Advertisement

This article describes the architecture of a power amplifier module for polar-modulated schemes. Polar-modulated power amplifiers use polar coordinates' representation of a signal-vector (vs. the traditional Cartesian I-Q representation used in the W-CDMA modulation scheme, for example). In this scheme, the recombination of coordinates is done at the output of the power amplifier. It is employed to make use of highly nonlinear (that is, very efficient) RF power amplifiers. However, the recombination of polar coordinates is essentially a nonlinear process. It means that the amplifier should stay nonlinear (with preferably unchanged level of nonlinearity) all the time and under all conditions.

In this article, a theory of a polar modulation is given. Based on it, the required degree of a power amplifier's nonlinearity is figured out. It is demonstrated that nonlinearity, while being a necessity, cannot become too large either. A straightforward method of keeping up the allowed nonlinearity window is shown and its shortcomings are pointed out. Finally, improvements are introduced and explained in detail. The resulting architecture allows a correct recombination of polar coordinates under any conditions of power amplifier operation.

Polar Modulation

The polar modulation concept is based on the fact that pure phase modulation does not create intermodulation products (sidelobes). The sidelobes are distorting a spectral mask. They are the reason for employing highly linear amplifiers. In case of phase-modulated signals, though, one can use very nonlinear (that is, very efficient) amplifiers. It will be shown that polar modulation actually requires a very nonlinear amplifier for proper operation.

Modern digital communications are characterized by their constellation diagrams, created by the positions of a signal-vector. Each modulation scheme defines a constellation diagram (QPSK, 3π/4 DQPSK, 64QAM, etc.), the transmitter produces it, and the receiver looks for signals in predetermined positions of the diagram. Traditionally, the positions of a signal-vector are described by Cartesian coordinates (Real and Imaginary, that is I-Q); however, the same positions can be exactly described by polar coordinates (amplitude and phase or magnitude and angle). The polar presentation allows at least the phase portion of a signal-vector representation to be amplified by very efficient amplifiers. The magnitude portion of a signal-vector is just a discrete DC variation, that is, requires no sophisticated amplifiers.