SlideShare a Scribd company logo
1 of 108
Download to read offline
In partial fulfilment of the requirements for the degree of
MASTER OF SCIENCE
in MICRO-ELECTRONICS
at the Delft University of Technology,
to be defended publicly on Wednesday September 24, 2014 at 15:00hrs
Supervisors : 1) Prof. Ir. Dr. C.J.M Chris Verhoeven,
Associate Professor,
TU Delft
2) Ir. Waldemar Lubbers,
RF & E head,
Innovative Solutions In Space B.V, Delft
Thesis Committee : 1) Prof. Ir. Dr. C.J.M Chris Verhoeven (TU Delft)
2) Ir. Waldemar Lubbers (ISIS BV)
3) Prof. Dr. Nick van der Meijs (TU Delft)
4) Ir. P.P.(Prem) Sundaramoorthy (TU Delft)
An electronic version of this thesis is available at http://repository.tudelft.nl/
The work in this thesis was supported by ISISpace B.V. Their cooperation is hereby gratefully
acknowledged.
Copyright c
All rights reserved.
[Dedication]
I dedicate this work to my mom, dad, vikku and my “Small Wall friends”….
Acknowledgement
I would like to express my sincere gratitude to the management of Innovative Solutions
In Space B.V for allowing me to work on such a fascinating research topic for my Master’s
thesis. They have provided me with some of the best possible resources to carry out my work
smoothly.
I would like to thank my university supervisor Prof. Chris and company supervisor Waldemar
for providing me the support, guidance and encouragement through the entire period of my
thesis. I am sure their guidance will go a long way in my career as a researcher.
The RF&E team at ISISpace B.V provided a great environment to exchange ideas. I
would like to thank Arancha, Eelco, Johan, Javier, Stefano, Wouter2
and Waldemar for making
me feel a part of the RF&E team and provide guidance and support and some very critical times
of my thesis.
I would like to express my gratitude to Ernst Habekotte from CATENA
microelectronics B.V for supporting this thesis at a crucial time by providing the LINC
development boards to understand the design and introducing me to the Mixed signal group at
TU Eindhoeven. I am grateful Prof. Baltus, P.G.M and Rainier from the mixed signal group at
TU Eindhoeven for allowing me to use the measurement setup there and providing an un-
conditional support through my entire stay.
I would like to thank my parent back in India who always had the confidence in me, no
matter what and always would be there for me as an emotional support. A very special thanks
to SKYPE and Facebook, I have not missed my home country!
My twin brother has played a very important part during my studies, he has been my
pillar of support ever since I came to Netherlands. I have almost never missed home.
It will not be complete without thanking my friends here in Delft. I surely did make a
lots of friends, they have supported me through the entire two years of my studies. A special
thanks to the “Small Wall” group of friends: Harshitha, Abhimanyu, Adithya, Arun, Arul,
Manjunath, Phani, Sriram and Sumedh. You guys were simply amazing. I would like to thank
Arun for helping me out in making the cover page of this thesis.
-Vishu
Preface
The fast growing interest in nano-satellite development and the use of nano-satellite for
complex missions has led to an increase in the downlink data-rate. The availability of free
bands (amateur frequency bands: VHF and UHF) is getting limited and more challenging to do
the frequency coordination. This can be attributed to the surge in number of nano-satellite
being launched. There has been 430% increase in the number of satellites launched in 2013
compared to 2012. This trend has led the nano-satellite developers to start looking into higher
frequency bands and S-band has been a popular choice. But, going by the trend, frequency
coordination in the S-band can also get challenging. Thus, there is a need to develop a nano-
satellite transmitter that uses spectrally efficient modulation scheme, at the same time it has to
be power efficient and be compatible with some of the existing communication standards.
This work deals with developing the requirements on a nano-satellite transmitter based on
mission case study, look at the need for efficiency and linearity enhancement technique for the
transmitter, select the most appropriate architecture for nano-satellite application and show a
proof of concept using a prototype.
Using system engineering approach the efficiency and linearity enhancement technique
that was chosen was LINC architecture. ADS simulations were performed to understand
different configurations of LINC and finally, measurements were performed on the chosen
configuration to characterize its efficiency and linearity. 16-QAM, 16-APSK, 32-QAM, 32-
APSK and 64-QAM modulation schemes were implemented and tested. The best EVM was
obtained for 16-APSK with 20 degrees compensation stub in the Chireix combiner which was
0.27% and the best efficiency was obtained for 16-APSK modulation scheme which was
implemented without any compensation stubs.
Based on the lessons learnt during simulations and measurements, recommendations
are provided to improve the LINC configuration and improve the reliability of the measurement
setup.
Visweswaran Karunanithi
Delft, September 2014
11
MSc Thesis, Visweswaran Karunanithi
Contents
[Dedication]...............................................................................................................................5
Acknowledgement.......................................................................................................................7
Preface.......................................................................................................................................9
Table of figures........................................................................................................................15
1. Introduction......................................................................................................................20
1.1. Current trends in nano-satellites......................................................................................21
1.2. Conclusion of the analysis................................................................................................24
1.3. Problem statement.............................................................................................................25
2. Nano-Satellite transmitters for the future........................................................................28
2.1. CCSDS and ECSS recommendations................................................................................28
2.1.1. Frequency band allocation & constraints ...........................................................29
2.1.2. Transmitter spurious emission and harmonic levels............................................31
2.1.3. Recommended modulation schemes.....................................................................31
2.1.4. Conclusion of the CCSDS and ECSS recommendations......................................32
2.2. Nano-satellite mission case study.....................................................................................33
2.2.1. Spacecraft housekeeping data downlink requirement .........................................34
2.2.2. Payload downlink data-rate requirement for a EES service ...............................34
2.2.3. Downlink data-rate requirement for nano-satellite providing data services ......36
2.3. Link budget analysis.....................................................................................................38
2.4. Summary of the case study and conclusion..................................................................40
3. High efficiency transmitter architectures ........................................................................43
3.1. Overview...........................................................................................................................43
3.1.1. Characterization of linearity and efficiency ..............................................................45
3.2. Doherty architecture....................................................................................................47
3.2.1. Working Principle................................................................................................47
3.2.2. Advantages and dis-advantages of Doherty architecture....................................49
3.3. Kahn/EER (Envelope Elimination and Restoration) ...................................................49
3.3.1. Working principle of Kahn/EER architecture......................................................49
3.3.2. Advantages and dis-advantages of Kahn architecture ........................................50
3.4. Envelope Tracking (ET) architecture ..........................................................................50
3.4.1. Working principle of ET.......................................................................................50
3.4.2. Advantages and disadvantages of ET architecture..............................................51
12
MSc Thesis, Visweswaran Karunanithi
3.5. Switched Capacitor Digital Power Amplifiers SCDPA...............................................51
3.5.1. Working principle of SCDPAs .............................................................................51
3.5.2. Advantages and dis-advantages of SCDPAs........................................................52
3.6. LINC (Linear amplification using non-linear components) architecture.........................53
3.6.1. Working principle ......................................................................................................53
3.6.2. Advantages and dis-advantages of LINC architecture..............................................54
3.7. Trade-off analysis .............................................................................................................55
3.8. Conclusion ........................................................................................................................56
4. LINC Architecture............................................................................................................57
4.1. Class of operation.............................................................................................................57
4.1.1. Class-F implementation.............................................................................................58
4.2. Different power combining techniques........................................................................62
4.3. Experimental setup............................................................................................................72
4.3.1. Single RF source measurement setup ........................................................................72
4.3.2. Modifications done to the original setup ...................................................................76
4.4. Catena LINC/Chireix PA architecture..............................................................................77
5. Measurement Results.......................................................................................................81
5.1. Modulation schemes tested...............................................................................................81
5.2. Performance characterization: EVM................................................................................82
5.2.1. 16-QAM modulation scheme................................................................................82
5.2.2. 16-APSK modulation scheme...............................................................................84
5.2.3. 32-QAM modulation scheme................................................................................86
5.2.4. 32-APSK modulation scheme...............................................................................88
5.2.5. 64-QAM modulation scheme................................................................................89
5.2.6. Summary of EVM and ACLP measurement. ........................................................90
5.3. Performance characterization: LINC/ Chireix efficiency.................................................91
6. Conclusion and Recommendations for future work.........................................................93
6.1 Recommendations on communication standards...............................................................93
6.2 Recommendations for a LINC architecture.......................................................................94
6.2.1. PA cell..................................................................................................................94
6.2.2. Combiner architecture.........................................................................................95
6.3 Recommendations for the experimental setup ...................................................................95
6.4 Future work........................................................................................................................96
Nomenclature...........................................................................................................................99
13
MSc Thesis, Visweswaran Karunanithi
Bibliography ..........................................................................................................................101
Table of figures
Figure 1. Number of nano-satellite launches..........................................................................21
Figure 2. (Left) Mission types between 2003 and 2012, (right) Mission types in 2013. ........22
Figure 3. (Left) data-rates used by nano-satellitemission between 2003 and 2012 (right)
data-rates used by nano-satellite missions in 2013.................................................................22
Figure 4. (Left) popular frequency bands used between 2003 and 2012. (Right) Frequency
bands used by nano-satellite missions in 2013........................................................................23
Figure 5. Maximum allowable bandwidth vs symbol rate......................................................30
Figure 6. Spectral emission limits...........................................................................................31
Figure 7. Pictorial representationof the relationbetween downlink data-rate, service
provided and type of payload...................................................................................................33
Figure 8. Pictorial representation of VDES [].......................................................................36
Figure 9. Link budget calculation of spacecraft housekeeping data downlink. .....................38
Figure 10. Link margin for different elevation angles.............................................................38
Figure 11. Link budget for an optical payload data downlink. ..............................................39
Figure 12. Link margin vs elevation angles for optical payload downlink. ............................39
Figure 13. Link budget for VED-Sat downlink.......................................................................40
Figure 14. Link margin vs elevation for VDE-Sat..................................................................40
Figure 15. Block-diagram of a transmitter with digital modulation. .....................................43
Figure 16. Linearity vs Efficiency trade-off for conventional PA classes..............................44
Figure 17. Efficiency and linearity enhancement techniques.................................................45
Figure 18. PDF of 16-APSK modulation scheme, implemented using an SRRC filter with a
roll-off of 0.5(4000 sps) ...........................................................................................................46
Figure 19. Doherty PA architecture........................................................................................47
Figure 20. Schematic representation of Doherty.....................................................................47
Figure 21. Input power vs Output power in linear scale [11]................................................48
Figure 22. Block diagram representationof a modern implementationof the Kahn
architecture..............................................................................................................................50
Figure 23. Block diagram of ET architecture.[11].................................................................51
Figure 24. Block diagram representation of SCDPA. ............................................................52
Figure 25. LINC block diagram..............................................................................................53
16
MSc Thesis, Visweswaran Karunanithi
Figure 26. 16-APSK decomposition into constant envelope signal........................................54
Figure 27. Output spectrum of 16-QAM for different classes of operation. [21]...................58
Figure 28. Functional representation of Class-F PA.............................................................59
Figure 29. Class-F harmonic matching circuit. .....................................................................60
Figure 30. Load impedance by the drain of the transistor at different harmonic frequencies.
..................................................................................................................................................60
Figure 31. IV characteristics of CGH27015F. .......................................................................61
Figure 32. Drain voltage and current waveform for a Class-F PA implemented using
CGH27015F GaN HEMT. .......................................................................................................61
Figure 33. RF power combining techniques for LINC. ..........................................................62
Figure 34. WPC as a power combining network for LINC ....................................................63
Figure 35. Plot of the two out-of-phase signals,combiner voltage waveform and efficiency of
WPC as a function of outphasing angle...................................................................................63
Figure 36. Improvement of isolationby introducing a quadrature coupler between the PA
and WPC. .................................................................................................................................64
Figure 37. Coupling between the two ports............................................................................64
Figure 38. Efficiency and voltage waveforms when a quadrature coupler is introduced
between the PA and combiner..................................................................................................65
Figure 39. Schematic representation of Chireix combiner.....................................................65
Figure 40. Chireix combiner as seen at the inputs when fed out-of-phase signals. ...............66
Figure 41. Chireix combiner design on FR4 using CPW. ......................................................67
Figure 42. S-parameter simulation of the Chireix combiner with compensation stubs placed
at 10 deg...................................................................................................................................67
Figure 43. Efficiency vs out-phasingangle comparison between Chireix combiner and WPC.
..................................................................................................................................................68
Figure 44. Comparison of coupling between Chireix combiner and WPC. ............................68
Figure 45. Radiation pattern of crossed dipole on a 3U cube-sat fed with an out-phasing
angle of 45 degrees..................................................................................................................69
Figure 46. Radiation pattern of the crossed dipole when fed with signals with 0 degree
outphasing angles. ...................................................................................................................70
Figure 47. Radiation pattern of crossed dipole fed with signals with 90 degrees out-phasing
angle.........................................................................................................................................70
Figure 48. Complete measurement setup................................................................................72
17
MSc Thesis, Visweswaran Karunanithi
Figure 49. Signal flow diagram..............................................................................................73
Figure 50. Initial setup for calibration...................................................................................74
Figure 51. Screenshot of calibration measurement.................................................................74
Figure 52. Setup to measure the non-linearity caused by mini-circuits PAs..........................75
Figure 53. Picture of the measurement setup.........................................................................75
Figure 54. Variation in the constant envelope signal due to the previous SCS algorithm.....76
Figure 55. Constant envelope signals after the modified SCS algorithm................................77
Figure 56. LINC/Chireix PA from CATENA Microelectronics..............................................78
Figure 57. Schematic representation of the LINC/Chireix PCB from CATENA....................79
Figure 58. Constellationdiagrams of 16-QAM, 16-APSK, 32-QAM, 32-APSK and 64-QAM.
..................................................................................................................................................81
Figure 59. Received constellationdiagram of 16-APSK when no compensation stubs (0
degrees compensation).............................................................................................................82
Figure 60. Output spectrum of amplified 16-QAM..................................................................83
Figure 61. 16-QAM constellation diagram with 20 degrees compensation stub. ..................83
Figure 62. Output spectrum of amplified 16-QAM with stubs................................................84
Figure 63. Constellation diagram of 16-APSK modulation without compensation stubs......84
Figure 64. Output spectrum of 16-APSK modulation without compensation stubs. ..............85
Figure 65. 16-APSK constellation diagram with a compensation stub of 20 degrees. ..........85
Figure 66. Output power spectrum of 16-APSK with a compensation stub of 20 degrees.....86
Figure 67. 32-QAM without compensation stubs. ..................................................................86
Figure 68. Output power spectrum of 32-QAM without compensation..................................87
Figure 69. Constellation diagram of 32-QAM with compensation stub at 20 degrees..........87
Figure 70. Output power spectrum of 32-QAM with a compensation at 20 degrees..............88
Figure 71. Constellation diagram of 32-APSK with 20 degrees compensation stubs............88
Figure 72. Output power spectrum of 32-APSK modulated signal with a compensation stub
at 20 degrees............................................................................................................................89
Figure 73. Constellationdiagram of 64-QAM with 20 degree compensation in the
LINC/Chireix amplifier............................................................................................................89
Figure 74. Output power spectrum of 64-QAM amplifiedby LINC/Chireix PA with 20
degrees compensation stubs.....................................................................................................90
Figure 75. Harmonic matching circuit for the Class-F PA on CATENA LINC board............94
Figure 76. Proposed measurement setup................................................................................95
18
MSc Thesis, Visweswaran Karunanithi
Figure 77. Hybrid-Chireix combiner......................................................................................96
20
MSc Thesis, Visweswaran Karunanithi
1
1. Introduction
he popularity of nano-satellites among universities and space research
organizations has grown quite rapidly over the last decade. Nano-satellite are a
class of small satellites that weigh less than 10 Kilograms, having almost all of the
subsystems present in a larger satellite and capable of mimicking larger satellites
with simpler payloads. The interest for nano-satellites among universities started
in 1999 through a CubeSat1
project that began at California Polytechnic State University (Cal
Poly) and Stanford University’s Space Systems Development Laboratory (SSDL). These
projects were mainly intended to be used by the Universities as a tool to learn satellite
development and understand the technology of developing hardware that can sustain the harsh
space environment. These missions mainly made use of Commercially Of-The-Shelf (COTS)
components which are industrial grade and easilyavailable. This reduced the development cost
and time of the mission. On the other hand, it reduced the life-time of a mission and also a
high probability of failure. The failures and lessons learnt from past nano-satellite missions
has helped in coming-up with mature designs that help in longer mission duration. An example
of this is Delfi-C32
that was developed by a team from Technical University of Delft and
launched on 28th
April 2008. This mission is a good example of a COTS based but robust
design that has survived multiple solar events and is still operational.
The focus of nano-satellite missions has slowly started to shift from simple technology
demonstration missions to more complex industry driven technology demonstration, science,
military and government missions. This clearly shows a paradigm shift in the way nano-
satellites are being perceived and the confidence in implementing more and more complex
missions on nano-satellites. One of the most important reason for this is the standardization in
shape, satellite-bus subsystems such as antenna systems, electrical power systems, On-board
computer, etc. This has helped the satellite developers to concentrate mainly on payload
development and integrate the payload with the satellite bus components and subsystems that
are available on the CubeSat market. They are developed by companies such as Innovative
Solutions In Space BV, GomSpace, etc.
As the complexity of nano-satellite missions grow, the amount of data generated by the
spacecraft (payload data and telemetry) has a direct influence on the downlink data-rate of the
communication link. The subsequent sections in Chapter-1 give an overview of the current
1 CubeSats is a standardization introduced by Calpoly that signifies a satellite that is in the shape of a cube. A
single unit of CubeSat is acube of 10 × 10 × 10 = 1000 cm3 byvolume andamass lesser than1.33 kg, also called
1U. This standardizationhas mainlyhelpedinusingthe same type of deployment system for various classes such
as 1U, 2U and 3U.
2 http://www.delfispace.nl/delfi-c3
T
21
MSc Thesis, Visweswaran Karunanithi
trends in nano-satellites, the paradigm shift in the nano-satellite mission types, how this
paradigm shift has caused the missions to start using high data rate transmitters and conclude
with the problem statement of this thesis work. In Chapter-2 a few mission cases are considered
to quantize the requirements posed by complex missions on the transmitter design with the help
of link budget calculations, state the CCSDS (Consultative Committee for Space Data Systems)
and ECSS (European Cooperation on Space Standardization) recommendations that have to be
followed in-order to design a transmitter that is compatible with a large number of ground
stations around the world that follow these standards and conclude with recommendations.
Chapter-3 will elaborate about the challenges in developing a high efficiency transmitter that
support the modulation schemes recommended in CCSDS space link protocol over ETSI
DVB.S2, some of the common efficiency and linearity enhancement techniques and conclude
with a trade-off that was performed on these techniques to find a suitable candidate for nano-
satellite transmitters. Chapter-4 deals with the LINC architecture which is one of the efficiency
and linearity enhancement techniques, where the design methodology is discussed with the help
of simulation results followed by an explanation of the measurement setup used to validate the
performance of the LINC architecture. Chapter-5 discusses the measurement results and
Chapter-6 provides the conclusions and future work.
1.1. Current trends in nano-satellites
In-order to understand the current trends in nano-satellite missions, a database of the nano-
satellite missions that were launched between 2003 and November 2013 was made and various
analysis were performed on this data. A total of 174 nano-satellites missions were launched
during this period, of which 183 transmitters were flown. The focus of the information
collected from these missions were: mission objective, year of launch, number of transmitters
flown, communication mode (half-duplex or full duplex), transmitter data-rate, modulation
schemes, transmit power and status of the mission. The database created can be found in
Figure 1. Number of nano-satellite launches.
22
MSc Thesis, Visweswaran Karunanithi
[Appendix]. Figure 1 shows the number of nano-satellites launched corresponding to the year.
It can be seen that there is a steep increase in the number of missions launched in 2013.
There is an increase of 437% from 2012 to 2013. These numbers were compared with
the Nano/microsatellite market assessment study that was later performed by SpaceWorks3
in
Feb 2014. The analysis converge and based on this analysis, the projections for 2014 show an
increase of 2 – 3 times higher than the number of launches in 2013.
Based on the mission objectives and the payload flown, the nano-satellite missions were
categorized into: education/technology demonstration, science/technology demonstration,
military, Industry/technology demonstration, education/science, remote sensing,
military/technology demonstration, technology demonstration, education and science missions.
The following Pie chart shows the percentage of missions that belong to each of the category:
Figure 2. (Left) Mission types between 2003 and 2012, (right) Mission types in 2013.
From Figure 2, it can be seen that 36% of the missions in 2013 comprise of industrial
technology demonstration, remote sensing, government/military missions compared to only
13% between 2003 and 2012. This change in trend shows that nano-satellites have started to
be perceived as a serious contender to replace the bigger satellites. As the mission type has a
3
http://www.sei.aero/eng/papers/uploads/archive/SpaceWorks_Nano_Microsatellite_Market_Assessment_Januar
y_2014.pdf
Figure 3. (Left) data-rates used by nano-satellite mission between 2003 and 2012 (right) data-rates used by nano-satellite
missions in 2013.
23
MSc Thesis, Visweswaran Karunanithi
direct influence on the downlink data-rate, it can be seen in Figure 3 the change in trend in
terms of popular data-rates used by nano-satellites between this period:
It can be seen that 43% of the missions used downlinkdata-rates below 1200 bps between2003
and 2012, compared to only 23% of them missions using this option in the year 2013. Looking
at the higher data-rate, only 9% of the mission used data-rates greater than 100 kbps between
2003 and 2012, compared to 24% of the mission in 2013 used a data-rate greater than 100 kbps.
Due to the limited availability of frequency bandwidth in some of the conventionally
used frequency bands such as VHF and UHF, as the downlink data-rate requirement increases,
there is also a need to go to the higher frequency bands such as, S-Band, X-band and Ku-band
where the larger bandwidths are available. The Figure 4 shows the trends in the frequency
bands used by the nano-satellites.
It can be seen that the number of missions using S-band and X-band downlink has slightly
increased in 2013. The amateur bands in UHF and VHF are a very popular choice for downlink.
The main reason for this can be attributed to space heritage and the popularity of this band
among radio amateurs around the world. This approach proves to be a win-win situation for
both the satellite developers in receiving the spacecraft from all around the world, as a result
increasing the reliability of a mission and at the same time benefit the amateur radio community
to perform small experiments as a return favour. OSCARs (Orbiting Satellites Carrying
Amateur Radios) in nano-satellites has gained popularity for the same reason. A good example
is FUNcube4
which was launched in November 2013 and has received 256.5 MB of telemetry
data so far (as of July 2014) from amateur radio operators around the world. This mission uses
a simple 1200 bps BPSK downlink. With the support of the amateur radio operators, the
telemetry of the satellite is available almost real-time, through the entire orbit.
4 FUNcube-1 (AO73) http://funcube.org.uk/
Figure 4. (Left) popular frequency bands used between 2003 and 2012. (Right) Frequency bands used by nano -satellite
missions in 2013.
24
MSc Thesis, Visweswaran Karunanithi
1.2. Conclusion of the analysis
Based on the study done so far to determine the current trend and to see if there is really a need
to start investigating the need for a high data-rate transmitter for a nano-satellite mission, the
following conclusions were derived:
- First and foremost, the analysis done in the previous section, converges with the 2014
Nano/microsatellite Market Assessment that was conducted by SpaceWorks in February 2014.
This market assessment also goes on to project a 2 – 3 times increase in the number of nano-
satellite that will be launched in 2014, thus between 160 and 240, and more than half of the
missions will be for Earth Observation/Remote Sensing.
- Apart from the total number of launches, nano-satellite cluster launches has also started to
gain popularity. A Minotaur-1 rocket put 29 satellites into orbit in a single launch on 19th
November 2013 and DNEPR placed 32 satellites in a single launch on 21st
November 2013.
This has made frequency coordination really challenging as most of the nano-satellites
preferred VHF and UHF bands which are very narrow bands and during initial days of the
mission after the satellite separates from the launch adaptor, the satellites are very close to each
other. This can result in interference issues. This would mean that satellites will have to start
choosing frequency bands where a large bandwidth is available to avoid interfering with other
satellites.
- As the complexity of the missions increase, the amount of data needed to be downlinked also
increases. This has made the nano-satellite developers to start using higher data-rates, and as
a result, start using S and X band.
- It was seen that a redundant low speed downlink transmitter (operating in amateur bands)
were flown alongside high data-rate transmitter in various missions. The use of amateur
frequency bands is likely to continue till a reliable standard is followed for high speed
downlinks and ground stations around the world upgrade to support such missions.
- Spectrally efficient modulation scheme is the need of the hour. Considering the number of
launches forecasted, there is a need to start investigating these modulation schemes for nano-
satellite missions.
- It is wise to adopt some of the existing recommendations laid down by CCSDS and ECSS.
The main advantage is, there are ground stations around the world that already support these
standards and the recommendations are popular among the bigger satellites.
25
MSc Thesis, Visweswaran Karunanithi
1.3. Problem statement
Based on the market study and preliminary analysis performed, it is evident that there is a need
to develop high speed transmitters for nano-satellites that implement spectrally efficient
modulation scheme and are complaint with ground station standards followed around the world
(especially ESA ground stations). Using this as the motivation, recommendations stated in
CCSDS and ECSS standards have to be studied and design constraints have to be derived.
CCSDS also provides recommendations for spectrally efficient modulations schemes. It has to
be investigated if these modulation schemes suit nano-satellite applications.
Mission case study needs to be done using some of the on-going and future projects to
see if there is really a need to implement such complex modulation schemes in nano-satellites.
Implementation challenges needs to be studied and the need to implement efficiency and
linearity enhancement technique in the transmitter design has to be investigated.
If efficiency and linearity enhancement technique needs to be implemented, using
system engineering approach, the most appropriate architecture for nano-satellite application
needs to be chosen. Once the architecture is chosen, a prototype needs to be designed to show
the proof of concept.
The following diagram shows the work-flow of this thesis:
26
MSc Thesis, Visweswaran Karunanithi
28
MSc Thesis, Visweswaran Karunanithi
2
2. Nano-Satellite transmitters for
the future
ano-satellites in the past were mainly thought of as practical learning tools by the
universities to teach engineering students about designing space hardware and
encourage the interest in pursuing a career in the space industry. The main
advantage of nano-satellites over the bigger satellites is: very short development
time and cheaper access to space. This attracted industries to start collaborating with the
universities and started using nano-satellites as a test-bed to quickly test and space-qualify
satellite hardware. A good example for this is Delfi-C3 which carried Thin Film Solar Cells
(TFSC) from Dutch space and two Autonomous Wireless Sun Sensors (AWSS) from TNO as
a part of its technology demonstration payloads [1]. Gradually over the period, a lot of
standardization has lead the nano-satellite developers to start coming-up with more complex
payloads. This change has posed the need to look at high data-rate transmitters that can meet
the needs of mission requirements. This chapter starts with an overview of the
recommendations laid by CCSDS and ECSS for space-to-earth links for satellites that fall under
category-A5
. Following the recommendations, a few nano-satellite mission cases are studied
to see how the recommendations fit, link budget analysis is performed to derive the design
requirements of the transmitter. This chapter concludes with some recommendations for the
roadmap for nano-satellite transmitter design.
2.1. CCSDS and ECSS recommendations
CCSDS was founded in 1982 by major space agencies of the world, it is a multi-national forum
for development of communications and data systems standards for spaceflight6
. The CCSDS
recommendations cover a broad spectrum: Space Internetworking Services, System
engineering, Mission Ops & Information Management services, Cross Support Services,
Spacecraft Onboard Interface Services and Space LinkServices. The scope of this thesis limits
to the following recommendations stated in Space Link Services: Radio Frequency and
modulation systems (Blue book, CCSDS 401.0-B) [2], Bandwidth-efficient modulations
(Green book, CCSDS 413.0-G-2) [3] and CCSDS Space link protocols over ETSI DVB-S2
standard (Blue book, CCSDS 131.3-B-1) [4].
5 Category-Aare those satellites whichorbit at an altitude of below 2 x 106 Km, Category-B are thosesatellites
which orbit at an altitude above 2 x 106 Km (This categoryalso includes deepspace probes)
6 http://public.ccsds.org/default.aspx
N
29
MSc Thesis, Visweswaran Karunanithi
2.1.1.Frequency band allocation & constraints
The frequency band allocationis done based on the service/operation provided by the satellites.
The different services are: Space Research (SR), Earth Exploration Satellite Service (EES),
Amateur Service and Space Operation (SO). The Table 1 gives the list of recommended
frequency bands corresponding to the service it can provide.
Table 1. Frequency allocation corresponding to service provided.
Frequency bands (MHz) Service7
(L-Band) 1215 - 1240 SR, EES
(L-Band) 1240 - 1300 SR, EES, Amateur
(L-Band) 1525 -1535 SO, EES
(S-Band) 2200 - 2290 SR, SO, EES
(X-Band) 8025 - 8400 EES
(X-Band) 8450 - 8500 SR
(Ka-Band) 25500 - 27000 SR, EES
(Ka-Band) 37000 - 38000 SR
The constraints in using each of the frequency bands is as follows:
- L-Band:
o This band shall not be used for feeder links of any service.
- S-Band (2200 – 2290 MHz):
o The maximum occupied bandwidth for spacecraft in this band shall not exceed
6MHz.
o The link shall be active only during the period when the satellite is in the
visibility cone of the ground station. (This will help frequency re-use)
o The reliability of the device used to switch-off the transmitter shall
commensurate with mission lifetime.
7 Purelyscience missions suchas radio-telescopes fall under the categoryof SR, Engineeringmissions suchas
communicationsatellitesfall under the categoryof SO and earth observationmissions suchas Remote sensing
fall under the categoryof EES.
30
MSc Thesis, Visweswaran Karunanithi
- X-Band (8025 – 8400 MHz): The constraints specified for S-Band applies for this band
as-well, except for the bandwidth constraints.
- X-Band (8400 – 8450 MHz):
o The maximum allowable bandwidth for a down-link in this band shall not
exceed the mask specified in Figure 5.
- Ka-Band (25.5 to 27 GHz & 37 to 38 GHz): There were no constraints formulated for
this band but there has to be an agreement with the frequency coordinator prior to using
this band.
- All the bands, up to X-band shall have a frequency stability not less than ± 2 × 10-5
(20ppm) under all conditions for the entire lifetime of the mission and for Ka-band, a
frequency stability of ± 0.02 ppm/o
C within the temperature range +10 o
C to +40 o
C.
The maximum allowed bandwidth corresponding to various data-rates is shown in the table
below:
Table 2. Maximum allowed bandwidth corresponding to data-rate and frequency bands.
Frequency
band (MHz)
Operation Occupied Bandwidth
2200–2290[Downlink]
And
8450–8500[Downlink]
Telemetry(Rs <
10ksps)
300 kHz
Telemetry (10 ksps
<Rs < 60 ksps)
1200 kHz or 30 × Rs
(smaller of the two)
Telemetry(60 ksps <
Rs < 2Msps)
1200 kHz or 12 × Rs,
whichever is larger upto
6 MHz at 2 GHz and 10
MHz at 8 GHz.
Telemetry(Rs >
2Msps)
1.1 x Rs up to 6 MHz at
S-band and 10MHz at
8GHz.
There are no bandwidth constraints stated for the frequency bands in Ka-band.
Figure 5. Maximum allowable bandwidth vs symbol rate.
31
MSc Thesis, Visweswaran Karunanithi
2.1.2.Transmitter spurious emission and harmonic levels
The emitted spectrum for all the space science services shall adhere to the spectral emission
mask shown in Figure 6. The total power contained in any single spurious emission shall not
exceed -60 dBc. This constraint is valid for frequency bands from VHF to Ka-band. In the
figure, the blue line indicated the boundary limits for data rate below 2 Msps, it has a slope of
43 dB per decade and the green line indicates the limit for data-rate above 2 Msps which has a
slope of 60 db per decade.
2.1.3.Recommended modulation schemes
The occupied bandwidth is directly dependent on the data rate, the modulation schemes are
recommended based on the data-rate is shown in Table 3
Table 3. Recommended modulation schemes.
Datarate<2
Msps
When using Sub-carrier:
- PCM/PM/Bi-phase.
- PCM/PM/NRZ.
Suppressed carrier modulation (medium rate <2Msps)
- BPSK
- QPSK
- OQPSK.
Datarate>
2Msps
Preferred modulationschemes when datarates are larger
than 2Msps:
- GMSK (Pre-coding necessary)
- Baseband filtered OQPSK
- TCM 8PSK
Figure 6. Spectral emission limits
32
MSc Thesis, Visweswaran Karunanithi
CCSDS space link protocols over ETSI DVB.S2
recommendations:
- QPSK
- OQPSK
- 16-APSK
- 32-APSK
- 64-APSK.
The downlink data-rates are categorized into high-speed downlink ( > 2Msps) and low-speed
data-rate (<2 Msps). The modulation schemes such as PCM/PM/Bi-phase and NRZ are
allowed only for data-rates below 2 Msps. The modulation schemes recommended for data
rates below 2 Msps are all constant envelope signals and does not influence the performance
of a HPA (High Power Amplifier) to a large extent when compared to non-constant envelope
modulation schemes. In the case of digital modulation techniques such as BPSK, QPSK and
OQPSK, although there is no inherent amplitude variation, but the use of SRRC (Squire Root
Raised Cosine) filter in-order to reduce the ISI (Inter Symbol Interference) causes an amplitude
variation, affecting the performance of the HPA. The non-linearity effects caused by OQPSK
is lesser compared to QPSK at larger SRRC roll-offs [5]. For data-rates above 2 Msps, the
modulation schemes proposed by CCSDS space link protocols over ETSI (European
Telecommunication Standards Institute) DVB-S2 (Digital Video Broadcast – Satellite) include
QPSK and OQPSK which are constant envelope schemes and a few additional modulation
schemes such as 16-APSK, 32-APSK and 64-APSK. APSK stands for Amplitude Phase Shift
Keying. As the name suggests, both phase and amplitude of the carrier are modulated in
accordance with the message signal. It is similar to QAM (Quadrature Amplitude Modulation)
but the PAPR of APSK modulation scheme is lesser than that of QAM. An in-depth analysis
of PAPR of these modulation schemes and its effects on the performance of the PA will be
discussed in later sections.
2.1.4.Conclusion of the CCSDS and ECSS recommendations
The recommendations stated in CCSDS and ECSS are a good starting point for standardization
in the communication system for nano-satellites. The standards have already been
implemented on the bigger satellites and ground stations around the world have the
infrastructure compatible with CCSDS standards.
Some quantitative analysis needs to be performed to determine if the modulation
schemes recommended for data-rates larger than 2 Msps is required for nano-satellites with the
help of mission case studies.
The main challenge in implementing the recommended modulation schemes for nano-
satellites is that, most of the spectral efficient schemes have a non-constant envelope profile.
This variation in the amplitude can cause a degradation in the performance of the PA in terms
of either efficiency or linearity, thus an analysis needs to be done to determine if one of the
conventional operating classes (Class – A, AB, B, C, D, E, F, F-1
or J) is sufficient or
efficiency/linearity enhancement technique needs to be performed. In case if an enhancement
33
MSc Thesis, Visweswaran Karunanithi
technique needs to be implemented, a trade-off between the different architectures needs to be
performed and the most suitable architecture needs to be chosen.
In the subsequent sections of this chapter, a few nano-satellite missions will be studied
to establish the requirements for a nano-satellite transmitter.
2.2. Nano-satellite mission case study
The downlink data-rate requirement for a space mission mainly depends on:
1) Number of sub-systems in the spacecraft
2) The amount of house-keeping data generated by each of the sub-system
3) The amount of payload data generated, which depends on the type of payload
4) The contact time with the ground station.
The amount of data generated by the payload depends on the service provided by the mission.
The Figure 7 shows the relation between the downlink data-rate requirement and the type of
the missions.
Figure 7. Pictorial representation of the relation between downlink data-rate, service provided and type of
payload.
34
MSc Thesis, Visweswaran Karunanithi
2.2.1.Spacecraft housekeeping data downlink requirement
In this case, a general 3U nano-satellite LEO mission is considered to calculate the total amount
of house-keeping data that will be generated by the spacecraft bus. The subsystems that are
assumed in a generic satellite bus are: On-board computer, antenna system, transceiver
(primary, secondary), electronic power system (Battery system, solar cells, etc.), attitude
control system (sensor and actuator) and payload telemetry. In order to get a more realistic
number, the satellite bus of TRITON-1 mission was considered. The house-keeping data from
each of the sub-systems is periodically sampled by the on-board computer and stored in the on-
board memory. Based on the available memory and mission requirements, either all of the
stored telemetry is downlinked or only the mission critical data is sent to the ground station. In
the case of TRITON-1, only some of the mission critical telemetry was downlinked. This case
considers a hypothetical situation where all the sampled telemetry needs to be downlinked to
the ground station.
The size of each of the telemetry packets is approximately 16 bytes and it is assumed
that each of the subsystems is sampled for telemetry on an average every 15 seconds. The total
amount of house-keeping data generated was calculated to be 7.7 Mbits per day. The Table 4
gives the calculation done in-order to find the required downlink data-rate.
Table 4. Required downlink data-rate to downlink all the generated house-keeping data.
TT&C data rate requirement. Units
Typical sampling time 1 sec
Data per subsystem 16 bytes
128 bits
Duty-cycle 100%
Data generated per day per subsystem 7.732E+05 bits
Number of subsystems (approx.) 10
Total data generated per day 7.732E+06 bits
Avg. GS contact time per day 50 min
Percentage of contact time to transmit
house-keeping data. 10 %
Effective TT&C downlink data rate 24576 bps
Data-overhead 50%
Data rate including over-head 36864 bps
Considering that a nano-satellite will not have a dedicated telemetry downlink, the same link
has to be shared between telemetry and payload data. Thus, in the above calculation it is
assumed that 10% (a rough estimate) of the total contact time is used to downlink the telemetry
data. The calculated downlink data-rate requirement is approximately 37 kbps.
2.2.2.Payload downlink data-rate requirement for a EES service
The Remote Sensing payloads mainly fall under the category of EES services. Some of the
commonly used payloads under this service include optical imaging payload, infra-red imaging
payload and SAR (Synthetic Aperture Radar). Among nano-satellites, optical imaging payload
35
MSc Thesis, Visweswaran Karunanithi
has been the most commonly used payload. As a case study, the mission requirements stated
for MISC-1 [6] mission will be considered. The mission was proposed to capture 7.5 m GSD
(Ground Spatial Distance) multispectral imagery from an altitude of 540 km. The payload for
this mission comprises of a 16 MP Kodak KAI-16000-CXA-JD CCD image sensor. The
mission overview are stated in Table 5 below:
Table 5. MISC-1 mission overview
MISC mission units
Payload Multi-spectral imaging
Volume 3U
Altitude 540 km
Image sensor KAI-1600-CXA
Number of Pixel 16 MP
Active pixel rows 4872
Active pixel column 3248
Number of Ground stations 13
Ground swath/image 702 Sq.km
Target area per day 137,500 Sq.km
Total number of images required per
day 196
Size of a single un-compressed image
(10 bit pixel depth) 19.78 MB
In the paper [6], it could be clearly seen that the mission was constrained by the downlink radio
capabilities (38.4 kbps downlink), as a result the image had to be compressed to 1/12th
its
original size and the mission required 13 ground stations to fulfill the requirement of
downlinking 196 images per day. The mission was finally not realized using the above
formulated mission requirements
Now, the required downlink data-rate will be calculated for different cases to see an
optimal solution for the mission to be feasible.
Table 6. Downlink data-rate requirement when un-compressedimage is transmittedand13 ground stations are
used.
MISC mission downlink data rate
requirement
Number of images needed to be
transmitted/day 196
Size of an image (Pixel depth assumed to
be 10bits) 1.58E+08 bits
Total image data generated/day 3.101E+10 bits
Avg. Ground station contact time/day 36 min
Total Ground station contact time (13
ground stations) 468 min
effective payload downlink data rate 1.10E+06 bps
Over-head 50%
Total required downlink data rate 1.65E+06 bps
36
MSc Thesis, Visweswaran Karunanithi
From the above calculation it can be seen that the downlink data-rate required to transmit un-
compressed image using 13 ground stations is 1.65 Mbps. Similarly, two other cases were
considered because the idea of using 13 ground stations to downlink the payload data from one
satellite means highly synchronized operation between the ground stations which is complex.
The two other scenarios that were considered are: 1) Downlink data rate requirement
when one satellite and one ground station is used, 2) Downlink data-rate requirement when 3
satellites and one ground station is used. In the first scenario, the calculation showed a required
downlink data rate of 21.52 Mbps and the second scenario required a down link data-rate of 7
Mbps per satellite.
2.2.3.Downlink data-rate requirement for nano-satellite providing data
services
For this case, VDE-Sat is considered. VDE stands for VHF Data Exchange Satellite. It is a
technological concept developed by IALA e-NAV8
committee and now widely discussed at
ITU, IMO and other organizations. VDES was originally developed to address emerging
indications of overload of VHF Data Link (VDL) of AIS and simultaneously enables a seamless
data exchange for the maritime community. This is an on-going project at ISISpace BV, which
is in a conceptual stage. The Figure 8 gives a pictorial representation of the mission concept.
Figure 8. Pictorial representation of VDES [7].
Some of the mission requirements are already formulated, the mission needs to have a
high-speed downlink at VHF (161MHz), the available bandwidth is 50 kHz and a downlink
symbol rate of 40 ksps. Based on these requirements, link budget was calculated for various
8 International Associationof Marine Aids to Navigation and Lighthouse Authorities
37
MSc Thesis, Visweswaran Karunanithi
possible modulation schemes and finally 16-APSK modulation was chosen. The link budget
analysis for this case is explained in the next section.
38
MSc Thesis, Visweswaran Karunanithi
2.3. Link budget analysis
For the cases discussed in the previous section, a link budget was calculated to check if the
assumed system design meets the requirements. The calculations were done by creating a MS
excel template with the relevant equations and the possibility to vary the various system
parameters (transmit power, transmit antenna gain, receiver antenna gain, etc.).
In the case of spacecraft house-keeping data, the required downlink data-rate was
calculated to be ~37 kbps. The link budget was calculated with realistic numbers from a
commercially available UHF transmitter and the antenna gain was considered to be 0 dBi. With
this, the link margin was calculated for different elevation angles and shown in Figure 10.
Figure 9. Link budget calculation of spacecraft housekeeping data downlink.
Figure 10. Link margin for different elevation angles.
39
MSc Thesis, Visweswaran Karunanithi
The modulation scheme chosen in this case is QPSK/OQPSK, and the occupied bandwidth of
the transmitted signal is 23 kHz.
In the case of an imaging payload discussed in the second case study, after various iterations,
an X-band downlink was chosen to downlink at 7 Mbps. The calculations and plot of the link
margin for various elevation angles are shown in Figure 11 and Figure 12.
Thus, it can be seen that it was possible to close the link for a downlink data-rate of 7 Mbps, a
transmit power of 2 Watts and an EIRP of 35.5 dBm.
In the case of VDE-Sat, again various options for modulation scheme and transmit power were
tried and it was possible to close the link with 16-APSK modulation scheme and a transmit
power of 6 Watts. The link calculations and plot are shown in Figure 13 and Figure 14.
Figure 11. Link budget for an optical payload data downlink.
Figure 12. Link margin vs elevation angles for optical payload downlink.
40
MSc Thesis, Visweswaran Karunanithi
2.4. Summary of the case study and conclusion
A summary of the case study is shown in Table 7, it can be seen that there is a direct relation
between the required downlink data-rate and the type of service the mission provides. It is not
possible to implement these high data-rates in the amateur bands where the available bandwidth
is limited. Thus, there is a need to move to the higher frequency bands and also use spectrally
efficient modulation schemes.
In general, missions with remote sensing payload needs the fastest downlink. In the
above case a 7 Mbps downlink was required to downlink all the generated payload data. The
Figure 13. Link budget for VED-Sat downlink.
Figure 14. Link margin vs elevation for VDE-Sat.
41
MSc Thesis, Visweswaran Karunanithi
frequency band for satellite house-keeping data downlink was chosen to be UHF and it can be
seen that it is possible to implement this relatively high-speed down link of 36 kbps within the
available bandwidth of 25 kHz.
Table 7. Summary of mission case study.
Spacecraft
housekeeping
data
MISC-1 payload
downlink
VDE-sat downlink.
Frequency Band UHF 8.4 GHz [X-band] 161.9 MHz [VHF]
Available
bandwidth
25 kHz 10 MHz 50kHz
Occupied
bandwidth
23.3 kHz 6.3 MHz 50 kHz
Downlink data rate 36 kbps 7 Mbps 40 ksps (160 kbps)
Modulation scheme QPSK/OQPSK 16-APSK 16-APSK
Transmit power 500 mW 3 W 6 W
PAPR [dB]9
3.4 4.5 4.5
The table also shows the PAPR of the spectral efficient modulation schemes used. The main
disadvantage in using a linear PA in the transmitter to amplify such non-constant envelope
signal is that, the efficiency drops drastically. Some of the transmitter components and
parameters that depend upon the efficiency and power consumption of the PA are:
- Battery life,
- Solar-cell and battery capacity,
- Heat-sink size and weight,
- Need for additional coolers.
Thus, the main challenge in implementing these modulation schemes for nano-satellite
missions is to implement an efficient amplifying architecture and at the same time not
compromise on the linearity. In order to achieve this, different efficiency and linearity
enhancement techniques need to be investigated and a suitable architecture needs to be chosen
for nano-satellite application.
9 Peak to Average Power Ratio calculatedincluding a SRRC filter witha roll-off of 0.5
43
MSc Thesis, Visweswaran Karunanithi
3
3. High efficiency transmitter
architectures
ome of the main constraints that needs to be considered while designing a subsystem
for nano-satellite mission are: power consumption, efficiency, low complexity,
efficient thermal design and sustain harsh space environment such as temperature
variations and radiation effects. Transmitter is one of the most power hungry
subsystems of a nano-satellite, based on the mission requirements the transmitter of a nano-
satellite can consume up to 30% of the total generated power. Thus, it is very important to
investigate a transmitter architecture that can efficiently transmit spectral efficient modulation
schemes that were discussed in the previous section, without compromising on the overall
performance. The main challenge involved in designing an efficient architecture is: the spectral
efficient modulation schemes have a non-constant envelope profile as a result of modulating
the message signal into both phase and amplitude of a carrier, the amplitude variation in the
modulated carrier is characterized by the Peak to Average Power Ratio (PAPR) of the signal.
The PAPR is the ratio of the peak power of the envelope to the average power of the signal
envelope. This is one of the critical parameters that influence the trade-off between linearity
and efficiency of a Power Amplifier in a transmitter. This chapter gives an overview of the
challenges posed by spectral efficient modulation schemes on the performance of the
transmitter, followed by different efficiency and linearity enhancement techniques from the
literature, a trade-off between the different architectures to determine a suitable one for nano-
satellites and a conclusion of this study.
3.1. Overview
The basic functionality of a transmitter is to collect the telemetry/payload data from the
On-board computer, put the data into packets, modulate the information in the packet, amplify
S
Figure 15. Block-diagram of a transmitter with digital modulation.
44
MSc Thesis, Visweswaran Karunanithi
the modulated signal to appropriate levels and pass it on to the antenna which radiates the
modulated RF signal. A generic block-diagram representation of a transmitter is shown in
Figure 15.
The baseband board is responsible for interacting with the on-board computer and
collect the data to be transmitted (telemetry/payload), implement the data link layer functions,
perform signal processing on the baseband digital data (Square Root Raised Cosine filtering)
and generates the stream of I (In-phase) and Q (Quadrature) data as an input to the modulator.
The IQ modulator, modulates the baseband IQ signal onto the carrier frequency and the output
power of the modulator is generally not enough to directly be transmitted and needs
amplification. Some of the IQ modulators currently available in the market produce an output
power between 0 dBm (1mW) to 10 dBm (10mW). Thus, the amplifier performs the operation
of converting the DC power (supply) into a significant amount of RF/microwave output power
that replicates the modulated signal that is inputted to the PA. The key design parameters of a
PA are efficiency and linearity. With the conventional class of operation such as Class-A, B,
AB, C, D. E and F, a trade-off has to be done between linearity and efficiency. There is no one
single PA or transmitter technique that suits all applications. A relation between efficiency and
linearity for different Classes of operation is shown in Figure 16.
When a PA is used to amplify constant envelope signals such as CW (Continuous Wave), FM,
classical FSK or GMSK, linearity of the PA is not critical and can be compromised for
efficiency. A Class C, D, E or F would be a logical choice for a constant envelope signal. On
the other hand, linearity of a PA becomes critical when the signal contains both amplitude and
phase modulation such as QAM or APSK. In this thesis the scope of modulation schemes that
will be considered to characterize the performance of the PA are: 16 and 32-APSK, which will
be compared with 16, 32 and 64-QAM modulation schemes. The PAPR of the QAM scheme
is more than the APSK modulation scheme. The PAPR contributed due to variation in the
amplitude of the symbols is shown in Table 8.
Figure 16. Linearity vs Efficiency trade-off for conventional PA classes.
45
MSc Thesis, Visweswaran Karunanithi
Table 8. PAPR contribution due to the envelope profile of various modulation schemes.
In addition to the PAPR contributed by the modulation scheme, additional amplitude variation
is caused due to passing the digital data through an SRRC filter. In order to minimize inter-
symbol interference, the digital data needs to be passed through a SRRC filter. The PAPR
depends on the roll-off factor of the filter. The following table (Table 9) shows the additional
PAPR contributed due to the SRRC filter for different roll-off factors [8].
Table 9. PAPR contribution due to the SRRC filter
Roll-off 0.1 0.2 0.3 0.4 0.5
PAPR 7.5dB 5.8dB 4.6dB 3.7dB 3.4dB
Thus the PAPR seen by the PA is the sum of the above two contributors. For example, if 32-
APSK modulation is implemented with a SRRC filter having a roll-off 0.5, then the combined
PAPR seen by the amplifier is 4.8 dB.
Peak efficiency of a PA is obtained when driven at saturation but this is also a non-
linear region. Thus, alternate architecture to amplify efficiently without compromising on the
linearity is studied in this chapter. Some of the techniques analyzed in this thesis are:
3.1.1. Characterization of linearity and efficiency
The non-linearity in a PA is caused mainly due to gain compression and saturation. The effects
of non-linearity can be characterized using various methods: carrier-to-intermodulation (C/I)
Modulation M-PSK 16-APSK 16-QAM 32-APSK 32-QAM
PAPR 0dB 1.1dB 2.6 dB 1.4 dB 3.9 dB
Figure 17. Efficiency and linearity enhancement techniques.
46
MSc Thesis, Visweswaran Karunanithi
ratio is used to compare the amplitude of the desired output carrier to the intermodulation
distortion (IMD) products, adjacent channel power ratio (ACPR) which is used to characterize
the quality of the spectrum, the power ratio in dBc is measured between the spectral
components within the allotted bandwidth and adjacent channel components. Another method
to characterize the linearity is by measuring EVM (Error Vector Magnitude), EVM gives the
vector distance between the received symbol and expected symbol for a given modulation
scheme in the constellation diagram.
Efficiency is one other important design parameter, from the literature there are various
definitions for efficiency of a PA. Drain efficiency: It is the ratio of the output power from the
PA (in Watts) to the supplied DC power (in Watts). It is given by:
out
d
DC
P
P
  (1)
Power added efficiency (PAE): The PAE is defined as the ratio of the RF power that is
amplified (difference between the RF output power and RF input power) to the DC power
consumed. It is given by:
( )out in
PAE
DC
P P
P

