Data is playing an increasingly important role in cellular handsets, requiring new designs to receive and transmit data as well as voice. The transition from circuit−switched to packet-based radios, demonstrated by the success of GPRS, has enabled further enhancements to data-driven applications. The availability of Enhanced Data rates for GSM Evolution (EDGE), a packet-based system, doubles the data-rate capability of GPRS for many handset functions, such as e-mail access, video cameras, and MP3 players. Fortunately, the POLARIS™ 2 TOTAL RADIO™ polar modulator from RF Micro Devices (Greensboro, NC) helps conserve power consumption and cut heat in compact EDGE handset designs even with these additional functions.

EDGE uses 3π/8-shifted 8PSK modulation, frequently referred to as eight-state phase-shift-keying (8PSK) modulation. Supporting 8PSK over normal GMSK adds reasonable but significant complexity to the development of cellular radio transceivers, power amplifiers, and the overall handset. The primary consideration is that 8PSK is a nonconstant envelope signal, which includes an amplitude signal component. As a result, good EDGE performance requires accurate phase and amplitude control. In contrast, GMSK uses only phase modulation.

Reduced battery life and heat dissipation are two of the most significant parameter for GPRS- and EDGE-based handsets, especially when using multislot transmit operation (two or more bursts are received or transmitted within a single frame). Problems arise from poor transmitter efficiency, especially for the power amplifier. In voice-only handsets, the duty cycle of the power amplifier is only 12.5 percent of the overall transmit/receive frame, which is one burst (time slot) out of eight. With Multi-Slot Class MSC12 multislot operation, this jumps to a 50-percent duty cycle and quickly increases the dynamic power consumption and the heat of the power amplifier. Many handset manufacturers are targeting MSC12 for near-term handsets, but a significant barrier to their production has been the problem of the heat dissipation.

Polar modulation is a relatively new approach to improving the efficiency of the transmit system. Polar modulation dates back several decades but has not been used commercially in handset developments until recently. The two primary forms of polar modulation are small-signal and large-signal polar modulation. Some systems are based o the use of one or more feedback loops or paths, while others employ at least a partially open loop. Each approach has trade-offs.

All current polar approaches for EDGE separately process or create the amplitude and phase signals, but ultimately recombine these signals prior to transmission. Large-signal polar modulation recombines the phase and amplitude signals at the power amplifier. In the case of the POLARIS™ 2 TOTAL RADIO™ solution, amplitude modulation is provided by varying the collector voltage of the amplifier, a method historically known as plate modulation. The POLARIS 2 transmit system is implemented using an all-digital design which provides excellent repeatability from device to device and enables pure digital interfaces with the baseband circuitry (Fig. 1). Some competing large-signal polar modulators use an analog approach, which is subject to process variations, which can lead to performance variations in large volumes.

Small-signal polar modulation (Fig. 2) strips off the AM signal at the output of the I/Q modulator using an amplitude-modulation (AM) detector. This signal is then fed into the voltage control input of a variable gain amplifier (VGA). The VGA recreates the modulation by varying the signal level to the input of a linear power amplifier. In this case, the phase and amplitude are recombined at the VGA. Some of these same techniques can be used with large-signal polar modulation. The primary difference is the VGA is where the recombination is done for small-signal polar modulation, and the power amplifier is where the recombination is performed for large-signal polar modulation.

The phase signal can be generated in several ways. Many systems create the phase signal at baseband and use a standard in-phase/quadrature (I/Q) modulator to provide frequency upconversion either to an intermediate frequency (IF) or to RF. The translational loop architecture is the most common approach for Gaussian minimum-shift-keying (GMSK) transmit systems, where the modulation is initiated at baseband and the radio performs the upconversion to RF. Adding amplitude capability to the standard translational loop is one method to used to implement polar modulation. In the POLARIS 2 solution, a fractional-N synthesizer generates the phase modulation.

The polar loop system represents a variation to the polar theme. The system can employ either large-signal or small-signal polar modulation and can have one or two feedback signals to represent the amplitude signal, the phase signal, or both. In the case of amplitude feedback, a power detector is often used at the output of the power amplifier. This further lowers the efficiency of the transmit path by a few percent. The feedback paths provide value by monitoring the output for changes in phase or amplitude driven by changes in VSWR. The feedback enables a real time method to compensate for these changes, but not without some penalty. In general systems with feedback have increased complexity, higher power consumption and lower efficiency, but can do well with VSWR changes. The POLARIS 2 solution uses a large-signal polar modulation system in an open-loop configuration between the transceiver and the power amplifier. This requires no post power-amplifier detector as in traditional polar loop systems, and is a benefit of the POLARIS 2 implementation.

A standard I/Q modulator can be used as an alternative to polar modulation (Fig. 3). Phase and amplitude information are sent as baseband signals to the I/Q modulator which upconverts both signal components and feeds them to the base of the power amplifier. This approach requires good carrier suppression to be effective. The POLARIS 2 solution avoids this requirement with no upconversion and no carrier signal.

