SATCOM GaN Power Amplifier Design

April 2019

S P O N S O R E D B Y

eBook

Table of Contents

2

3 Introduction Pat Hindle Microwave Journal, Editor

9 Norsat GaN and GaAS – Comparing the Technology Dr. Mehdi Ardavan, RF Antenna Systems Engineer, Norsat International Norsat

18 GaN SSPA Technology for Space-Based Applications Mario LaMarche Mercury Systems, San Jose, Calif.

Military Satellite Terminals RF Technology Trends 4 and Outlook Asif Anwar Strategy Analytics, London, UK

Designing A Broadband, Highly Efficient, GaN RF 12 Power Amplifier J. Brunning and R. Rayit SARAS Technology, Leeds, U.K.

SATCOM GaN Power Amplifier Design

According to Strategy Analytics, software defined radio architecture, radio-satellite-network integration as well as enabling un-interrupted and secured communications operations down to the tactical edge in a congested and contested spectrum environment will drive spending on military communications systems. They forecast spending on global military communications systems and services will grow to over $36.7 billion in 2026, representing a CAGR of 3.5%. Spending on global communications systems and services comprises radio, satellite communications (SATCOM), datalink, network and other communications systems. SATCOM systems (incorporating satellite payloads and satellite terminals) will account for 37.2% in 2026 representing the largest market opportunity in terms of military communications systems, and will be worth $13.7 billion.

The first article in this eBook is written by Strategy Analytics’ Asif Anwar covering the military satellite terminal basics, market trends and the outlook for this market. Typical system configurations and terminal setups are reviewed along with the major players in the industry.

With a major shift from GaAs to GaN as the semiconductor device material of choice, Norsat compares the two technologies covering the reasons for this shift and the advantages of GaN technology for SATCOM amplifiers. According to Norsat, GaN amplifiers can be up to 70% smaller than GaAs amplifiers.

Authors from SARAS Technology review the design of a broadband, highly efficient, GaN power amplifier. They describe their design flow that minimizes design uncertainties for first pass success and demonstrate this process with a 10 W GaN amplifier that achieves peak drain efficiency greater than 54 percent in the 2.0 to 2.5 GHz range.

Finally, Mercury Systems discusses GaN SSPA technology for space-based applications reviewing the challenges to device design and reliability as they start to replace TWTs in this area. They include the methods used to qualify GaN for these applications.

This eBook provides an overview of SATCOM systems and terminals, military SATCOM market outlook, the advantages of GaN-based amplifiers for SATCOM, GaN device design and an overview of space qualified GaN amplifiers. This eBook is sponsored by Norsat so thanks to them for supporting this effort.

3

Introduction

Pat Hindle, Microwave Journal Editor

WWW.MWJOURNAL.COM/ARTICLES/319204

Military Satellite Terminals RF Technology Trends and OutlookAsif AnwarStrategy Analytics, London, UK

Software defined radio architecture, radio-satellite-network integration as well as enabling un-inter-rupted and secured communications operations

down to the tactical edge in a congested and contested spectrum environment will drive spending on military communications systems. Strategy Analytics forecasts spending on global military communications systems and services will grow to over $36.7 billion in 2026, representing a CAGR of 3.5%. Spending on global communications systems and services comprises radio, satellite communications (SATCOM), datalink, network and other communications systems. Satellite commu-nications (SATCOM) systems (incorporating satellite payloads and satellite terminals) will account for 37.2% in 2026 representing the largest market opportunity in terms of military communications systems, and will be worth $13.7 billion. What are the implications for en-abling RF power technologies? Which companies are positioned to benefit?

INTRODUCTIONMilitary communications networks provide for the

exchange of voice, video and data between geographi-cally dispersed elements of a battle force. Trends driv-ing spending on the military communications sector will be underpinned by software defined radio, satellite

connectivity and network-centric IP-based communica-tions.

Satellite communications networks consist of user terminals, satellites and a ground network that provides control and interface functions. The benefit of a satellite communications network lies in its ability to link users to voice, video and data information where other forms of terrestrial networks may not be feasible. The advan-tages and disadvantages of this form of communication are highly dependent on the satellite and network con-figuration.

Typical networks are described as one-way or two-way. In a one-way network, the communication origi-nates at the remote transmitting terminal. It is received and re-transmitted from the satellite to the network operations center or hub. From here, the signal is re-transmitted to the satellite and then to the remote re-ceiving terminal. This type of communication is known as “double-hop” and the network topology is called a star. The hub provides the central routing and switch-ing functions of the network between the satellite and remote terminals, as well as command and control of the satellite itself.

A network can also be configured so the remote ter-minals bypass the hub and communicate directly with each other through the satellite. This is called two-way

5

communication and the satellite network is in a mesh configuration. In this case, the hub only provides moni-toring and control functions to the satellite.

A satellite will use an array of transponders to translate the uplink signal from the ground to a lower frequency for the downlink transmission back to the ground ter-minals. This frequency shifting technique reduces inter-ference and feedback and the resulting architecture, is known as“bent pipe” architecture, that also helps re-duce the cost and complexity of the satellite. Typically, the terminals, gateways and network operations center must be within the coverage of the same beam.

To meet ongoing requirements for global coverage coupled with a move towards higher bandwidth re-quirements has resulted in networks being operated at higher frequency underpinned by satellites that feature greater onboard functionality. Such satellites feature sophisticated on-board processors (OBP), switch matrix and phased array technology to place the routing intel-ligence in the satellite. This allows the ground terminals to communicate directly with one another rather than every transmission going through the network opera-tions center.

These single-hop mesh networks offer greater avail-able channel bandwidth and more efficient modula-tion schemes with onboard satellite processing and electronics allowing the networks to be comprised of many smaller spot beams to target usage areas more efficiently. These spot beams link in a mesh architecture where terminals in any two beams can communicate

with one another directly. This enables frequency re-use and dynamic allocation of bandwidth among beams.

Satellites communications use several standard no-menclatures to reference operating frequency. The common International Telecommunications Union (ITU) designation of satellites frequencies of operation are classified as UHF, SHF and EHF.• Ultra High Frequency (UHF) - 300 MHz - 3 GHz range• Super High Frequency (SHF) - 3 – 30 GHz• Extremely High Frequency (EHF) in the 30 – 300 GHz

The most commonly used frequencies for satellite communications can also be categorized using IEEE nomenclature. In addition to the C-, Ku- and Ka-band frequency bands, military satellite systems also make extensive use of X-band communications frequencies. While the nomenclature “C-band”, “X-Band”, “Ku-band” and “Ka-band” are used universally, the fre-quency range used depends on the area of the world and end usage with the actual frequency ranges often classified to make them more difficult to disrupt.

Military satellite communications have typically fo-cused on C-band and X-band operations, but as use of satellite terminals has increased, so these bands have become increasingly capacity constrained and increas-ingly expensive. Bandwidth demand over the years has increased driven in part by continued growth in intel-ligence requirements and the expansion of UAS plat-form usage to incorporate BLOS (beyond line of sight) operations. This has led to the use of systems operating at higher frequencies such as Ku-band and Ka-band.

