Many microwave radar transmitters require active devices that can produce RF output power in the order of kilowatts to even megawatts. Routinely, microwave traveling-wave tube devices are utilized for this application. However, the currently used traveling-wave tubes are inefficient, large, expensive and have suspect reliability. While semiconductor-based amplifiers in principle can offer a more effective solution, semiconductor transistors have up until recently been limited in the DC voltage that can be applied to the device by the inherent critical breakdown field that the material can sustain. Since limited DC voltage can be applied, high-RF power operation requires large DC and RF current, which in turn requires large-area devices.1 High-current operation is inefficient due to series losses and the fact that large-area devices have inherently high capacitance and very low impedance, which limit operating frequency and bandwidth.1 GaN technology now offers a solution to this dilemma.
Solid-state amplifiers are already replacing traveling-wave-tube amplifiers (TWTA) for a variety of microwave power applications. However, the low operating voltages of Si and GaAs devices lead to a large device periphery, resulting in high device and circuit complexity and reducing production yield and reliability. Wide bandgap technologies like GaN can achieve a power density five times higher than that of conventional GaAs-based metal-semiconductor field effect transistors (MESFET) and heterojunction bipolar transistors (HBT). This will ultimately result in reduced circuit complexity, improved gain and efficiency and higher reliability. In particular, radar systems will benefit from the development of this technology.
GaN is the Future
The development of wide bandgap semiconductor materials, such as GaN and GaN-based alloys, offers the ability to fabricate RF active devices, specifically AlGaN/GaN power high electron mobility transistors (HEMT), with significantly improved output power performance.1 The improved RF output power is made possible due to the much improved material properties: high electric breakdown field, high saturated electron drift velocity, and when epitaxially grown on semi-insulating SiC substrates, improved thermal conductivity. The data in Table 12 allows a comparison of the most important performance metrics of Si, GaAs, SiC and GaN. The larger thermal conductivity of SiC and GaN enables lower temperature rise due to self-heating. The five to six times higher breakdown field of SiC and GaN is what gives those materials the advantage over Si and GaAs for RF power devices.2 SiC is a wide bandgap material (3.26eV), but suffers from poor electron transport properties, which hinders its use in very-high-frequency amplifiers. SiC has also been limited by expensive, small and low-quality substrate wafers.
Even though the mobility of the carriers is significantly better for GaAs devices, the higher peak and saturation velocity for GaN HEMTs compensates for its relatively lower mobility, enabling high-frequency performance. These advantages for GaN, with the high linearity and low noise of the HEMT architecture, open the application of these devices for high-power radar amplifiers.
An additional advantage of GaN HEMTs stems from the large conduction band energy offset between the GaN channel and the AlGaN barrier layer. This allows a significant increase in the channel carrier density in GaN-based HEMTs with respect to other materials (up to 1013cm–2 and above). This, along with higher voltage handling, increases power density. Power density is a very important metric for high-power devices because a higher power density yields a smaller die size and more easily realized input and output matches. The rapid progress in saturated CW RF power densities versus time for GaN FETs at X-band is shown in Figure 1.
The high operating voltages and high power densities that are possible with the wide-bandgap RF devices offer a number of advantages for power amplifier design, manufacture and assembly in comparison to Si LDMOS and GaAs MESFET technologies. GaN HEMT technology offers high power per unit channel width that translates into smaller, less expensive devices for the same output power, which are not only easier to fabricate but also increase the impedance of the devices. This enables lower complexity and lower cost impedance matching in amplifiers. The high voltage operation possible with GaN helps in eliminating the need for voltage converters, further reducing system cost.
The Path is Clear
The present day state-of-the-art output power vs. frequency for microwave solid-state devices and microwave tubes is shown in Figure 2.1 Historically, tube amplifiers, such as grid-controlled tubes, magnetrons, klystrons, TWTs and crossed field amplifiers (CFA), have been used as power amplifiers for radar transmitters. These amplifiers generate high power, but usually operate with a low duty cycle. The klystron amplifier offers higher power than the magnetron at microwave frequencies, and also allows the use of more complex waveforms. The TWT is similar to the klystron, but with wider bandwidth. CFAs are characterized by wide bandwidth, modest gain and compactness. Solid-state power amplifiers (SSPA) support long pulses and high-duty-cycle waveforms. Individual SSPA elements can be combined to produce sufficient amplification, despite the fact that individual SSPA elements have low power amplification. Silicon bipolar transistors, gallium arsenide MESFETs and GaAs PHEMTs are some of the solid-state elements used in SSPAs. GaN HEMTs, however, can be combined to create a SSPA with a significant improvement in mean output power, resulting in an improvement of the radar detection range.
As shown, solid-state transistors produce RF power levels less than about 100 watts at S-band frequencies and below, and their output continuously decreases with increasing frequency.1 The RF output power capability of GaAs FETs approaches 50 watts at S-band and at Ka-band is limited to approximately one watt.1 The GaAs FETs are limited in RF output power capability primarily due to the low drain bias breakdown voltage.1 Semiconductor devices fabricated from wide-bandgap material, such as GaN, offer significantly improved RF performance.
