Power density of the new communications systems supporting the networked battlefield can be four times lower than that of the AM/FM radios they replace due to the use of advanced modulation techniques to improve data throughput of Net-Centric communications systems. These software-defined radios must operate over wide transmit bandwidths, which combined with the newer waveforms compound the already challenging task of making these radios lighter and last longer on fewer batteries. A power management technique described by Bell Labs in 1937, but only recently successfully implemented in commercial cellular and broadcast communication systems, is attracting interest. This article examines the latest developments in military communications technology.
Upgrading Military Communications
To put the current situation into context, it is believed that military communications systems will go through a dramatic upgrade cycle over the next five to ten years. The upgrade, which has already started, will create a Net-Centric communication system where voice, data and video information can flow securely and rapidly throughout all theatre elements in the battlefield.
The USA’s Joint Tactical Radio System (JTRS) program is the vanguard of this change in how the military communicates in the battlefield. New Handheld, Manpack and Small Form Factor (HMS) radios under the JTRS program offer tactical vehicles and dismounted units with reliable, good quality connectivity over a wide bandwidth even in rugged and urban environments.
Such secure, reliable high bandwidth voice, data and video communications technology is essential to deliver the required speed of command in today’s conflict environment. It is the two-way backbone that connects the whole chain of command from the top down to the lowest level, giving critical and immediate situational awareness and maximizing combat effectiveness.
These new standards derive their higher throughput by using more complex modulation schemes, but these reduce the efficiency of the RF power transmitters. The 2 MHz to 2 GHz bandwidth specification for JTRS radios imposes further challenges to the designer of the RF transmission circuits. Power density of the new radios can be worse than the established frequency hopping AM/FM technology by a factor of four, with a negative impact on the Size, Weight and Power (SWaP) envelope of the communications systems a vehicle-borne or dismounted unit needs to carry.
These considerations are driving a major push to improve the efficiency of battlefield communications technologies. A major area of focus is the RF amplifier, which can end up consuming as much as half the power in a high speed modem. Power amplifier designers are faced with a difficult optimization challenge that must balance size and efficiency yet work over the demanded transmit bandwidth.
Advances in power transistor technologies have allowed designers to meet their size goals by addressing the wide transmit bandwidth in only one or two power amps; however, the efficiency has suffered accordingly. It is time to take a fresh look at the design of the RF transmission amplifier chain.
RF Power Amplifier Efficiency
The waveforms used in the new networked battlefield communications protocols are usually OFDM- or QAM-based and support frequency hopping and adaptive signal to noise encoding schemes. For example, the Wideband Networking Waveform (WNW) specified by JTRS for ground to ground and ground to air communications and the Tactical Targeting Network Technology (TTNT) used for airborne sensor, shooter and ordnance communication is based on the OFDM modulation scheme.
The Soldier Radio Waveform (SRW) for soldier to soldier communications is based on QAM modulation. The Mobile User Objective System (MUOS) for satellite to ground, sea or air communication uses both OFDM and QAM, and leverages the W-CDMA technology developed commercially for existing mobile phone networks.
Figure 1 Conventional Class AB power amplifier configuration.
Channel coding and modulation techniques like QAM and OFDM require faithful reproduction of the amplitude of the transmitted RF signal. RF PAs are classic Class AB amplifiers (see Figure 1), which offer the most efficient operation when the RF envelope waveform is closest to peak power.
Efficiency is a function of the RF signal crest factor (peak-to-average power ratio or PAPR) where the higher the peak power with respect to the mean power, the lower the efficiency. This in turn is determined by the type of modulation and coding scheme. There is no single formula that defines that relationship, but a good rule of thumb is that every dB of crest factor reduction provides a 2 to 2.5 percent efficiency change.
In a W-CDMA transmitter, the PA peak power is usually 4 to 6.5 dB above the mean power. OFDM signals are composed of a large number of individual components, the power of each varying with time. The resultant amplitude of the composite signal over time is therefore not constant, but ‘peaky’ in nature and results in even higher crest factors—up to 9.5 dB—and even lower PA efficiencies.
In general, the higher the data rates, the higher the PAPR and the more difficult the amplification design process becomes. This non constant amplitude modulation means that a typical amplifier rarely runs up to its saturated output power capability resulting in lower efficiency.
