In just a few years’ time, so many things about wireless communication will be different from the way we know them today. Cellular networks, for example, traditionally made up primarily of larger base stations, will add huge numbers of ‘small cells.' The growth of the Internet of Things (IoT) will mean countless wireless sensors and controllers being deployed all around the world. Meanwhile, wireless carriers will be making much greater use of the millimetre-wave spectrum region.
All these scenarios demonstrate the need for optimum efficiency in radio frequency (RF) power transistors and amplifiers – after all, the transmit element of a transceiver accounts for a significant chunk of the overall operating costs.
While 100% efficiency in RF power generation is impossible, even relatively small increases help cut power consumption or the number of devices needed to achieve a particular RF output – which in turn means your amplifier can be smaller. In small cells and IoT kits, in particular, minimizing power consumption is essential. This is why considerable effort is being spent on increasing efficiency, both by RF equipment manufacturers and academic researchers.
As a result, we see more-efficient devices, higher classes of amplifiers and the use of design techniques including digital pre-distortion, crest-factor-reduction, and envelope tracking. We will look at these in more detail shortly.
First, however, there is gallium-nitride (GaN) technology. The CGHV14800F HEMT power transistor from Wolfspeed is an excellent example of the possible efficiencies you can achieve with GaN. This particular product is rated at 800W between 1200MHz and 1400MHz, but typically performances (with a pulse width of 100µs and a 5% duty cycle) in the ranges between 910W at 1400MHz (67% efficient) and 1000W at 1200MHz (74% efficient). This product is aimed at L-band radars, such as those used in long-range surveillance, weather systems, air traffic control and anti-missile defense.
The Doherty architecture rises once more
The return to the Doherty architecture design is also helping improve RF power efficiency. Created in 1936, this approach lay dormant for many years before making a comeback and is now used in the majority of base station amplifiers, thanks to recent advances in its design. The Doherty architecture enables designers to create amplifiers with high power-added efficiency from input signals with high peak-to-average ratios (PARs). Compared to a traditional parallel Class-AB amplifier, a well-designed Doherty amplifier can be up to 14% more efficient.
The reason the Doherty architecture was little used for many years was the prominence of the amplitude modulation (AM) and frequency modulation (FM) schemes. For a long time, only a few types of signal had the right characteristics for Doherty. Today, things are different: high-PAR signals are ubiquitous, including in Wi-Fi, LTE, WCDMA and CDMA 2000.
A typical Doherty amplifier is classed as a load-modulation architecture and is made up of two amplifiers. One is a carrier that is biased to work in Class AB mode; the second is a peaking amplifier, biased for Class C operation. The input signal is divided equally between the two, with a 90-degree phase difference, and is re-joined post-amplification. Both amplifiers get used during input peaks. In addition, by presenting each with a load impedance, you get maximum power output. The architecture is shown in Figure 1, below.
A one-ended, two-way Class AB Doherty amplifier. Source: Slipstream Engineering Design Ltd.
When the power of the input signal drops, the Class C amplifier switches off, leaving the Class AB carrier in sole operation. Because it is presented with modulated load-impedance, it delivers better gain and efficiency.
Why we need linearization
One drawback of Doherty amplifiers, compared to dual Class AB amplifiers, is their reduced linearity. To address this, you need to employ digital and analog linearization. Most commonly used is digital pre-distortion (DPD), or a combination of DPD and crest-factor-reduction (CFR). These techniques significantly cut distortion, while the right design of the amplifier and overall device keeps any loss of linearity to a minimum. Note that these techniques can also be applied to non-Doherty architectures.
Amplifiers work best when used close to their saturation point. Go any higher, and your output power will begin to drop off, regardless of increases to the input power. You also get distortion, which can interfere with other channels and services. This distortion is why many designers reduce RF power output to a ‘safe zone,' to maintain linearity, even though this results in reduced efficiency. Consequently, you may need more RF transistors to achieve a particular level of RF output, which drives up power requirements (and hence operational costs), as well as the bill of materials.
DPD inversely distorts the signal going into the amplifier, which cancels out the amplifier’s nonlinearity. This means designers do not need to stay within the aforementioned sub-optimal ‘safe zone,' so can deliver higher RF output, without necessarily needing more RF power devices. A more efficient amplifier also means lower power consumption and less cooling requirement, with all the benefits that these characteristics bring.
CFR reduces the input signals’ PARs, thereby cutting peaks so that signals can go through the amplifier without distorting or being clipped. Combine DPD and CFR, and the benefits can be even greater.
Doherty is not the only amplifier architecture enjoying a resurgence. Outphasing patented almost 80 years ago, can boost amplifier efficiency and operating bandwidth. It works by producing a modulated RF output, which it does by controlling the phase-shift of switched-mode or saturated RF power amplifiers. The output modulation is achieved using a lossless, non-isolating power combiner.
Outphasing is ideally suited to modern, high-PAR communication systems because it can boost efficiency across a wide range of power levels. Research into outphasing is ongoing to minimize its drawbacks, most notably the need to split the signal into many amplitude-modulated and phase-modulated signals. This has traditionally put outphasing at a disadvantage compared to Doherty designs, because of the increased component count and cost. However, with progress being made in minimizing these issues, outphasing could well be used in combination with Doherty amplifiers.
Another efficiency-boosting technique that amplifier designers can use is envelope tracking. This ensures the amplifier continually operates in its peak region (and hence at maximum power) by dynamically adjusting the applied voltage.
In contrast, conventional designs use fixed supply voltages and are only at maximum efficiency when they operate in compression. Consequently, the efficiency of this type of amplifier drops as crest factor increases, because the amplifier mostly operates below its peak power level and optimum efficiency.
Envelope tracking gathers envelope information from the IQ modem’s properties and sends it to a specialist power supply, which delivers the right voltage. Qualcomm was first to build envelope tracking into a chip in 2013, and the technique is now found in many smartphones.
Its main drawback is its limited signal bandwidth, between 20MHz and 40MHz. This is far below the 160MHz used in IEEE 802.11ac Wi-Fi and even LTE Advanced’s 100MHz. That said, newly developed Envelope Tracking Advanced (ETAdvanced), which is based on long-term research into Asymmetrical Multilevel Outphasing (AMO) by two MIT professors, looks to have eliminated this issue. The professors’ company, Eta Devices, boasts a real depth and breadth of expertise, with those involved including Analog Devices co-founder Ray Stata, serial entrepreneur Mattias Astrom and Dr. Mark Briffa, formerly of Huawei and Ericsson. Nokia acquired Eta Devices in late 2016.
According to Eta Devices, ETAdvanced enables designers to deliver greater efficiency from their base stations than any other technique – both at peak power and back-off. It claims to be 25% more efficient than envelope tracking in narrowband scenarios and requires fewer components, resulting in a simpler design. Moreover, it supports channel bandwidths up to the 160MHz level, meaning it can be used in 802.11ac and LTE-Advanced. It could also be useful for 5G devices.
Eta Devices has demonstrated an 802.11ac Wi-Fi RF power amplifier, using ETAdvanced at a bandwidth of 160MHz. This device used 80% less power than a traditional amplifier in the same use case. According to Eta Devices, this represents a breakthrough, in that it is the first demonstration of advanced supply modulation in a scenario with wide bandwidths.
The continuing RF power efficiency drive
These examples illustrate the breadth and depth of research going into improving the efficiency of RF power generation. There is much more besides, including combining different classes of amplifiers.
Moreover, with the IoT and 5G communications driving the need for increased data rates, this push to enhance RF power efficiency will continue unabated.