Shrinking wireless and mobile electronic devices are impacting antenna design and performance. Portable device designers must work with miniature antennas covering 824 to 2170 MHz or more, different modulation schemes, and high data rates. With shrinking handsets, the area for the antenna is limited, often with the antenna wrapped around peripheral functions in the handset. Such a solution often makes the antenna more susceptible to detuning by environmental effects and lowers the antenna's efficiency. An optimum antenna tuning technology is needed to offset these effects: it must be low loss, linear, with low power consumption, while being capable of handling RF levels to +40 dBm (30 V peak).

Wireless handsets, of course, are no longer just cellular telephones, but must also support noncellular services such as mobile television, Bluetooth, wireless-local-area-network (WLAN), and Global Positioning System (GPS) applications. Because of new features, functionality, and industrial design requirements (such as cameras, keyboards, and thin handsets), little space is left for the antenna. As antennas are wrapped and repathed, they lose efficiency. Fortunately, some of this lost performance can be recovered with antenna tuning, in which the system uses dynamic impedance tuning techniques to optimize the antenna performance.

An open-loop antenna tuning system is often used when a passive antenna cannot meet the desired performance requirements. The tunable element fine-tunes the performance of the antenna at set frequency bands and modes of operation, and it can also take into account information regarding transmit/receive frequencies, modulation scheme, or use case (e.g. slide open or closed). This information is stored in a lookup table in the baseband memory when the handset is designed. However, an open-loop system does not measure the operation of the antenna in real time so it cannot take into account changing environmental conditions (Fig. 1a).

In a mobile device, the environment is constantly changing as an operator walks, drives, or moves their fingers while holding the handset. To compensate in real time, adaptive closedloop antenna tuning can be used, where an impedance mismatch sensor provides constant feedback by tracking the operation of the antenna. The mismatch sensor looks at the power that is reflected back to the antenna, which is measured as the antenna's voltage standing wave ratio (VSWR). The sensor then compares the amplitude of this reflected power to the transmit power and makes an adjustment to the impedance tuning circuitry, allowing the closed-loop antenna tuner to track the optimal frequency and matching for the antenna in all use cases. The tuning algorithm forces the tunable elements to constantly track and adjust to the optimal setting (Fig. 1b).

The biggest roadblock to implementing antenna tuning in cellular handsets has been the absence of a high-performance, electronically tunable reactive component that is low loss and has a wide tuning ratio. The most challenging component requirements have been power handling and linearity. Antennas for GSM transmit at power levels to +33 dBm, but under mismatch conditions, the tunable component must handle RF signal levels to +40 dBm (30 V peak).

Designers have experimented with microelectromechanical systems (MEMS) and ferroelectric materials technologies, such as barium strontium titanate (BST) technology, to implement tunable antennas and filters. While these techniques show promise, they are not yet ready for high-volume production. To solve the problem of antenna tunability in handsets, designers need a new technique that uses proven technology and can immediately support highvolume manufacturing.

The RF transceiver portion of a mobile handset is designed for an impedance of 50 ohms. Ideally, its antenna would exhibit a 50-ohm impedance across the full frequency band. This rarely occurs, however, because small wireless handsets have an inherently narrow antenna bandwidth, poor matching and low radiation efficiency. The bandwidth limitation forces handset antenna designers to aim at a non-50-ohm impedance across the full operating band, with typical VSWR specification of 2.0:1 or 3.0:1 for multiband antennas.

A handset antenna's impedance is also affected by environmental factors, such as how users hold their handsets. The proximity of a user's body also causes significant absorption of the power radiated by the handset, further reducing the radiation efficiency of the antenna. Typically, handset antennas operate at a VSWR of less than 3.0:1, but when a user places a finger on top of the antenna radiator, the VSWR may degrade to as much as 9.0:1. This performance degradation, even in nominal free-space operation, is significant, considering that all devices in the signal chain were designed to operate at an ideal VSWR of 1.0:1. Figure 2 demonstrates the "hand effect," which refers to the detuning of the antenna when the user holds the handset with his or her hand in close proximity to the antenna radiator. This tends to decrease the resonant frequency of the antenna, causing it to be badly mismatched at its intended operating frequency.

Front-end power is lost with antenna mismatch. For an antenna with VSWR of 3.0:1, about 1.25 dB power is lost due to power reflection. At a VSWR of 5.0:1, the mismatch loss increases to 2.55 dB. This mismatch also causes the handset power amplifier's (PA's) output to drop, further reducing the handset transceiver's radiated power.

