Title

NOVEL COMBINER CIRCUITS FOR A DOHERTY RF POWER AMPLIFIER

Author

James E. Mitzlaff

Abstract

Using a Doherty circuit is one good way to improve RF Power amplifier (PA) efficiency, but it has found only limited application in practice because there are two critical transmission line lengths which must be set correctly in order to minimize distortion products such as IMD and spectral regrowth. While there are methods which can be used to approximately predict the required transmission line lengths, these methods are usually not accurate enough to give first pass success, so the design must be iterated. Iteration of this type of design is complicated by the fact that changing the length these transmission lines forces all the components before or after them to move on the layout.

This paper describes several ways to eliminate these critical transmission lines by using various types of quadrature hybrid combiners to replace the combiner structures in prior art Doherty PAs. This, in turn, allows the Doherty PA to be "tuned" for minimum distortion by varying the length of a transmission line stub or changing the value of a reactive component, which can be done without significantly altering the existing circuit board layout.

Body

INTRODUCTION - PRIOR ART

A block diagram of a prior art Doherty PA circuit is shown in Fig. 1, on the following page, and its operation is described below. This circuit is also used in the ADS simulation shown in Fig. 2, which produced the results shown in Fig. 3.

The input signal is split into two quadrature (90º phase difference) components by HYB1. These two signal components are applied to two amplifiers, the carrier amplifier (CA) and the peaking amplifier (PkA),  which are generally identical except for their gate bias levels. In this particular embodiment, the CA is connected to the in-phase (0º) output of HYB1 and the PkA is connected to the quadrature (‑90º) output.

The CA is biased as it normally would be when used as a standalone Class AB amplifier. The PkA is biased at, or just below, cutoff (i.e. Class C bias), so that it draws no idling current, and little or no current at low input power levels, but will draw current and amplify signals at high input power levels. Because both the CA and the PkA used in this example are multi stage amplifiers (3 stage RFIC models) the cutoff bias for the PkA can be applied to just it's 3rd (final) stage, or it can be applied to multiple stages. In this case, the preferred method is to apply cutoff bias to the 2nd and 3rd stages of the PkA, via the ports g2p and g3p, while ports g2c and g3c are used to provide Class AB bias to the CA. The port g1 applies Class AB bias to both the CA and PkA stages.

Transmission line elements TL1 and TL2 are used modify the output characteristics of both the CA and the PkA in order to maximize overall efficiency and minimize distortion products such as IMD and spectral regrowth. In order to maximize efficiency, element TL2 is used to rotate...