Figure 1

Figure 1 HMC1099 saturated output power vs. frequency for various temperatures.

Improving amplifier efficiency is a key goal in many systems as PAs typically consume the largest amount of energy of any component. GaN has several advantages over other technologies such as high power density, high power, high gain and high efficiency.

In order to achieve maximum efficiency, manufacturers are using modulation techniques like envelope tracking or outphasing plus DPD to reduce distortion. Applying these techniques to GaN has taken high power amplifiers to the next level for many applications and promises to greatly reduce power consumption. We asked several leading GaN device manufacturers to provide examples of their highest efficiency GaN amplifier designs as we look to a greener future.

Analog Devices - Broadband and High Power A&D GaN
Norwood, Mass.

GaN is changing the RF and microwave landscape across communications systems ranging from mobile wireless networks to aerospace and defense, with most future radar, military communications and electronic warfare systems investigating its benefits. By reducing device parasitic elements, using shorter gate lengths and using higher operating voltages, GaN transistors have reached higher output power densities, wider bandwidth and improved DC to RF efficiencies. As the size of modern systems become increasingly important, GaN, with its improved power density, is up to 6´better than GaAs and provides a reduced system footprint with higher reliability.

One example of the benefit of leveraging GaN is in modern phased array radar systems. With thousands of active elements, GaN technology is providing comprehensive solutions for transmit-receive modules, enabling increased power density and integration with the PA, LNA and T/R switch all developed in GaN. The high breakdown voltage also potentially eliminates the need for the limiter — traditionally used to protect the LNA — reducing component count and area, which is critical at higher frequencies with minimal antenna aperture spacing.

Figure 2

Figure 2 HMC1099 power-added efficiency vs. frequency for various temperatures.

GaN’s higher operating impedance also enables optimized solutions for electronic surveillance and countermeasures, with a single broadband power amplifier now able to cover multi-octave bandwidths. For example, ADI’s HMC7149, based on a 0.25 micron process, is able to support amplification across 6 to 18 GHz with minimal variation of output power and gain. A single PA can now be used where previous PA performance necessitated band partitioning.

Advanced power combining solutions incorporated with GaN yield even stronger benefits to users. Using proprietary power combining methods and advanced bias control/monitoring circuits in core power combined modules, ADI’s SSPAs provide solutions from 100 W broadband to 8 kW at X-Band, combining up to 256 MMICs. These new system-optimized solutions for radar and electronic warfare have high efficiency and RF power density and provide alternative solutions to traditional TWTs.

Another example is a 10 W GaN amplifier which operates over an instantaneous bandwidth of 0.01 to 1.1 GHz with ± 0.5 dB of gain flatness (see Figure 1). It offers high saturated output power with 69 percent typical power-added efficiency (PAE) and 18.5 dB of small signal gain (see Figure 2). The device is designed for industrial and portable applications where GaN technology is needed to meet the increasing demands on battery lifetime. The device utilizes internal prematching to realize a single external matching network to provide high PAE over the full bandwidth and is packaged in a compact, low cost 5 mm × 5 mm QFN package.

After its acquisition of Hittite, ADI is taking aim at many high performance defense and aerospace applications. The company offers a broad portfolio of experience in everything from device to subsystem products and dedicated labs, production and testing facilities.

Figure 3

Figure 3 Static CW measurements for various values of input power and phase angle.

Ampleon – High Efficiency Cellular Base station GaN
Nijmegen, The Netherlands

In cellular base stations, the need to have higher throughput forces the use of higher modulation coding schemes, which leads to higher peak-to-average signals. As a consequence, the RF PA’s average efficiency is reduced. One candidate architecture suitable to minimize this issue is out-phasing, where highly efficient switched-mode PAs can be used. However, its pure version has some drawbacks, namely the nulling problem and response to DPD algorithms. A solution for these issues is to split the operation into two ranges: pure out-phasing and a linear mode in the highest and lowest input power levels, respectively. The nulling problem is solved by this method, as the zero output power is achieved when the input power is also zero and not at the expense of subtracting two high power signals with opposite phase.