The use of Ka-band for military satellite communi-cations in particular has received growing attention as the frequency band offers a number of benefits which include:• Higher upload and download data rates• Better spectral efficiencies• Less congestion in the spectrum band• Lower bandwidth costs for the user

Furthermore, antenna gain is proportional to area and frequency, so higher operating frequencies trans-late to smaller antennas for the same gain enabling smaller satellite terminal equipment and pushing the concept of COTM/SOTM (communications on the move/satellite on the move) into the hands of individual soldiers. The narrower beam width of a higher frequen-cy signal allows for more and narrower spot-beam op-eration which can be further enabled through the use of phased arrays.

s Fig. 1 Satellite Communication Network Topologies.

Signal

HubStar Topology

RemoteTerminals

SignalControl

Hub

Mesh Topology

RemoteTerminals

Double-Hop Mesh

OBPSatelite

BentPipe

s Fig. 2 IEEE Frequency Designations.

6

These trends will support the continued use of military satellite communications (milsatcoms) as key enablers in completing the C4ISR jigsaw and acting as critical nodes in the net-centric communications environment. This is reflected by the broad range of terminal solutions avail-able, targeting requirements across the land, air and na-val domains.

THE MILSATCOM TERMINAL SUPPLIER LANDSCAPE

The following graphic provides an illustrative snap-shot (and is by no means exhaustive) of some of the companies that are active in supplying military satellite terminals to the defense sector.

Land-based systems can be categorized based on size and mission (strategic vs. tactical) and incorporate fixed and transportable systems, mobile/vehicular- based systems and portable and dismounted form factors.• Elta Systems ELTA’s ELK-1895 is a full duplex light-

weight transportable tactical SATCOM terminal that is designed to be transported and operated by a single soldier.

• The Low Cost Terminal (LCT) is an industry-funded terminal which takes advantage of Northrop Grum-

man and Lockheed Martin system knowledge along-side Comtech TCS to provide Protected Communi-cations on the Move (P-COTM) and Protected SIPR/NIPR Access Point (P-SNAP) communications at the halt capabilities with a focus on supporting commu-nications over the AEHF (Advanced Extremely High Frequency) satellite network.Airborne military satellite terminals are used across a

broad range of platforms including combat aircraft, spe-cial mission platforms, helicopters and UAS.• Viasat offers a broad range of terminals designed to

cover the full complement of airborne missions. The Ku/Ka-band VMT-1220HM airborne satellite terminal is designed for COTM on C-130 platforms, and the company also offers terminals for light aircraft as well as helicopter platforms.

• Airbus offers the AirPatrol satcom terminal which is designed to be installed on fixed wing, rotary wing or UAS platforms and can be configured for operation across X, Ku or Ka frequency bandsMilitary satellite terminals in the naval domain span

the full spectrum of platforms both surface and subsea, including aircraft carriers, destroyers, corvette, frigate platforms as well as the emerging demand from USVs (unmanned surface vehicles).• Raytheon’s NMT (Navy Multiband Terminal) is ex-

pected to be installed in approximately 300 U.S. Navy ships, submarines and shore stations, replac-ing several existing SATCOM systems. NMT variants offer X-, Ka- and Q-band, enabling communications across current and legacy US military satellite com-munications networks including AEHF, Milstar, Ultra High Frequency Follow-on (UFO/E/EE), Interim Polar, Enhanced Polar System (EPS), Defense Satellite Com-munications System (DSCS) and Wideband Global SATCOM System (WGS).

• Indra offers the TNX-100 terminal which is designed to operate at X-band through through a range of sat-ellites including SPAINSAT, SKYNET, SYRACUSE, SY-CRAL and XTAR.

ANATOMY OF A SATELLITE TERMINALA satellite terminal can be broken down into a num-

ber of constituents, as shown by a representative break-down of the Norsat GLOBETrekker 2.0 Flyaway Satellite Terminal.

From a RF and microwave perspective there a num-ber of core components that enable connectivity in a satellite terminal.• A BUC (block upconverter) converts a band of fre-

quencies from a lower frequency to a higher frequen-cy and is used in the uplink.

• An RF power amplifier which provides the power am-plification of the signal.

• The LNB (low-noise block downconverter) is used in the receive path of a transmission, and typically com-bines a LNA (low-noise amplifier), as well as other components to that enable the received signal to be down converted for the modem.The above is an extreme oversimplification as dem-

onstrated by the block diagram shown in Figure 5, which

s Fig. 3 Milsatcom Terminal Supplier Landscape

s Fig. 4 Anatomy of a Satellite Terminal.

UniversalLNB

SSPA

SegmentedBoom Arm

Carbon FiberAntenna

High PerformanceFeed

SunlightReadableDisplay

Quad-PodLegs

Base UnitLinkControl™ Software

IATA CompliantPackaging

Auto Levelling Feature

7

shows some of the other components that make up a satellite terminal system.

Requirements for higher data rates with a focus on IP-centric communications that encompass video and data as well as voice will push demand for military sat-ellite terminal systems. Communicating securely faster over multiple channels and wider spectrum in an in-creasingly complex spectrum environment will under-pin the current and future trends for system design ar-chitectures which will dictate the underlying changes in component technology demand.

This will be especially true for the RF power amplifier where the choice between vacuum tube-based solu-tions based around klystrons and TWT (travelling wave tubes) based power amplifiers has been expanded over the years to encompass SSPAs (solid-state power ampli-fiers). This latter category has fragmented further with the maturation of GaN technology offering an alterna-tive to both vacuum tube and GaAs-based solutions.

MILITARY SATELLITE TERMINAL DEMAND AND OUTLOOK

Strategy Analytics forecasts that the global military satellite terminal market will grow from $4.3 billion in 2016 to reach $6.2 billion in 2026, a CAGR of 3.6%. The total number of satellite terminal shipments is forecast to grow at a CAGR of 3.7% through 2026 to reach 8,376 units from 5,829 units in 2016.

Land-based satellite terminals will continue to rep-resent the largest market both in dollar as well as ship-ment terms, with the segment forecast to account for 49.7% of the total satellite terminal communications spend and 77.0% of total shipments in 2026. The mar-ket for airborne satellite terminals is forecast to grow the fastest, from $1.0 billion in 2016 to $1.5 billion in 2026, at a CAGR of 4.0%. Shipborne military satellite terminal communications system demand will grow at a CAGR of 3.1% to be worth $1.6 billion in 2026.

While the traditional frequencies including C-band and X-band will remain a staple component of satellite

communications, bandwidth constraints and a push to-wards higher data rates focused on IP-centric communi-cations that encompass video and data as well as voice will push demand for military satellite terminal systems operating at higher Ku- and Ka-bands with the subse-quent market for systems operating at these frequen-cies forecast to grow at a CAGR of almost 5% through 2026. This will be coupled with the emergence of multi-band/wideband capable systems to enable true global roaming capabilities.

