The Demands of 5G Drive RF Power Amplifier Enhancement
RF power amplifier designers arguably have some of the most vexing challenges, many of which have been made even more difficult since wireless networks transitioned from analog to digital modulation techniques. These issues have also been exacerbated by densely populated spectrum in which their designs operate. To tackle these and other problems, designers are relying on techniques designed to optimize the conflicting requirements for achieving optimum RF output power over wide channel bandwidths without sacrificing efficiency or linearity.
These tasks take on ever greater significance when viewed in the context of the contribution to the electric grid of small cells and IoT for 5G and other communications services (Figure 1). This fact has seemingly become lost in the flag-waving about 5G and IoT’s presumably enormous benefits. The broad umbrella of IoT includes everything from home automation systems to industrial production, “smart” cities and buildings, and eventually, autonomous vehicles and infrastructure that supports it, and 5G will play a major role along with Wi-Fi to deliver the required connectivity.
For example, every IoT application relies on sensors on or within which are short-range transceivers, and even though they consume very little DC power, when there 50 billion of them they will have an impact on the national electricity grid. Of course, some of their contribution to power consumption can be negated because their transmitters are not always operating and because they should increase the efficiency of the applications they serve.
However, there is no guarantee IoT will succeed in offsetting its unbounded energy consumption. The power consumed by information communications technology devices grows by about 7% per year, and as they are eventually connected to cloud data centers, it will increase their power requirements as well. While it is too soon to know how much IoT devices and small cells will demand from the grid, the efficiency of their tiny RF power amplifiers will be extraordinarily important.
The Challenges of High-Order Modulation Schemes
Signal modulation techniques such as orthogonal frequency-division multiplexing (OFDM), multivariate linear amplitude modulation, and others are driving amplifier performance requirements to new levels. All produce modulated signals that have high Peak to Average Power Ratios (PAPRs), so the signals tend to operate in their nonlinear region that produces distortion and interference. Consequently, high linearity and efficiency are essential as power amplifiers consume more power than any other component in a communication system (Figure 2).
OFDM, widely used in wireless systems today, has the advantage of eliminating Inter-Carrier Interference (ICI) and Inter Symbol Interference (ISI) in the signal while being more resistant than a single-carrier system to narrowband interference and frequency-selective fading. In addition, OFDM’s high spectral efficiency accommodates a large number of subcarriers in a very narrow channel.
However, OFDM and other higher-order modulation schemes pose major challenges for designers. When high-PAPR signals pass through a nonlinear power amplifier, the result is intermodulation distortion that creates out-of-band emissions, which results in interference with services in adjacent bands while degrading bit error rate (BER) performance.
To remedy this, amplifiers must have very high linearity, which is typically achieved by “backing-off” the level of the input signal so it remains within the linear operating range of the amplifier. This significantly reduces their efficiency, which is highest at its maximum output power when operated in compression.
Two approaches to avoid back-off include reducing the baseband signal’s PAPR and PA linearization. There are many PAPR reduction methods such as clipping and filtering, coding, active constellation extension, tone reservation, companding, tone injection, and others. All have their advantages and disadvantages, but no technique is fully effective in the reduction of PAPR.
The most widely used linearization technique is digital predistortion (DPD), which makes it possible to obtain more power from an amplifier without resorting to a larger, less efficient, and more expensive one. With DPD, a digital pre-distorter at baseband creates an expanding nonlinearity that is complementary to the compressing characteristic of the power amplifier. In essence, “inverse distortion” is introduced into the input of the amplifier, canceling non-linearities the amplifier might have.
Predistortion works by either correcting gain and phase distortions or canceling inter-modulation products, either of which produces impressive results. That is, a predistorter designed to correct gain and phase non-linearities also improves IMD, and a design focused on reducing intermodulation products also reduces gain and phase fluctuations. The end result is an amplifier that is more linear and reduces the amplifier’s distortion.
Achieving Higher Efficiency
Techniques such as envelope tracking (ET), the Doherty amplifier, and outphasing improve efficiency while maintaining linearity. What is surprising is that most of the techniques were invented many years ago, and while very appealing, never saw widespread use because signals only relatively recently have the high levels of PAPR that require them.
