A key requirement for military and aerospace communications is the ability to avoid eavesdropping. As interception techniques become more sophisticated so must the modulation and encryption schemes employed by transmission systems. Military and avionics communications systems now use highly agile radio communications, hopping from frequency band to band, as well as employing wideband protocols to allow efficient scrambling of the data and, in effect, make the transmission look increasingly like noise to an eavesdropper.
However, the shift to more complex modulation schemes and wideband, spread-spectrum communications technologies places increased pressure on the power amplifiers used for RF, which are typically tuned for efficiency over a relatively narrow band. Trying to use these devices for wideband operation results in low energy efficiency. A similar process has affected the commercial environment.
For example, in commercial protocols such as GSM, Gaussian minimum shift keying technique was chosen because it allows a linear power amplifier to operate in its most efficient, saturated region. More complex modulation schemes, such as quadrature amplitude modulation, provide greater spectral efficiency, allowing more data to be packed into a smaller bandwidth or provide greater headroom for encryption. As these modulation schemes alter phase and amplitude rather than just the phase of GMSK, the power amplifier needs to drop further into its linear region, which is less power-efficient than working close to the saturated region.
As the modulation schemes get more complex, the peak-to-average ratio or crest factor worsens. For comparisons from the commercial environment, LTE takes peak-to-average values to 10 dB, compared to the 7 dB of 3G UMTS and the 3 dB of GSM. This in turn has led to the development of circuit-level techniques that make it possible to deliver wideband communications without loss of efficiency.
One way of dealing with the issue is to take a modern form of a power-amplifier design developed in the 1930s at Bell Labs by William Doherty. The circuit uses two amplifiers in parallel. One is active all of the time, with a second amplifier adding additional energy for modulation peaks. Because the main amplifier can be run for longer under ‘back-off’ conditions at points close to saturation, the overall efficiency increases. The secondary amplifier provides fine-tuned control over the signal.
In principle, the use of more amplifiers than two can improve efficiency but will also increase cost and complexity. As communications techniques push into greater levels of digital complexity and spectral efficiency, the N-way Doherty design is likely to become more common.
Figure 1: The Doherty amplifier architecture.
Although the architecture of the Doherty amplifier is elegant and conceptually simple, details of the design can make large differences to performance. Operation is influenced by the coupling factor of the input divider and the way in which the carrier and peaking amplifier stages are biased. The way the peaking amplifier is turned on depends on both input power level and the gate bias voltage, which also determine the power efficiency of the low-output and peaking conditions. How these parameters are set depend on the type of signal that is to be applied.
For example, if the gate-bias voltage for the peaking amplifier is made more negative, the later this amplifier will turn on, which should increase efficiency under back-off conditions, but actual efficiency results may not follow for complex signal modulation schemes.
For a signal with a crest factor in excess of 10 dB, more than 98 percent of the RF power from a Doherty amplifier is generated by the carrier amplifier. For less than 2 percent of the time, the peaking amplifier does not need to be used.
First described a year after Doherty’s proposal, envelope tracking is another approach that can provide increased efficiency with complex, wideband modulation schemes that have high crest factors.
Envelope tracking uses digital controllers to continually adjust the supply voltage to the power amplifier. This avoids the situation of the amplifier being supplied with too high a voltage that will cause excess energy to be simply dissipated as heat. The efficiency improvement from envelope tracking tends to increase the further the amplifier operates from its ideal range as the voltage reductions allow the transistor to operate as fully saturated for a greater proportion of the time; its actual output is reduced thanks to the voltage reductions applied by the envelope-tracking circuitry.
Figure 2: Envelope tracking modulates the voltage to follow the required output amplitude of the modulation scheme in real time.
Doherty and envelope tracking techniques can be combined in a single circuit to provide the advantages of each. Typically, envelope tracking is applied to the carrier amplifier while the peaking amplifier is driven conventionally to take care of peaks that cannot be satisfied when the voltage supplied to the carrier amplifier is at its maximum. There are tradeoffs in terms of efficiency as losses in the divider/combiner and due to the peaking amplifier only being driven into saturation for very short periods result in an efficiency loss for high-amplitude signals. However, for lower amplitude signals that will be common in very-high crest-factor modulation schemes, the efficiency of the carrier amplifier can be maintained at a high level thanks to its long-term operation in the saturated region.