MILITARY SATELLITE TERMINAL SYSTEM COMPONENT TRENDS AND OUTLOOK

Communicating securely at faster rates over multiple channels and wider bandwidth in an increasingly com-plex spectrum environment will underpin the current and future trends for system design architectures which will dictate the underlying changes in component tech-nology demand. In general, these can be distilled into four key issues that are largely in common with the oth-er military and commercial technology sectors. These include dealing with a growing data tsunami, identify-ing the optimal technologies from the antenna to the baseband, bridging the energy gap to allow equipment to be SWaP-optimized and recognizing the increasing demands that will be placed on semiconductor perfor-mance as the emphasis shifts towards higher frequen-cies and broadband performance.• Communications are no longer confined to voice

transmission, but are focused on IP-centric delivery of data in a wide range of formats including video, imagery, messaging with the delivery of voice now also moving into the IP-based domain. This increas-ing use of data translates into ever larger chunks of bandwidth being consumed at faster and faster rates. As spectrum becomes an increasingly sparse resource, dealing with this data tsunami across mili-tary communications is tightening up the require-ments for better spectrum management. Optimizing spectrum use will require use of more complex mod-ulation while the use of AESA-based architectures will also migrate into satellite terminals longer term.At the RF front-end, satellite terminals will take in-

creasingly take advantage high power wideband RF technologies that can bridge the traditional gap that has existed between solid-state and vacuum tube tech-nologies. Similarly, the receive side will be underpinned by wideband receive capabilities, coupled with bring-

s Fig. 5 Qorvo Block Diagram of a Multi-Band VSATSource: Qorvo

PLL

VCODiplexer

FrequencyResponse

Antenna

Mixer

DriverAmp.

IF

Gain Block

PA

Upconverter

Gain Block

Mixer LNA

Downconverter

IF

s Fig. 6 Military Satellite Terminal Market Outlook.

7

6

5

4

3

2

1

0

$ B

illio

n

2016 2017 2018 2019

Land Airborne Shipborne

2020 2021 2022 2023 2024 2025 2026

Global Military Satellite Terminal Market by Domain

8

There are numerous other companies that are active-ly pursuing the benefits of GaN over GaAs-based solu-tions including Comtech Xicom, General Dynamics and Mission Microwave and are offering solutions for the military satellite terminal market.

The primary benefits of GaN technology can be dis-tilled down to several primary attributes:• Linearity• Power• Efficiency• Reliability• Size• Weight

Strategy Analytics forecasts demand for high power RF and related semiconductor components and tech-nologies through to the digital backend from the mili-tary satellite terminal market will grow at CAGR of 1.7% from $175 million in 2016 to reach $206 million in 2026, and the increasing use of solid-state technologies will translate to the penetration of GaN technology growing by over 500% through to 2026.

CONCLUSIONS AND IMPLICATIONSStrategy Analytics forecasts spending on global mili-

tary communications systems and services will grow to over $36.7 billion in 2026, representing a CAGR of 3.5%. Satellite communications will account for 37.2% of this opportunity. These trends will support the continued use of military satellite communications (milsatcoms) as key enablers in completing the C4ISR jigsaw and act-ing as critical nodes in the net-centric communications environment. This is reflected by the broad range of ter-minal solutions available, targeting requirements across the land, air and naval domains.

Communicating securely and faster over multiple channels and wider spectrum in an increasingly complex spectrum environment will underpin the current and fu-ture trends for system design architectures which will dictate the underlying changes in component technol-ogy demand. At the RF front-end, satellite terminals will take increasingly take advantage high power wideband RF technologies that can bridge the traditional gap that has existed between solid-state and vacuum tube tech-nologies and this is encapsulated in the attributes of-fered by GaN technology.

Strategy Analytics forecasts the penetration of GaN technology will grow by over 500% through to 2026. This represents a growing opportunity for GaN semicon-ductor technology suppliers including Northrop Grum-man, Qorvo, Wolfspeed, a Cree Company, UMS and Win Semiconductor. Companies that succeed in this market will need to combine the linearity, power and efficiency offered by GaN into MMIC-based solutions offered in cost effect packaging that enables the con-tinuing requirements for smaller sized and lower weight satellite terminals.n

ing the signal into the digital domain using faster ADCs and DACs to enable faster digital processing.

To meet these challenges, military satellite terminals will see increasing use of solid state technologies. These solutions were initially based around GaAs technology and offered potential advantages over vacuum tube based solutions in terms of cost and size with power outputs reaching 100 to 200 W. As GaN technology has matured, so SSPAs based on GaAs are being displaced by GaN solutions that can offer higher power and great-er linearity enabled in smaller form factors.

The caveat is that despite solid-state semiconductor technologies such as GaN seeing increasingly robust adoption, a continued emphasis on high power, long range communications, especially in the shipborne mili-tary satellite terminal market, as efficient performance above 200 W at X-, Ku- and Ka-band frequency bands remains a differentiator for vacuum tube based RF pow-er amplification. However, the performance attributes of GaN are now enabling solutions that offer power out-puts that start to approach TWTA levels. This will mean that while existing equipment repairs and retrofits will maintain a market for vacuum tube-based components, the emphasis moving forwards will shift towards solid state solutions

The trend towards solid state solution based offer-ings is being reflected in the offerings from major suppli-ers of BUCs and SSPAs supporting the satellite terminal market.• Advantech Wireless offers a complete line of GaN-

based SSPA, Solid State Power Block (SSPB) and block upconverters. At X-band, for example, the company’s SapphireBlu™ series of GaN-based prod-ucts offer power levels reaching up to 600 W.

• Communications & Power Industries (CPI) has offered solid state amplifiers for over four decades, and has been increasingly focusing solutions on GaN tech-nology to replace the company’s GaAs-based of-ferings. For example, the company’s C-band 100W GaN-based Model 4710H offers over 35% reduction in weight over the GaAs-based 7720H 100W C-band GaAs BUC, as well as offering reduction in volume and an increase in efficiency. The company has de-tailed even better performance metrics when com-paring Ku-band solutions whilst using GaN for Ka-band has yielded the Model B5KO which offers four times the output power of the previous generation of GaAs amplifier.

• Norsat is also developing solutions based on GaN with its ATOM series of Ku- and Ka-band series of products citing several benefits that include reduced production price, low conductance losses due to low resistance, quicker devices yielding fewer switching losses, lower power requirements and smaller de-vices. Furthermore, Norsat’s comparison of Ka-band GaAs-based versus GaN-based offerings, find GaN offering advantages in other parameters such as ACPR (spectral regrowth) and TTIM (two-tone inter-modulation).

9

Gallium nitride (GaN) technology has been gradually replacing the traditional gallium arsenide (GaAs) semiconductor device in the RF and microwave field for high power applications. The higher power density of GaN devices have made it possible to manufacture single transistors or MMICs with higher output powers leading to the elimination of power-combining and thus reducing the overall cost of a product. GaN devices also have higher power efficiency compared to the incumbent GaAs devices, making them an unbeatable rival when the overall power consumption is important. Moreover, GaN devices require higher voltage and lower current for biasing, leading to narrower traces on the printed circuit boards, reducing the PCB size. The higher thermal conductivity and tolerance of the GaN devices makes the power dissipation in equal power levels easier compared to GaAs devices. Also, the high electron velocity in their wide band gap make GaN devices more suitable for higher frequency applications as well. However, to achieve its best linearity performance, a GaN device must operate a few dB below its highest power, defeating the purpose of maximizing power. The nonlinear performance can be mitigated at the system level using a different biasing point which does not affect the output power significantly. On the other hand, there are ongoing attempts to improve the linearity at the device level too. Also, GaN devices have larger gain variation over temperature which can be mitigated using controllable attenuators.