For example, ET (Figure 3) did not become popular until decades after it was first created in the 1930s by Loy Barton at the University of Arkansas to meet the challenges of high-power AM broadcast transmitters. It was the result of several years of work, and Barton’s name never became a household word even though he also invented the Class B amplifier. It was resurrected only when the third generation of cellular systems required a way to achieve higher efficiency when faced with complex modulation techniques. ET was first deployed in base stations in 2008 and in mobile phones in 2014.
ET has become very popular for wireless devices, especially those in which power consumption is a critical metric. It can be less complex than other techniques, uses a single power amplifier, and requires minimal additional components. ET achieves its benefits by continuously monitoring signal amplitude and rapidly varying supply voltage as a function of instantaneous output power to ensure that the device is always operating in the region where it achieves its highest efficiency.
This means that the EV circuitry ensures the amplifier is supplied only the voltage it needs when it needs it, which has a positive effect on power consumption. In contrast, an amplifier using a fixed supply voltage achieves its highest efficiency only when operating in gain compression, which it cannot do with OFDM signals because of the aforementioned creation of distortion. It thus becomes less efficient as the signal’s crest factor increases, and as a result, the amplifier operates mostly below its peak power and highest efficiency.
The Doherty architecture, now almost universally used in base station amplifiers, was invented in 1936 by W.H. Doherty of Bell Labs (then a part of Western Electric). When properly designed, a Doherty amplifier can significantly increase efficiency when compared to standard parallel Class AB amplifiers. A classic Doherty amplifier (Figure 4) consists of two amplifiers: a carrier amplifier biased to operate in Class AB mode and a peaking amplifier biased to operate in Class C mode. A power divider splits the input signal equally to each amplifier with a 90º. difference in phase. After amplification, the signals are rejoined with a power combiner. Both amplifiers operate during the peaks of the input signal and are each presented with a load impedance that enables maximum power output.
As input signal power decreases, the Class C peaking amplifier turns off and only the Class AB carrier operates. At these lower power levels, the Class AB carrier amplifier is presented with a modulated load impedance that enables higher efficiency and gain to be produced.
There are several types of Doherty power amplifier schemes: symmetric, asymmetric, and digital. The symmetric Doherty amplifier is the most basic type and uses two identical amplifiers. The asymmetric type is very popular as it has two different amplifiers, of which the peaking amplifier can handle higher power, which allows it to handle higher signal peaks, and the other amplifier focuses on lower signal levels. Together they can provide better performance than the basic symmetric type.
The digital Doherty amplifier (Figure 5) offers the most flexible approach and the ability to employ digital signal processing in both amplifier paths optimizes the performance that can be achieved when compared to the standard Doherty implementation. Instead of using a fixed analog splitter, the two signals come from the digitized baseband signals.
In the digital Doherty amplifier, a lookup table can be used for dynamic phase alignment between the carrier and peaking amplifiers, and DPD employed within an open-loop peaking amplifier signal path allows the peaking amplifier’s amplitude-modulation and phase-modulation response to be fairly constant. The phase difference issues between the two transmission paths can be corrected relatively easily by adding a constant phase shift at the input of the phase-lagging signal path.
A huge number of different Doherty amplifier designs have been created over the years, employing a wide variety of enhancement techniques ranging from placing multiple gain stages in the Doherty splitter and combiner to using programmable splitters, and envelope shaping, to name just a few.
Another amplifier architecture developed and patented 85 years ago by Henri Chireix is called outphasing (Figure 6) of AM broadcast transmitters to improve their average efficiency and linearity. The concept was revived in 1974 by D.C. Cox, who introduced the term Linear Amplification with Nonlinear Components (LINC) to realize a linear amplifier where the intermediate stages of RF power amplification could employ highly nonlinear devices. It has demonstrated its ability to increase amplifier efficiency and increase operating bandwidth. That allows the power amplifiers to continuously operate at their peak power efficiency while providing an almost undistorted output signal.
In theory, at least, the concept is straightforward. An amplitude-modulated signal is separated by a signal component separator (SCS) into two phase-modulated (PM) signals that have equal constant envelopes and opposite modulated phase variations. These two constant-envelope PM signals are then amplified separately by two independent identical power amplifiers after which the two amplified signals are combined at the amplifiers’ output to produce an amplified version of the original AM signal.
The crucial element in the Chireix outphasing system is the SCS that converts the AM signal into two outphased component PM signals with constant envelopes. This modulation conversion makes it possible to produce very efficient and very linear amplification, as the envelopes of the signals are now fixed while their magnitude envelopes have no signal information. That is, the amplitude information of the original AM signal is contained in the phase of the component PM signals, so amplifiers can be used in each branch to achieve very high spectral efficiency.