Over time, various figures of merit (FOM) for evaluating semiconductors for high power and high frequency potential have emerged. These FOMs for high performance RF semiconductors attempt to combine the most relevant material properties into a number that provides an estimation of the comparative strengths of the associated materials. The Johnson FOM (JFOM = ECR vsat/π) takes into account the breakdown electric field ECR and the saturated electron velocity vsat and corresponds approximately to the product of device cut-off frequency ft times the corresponding breakdown voltage. As can be observed from Figure 3,3 the Johnson FOM for GaN is larger by at least x15 than that for GaAs.
Aethercomm believes that if the trends in GaN advancement are maintained at their current rate, the predicted performance of GaN HEMTs in the year 2010 will be as depicted in Figure 4. As shown, GaN will soon overtake all of its competitors in every category.
Efficiency is the Key
Modern radar systems for military applications place new requirements on RF power amplifiers due to the desire to reduce system size, weight and cost. A major shift in amplifier specifications focuses more on amplifier efficiency to reduce DC power requirements and improve system reliability through lower component power dissipation. Furthermore, wide-bandgap, broadband, high-efficiency microwave devices will enable enhanced system capability.
The low parasitic capacitance and high breakdown voltage of GaN HEMTs makes them ideal for realizing the class-E and class-F high efficiency amplifier modes. Both modes have theoretical efficiencies of 100 percent. Recently, several GaN transistor vendors have implemented class-E amplifiers in hybrid form. Typical results are ten watts output power at L-band with efficiencies from 80 to 90 percent.
Aethercomm recently delivered a class-F high efficiency amplifier module operating at L-band to a major defense contractor. The desired output power was to exceed 50 W with an efficiency of 60 percent for the entire amplifier. Due to the tight delivery schedule, it was necessary to use off-the-shelf packaged transistors rather than developing a custom hybrid solution.
The power amplifier final stage was implemented using a balanced pair of packaged GaN HEMTs operating in class-F. Matching networks supplying the harmonic terminations necessary for class-F operation were designed by starting with an idealized model of the transistor. The parasitic capacitance and inductance of the transistor package was then added and the matching networks were modified appropriately to maintain the required harmonic terminations at the transistor die. The amplifier was then simulated using a nonlinear model of the transistor and the matching networks were modified to optimize efficiency and power.
A single-ended prototype of the class-F output stage was constructed. Drain efficiency of 75 percent, output power of 40 W and gain of 16 dB were obtained with only minor bench tuning. The results tracked the simulation closely. Low power GaN devices suitable for the driver stage were not available. The three-stage driver was designed using GaAs MESFETs, which were operated in class-A. Initially, it was believed that the driver stages would have to be operated in a high efficiency mode in order to achieve the required PAE; however, analysis indicated that with proper sizing of the transistors, class-A operation was permissible. The driver had 40 dB gain and 10 W power consumption.
The final configuration of the power amplifier exhibited a peak PAE of 63 percent at an output power of 75 W. At P2dB, the amplifier had an output power of 65 W and 61 percent PAE. Table 2 gives the characteristics of the power amplifier under different drive levels. Since the class-F final stage is biased at threshold, with no quiescent current, the amplifier has gain expansion at low drive levels. The amplifier gain peaks then compresses as the maximum output power is approached. Table 2 demonstrates the efficiency of this design.
Aethercomm has also tested a 200 W GaN HEMT device on SiC substrate designed to maximize power-added efficiency while maintaining a high output power for an operating frequency of 1215 to 1390 MHz and has observed efficiencies greater than 56 percent while maintaining output power levels in excess of 205 W at P3dB.
Many SSPAs intended for radar applications are designed with semiconductor RF devices configured for class-C operation. This form of bias yields very efficient operation for a single transistor stage; however, the class-C transistor has such low gain, typically 6 dB, that the efficiency advantage is lost due to the fact that many additional stages of gain are required to achieve the desired output power.
Conclusion
Future radar systems such as the active phased-array radar (APAR) will require increasingly smaller, more highly efficient SSPAs. The desire for extremely fast scanning rates, much higher range, the ability to track and engage a tremendous number of targets, low probability of intercept, and the ability to function as a radio/jammer will require an innovative and cost-effective transistor technology. Recent developments in the GaN HEMT have made it possible to realize highly efficient amplifiers at microwave frequencies. GaN HEMT devices provide a very high ratio of peak current to output capacitance, as well as an extremely high breakdown voltage and power density capability. This unique combination of characteristics allows designers freedom to achieve higher overall amplifier performance compared to competing devices.
References
1. R.J. Trew, “Wide Bandgap Transistor Amplifiers for Improved Performance Microwave Power and Radar Applications,” 15th International Conference on Microwaves, Radar and Wireless Communications, 2004 MIKON 2004, Vol. 1, May 17–19, 2004, pp. 18–23.
2. C.E. Weitzel, “RF Power Devices for Wireless Communications,” IEEE MTT-S Digest, Vol. 1, June 2–7, 2002, pp. 285–288.
3. T. Kikkawa, K. Imanishi, M. Kanamura and K. Joshin, “Recent Progress of Highly Reliable GaN-HEMT for Mass Production,” CS MANTECH Conference, April 24–27, 2006, pp. 171–174.