G aN technology has been gradually replacing the traditional GaAs in the RF and microwave field, especially as manufacturers of FETs and MMICs,

to create higher power devices. Consequently, new BUCs and SSPAs are using GaN technology in satellite applications.

GaN devices have higher power density of 4 to 8 W/mm compared to GaAs devices which usually are in the range 0.5 to 1.5 W/mm.1,2 Usually the size of GaAs MMICs are restricted to a maximum 20 mm2 which makes the total potential power limited.6 The highest power GaAs MMICs at Ku-band are typically rated with a P1d of 4-5 W range while GaN MMICs can provide output powers as high as 80 or 100 W at Ku-band. GaAs FETs can be 30 W or higher.

GaN transistors and MMICs have higher power effi-ciency and can deliver the same output power and gain

Norsat GaN and GaAS – Comparing the Technology Dr. Mehdi Ardavan, RF Antenna Systems Engineer, Norsat International Norsat

s Fig. 1 Size comparison of GaAs versus GaN amplifier.

10

consuming less power than their GaAs counterparts. At Ka-band, power added efficiency of above 50% were measured with GaN HEMT technology and, at 10 GHz, the PAE increased to 67%3 which is significantly higher than the 38 to 44% range which a similar research had achieved for GaAs devices.7

GaN devices have a higher breakdown voltage lead-ing to bias voltages in the 22-50 V range while the GaAs devices are usually biased at 7-8 V or even lower result-ing in higher biasing currents for GaAs devices. The high voltage and low current bias conditions for GaN devices lead to less restrictions in PCB design. To carry the high-er currents required for GaAs devices the PCBA trace widths must be greater which occupies more space and if not designed well, may lead to a higher operating temperature for the PCB and its components, reducing the MTBF and lifetime.

The substrate used for GaN on SiC devices have a higher thermal conductivity which facilitates heat re-moval and cooling of the device. The thermal conduc-tivity of SiC is three times higher than that of Si (GaAs).4,5 Better thermal conductivity can lead to smaller and less expensive cooling solutions and to increased MTBF as the devices will run cooler.

GaN devices have higher electron mobility, and this allows them to have higher output powers at higher fre-quencies than GaAs devices. Although it is possible to build GaAs HEMT devices operating at higher frequen-cies, but their output power quickly drops after 10 GHz. At Ka-band the output power of GaAs device is usu-ally about 1 or 2 W or less, while the GaN technology can offer tens of watts at Ka-band.5 To build a 10 W amplifier at or above Ka-band, a designer may have to combine the power of several GaAs amplifiers, where a single GaN device is capable of providing the same output power.

Designing SSPAs and BUCs is simpler with GaN MMICs because of their greater output power and high gain. The highest power GaAs devices are low gain FETs and creating a SSPA or BUC with these devices requires three or four devices in cascade and creating higher power products will require power combining several FETs. Power-combining using microstrip increases the size of the PCBA board and is quite lossy. Power com-bining in waveguide, has much lower losses than mi-crostrip, but it still requires extra space and directly adds to the cost. Conversely, using high gain, high power GaN MMICs reduces size and cost (see Figure 1).

While GaN technology simplifies high power BUC design, it is not without its challenges. Gain stability over temperature is an important BUC specification and is typically about 3 dB p-p over -40 deg C to 60 deg C. While the gain of a GaAs device may vary only a few dB over this temperature range, a GaN device can easily have more than 10 dB variation per device. Each device in the RF chain adds to this variation. Designing for a large gain variation can be challenging. To control the gain variation, temperature-controlled attenuators are incorporated into the RF chain and GaAs MMICs are of-ten used in the lower power portions of the RF chain to limit the gain variation.

The 1-dB compression point is not a good metric to compare GaAs and GaN devices. The 1 dB compres-sion point is usually around 1 dB below saturation for GaAs devices while it could be 4 or even 10 dB below saturation for GaN devices. A better metric for com-parison is the usable power which is often defined as linear power. GaN devices offer high saturated output power, but the usable power defined as linear power is often much lower than the rated saturated power. It is seen in Figures 2 and 3 that the linear power of a 40 W (46 dBm) GaAs BUC is about 0.5 dB higher than the GaN counterpart where the current biasing of the main amplifier has already been optimized for better linearity. Whether it is related to gain compression or intermodula-tion products, the linear power of GaAs devices is closer to the saturated power than that of GaN devices.8 Linear power of 2-3 dB below saturated power can be achieved with GaN devices by optimizing the bias current.

s Fig. 2 Third order Intermodulation for 40 W GaN ATOM with tone spacing of 2.6 MHz. Plinear is 43.3 dBm.

s Fig. 3 Third order Intermodulation for 40 W GaAs ATOM with tone spacing of 2.6 MHz. Plinear is 43.8 dBm.

Despite the challenges of gain variation and linear power, the availability of high power GaN devices, means the cost of GaN technology BUC or SSPAs is sig-nificantly reduced compared to GaAs technology BUCs because, in a GaN-based product, less components are used, power-combining is either not required or done at a much smaller scale, the size of the PCB and housing is smaller, the overall weight of the product is smaller, and the heat dissipation system is smaller and cheaper. n

References1. RF Wireless World, Retrieved March 14, 2019 from http://www.rfwireless-

world.com/Terminology/Difference-between-GaN-and-GaAs-Power-Am-plifiers.html

2. Yi-Feng Wu, D. Kapolnek, J. P. Ibbetson, P. Parikh, B. P. Keller and U. K. Mishra, "Very-high power density AlGaN/GaN HEMTs," in IEEE Transac-tions on Electron Devices, vol. 48, no. 3, pp. 586-590, March 2001.

3. M. Kao, C. Lee, R. Hajji, P. Saunier and H. Tserng, "AlGaN/GaN HEMTs with PAE of 53% at 35 GHz for HPA and Multi-Function MMIC Applications," 2007 IEEE/MTT-S International Microwave Symposium, Honolulu, HI, 2007, pp. 627-629.

4. H Amano et al 2018 J. Phys. D: Appl. Phys. 51 1630015. Dasgupta, Avirup & Chauhan, Yogesh. (2018). Compact Modeling of Elec-

trostatics, Quasi-ballistic Transport and Noise in MOSFETs and Magnetic Tunnel Junctions.

6. A. K. Ezzeddine, "Advances in Microwave & Millimeter-wave Integrated Circuits," 2007 National Radio Science Conference, Cairo, 2007, pp. 1-8.

7. M. Kao, S. Nayak, R. Hajji, S. E. Hillyard and A. A. Ketterson, "High Per-formance Dual Recess 0.15-μm pHEMT for Multi-Function MMIC Applica-tions," 2006 IEEE Compound Semiconductor Integrated Circuit Sympo-sium, San Antonio, TX, 2006, pp. 129-132.

8. M. Ardavan, “A Comparison of GaN vs. GaAs System Performance”, Micro-wave Journal, July 2018 http://www.microwavejournal.com/articles/30705-a-comparison-of-gan-vs-gaas-system-performance

Norsat's ATOM series of Ku and Ka-band GaN and GaAs Block Upconverters are small, light,

and extremely efficient. With industry leading size, weight, power and MIL-STD specs, these BUCs can

be integrated into a variety of systems with the most stringent specifications and extreme form factors,

making them ideal for applications such as Airborne, RADAR, TWTA replacement, SOTM, COTP,

Portable VSAT, CDL and more. These BUCs are highly customizable , weather-proof and feature

excellent thermal management.