   (2)
The PAE is a more widely used metric as it takes into account the gain of the PA.
When a non-constant envelope signal is amplified by a PA, a more useful metric to measure is
the average efficiency [9], it is mathematically expressed as
oAVG
AVG
dcAVG
P
P
   (3)
PoAVG and PdcAVG are a function of the Probability Density Function (PDF) of the input non-
constant envelope signal. The PDF gives the relative amount of time an envelope spends at
various amplitude levels. Figure 18 gives the PDF of a 16-APSK signal when passed through
a SRRC filter that has a roll-off of 0.5.
Figure 18. PDF of 16-APSK modulation scheme, implemented using an SRRC filter with a roll-off
of 0.5(4000 sps)
47
MSc Thesis, Visweswaran Karunanithi
Although this matric is straight forward to measure for the conventional PA classes (Classes
A-F), the above equation cannot be directly used in all the efficiency and linearity enhancement
techniques that will be discussed later in this chapter. In the case of efficiency and linearity
enhancement techniques such LINC, EER and ET the message signal that is fed to the PA is
modified so that the PA is driven at its peak efficiency. In such cases PAE is a more preferred
metric to gauge the performance of the architecture.
3.2. Doherty architecture
The Doherty architecture (shown in Figure 19 ) was first proposed by W.H. Doherty in 1936
[10]. This architecture makes use of active load-pull technique where the load seen by one PA
can dynamically be varied by applying current from another PA which is coherent in phase.
3.2.1.Working Principle
Figure 19. Doherty PA architecture.
Figure 20. Schematic representation of Doherty.
48
MSc Thesis, Visweswaran Karunanithi
This architecture comprises of two10
signal paths, amplified by two non-identical PAs called
Carrier PA and Peaking PA (sometimes also called Auxiliary PA). At low input levels the
peaking PA is in cut-off (consuming no current) and only the carrier PA is operational. Once
the input level reaches a certain threshold, the peaking PA turns ON and the output power is
contributed by both the PAs. The transfer curves of the two PAs is shown in Figure 21 [11].
The carrier PA operates linearly till the point output power reaches Pmax/4 as there is no
influence from the peaking PA (not yet turned ON), at Pmax/4 (which is 6 dB below the peak
envelope power), Peaking PA turns ON and current flows from both the PAs to a common load
leading to load modulation. From this point onwards the transfer curve of both the PAs are
non-linear. The process of active load pulling can be better explained using the schematic
representation of the architecture shown in Figure 20.
Both the PAs are considered to be current sources. When the Peaking PA is OFF, I2 = 0 and
only I1T flows through the load resistor RL. The voltage drop across the resistor is a linear
function of the current flowing through the resistor RL.
But when the Peaking PA turns ON, both I1L and I2 flows through the load resistor and the
impedance seen by the individual PAs is given by:
10 It is possible to have N paths, but the architecture was introducedwith two paths.
Figure 21. Input power vs Output power in linear scale [11]
49
MSc Thesis, Visweswaran Karunanithi
2
1
1
2
1
1
2
1
2
1
1
2
2
1
1
1
T L
T
T
T
T
T
L
T
L
I
Z R
I
Z
Z
Z
Z
Z
I Z
R
V
I
Z R
I
 
