IEEE Circuits and Systems Magazine - Q4 2020 - 48

a load modulation network (LMN) was implemented in
the Doherty PA. One of the important parameters is the
transmission line characteristic impedances in the LMN,
which is adjusted to minimize degradation in efficiency
caused by the current difference between active devices. To investigate the impact of current difference on
the entire efficiency range, extensive theoretical analyses were performed, resulting in promising solutions.
In [10] and [11], compact three-way Doherty PAs were
designed. The compact size and cost reduction could be
obtained simultaneously by sharing the current of one
device with the remaining two devices.
In recent years, the techniques of carrier aggregation
(CA) and multi-input multi-output (MIMO) have been
widely adopted in wireless communication systems,
such as 4G and 5G. This required the transmitter to operate over a wide frequency range. Hence, broadband
Doherty PAs were urgently needed [22]-[62]. In [22], an
architecture for load modulation and its related design
procedures were proposed. The architecture neither
required a quarter-wavelength output impedance transformer, nor offset lines, which were considered as the
crime culprit of narrowband Doherty PAs. A broadband
Doherty PA was also proposed in [23], using the " L-type "
low-order impedance inverter named " post-matching " .
This topology, allowed the relationship between optimal drain impedances and load conditions of active
devices to be derived and an exhaustive design procedure for broadband Doherty PAs was given. However,
most of these broadband Doherty PAs sacrificed backoff efficiency and power utilization factor, which limited
further development on increasing the bandwidth. In
[31]-[39], some efficiency enhanced broadband Doherty
PAs were designed by the authors based on harmonic
manipulation. In [32], a broadband efficiency enhancement technique for post-matching Doherty PA was proposed. The concept of mutual coupling was introduced
between two second harmonic short circuit networks,
which achieved a short-circuit of the second harmonic
over a broad bandwidth at back-off point and saturation. As a result, the efficiency of the proposed Doherty
PA was enhanced over a wide bandwidth. In [36], a
mixed topology for high efficiency, broadband Doherty
PA was proposed. A harmonic injection network (HIN)
was integrated into the Doherty PA between the two active devices. This allowed manipulation of the waveform

through use of the second harmonic components, which
enhanced the peaking amplifier output power. However,
the HIN led to deterioration in bandwidth due to mismatch at the fundamental frequency. To mitigate this reduction in bandwidth, a bandwidth compensation technique was proposed to reduce the peaking device drain
impedance variation. In [55], at 60 GHz, a Doherty PA
based on 65-nm bulk CMOS processes, was designed to
solve the low trans-conductance which was caused by
class-C operation. An adaptive biasing network was devised to dynamically adjust the bias voltage of the peaking PA to overcome this. In [56], a transformer-based
Doherty PA technique was exploited to improve the linearity and back-off efficiency at mm-Wave frequencies.
An asymmetrical series power combiner with LC tuning
circuits was proposed for Doherty operation. A parallel
combiner at the PA output was then used to combine
the power from two identical PA units that used the proposed Doherty technique. In [59], the 28-/37-/39-GHz linear Doherty PA based on silicon was presented. A new
on-chip, transformer based power combiner reduced
the impedance transformation ratio at back-off, thus improving the bandwidth and efficiency.
In this article, Section II describes the operating principles of the Doherty PA. The basic circuit configuration
of the Doherty PA is also shown. Section III describes
the most recently developed broadband Doherty PAs.
While sections IV and V describe current technical challenges and application of the Doherty PAs in 5G wireless
communication systems, respectively. Next, process
technologies for the implementation of Doherty PAs are
compared in Section VI, followed finally by a conclusion.
II. Doherty PA Overview
The Doherty PA was first proposed for the amplitude
amplification of modulated signals in radio broadcasting systems using vacuum tubes in 1936. More recently,
the phase and frequency modulations were done to the
constant envelope modulated signals, which was considered the best solution for maximizing the PA output
power and efficiency. For this purpose, Doherty PA architectures are not the best choice. With the release
of new communication standards, this situation has
changed significantly. For higher data rates, spectral efficiencies required a mixed modulation techniques, i.e.,
both amplitudes and phase/frequency are modulated.

(Corresponding authors: W. S. Chan, S. Chen).
X. Y. Zhou is with the Department of Electrical Engineering, Princeton University, New Jersey, USA. W. S. Chan is with the Department of Electrical Engineering, City University of Hong Kong, Kowloon, Hong Kong SAR. China. (e-mail: eeej2710@cityu.edu.hk). S. Chen is with the Key Lab of RF Circuits and Systems of the Ministry of Education of China, Microelectronic CAD Center, Hangzhou Dianzi University, Xiasha High Education Park, Hangzhou 310018,
China. W. Feng is with the School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510006, China, Â and also
with the Department of Communication Engineering, Nanjing University of Science & Technology, 210094 Nanjing, China.

48â

IEEE CIRCUITS AND SYSTEMS MAGAZINE

FOURTH QUARTER 2020

IEEE Circuits and Systems Magazine - Q4 2020

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