Learn More

Norsat ATOM Series

12

Demand for linear RFPAs covering the fre-quency range from 1.5 to 2.8 GHz is driving new design methods for broadband, linear and highly efficient RFPAs operating in out-

put back-off mode. Improving efficiency in PAs has long been a challenge for designers, in part due to poor control of harmonic load impedances. The dif-ficulty measuring waveforms at microwave frequen-cies makes it hard to determine if optimum wave-shaping has been achieved. Broadband design adds a challenge when a harmonic of a lower operating frequency lies in the operating band. These inherent difficulties can be compounded by imprecise design techniques, leading to multiple time-consuming and expensive iterations.

In this article, a design flow is described that uses NI AWR Design Environment platform, specifically Micro-wave Office circuit design software, as well as a mea-surement technique for determining the input and out-put impedances of the matching networks, prior to RFPA turn on. Several approaches to the problems inherent in PA design are presented with the aim of minimizing un-certainty and achieving first-pass success.

The effectiveness of this approach is demonstrated using a commercially available discrete 10 W GaN on SiC, packaged, high electron mobility transistor fabricat-

Designing A Broadband, Highly Efficient, GaN RF Power AmplifierJ. Brunning and R. RayitSARAS Technology, Leeds, U.K.

A design approach for a broadband, linear, efficient output back-off mode RF power amplifier (RFPA) emphasizes the importance of minimizing design uncertainties. Using this approach, excellent agreement between modeled and measured performance is achieved with a first- pass design.

ed with a 0.25 µm process (Qorvo’s T2G6000528) and a 20 mil RO4350B printed circuit board. The fabricated RFPA achieves a peak power greater than 40 dBm and a peak drain efficiency greater than 54 percent over its op-erating bandwidth. In back-off mode, the RFPA achieves an uncorrected linearity of 30 dBc and drain efficiency of 34 percent or higher when driven with a 2.5 MHz, 9.5 dB peak-to-average power ratio (PAPR) COFDM signal in the 2.0 to 2.5 GHz band.

RFPA DESIGN FLOW

Device SelectionThe first step begins with a thorough device/technol-

ogy selection process to determine the best candidate device to meet a specific set of criteria prior to the time-consuming tasks of load- and source-pull and network synthesis. Several candidates are acceptable on the ba-sis of claimed frequency and power. In addition to the more common characteristics such as Vds, gain, operat-ing frequency and power rating, other parameters such as Cds, Cgs and transformation ratio are considered.

Optimal Load Impedance ExtractionOnce a device is selected and a nonlinear model

obtained, optimal source and load impedances are determined. The required load impedances to achieve

WWW.MWJOURNAL.COM/ARTICLES/30456

13

maximum power, efficiency and gain—or an acceptable trade-off between these performance metrics—are fre-quency dependent and vary substantially over the oper-ating bandwidth of a broadband design.

To determine the correct load impedance, a com-bination of load-pull plotting at the fundamental and harmonic frequencies and waveform engineering (cir-cuit design techniques based on shaping the transis-tor voltage and current waveforms) are performed in Microwave Office. The use of waveform engineering re-lies on having access to the intrinsic device nodes across the current generator of the device plane, rather than at the package reference plane. Assuming the nonlinear device model provides these nodes, a waveform engi-neering approach enables the visual observation of volt-age and current swing, clipping and amplifier class of operation.

For this example, a load-pull simulation is run at Vds = +28 V and Idq = 90 mA across the operating band, and the impedances for optimal power and ef-ficiency are extracted, with the mid-band results shown in Figure 1. A target load region based on the overlap between Pmax ‐1 dB and drain efficiency max (effmax) ‐5 percent is defined. Clearly, the larger this target area is, the easier the matching problem becomes. In this case Pmax occurs on a tightly-packed clockwise rotating locus over the operating bandwidth, which is helpful in the case of a broadband amplifier. Load-pull is per-formed at the fundamental frequency due to the broad-band nature of the RFPA and consequent difficulties in achieving optimal harmonic terminations1 without using transmission zeros in the network.2 Load-pull at the sec-ond harmonic is also performed, with a region of high efficiency identified1 that can be controlled in the net-work synthesis.

Network SynthesisNarrowband RFPAs have the advantage of little varia-

tion of the optimal load impedance over their operating bandwidth, making the task of network design less com-plex. This is not to say that a low fractional bandwidth match is always trivial. Indeed, an investigation of source and load impedances will reveal that for very high per-formance, the network fundamental impedance must often be precisely controlled to a single gamma point, with significant sub-optimal performance penalties if the network locus misses its target load impedance. Precise control of harmonic termination impedances for F and F-1 amplifier classes increases the complexity of the task beyond what is required for an average PA design.

In the case of a broadband amplifier, particularly one with high performance specifications, the network is re-quired to control its impedance variation over a far larg-er fractional bandwidth. After defining optimal imped-ances and target areas, the load network is developed using a simplified real-frequency technique (SRFT)3 to design the ideal lumped-element network and convert it to a distributed stepped-impedance format,4 before performing electromagnetic (EM) simulation. In this ex-ample, EM simulation results agree closely with model predictions; however, for less conventional matching topologies, this might not be the case. In general, EM

s Fig. 1 Fundamental frequency load-pull analysis showing power (red) and efficiency (blue) contours over the operating bandwidth.

Zoptpwr

Zopteff

1.0

0.5

0.5 1.0 2.0

0.2

–0.2

–0.5

–1.0

–2.0

–5.0

5.0

2.0

s Fig. 2 Distributed load network loss and match (a) and transducer and operational power gain vs. frequency (b).

0

0.2

0.4

0.6

0.8

1.0

0

8

16

24

32

40

1400 1900 2400 2900

Frequency (MHz)

Inse

rtio

n Lo

ss (d

B) R

eturn Loss (dB

)

(a)

1.0

0.9

0.81400 1900 2400 2800

Frequency (MHz)(b)

Power Gain

Transducer Gain

Load

Net

wor

k E

f�ci

ency

14

ance seen by the device drain. Although the output is matched for compressed power and efficiency, not for minimum reflection at the drain, the use of a conjugate match is found to agree well with the predicted reduc-tion in compressed power due to imperfect realization of the target load impedance. Thus, the plotted trans-ducer gain is a good measure of the overall quality of the output match.

Achieving an optimal broadband match using this transistor is relatively straightforward for several reasons. First, the transformation ratio is relatively low over the operating bandwidth (about 2:1); second, the load im-pedance for optimal Pmax are tightly packed; finally, the optimal impedance varies with increasing frequency in a clockwise rotating locus. The fairly low transformation ra-tio is a useful criterion favoring the selection of this GaN device for a broadband RFPA application.

Source NetworkControl of source impedance variation over the op-

erating bandwidth is achieved through the use of a bandpass filter network, which also has the advantage of reducing low frequency gain, where the transistor’s inherent gain is very high. This particular source imped-ance matching network is also responsible for improv-ing the amplifier’s low frequency stability. The imped-ance transformation ratio of about 15:1 requires a more elaborate network. Although not used here, matching networks with a positive slope, or equalization, can be conveniently introduced into the source matching cir-cuit, as well.

Stability is achieved using a shunt connected series RC pair adjacent to the input port followed by a series R. Although this is a severe approach, analysis shows the transistor to be potentially unstable in the operating band, and some gain must be sacrificed to achieve un-conditional stability from 1 MHz to greater than 6 GHz, where the transistor ceases to have gain (Fmax).

Waveform EngineeringWaveform engineering5 is also used to analyze the

RFPA, using both the load-pull tuner and, more critically, the realized load network. Recent device models giving access to the voltage and current nodes at the intrinsic current generator plane allow accurate observation of both the V and I waveforms and the dynamic load line (DLL). This enables analysis of clipping and the RFPA mode of operation, as well as the peak voltages and cur-rents generated.

Prior to these nodes being available, the only option was to monitor waveforms at the package plane, which clearly has limitations due to package parasitics. Nega-tion of the parasitic network is feasible, but only if the topology and component values are known and their electrical impact removed through de-embedding dur-ing simulation. Although care has been taken to control the second harmonic load impedance, analysis of the waveforms (see Figure 3) shows that the third harmonic impedance is favorable without further optimization.

These waveforms show a peak voltage of less than 60 V and a peak current of less than 1500 mA at

simulation is seen as an important step in reducing un-certainty in the design flow.

One design technique is to represent the conjugate of the optimal impedance as that of a two-terminal gen-erator (port 1), after which the matching network design can be viewed as a problem of reducing the mismatch loss that exists between this complex-valued load and a 50 Ω termination over the amplifier’s operating band-width. This mismatch can, however, be evaluated at the 50 Ω side (port 2) of the network, as shown in Figure 2a. As a passive network, the output matching circuit has an operating power gain less than 1, equal to its efficiency determined only by internal dissipative loss. The neces-sarily smaller transducer gain is the product of this ef-ficiency with the effect of loss due to reflection at the input. These quantities are shown as percentage efficien-cies in Figure 2b. The efficiency of the load network is calculated to be 96.6 percent at 2800 MHz, close to the value calculated from return loss at the same frequency. For comparison, the operational power gain, which con-siders purely ohmic loss in the network, is calculated to have an efficiency of 97.7 percent. Although this does not directly include reflection losses, its value does de-pend on the termination impedances, as these affect the distribution of current and voltage within the network, hence the copper and dielectric losses, respectively.

Transducer gain is evaluated for a generator whose impedance is the conjugate of the target load imped-

s Fig. 3 DLL (a) and IV waveforms (b) at the intrinsic device nodes, with a 1500 MHz CW signal and 10 W output power.

1500

1300

1100

900

700

500

300

100

–1000 20 40 60

Intrinsic Drain Voltage (V)

Intr

insi

c D

rain

Cur

rent

(mA

)

(a)

Transition Region

A

B

60

50

40

30

20

10

00 0.5 1

Time (ns)(b)

1500

1250

1000

750

500

250

01.333

A

B

5.5 V 47.9 V

Intrinsic Drain C

urrent (mA

)Intr

insi

c D

rain

Vol

tag

e (V

)

0.44 ns 1.11 ns

15

source network (INMAT), load network (OUTMAT) and a copper center section to mount the device (see Figure 4). The device source was soldered down.

Passive MeasurementsPrior to complete assembly, the impedances of the

INMAT and OUTMAT circuits, as presented to the tran-sistor tabs, were measured to correlate the modeled and measured datasets. The measured data shows ex-cellent agreement with the modeled impedance from 1000 to 3000 MHz with no tuning (see Figure 5a). A measurement of the INMAT and OUTMAT circuits over a wider band from 20 MHz to 10 GHz still shows very good agreement between modeled and measured im-pedance (see Figures 5b and 5c). With the aid of the modular three-piece jig, impedances seen by the de-vice can be measured directly and accurately without us-ing mechanically awkward probes, which can introduce electrical parasitic—notably stray inductance—at the at-tachment point. The jig is not the production version of the amplifier but is an important step in the design flow, to eliminating uncertainties at each design stage.

Small-Signal Measurements Initial small-signal gain measurements used a drain

bias of Vds = +28 V and an Idq = 90 mA. Measured and modeled gain and impedance match are closely cor-

1500 MHz, which are well within device ratings. More significant, in terms of efficiency, is near-ideal class F operation, with the half-wave rectified current waveform exactly 180 degrees out of phase with the voltage wave-form and very little voltage/current overlap. Using a DLL analysis, three regions are defined: region A (Vmin and Imax), region B (Vmax and Imin) and the transition region. Over one period, the waveform remains in region A or B for 63.8 percent of the time, while in the transition region for only 36.2 percent of the period.

RFPA VALIDATIONTo validate the approach, the RFPA was fabricated on

Rogers 4350B 20 mil board (εr = 3.48). The circuit was mounted on a jig consisting of three pieces containing the

s Fig. 4 Fabricated RFPA.

s Fig. 5 5 Measured vs. modeled INMAT and OUTMAT impedances from 1000 to 3000 MHz (a); measured vs. modeled impedances from 20 MHz to 10 GHz for the INMAT (b) and OUTMAT (c) circuits.

1.0

0.5

0.5 1.0 2.0

0.2

–0.2

–0.5

–1.0–2.0

–5.0

5.0

2.0

INMAT

(a)

1.0

0.5

0

–0.5

–1.0

20 2020 6020 10,000

Rea

lIm

agin

ary

4020 8020

Frequency (MHz)

(b)

OUTMAT

1.0

0.5

0

–0.5

–1.0

20 2020 6020 10,000

Rea

lIm

agin

ary

0.5

0

–0.5

–1.04020 8020

(c)Frequency (MHz)

1.0

0.5

0

–0.5

–1.0

16

network synthesis, using an SRFT technique combined with analysis using mismatch loss and transducer power gain, provides a broadband match with relatively simple matching networks.n

To view the companion webinar to this design project visit: bit.ly/2IqFOqg

related (see Figure 6) with a small-signal gain greater than 16 dB and an input return loss greater than 7.5 dB over the operating band. The amplifier is stable when subjected to practical stability tests such as varying the drain rail voltage and using an external tuner to vary the source impedance seen by the device.

Large-Signal MeasurementsLarge-signal measurements used a drain bias of Vds =

+28 V and an Idq = 90 mA. A continuous wave signal source was fed to the amplifier through a driver amplifier. RF in-put and output power measurements were corrected for any compression in the driver. Power gain, drain efficiency and power delivered to the load were measured at 3 dB compression. The modeled results show a maximum P3dB of 41 dBm, maximum drain efficiency of 63.2 percent and a maximum gain of 16.4 dB. The measured results show a P3dB of 40.6 dBm, maximum drain efficiency of 59.1 per-cent and a maximum gain of 15.7 dB (see Figure 7). The RFPA delivers more than 10 W down to 1300 MHz and up to 2900 MHz, extending its range to a fractional bandwidth of 76.2 percent.

To evaluate efficiency in output back-off mode and in-termodulation sideband performance, a 2.5 MHz chan-nel bandwidth COFDM signal with 9.5 dB PAPR was used over the band from 2.0 to 2.5 GHz. As a single-ended am-plifier at 34.5 dBm output power, the average efficiency was 34 to 35.9 percent, with a linearity of 30 dBc measured ±1.25 MHz about the center frequency (see Figure 8). Similar results were obtained in the band from 1.805 to 1.88 GHz using a WCDMA test signal with PAPR = 7.8 dB.

A balanced version of the amplifier is under con-struction. Including imperfect hybrids, it is predicted to achieve +37 dBm with an average efficiency of approxi-mately 34 percent and a linearity of 30 dBc at ±1.25 MHz from the center frequency. Linearity could be im-proved using linearization techniques such as digital predistortion or envelope tracking. Achieving high effi-ciency at signal peaks enables operation at greater peak compression, so the amplifier can be operated at higher relative power over the whole dynamic range. Hence, efficiency and linearity are improved even on high PAPR signals.

CONCLUSIONAn approach for the design of broadband, linear and

highly efficient RFPAs minimizes uncertainty to achieve first-pass success. The design methodology comprises four stages: device selection using qualitative and quan-titative analysis, optimization of load and source imped-ance matching networks using load- and source-pull, passive network synthesis including EM verification and waveform engineering using intrinsic voltage and current nodes. Together, these techniques provide a proven sys-tematic approach to designing the entire RFPA.

A measurement technique for fabricated source and load networks, enabling comparison of modeled and measured impedances at the transistor tabs, has also been demonstrated using a three-piece jig. Passive

s Fig. 7 Modeled vs. measured large-signal CW power, gain and efficiency.

42

38

34

30

26

22

18

14

10

100

90

80

70

60

50

40

30

20

10

01500 1700 250023002100 2700

Frequency (MHz)

3 d

B G

ain

(dB

) and

Pow

er (d

Bm

)

Drain E

f�ciency (%)

Modeled Measured

Modeled Measured

1900

Modeled Measured

s Fig. 8 Single-ended amplifier intermodulation performance with a 2.5 MHz, 9.5 dB COFDM signal.

s Fig. 6 Modeled vs. measured small-signal gain and input return loss.

25

20

15

10

5

0

–5

–10

0

2.9

5.7

8.6

11

14

17

201400 1900 2400 2900

Frequency (MHz)

Line

ar G

ain

(dB

) Return Loss (d

B)

Modeled Measured

Modeled Measured

17

ACKNOWLEDGMENTSThe authors would like to thank Andy Wallace of AWR Group, NI and Qorvo/Modelithics for the device model.

References1. D. T. Wu, F. Mkadem and S. Boumaiza, “Design of a Broadband and Highly

Efficient 45 W GaN Power Amplifer via Simplified Real Frequency Tech-nique,” IEEE MTT-S International Microwave Symposium, May 2010, pp. 1091–1092.

2. R. A. Beltran, “Class F and Inverse Class F Power Amplifier Loading Net-works Design Based upon Transmission Zeros,” IEEE MTT-S International Microwave Symposium, June 2014.

3. P. L. D. Abrie, "Design of RF and Microwave Amplifiers and Oscillators, 1st edition," Artech House, 1999.

4. D. M. Pozar, "Microwave Engineering, 2nd edition," Wiley, 1998.5. S. C. Cripps, "RF Power Amplifiers for Wireless Communications, 2nd edi-

tion," Artech House, 2006.

Norsat’s latest 5G Interference Solution consists of a C-band interference suppression LNB and Band

Pass Filter. With the rollout of 5G infrastructure, new cellular base station signals will be powerful enough

to saturate sensitive C-band satellite receiving systems. Norsat’s advanced LNB with internal filtering

can mitigate 5G signals and is the perfect solution for broadcasters, media houses and other C-band

operators. For additional filtering, the band pass filter can be used to reject terrestrial interference within

the C-band. Please Contact us for additional custom solutions and spectrum consulting services.

Learn More

Norsat C-band 5G Interference Solution

18

Few technology applications are positioned to ben-efit from high-power RF GaN device insertion to the extent as space payloads. Costing roughly

$10,000 per pound of payload to launch a satellite into space, the benefit of small, lightweight hardware is ob-vious. The trend toward low Earth orbit (LEO) satellite constellations is increasing the pressure to develop cost-saving technologies. While GaN is well-positioned to deliver these benefits, its use is not without challenges. To maximize the mean time between failures (MTBF), the thermal conduction path away from the device must be carefully designed. As an added challenge, the lack of industry heritage using GaN in space requires thor-ough analysis and additional qualification testing.

GaN is a III-V direct bandgap semiconductor. Similar to GaAs, its high electron mobility makes it well-suited for RF/microwave applications. Compared to GaAs, the wider bandgap of GaN—3.4 vs. 1.4 eV for GaAs—en-ables operation at very high-power densities. Instead of using bulky combining networks to sum the power of many GaAs devices, a small number of GaN devices will efficiently produce high output power. As GaN technol-ogy continues to mature (see Figure 1), it is replacing some traveling wave tube (TWT) amplifiers, which have been the primary technology for satellite power ampli-fiers for years.

SATELLITE AMPLIFIER TECHNOLOGYAs with nearly all communications systems, satellite

transponders include transmit and receive modules. In the traditional architecture, the uplink signal is passed through a low noise amplifier to a frequency converter, then to the transmit module. Amplifying the signal to the required output level is typically the role of a TWT

GaN SSPA Technology for Space-Based ApplicationsMario LaMarcheMercury Systems, San Jose, Calif.

amplifier. While tube amplifiers produce high-power at Ka-Band, their large size and high-cost are challenging, especially evident with the new generation of LEO satel-lites. Since these satellites must be smaller and less ex-pensive than traditional satellites, relying on expensive and large TWT amplifiers is problematic.

TWTs amplify RF signals through the interaction between an electron beam and the RF signal. While this is an efficient method for generating high output power, TWT amplifiers are inherently complex assem-blies, requiring the mechanical integration of multiple,

WWW.MWJOURNAL.COM/ARTICLES/31995

s Fig. 1 Increasingly used in radar, EW and communications application, GaN power amplifiers offer size, weight and power benefits for satellites.

19

high-precision components. This complexity drives the high price of TWTs and increases the risk of failure. TWT amplifiers also require very high bias voltage—usually thousands of volts—generated by a high voltage sup-ply, which is also large and expensive. As a rough order-of-magnitude, the size of a Ka-Band TWT amplifier with 500 W output power is about 18 in. × 3 in. × 3 in., with an equally large power supply. Not only does the large size of the TWT amplifier restrict its use in LEO satellites, even traditional satellites have strict size and weight lim-its on their payload systems and will benefit from smaller components.

A solid-state solution offers a more robust, compact option. A GaN power amplifier uses standard IC manu-facturing processes, producing small devices only a few millimeters on each side. Instead of using artisan-style manufacturing, GaN devices are produced using auto-mated semiconductor processes at low-cost. While a single GaN device is unable to deliver the same output power as a TWT amplifier, multiple GaN devices can be combined in a small package. As an added benefit, GaN amplifiers only require bias voltages of 28 to 50 V.

Given the differences between GaN and TWT ampli-fiers, GaN is particularly attractive for applications sensi-tive to size, weight and cost, as well as those that require less transmit power. This precisely describes LEO satel-lites.