High PAPR signals make the design of PAs difficult for two reasons: Firstly, the amplifier must be linear over a wide dynamic range to preserve modulation accuracy and spurious performance. It is possible to use a technique called Crest Factor Reduction (CFR) to allow the PA to operate closer to peak power for most of the time by limiting the peaks of the signal using DSP techniques. However, this needs to be done with care to minimize distortion and maintain adequate signal Error Vector Magnitude (EVM). Typical CFR will reduce the PAPR to around 8.0 to 8.5 dB.
Figure 2 Drain efficiency vs. power output and probability distribution of the instantaneous output power value.
Secondly, the variation with time of the PA output power results in poor overall power efficiency. The reason for this is shown in Figure 2. A Class AB (linear) PA is at its most efficient at peak power, but the drain (power conversion) efficiency, as shown by the solid line, drops off rapidly as the output power decreases. The probability distribution of instantaneous output power for a typical OFDM signal (dashed curve [not to a specific scale]) shows that for much of the time the signal power lies well below the peak power and hence the device is operating at low (average) efficiency. Note that the PAPR value shown in this diagram assumes that CFR has been used to reduce the PAPR of the transmitted signal; without this, overall efficiency would be even lower.
Improving PA Efficiency
A number of techniques are now being used to improve PA efficiency. The majority of these have found their first use in the cellular industry, where the problems of high network power consumption and environmental impact have already caused many network operators to force the pace of change and demand significantly improved equipment efficiency from their suppliers.
The three major techniques are:
- Digital Pre-Distortion (DPD) and Linearization
- Envelope Tracking
DPD and Linearization
As already noted, CFR can make a useful contribution to improving PA efficiency by allowing controlled compression of peak signals, effectively allowing the PA to operate nearer peak power and hence at a higher efficiency. DPD and Linearization techniques build on this by compensating for nonlinearities in the final RF output stage. In the process they also improve adjacent channel and EVM performance, and by allowing some compensation for the distortion caused by nonlinearities near compression, the PA can be driven harder, resulting in an improvement in power efficiency.
The best improvements result when DPD and Linearization are used as part of a system architecture incorporating active sampling of the output signal as part of a feedback loop. Only then can the system fully compensate for changes in amplifier characteristics with time, temperature and signal characteristics.
The Doherty PA configuration uses two amplifying devices driven in parallel, with their outputs combined. One amplifier (the ‘carrier’, typically a Class AB amplifier) provides all the output power (with the second device turned off) until the power required causes it to enter its nonlinear region, at which point the second (‘peaking’) amplifier (typically operating in Class C) is switched on and provides additional power. While several academic papers have quoted impressively high efficiency capabilities for Doherty amplifiers, in practice the typical efficiency being achieved with these designs is around 25 to 30 percent. However, this improved efficiency comes with a number of drawbacks, including the difficulty in maintaining matching and linearity over time and with temperature and device variations.
The most important limitation, however, is the limited PA bandwidth due to the complicated and essentially narrow-band matching required between the two amplifiers. Whilst the bandwidth available is adequate for cellular systems, it does not address all battlefield communications requirements.
Envelope Tracking, as a technique for improving power efficiency of RF power amplifiers, was first described by Bell Labs in 1937. Instead of optimizing a final RF stage power transistor supplied by constant voltage, the supply voltage to the power amplifier output transistor is adjusted dynamically, in synchronism with the envelope of the modulated RF signal passing through the device.
Figure 3 Envelope tracking (right) reduces the voltage difference between the supply voltage and the signal envelope significantly reducing the energy dissipated as heat.
This ensures that the output device remains in its most efficient operating region (i.e. in saturation), dramatically reducing the energy dissipated. Figure 3 shows envelope tracking in operation. Without envelope tracking, the difference between the constant power into the RF amplifier and the RF output waveform is dissipated in the RF power transistor as heat. With envelope tracking, the supply voltage tracks the signal envelope, dramatically reducing the energy dissipated.
Figure 4 Efficiency of a power amplifier driven by variable voltage.
Figure 4 demonstrates the high efficiency of an envelope tracking amplifier throughout the high-probability region of continuous output power. It is essentially a superposition of Figure 2, showing a series of drain efficiency versus RF power output curves as the supply voltage is varied. The locus of these curves represents the efficiency of a power amplifier driven by variable voltage.
Implementation of Envelope Tracking
Although the principles of envelope tracking have been known for some time, the practical difficulties of implementing a working system have prevented the concept from being employed until recently. The challenge is to make a power supply modulator capable of achieving the accuracy, bandwidth and noise specifications necessary at a level of conversion efficiency that delivers significant energy saving for the system as a whole. Critical performance issues include tracking accuracy, modulator efficiency, stability, compliance with spurious-signal and noise specifications, and bandwidth for multi-carrier support.
