Thèse n° 7489

Antennas for CubeSat Communication

Présentée le 6 mars 2020

à la Faculté des sciences et techniques de l’ingénieur

Groupe SCI STI AS
Programme doctoral en génie électrique

pour l’obtention du grade de Docteur ès Sciences

par

Miroslav VELJOVIC

Acceptée sur proposition du jury

Dr S.-R. Cherkaoui, président du jury

Prof. A. Skrivervik, directrice de thèse
Prof. J. Encinar, rapporteur
Prof. C. Luxey, rapporteur
Prof. R. Fleury, rapporteur

2020
Two antennas got married on a roof.
The ceremony wasn’t much,
but the reception was excellent!

(unknown author)
Abstract
CubeSats are a type of small satellites (< 500 kg) that weigh several kilograms and consist of
multiple Units (U) measuring 100 × 100 × 113.5 mm3. CubeSats emerged as a low-cost alternative
to conventional large satellites, and have since demonstrated capabilities for communication,
Earth observation, technology demonstration and many other. Following their success in the
research and education domains, for which they were developed, CubeSats have attracted the
attention of industry thanks to their cost-efficient, yet promising, features.

The primary goal of CubeSats, as it is the case with any miniaturized satellite, is to reduce the cost
of orbital deployment. CubeSats are commonly launched as a secondary payload on large launch
vehicles. One or several satellites are placed in a dedicated deployment system, which typically
accommodates three CubeSat Units, and deployed with a loaded spring when the satellite is in
orbit. The deployment system strictly limits the dimensions of any features on the CubeSat
surface.

CubeSat antennas perform the same functions as antennas on conventional satellites, such as
telemetry and command, communication, navigation or inter-satellite links (ISL). However, most
traditional antennas for small satellites are not suitable for CubeSats due to the profile constraints
of the deployment system. Two approaches for antenna design are generally adopted. In the first,
one or several low-profile antennas are placed on the CubeSat exterior. In the second, deployable
antenna structures are used if the CubeSat platform does not offer sufficient dimensions. At low
frequencies, the additional space is required to achieve a good radiation efficiency, whereas at
high frequencies an increased antenna aperture provides a high gain. In any case, the stowed
volume of the antenna may not violate the profile limitation specified by the standard.

This thesis describes different antenna geometries, both low-profile and deployable, which
enable several communication aspects of an Internet-of-Things (IoT) constellation of 3U
CubeSats in the Low Earth Orbit (LEO). Wideband aperture-coupled patch antennas for TT&C
and data downlink in S and X bands are presented in the beginning. A thoroughly investigated
shielded-stripline feeding structure enables a wideband circular polarization (CP) with a
unidirectional radiation, and demonstrates capabilities such as interference suppression at
adjacent frequencies and efficient CubeSat integration. Patch-antenna arrays in L band are then
presented, which radiate several independent beams for the capacity increase of an IoT machine-
to-machine (M2M) communication system. The high permittivity of the antenna substrates lead
to a strong coupling to the 3U CubeSat structure, emphasizing the importance of the antenna
placement. Finally, different configurations of high-gain fixed-beam reflectarrays (RA) and
transmitarrays (TA) are for the first time proposed for CubeSat ISL LEO communications in K
band. Two novel unit cells are presented for RA and TA antennas, based on coupled loops and
aperture-coupled patches, respectively. An axially corrugated CP horn antenna is designed as the
feeding element for the arrays. A prototype of this antenna is fabricated using additive
manufacturing in aluminum.

The main novelty of the proposed solutions is in their electromagnetic performance, as opposed
to most CubeSat antenna designs where the mechanical design is the main achievement. The
presented antennas succeed to exhibit a wide bandwidth or a specified radiation under strict size
constraints. The performance of all antennas, presented in this thesis, is validated by agreements
between the calculation and 3D-simulation results and the VNA and anechoic-chamber
measurements of several prototypes.

Keywords: 3D printing, additive manufacturing, antenna, antenna array, aperture coupling,

CubeSat, feeding network, Internet-of-Things, inter-satellite link, nanosatellite, patch antenna,
reflectarray, satellite communication, transmitarray, wideband
Résume
Le CubeSat est une catégorie de petits satellites (< 500 kg) qui pèsent quelques kilogrammes et se
compose d’une ou plusieurs unités (U) mesurant 100 × 100 × 113,5 mm3. Le CubeSat est apparu
comme une alternative à faible coût aux satellites conventionnels et a depuis démontré ses
capacités de communication, d'observation de la Terre, de démonstration technologique etc.
Suite à son succès dans les domaines de la recherche et de l'éducation pour lesquels il a été
développé, le CubeSat a attiré l'attention de l'industrie grâce à ses capacités prometteuses et peu
coûteuses.

L'objectif principal du CubeSat, comme pour tous les satellites miniatures, est de réduire le coût
du déploiement orbital. Les CubeSats sont généralement lancés en tant que charge utile
secondaire sur les grandes fusées. Un ou plusieurs satellites sont placés dans un système de
déploiement dédié, qui accueille généralement trois unités de CubeSat. Ils sont déployés avec un
ressort lorsque l’orbite visée est atteinte. Le système de déploiement limite strictement les
dimensions de tous les dispositifs placés sur le châssis d’un CubeSat.