CHALLENGES OF GaN IN SPACEWhile GaN amplifiers offer compelling benefits for sat-

ellite applications, specific challenges must be overcome to successfully use GaN in space-qualified hardware. The first and most obvious challenge arises from the high power density of the device. While TWT amplifiers also require a complex cooling system, a GaN IC generates significant heat in a very small space. For example, a 30 W solid-state GaN amplifier can easily draw 2.5 A bi-ased at 28 V, resulting in 40 W power dissipation in an area not much larger than 10 mm2. If the thermal transfer is inadequate to cool the device, the elevated junction temperature will lower output power and reduce MTBF—possibly even causing catastrophic failure.

This raises the second challenge: reliability. The high-power dissipation common with GaN devices results in a significant temperature rise in the active region; as the temperature in the device increases, the reliability of the amplifier degrades. The temperature rise depends on the power dissipation in the GaN and the thermal resis-tance between the device and the case—both difficult to model and control. Power dissipation depends on multiple factors such as RF drive and load impedance, and the thermal resistance is highly dependent on minor variations in the assembly process.

Even under ideal circumstances, where the tem-perature is carefully controlled, high RF drive levels can cause permanent damage to the GaN lattice, resulting in degraded output power. Compared to GaAs, GaN is a much newer technology, and the lack of heritage raises reliability concerns. While this applies to all ap-plications using GaN, operating in a space environment requires an extra focus to assure reliability. Since repair is generally not an option, a single device failure can be

extremely expensive.The design of the GaN ICs and amplifier modules

is also a challenge, especially for space-based applica-tions requiring custom designs for specific programs, rather than using standard, off-the-shelf products. One critical element to first-pass design success is accurate device modeling. Since even class A amplifier design requires nonlinear models, modeling a GaN amplifier is considerably more complex than simply using an S-parameter file.

This discussion highlights several key challenges to implement GaN technology in space-qualified power amplifiers. Addressing these requires multi-disciplinary expertise including, RF design, mechanical design, man-ufacturing and quality. The following sections discuss possible approaches to managing these challenges.

GaN AMPLIFIER THERMAL MANAGEMENTThe high power density in GaN semiconductors pres-

ents a major thermal management challenge. Pulling the heat away from the active region of the device is critical to maximizing the output power and reliability. Starting with the bare die, proper thermal management requires an optimal die attach process. Since even a small increase in thermal resistance results in a significant temperature rise, use of a high thermal conductivity material for die attach is critical. For example, using a gold-tin eutectic die attach process provides much better thermal conduc-tivity than silver epoxy. However, achieving good die at-tach with high thermal conductivity requires more than simply choosing the correct material. The process must be carefully controlled. Since even small air voids under the die can greatly increase thermal resistance, they must be minimized, which requires experience, careful process control and techniques such as performing die attach in a vacuum. Validating die attach is also critical to ensur-ing proper heat transfer. This can be accomplished us-ing scanning acoustic microscopy (C-SAM), which identi-fies voids between the die and the thermal spreader or baseplate. Figure 2 shows typical C-SAM images of die attachment, comparing with good solder coverage and excessive voids.

The thermal conductivity of the baseplate mate-rial holding the die must also be maximized. For lower power applications, die is often installed on a Kovar™ baseplate, chosen because of its matched coefficient of thermal expansion (CTE). However, when thermal con-duction is critical, a material such as copper molybde-

s Fig. 2 C-SAM images showing GaN die attach with largely void-free solder coverage (a) and excessive voiding (b).

(a) (b)

20

num (CuMo) is a better choice. The process of choosing materials to optimize the thermal conductivity of each interface continues through the entire design, from de-vice to system packaging.

While this thermal design approach is used for GaN amplifier designs regardless of application, it is particu-larly important for space-qualified hardware. The size and weight constraints common to space programs in-crease power density by limiting the volume, while the reliability requirements for space operation require max-imum cooling of the active devices.

GaN RELIABILITYReliability is often characterized by the failure rate

versus time, which often looks like a “bathtub” and has been called the “bathtub curve” (see Figure 3). Typi-cally, the majority of failures occur early in the product’s life or after considerable use. Early failures are usually caused by a manufacturing defect, either during device fabrication or subsequent assembly. On the other side of the graph, the uptick in failures represents the de-vice wearing out near the end of its lifetime. To optimize and assure reliability, each of these failure types must be considered.

In the case of GaN, early failures are reduced through manufacturing process control, wafer screening and burn-in. Process control includes repeatable die attach, discussed above, and control of all aspects of the manu-facturing process. Clear documentation and operator training are critical, as well as environmental controls, such as reducing the risk for damage caused by electro-static discharge (ESD). With minimal performance varia-tion across a GaN wafer, sample testing can be used to qualify a wafer, improving confidence in the device’s reliability before committing the devices from a wafer to assembly. Depending on the program, wafer screen-ing may require accelerated life testing and destructive physical analysis. The risk of early failure can be reduced further through 100 percent burn-in screening. Using burn-in, amplifiers are biased and placed in an oven at elevated temperature for a specified time to stress the active devices. Burn-in screening will weed out the early failures, reducing the probability of failure in the field.

To quantify the length of operational time before wear-out failures occur, the MTBF of the amplifier is calculated. This calculation uses multiple factors such

as device temperature, bias and environmental condi-tions and is based on empirical data from accelerated life testing. To maximize the MTBF, the device cooling must be optimized, as described earlier.

While the radiation encountered in space creates an-other reliability risk, the high molecular bond strength of GaN results in a higher radiation tolerance than GaAs and silicon.

CUSTOM GaN AMPLIFIER DESIGNSince space missions usually require amplifiers with

unique frequencies, bandwidths, output power and reli-ability, they usually require custom designs rather than off-the-shelf products. Since the production volumes are typically low, the cost of the development must be mini-mized as well as the unit cost, to keep the total program cost low. One of the best ways to reduce design time and cost is to improve first-pass success, which requires accurate device modeling.

Nonlinear device modeling for power amplifier de-sign is always a challenge; since GaN is a newer technol-ogy, its models are less mature. To address this lack of accurate models, the design engineer has several op-tions:• Rely on measured load-pull data to determine the

ideal output impedance match to optimize the RF power.

• Through experience, adjust the simulation models to improve the accuracy for the specific design condi-tions.

• Substitute a measured, small-signal S-parameter file for the nonlinear model to confirm similar results for linear operation.

SUMMARYSatellite applications will clearly benefit from space-

qualified GaN power amplifiers, which offer size, weight and thermal benefits over TWT amplifiers. However, producing GaN power amplifiers is challenging, par-ticularly with the lack of space heritage. Successful in-sertion requires experience with both the electrical and mechanical aspects of GaN amplifier design, comple-menting space quality levels and requirements such as MIL-PRF-38534 class K. For organizations with both competencies, space-qualified GaN technology offers exciting new market opportunities.n

DecreasingFailureRate

Early”Infant Mortality” Failure

ConstantFailureRate

IncreasingFailureRate

Failu

re R

ate

Constant (Random)Failures

Observed FailureRate

Wear OutFailures

Time

s Fig. 3 Reliability bathtub curve. Source: Wikimedia.

http://www.norsat.com/atom-series/