The I/Q modulator approach requires surface-acoustic-wave (SAW) filters for each band in the transmit path, adding cost and size to the overall solution. With tri- or quadband handset designs, the number of SAWs (and the cost of this approach) increases. The filters are needed to reject noise several megahertz from the channel frequency in order to meet the ETSI receive-band noise requirement. (Transmit-band SAW filters are not required with the POLARIS 2 system as the noise in the receive band is low enough without them.) Additionally, the I/Q modulator line-up requires use of a linear power amplifier where the efficiency is less than a saturated power amplifier (Fig. 7).

Page Title

From this comparison, it can be seen that the primary benefit across all polar modulation approaches is the reduction in filtering required to meet the receive-band noise specifications. SAW filters add insertion loss and reduce power efficiency when placed in a handset transmitter's final stages, adding heat and reducing battery life

The POLARIS 2 approach employs predistortion (sometimes referred to as a feedforward technique) as a linearization scheme to compensate for the AM-to-AM and AM-to-PM distortion that occurs in the transmit power amplifier. With an optimal predistortion scheme, the power amplifier can be used in its saturated state while still providing adequate linearity (a practice in modern GSM handsets). This provides for additional transmit power efficiencies reducing heat and increasing battery life.

The POLARIS 2 architecture employs a low-power polar modulation approach. Baseband transmit data are split into amplitude and phase components (Fig. 4). Phase components are predistorted to account for the PLL loop-filter rolloff. They are then combined with the channel selection word of the fractional-N synthesizer. The fractional-N synthesizer traces out the frequency-versus-time waveform to provide the phase modulation for the 8PSK signal.

The amplitude components are scaled according to the power-amplifier ramping control signal applied to the power-amplifier collector voltage control system. Thus, the amplitude components are "plate" modulated directly at the power amplifier. This is possible because the amplifier's power control system provides a highly linear amplitude transfer function between the input control signal voltage and the output RF voltage. The amplifier operates in saturation at all times for superior power efficiency.

EDGE output power for power class E2 is limited to +26 dBm (−4/+3 dB) in the high bands and +27 dBm (±3 dBm) for normal operation in the low bands. Standard GMSK output power is limited to +33 and +30 dBm for the low and high bands, respectively, with a ±2-dBm window. Power amplifiers are usually designed to achieve maximum power-added efficiency (PAE) at the maximum output power since this is when the highest power consumption occurs. Due to the fact that 8PSK is specified to a lower maximum output power, the power amplifier is less efficient when in this mode. Where modern GMSK (GSM) power amplifiers provide around 35-to-40-percent efficiency, when measured at the antenna, the same amplifier in EDGE mode provides less than 30-percent efficiency.

The reduced output power for EDGE is driven by a number of reasons, but for linear systems, a primary goal is to enable the power amplifier to be in saturation for high-power GMSK signals and in a linear region when in 8PSK mode. This allows use of a single amplifier for both modes of operation. For improved efficiency, the POLARIS 2 approach maintains the power amplifier in saturation in both modes. This enables the overall current consumption to be lower and has been shown to enable seamless switching between modes in a multislot burst.

The use of predistortion in the POLARIS 2 approach helps eliminate all power detectors or couplers, feedback circuits, and many other functions required to support a feedback loop. Power consumption is reduced in open-loop systems due to reduced complexity and less insertion loss after the power amplifier.

Technical concerns with an open loop system include the need to meet critical parameters including error vector magnitude (EVM), output RF spectrum, and output power. Figure 5 shows the EVM performance, Fig. 6 shows the output RF spectrum, and Fig. 7 shows the output power of the POLARIS 2 modulator. The solution has been demonstrated to meet all ETSI requirements, has achieved Full Type Approval and has passed all FCC requirements and testing.

The POLARIS 2 polar modulation approach offers savings in power consumption compared to a direct-conversion system or I/Q modulator using a linear power amplifier (Fig. 8). In comparing systems, it is critical to achieve the best efficiency at the highest power levels, since power consumption is greatest at these levels. The efficiency of the power amplifier in the polar modulation approach gives that method a clear advantage in power savings, since the power amplifier dominates the overall power consumption of the transmitter at the higher power levels.

This improvement in efficiency applies to both GMSK and EDGE modes. The linear power amplifier cannot be driven into deep saturation while in GMSK mode like the polar saturated power amplifier can, as the linear PA must be biased to come out of saturation to meet the linear requirements in EDGE mode, limiting efficiency in both modes.

Although the pure percentage differences appear small, approximately 5 percent, this is 5 percent out of a 25-percent absolute figure. The relative improvement of the polar modulation method over the linear I/Q method shows about 20-percent relative improvement in power-added efficiency. This will result in improved battery life of a similar percentage for a voice-based handset. The actual battery life improvement will depend on many other considerations with overall handset functionality being the dominant variable.

The value of polar modulation has been clearly presented. By eliminating the need for SAW filters savings are achieved in cost, size, and power consumption when compared to linear modulators and power amplifiers. Large-signal polar modulation brings the greatest benefit by use of a saturated power amplifier and reduced insertion loss after the power amplifier providing the higher transmit efficiencies. These benefits combine for reduced heat and increased battery life compared to other methods. RF Micro Devices, 7628 Thorndike Rd., Greensboro, NC 27409-9421; (336) 678-5567, FAX: (336) 678-0091, e-mail:, Internet: