Improving Power Efficiency of WLAN Systems Using Outphasing Power Amplifiers

As wireless communication technology continues to develop, the requirements for power amplifiers are becoming increasingly higher, and traditional amplifiers can no longer meet the needs of practical applications. This article introduces an out-of-phase power amplifier that can not only increase power but also maintain high efficiency at higher power levels.

The high power consumption and performance limitations of legacy 802.11a orthogonal frequency division multiplexing (OFDM) systems have hampered the adoption of 802.11a and dual-band WLAN products. The physics involved in processing multi-carrier waveforms such as OFDM and wideband CDMA (W-CDMA) fundamentally limits the efficiency, output power, and signal quality of linear power amplifiers, especially those that power legacy 802.11a systems. To fully realize the performance of these systems in a wide range of WLAN-enabled devices, including small, power-constrained devices, a new modem architecture and power amplifier design is required.

The 802.11a standard is based on OFDM modulation, in which data is multiplexed across 52 carriers, each of which can be modulated using BPSK, QPSK, 16QAM, or 64QAM. This type of propagation improves immunity to multipath fading and certain interfering waveforms, but its disadvantage is that the resulting RF signal has a large peak-to-average power ratio. In addition, high-level modulation methods require low amplification distortion to avoid increasing the error vector magnitude (EVM).

Traditionally, the output power (also

There is a complex trade-off between range, data rate, and power consumption. To achieve high data rates, good linearity is required, which is usually achieved by stepping back to a Class AB power amplifier, which results in reduced transmit power. Low transmit power results in a poor link, which in turn reduces the operating range. High power and long operating range are also achievable, but at the expense of reduced data rates or reduced battery life. In other words, users like low power, high data rates, and long operating range, but can only achieve two of the three requirements at the same time due to the linear Class AB amplifier responsible for handling signal amplification.

When operating at peak power, traditional Class AB amplifiers are very efficient (theoretical efficiency is 78.5%), but their efficiency drops very quickly at low power. When such amplifiers are used for 802.11a OFDM signals, the amplifiers must be adjusted to handle the peak power level, but on average they operate at a level 8dB lower than the peak, so most of the time they operate at very low efficiency, with an average efficiency of only about 10%. If the amplifier is set back to support a 54Mbps data rate, the efficiency will be even lower.

Therefore, a technique is needed to enable the amplifier to operate at peak power while being at peak efficiency most of the time. The answer is outphasing. Let's take a look at how the outphasing structure is established and its impact on 802.11a power amplifiers.

Out-of-Phase Amplifier

Linear amplification using nonlinear elements, called out-of-phase amplifier technology, can provide WLAN designers with another way to achieve high efficiency over a wider output power range. In an out-of-phase amplifier, two signals with fixed amplitude but different phases ("phase segments") are amplified in two separate amplifiers ("dividing amplifiers") and then combined to form a signal with different phase and amplitude. When the phase segments are in phase, the envelope power is maximum; when they are out of phase, the envelope power is minimum.

FIG1 uses a vector diagram to show how two signals α and β with constant voltage and different phases are combined to form an arbitrary voltage signal R. FIG2 is a schematic diagram of the structure of a power amplifier using out-of-phase technology.

Because the sub-amplifiers are always operating at their most appropriate and widest swing, each amplifier always has peak efficiency. If there is a combiner to provide isolation between the two amplifiers, the efficiency of the system will deteriorate due to the loss of the combiner. If a low-loss combiner (which does not provide isolation) is used, the overall system efficiency can be very high.

A special variation of outphasing amplifier technology is the Chireix technique, which uses a passive combiner. This combiner applies a load impedance to the sub-amplifier that varies with the envelope power, so that when lower output power is required, the sub-amplifier will drive a high impedance load. The varying impedance forces the amplifier to draw less current when low RF power is required, thus maintaining high efficiency even when it drops. It is important to note that the voltage swing at the output of the sub-amplifier is fixed, but the varying output impedance causes the current to swing, and the DC current requirement will also change.

When using the outphasing technique, the choice of the sub-amplifier is critical, and the Class F amplifier is particularly well suited to operating in this mode. The Class F amplifier is not linear, but as the sub-amplifier in the outphasing amplifier system operates at a fixed amplitude, this is not important. The Class F amplifier uses special termination methods at the second and third harmonics to minimize the voltage across the amplifier transistor when it is "on", thereby reducing power losses in the switching device. The peak output power of this type of amplifier is proportional to the square of the drain voltage, so the average output power can be set by the Vdd supply voltage, which can be set to operate at the maximum efficiency possible for any average output power amplifier.

