For most amplifier technologies and architectures, the linearizer adds the maximum benefit at the highest operating powers where inherent distortions are greatest. Depending on the amplifier type, up to 3 dB of additional RF power can be realized for the same distortion products. For the higher-order modulation schemes linearization is a common approach especially for macro applications.

Generally, the linearity of an amplifier improves for low-power signals and this results in smaller distortion products. When an amplifier is backed off from its maximum-rated power, the effect of any linearization is much less hence, linearization in general, does not provide improvements in amplifier performance for low-power operation. (See Figure 1.)

It is clear from Figure 1 that the linearizer adds benefit at the upper and lower excursions of the RF signal envelope. When the RF signal is backed off by 3 dB from maximum rated there is no real benefit from linearization. As the RF power is further reduced efficiency drops. Although linearizers offer little benefit at lower RF powers, they can be used to indicate the level of distortion occurring.

For linearizers using negative feedback, there will be an error signal that relates to the error induced in the power amplifier. This can be used to determine if the bias is too high or two low for the particular application. For 3G applications, closed loop linearization is a significant challenge given the wide loop bandwidths required. Some of this functionality would need to be performed in the digital domain.

In situations where the power amplifier is operating at significantly lower power than maximum, there is another option of using adaptive bias and power supply modulation to improve efficiency. Assume that a 3G WCDMA class amplifier produces 12% efficiency for maximum RF power. If we wish to produce an RF output, which is a quarter of maximum, the dc dissipation remains the same with efficiency dropping to 3% (0.25 * 12).

With adaptive bias for Class A, for an amplifier 6 dB below maximum RF power, this would result in a halving of bias current and supply voltage when compared with the full RF power case.

Figure 2 shows this technique can maintain amplifier efficiency over a range of powers. This approach is of benefit if the RF power amplifier is only expected to run at full-rated power for a short time during a 24-hour period. The bias circuitry needs some knowledge of the RF power requirement in order to adjust itself. The benefit to be realized depends on the amplifier characteristics. For example, a device operating in Class A could realize the full benefit.

A device operating in Class AB would potentially have less benefit as the quiescent current varies with RF power automatically. The bias current does need to be modified to track temperature changes. However, the significant change would be modulation of the supply voltage. This is shown in Figure 3 (based on real measurement).

In the case of a Class AB amplifier, the efficiency with adaptive bias does decrease with falling drive as Class AB transistor are sensitive to external biasing arrangements and efficiencies fall away more rapidly than theory would suggest. Under these conditions, the adaptive biasing can set the correct quiescent current for a particular supply voltage to ensure optimum linearity. This is often referred as the sweet spot and maximizes the device efficiency for the given operating conditions.

This can be taken a step further where the adaptive bias is dynamic and tracks the modulation envelope in real time. It is possible to use dynamic modulation of the power supply to track the signal but this generates distortion products with the systems that are problematic (alteration of the bias and power supply results in unwanted modulation of the RF signal). In addition, a significant safety margin is required to ensure that the amplifier can deliver the RF power required at that instance that degrades efficiency.

Therefore, dynamic power supply modulation cannot achieve the low distortion products required for most modern communication systems.

It should be noted that a power supply unit (PSU) capable of dynamic voltage adjustment is a significant challenge. Care needs to be taken to ensure the PSU maintains high efficiency and stability while achieving rapid change in output voltage.

However, by combining linearization and adaptive bias, one can optimize amplifier efficiency over a range of RF powers. In this hybrid application, both adaptive bias and linearization are included. Figure 4 shows such an arrangement.

Both systems are wrapped around a power amplifier, which could be a single or multistage unit. The areas of interest are the linearizer and adaptive bias system. Such a system is possible for narrowband systems such as GSM or EDGE. Nonetheless, different linearization strategies would be required to meet the signal bandwidth for 3G.

In this design, the linearizer is a negative feedback system (for example, a polar loop) comparing the amplitude and phase distortions caused in the PA. These comparisons are used to derive error signals, which drive modulators that oppose the distortions occurring within the amplifier. The operation of the polar loop linearizer is discussed in^[1].

For adaptive bias, the power amplifier block (two stage) is connected to an active power supply that can alter its output voltage in a dynamic manner. The bias control block sets the quiescent current for the amplifiers. For a Class A amplifier, both of these can vary in sympathy to maintain the correct load line for different RF power requirements.

Previous 1 2 3 Next

RSS    Save to Del.icio.us  Digg This