However, an evolution of this principle, High Accuracy Tracking (HAT™), is showing an improvement in efficiency, going from typically 30 percent for a standard Class AB amplifier to beyond 60 percent with HAT. Japanese cellular infrastructure vendor Sumitomo has recently launched a product based on the HAT principle, and multiple other base station and Digital TV transmitter manufacturers are at an advanced stage of adopting this new technology in their products. A solution for handsets is in development.
HAT technology has been demonstrated on a GaN PA. With QAM-based waveforms similar to SRW, they have shown a potential for 30 percent less power consumption and 42 percent more battery life for a manpack radio based on SRW. This was achieved over a wide frequency range and across multiple modulation modes. The power dissipation of the PA transistor itself is reduced by two-thirds, leading to a significant reduction in device thermal management requirements. Significant reduction in device temperature also leads to increased PA device reliability. Also, for handset applications, demonstrations have shown an improvement in linearity with HAT, eliminating the requirement for DPD altogether.
Figure 5 Application of a HAT power modulator to a standard power amplifier.
HAT implementation is relatively straightforward (see Figure 5), involving the addition of a HAT modulator module. The only addition required to the standard PA architecture is an output from the DPD/Linearization function to drive the HAT Power Modulator with a digital representation of the modulation envelope. The module can be a small box (70 mm × 70 mm × 18 mm for the commercial units supporting 40 W cellular base station transmitters, as shown in Figure 6), or can be a silicon chip for lower power handheld transmitters. In the future, the HAT algorithm may be integrated into power management chips already used in a radio transmitter. In addition, some minor redesign of the PA layout is needed to ensure optimal matching and hence efficiency.
Figure 6 Nujira HAT power modulator integrated with a typical PA.
To retain compliance with demanding noise and spurious specification, the power modulator tracks the RF signal envelope with utmost accuracy in both timing and amplitude. It does so by calculating the amplitude from the digital signal (√(I² + Q²)) and applying a simple function to arrive at the optimum instantaneous drain voltage. In parallel, a delay is calculated and applied to the RF signal before it is input to the amplifier, cancelling out the delay in the modulator.
Efficiency Improvement Across UHF Band
Figure 7 Level of energy efficiency improvement obtainable when using an envelope tracking HAT power modulator in a wideband UHF system.
Figure 7 shows the level of energy efficiency improvement that can be obtained when using an envelope tracking HAT power modulator in a wideband UHF system. It can be seen that the efficiency improvement exactly tracks the normal Class AB ‘fixed drain’ solution across the whole band, whilst maintaining strong linearity and providing some additional benefit in the form of increased power output from the transistor (due to thermal management improvement).
Note that the efficiency enhancement is maintained across a very wide band—a characteristic of envelope tracking. This need to provide adequate linearity with enhanced efficiency over a wide band with high peak to average signals is the type of challenge now facing designers for the next generation of networked battlefield communications systems. It has been shown that with envelope tracking, linearity is intrinsically improved compared to Class AB designs and the result can be achieved without the use of DPD and the added system complexity that this would bring.
To put the above discussion into context, the 40 percent of the 50 to 55 kg kit carried by the 21st century infantryman can be power related, and 30 percent of the load carried by a platoon can now be related to powering the communications and other electronics it carries. There is high level realization that the benefits of new communications standards need to be realized while reducing, rather than adding to this burden.
In consequence, western armed forces are giving focus to the power density of their systems, defined as watt hours per kilogram. The vision is to drive power density upwards from today’s 200 Wh/kg through 400 Wh/kg to a goal of 600 Wh/kg by 2011. Though DPD, linearization and Doherty can all make a contribution towards this target, HAT has demonstrated that it is capable of fully compensating for the inherent inefficiencies of transmitting OFDM, QAM and similar signals, and reversing the trend of rising RF transmitter energy use.
Shaun Cummins graduated from Nottingham Trent University in 1994 with a degree in Electrical and Electronic Engineering. He is presently the advanced technology director at Nujira looking at new architectures and systems where envelope tracking technology can be applied. Cummins is one of the earliest members of Nujira and has been active in a broad range of envelope tracking technology development. He has also been very active in the area of linearization technology, particularly Digital Pre-Distortion, as it applies to ET systems.