Practical implementation of 5GHz

A 5 GHz out-of-phase amplifier has been developed that uses a pair of Class F amplifiers on a single GaAs die. For a Class F amplifier to be efficient, the active device must behave like an ideal switch, with minimal “on” impedance, low capacitance , and should switch quickly from the “on” state to the “off” state. In addition, the device must support high voltages to output sufficient power without large impedance transformations, which complicate harmonic termination and combiners. In a 50-ohm system with power greater than 1 watt, the load line requires that each amplifier must produce a voltage root mean square (RMS) of 5 volts, or 15 volts peak-to-peak, and the switch must tolerate peak voltage excursions far in excess of the peak-to-peak voltage. 0.5 μm GaAs PHEMTs can meet these requirements simultaneously, and power PHEMTs that support voltages greater than 17 volts and Fmax approaching 100 GHz are now available on the market.

The amplifiers are fabricated in pairs on a single GaAs monolithic IC with driver stages and biasing circuits . The amplifier die contains no combiners or terminations that follow the final devices . The combiners require low-loss transmission lines operating at 5GHz, while the Class F amplifiers require low-loss terminators operating at 10GHz and 15GHz. Since low loss is not possible by fabricating these components on GaAs chips, it is achieved through a precisely controlled wire connection with passive components fabricated on the module ceramic substrate. The final module is 8×8mm in area and is fabricated on 0.015-inch alumina using a thick-film-like process.

Adaptive Predistortion

Like all amplifiers, outphasing power amplifiers generate distortion. Distortion primarily comes from AM-to-AM conversion (gain compression) and AM-to-PM conversion, which results in an increase in the error vector magnitude (EVM) of the modulation constellation and increased out-of-band emissions.

Predistortion is a method of compensating for these distortions by providing magnitude enhancement and phase correction for large amplitude signals. In practice, adaptive predistortion is required to accurately reduce EVM and out-of-band emissions. An adaptive predistorter combines the desired transmitted signal (before digital-to- analog conversion, upconversion, and power amplification) with the downconverted digitized actual transmitted signal.

The signals are compared and the difference between the two signals is then used to update the complex coefficients in the predistortion lookup table.

For applications such as 802.11a WLAN, the EVM of the out-of-phase power amplifier can be reduced to about -30dB after adaptive pre-distortion. The adjacent channel emission level of the out-of-phase power amplifier can also be reduced to about -60dBc using the same method.

Phase segment

In order for the outphasing amplifier to work, a system is needed that can generate a constant envelope phase vector segment signal. Any signal can be decomposed into phase segments, which has been difficult to do in the past, but modern DSP technology makes it feasible, even for complex OFDM signals. For example, a single-chip physical layer (PHY) integrated circuit has been developed that can generate phase segments that are fully compatible with 802.11a signals when amplified by an outphasing power amplifier. In short, even if phase segments are used, the output formed is an interoperable 802.11a signal.

Figure 3 shows the relationship between output power and output stage source current versus the phase angle of the amplifier drive signal, with data from an amplifier running at 5.25GHz with a Vdd of 5V. Remember that the amplifier is in deep saturation and therefore operates at a constant voltage amplitude, but note that the source current is highly dependent on the out-of-phase angle. This shows that the impedance of each amplifier (looking toward the combiner) does increase at low output powers, while the source current decreases.

The exact amount that the source current drops can be seen in Figure 4, which shows the measured efficiency of the out-phasing PA at various output powers. It also shows the maximum theoretical efficiency for the ideal Class B and ideal Class A amplifiers, and that the actual Class AB amplifier will be between the ideal Class A and ideal Class B curves. Note that the measured efficiency of the actual out-phasing PA is better than the theoretically perfect Class B amplifier. At full power, the amplifier is operating completely in-phase, and 80% efficiency is observed. As the phase of the signal entering the amplifier decreases, the output power also decreases, but the efficiency decreases much more slowly than in a typical Class AB amplifier. At 7.8dB power below the peak, which is a typical peak-to-average ratio for an 802.11a signal, the amplifier is 46% efficient.

The driver stage also contributes to the overall power consumption. The power added efficiency (PAE) including the driver stage is greater than 33% at the 7.8dB back-off point. This back-off efficiency can be achieved over a wide range of supplies. Figure 5 shows the efficiency of the amplifier at 7.8dB back-off for various supply voltages. Note that there are really two ways to control the instantaneous output power of the amplifier, out-phasing and changing the supply voltage. The supply voltage is usually used to slowly change the average output power, and the phase angle is used to quickly change the instantaneous envelope of the signal.

Figure 6 shows that the PA achieves very high back-off efficiency over a wide range of output power levels. Comparing this performance to a conventional Class AB amplifier, whose efficiency drops off sharply outside the optimized operating point, the outphasing PA maintains an excellent PAE/power consumption ratio over a wide range of output power levels.

Conclusion

Traditional amplifier structures limit the range and efficiency of WLAN systems. The Chireix structure presented in this article has advantages that are particularly suitable for the 802.11a WLAN standard when working with extremely low loss combiners. This implementation has demonstrated unprecedented peak efficiency of 80% and average power added efficiency of more than 33% on real OFDM signals. The baseband processor that generates the correction signal to drive the Chireix amplifier is very practical and does not increase the complexity of the transmission structure. This innovation will enable the vitality of 802.11a WLAN and make full use of energy in portable devices and low-power applications.