  
 


 
 
 
 
  
 
(4)
It can be seen from the above equation that the impedance seen by each of the PA is a function
of the current flowing through the other PA.
3.2.2.Advantages and dis-advantages of Doherty architecture
Some of the advantages of Doherty architecture are:
- It is possible to achieve a good efficiency enhancement in the power back-off region.
- Suitable to amplify signals with a PAPR as large as 10dB.
- The architecture is relatively easy to implement compared to EER and Envelope
Tracking architectures.
- The performance in the power back-off region can be significantly improved by using
more stages (N-way Doherty) [12].
Dis-advantages of Doherty are:
- Gain degradation: Due to the low bias voltage, Peaking PA contributes lower output
compared to Carrier PA.
- Peaking PA has to switch ON at exactly Pmax/4.
- Both PAs have to follow the nonlinearity of Figure 21 and generally pre-distortion
circuit is necessary.
- Higher intermodulation distortion: It is caused again due to low biasing of the Peaking
PA.
- Narrow bandwidth: The quarter wave lines used for impedance transformation reduces
the bandwidth of operation.
3.3. Kahn/EER (Envelope Elimination and Restoration)
3.3.1.Working principle of Kahn/EER architecture
This technique of amplification was first proposed by Kahn in 1952. In a non-constant envelope
signal, message is modulated onto both amplitude and phase of the carrier signal. In the case
of Kahn/EER method, the amplitude of the message is first eliminated from the modulated
signal, leaving behind a constant envelope phase modulated signal and the amplitude envelope
profile. Now, the constant envelope signal can be amplified using a high-efficiency, non-linear
PA such as class C, D and the envelope profile is separately amplified and fed in as the supply
50
MSc Thesis, Visweswaran Karunanithi
voltage of the high efficiency PA. A block diagram representation of this architecture is shown
in [figure]. Thus, it is possible to achieve high efficiency and linearity regardless of the PAPR
of the input signal. From the literature, it can be seen that very highefficiencies can be achieved
but the design is fairly complex to implement.
3.3.2.Advantages and dis-advantages of Kahn architecture
Some of the advantages of the architecture are:
- Ideally can be operated at the peak efficiency of the final PA (depends on the
operating Class chosen for the final stage).
- Provides excellent linearity as the performance is not dependent on the linearity of the
amplifying transistor.
Some of the dis-advantages of this architecture are:
- Circuit complexity: Implementation of the envelope restoration circuitry is complex,
as perfect synchronization has to be achieved between the two signals.
- Needs pre-distortion at higher frequency.
- The switching frequency of the class-Smodulator should be at least 6 times the RF
bandwidth.
3.4. Envelope Tracking (ET) architecture
3.4.1.Working principle of ET
ET and EER technique use the concept of supply modulation, where EER uses a combination
of non-linear PA and an envelope re-modulation circuit but in the case of ET, a combination of
linear PA and a supply modulation circuit which tracks the envelope profile of the input signal.
The envelope profile is converted to discrete DC levels using the power conditioner and fed as
the drain voltage to the final PA. The supply voltage is varied dynamically with sufficient
Figure 22. Block diagram representation of a modern implementation of the Kahn architecture.
51
MSc Thesis, Visweswaran Karunanithi
headroom to allow the RF PA to operate in a linear mode. The block diagram representation
of an ET architecture is shown in Figure 23
3.4.2.Advantages and disadvantages of ET architecture
The advantages of ET architecture are as follows:
- The implementation is slightly less complex when compared to EER.
- The efficiency of linear amplifiers such as Class-A or AB can be enhanced using this
technique.
- Constant efficiency can be maintained in the power back-off region.
The main disadvantages of this architecture are:
- The peak efficiency achieved cannot exceed that of Class-AB.
- Circuit overhead is high.
- Synchronization between the carrier containing phase information and the time varying
supply is the main challenge in this technique.
- The PAE is poor compared to other efficiency enhancement techniques due to the
additional circuitry.
3.5. Switched Capacitor Digital Power Amplifiers SCDPA
3.5.1.Working principle of SCDPAs
Instead of making use of the linear trans-conductance property of a CMOS transistor, SCDPA
makes use of its switching property. The block diagram of SCDPA is shown in Figure 24.
Deeply scaled CMOS transistors are poor trans-conductors but very good switches. SCDPAs
exploit this property. SCDPAs incorporates the functionality of a Digital-to-analog converter
and a PA into the same circuitry. Each of the CMOS transistors are represented by switches
(C0, C1, C2…. C7) and they are either at toggle (between Vdd and ground) or only ground
potential. This is decided by the control signal given to the gate from the micro-controller/DSP.
Figure 23. Block diagram of ET architecture.[11]
52
MSc Thesis, Visweswaran Karunanithi
The output voltage depends on the ratio of the switches that are ON (toggle state) to the total
number of switches. For example, C4, C5, C6 and C7 are at ground potential (representing
OFF) C0, C1, C2 and C3 are in toggle mode (ON state) then Vout will be 0.5Vdd. Similarly,
when all the switches are in toggle mode, maximum output power can be achieved. More
details about this architecture can be found in [13].
3.5.2.Advantages and dis-advantages of SCDPAs
Some of the advantages of this architecture are:
- Low power consumption.
- High signal bandwidth.
- Good performance in the power back-off region. (flat efficiency uo-to 13dB PAPR
reported [14])
Some of the disadvantages of this architecture are:
- Not highly linear: AM-PM distortions are high.
- Parasitic capacitance in the current technology still limit its performance.
- Low PAE (45% reported[14])
- Low output power.
Figure 24. Block diagram representation of SCDPA.
53
MSc Thesis, Visweswaran Karunanithi
3.6. LINC (Linear amplification using non-linear components)
architecture
3.6.1. Working principle
LINC amplifiers, also known as out-phasing amplifiers were first developed by Henri Chireix
in 1935 [15] followed by L. F. Gaudernack’s work in 1938 [16]. These systems were proposed
to improve both efficiency and linearity of an AM- broadcast transmitter. The concept slowly
gained popularity over the years and in 1974 D. C. Cox [17] introduced the term LINC. As the
name suggests, linear amplification was obtained using non-linear components. The working
principle of this architecture can be explained using the block diagram shown in Figure 25.
An amplitude modulated (AM) and phase modulated (PM) signal (non-constant
envelope signal) such as 16-APSK is split using a signal processing unit into two PM (constant
envelope) signals S1(t) and S2(t). The signals S1(t) and S2(t) are then amplified independently,
denoted by G.S1(t) and G.S2(t). The amplified signals are then combined using a power
combiner to replicate the amplified AM and PM signal (in this case 16-APSK). This method
has gained popularity due to the recent developments in signal processing units. The main
advantage in using this system is that, as the signals being amplified by the PA blocks (amplifier
1&2) are constant envelope signals, thus it is possible to use high efficiency PAs and not worry
about the non-linearity caused by the PAs. The operation can be explained using the following
mathematical expressions:
The AM and PM modulated signal can be represented using the expression:
j( ( ))
( ) r(t) t t
ins t e  
 (5)
Where r(t) is the AM and can be represented as r(t) = rmax cos(θ(t)).
 ( ) ( )maxr
r(t)
2
j t j t
e e 
  (6)
Thus, the input non-constant envelope signal can be represented as a sum of two constant
envelope signals:
Figure 25. LINC block diagram.
54
MSc Thesis, Visweswaran Karunanithi
 ( ( ) ( )) ( ( ) ( ))maxr
( )
2
( ) 1( ) 2( )
j t t t j t t t
in
in
s t e e
s t S t S t
        
