Efficiency ratings for an amplifier can be the difference between battery operation and the need for a more robust power supply. High efficiency is a design goal for any power amplifier, although achieving efficiency typically comes at the expense of linearity. In addition, as the efficiency of an amplifier increases, the amount of bias power dissipated as heat decreases, simplifying the

thermal management concerns. Efficiency is defined in several different ways for an amplifier. The drain efficiency is simply the ratio of the RF output power to the DC input power, usually designated by

η = Pout/Pin

where
Pout = the RF output power (in W) and
Pin = the DC bias power (in W).

Another version of efficiency is poweradded efficiency (PAE) which includes the RF drive power (Pdr) at the input of the amplifier, rather than just DC bias, and subtracts it from the output power in the manner of (Pout Pdr)/ Pin to determine PAE. When an amplifier's gain is high, this approach provides valid results. But when the gain is low, the subtraction of the RF drive level can sometimes lead to negative PAE values. In addition, a form of efficiency known as the overall efficiency, which is in the form of Pout/(Pin+ Pdr) can be used in many cases to compare the performance of different amplifiers, and can even be modified by including power consumed by supporting circuits.

Another way to consider an amplifier's efficiency performance is by means of the instantaneous efficiency, which is the efficiency measured at a specific output level. For an amplifier operating with amplitude modulation (AM), the output level will vary over time as will efficiency. For many amplifiers, maximum efficiency will occur at maximum output level, which is generally the peak output power (PEP); the efficiency will decrease as the output level decreases. Because of these variations in output power and efficiency with modulated waveforms, it may be helpful to use a term known as average efficiency to represent the efficiency averaged over time:

ηAVG = PoutAVG/PinAVG

where

PoutAVG = average RF output power (in W) and
PinAVG = average DC bias power (in W).

For an amplifier with time-varying output power and efficiency, its output envelope can be described by a probability density function (PDF), which relates the amount of time spent for different output levels. Similarly, a cumulative density function (CDF) shows the probability that an amplifier's output envelope doesn't exceed a specified amplitude value. The nature of the PDF or CDF will depend on the modulation format, with complex modulation formats tending to yield nonconstant envelopes whereas more traditional modulation formats, such as FM, tend to yield constant envelopes.

Over the years, a number of different biasing schemes have been developed for amplifiers, from extremely linear Class A operation, in which active devices remain powered on through the full sinewave cycle of an input signal, to more power-efficient schemes, such as Class D, in which multiple transistors are switched on and off to conduct different portions of an input waveform.

Although Class A amplifiers achieve poor efficiency in converting the DC input power to RF output power, this approach provides outstanding linearity. A transistor Class A amplifier has continuous collector-current flow throughout each RF cycle, although only a fraction of the DC source power is converted into RF output power; the rest is dissipated as heat.

In a Class B amplifier, a transistor's base is biased near the collector-current cutoff point, causing the collector current to flow only for 180 deg. of each RF signal cycle. Since the DC source power is only on for one-half of each RF cycle, the DC-source-power to RF output power efficiency is higher than for a Class A amplifier. The penalty, however, is in increased output waveform distortion and a sacrifice in linearity.

Some designs use a pair of amplifiers in a "push-pull" configuration to provide amplification for each half of the waveform, although switching transitions from one amplifier to the other can result in some nonlinearities in the output waveform. To improve linearity, Class AB amplifiers combine the traits of the two approaches. In this arrangement, the collector current flows less than 360 deg. but more than 180 deg. for each RF cycle. Because of the large range of bias options, a designer can optimize the amplifier between the extremes of low distortion and low efficiency and high distortion and high efficiency. In a Class C amplifier, a transistor's collector current is biased to flow less than 180 deg. of each RF signal cycle. This can yield higher efficiency than a Class B amplifier, but with degraded linearity.

Higher-efficiency amplifiers have been developed in recent years, such as Class D and Class E designs, that employ switching between multiple active devices to achieve efficiency levels of 90 percent or more. In addition, some designs, such as Class G and Class H amplifiers, rely on variation of the voltage supply rails to achieve high efficiency. The supply rails can be varied continuously or in discrete steps. The penalty for higher efficiency is the higher circuit complexity needed to achieve the variable voltage supplies.