IEEE Circuits and Systems Magazine - Q4 2020 - 53

In recent years, various approaches and different techniques have been
used to increase the bandwidth of Doherty power amplifiers.
back-off efficiency also deteriorates with decreasing
RL . Based on the topology mentioned in [23], section
III-B and Section III-C, will describe methods on how to
achieve higher efficiency from OBO point to saturation
without sacrificing other parameters (such as bandwidth, output power and gain) will be discussed.
In [24], a high-efficiency Doherty PA using an integrated compensating reactance (CR) for broadband operation was proposed. The influence of the reactance in
the combining load for the carrier load impedance at the
OBO point in wideband operation was analyzed. An additional quarter-wavelength transmission line was integrated into the peaking amplifier output, which resulted
in the generation of CR for the load carrier amplifier load
impedance. Consequently, enhancement of back-off efficiency could be obtained over a wideband and without
the expense of load modulation at saturation. That is
because the characteristic impedance of quarter-wavelength transmission line is Z0, which did not affect the
load modulation at saturation. To satisfy the impedance
matching requirement simultaneously at back-off and
saturation, a two-point matching OMN design method
using ABCD matrix was proposed and followed a systematic design procedure.
In [25], various kinds of output combiners allowed
Doherty operation, which gave rise to a continuum of
Doherty amplifiers. In general, key circuit parameters
were determined for each configuration so that adequate load modulation mechanism could be achieved
over an extended bandwidth. An extra degree of freedom ( a ) was given in the Doherty PA design making
it suitable for different solutions and trade-offs between them highlighted. Subsequently the parameters for the output combiner and input splitter, and
the current and voltage profiles of carrier and peaking amplifiers could be calculated using the given design equations.
In [26], a paralleled Doherty amplifier architecture
was proposed for dual-band/broadband operation. For
the paralleled Doherty PA, two quarter-wavelength
transmission line impedance inverters were used in the
peaking branch. This results in a large variation of power
and efficiency of the peaking device thus deteriorating
the Doherty efficiency behavior in broadband configuration. For the post-matching Doherty PA, one low-pass
impedance inverter and one quarter-wavelength transmission line impedance inverter were used in the peaking branch. This resulted in a more consistent efficiency
FOURTH QUARTER 2020

and power of the peaking device over entire operating
frequency band. Consequently, a more typical Doherty
efficiency behavior could be seen.
B. Broadband Doherty Power Amplifier Focusing on
the Fundamental and Harmonic Terminations
For the Doherty PA works of [30], [32], [35], [36] and [50],
they consider the fundamental and harmonic terminations, simultaneously. It should be noted that the efficiency performance of these Doherty PAs were improved
but not at the expense of the bandwidth, which was in
general a result of the harmonic manipulating networks.
In [30], the continuous-mode technique was proposed and used to design a novel broadband Doherty
PA. Harmonic isolation was typically required between
the two transistors in the conventional Doherty PA.
However, this requirement restricts further development of broadband Doherty PA with high efficiency. On
the contrary, with the help of a post harmonic tuning
network, the two transistors allow to modulate each
other at harmonic frequencies. Consequently, a series of
highly efficient Doherty PA modes could be created over
a continuous frequency band.
For efficiency enhancement of Doherty PAs, the second harmonic short-circuit condition is a good method.
By placing in shunt, a series LC resonant circuit before
the FII, harmonic termination can be achieved. However,
there are some drawbacks when introducing the shunt
second harmonic short-circuited network (SHSN) into
the post-matching Doherty PA. Firstly, the bandwidth of
the Doherty PA is sacrificed due to the mismatch caused
by the SHSN. Secondly, each SHSN requires a large
equivalent L at low power and a small equivalent L in the
Doherty region. That is because the imaginary term for
optimal impedance of the device will move towards the
real axis.
In [32], a pair of mutually coupled SHSN is introduced
into the LMN for efficiency enhancement in the post-matching Doherty PA (shown in Fig. 7). This approach achieves
the " short-circuited impedance " at the second harmonic
terminations but not at the expense of bandwidth.
Influenced by both magnetic coupling (Lm) and electric coupling (Cm), the equivalent capacitor of single SHSN
becomes C 1 - C m and C 2 - C m . The equivalent inductor of
single SHSN becomes L 1 + L C 1 + L m and L 2 + L C 2 + L m .
In the OBO point, only the carrier PA conducts. Both
magnetic and electric coupling co-exists in the SHSNs,
and is shown in Fig. 8. According to the -equivalent -circuit,
IEEE CIRCUITS AND SYSTEMS MAGAZINE

IEEE Circuits and Systems Magazine - Q4 2020

Table of Contents for the Digital Edition of IEEE Circuits and Systems Magazine - Q4 2020