Figure 3: A comparison of Doherty and envelope-tracked Doherty against traditional power amplifier efficiency. (Source: Moon et al, 2010)
To support wideband modulation schemes, the power transistors and amplifiers themselves have evolved. Silicon and gallium arsenide are coming under increased competition from technologies such as silicon germanium and gallium nitride.
GaN supports the fabrication of high electron mobility transistor (HEMT) structures to deliver higher electron velocity than is possible with silicon devices. The high mobility results from the way in which a two-dimensional electron gas forms at the interfaces between the component materials. The carriers in this gas move far more freely than in materials such as silicon. As a result, GaN transistors are far more suitable for high-frequency power switching circuits.
A further advantage of GaN is that devices are able to operate in high-temperature conditions that would be challenging to silicon devices, suiting them to the extreme environments required for military applications.
As an example of higher-power, high-frequency transistors, Cree’s CGH40006P is a GaN high electron mobility transistor (HEMT). The CGH40006P, operating from a 28 V rail, offers a general-purpose, broadband solution to a variety of RF and microwave applications. GaN HEMTs offer high efficiency, high gain and wide bandwidth capabilities, making the CGH40006P ideal for linear and compressed amplifier circuits, and a number of circuits have been demonstrated that use the technology for both Doherty and envelope-tracking Doherty applications (1, 2). The transistor is available in a solder-down, pill package. The GaN-based transistor can operate from DC to 6 GHz to support the demands of wideband communications systems.
The RF3931 from RFMD also uses GaN to implement a 48 V 30 W high-power discrete amplifier designed for wireless infrastructure as well as general-purpose broadband amplifier applications. The amplifier’s process technology allows high efficiency and flat gain over a broad frequency range in a single amplifier design.
The RF3931 itself is an unmatched GaN transistor packaged in a hermetic, flanged ceramic package to provide excellent thermal stability through the use of advanced heatsink and power dissipation technologies. The use of simple matching networks external to the package allows the use of wideband gain and power performance in a single amplifier. Offering operation from DC to 3.5 GHz, the transistor can supply output power of up 50 W.
M/A-Com has designed GaN-based devices for both pulsed-power operation suitable for radar systems as well as continuous wave for communications applications. The MAGX series includes the MAGX-000035
family, a 30 MHz to 3.5 GHz power transistor that deploys GaN on top of a silicon-carbide substrate for additional ruggedness. The unmatched device employs a depletion-mode structure and is packaged in a flanged or flangeless ceramic package to provide excellent thermal performance. Combined with the SiC substrate and GaN technology, this type of device is highly suited to high-heat environments and can function with a junction temperature of 200°C.
For lower-power transmitter applications, the ATF-50189
from Avago Technologies is a high linearity, low-noise enhancement-mode pseudomorphic HEMT packaged in a low-cost surface-mount SOT89. The enhancement-mode operation allows simple, single-supply biasing and, through the use of patch resistors, allows the bias current to be fine-tuned for unit-to-unit variation. By providing the combination of low noise and high output IP3 at the same bias point means that the device is suited to both receiver and transmitter applications, simplifying the design of smaller radio systems. Its operating frequency range is from 400 MHz to 3.9 GHz.
Thanks to advances in both power devices and circuit architectures, it is now possible to achieve high spectral and energy efficiency in heavily encrypted wideband communications systems. As processes such as GaN improve and teams gain greater understanding of the tradeoffs in Doherty and envelope-tracking schemes, we can expect to see further enhancements.
- Simon Wood, Ray Pengelly. A High-Efficiency Doherty Amplifier with Digital Predistortion for Wimax. High-Frequency Design, December 2008.
- Junghwan Moon et al. Doherty Amplifier with Envelope Tracking for High Efficiency. IEEE International Microwave Symposium. May 2010.