 
 
(7)
Where,
 
 
( ( ) ( ))max
( ( ) ( ))max
r
1( ) ,
2
r
2( )
2
j t t t
j t t t
S t e
S t e
  
  
 
 


(8)
The above mathematical expressions can be described using 16-APSK constellation diagram
as shown in Figure 26. In this case, the symbol 1100 is a non-constant envelope signal with its
amplitude 2.84 times lesser than the peak amplitude. This symbol is decomposed into constant
envelope signals S1(t) and S2(t) with appropriate out phasing angle θ(t) and constant amplitude
rmax/2.
max
( )
( ) arccos
r t
t
r

 
  
 
(9)
In the above case, the ratio between instantaneous amplitude and peak amplitude is (1/2.84),
thus using the above relation, θ(t) = 69.38 deg. In order to represent a constellation point on
the outermost ring (1000, 1001, …) the out-phasing angle θ(t) = 0 deg.
3.6.2. Advantages and dis-advantages of LINC architecture
Some of the advantages of LINC architecture are as follows:
- AM/AM and AM/PM distortions caused by individual PA blocks does not affect the
overall performance of the architecture.
Figure 26. 16-APSK decomposition into constant envelope signal.
55
MSc Thesis, Visweswaran Karunanithi
- Identical PAs are used in the two signal paths, leading to symmetrical designing of the
two paths. Thus, the matching network and the biasing network designed for one of the
paths can simply be replicated for the other.
- Most of the complexity is in splitting the signal component and synchronizing them,
which is taken care by the signal processing unit, the RF design complexity of the PAs
is much lesser compared to the other efficiency/linearity enhancement techniques.
Some of the disadvantages of the LINC architecture are as follows:
- Although using isolating power combiners gives the best performance in terms of
linearity, the overall efficiency of the system drops considerably in an isolating power
combiner such as WPC. Half of the power is dissipated in the isolation resistor,
bringing down the overall efficiency of the system.
- In the case of non-isolating power combiners (Chireix combiner), the linearity is
compromised for efficiency at large out-phasing angles as a result of load modulation.
3.7. Trade-off analysis
All of the efficiency and linearity enhancement techniques discussed so far have both
advantages and dis-advantages, there is no one architecture that suits all applications. Thus, a
trade-off needs to be performed to determine the most suitable architecture for nano-satellite
application. The parameters and its weights that are considered for a trade-off can vary based
on the application. The parameters considered for this trade-off are:
Complexity: This parameter is given the maximum weight of 5. While designing a system for
space application, it is very important to that the system is robust and less complex.
Circuit over-head: In some cases the over-head in accurately designing the biasing circuitry
or synchronization circuitry adds on to the complexity of the design. Thus, an architecture with
minimal over-head is preferred over the rest. This parameter is given a weight of 4.
Form-factor and performance in the power back-off: The next highest priority is given to
form-factor and the performance in power back-off with a weight of 3. As mass and volume is
a major constraint in nano-satellite design, it is important to choose an architecture that is less
voluminous. The form-factor of all of the architectures discussed above is considerably large
compared to the conventional operationclasses such as Class-A, B, C, etc. Thus, this parameter
is given a weight of 3. All of the architectures discussed are capable of performing equally
well in the power back-off levels associated with 16 and 32 APSK modulation schemes, thus
the weight is not high.
Efficiency and Linearity: Although the discussion so far has been mainly towards choosing a
highly efficient and linear architecture, based on the literature study it was seen that there was
not much to differentiate between the architectures. Thus, when it comes to trade-off analysis,
these parameters are not given a very high weight.
56
MSc Thesis, Visweswaran Karunanithi
Based on the weights assigned to each of the parameter, the highest an architecture can
score is 95. This following table gives the trade-off:
It can be seen from the above table that LINC architecture proves to be a better choice
for nano-satellites compared to the other efficiency and linearity enhancement architectures.
3.8. Conclusion
An analysis of different efficiency and linearity enhancement architectures were performed and
it could be seen that there was no one architecture that suited all applications. All the
architectures had a certain advantage over the other, thus a trade-off was performed on the
different architectures to see which one would suit nano-satellite application the best. The
parameters considered for the trade-off were: complexity/cost, design over-head, form-factor,
performance, linearity and efficiency.
Based on the trade-off, LINC architecture proved to be the most suitable architecture
for nano-satellite application due to its reduced complexity in the RF design, lower over-head
and smaller form factor.
Thus, it was decided to investigate the LINC PA in more detail to understand some of
the implementation challenges.
Table 10. Performance trade-off
57
MSc Thesis, Visweswaran Karunanithi
4
4. LINC Architecture
s explained in the previous section, the principle of operation of the LINC
architecture is by splitting the non-constant envelope signal (AM and PM) into two
constant envelope signals (PM signal) and amplify them individually. By doing
this, it is possible to drive the PAs at its peak efficiency. The efficiency of the
complete system is dependent on: efficiency of the individual PA blocks and the combiner
efficiency. On the other hand, as the PAs are driven by constant envelope signals, the linearity
of the system is mainly dependent on the linearity of the combiner. There are various choices
one can make while selecting an appropriate operating class for a PA and a power combining
technique. The subsequent sectionwill give an overview of how an appropriate operatingclass
and power combining technique was decided.
4.1. Class of operation
The operation classes of PAs in class AB, B, C, D, E, F, F-1
and J are possible to
implement on LINC architecture. Based on the literature, it was seen that LINC/Chireix
amplifiers were implemented using Class-B [18], C [19], AB and F [20]. The analysis
performed in [21] was taken as a reference to choose the best mode of operation. In the
analysis, the operating modes considered were Class-AB, B, C, CMCD (current mode Class-
D), VMCD (Voltage mode Class-D), E, F, F-1
and J. Simulations were performed on these
Classes using CREE GaN HEMT (CGH27015), it was designed for an output power of 40 dBm
(10 Watts) at 900 MHz for 16-QAM modulation. Details about the simulation models can be
found in [21].
The simulations were done for a symbol rate of 3.84 Msps and roll-off of the filter equal to
0.35.
The drain efficiencies of the different classes are as follows:
Table 11. Drain Efficiency (DE) for different Classes [21]
Class AB B C CMCD VMCD E F F-1
J
DE
(%)
60 62 64 83 71 88 71 69 75
The PAE for these classes were as follows:
A
58
MSc Thesis, Visweswaran Karunanithi
Table 12. Simulation results from [21 ] for PAE of the different amplifier classes.
Class AB B C CMCD VMCD E F F-1
J
PAE
(%)
56 55 53 79 69 84 67 65 69
The publication gives a detailed analysis of efficiencies simulated for two different stub
compensation angles: 10 deg and 30 deg.
The simulated output spectrum which helps is shown in Figure 27.
Thus, based on the analysis and results from this publication, it was seen that Class-F, F-1
and
CMCD give the best performance in terms of both efficiency and linearity, but CMCD requires
two transistors for their implementation whereas Class-F or F-1
require a single transistor per
PA.
For this work, Class-F implementation was chosen and simulations were performed to
verify the results from literature.
4.1.1. Class-F implementation
Class-F PAs use the technique of shaping the drain/collector waveform to achieve a better
efficiency. Shaping of the waveform is done by appropriate harmonic termination. In the case
of class-F mode of operation, the drain is presented with a short-circuit termination at the even
harmonics of drain voltage/current, open-circuit termination at odd harmonics and the desired
load based on the required output power at the fundamental frequency. This helps in flattening
of the voltage waveform, allowing the majority of the drain current to flow when the drain
Figure 27. Output spectrum of 16-QAM for different classes of operation. [21]
59
MSc Thesis, Visweswaran Karunanithi
voltage is low, resulting in reduced power dissipation. By lowering the power dissipation, the
efficiency is improved. The increase in efficiencyis directlyrelated to the number of harmonics
handled [22]. Figure 28 shows a functional representation of Class-F mode of operation. By
implementing such harmonic tuning at the output, it is possible to shape the drain voltage into
a square wave, which helps in reducing the overlap between the current and voltage waveform.
An ideal case drain voltage and current waveform is shown in Figure 29. The expression for
drain voltage and current can be obtained as explained in [22], but in practical implementations
it is not possible to handle all the harmonic components. From [23] it is seen that, by handling
only the first four harmonics, it was possible to attain a PAE larger than 80%.
Based on the equations derived in [22], γv and γi were used to fine tune the load resistance
corresponding to the desired output power.
Figure 28. Functional representation of Class-F PA
60
MSc Thesis, Visweswaran Karunanithi
The Figure 29 shows a schematic of the output harmonic matching circuit simulated on ADS
using ideal components and [figure] shows the output impedance seen by the PA at different
harmonic frequencies. The circuit is designed for a fundamental frequency at 900 MHz.
The implementation of the Class-F amplifier was done using a GaN HEMT from CREE
CGH27015F.
Figure 29. Class-F harmonic matching circuit.
Figure 30. Load impedance by the drain of the transistor at different harmonic frequencies.
61
MSc Thesis, Visweswaran Karunanithi
The IV characteristics the used HEMT is as shown in [figure]. Using the large signal model of
CGH27015F from CREE, the simulations were performed on ADS.
The harmonic impedance matching circuit shown at the output was implemented at the output
of the HEMT and the following figure [figure] shows the drain current and voltage waveforms.
The final implementation of the class-F PA was done based on the [20]. More details about
the final design integrated with the Chireix combiner is elaborated in subsequent sections.
Figure 31. IV characteristics of CGH27015F.
Figure 32. Drain voltage and current waveform for a Class-F PA implemented using
CGH27015F GaN HEMT.
62
MSc Thesis, Visweswaran Karunanithi
4.2. Different power combining techniques
Once the out-of-phase signals are independently amplified using Class-F PAs, the signals are
then combined to produce the amplified non-constant envelope signal. The combiner network
acts as an adder, enabling the net output amplitude to be controlled via the relative phase of the
two non-constant envelope signals. There are various methods of power combining from the
literature. The various methods of power combining for LINC system are shown in Figure 33
The method of combining can broadly be classified into two types: using matched combiners
and using un-matched combiners. A matched combiner would provide high isolation between
the two input ports and provides a constant load impedance to each of the PA for all out-phasing
angles. Examples of matched combines are Wilkinson Power Combiner (WPC), Hybrid
couplers, rat-race, etc. Although the performance of such combiners are very good in-terms of
linearity, the efficiency of combining is traded for the isolation. Almost half the power is
dissipated in the isolation resistor. Thus, these techniques were not very popular for the final
implementation of LINC, they were mainly used for calibration purpose. In recent time, use of
such combiners have re-gained popularity as methods such as RF-DC conversion can be used
to recycle the power dissipated in the isolation resistor [24]. Although this method shows
promising performance, the circuit over head is quite large and best suited for more than four-
way power combining. Due to these reason, this method was not incorporated as the power
combining network but ADS simulations were performed to quantify the performance of
different matched power combining techniques.
Figure 33. RF power combining techniques for LINC.
63
MSc Thesis, Visweswaran Karunanithi
Figure 34 shows the two out-of phase input signals V1(t) & V2(t), combined signal (vload) and
efficiency as a function of out-phasing angle.
Although WPC can provide a good isolation between the two input ports, sometime it is not
good enough to completely isolate the two ports. Some of the commercially available WPC s
such as ZAPD-30-S+ (from Mini-circuits) provide an isolationbetween 12 and 15 dB. The can
Figure 34. WPC as a power combining network for LINC
Figure 35. Plot of the two out-of-phase signals, combiner voltage waveformand efficiency of WPC as a function of
outphasing angle.
64
MSc Thesis, Visweswaran Karunanithi
lead to power getting coupled into the other port and seen at the output of the PA, this leads to
degradation in the performance of the PA. This was noticed during the measurement done
latter. In-order to over-come this, a quadrature coupler stage can be implemented before a
WPC and simulation results show a significant improvement in the isolation. Figure 36 shows
the schematic of this setup,
Figure 37 shows the coupling between the ports.
The simulation was done for a CPW (co-planar waveguide) on a FR-4 substrate. The substrate
parameters can be seen in Figure 38. The efficiency and voltage waveforms can be seen in
Figure 39
Figure 36. Improvement of isolation by introducing a quadrature coupler between the PA and WPC.
Figure 37. Coupling between the two ports.
65
MSc Thesis, Visweswaran Karunanithi
Apart from these methods where it is possible to attain high isolation at the cost of efficiency
at larger out-phasing angles, there are methods of power combining using unmatched networks.
Chireix combiners is one such example of a lossless combiners with low isolation. A Chireix
combiner mainly comprises of two simple λ/4 lines with one end having a compensation
element (capacitive/inductive) and the other end connected to the load. The concept of using
Chireix combiner in LINC can be understood using the schematic representation of the
combiner shown in Figure 39. [25].
The output of the PAs are represented by voltage sources V1 and V2. These voltage sources
can be expressed in phasor form as follows:
( )
1 .
(cos sin )
j
o
o
V V e
V j

 

 
(10)
Figure 38. Efficiency and voltage waveforms when a quadrature coupler is introduced between the PA and combiner.
Figure 39. Schematic representation of Chireix combiner.
66
MSc Thesis, Visweswaran Karunanithi
( )
2 .
(cos sin )
j
o
o
V V e
V j

 


 
(11)
Here, θ is the out-phasing angle that can vary between 0 and 90 deg.
Thus, the voltage across the resistor is:
| 1 2 | 2 . sin( )L oV V V V j    (12)
Based on the derivation from [25], the impedance seen by the two amplifiers can be expressed
as:
 1 1 cot
2
LR
Z j   (13)
 2 1 cot
2
LR
Z j   (14)
The corresponding admittance are:
2
1
2.sin sin2
L L
Y j
R R
 
  (15)
2
2
2.sin sin2
L L
Y j
R R
 
  (16)
It can be seen that the susceptance seen by both the PAs are different and is a function of the
out-phasing angle. The susceptance seen by the PA is depicted in Figure 40.
As these susceptances are a function of the out-phasing angle, compensation can be provided
based on the PDF of the modulated signal. The relation between out-phasing angle and
amplitude is described in equation (9). The value of the inductor and capacitors can be
calculated using the imaginary part of the conjugate of admittance from equation (15) and (16)
.
Figure 40. Chireix combiner as seen at the inputs when fed out-of-phase signals.
67
MSc Thesis, Visweswaran Karunanithi
Based on the above equations, a chireix combiner was designed and simulated on ADS, the
substrate used was FR4 and the transmission lines used were CPW lines. LineCalc tool inADS
was used to calculate the dimensions of the CPW corresponding to the electrical length and
characteristic impedance of the line. Figure 41 shows the schematic of the Chireix combiner.
The above design is implemented for a compensation angle of 10 degrees. This can be seen
from the S-parameter simulation results shown in Figure 42. The initial phase difference
without compensation stubs between the input port-1/2 to output port-3 was measured to be
Figure 41. Chireix combiner design on FR4 using CPW.
Figure 42. S-parameter simulation of the Chireix combinerwith compensation stubs placed
at 10 deg.
68
MSc Thesis, Visweswaran Karunanithi
118.2 deg. It can be seen that by adding 10 degree compensation stub, Phase(S(2,3)) has
reduced by 10 degrees and the Phase(S(1,3)) has increased by 10 degrees.
A comparison was done between WPC and Chireix combiner to quantify the efficiency of the
two combiners as a function of out-phasing angle, this plot is shown in Figure 43.
It can be seen from the plot in figure that the efficiency is improved for larger out-phasing
angles when a Chireix combiner is used. The main drawback with Chierix combiner while
compared to WPC is that the linearity is poor at larger out-phasing angles due to the fact that
Chireix combiner provides poor isolation between the two input ports. This can be seen in
Figure 44 which shows the poor isolation exhibited by the Chierix combiner.
Figure 43. Efficiency vs out-phasing angle comparison between Chireix combiner and
WPC.
Figure 44. Comparison of coupling between Chireix combiner and WPC.
69
MSc Thesis, Visweswaran Karunanithi
One other power combining technique that has surfaced in recent time is, spatial power
combining. In this technique, the output of the PAs are fed directly to two separate antennas
and the radiated power from the individual antennas combine in the air to form the non-constant
envelope signal. Using the idea from [26], simulations were performed on a UHF antenna
combiner on nano-satellites. A simulation on FEKO was done on crossed dipole for different
out-phasing angles to see its effect on the radiation pattern.
The radiation pattern when the dipoles are fed with signals having an out-phasing angle of 45
degrees is shown in Figure 45
The radiation pattern of the crossed dipole when fed with zero out-phasing angle is shown in
Figure 46.
Figure 45. Radiation pattern of crossed dipole on a 3U cube-sat fed with an out-phasing angle of 45 degrees.
70
MSc Thesis, Visweswaran Karunanithi
It can be seen that the beam-width has increased and the peak gain has decreased from 3.4 dBi
to 3 dBi.
Thus, it could be seen that the main effect on the radiation pattern due to the change in out-
phasing angle is the variation in the beam-width. The gain variation in the bore side was not
much. The crossed dipole arrangement provides fair isolation between the two antennas.
Figure 46. Radiation pattern of the crossed dipole when fed with signals with 0 degree outphasing angles.
Figure 47. Radiation pattern of crossed dipole fed with signals with 90 degrees out-phasing angle.
71
MSc Thesis, Visweswaran Karunanithi
Although, the simulation results show promising results, further analysis needs to be
performed to use it in a LINC architecture, making this an interesting topic for future work.
After analyzing various amplifier cells and different power combining techniques, it was
decided implement the LINC architecture using Class-F PA and a Chireix combiner. In order
to show a proof of concept, a LINC architecture designed by CATENA microelectronics BV
was used [27] This board had implemented Class-F PA cells and a Chireix combiner, thus this
board was used to validate the above simulation results and propose possible improvement to
the design. The further sections will describe the measurements performed on the LINC PA
for different modulation schemes.
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)
Final_29_09_v3 (1)

More Related Content

Similar to Final_29_09_v3 (1)

GLASSNER_FINAL_REPORT_(08-13-14)
GLASSNER_FINAL_REPORT_(08-13-14)GLASSNER_FINAL_REPORT_(08-13-14)
GLASSNER_FINAL_REPORT_(08-13-14)Austin Glassner
 
110015978_WirelessChannelsInInceOS
110015978_WirelessChannelsInInceOS110015978_WirelessChannelsInInceOS
110015978_WirelessChannelsInInceOSTejas Unnikrishnan
 
final year project
final year projectfinal year project
final year projectshiola kofi
 
elec_2016_nguyen_huy
elec_2016_nguyen_huyelec_2016_nguyen_huy
elec_2016_nguyen_huyNguyen Huy
 
Thesis_Underwater Swarm Sensor Networks
Thesis_Underwater Swarm Sensor NetworksThesis_Underwater Swarm Sensor Networks
Thesis_Underwater Swarm Sensor NetworksGunilla Burrowes
 
Audio signal transmission using Li-Fi.
Audio signal transmission using Li-Fi.Audio signal transmission using Li-Fi.
Audio signal transmission using Li-Fi.Divahar Thangavel
 
Performance improvement of mimo mc cdma system using equalization, beamformin...
Performance improvement of mimo mc cdma system using equalization, beamformin...Performance improvement of mimo mc cdma system using equalization, beamformin...
Performance improvement of mimo mc cdma system using equalization, beamformin...Tamilarasan N
 
A FAULT-TOLERANCE LINGUISTIC STRUCTURE FOR DISTRIBUTED APPLICATIONS
A FAULT-TOLERANCE LINGUISTIC STRUCTURE FOR DISTRIBUTED APPLICATIONSA FAULT-TOLERANCE LINGUISTIC STRUCTURE FOR DISTRIBUTED APPLICATIONS
A FAULT-TOLERANCE LINGUISTIC STRUCTURE FOR DISTRIBUTED APPLICATIONSVincenzo De Florio
 
Vezzoli system design for sustainable energy for all nanjing_(37)_2014.05
Vezzoli system design for sustainable energy for all nanjing_(37)_2014.05Vezzoli system design for sustainable energy for all nanjing_(37)_2014.05
Vezzoli system design for sustainable energy for all nanjing_(37)_2014.05LeNS_slide
 
Design horn-antenna using hfss
Design horn-antenna using hfssDesign horn-antenna using hfss
Design horn-antenna using hfssMusbiha Binte Wali
 
Presentation by Heather Wharrad
Presentation by Heather WharradPresentation by Heather Wharrad
Presentation by Heather WharradVideoguy
 
Ph d model-driven physical-design for future nanoscale architectures
Ph d model-driven physical-design for future nanoscale architecturesPh d model-driven physical-design for future nanoscale architectures
Ph d model-driven physical-design for future nanoscale architecturesCiprian Teodorov
 
Analysis Of Image Compression Methods Based On Transform And Fractal Coding
Analysis Of Image Compression Methods Based On Transform And Fractal CodingAnalysis Of Image Compression Methods Based On Transform And Fractal Coding
Analysis Of Image Compression Methods Based On Transform And Fractal CodingJim Webb
 
A Dynamic Middleware-based Instrumentation Framework to Assist the Understand...
A Dynamic Middleware-based Instrumentation Framework to Assist the Understand...A Dynamic Middleware-based Instrumentation Framework to Assist the Understand...
A Dynamic Middleware-based Instrumentation Framework to Assist the Understand...Luz Martinez
 

Similar to Final_29_09_v3 (1) (20)

GLASSNER_FINAL_REPORT_(08-13-14)
GLASSNER_FINAL_REPORT_(08-13-14)GLASSNER_FINAL_REPORT_(08-13-14)
GLASSNER_FINAL_REPORT_(08-13-14)
 
Cse 2008 7
Cse 2008 7Cse 2008 7
Cse 2008 7
 
110015978_WirelessChannelsInInceOS
110015978_WirelessChannelsInInceOS110015978_WirelessChannelsInInceOS
110015978_WirelessChannelsInInceOS
 
SI Thesis
SI ThesisSI Thesis
SI Thesis
 
Holo zoo
Holo zooHolo zoo
Holo zoo
 
final year project
final year projectfinal year project
final year project
 
V01 i010402
V01 i010402V01 i010402
V01 i010402
 
elec_2016_nguyen_huy
elec_2016_nguyen_huyelec_2016_nguyen_huy
elec_2016_nguyen_huy
 
Thesis_Underwater Swarm Sensor Networks
Thesis_Underwater Swarm Sensor NetworksThesis_Underwater Swarm Sensor Networks
Thesis_Underwater Swarm Sensor Networks
 
Audio signal transmission using Li-Fi.
Audio signal transmission using Li-Fi.Audio signal transmission using Li-Fi.
Audio signal transmission using Li-Fi.
 