Ampleon has developed an out-phasing PA prototype board that uses two IAF 0.25 mm GaN transistors. A quasi load insensitive (QLI) class-E topology is chosen for its high efficiency against load modulation. The higher harmonic impedances of the transistors were matched inside the commercial RF package, SOT1135, which eases the combiner design. Since the packaged devices are not very sensitive to higher harmonic load impedances, the combiner is easier to design. A standard 30 mil Rogers substrate, RO4350B, is used for the board design.

Figure 4

Figure 4 W-CDMA single carrier spectrum before linearization (a) and after linearization (b).

The static CW measurements for several values of input power and phase angle using a dedicated dual input measurement system are shown in Figure 3. Each line corresponds to an input power level and each dot to an out-phasing angle. It is evident that in the pure out-phasing regime, at the back-off levels, the efficiency degrades considerably. Additionally, it is interesting to see that it is possible to obtain the same output power level using different values of input power and phase angle. With this unique property, the best efficiency performance can be selected (especially in back-off) using the appropriate combination of input power and phase angle using a lookup table approach. The unique amplitude and phase points corresponding to the highest efficiency are chosen and lookup tables (phase, amplitude) are formed. This method allows not only the extension of the dynamic range (nulling issue), but also efficiency optimization.

The mixed mode, out-phasing PA was tested with a single
carrier W-CDMA signal seen in Figures 4a (non-linearized) and 4b (linearized). Before the linearization, the starting value of ACLR is equal to -28 dBc and the EVM 9 percent. After a single pass DPD cycle, it was possible to improve these values to -51 dBc and 1.1 percent, respectively. The achieved efficiency is high (>65 percent) and the PA responded to linearization algorithms, which showed the potential of the mixed mode, out-phasing concept for future high efficiency base stations.

Ampleon is the spinoff of the NXP RF power business, as NXP acquired Freescale and kept their RF power products. Ampleon is a major player in the wireless infrastructure power market, developing many unique designs to meet the next generation of RF power: the RF energy market for applications such as microwave ovens, lighting, automobile ignition systems and food processing.

MACOM – Breaking the Mold with High Efficiency GaN on Si
Lowell, Mass.

MACOM is the only RF GaN manufacturer that produces GaN on Si devices (the others use GaN on SiC). MACOM does produce devices on both substrates, putting significant focus on GaN on Si for low cost, high volume markets such as wireless infrastructure and RF energy. MACOM recently released a GaN wideband
D-mode transistor optimized for 1.8 to 2.2 GHz modulated signal operation in cellular base station applications and housed in an over molded plastic package. Using MACOM’s Gen4 GaN technology, the MAGB-101822-120B0P is one in a family of products in MACOM’s MAGB series. The series enables competitive products with state-of-the-art performance for LTE base station applications at LDMOS-like cost structures (at volume). These products have been optimized to deliver high drain efficiency and linear gain and are easy to linearize using digital predistortion (DPD) according to some customers.

These products target all the cellular bands within the 1.8 to 3.8 GHz range and deliver significant power efficiency improvements, in addition to package size reduction over legacy LDMOS. The products are housed in a plastic TO-272 package, operate over 400 MHz of bandwidth, and support 30, 40 and 60 W cellular infrastructure applications. The ability of this single device to cover the cellular bands and power levels from 1.8 to 2.2 GHz would require multiple LDMOS-based products.

The device delivers 160 W of peak output power on the load-pull system (fundamental tuning only), has linear gain of 20 dB and peak efficiency of 75 percent across the full band, similar performance to the ceramic version (see Figure 5). Peak efficiency can be improved to well above 80 percent when the device is presented with the proper harmonic terminations. A 2x MAGB-101822-120B0P symmetric Doherty amplifier optimized for Band 1 is capable of delivering 55 dBm of peak power. When the Doherty amplifier is measured with a two carrier 20 MHz LTE signal, a total of 40 MHz carrier, and at 7.5 dB back-off, the gain is 15.5 dB and efficiency is 55 percent. Using a commercially available DPD kit and without any special optimization, the adjacent channel power ratio (ACPR) can be easily corrected to less than -55 dBc. The Doherty amplifier achieves a video bandwidth of 200 MHz.

Figure 5

Figure 5 Output power, gain and efficiency vs. frequency.