Narrowband duplex or receive filters in handsets also experience ripples on the passband when not terminated to their characteristic impedance. Depending on the phase angle, this may cause 1 to 2 dB of incremental loss in addition to the mismatch loss. Figure 3 shows a typical Band V WCDMA duplexer transmit filter response. The green line shows how the filter performs at 50 ohms. The red line is the specification, and the blue lines show the filter response when the antenna has VSWR of 5.0:1 across all phase angles. In the worst case, the filter's insertion loss is about 5 dB.

The absorption loss in the body, mismatch loss in the antenna, ripples in the RF filter passband, and reduction in the PA output power combine to severely reduce the power radiated by the handset. These effects are directly visible to the handset user as a decrease in battery life, degradation of the link budget and call quality, and an increase in the number of dropped calls. In response to this, many network operators have developed radiated power requirements for handset antennas. Antenna specifications such as total radiated power (TRP) and total isotropic sensitivity (TIS) are tested by simulating actual use cases with head and hand rather than simply testing a handset in free space or performing a conducted emissions measurement in a 50-ohm environment.

In order to meet these stringent power specifications, adaptive antenna tuning may be the only option. The approach forces the antenna to appear 50 ohms despite environmental effects, and the rest of the system operates optimally. Although an antenna tuner adds its own insertion loss to the circuit when the antenna is at 50 ohms (a 1.0:1 VSWR), adaptive antenna tuning can significantly improve the overall insertion loss from the tuner input to the antenna input compared to an uncorrected system (Fig. 4), with performance improvements in the PA and RF filters.

Designers have used passive matching networks to increase antenna bandwidths for multiband handsets or to improve the matching of a constrained antenna. Such matching networks are comprised of discrete capacitors and inductors. They are fixed in value, so the matching network is also fixed to match the antenna for a certain frequency band and/or set of conditions. The matching cannot be adjusted for changing conditions. In some cases, designers may attempt to overcome antenna detuning in multiband designs by using "brute force" approaches, such as increasing the distance between the internal antenna and the handset covers, or placing the antenna in a different location inside the handset. RF switches may also be used to optimize the antenna from one band to another, by switching different load impedances or by changing the electrical length of the antenna radiator. But such approaches typically do not allow enough resolution in tuning control.

One additional challenge facing handset antenna designers is the fact that different frequencies are used for transmit and receive operations. For example, a time-divisionduplex (TDD) system such as GSM transmits one burst in time and then quickly receives a second burst in time. This requires the transceiver to quickly switch back and forth between transmit and receive functions. When the antenna is tuned for transmission, its performance may not be optimum for reception.2 This limits the handset's sensitivity and may counteract the benefits of antenna tuning. If the tuning element can change reactance fast enough (in typically 5 microseconds), the antenna can be quickly retuned between transmit and receive functions. For frequency-division-duplex (FDD) systems such as WCDMA, the transmitter and receiver operate simultaneously, requiring additional intelligence for the tuning algorithm to optimize the antenna simultaneously for transmit and receive frequencies.

Continue to page 2

Page Title

Multiband antennas require a circuit that provides good performance at the band edges, actively tracking any detuning in the antenna, and retuning the antenna to the correct position in a short period of time. Since it is connected to the antenna, this circuit must be extremely linear so as not to generate harmonics or intermodulation distortion (IMD). Any tunable element, such as a variable capacitor, must have a tuning ratio of at least 3:1. In addition, the whole circuit must consume less than 1 mA current to minimize overall system power requirements. Finally, the tuning element must be small, rugged, and reliable, with low insertion loss and a quality factor (Q) of 50 or better, otherwise it will not improve system performance.

Several technologies have been suggested as antenna tuning solutions, including the use of MEMS switched capacitors and barium strontium titanate (BST) capacitors. The challenge with both of these emerging technologies is that neither has been proven in real-world deployments, let alone high-volume production. For example, MEMS switched capacitors are adjustable capacitors consisting of one metallic plate that is suspended several micrometers above a second metallic plate. By altering the separation space, the capacitance is adjusted. These devices can be arranged in an array, creating a tuning matrix to achieve precise control of the system capacitance. Despite the potential for high Q and linear performance, there are significant challenges remaining for this approach. For instance, the high power levels in a GSM handset may cause the two metal electrodes to latch or self-actuate. They are susceptible to the effects of microphonics (mechanical shocks modulating the capacitance) and electrostatic discharge (ESD); cannot meet the GSM transmit/receive switching time requirement; and must be housed in a hermetic package. Even though the MEMS capacitors have high Q, the parasitic inductance and capacitance of the lossy bulk silicon substrate often degrade the overall Q. Work now focuses on improved control of the capacitance values and better accuracy. Despite the promise, researchers are still working to demonstrate this technology, so it is not quite ready for high-volume production.