Performance improvement of mimo mc cdma system using equalization, beamformin...
Performance improvement of mimo mc cdma system using equalization, beamformin...Performance improvement of mimo mc cdma system using equalization, beamformin...
Performance improvement of mimo mc cdma system using equalization, beamformin...
 
1324549
13245491324549
1324549
 
A FAULT-TOLERANCE LINGUISTIC STRUCTURE FOR DISTRIBUTED APPLICATIONS
A FAULT-TOLERANCE LINGUISTIC STRUCTURE FOR DISTRIBUTED APPLICATIONSA FAULT-TOLERANCE LINGUISTIC STRUCTURE FOR DISTRIBUTED APPLICATIONS
A FAULT-TOLERANCE LINGUISTIC STRUCTURE FOR DISTRIBUTED APPLICATIONS
 
Vezzoli system design for sustainable energy for all nanjing_(37)_2014.05
Vezzoli system design for sustainable energy for all nanjing_(37)_2014.05Vezzoli system design for sustainable energy for all nanjing_(37)_2014.05
Vezzoli system design for sustainable energy for all nanjing_(37)_2014.05
 
Design horn-antenna using hfss
Design horn-antenna using hfssDesign horn-antenna using hfss
Design horn-antenna using hfss
 
Presentation by Heather Wharrad
Presentation by Heather WharradPresentation by Heather Wharrad
Presentation by Heather Wharrad
 
Thesis
ThesisThesis
Thesis
 
Ph d model-driven physical-design for future nanoscale architectures
Ph d model-driven physical-design for future nanoscale architecturesPh d model-driven physical-design for future nanoscale architectures
Ph d model-driven physical-design for future nanoscale architectures
 
Analysis Of Image Compression Methods Based On Transform And Fractal Coding
Analysis Of Image Compression Methods Based On Transform And Fractal CodingAnalysis Of Image Compression Methods Based On Transform And Fractal Coding
Analysis Of Image Compression Methods Based On Transform And Fractal Coding
 
A Dynamic Middleware-based Instrumentation Framework to Assist the Understand...
A Dynamic Middleware-based Instrumentation Framework to Assist the Understand...A Dynamic Middleware-based Instrumentation Framework to Assist the Understand...
A Dynamic Middleware-based Instrumentation Framework to Assist the Understand...
 

Final_29_09_v3 (1)