The device enables the implementation of a simple symmetric Doherty amplifier design without compromising RF performance, compared to less performing and complex asymmetric Doherty topologies needed when using LDMOS based transistors. The symmetric Doherty amplifier, using the MAGB-101822-120B0P, is also easy to DPD linearize, which has been a challenge for users of other GaN-based products in the market. By overcoming all the RF and thermal challenges of designing GaN-based products in a plastic molded package, this product replaces high cost ceramic air cavity packages without compromising RF and thermal performance. MACOM’s improved package offering provides further system level cost savings to the customer and eliminates another barrier to full GaN adoption.

The MACOM Gen4 technology is enabling wireless carriers to deploy the latest LTE releases and significantly reduce system operating expenses at highly competitive price points, with a scalable supply chain combined with experienced applications and design support team. MACOM is qualifying this process for widescale production and also taking aim at the RF energy market.

Qorvo – High Performance GaN for Radar
Greensboro, N.C.

Airborne, land and naval radar platforms continue to push for higher PAE for the transmit PAs. Improving the PAE of the system can result in simplified thermal management; power supply requirement relaxation and longer Tx path pulse width operation. To meet the need for higher PAE in commercial and military radar applications at S-Band, Qorvo has developed a family of GaN PAs. All Qorvo S-Band GaN PAs have > 50 percent PAE. The latest and highest efficiency member of the S-Band family is the QPA1000, well suited to meet the needs of radar applications in the 2.8 to 3.2 GHz frequency band.

Figure 6a

Figure 6a PAE vs. frequency: Vd = 25 V; Idq = 200 mA; PW = 100 µs, Duty Cycle = 10%.

Figure 6b

Figure 6b Output power vs. frequency: Vd = 25 V; Idq = 200 mA; PW = 100 µs, Duty Cycle = 10%.

 

Figure 7

Figure 7 Large signal gain vs. input power: Vd = 25 V; Idq = 200 mA; PW = 100 µs, Duty Cycle = 10%.

It is designed using the Qorvo QGaN25 0.25 mm GaN on SiC production process to provide >58 percent PAE and >50 W saturated power pulsed with an input power of 25 dBm (see Figure 6). Measured PAE exceeds 60 percent in portions of the operating band with minimal changes to the saturated output power. The part is a two stage, near class B design with >22 dB large signal gain (see Figure 7). The PA is mounted in a 7 mm × 7 mm × 0.85 mm, 48 pin molded plastic QFN surface-mount package. The PA is designed to operate at a quiescent bias point of Vd = 25 V and Idq = 200 mA. For verification testing the pulse width and duty cycle are 100 µs and 10 percent, respectively, but it is capable of supporting a variety of operating conditions and pulse applications due to the good thermal properties.

The high PAE is achieved by optimizing the load-side impedance termination for both gain stages, and the transistor layout for both electrical and thermal performance, as well as compensating for the package parasitics. Measured load-pull data, package interface models, nonlinear transistor models, electromagnetic simulation and thermal modeling are all used to optimize the circuit. Package parasitics are determined through similar design testing, modeling and calculation.

Qorvo continues to push amplifier PAE as a critical parameter for radar applications. Operating class choice, form factor, load-pull data, model accuracy and thermal considerations all play an important role in achieving optimal overall amplifier performance.

Figure 8

Figure 8 Ouput power and PAE vs. input power: Vd = 50 V; Idq = 500 mA; frequency 1.55 to 1.6 GHz.

Sumitomo Electric – High Efficiency GaN for Space Applications
Osaka, Japan

Sumitomo Electric has recently qualified the SGN15H150IV GaN device at high reliability levels for space applications. This third generation GaN device is allowing an advantage over previous technologies in gain and power levels with excellent reliability. It has 150 W (51.5 dBm) typical output power and 117 W (50.7 dBm) minimum power output at 4 dB gain compression when operated at a drain voltage of 50 V. That is the highest power level of space qualified GaN device available at this frequency, covering 1.55 to 1.6 GHz according to the company. High impedance levels allow for simple external matching circuits to meet all performance goals.  A high linear gain of 17.5 dB minimum, 18.8 dB typical, makes driving the device easier. It has excellent PAE of 67 percent minimum (71 percent typical has been measured in an application engineering breadboard). Figure 8 shows output power and PAE vs.

input power.