BST capacitors are based on thinfilm ferroelectric materials. For implementing a variable capacitor, the idea with this technology is to apply a high-voltage DC bias to the material, change its dielectric constant, and thereby change the capacitance. This material has been in development for decades, and it still faces technical hurdles for handset applications, including linearity, ease of use, and performance reliability over temperature. BST does not tune in a linear fashion and suffers from hysteresis, making the technology difficult to use. In addition, the material's key properties such as linearity and losses may change over time and temperature.

A major challenge with both MEMS and BST technologies is that they need a high bias voltage (up to 30 V or higher) to tune, typically requiring a separate CMOS charge pump and controller chip. Usually, these exotic technologies cannot enable monolithic integration of additional RF, analog and digital circuitry on the same die, and therefore require multiple ICs to implement adaptive antenna tuners. Typically, adaptive antenna tuners based on MEMS and BST technologies are being implemented using multichip- module (MCM) technology.

As an alternative to MEMS and BST approaches, designers at Peregrine Semiconductor have applied the firm's UltraCMOS process technology and HaRP design innovations1-3 to create DuNE technology, a patentpending design methodology for producing practical digitally tunable capacitors (DTCs). These are variable capacitors with digital control (see table) that meet the challenging cellular handset antenna tuning requirements for low bias voltage, high linearity, and high tuning accuracy. DTCs (Fig. 5a) will be available in flip-chip form measuring just 1.36 x 0.81 mm (Fig. 5b).

Unlike bulk CMOS and silicon-oninsulator (SOI) technologies, Ultra- CMOS field-effect transistors (FETs) can be stacked to handle high RF power levels, due to the fully insulating sapphire substrate. This allows power capabilities to be scaled from +20 dBm to more than +40 dBm to handle the RF power levels in GSM and WCDMA operation without degrading Q or tuning ratio.

For cellular applications, DuNE DTCs can be designed with capacitance values from about 0.5 to 10 pF, with typical tuning ratios ranging from 3:1 to 6:1 with 5-b resolution or 32 states of resolution (Fig. 6).Typical Q values range from 40 to 80 at 1 to 2 GHz (Fig. 7). In addition to better than +38 dBm of power handling at 50 ohms (Fig. 8) and switching speed of better than 5 microseconds, this new technology meets other necessary specifications for a cellular antenna tuning circuit. For instance, the power consumption is only about 100 microamps (orders of magnitude lower than some of the alternative tuning technologies mentioned).

Since miniaturization and parts counts reduction is crucial in cellular handsets, it would be highly desirable to be able to integrate all of the key functions of an adaptive antenna tuner on a single die. Figure 9 shows conceptually how DuNE technology would allow for the monolithic integration of a DTC RF core with digital functions (serial interface, tuning algorithm, digital mismatch sensor), analog functions (charge pumps, biasing circuitry), as well as RF active (power detectors) circuitry. This would allow for a self-contained and autonomous solution with minimal parts count and area.

DuNE Technology offers a new and intriguing alternative to the MEMS and BST technologies for implementing cellular antenna tuning in handsets. It is a technology based on proven building blocks and process technologies that are already shipping millions of units per week to the handset industry. Standalone, self-contained DTCs communicate directly with the transceiver or baseband, and can be used in a wide range of applications, from open-loop optimization of antenna match in both the cellular and non-cellular antennas, to implementing a closed-loop adaptive antenna tuner for stringent multimode mobile handset designs. All parameters of the DTC (capacitance values, tuning ratio, quality factor and power handling) can be changed by circuit design instead of materials engineering, making it very fast to generate new application-specific designs. DuNE DTCs for cellular as well as mobile TV applications are sampling to strategic customers around the world, and are expected to be in volume production in the 2009-to-2010 time frame. Ultimately, DuNE technology allows for monolithic integration of the complete adaptive antenna tuner system to help handset designers to meet ever-demanding antenna requirements.

REFERENCES

1. Refugio Jones, "RF-MEMS Enable Tunable Antennas," Semiconductor International, May 1, 2007.

2. A. Tombak, "A Ferroelectric-Capacitor-Based Tunable Matching Network for Quad-Band Cellular Power Amplifiers," IEEE Transactions on Microwave Theory & Techniques, Vol. 55, No. 2, February 2007, pp. 370-375.

3. Peregrine Semiconductor, "UltraCMOS Process Technology," http://www.psemi.com and "HaRP Technology Innovations, http://www.psemi.com/content/foundry/foundry_harp.html.