  • 1.
  • 2.
  • 3. In partial fulfilment of the requirements for the degree of MASTER OF SCIENCE in MICRO-ELECTRONICS at the Delft University of Technology, to be defended publicly on Wednesday September 24, 2014 at 15:00hrs Supervisors : 1) Prof. Ir. Dr. C.J.M Chris Verhoeven, Associate Professor, TU Delft 2) Ir. Waldemar Lubbers, RF & E head, Innovative Solutions In Space B.V, Delft Thesis Committee : 1) Prof. Ir. Dr. C.J.M Chris Verhoeven (TU Delft) 2) Ir. Waldemar Lubbers (ISIS BV) 3) Prof. Dr. Nick van der Meijs (TU Delft) 4) Ir. P.P.(Prem) Sundaramoorthy (TU Delft) An electronic version of this thesis is available at http://repository.tudelft.nl/
  • 4. The work in this thesis was supported by ISISpace B.V. Their cooperation is hereby gratefully acknowledged. Copyright c All rights reserved.
  • 5. [Dedication] I dedicate this work to my mom, dad, vikku and my “Small Wall friends”….
  • 6.
  • 7. Acknowledgement I would like to express my sincere gratitude to the management of Innovative Solutions In Space B.V for allowing me to work on such a fascinating research topic for my Master’s thesis. They have provided me with some of the best possible resources to carry out my work smoothly. I would like to thank my university supervisor Prof. Chris and company supervisor Waldemar for providing me the support, guidance and encouragement through the entire period of my thesis. I am sure their guidance will go a long way in my career as a researcher. The RF&E team at ISISpace B.V provided a great environment to exchange ideas. I would like to thank Arancha, Eelco, Johan, Javier, Stefano, Wouter2 and Waldemar for making me feel a part of the RF&E team and provide guidance and support and some very critical times of my thesis. I would like to express my gratitude to Ernst Habekotte from CATENA microelectronics B.V for supporting this thesis at a crucial time by providing the LINC development boards to understand the design and introducing me to the Mixed signal group at TU Eindhoeven. I am grateful Prof. Baltus, P.G.M and Rainier from the mixed signal group at TU Eindhoeven for allowing me to use the measurement setup there and providing an un- conditional support through my entire stay. I would like to thank my parent back in India who always had the confidence in me, no matter what and always would be there for me as an emotional support. A very special thanks to SKYPE and Facebook, I have not missed my home country! My twin brother has played a very important part during my studies, he has been my pillar of support ever since I came to Netherlands. I have almost never missed home. It will not be complete without thanking my friends here in Delft. I surely did make a lots of friends, they have supported me through the entire two years of my studies. A special thanks to the “Small Wall” group of friends: Harshitha, Abhimanyu, Adithya, Arun, Arul, Manjunath, Phani, Sriram and Sumedh. You guys were simply amazing. I would like to thank Arun for helping me out in making the cover page of this thesis. -Vishu
  • 8.
  • 9. Preface The fast growing interest in nano-satellite development and the use of nano-satellite for complex missions has led to an increase in the downlink data-rate. The availability of free bands (amateur frequency bands: VHF and UHF) is getting limited and more challenging to do the frequency coordination. This can be attributed to the surge in number of nano-satellite being launched. There has been 430% increase in the number of satellites launched in 2013 compared to 2012. This trend has led the nano-satellite developers to start looking into higher frequency bands and S-band has been a popular choice. But, going by the trend, frequency coordination in the S-band can also get challenging. Thus, there is a need to develop a nano- satellite transmitter that uses spectrally efficient modulation scheme, at the same time it has to be power efficient and be compatible with some of the existing communication standards. This work deals with developing the requirements on a nano-satellite transmitter based on mission case study, look at the need for efficiency and linearity enhancement technique for the transmitter, select the most appropriate architecture for nano-satellite application and show a proof of concept using a prototype. Using system engineering approach the efficiency and linearity enhancement technique that was chosen was LINC architecture. ADS simulations were performed to understand different configurations of LINC and finally, measurements were performed on the chosen configuration to characterize its efficiency and linearity. 16-QAM, 16-APSK, 32-QAM, 32- APSK and 64-QAM modulation schemes were implemented and tested. The best EVM was obtained for 16-APSK with 20 degrees compensation stub in the Chireix combiner which was 0.27% and the best efficiency was obtained for 16-APSK modulation scheme which was implemented without any compensation stubs. Based on the lessons learnt during simulations and measurements, recommendations are provided to improve the LINC configuration and improve the reliability of the measurement setup. Visweswaran Karunanithi Delft, September 2014
  • 10.
  • 11. 11 MSc Thesis, Visweswaran Karunanithi Contents [Dedication]...............................................................................................................................5 Acknowledgement.......................................................................................................................7 Preface.......................................................................................................................................9 Table of figures........................................................................................................................15 1. Introduction......................................................................................................................20 1.1. Current trends in nano-satellites......................................................................................21 1.2. Conclusion of the analysis................................................................................................24 1.3. Problem statement.............................................................................................................25 2. Nano-Satellite transmitters for the future........................................................................28 2.1. CCSDS and ECSS recommendations................................................................................28 2.1.1. Frequency band allocation & constraints ...........................................................29 2.1.2. Transmitter spurious emission and harmonic levels............................................31 2.1.3. Recommended modulation schemes.....................................................................31 2.1.4. Conclusion of the CCSDS and ECSS recommendations......................................32 2.2. Nano-satellite mission case study.....................................................................................33 2.2.1. Spacecraft housekeeping data downlink requirement .........................................34 2.2.2. Payload downlink data-rate requirement for a EES service ...............................34 2.2.3. Downlink data-rate requirement for nano-satellite providing data services ......36 2.3. Link budget analysis.....................................................................................................38 2.4. Summary of the case study and conclusion..................................................................40 3. High efficiency transmitter architectures ........................................................................43 3.1. Overview...........................................................................................................................43 3.1.1. Characterization of linearity and efficiency ..............................................................45 3.2. Doherty architecture....................................................................................................47 3.2.1. Working Principle................................................................................................47 3.2.2. Advantages and dis-advantages of Doherty architecture....................................49 3.3. Kahn/EER (Envelope Elimination and Restoration) ...................................................49 3.3.1. Working principle of Kahn/EER architecture......................................................49 3.3.2. Advantages and dis-advantages of Kahn architecture ........................................50 3.4. Envelope Tracking (ET) architecture ..........................................................................50 3.4.1. Working principle of ET.......................................................................................50 3.4.2. Advantages and disadvantages of ET architecture..............................................51
  • 12. 12 MSc Thesis, Visweswaran Karunanithi 3.5. Switched Capacitor Digital Power Amplifiers SCDPA...............................................51 3.5.1. Working principle of SCDPAs .............................................................................51 3.5.2. Advantages and dis-advantages of SCDPAs........................................................52 3.6. LINC (Linear amplification using non-linear components) architecture.........................53 3.6.1. Working principle ......................................................................................................53 3.6.2. Advantages and dis-advantages of LINC architecture..............................................54 3.7. Trade-off analysis .............................................................................................................55 3.8. Conclusion ........................................................................................................................56 4. LINC Architecture............................................................................................................57 4.1. Class of operation.............................................................................................................57 4.1.1. Class-F implementation.............................................................................................58 4.2. Different power combining techniques........................................................................62 4.3. Experimental setup............................................................................................................72 4.3.1. Single RF source measurement setup ........................................................................72 4.3.2. Modifications done to the original setup ...................................................................76 4.4. Catena LINC/Chireix PA architecture..............................................................................77 5. Measurement Results.......................................................................................................81 5.1. Modulation schemes tested...............................................................................................81 5.2. Performance characterization: EVM................................................................................82 5.2.1. 16-QAM modulation scheme................................................................................82 5.2.2. 16-APSK modulation scheme...............................................................................84 5.2.3. 32-QAM modulation scheme................................................................................86 5.2.4. 32-APSK modulation scheme...............................................................................88 5.2.5. 64-QAM modulation scheme................................................................................89 5.2.6. Summary of EVM and ACLP measurement. ........................................................90 5.3. Performance characterization: LINC/ Chireix efficiency.................................................91 6. Conclusion and Recommendations for future work.........................................................93 6.1 Recommendations on communication standards...............................................................93 6.2 Recommendations for a LINC architecture.......................................................................94 6.2.1. PA cell..................................................................................................................94 6.2.2. Combiner architecture.........................................................................................95 6.3 Recommendations for the experimental setup ...................................................................95 6.4 Future work........................................................................................................................96 Nomenclature...........................................................................................................................99
  • 13. 13 MSc Thesis, Visweswaran Karunanithi Bibliography ..........................................................................................................................101
  • 14.
  • 15. Table of figures Figure 1. Number of nano-satellite launches..........................................................................21 Figure 2. (Left) Mission types between 2003 and 2012, (right) Mission types in 2013. ........22 Figure 3. (Left) data-rates used by nano-satellitemission between 2003 and 2012 (right) data-rates used by nano-satellite missions in 2013.................................................................22 Figure 4. (Left) popular frequency bands used between 2003 and 2012. (Right) Frequency bands used by nano-satellite missions in 2013........................................................................23 Figure 5. Maximum allowable bandwidth vs symbol rate......................................................30 Figure 6. Spectral emission limits...........................................................................................31 Figure 7. Pictorial representationof the relationbetween downlink data-rate, service provided and type of payload...................................................................................................33 Figure 8. Pictorial representation of VDES [].......................................................................36 Figure 9. Link budget calculation of spacecraft housekeeping data downlink. .....................38 Figure 10. Link margin for different elevation angles.............................................................38 Figure 11. Link budget for an optical payload data downlink. ..............................................39 Figure 12. Link margin vs elevation angles for optical payload downlink. ............................39 Figure 13. Link budget for VED-Sat downlink.......................................................................40 Figure 14. Link margin vs elevation for VDE-Sat..................................................................40 Figure 15. Block-diagram of a transmitter with digital modulation. .....................................43 Figure 16. Linearity vs Efficiency trade-off for conventional PA classes..............................44 Figure 17. Efficiency and linearity enhancement techniques.................................................45 Figure 18. PDF of 16-APSK modulation scheme, implemented using an SRRC filter with a roll-off of 0.5(4000 sps) ...........................................................................................................46 Figure 19. Doherty PA architecture........................................................................................47 Figure 20. Schematic representation of Doherty.....................................................................47 Figure 21. Input power vs Output power in linear scale [11]................................................48 Figure 22. Block diagram representationof a modern implementationof the Kahn architecture..............................................................................................................................50 Figure 23. Block diagram of ET architecture.[11].................................................................51 Figure 24. Block diagram representation of SCDPA. ............................................................52 Figure 25. LINC block diagram..............................................................................................53
  • 16. 16 MSc Thesis, Visweswaran Karunanithi Figure 26. 16-APSK decomposition into constant envelope signal........................................54 Figure 27. Output spectrum of 16-QAM for different classes of operation. [21]...................58 Figure 28. Functional representation of Class-F PA.............................................................59 Figure 29. Class-F harmonic matching circuit. .....................................................................60 Figure 30. Load impedance by the drain of the transistor at different harmonic frequencies. ..................................................................................................................................................60 Figure 31. IV characteristics of CGH27015F. .......................................................................61 Figure 32. Drain voltage and current waveform for a Class-F PA implemented using CGH27015F GaN HEMT. .......................................................................................................61 Figure 33. RF power combining techniques for LINC. ..........................................................62 Figure 34. WPC as a power combining network for LINC ....................................................63 Figure 35. Plot of the two out-of-phase signals,combiner voltage waveform and efficiency of WPC as a function of outphasing angle...................................................................................63 Figure 36. Improvement of isolationby introducing a quadrature coupler between the PA and WPC. .................................................................................................................................64 Figure 37. Coupling between the two ports............................................................................64 Figure 38. Efficiency and voltage waveforms when a quadrature coupler is introduced between the PA and combiner..................................................................................................65 Figure 39. Schematic representation of Chireix combiner.....................................................65 Figure 40. Chireix combiner as seen at the inputs when fed out-of-phase signals. ...............66 Figure 41. Chireix combiner design on FR4 using CPW. ......................................................67 Figure 42. S-parameter simulation of the Chireix combiner with compensation stubs placed at 10 deg...................................................................................................................................67 Figure 43. Efficiency vs out-phasingangle comparison between Chireix combiner and WPC. ..................................................................................................................................................68 Figure 44. Comparison of coupling between Chireix combiner and WPC. ............................68 Figure 45. Radiation pattern of crossed dipole on a 3U cube-sat fed with an out-phasing angle of 45 degrees..................................................................................................................69 Figure 46. Radiation pattern of the crossed dipole when fed with signals with 0 degree outphasing angles. ...................................................................................................................70 Figure 47. Radiation pattern of crossed dipole fed with signals with 90 degrees out-phasing angle.........................................................................................................................................70 Figure 48. Complete measurement setup................................................................................72
  • 17. 17 MSc Thesis, Visweswaran Karunanithi Figure 49. Signal flow diagram..............................................................................................73 Figure 50. Initial setup for calibration...................................................................................74 Figure 51. Screenshot of calibration measurement.................................................................74 Figure 52. Setup to measure the non-linearity caused by mini-circuits PAs..........................75 Figure 53. Picture of the measurement setup.........................................................................75 Figure 54. Variation in the constant envelope signal due to the previous SCS algorithm.....76 Figure 55. Constant envelope signals after the modified SCS algorithm................................77 Figure 56. LINC/Chireix PA from CATENA Microelectronics..............................................78 Figure 57. Schematic representation of the LINC/Chireix PCB from CATENA....................79 Figure 58. Constellationdiagrams of 16-QAM, 16-APSK, 32-QAM, 32-APSK and 64-QAM. ..................................................................................................................................................81 Figure 59. Received constellationdiagram of 16-APSK when no compensation stubs (0 degrees compensation).............................................................................................................82 Figure 60. Output spectrum of amplified 16-QAM..................................................................83 Figure 61. 16-QAM constellation diagram with 20 degrees compensation stub. ..................83 Figure 62. Output spectrum of amplified 16-QAM with stubs................................................84 Figure 63. Constellation diagram of 16-APSK modulation without compensation stubs......84 Figure 64. Output spectrum of 16-APSK modulation without compensation stubs. ..............85 Figure 65. 16-APSK constellation diagram with a compensation stub of 20 degrees. ..........85 Figure 66. Output power spectrum of 16-APSK with a compensation stub of 20 degrees.....86 Figure 67. 32-QAM without compensation stubs. ..................................................................86 Figure 68. Output power spectrum of 32-QAM without compensation..................................87 Figure 69. Constellation diagram of 32-QAM with compensation stub at 20 degrees..........87 Figure 70. Output power spectrum of 32-QAM with a compensation at 20 degrees..............88 Figure 71. Constellation diagram of 32-APSK with 20 degrees compensation stubs............88 Figure 72. Output power spectrum of 32-APSK modulated signal with a compensation stub at 20 degrees............................................................................................................................89 Figure 73. Constellationdiagram of 64-QAM with 20 degree compensation in the LINC/Chireix amplifier............................................................................................................89 Figure 74. Output power spectrum of 64-QAM amplifiedby LINC/Chireix PA with 20 degrees compensation stubs.....................................................................................................90 Figure 75. Harmonic matching circuit for the Class-F PA on CATENA LINC board............94 Figure 76. Proposed measurement setup................................................................................95
  • 18. 18 MSc Thesis, Visweswaran Karunanithi Figure 77. Hybrid-Chireix combiner......................................................................................96
  • 19.
  • 20. 20 MSc Thesis, Visweswaran Karunanithi 1 1. Introduction he popularity of nano-satellites among universities and space research organizations has grown quite rapidly over the last decade. Nano-satellite are a class of small satellites that weigh less than 10 Kilograms, having almost all of the subsystems present in a larger satellite and capable of mimicking larger satellites with simpler payloads. The interest for nano-satellites among universities started in 1999 through a CubeSat1 project that began at California Polytechnic State University (Cal Poly) and Stanford University’s Space Systems Development Laboratory (SSDL). These projects were mainly intended to be used by the Universities as a tool to learn satellite development and understand the technology of developing hardware that can sustain the harsh space environment. These missions mainly made use of Commercially Of-The-Shelf (COTS) components which are industrial grade and easilyavailable. This reduced the development cost and time of the mission. On the other hand, it reduced the life-time of a mission and also a high probability of failure. The failures and lessons learnt from past nano-satellite missions has helped in coming-up with mature designs that help in longer mission duration. An example of this is Delfi-C32 that was developed by a team from Technical University of Delft and launched on 28th April 2008. This mission is a good example of a COTS based but robust design that has survived multiple solar events and is still operational. The focus of nano-satellite missions has slowly started to shift from simple technology demonstration missions to more complex industry driven technology demonstration, science, military and government missions. This clearly shows a paradigm shift in the way nano- satellites are being perceived and the confidence in implementing more and more complex missions on nano-satellites. One of the most important reason for this is the standardization in shape, satellite-bus subsystems such as antenna systems, electrical power systems, On-board computer, etc. This has helped the satellite developers to concentrate mainly on payload development and integrate the payload with the satellite bus components and subsystems that are available on the CubeSat market. They are developed by companies such as Innovative Solutions In Space BV, GomSpace, etc. As the complexity of nano-satellite missions grow, the amount of data generated by the spacecraft (payload data and telemetry) has a direct influence on the downlink data-rate of the communication link. The subsequent sections in Chapter-1 give an overview of the current 1 CubeSats is a standardization introduced by Calpoly that signifies a satellite that is in the shape of a cube. A single unit of CubeSat is acube of 10 × 10 × 10 = 1000 cm3 byvolume andamass lesser than1.33 kg, also called 1U. This standardizationhas mainlyhelpedinusingthe same type of deployment system for various classes such as 1U, 2U and 3U. 2 http://www.delfispace.nl/delfi-c3 T
  • 21. 21 MSc Thesis, Visweswaran Karunanithi trends in nano-satellites, the paradigm shift in the nano-satellite mission types, how this paradigm shift has caused the missions to start using high data rate transmitters and conclude with the problem statement of this thesis work. In Chapter-2 a few mission cases are considered to quantize the requirements posed by complex missions on the transmitter design with the help of link budget calculations, state the CCSDS (Consultative Committee for Space Data Systems) and ECSS (European Cooperation on Space Standardization) recommendations that have to be followed in-order to design a transmitter that is compatible with a large number of ground stations around the world that follow these standards and conclude with recommendations. Chapter-3 will elaborate about the challenges in developing a high efficiency transmitter that support the modulation schemes recommended in CCSDS space link protocol over ETSI DVB.S2, some of the common efficiency and linearity enhancement techniques and conclude with a trade-off that was performed on these techniques to find a suitable candidate for nano- satellite transmitters. Chapter-4 deals with the LINC architecture which is one of the efficiency and linearity enhancement techniques, where the design methodology is discussed with the help of simulation results followed by an explanation of the measurement setup used to validate the performance of the LINC architecture. Chapter-5 discusses the measurement results and Chapter-6 provides the conclusions and future work. 1.1. Current trends in nano-satellites In-order to understand the current trends in nano-satellite missions, a database of the nano- satellite missions that were launched between 2003 and November 2013 was made and various analysis were performed on this data. A total of 174 nano-satellites missions were launched during this period, of which 183 transmitters were flown. The focus of the information collected from these missions were: mission objective, year of launch, number of transmitters flown, communication mode (half-duplex or full duplex), transmitter data-rate, modulation schemes, transmit power and status of the mission. The database created can be found in Figure 1. Number of nano-satellite launches.
  • 22. 22 MSc Thesis, Visweswaran Karunanithi [Appendix]. Figure 1 shows the number of nano-satellites launched corresponding to the year. It can be seen that there is a steep increase in the number of missions launched in 2013. There is an increase of 437% from 2012 to 2013. These numbers were compared with the Nano/microsatellite market assessment study that was later performed by SpaceWorks3 in Feb 2014. The analysis converge and based on this analysis, the projections for 2014 show an increase of 2 – 3 times higher than the number of launches in 2013. Based on the mission objectives and the payload flown, the nano-satellite missions were categorized into: education/technology demonstration, science/technology demonstration, military, Industry/technology demonstration, education/science, remote sensing, military/technology demonstration, technology demonstration, education and science missions. The following Pie chart shows the percentage of missions that belong to each of the category: Figure 2. (Left) Mission types between 2003 and 2012, (right) Mission types in 2013. From Figure 2, it can be seen that 36% of the missions in 2013 comprise of industrial technology demonstration, remote sensing, government/military missions compared to only 13% between 2003 and 2012. This change in trend shows that nano-satellites have started to be perceived as a serious contender to replace the bigger satellites. As the mission type has a 3 http://www.sei.aero/eng/papers/uploads/archive/SpaceWorks_Nano_Microsatellite_Market_Assessment_Januar y_2014.pdf Figure 3. (Left) data-rates used by nano-satellite mission between 2003 and 2012 (right) data-rates used by nano-satellite missions in 2013.
  • 23. 23 MSc Thesis, Visweswaran Karunanithi direct influence on the downlink data-rate, it can be seen in Figure 3 the change in trend in terms of popular data-rates used by nano-satellites between this period: It can be seen that 43% of the missions used downlinkdata-rates below 1200 bps between2003 and 2012, compared to only 23% of them missions using this option in the year 2013. Looking at the higher data-rate, only 9% of the mission used data-rates greater than 100 kbps between 2003 and 2012, compared to 24% of the mission in 2013 used a data-rate greater than 100 kbps. Due to the limited availability of frequency bandwidth in some of the conventionally used frequency bands such as VHF and UHF, as the downlink data-rate requirement increases, there is also a need to go to the higher frequency bands such as, S-Band, X-band and Ku-band where the larger bandwidths are available. The Figure 4 shows the trends in the frequency bands used by the nano-satellites. It can be seen that the number of missions using S-band and X-band downlink has slightly increased in 2013. The amateur bands in UHF and VHF are a very popular choice for downlink. The main reason for this can be attributed to space heritage and the popularity of this band among radio amateurs around the world. This approach proves to be a win-win situation for both the satellite developers in receiving the spacecraft from all around the world, as a result increasing the reliability of a mission and at the same time benefit the amateur radio community to perform small experiments as a return favour. OSCARs (Orbiting Satellites Carrying Amateur Radios) in nano-satellites has gained popularity for the same reason. A good example is FUNcube4 which was launched in November 2013 and has received 256.5 MB of telemetry data so far (as of July 2014) from amateur radio operators around the world. This mission uses a simple 1200 bps BPSK downlink. With the support of the amateur radio operators, the telemetry of the satellite is available almost real-time, through the entire orbit. 4 FUNcube-1 (AO73) http://funcube.org.uk/ Figure 4. (Left) popular frequency bands used between 2003 and 2012. (Right) Frequency bands used by nano -satellite missions in 2013.