The device has excellent performance over temperature with the classic HEMT two-slope gain shape. The inflection point is below -20ºC, making gain constant from -40º to -10ºC, and a slope of -0.012 dB/ºC above that point. The same phenomenon makes the power constant from -40º to 25ºC. The power has the same slope as the gain of -0.012 dB/ºC from 25º to 85ºC. PAE is approximately linear, with a slope of -0.068 percent per ºC.

At 150 W output power, thermal considerations are very important. The thermal characteristics have been optimized in the device and housing structure, resulting in very low thermal resistance of 0.6 ºC/W typical and 0.7 ºC/W maximum. Each device is measured for thermal resistance individually and that data is delivered with the device. The maximum junction temperature is 250ºC. This high maximum junction temperature allows for a conservative de-rating to 160ºC operating junction temperature for safe and reliable operation.

The package is the same as the previous generation of GaAs high reliability designs, and the pre-match offers similar impedance levels as previous lower power devices. This eases upgrading current designs to take advantage of the higher output power.

For those who are attempting even higher PAE, low Cds and excellent performance at lower voltages make this device is a good candidate for envelope tracking schemes. Low on resistance along with high reverse breakdown voltage (BVgd) also makes it attractive for switch-mode designs. Nonlinear models are available as well, for more precise first-pass design and evaluation of nonlinear behavior before building the device into a design.

Primary applications would be in navigation satellites, GPS L1 Band or Galileo E1 Band. Sumitomo continues to focus on high performance satellite, radar and space applications for its GaN products.

Wolfspeed, a Cree Company Energy Efficient GaN for Base Stations
Durham, N.C.

Wolfspeed is enabling designers to invent wireless systems for a responsible, energy-efficient future. Eta Devices, an MIT spin-out based in Cambridge, Mass., is pioneering the use of supply modulation through its technology called ETAdvanced. ETAdvanced enables base station transmitters to achieve the highest efficiency at peak power and the highest efficiency at back-off of any known technique according to the companies. And, unlike previous technologies using supply modulation, ETAdvanced has proven to retain its advantages for both high bandwidth (multi-carrier LTE) and high linearity (MC-GSM) applications. When combined with Wolfspeed’s GaN devices, the result has been reported to be the highest efficiency base station amplifiers.

Figure 9

Figure 9 ETAdvanced uses discrete supply levels to modulate the drain of a PA, making millions of transitions per second to optimize power consumption.

ETAdvanced solves the key power challenge facing the wireless industry today. As shown in Figure 9, ETAdvanced uses discrete supply levels to modulate the drain of a PA, making millions of transitions per second in order to optimize power consumption. At almost every place in the wireless ecosystem, the power consumption and the heat dissipation of the PA are limiting factors that increase cost and size, limit output power and range, and overtax energy supply resources.

In February 2016 at Mobile World Congress, Eta Devices demonstrated what could be the highest efficiency GaN high power amplifier compliant to MC-GSM/LTE specifications. The linearity specification for MC-GSM is that distortion products must be lower than -60 dBc. According to Eta Devices, Cree GaN HEMTs can be used over a very wide range of drain bias voltages. They exploit this to achieve not only high efficiency at maximum average power, but also achieve high efficiency at back-off. Their demonstration uses Cree’s high performance CGH40025F with high efficiency, high reliability and consistent performance.

Figure 10

Figure 10 Eta Devices’ demonstration of LTE and MC-GSM performance using ETAdvanced to modulate PAs built around the CGH40025F with 70% average final stage efficiency.

Figure 10 shows measurements from Eta Devices’ MWC 2016 demonstration. Eta Devices was able to demonstrate 70 percent average final stage efficiency for a fully modulated LTE carrier and >60 percent average final stage efficiency for MC-GSM transmissions. These efficiency numbers include the loss of the supply modulator.

Wolfspeed and Eta Devices have teamed together to design highly efficient amplifiers for base station applications that promise to greatly reduce power consumption, saving wireless communications providers millions in Op Ex for a greener future.

Conclusion

It was not long ago that 50 percent efficient power amplifiers were the best in the industry, but now, many manufacturers are obtaining 65 to 70 percent efficiency. GaN and advanced modulation and matching techniques are pushing efficiencies higher every year.