And as the “branch” amplifiers have fixed envelopes, the nonlinearity of the input-output power characteristic typical of most high-efficiency amplifiers has very little impact on the overall input-output transfer function of the outphasing system. This means that the system can be very linear across a wide range of signal levels, assuming that the SCS and power combiner do not introduce nonlinear signal distortion. The high-efficiency amplifiers realize high linearity, thus the LINC acronym that is typically used for these types of amplifiers.
In general, the approach controls the phase shift of multiple saturated or switched-mode RF power amplifiers to create a modulated RF output. A lossless, non-isolating power combiner produces load modulation of the amplifiers that modulate the output. As it can increase efficiency over a broad range of power levels, it is well suited for use in today’s high-PPAR communication systems.
To summarize, the Chireix (LINC) amplifier consists of the SCS, the power amplifiers, and a power combiner that achieve signal component separation, amplification, and signal component recombination in through outphasing. The approach makes it possible to use nonlinear amplifiers driven by constant-envelope signals that can produce much higher efficiency than linear amplifiers. Even though these nonlinear amplifiers have very high efficiency, they do not have an impact on distortion at the output as they operate at constant envelope signals. The remaining important consideration is the power combiner in which an inevitable trade-off between linearity and efficiency must be made.
Research into outphasing has been conducted by both device manufacturers and academia to eliminate some of its disadvantages, perhaps the most important of which is the need to separate the signal into multiple phase-and amplitude-modulated signals. This increases the number of required components (and thus cost) when compared with the Doherty approach that does not need this separation. However, ongoing research continues to mitigate these problems.
Suffice it to say that the design challenges posed by today’s digital communications schemes are far from trivial and require a continuous stream of advances from materials to system design. The techniques described in this article are just a smattering of the massive amounts of research being conducted into making amplifiers more efficient, linear, broadband, and affordable, all within the constraints posed by today’s advanced modulation schemes.
For proof, consider that every technical paper or presentation covering Doherty amplifiers, envelope tracking, outphasing, and other techniques, represents enormous levels of effort from industry and academia throughout the world. Even a decade ago, many of these technologies were either embryonic or remained unexplored. Today, any technology that shows promise is fair game for exploration, as requirements will only get more challenging as 5G moves to 6G…and beyond.
1. Thesis, Chireix’s/LINC Power Amplifier for Base Station Applications Using GaN Devices with Load Compensation, Jijun Bi, Delft University of Technology, September 2008.
2. Improvement of Linear Amplification with Nonlinear Components Based on COX’s Theory, Chengguo Liu, 3rd International Conference on Electric and Electronics (EEIC 2013).
3. Outphasing, Envelope & Doherty Transmitter Test & Measurement Application Note, Rohde & Schwarz, 2018.
4. Recent Developments of Dual-Band Doherty Power Amplifiers for Upcoming Mobile Communications Systems, Ahmed M. Abdulkhaleq, et al, MDPI, June 2019.
5. Optimizing the Perennial Doherty Power Amplifier, Gareth Lloyd, Rohde & Schwarz, Microwave Journal, March 2019.
6. A High Efficiency and Wideband Doherty Power Amplifier for 5G, Halil Voolkan Hunerli, Master’s thesis, Photonics and Space Engineering, Department of Microtechnology and Nanoscience, Chalmers University of Technology, Gothenburg, Sweden, 2017.
7. Techniques to Improve Power Amplifier Energy Efficiency for 5G, Erdem Bala, Leonid Kazakevich, Rui Yang, InterDigital Communications, Inc., November 2014.
8. A 28/37/39 GHz Multiband Linear Doherty Power Amplifier in Silicon for 5G Applications, Song Hu, Fei Wang, and Hua Wang, IEEE Journal of Solid-State Circuits, 2019.
9. Integration of RF Circuits with High-Speed GaN Switching on Silicon Substrates, Cambridge (University) Centre for Gallium Nitride, GaN power Electronics on Si.
10. A Review of 5G Power Amplifier Design at cm-Wave and mm-Wave Frequencies, Y. C. Lie, J. C. Mayeda, Y. Li, and J. Lopez, Hindawi, Wireless Communications and Mobile Computing, Volume 2018.