  • 24. 24 MSc Thesis, Visweswaran Karunanithi 1.2. Conclusion of the analysis Based on the study done so far to determine the current trend and to see if there is really a need to start investigating the need for a high data-rate transmitter for a nano-satellite mission, the following conclusions were derived: - First and foremost, the analysis done in the previous section, converges with the 2014 Nano/microsatellite Market Assessment that was conducted by SpaceWorks in February 2014. This market assessment also goes on to project a 2 – 3 times increase in the number of nano- satellite that will be launched in 2014, thus between 160 and 240, and more than half of the missions will be for Earth Observation/Remote Sensing. - Apart from the total number of launches, nano-satellite cluster launches has also started to gain popularity. A Minotaur-1 rocket put 29 satellites into orbit in a single launch on 19th November 2013 and DNEPR placed 32 satellites in a single launch on 21st November 2013. This has made frequency coordination really challenging as most of the nano-satellites preferred VHF and UHF bands which are very narrow bands and during initial days of the mission after the satellite separates from the launch adaptor, the satellites are very close to each other. This can result in interference issues. This would mean that satellites will have to start choosing frequency bands where a large bandwidth is available to avoid interfering with other satellites. - As the complexity of the missions increase, the amount of data needed to be downlinked also increases. This has made the nano-satellite developers to start using higher data-rates, and as a result, start using S and X band. - It was seen that a redundant low speed downlink transmitter (operating in amateur bands) were flown alongside high data-rate transmitter in various missions. The use of amateur frequency bands is likely to continue till a reliable standard is followed for high speed downlinks and ground stations around the world upgrade to support such missions. - Spectrally efficient modulation scheme is the need of the hour. Considering the number of launches forecasted, there is a need to start investigating these modulation schemes for nano- satellite missions. - It is wise to adopt some of the existing recommendations laid down by CCSDS and ECSS. The main advantage is, there are ground stations around the world that already support these standards and the recommendations are popular among the bigger satellites.
  • 25. 25 MSc Thesis, Visweswaran Karunanithi 1.3. Problem statement Based on the market study and preliminary analysis performed, it is evident that there is a need to develop high speed transmitters for nano-satellites that implement spectrally efficient modulation scheme and are complaint with ground station standards followed around the world (especially ESA ground stations). Using this as the motivation, recommendations stated in CCSDS and ECSS standards have to be studied and design constraints have to be derived. CCSDS also provides recommendations for spectrally efficient modulations schemes. It has to be investigated if these modulation schemes suit nano-satellite applications. Mission case study needs to be done using some of the on-going and future projects to see if there is really a need to implement such complex modulation schemes in nano-satellites. Implementation challenges needs to be studied and the need to implement efficiency and linearity enhancement technique in the transmitter design has to be investigated. If efficiency and linearity enhancement technique needs to be implemented, using system engineering approach, the most appropriate architecture for nano-satellite application needs to be chosen. Once the architecture is chosen, a prototype needs to be designed to show the proof of concept. The following diagram shows the work-flow of this thesis:
  • 27.
  • 28. 28 MSc Thesis, Visweswaran Karunanithi 2 2. Nano-Satellite transmitters for the future ano-satellites in the past were mainly thought of as practical learning tools by the universities to teach engineering students about designing space hardware and encourage the interest in pursuing a career in the space industry. The main advantage of nano-satellites over the bigger satellites is: very short development time and cheaper access to space. This attracted industries to start collaborating with the universities and started using nano-satellites as a test-bed to quickly test and space-qualify satellite hardware. A good example for this is Delfi-C3 which carried Thin Film Solar Cells (TFSC) from Dutch space and two Autonomous Wireless Sun Sensors (AWSS) from TNO as a part of its technology demonstration payloads [1]. Gradually over the period, a lot of standardization has lead the nano-satellite developers to start coming-up with more complex payloads. This change has posed the need to look at high data-rate transmitters that can meet the needs of mission requirements. This chapter starts with an overview of the recommendations laid by CCSDS and ECSS for space-to-earth links for satellites that fall under category-A5 . Following the recommendations, a few nano-satellite mission cases are studied to see how the recommendations fit, link budget analysis is performed to derive the design requirements of the transmitter. This chapter concludes with some recommendations for the roadmap for nano-satellite transmitter design. 2.1. CCSDS and ECSS recommendations CCSDS was founded in 1982 by major space agencies of the world, it is a multi-national forum for development of communications and data systems standards for spaceflight6 . The CCSDS recommendations cover a broad spectrum: Space Internetworking Services, System engineering, Mission Ops & Information Management services, Cross Support Services, Spacecraft Onboard Interface Services and Space LinkServices. The scope of this thesis limits to the following recommendations stated in Space Link Services: Radio Frequency and modulation systems (Blue book, CCSDS 401.0-B) [2], Bandwidth-efficient modulations (Green book, CCSDS 413.0-G-2) [3] and CCSDS Space link protocols over ETSI DVB-S2 standard (Blue book, CCSDS 131.3-B-1) [4]. 5 Category-Aare those satellites whichorbit at an altitude of below 2 x 106 Km, Category-B are thosesatellites which orbit at an altitude above 2 x 106 Km (This categoryalso includes deepspace probes) 6 http://public.ccsds.org/default.aspx N
  • 29. 29 MSc Thesis, Visweswaran Karunanithi 2.1.1.Frequency band allocation & constraints The frequency band allocationis done based on the service/operation provided by the satellites. The different services are: Space Research (SR), Earth Exploration Satellite Service (EES), Amateur Service and Space Operation (SO). The Table 1 gives the list of recommended frequency bands corresponding to the service it can provide. Table 1. Frequency allocation corresponding to service provided. Frequency bands (MHz) Service7 (L-Band) 1215 - 1240 SR, EES (L-Band) 1240 - 1300 SR, EES, Amateur (L-Band) 1525 -1535 SO, EES (S-Band) 2200 - 2290 SR, SO, EES (X-Band) 8025 - 8400 EES (X-Band) 8450 - 8500 SR (Ka-Band) 25500 - 27000 SR, EES (Ka-Band) 37000 - 38000 SR The constraints in using each of the frequency bands is as follows: - L-Band: o This band shall not be used for feeder links of any service. - S-Band (2200 – 2290 MHz): o The maximum occupied bandwidth for spacecraft in this band shall not exceed 6MHz. o The link shall be active only during the period when the satellite is in the visibility cone of the ground station. (This will help frequency re-use) o The reliability of the device used to switch-off the transmitter shall commensurate with mission lifetime. 7 Purelyscience missions suchas radio-telescopes fall under the categoryof SR, Engineeringmissions suchas communicationsatellitesfall under the categoryof SO and earth observationmissions suchas Remote sensing fall under the categoryof EES.
  • 30. 30 MSc Thesis, Visweswaran Karunanithi - X-Band (8025 – 8400 MHz): The constraints specified for S-Band applies for this band as-well, except for the bandwidth constraints. - X-Band (8400 – 8450 MHz): o The maximum allowable bandwidth for a down-link in this band shall not exceed the mask specified in Figure 5. - Ka-Band (25.5 to 27 GHz & 37 to 38 GHz): There were no constraints formulated for this band but there has to be an agreement with the frequency coordinator prior to using this band. - All the bands, up to X-band shall have a frequency stability not less than ± 2 × 10-5 (20ppm) under all conditions for the entire lifetime of the mission and for Ka-band, a frequency stability of ± 0.02 ppm/o C within the temperature range +10 o C to +40 o C. The maximum allowed bandwidth corresponding to various data-rates is shown in the table below: Table 2. Maximum allowed bandwidth corresponding to data-rate and frequency bands. Frequency band (MHz) Operation Occupied Bandwidth 2200–2290[Downlink] And 8450–8500[Downlink] Telemetry(Rs < 10ksps) 300 kHz Telemetry (10 ksps <Rs < 60 ksps) 1200 kHz or 30 × Rs (smaller of the two) Telemetry(60 ksps < Rs < 2Msps) 1200 kHz or 12 × Rs, whichever is larger upto 6 MHz at 2 GHz and 10 MHz at 8 GHz. Telemetry(Rs > 2Msps) 1.1 x Rs up to 6 MHz at S-band and 10MHz at 8GHz. There are no bandwidth constraints stated for the frequency bands in Ka-band. Figure 5. Maximum allowable bandwidth vs symbol rate.
  • 31. 31 MSc Thesis, Visweswaran Karunanithi 2.1.2.Transmitter spurious emission and harmonic levels The emitted spectrum for all the space science services shall adhere to the spectral emission mask shown in Figure 6. The total power contained in any single spurious emission shall not exceed -60 dBc. This constraint is valid for frequency bands from VHF to Ka-band. In the figure, the blue line indicated the boundary limits for data rate below 2 Msps, it has a slope of 43 dB per decade and the green line indicates the limit for data-rate above 2 Msps which has a slope of 60 db per decade. 2.1.3.Recommended modulation schemes The occupied bandwidth is directly dependent on the data rate, the modulation schemes are recommended based on the data-rate is shown in Table 3 Table 3. Recommended modulation schemes. Datarate<2 Msps When using Sub-carrier: - PCM/PM/Bi-phase. - PCM/PM/NRZ. Suppressed carrier modulation (medium rate <2Msps) - BPSK - QPSK - OQPSK. Datarate> 2Msps Preferred modulationschemes when datarates are larger than 2Msps: - GMSK (Pre-coding necessary) - Baseband filtered OQPSK - TCM 8PSK Figure 6. Spectral emission limits
  • 32. 32 MSc Thesis, Visweswaran Karunanithi CCSDS space link protocols over ETSI DVB.S2 recommendations: - QPSK - OQPSK - 16-APSK - 32-APSK - 64-APSK. The downlink data-rates are categorized into high-speed downlink ( > 2Msps) and low-speed data-rate (<2 Msps). The modulation schemes such as PCM/PM/Bi-phase and NRZ are allowed only for data-rates below 2 Msps. The modulation schemes recommended for data rates below 2 Msps are all constant envelope signals and does not influence the performance of a HPA (High Power Amplifier) to a large extent when compared to non-constant envelope modulation schemes. In the case of digital modulation techniques such as BPSK, QPSK and OQPSK, although there is no inherent amplitude variation, but the use of SRRC (Squire Root Raised Cosine) filter in-order to reduce the ISI (Inter Symbol Interference) causes an amplitude variation, affecting the performance of the HPA. The non-linearity effects caused by OQPSK is lesser compared to QPSK at larger SRRC roll-offs [5]. For data-rates above 2 Msps, the modulation schemes proposed by CCSDS space link protocols over ETSI (European Telecommunication Standards Institute) DVB-S2 (Digital Video Broadcast – Satellite) include QPSK and OQPSK which are constant envelope schemes and a few additional modulation schemes such as 16-APSK, 32-APSK and 64-APSK. APSK stands for Amplitude Phase Shift Keying. As the name suggests, both phase and amplitude of the carrier are modulated in accordance with the message signal. It is similar to QAM (Quadrature Amplitude Modulation) but the PAPR of APSK modulation scheme is lesser than that of QAM. An in-depth analysis of PAPR of these modulation schemes and its effects on the performance of the PA will be discussed in later sections. 2.1.4.Conclusion of the CCSDS and ECSS recommendations The recommendations stated in CCSDS and ECSS are a good starting point for standardization in the communication system for nano-satellites. The standards have already been implemented on the bigger satellites and ground stations around the world have the infrastructure compatible with CCSDS standards. Some quantitative analysis needs to be performed to determine if the modulation schemes recommended for data-rates larger than 2 Msps is required for nano-satellites with the help of mission case studies. The main challenge in implementing the recommended modulation schemes for nano- satellites is that, most of the spectral efficient schemes have a non-constant envelope profile. This variation in the amplitude can cause a degradation in the performance of the PA in terms of either efficiency or linearity, thus an analysis needs to be done to determine if one of the conventional operating classes (Class – A, AB, B, C, D, E, F, F-1 or J) is sufficient or efficiency/linearity enhancement technique needs to be performed. In case if an enhancement
  • 33. 33 MSc Thesis, Visweswaran Karunanithi technique needs to be implemented, a trade-off between the different architectures needs to be performed and the most suitable architecture needs to be chosen. In the subsequent sections of this chapter, a few nano-satellite missions will be studied to establish the requirements for a nano-satellite transmitter. 2.2. Nano-satellite mission case study The downlink data-rate requirement for a space mission mainly depends on: 1) Number of sub-systems in the spacecraft 2) The amount of house-keeping data generated by each of the sub-system 3) The amount of payload data generated, which depends on the type of payload 4) The contact time with the ground station. The amount of data generated by the payload depends on the service provided by the mission. The Figure 7 shows the relation between the downlink data-rate requirement and the type of the missions. Figure 7. Pictorial representation of the relation between downlink data-rate, service provided and type of payload.
  • 34. 34 MSc Thesis, Visweswaran Karunanithi 2.2.1.Spacecraft housekeeping data downlink requirement In this case, a general 3U nano-satellite LEO mission is considered to calculate the total amount of house-keeping data that will be generated by the spacecraft bus. The subsystems that are assumed in a generic satellite bus are: On-board computer, antenna system, transceiver (primary, secondary), electronic power system (Battery system, solar cells, etc.), attitude control system (sensor and actuator) and payload telemetry. In order to get a more realistic number, the satellite bus of TRITON-1 mission was considered. The house-keeping data from each of the sub-systems is periodically sampled by the on-board computer and stored in the on- board memory. Based on the available memory and mission requirements, either all of the stored telemetry is downlinked or only the mission critical data is sent to the ground station. In the case of TRITON-1, only some of the mission critical telemetry was downlinked. This case considers a hypothetical situation where all the sampled telemetry needs to be downlinked to the ground station. The size of each of the telemetry packets is approximately 16 bytes and it is assumed that each of the subsystems is sampled for telemetry on an average every 15 seconds. The total amount of house-keeping data generated was calculated to be 7.7 Mbits per day. The Table 4 gives the calculation done in-order to find the required downlink data-rate. Table 4. Required downlink data-rate to downlink all the generated house-keeping data. TT&C data rate requirement. Units Typical sampling time 1 sec Data per subsystem 16 bytes 128 bits Duty-cycle 100% Data generated per day per subsystem 7.732E+05 bits Number of subsystems (approx.) 10 Total data generated per day 7.732E+06 bits Avg. GS contact time per day 50 min Percentage of contact time to transmit house-keeping data. 10 % Effective TT&C downlink data rate 24576 bps Data-overhead 50% Data rate including over-head 36864 bps Considering that a nano-satellite will not have a dedicated telemetry downlink, the same link has to be shared between telemetry and payload data. Thus, in the above calculation it is assumed that 10% (a rough estimate) of the total contact time is used to downlink the telemetry data. The calculated downlink data-rate requirement is approximately 37 kbps. 2.2.2.Payload downlink data-rate requirement for a EES service The Remote Sensing payloads mainly fall under the category of EES services. Some of the commonly used payloads under this service include optical imaging payload, infra-red imaging payload and SAR (Synthetic Aperture Radar). Among nano-satellites, optical imaging payload
  • 35. 35 MSc Thesis, Visweswaran Karunanithi has been the most commonly used payload. As a case study, the mission requirements stated for MISC-1 [6] mission will be considered. The mission was proposed to capture 7.5 m GSD (Ground Spatial Distance) multispectral imagery from an altitude of 540 km. The payload for this mission comprises of a 16 MP Kodak KAI-16000-CXA-JD CCD image sensor. The mission overview are stated in Table 5 below: Table 5. MISC-1 mission overview MISC mission units Payload Multi-spectral imaging Volume 3U Altitude 540 km Image sensor KAI-1600-CXA Number of Pixel 16 MP Active pixel rows 4872 Active pixel column 3248 Number of Ground stations 13 Ground swath/image 702 Sq.km Target area per day 137,500 Sq.km Total number of images required per day 196 Size of a single un-compressed image (10 bit pixel depth) 19.78 MB In the paper [6], it could be clearly seen that the mission was constrained by the downlink radio capabilities (38.4 kbps downlink), as a result the image had to be compressed to 1/12th its original size and the mission required 13 ground stations to fulfill the requirement of downlinking 196 images per day. The mission was finally not realized using the above formulated mission requirements Now, the required downlink data-rate will be calculated for different cases to see an optimal solution for the mission to be feasible. Table 6. Downlink data-rate requirement when un-compressedimage is transmittedand13 ground stations are used. MISC mission downlink data rate requirement Number of images needed to be transmitted/day 196 Size of an image (Pixel depth assumed to be 10bits) 1.58E+08 bits Total image data generated/day 3.101E+10 bits Avg. Ground station contact time/day 36 min Total Ground station contact time (13 ground stations) 468 min effective payload downlink data rate 1.10E+06 bps Over-head 50% Total required downlink data rate 1.65E+06 bps
  • 36. 36 MSc Thesis, Visweswaran Karunanithi From the above calculation it can be seen that the downlink data-rate required to transmit un- compressed image using 13 ground stations is 1.65 Mbps. Similarly, two other cases were considered because the idea of using 13 ground stations to downlink the payload data from one satellite means highly synchronized operation between the ground stations which is complex. The two other scenarios that were considered are: 1) Downlink data rate requirement when one satellite and one ground station is used, 2) Downlink data-rate requirement when 3 satellites and one ground station is used. In the first scenario, the calculation showed a required downlink data rate of 21.52 Mbps and the second scenario required a down link data-rate of 7 Mbps per satellite. 2.2.3.Downlink data-rate requirement for nano-satellite providing data services For this case, VDE-Sat is considered. VDE stands for VHF Data Exchange Satellite. It is a technological concept developed by IALA e-NAV8 committee and now widely discussed at ITU, IMO and other organizations. VDES was originally developed to address emerging indications of overload of VHF Data Link (VDL) of AIS and simultaneously enables a seamless data exchange for the maritime community. This is an on-going project at ISISpace BV, which is in a conceptual stage. The Figure 8 gives a pictorial representation of the mission concept. Figure 8. Pictorial representation of VDES [7]. Some of the mission requirements are already formulated, the mission needs to have a high-speed downlink at VHF (161MHz), the available bandwidth is 50 kHz and a downlink symbol rate of 40 ksps. Based on these requirements, link budget was calculated for various 8 International Associationof Marine Aids to Navigation and Lighthouse Authorities
  • 37. 37 MSc Thesis, Visweswaran Karunanithi possible modulation schemes and finally 16-APSK modulation was chosen. The link budget analysis for this case is explained in the next section.
  • 38. 38 MSc Thesis, Visweswaran Karunanithi 2.3. Link budget analysis For the cases discussed in the previous section, a link budget was calculated to check if the assumed system design meets the requirements. The calculations were done by creating a MS excel template with the relevant equations and the possibility to vary the various system parameters (transmit power, transmit antenna gain, receiver antenna gain, etc.). In the case of spacecraft house-keeping data, the required downlink data-rate was calculated to be ~37 kbps. The link budget was calculated with realistic numbers from a commercially available UHF transmitter and the antenna gain was considered to be 0 dBi. With this, the link margin was calculated for different elevation angles and shown in Figure 10. Figure 9. Link budget calculation of spacecraft housekeeping data downlink. Figure 10. Link margin for different elevation angles.
  • 39. 39 MSc Thesis, Visweswaran Karunanithi The modulation scheme chosen in this case is QPSK/OQPSK, and the occupied bandwidth of the transmitted signal is 23 kHz. In the case of an imaging payload discussed in the second case study, after various iterations, an X-band downlink was chosen to downlink at 7 Mbps. The calculations and plot of the link margin for various elevation angles are shown in Figure 11 and Figure 12. Thus, it can be seen that it was possible to close the link for a downlink data-rate of 7 Mbps, a transmit power of 2 Watts and an EIRP of 35.5 dBm. In the case of VDE-Sat, again various options for modulation scheme and transmit power were tried and it was possible to close the link with 16-APSK modulation scheme and a transmit power of 6 Watts. The link calculations and plot are shown in Figure 13 and Figure 14. Figure 11. Link budget for an optical payload data downlink. Figure 12. Link margin vs elevation angles for optical payload downlink.
  • 40. 40 MSc Thesis, Visweswaran Karunanithi 2.4. Summary of the case study and conclusion A summary of the case study is shown in Table 7, it can be seen that there is a direct relation between the required downlink data-rate and the type of service the mission provides. It is not possible to implement these high data-rates in the amateur bands where the available bandwidth is limited. Thus, there is a need to move to the higher frequency bands and also use spectrally efficient modulation schemes. In general, missions with remote sensing payload needs the fastest downlink. In the above case a 7 Mbps downlink was required to downlink all the generated payload data. The Figure 13. Link budget for VED-Sat downlink. Figure 14. Link margin vs elevation for VDE-Sat.
  • 41. 41 MSc Thesis, Visweswaran Karunanithi frequency band for satellite house-keeping data downlink was chosen to be UHF and it can be seen that it is possible to implement this relatively high-speed down link of 36 kbps within the available bandwidth of 25 kHz. Table 7. Summary of mission case study. Spacecraft housekeeping data MISC-1 payload downlink VDE-sat downlink. Frequency Band UHF 8.4 GHz [X-band] 161.9 MHz [VHF] Available bandwidth 25 kHz 10 MHz 50kHz Occupied bandwidth 23.3 kHz 6.3 MHz 50 kHz Downlink data rate 36 kbps 7 Mbps 40 ksps (160 kbps) Modulation scheme QPSK/OQPSK 16-APSK 16-APSK Transmit power 500 mW 3 W 6 W PAPR [dB]9 3.4 4.5 4.5 The table also shows the PAPR of the spectral efficient modulation schemes used. The main disadvantage in using a linear PA in the transmitter to amplify such non-constant envelope signal is that, the efficiency drops drastically. Some of the transmitter components and parameters that depend upon the efficiency and power consumption of the PA are: - Battery life, - Solar-cell and battery capacity, - Heat-sink size and weight, - Need for additional coolers. Thus, the main challenge in implementing these modulation schemes for nano-satellite missions is to implement an efficient amplifying architecture and at the same time not compromise on the linearity. In order to achieve this, different efficiency and linearity enhancement techniques need to be investigated and a suitable architecture needs to be chosen for nano-satellite application. 9 Peak to Average Power Ratio calculatedincluding a SRRC filter witha roll-off of 0.5
  • 42.
  • 43. 43 MSc Thesis, Visweswaran Karunanithi 3 3. High efficiency transmitter architectures ome of the main constraints that needs to be considered while designing a subsystem for nano-satellite mission are: power consumption, efficiency, low complexity, efficient thermal design and sustain harsh space environment such as temperature variations and radiation effects. Transmitter is one of the most power hungry subsystems of a nano-satellite, based on the mission requirements the transmitter of a nano- satellite can consume up to 30% of the total generated power. Thus, it is very important to investigate a transmitter architecture that can efficiently transmit spectral efficient modulation schemes that were discussed in the previous section, without compromising on the overall performance. The main challenge involved in designing an efficient architecture is: the spectral efficient modulation schemes have a non-constant envelope profile as a result of modulating the message signal into both phase and amplitude of a carrier, the amplitude variation in the modulated carrier is characterized by the Peak to Average Power Ratio (PAPR) of the signal. The PAPR is the ratio of the peak power of the envelope to the average power of the signal envelope. This is one of the critical parameters that influence the trade-off between linearity and efficiency of a Power Amplifier in a transmitter. This chapter gives an overview of the challenges posed by spectral efficient modulation schemes on the performance of the transmitter, followed by different efficiency and linearity enhancement techniques from the literature, a trade-off between the different architectures to determine a suitable one for nano- satellites and a conclusion of this study. 3.1. Overview The basic functionality of a transmitter is to collect the telemetry/payload data from the On-board computer, put the data into packets, modulate the information in the packet, amplify S Figure 15. Block-diagram of a transmitter with digital modulation.
  • 44. 44 MSc Thesis, Visweswaran Karunanithi the modulated signal to appropriate levels and pass it on to the antenna which radiates the modulated RF signal. A generic block-diagram representation of a transmitter is shown in Figure 15. The baseband board is responsible for interacting with the on-board computer and collect the data to be transmitted (telemetry/payload), implement the data link layer functions, perform signal processing on the baseband digital data (Square Root Raised Cosine filtering) and generates the stream of I (In-phase) and Q (Quadrature) data as an input to the modulator. The IQ modulator, modulates the baseband IQ signal onto the carrier frequency and the output power of the modulator is generally not enough to directly be transmitted and needs amplification. Some of the IQ modulators currently available in the market produce an output power between 0 dBm (1mW) to 10 dBm (10mW). Thus, the amplifier performs the operation of converting the DC power (supply) into a significant amount of RF/microwave output power that replicates the modulated signal that is inputted to the PA. The key design parameters of a PA are efficiency and linearity. With the conventional class of operation such as Class-A, B, AB, C, D. E and F, a trade-off has to be done between linearity and efficiency. There is no one single PA or transmitter technique that suits all applications. A relation between efficiency and linearity for different Classes of operation is shown in Figure 16. When a PA is used to amplify constant envelope signals such as CW (Continuous Wave), FM, classical FSK or GMSK, linearity of the PA is not critical and can be compromised for efficiency. A Class C, D, E or F would be a logical choice for a constant envelope signal. On the other hand, linearity of a PA becomes critical when the signal contains both amplitude and phase modulation such as QAM or APSK. In this thesis the scope of modulation schemes that will be considered to characterize the performance of the PA are: 16 and 32-APSK, which will be compared with 16, 32 and 64-QAM modulation schemes. The PAPR of the QAM scheme is more than the APSK modulation scheme. The PAPR contributed due to variation in the amplitude of the symbols is shown in Table 8. Figure 16. Linearity vs Efficiency trade-off for conventional PA classes.
  • 45. 45 MSc Thesis, Visweswaran Karunanithi Table 8. PAPR contribution due to the envelope profile of various modulation schemes. In addition to the PAPR contributed by the modulation scheme, additional amplitude variation is caused due to passing the digital data through an SRRC filter. In order to minimize inter- symbol interference, the digital data needs to be passed through a SRRC filter. The PAPR depends on the roll-off factor of the filter. The following table (Table 9) shows the additional PAPR contributed due to the SRRC filter for different roll-off factors [8]. Table 9. PAPR contribution due to the SRRC filter Roll-off 0.1 0.2 0.3 0.4 0.5 PAPR 7.5dB 5.8dB 4.6dB 3.7dB 3.4dB Thus the PAPR seen by the PA is the sum of the above two contributors. For example, if 32- APSK modulation is implemented with a SRRC filter having a roll-off 0.5, then the combined PAPR seen by the amplifier is 4.8 dB. Peak efficiency of a PA is obtained when driven at saturation but this is also a non- linear region. Thus, alternate architecture to amplify efficiently without compromising on the linearity is studied in this chapter. Some of the techniques analyzed in this thesis are: 3.1.1. Characterization of linearity and efficiency The non-linearity in a PA is caused mainly due to gain compression and saturation. The effects of non-linearity can be characterized using various methods: carrier-to-intermodulation (C/I) Modulation M-PSK 16-APSK 16-QAM 32-APSK 32-QAM PAPR 0dB 1.1dB 2.6 dB 1.4 dB 3.9 dB Figure 17. Efficiency and linearity enhancement techniques.
