUPA applications is shown in Figure 3. This MCM module consists of a single HBT PA die, an integrated coupler/detector on a GaAs die, an inter-stage filter and a duplexer, along with other SMT components. The other example, as shown in Figure 4, consists of two HBT PA dies with on-chip detectors and two embedded laminate couplers for dual-band applications.

Handset talk time is a very critical parameter in particular for HSDPA/HSUPA applications since the higher PAR signal waveforms impose more stringent demand on linearity performance, which lead to PA back-off operation and inherently lower efficiency. The talk time for WCDMA handsets is determined by the power distribution function. While dependent on the PA peak power efficiency, talk time is heavily weighted by the level of current consumption associated at the low/mid-power level where the FEM operates, as shown in Figure 5. With the use of DC-to-DC converter, the PA current drawn from a handset battery can be reduced quite significantly, especially at the low/mid-power levels, with little impact to FEM architecture except that the design needs to be compatible with the use of external DC-to-DC converter and the FEM performance may suffer at peak power due to the occurrence of certain Ohmic drop, i.e. 100mV, at the bypass mode. Another approach is to significantly improve efficiency at low power level by modifying PA load line and/or reducing bias level, leading to lower current consumption and DC quiescent current [4, 9]. A digital mode control pin will be needed for the switching between low- and high-power modes with different load line and bias level, which is also shown in the FEM block diagram shown in Figure 1.

With the rapid transition to more complex multi-band, multi-mode handset architectures, more stringent requirements and greater challenges are imposed upon the development of 3G FEMs in particular for WCDMA/HSDPA/HSUPA applications. The FEM solutions provide higher level of integration, compact size, increased functionality and more enhanced performance, enabling handset designers with shorter development cycle time and more design flexibility. The combination of the SIP approach and the selection of best front-end topology will enable handset FEM vendors to deliver a solution with low current consumption, excellent linearity margin and minimized performance variation/degradation under mismatch conditions, while meeting the requirements of small form factor, low bill of material costs, increased functionalities, robust product quality and high production yield.

References
[1] R. Jos, “Technology Developments Driving an Evolution of Cellular Phone Power Amplifiers to Integrated RF Front-End Modules,” IEEE Journal of Solid-Stage Circuits, Vol. 36, No. 9, September 2001, pp.1382-1389.
[2] B. Daly, “Handset Front-End Design Increases in Complexity with Move to Multi-Band, Multi-Featured Cellphones,” Microwave Product Digest, November 2007.
[3] A. Metzger, et al, “An InGaP/GaAs Merged HBT-FET (BiFET) Technology and Applications to the Design of Handset Power Amplifiers,” IEEE Journal of Solid-Stage Circuits, Vol. 42, No. 10, October 2007, pp.2137-2148.
[4] G. Zhang, S. Chang and Z. Alon, “A High Performance Balanced Power Amplifier and Its Integration into a Front-end Module at PCS Band,” IEEE MTT-S International Microwave Symposium Digest, Honolulu, HI, USA, 3-8 June 2007, pp.251-254.
[5] G. Zhang, S. Chang and A. Wang, “WCDMA PCS Handset Front End Module,” IEEE MTT-S International Microwave Symposium Digest, San Francisco, CA, USA, 11-16 June 2006, pp.304-307.
[6] S. Chang, N. S. Cheng, H. Finlay and B. Park, “Load Variation Tolerant Radio Frequency Amplifier,” U. S. Patent No. 6,954,623 B2.
[7] S. Machuga, “Semiconductor Innovation and Integration Drives Next-Generation Handsets,” Microwave Product Digest, November 2006.
[8] http://www.gsmworld.com/documents/twg/blm_prd_45.pdf.
[9] T. Fowler, et al, “Efficiency Improvement Techniques at Low Power Levels for Linear CDMA and WCDMA Power Amplifiers,” IEEE RFIC Symposium Digest, Seattle, WA, USA, 2-4 June 2002, pp.41-44.

SKYWORKS SOLUTIONS, INC.
www.skyworksinc.com
TXTLINX.COM101
Email this article to a friend!