Les antennes pour CubeSats fonctionnent de la même manière que les celles des satellites
conventionnels : elles assurent la télémétrie et la commande, la communication, la navigation ou
les liaisons inter-satellites (ISL). Cependant, la plupart des antennes traditionnelles pour petits
satellites ne conviennent pas aux CubeSats à cause des contraintes de taille du système de
déploiement. Deux approches pour la conception d'antennes sont généralement adoptées. Dans
la première, une ou plusieurs antennes à bas profil sont placées sur la surface externe du CubeSat.
Dans la seconde, quand la plateforme CubeSat n'offre pas les dimensions suffisantes pour un bon
rendement à basses fréquences, ou pour un gain élevé à hautes fréquences, des antennes
déployables sont utilisées. Dans les deux cas, les dimensions de l'antenne non déployée ne
doivent pas dépasser les limitations de volume.

Cette thèse adresse les géométries d'antennes minces et/ou déployables qui couvrent plusieurs
aspects de communication d'une constellation « Internet of Things » (IoT) de CubeSats 3U sur
orbite terrestre basse (LEO). La thèse présente plusieurs antennes patch à large bande excitées au
travers d'une fente dans le plan de masse pour la télémétrie et la commande (TT&C) ainsi que
pour la liaison descendante de données dans les bandes S et X. Une structure d'alimentation
triplaque blindée assure une performance à polarisation circulaire dans une large bande avec un
rayonnement unidirectionnel. Cette structure permet aussi la suppression des interférences dans
les fréquences voisines et l'intégration efficace dans un CubeSat. Des réseaux d'antennes patch
en bande L sont ensuite présentés, qui permettent de rayonner plusieurs faisceaux indépendants
pour l'augmentation de capacité d'un système de communication IoT. La permittivité élevée des
substrats d'antennes conduit à un fort couplage avec le châssis du CubeSat 3U, mettant en
évidence l'importance du placement de l'antenne. Enfin, différentes configurations d’antennes
type « reflectarray » (RA) et « transmitarray » (TA) à gain élevé sont proposées pour la première
fois pour la communication ISL en bande K pour les CubeSats. Deux nouveaux éléments unitaires
sont présentés pour les antennes RA et TA respectivement, basées soit sur des boucles couplées
soit sur des patchs couplés par une fente dans le plan de masse. Une antenne cornet corrugué à
polarisation circulaire est conçue pour alimenter les réseaux. Un prototype de cette antenne est
réalisé en utilisant la fabrication additive en aluminium.

La principale originalité des résultats obtenus est dans la qualité des performances
électromagnétiques, contrairement à la plupart des études où l’innovation se trouve dans la
conception mécanique des antennes. Les antennes développées possèdent une large bande
passante pour une performance de rayonnement améliorée par rapport à l’état de l’art. De plus,
ces caractéristiques sont obtenues sous des contraintes géométriques strictes. Les performances
de toutes les antennes présentées dans cette thèse sont validées par la comparaison entre les
résultats numériques et les mesures de plusieurs prototypes.

Mots-clés : impression 3D, fabrication additive, antenne, réseau d'antennes, antennes excitée par
fente, CubeSat, réseau d'alimentation, Internet of Things, liaison inter-satellites, nanosatellite,
antenne patch, reflectarray, communication par satellite, transmitarray, large bande
Сажетак
CubeSat је тип наносателита који се састоји из запреминских јединица (U) димензија
100 × 100 × 113.5 mm3. Сателити типа CubeSat представљају економичну алтернативу
великих сателита и од свог настанка су демонстрирали функције као што су радио‐
комуникације, фотографисање Земље, итд. На основу успеха у областима образовања и
истраживања, за које су првенствено намењени, CubeSat сателити су привукли пажњу
индустријског сектора захваљујући својим економичним, али обећавајућим особинама.

Главни разлог за развој наносателита је смањење трошкова њиховог лансирања у орбиту.

CubeSat сателити се најчешће лансирају као секундарни товар на великим ракетама. Један
или неколико сателита, укупне запремине до 3U, се постављају у специјалан одељак који
избацује сателите помоћу опруге када достигну одговарајућу орбиту. Овакви системи за
лансирање строго ограничавају максималне дозвољене димензије CubeSat сателита.

Антене на CubeSat сателитима врше исте функције као и антене на конвенционалним

сателитима. Међутим, већина антена за друге врсте малих сателита нису одговарајуће за
CubeSat сателите због физичких ограничења система за лансирање. Антене за CubeSat
сателите се могу поделити у две групе. У прву групу спадају антене ниског профила које се
налазе на површини сателита. Антене друге групе користе различите врсте механизама за
повећање укупних димензија антене. Такво повећање је неопходно како би се остварио
задовољавајућ степен корисног дејства или добитак антене на ниским или високим
фреквенцијама, респективно.