  • 46. 46 MSc Thesis, Visweswaran Karunanithi ratio is used to compare the amplitude of the desired output carrier to the intermodulation distortion (IMD) products, adjacent channel power ratio (ACPR) which is used to characterize the quality of the spectrum, the power ratio in dBc is measured between the spectral components within the allotted bandwidth and adjacent channel components. Another method to characterize the linearity is by measuring EVM (Error Vector Magnitude), EVM gives the vector distance between the received symbol and expected symbol for a given modulation scheme in the constellation diagram. Efficiency is one other important design parameter, from the literature there are various definitions for efficiency of a PA. Drain efficiency: It is the ratio of the output power from the PA (in Watts) to the supplied DC power (in Watts). It is given by: out d DC P P   (1) Power added efficiency (PAE): The PAE is defined as the ratio of the RF power that is amplified (difference between the RF output power and RF input power) to the DC power consumed. It is given by: ( )out in PAE DC P P P     (2) The PAE is a more widely used metric as it takes into account the gain of the PA. When a non-constant envelope signal is amplified by a PA, a more useful metric to measure is the average efficiency [9], it is mathematically expressed as oAVG AVG dcAVG P P    (3) PoAVG and PdcAVG are a function of the Probability Density Function (PDF) of the input non- constant envelope signal. The PDF gives the relative amount of time an envelope spends at various amplitude levels. Figure 18 gives the PDF of a 16-APSK signal when passed through a SRRC filter that has a roll-off of 0.5. Figure 18. PDF of 16-APSK modulation scheme, implemented using an SRRC filter with a roll-off of 0.5(4000 sps)
  • 47. 47 MSc Thesis, Visweswaran Karunanithi Although this matric is straight forward to measure for the conventional PA classes (Classes A-F), the above equation cannot be directly used in all the efficiency and linearity enhancement techniques that will be discussed later in this chapter. In the case of efficiency and linearity enhancement techniques such LINC, EER and ET the message signal that is fed to the PA is modified so that the PA is driven at its peak efficiency. In such cases PAE is a more preferred metric to gauge the performance of the architecture. 3.2. Doherty architecture The Doherty architecture (shown in Figure 19 ) was first proposed by W.H. Doherty in 1936 [10]. This architecture makes use of active load-pull technique where the load seen by one PA can dynamically be varied by applying current from another PA which is coherent in phase. 3.2.1.Working Principle Figure 19. Doherty PA architecture. Figure 20. Schematic representation of Doherty.
  • 48. 48 MSc Thesis, Visweswaran Karunanithi This architecture comprises of two10 signal paths, amplified by two non-identical PAs called Carrier PA and Peaking PA (sometimes also called Auxiliary PA). At low input levels the peaking PA is in cut-off (consuming no current) and only the carrier PA is operational. Once the input level reaches a certain threshold, the peaking PA turns ON and the output power is contributed by both the PAs. The transfer curves of the two PAs is shown in Figure 21 [11]. The carrier PA operates linearly till the point output power reaches Pmax/4 as there is no influence from the peaking PA (not yet turned ON), at Pmax/4 (which is 6 dB below the peak envelope power), Peaking PA turns ON and current flows from both the PAs to a common load leading to load modulation. From this point onwards the transfer curve of both the PAs are non-linear. The process of active load pulling can be better explained using the schematic representation of the architecture shown in Figure 20. Both the PAs are considered to be current sources. When the Peaking PA is OFF, I2 = 0 and only I1T flows through the load resistor RL. The voltage drop across the resistor is a linear function of the current flowing through the resistor RL. But when the Peaking PA turns ON, both I1L and I2 flows through the load resistor and the impedance seen by the individual PAs is given by: 10 It is possible to have N paths, but the architecture was introducedwith two paths. Figure 21. Input power vs Output power in linear scale [11]
  • 49. 49 MSc Thesis, Visweswaran Karunanithi 2 1 1 2 1 1 2 1 2 1 1 2 2 1 1 1 T L T T T T T L T L I Z R I Z Z Z Z Z I Z R V I Z R I                       (4) It can be seen from the above equation that the impedance seen by each of the PA is a function of the current flowing through the other PA. 3.2.2.Advantages and dis-advantages of Doherty architecture Some of the advantages of Doherty architecture are: - It is possible to achieve a good efficiency enhancement in the power back-off region. - Suitable to amplify signals with a PAPR as large as 10dB. - The architecture is relatively easy to implement compared to EER and Envelope Tracking architectures. - The performance in the power back-off region can be significantly improved by using more stages (N-way Doherty) [12]. Dis-advantages of Doherty are: - Gain degradation: Due to the low bias voltage, Peaking PA contributes lower output compared to Carrier PA. - Peaking PA has to switch ON at exactly Pmax/4. - Both PAs have to follow the nonlinearity of Figure 21 and generally pre-distortion circuit is necessary. - Higher intermodulation distortion: It is caused again due to low biasing of the Peaking PA. - Narrow bandwidth: The quarter wave lines used for impedance transformation reduces the bandwidth of operation. 3.3. Kahn/EER (Envelope Elimination and Restoration) 3.3.1.Working principle of Kahn/EER architecture This technique of amplification was first proposed by Kahn in 1952. In a non-constant envelope signal, message is modulated onto both amplitude and phase of the carrier signal. In the case of Kahn/EER method, the amplitude of the message is first eliminated from the modulated signal, leaving behind a constant envelope phase modulated signal and the amplitude envelope profile. Now, the constant envelope signal can be amplified using a high-efficiency, non-linear PA such as class C, D and the envelope profile is separately amplified and fed in as the supply
  • 50. 50 MSc Thesis, Visweswaran Karunanithi voltage of the high efficiency PA. A block diagram representation of this architecture is shown in [figure]. Thus, it is possible to achieve high efficiency and linearity regardless of the PAPR of the input signal. From the literature, it can be seen that very highefficiencies can be achieved but the design is fairly complex to implement. 3.3.2.Advantages and dis-advantages of Kahn architecture Some of the advantages of the architecture are: - Ideally can be operated at the peak efficiency of the final PA (depends on the operating Class chosen for the final stage). - Provides excellent linearity as the performance is not dependent on the linearity of the amplifying transistor. Some of the dis-advantages of this architecture are: - Circuit complexity: Implementation of the envelope restoration circuitry is complex, as perfect synchronization has to be achieved between the two signals. - Needs pre-distortion at higher frequency. - The switching frequency of the class-Smodulator should be at least 6 times the RF bandwidth. 3.4. Envelope Tracking (ET) architecture 3.4.1.Working principle of ET ET and EER technique use the concept of supply modulation, where EER uses a combination of non-linear PA and an envelope re-modulation circuit but in the case of ET, a combination of linear PA and a supply modulation circuit which tracks the envelope profile of the input signal. The envelope profile is converted to discrete DC levels using the power conditioner and fed as the drain voltage to the final PA. The supply voltage is varied dynamically with sufficient Figure 22. Block diagram representation of a modern implementation of the Kahn architecture.
  • 51. 51 MSc Thesis, Visweswaran Karunanithi headroom to allow the RF PA to operate in a linear mode. The block diagram representation of an ET architecture is shown in Figure 23 3.4.2.Advantages and disadvantages of ET architecture The advantages of ET architecture are as follows: - The implementation is slightly less complex when compared to EER. - The efficiency of linear amplifiers such as Class-A or AB can be enhanced using this technique. - Constant efficiency can be maintained in the power back-off region. The main disadvantages of this architecture are: - The peak efficiency achieved cannot exceed that of Class-AB. - Circuit overhead is high. - Synchronization between the carrier containing phase information and the time varying supply is the main challenge in this technique. - The PAE is poor compared to other efficiency enhancement techniques due to the additional circuitry. 3.5. Switched Capacitor Digital Power Amplifiers SCDPA 3.5.1.Working principle of SCDPAs Instead of making use of the linear trans-conductance property of a CMOS transistor, SCDPA makes use of its switching property. The block diagram of SCDPA is shown in Figure 24. Deeply scaled CMOS transistors are poor trans-conductors but very good switches. SCDPAs exploit this property. SCDPAs incorporates the functionality of a Digital-to-analog converter and a PA into the same circuitry. Each of the CMOS transistors are represented by switches (C0, C1, C2…. C7) and they are either at toggle (between Vdd and ground) or only ground potential. This is decided by the control signal given to the gate from the micro-controller/DSP. Figure 23. Block diagram of ET architecture.[11]
  • 52. 52 MSc Thesis, Visweswaran Karunanithi The output voltage depends on the ratio of the switches that are ON (toggle state) to the total number of switches. For example, C4, C5, C6 and C7 are at ground potential (representing OFF) C0, C1, C2 and C3 are in toggle mode (ON state) then Vout will be 0.5Vdd. Similarly, when all the switches are in toggle mode, maximum output power can be achieved. More details about this architecture can be found in [13]. 3.5.2.Advantages and dis-advantages of SCDPAs Some of the advantages of this architecture are: - Low power consumption. - High signal bandwidth. - Good performance in the power back-off region. (flat efficiency uo-to 13dB PAPR reported [14]) Some of the disadvantages of this architecture are: - Not highly linear: AM-PM distortions are high. - Parasitic capacitance in the current technology still limit its performance. - Low PAE (45% reported[14]) - Low output power. Figure 24. Block diagram representation of SCDPA.
  • 53. 53 MSc Thesis, Visweswaran Karunanithi 3.6. LINC (Linear amplification using non-linear components) architecture 3.6.1. Working principle LINC amplifiers, also known as out-phasing amplifiers were first developed by Henri Chireix in 1935 [15] followed by L. F. Gaudernack’s work in 1938 [16]. These systems were proposed to improve both efficiency and linearity of an AM- broadcast transmitter. The concept slowly gained popularity over the years and in 1974 D. C. Cox [17] introduced the term LINC. As the name suggests, linear amplification was obtained using non-linear components. The working principle of this architecture can be explained using the block diagram shown in Figure 25. An amplitude modulated (AM) and phase modulated (PM) signal (non-constant envelope signal) such as 16-APSK is split using a signal processing unit into two PM (constant envelope) signals S1(t) and S2(t). The signals S1(t) and S2(t) are then amplified independently, denoted by G.S1(t) and G.S2(t). The amplified signals are then combined using a power combiner to replicate the amplified AM and PM signal (in this case 16-APSK). This method has gained popularity due to the recent developments in signal processing units. The main advantage in using this system is that, as the signals being amplified by the PA blocks (amplifier 1&2) are constant envelope signals, thus it is possible to use high efficiency PAs and not worry about the non-linearity caused by the PAs. The operation can be explained using the following mathematical expressions: The AM and PM modulated signal can be represented using the expression: j( ( )) ( ) r(t) t t ins t e    (5) Where r(t) is the AM and can be represented as r(t) = rmax cos(θ(t)).  ( ) ( )maxr r(t) 2 j t j t e e    (6) Thus, the input non-constant envelope signal can be represented as a sum of two constant envelope signals: Figure 25. LINC block diagram.
  • 54. 54 MSc Thesis, Visweswaran Karunanithi  ( ( ) ( )) ( ( ) ( ))maxr ( ) 2 ( ) 1( ) 2( ) j t t t j t t t in in s t e e s t S t S t              (7) Where,     ( ( ) ( ))max ( ( ) ( ))max r 1( ) , 2 r 2( ) 2 j t t t j t t t S t e S t e             (8) The above mathematical expressions can be described using 16-APSK constellation diagram as shown in Figure 26. In this case, the symbol 1100 is a non-constant envelope signal with its amplitude 2.84 times lesser than the peak amplitude. This symbol is decomposed into constant envelope signals S1(t) and S2(t) with appropriate out phasing angle θ(t) and constant amplitude rmax/2. max ( ) ( ) arccos r t t r         (9) In the above case, the ratio between instantaneous amplitude and peak amplitude is (1/2.84), thus using the above relation, θ(t) = 69.38 deg. In order to represent a constellation point on the outermost ring (1000, 1001, …) the out-phasing angle θ(t) = 0 deg. 3.6.2. Advantages and dis-advantages of LINC architecture Some of the advantages of LINC architecture are as follows: - AM/AM and AM/PM distortions caused by individual PA blocks does not affect the overall performance of the architecture. Figure 26. 16-APSK decomposition into constant envelope signal.
  • 55. 55 MSc Thesis, Visweswaran Karunanithi - Identical PAs are used in the two signal paths, leading to symmetrical designing of the two paths. Thus, the matching network and the biasing network designed for one of the paths can simply be replicated for the other. - Most of the complexity is in splitting the signal component and synchronizing them, which is taken care by the signal processing unit, the RF design complexity of the PAs is much lesser compared to the other efficiency/linearity enhancement techniques. Some of the disadvantages of the LINC architecture are as follows: - Although using isolating power combiners gives the best performance in terms of linearity, the overall efficiency of the system drops considerably in an isolating power combiner such as WPC. Half of the power is dissipated in the isolation resistor, bringing down the overall efficiency of the system. - In the case of non-isolating power combiners (Chireix combiner), the linearity is compromised for efficiency at large out-phasing angles as a result of load modulation. 3.7. Trade-off analysis All of the efficiency and linearity enhancement techniques discussed so far have both advantages and dis-advantages, there is no one architecture that suits all applications. Thus, a trade-off needs to be performed to determine the most suitable architecture for nano-satellite application. The parameters and its weights that are considered for a trade-off can vary based on the application. The parameters considered for this trade-off are: Complexity: This parameter is given the maximum weight of 5. While designing a system for space application, it is very important to that the system is robust and less complex. Circuit over-head: In some cases the over-head in accurately designing the biasing circuitry or synchronization circuitry adds on to the complexity of the design. Thus, an architecture with minimal over-head is preferred over the rest. This parameter is given a weight of 4. Form-factor and performance in the power back-off: The next highest priority is given to form-factor and the performance in power back-off with a weight of 3. As mass and volume is a major constraint in nano-satellite design, it is important to choose an architecture that is less voluminous. The form-factor of all of the architectures discussed above is considerably large compared to the conventional operationclasses such as Class-A, B, C, etc. Thus, this parameter is given a weight of 3. All of the architectures discussed are capable of performing equally well in the power back-off levels associated with 16 and 32 APSK modulation schemes, thus the weight is not high. Efficiency and Linearity: Although the discussion so far has been mainly towards choosing a highly efficient and linear architecture, based on the literature study it was seen that there was not much to differentiate between the architectures. Thus, when it comes to trade-off analysis, these parameters are not given a very high weight.
  • 56. 56 MSc Thesis, Visweswaran Karunanithi Based on the weights assigned to each of the parameter, the highest an architecture can score is 95. This following table gives the trade-off: It can be seen from the above table that LINC architecture proves to be a better choice for nano-satellites compared to the other efficiency and linearity enhancement architectures. 3.8. Conclusion An analysis of different efficiency and linearity enhancement architectures were performed and it could be seen that there was no one architecture that suited all applications. All the architectures had a certain advantage over the other, thus a trade-off was performed on the different architectures to see which one would suit nano-satellite application the best. The parameters considered for the trade-off were: complexity/cost, design over-head, form-factor, performance, linearity and efficiency. Based on the trade-off, LINC architecture proved to be the most suitable architecture for nano-satellite application due to its reduced complexity in the RF design, lower over-head and smaller form factor. Thus, it was decided to investigate the LINC PA in more detail to understand some of the implementation challenges. Table 10. Performance trade-off
  • 57. 57 MSc Thesis, Visweswaran Karunanithi 4 4. LINC Architecture s explained in the previous section, the principle of operation of the LINC architecture is by splitting the non-constant envelope signal (AM and PM) into two constant envelope signals (PM signal) and amplify them individually. By doing this, it is possible to drive the PAs at its peak efficiency. The efficiency of the complete system is dependent on: efficiency of the individual PA blocks and the combiner efficiency. On the other hand, as the PAs are driven by constant envelope signals, the linearity of the system is mainly dependent on the linearity of the combiner. There are various choices one can make while selecting an appropriate operating class for a PA and a power combining technique. The subsequent sectionwill give an overview of how an appropriate operatingclass and power combining technique was decided. 4.1. Class of operation The operation classes of PAs in class AB, B, C, D, E, F, F-1 and J are possible to implement on LINC architecture. Based on the literature, it was seen that LINC/Chireix amplifiers were implemented using Class-B [18], C [19], AB and F [20]. The analysis performed in [21] was taken as a reference to choose the best mode of operation. In the analysis, the operating modes considered were Class-AB, B, C, CMCD (current mode Class- D), VMCD (Voltage mode Class-D), E, F, F-1 and J. Simulations were performed on these Classes using CREE GaN HEMT (CGH27015), it was designed for an output power of 40 dBm (10 Watts) at 900 MHz for 16-QAM modulation. Details about the simulation models can be found in [21]. The simulations were done for a symbol rate of 3.84 Msps and roll-off of the filter equal to 0.35. The drain efficiencies of the different classes are as follows: Table 11. Drain Efficiency (DE) for different Classes [21] Class AB B C CMCD VMCD E F F-1 J DE (%) 60 62 64 83 71 88 71 69 75 The PAE for these classes were as follows: A
  • 58. 58 MSc Thesis, Visweswaran Karunanithi Table 12. Simulation results from [21 ] for PAE of the different amplifier classes. Class AB B C CMCD VMCD E F F-1 J PAE (%) 56 55 53 79 69 84 67 65 69 The publication gives a detailed analysis of efficiencies simulated for two different stub compensation angles: 10 deg and 30 deg. The simulated output spectrum which helps is shown in Figure 27. Thus, based on the analysis and results from this publication, it was seen that Class-F, F-1 and CMCD give the best performance in terms of both efficiency and linearity, but CMCD requires two transistors for their implementation whereas Class-F or F-1 require a single transistor per PA. For this work, Class-F implementation was chosen and simulations were performed to verify the results from literature. 4.1.1. Class-F implementation Class-F PAs use the technique of shaping the drain/collector waveform to achieve a better efficiency. Shaping of the waveform is done by appropriate harmonic termination. In the case of class-F mode of operation, the drain is presented with a short-circuit termination at the even harmonics of drain voltage/current, open-circuit termination at odd harmonics and the desired load based on the required output power at the fundamental frequency. This helps in flattening of the voltage waveform, allowing the majority of the drain current to flow when the drain Figure 27. Output spectrum of 16-QAM for different classes of operation. [21]
  • 59. 59 MSc Thesis, Visweswaran Karunanithi voltage is low, resulting in reduced power dissipation. By lowering the power dissipation, the efficiency is improved. The increase in efficiencyis directlyrelated to the number of harmonics handled [22]. Figure 28 shows a functional representation of Class-F mode of operation. By implementing such harmonic tuning at the output, it is possible to shape the drain voltage into a square wave, which helps in reducing the overlap between the current and voltage waveform. An ideal case drain voltage and current waveform is shown in Figure 29. The expression for drain voltage and current can be obtained as explained in [22], but in practical implementations it is not possible to handle all the harmonic components. From [23] it is seen that, by handling only the first four harmonics, it was possible to attain a PAE larger than 80%. Based on the equations derived in [22], γv and γi were used to fine tune the load resistance corresponding to the desired output power. Figure 28. Functional representation of Class-F PA
  • 60. 60 MSc Thesis, Visweswaran Karunanithi The Figure 29 shows a schematic of the output harmonic matching circuit simulated on ADS using ideal components and [figure] shows the output impedance seen by the PA at different harmonic frequencies. The circuit is designed for a fundamental frequency at 900 MHz. The implementation of the Class-F amplifier was done using a GaN HEMT from CREE CGH27015F. Figure 29. Class-F harmonic matching circuit. Figure 30. Load impedance by the drain of the transistor at different harmonic frequencies.
  • 61. 61 MSc Thesis, Visweswaran Karunanithi The IV characteristics the used HEMT is as shown in [figure]. Using the large signal model of CGH27015F from CREE, the simulations were performed on ADS. The harmonic impedance matching circuit shown at the output was implemented at the output of the HEMT and the following figure [figure] shows the drain current and voltage waveforms. The final implementation of the class-F PA was done based on the [20]. More details about the final design integrated with the Chireix combiner is elaborated in subsequent sections. Figure 31. IV characteristics of CGH27015F. Figure 32. Drain voltage and current waveform for a Class-F PA implemented using CGH27015F GaN HEMT.
  • 62. 62 MSc Thesis, Visweswaran Karunanithi 4.2. Different power combining techniques Once the out-of-phase signals are independently amplified using Class-F PAs, the signals are then combined to produce the amplified non-constant envelope signal. The combiner network acts as an adder, enabling the net output amplitude to be controlled via the relative phase of the two non-constant envelope signals. There are various methods of power combining from the literature. The various methods of power combining for LINC system are shown in Figure 33 The method of combining can broadly be classified into two types: using matched combiners and using un-matched combiners. A matched combiner would provide high isolation between the two input ports and provides a constant load impedance to each of the PA for all out-phasing angles. Examples of matched combines are Wilkinson Power Combiner (WPC), Hybrid couplers, rat-race, etc. Although the performance of such combiners are very good in-terms of linearity, the efficiency of combining is traded for the isolation. Almost half the power is dissipated in the isolation resistor. Thus, these techniques were not very popular for the final implementation of LINC, they were mainly used for calibration purpose. In recent time, use of such combiners have re-gained popularity as methods such as RF-DC conversion can be used to recycle the power dissipated in the isolation resistor [24]. Although this method shows promising performance, the circuit over head is quite large and best suited for more than four- way power combining. Due to these reason, this method was not incorporated as the power combining network but ADS simulations were performed to quantify the performance of different matched power combining techniques. Figure 33. RF power combining techniques for LINC.
  • 63. 63 MSc Thesis, Visweswaran Karunanithi Figure 34 shows the two out-of phase input signals V1(t) & V2(t), combined signal (vload) and efficiency as a function of out-phasing angle. Although WPC can provide a good isolation between the two input ports, sometime it is not good enough to completely isolate the two ports. Some of the commercially available WPC s such as ZAPD-30-S+ (from Mini-circuits) provide an isolationbetween 12 and 15 dB. The can Figure 34. WPC as a power combining network for LINC Figure 35. Plot of the two out-of-phase signals, combiner voltage waveformand efficiency of WPC as a function of outphasing angle.
  • 64. 64 MSc Thesis, Visweswaran Karunanithi lead to power getting coupled into the other port and seen at the output of the PA, this leads to degradation in the performance of the PA. This was noticed during the measurement done latter. In-order to over-come this, a quadrature coupler stage can be implemented before a WPC and simulation results show a significant improvement in the isolation. Figure 36 shows the schematic of this setup, Figure 37 shows the coupling between the ports. The simulation was done for a CPW (co-planar waveguide) on a FR-4 substrate. The substrate parameters can be seen in Figure 38. The efficiency and voltage waveforms can be seen in Figure 39 Figure 36. Improvement of isolation by introducing a quadrature coupler between the PA and WPC. Figure 37. Coupling between the two ports.
  • 65. 65 MSc Thesis, Visweswaran Karunanithi Apart from these methods where it is possible to attain high isolation at the cost of efficiency at larger out-phasing angles, there are methods of power combining using unmatched networks. Chireix combiners is one such example of a lossless combiners with low isolation. A Chireix combiner mainly comprises of two simple λ/4 lines with one end having a compensation element (capacitive/inductive) and the other end connected to the load. The concept of using Chireix combiner in LINC can be understood using the schematic representation of the combiner shown in Figure 39. [25]. The output of the PAs are represented by voltage sources V1 and V2. These voltage sources can be expressed in phasor form as follows: ( ) 1 . (cos sin ) j o o V V e V j       (10) Figure 38. Efficiency and voltage waveforms when a quadrature coupler is introduced between the PA and combiner. Figure 39. Schematic representation of Chireix combiner.
  • 66. 66 MSc Thesis, Visweswaran Karunanithi ( ) 2 . (cos sin ) j o o V V e V j        (11) Here, θ is the out-phasing angle that can vary between 0 and 90 deg. Thus, the voltage across the resistor is: | 1 2 | 2 . sin( )L oV V V V j    (12) Based on the derivation from [25], the impedance seen by the two amplifiers can be expressed as:  1 1 cot 2 LR Z j   (13)  2 1 cot 2 LR Z j   (14) The corresponding admittance are: 2 1 2.sin sin2 L L Y j R R     (15) 2 2 2.sin sin2 L L Y j R R     (16) It can be seen that the susceptance seen by both the PAs are different and is a function of the out-phasing angle. The susceptance seen by the PA is depicted in Figure 40. As these susceptances are a function of the out-phasing angle, compensation can be provided based on the PDF of the modulated signal. The relation between out-phasing angle and amplitude is described in equation (9). The value of the inductor and capacitors can be calculated using the imaginary part of the conjugate of admittance from equation (15) and (16) . Figure 40. Chireix combiner as seen at the inputs when fed out-of-phase signals.
  • 67. 67 MSc Thesis, Visweswaran Karunanithi Based on the above equations, a chireix combiner was designed and simulated on ADS, the substrate used was FR4 and the transmission lines used were CPW lines. LineCalc tool inADS was used to calculate the dimensions of the CPW corresponding to the electrical length and characteristic impedance of the line. Figure 41 shows the schematic of the Chireix combiner. The above design is implemented for a compensation angle of 10 degrees. This can be seen from the S-parameter simulation results shown in Figure 42. The initial phase difference without compensation stubs between the input port-1/2 to output port-3 was measured to be Figure 41. Chireix combiner design on FR4 using CPW. Figure 42. S-parameter simulation of the Chireix combinerwith compensation stubs placed at 10 deg.
  • 68. 68 MSc Thesis, Visweswaran Karunanithi 118.2 deg. It can be seen that by adding 10 degree compensation stub, Phase(S(2,3)) has reduced by 10 degrees and the Phase(S(1,3)) has increased by 10 degrees. A comparison was done between WPC and Chireix combiner to quantify the efficiency of the two combiners as a function of out-phasing angle, this plot is shown in Figure 43. It can be seen from the plot in figure that the efficiency is improved for larger out-phasing angles when a Chireix combiner is used. The main drawback with Chierix combiner while compared to WPC is that the linearity is poor at larger out-phasing angles due to the fact that Chireix combiner provides poor isolation between the two input ports. This can be seen in Figure 44 which shows the poor isolation exhibited by the Chierix combiner. Figure 43. Efficiency vs out-phasing angle comparison between Chireix combiner and WPC. Figure 44. Comparison of coupling between Chireix combiner and WPC.
  • 69. 69 MSc Thesis, Visweswaran Karunanithi One other power combining technique that has surfaced in recent time is, spatial power combining. In this technique, the output of the PAs are fed directly to two separate antennas and the radiated power from the individual antennas combine in the air to form the non-constant envelope signal. Using the idea from [26], simulations were performed on a UHF antenna combiner on nano-satellites. A simulation on FEKO was done on crossed dipole for different out-phasing angles to see its effect on the radiation pattern. The radiation pattern when the dipoles are fed with signals having an out-phasing angle of 45 degrees is shown in Figure 45 The radiation pattern of the crossed dipole when fed with zero out-phasing angle is shown in Figure 46. Figure 45. Radiation pattern of crossed dipole on a 3U cube-sat fed with an out-phasing angle of 45 degrees.
  • 70. 70 MSc Thesis, Visweswaran Karunanithi It can be seen that the beam-width has increased and the peak gain has decreased from 3.4 dBi to 3 dBi. Thus, it could be seen that the main effect on the radiation pattern due to the change in out- phasing angle is the variation in the beam-width. The gain variation in the bore side was not much. The crossed dipole arrangement provides fair isolation between the two antennas. Figure 46. Radiation pattern of the crossed dipole when fed with signals with 0 degree outphasing angles. Figure 47. Radiation pattern of crossed dipole fed with signals with 90 degrees out-phasing angle.
  • 71. 71 MSc Thesis, Visweswaran Karunanithi Although, the simulation results show promising results, further analysis needs to be performed to use it in a LINC architecture, making this an interesting topic for future work. After analyzing various amplifier cells and different power combining techniques, it was decided implement the LINC architecture using Class-F PA and a Chireix combiner. In order to show a proof of concept, a LINC architecture designed by CATENA microelectronics BV was used [27] This board had implemented Class-F PA cells and a Chireix combiner, thus this board was used to validate the above simulation results and propose possible improvement to the design. The further sections will describe the measurements performed on the LINC PA for different modulation schemes.