Ова дисертација описује неколико антена које омогућавају различите радио‐

комуникационе функције унутар мреже 3U‐CubeSat сателита у ниској Земљиној орбити.
Неколико широкопојасних микротракастих антена у S и X фреквенцијским опсезима,
спрегнутих са водом за напајање путем прореза, су представљене на почетку. Оклопљена
структура за напајање у техници асиметричних тракастих водова омогућава код антена
овог типа широкопојасну кружну поларизацију израчених таласа, потискивање
интерференције из суседних фреквенцијских опсега и ефикасну интеграцију са металном
CubeSat структуром. Затим су представљени низови микротракастих антена у L опсегу који
зраче неколико независних снопова у циљу повећања капацитета радио‐комуникационог
система. Висока пермитивност супстрата изазива јаку спрегу антена са структуром
сателита, због чега се положај антене на CubeSat структури мора узети у обзир.
Последњи део ове дисертације предлаже нове конфигурације антенских низова великог
добитка у K опсегу за радио‐комуникацију између CubeSat сателита, типа reflectarray и
transmitarray. По један елемент је дизајниран за сваки од ових низова, на бази спрегнутих
микротракастих петљи и микротракастих антена спрегнутих помоћу прореза. Кружно
поларисана левак антена је дизајнирана специфично за напајање датих низова. Прототип
левак антене је направљен помоћу технике 3D штампе алуминијумског прашка.

Електромагнетске карактеристике антена описаних у оквиру ове дисертације

представљају њихов главни новитет, насупрот већини досадашњих CubeSat антена, чији
допринос лежи у механичком дизајну. Описане антене испoљавају широкопојасне
карактеристике упркос строгим ограничењима укупних димензија. Поклапање резултата
нумеричких прорачуна и симулација са једне стране, те мерења у анехоичној комори путем
векторског анализатора мрежа са друге стране, потврђује рад свих антена представљених
у оквиру ове дисертације.

Кључне речи: 3D штампа, CubeSat, антена, антенски низ, међу‐сателитска веза,

микротракаста антена, мрежа за напајање, наносателит, сателитске радио‐комуникације,
спрега путем прореза, тракасти вод, широкопојасни
Acknowledgment
I would like to thank the members of the jury, Prof. Jose Encinar, Prof. Cyril Luxey, Prof. Romain
Fleury, and the jury president, Dr. Sidi-Rachid Cherkaoui, for their participation in the oral exam
of this thesis. Their useful comments and suggestions have improved the quality of the final
manuscript and have sparked new ideas for the future work on this topic. I am thankful to
Astrocast for supporting the work on this thesis, and I especially enjoyed working with Mehrdad,
Ester and Kevin during this time. The friendly ACI team at EPFL, notably José, Jean-Marie, Jean-
Marc, performed the fabrication of many prototypes, and their work is greatly appreciated. I wish
to express my gratitude to Tomislav and Antoine of SwissTo12, and Frank, Przemyslaw and
Alberto of Viasat, for their help and advice during prototype measurements.

Working on a PhD thesis is a stressful task, unless your thesis advisor is Anja Skrivervik. Her
scientific and diplomatic skills, devotion to PhD students and a warm personality have allowed
me to be optimistic and productive, while enjoying my work in the previous years. Together with
Prof. Juan Mosig, she created a friendly, supportive atmosphere in LEMA, and afterwards MAG,
which I will always remember for the scientific and inter-cultural exchanges during many social
events. A big thank you goes to all MAG, LEMA and LWE members, with whom I had a big pleasure
working and interacting over the past four years. For the most of this period, my office mates were
two wonderful people, Danelys and Anton. Known as ‘The Skriverviks’, the three of us had many
jokes and interesting talks, and they were the main people responsible for my good mood. I also
wish to thank Santiago for his unconditional assistance and scientific discussions we had in the
early years of my PhD.

The most valuable aspect of the PhD at EPFL is the chance to meet people from all around the
world and learn about new cultures, while challenging your views at the same time. Among many
others, I mostly wish to thank my Greek friends and the Satellite team, for many inspiring
discussions and relaxing events. Over the years, however, I have received the largest support in
every aspect from the Serbian community in Lausanne, especially the notorious group ‘Kafana na
Balkanu’. They played a big role in this thesis, and for that, they have my sincere gratitude.

During the PhD, one person went alongside me through every achievement and obstacle we
encountered together. For the love you have given me every single day, Katarina, you deserve to
share this success as your own.

My family was and continues to be my biggest support. In spite of having spent over two decades
in education, my family has taught me the most valuable lessons in life. Mama, Tata, Milena,
Jovana, Luka, to you I dedicate this thesis.
Acronyms and Abbreviations
ADCS Attitude Determination and Control System
AR Axial ratio
COTS Commercial off-the-shelf
CP Circular polarization
DMLS Direct laser metal sintering
EM Electromagnetic
FEM Finite-element method
FN Feeding network
FSS Frequency-selective surface
FTBR Front-to-back ratio
GNSS Global navigation satellite system
IoT Internet of Things
ISL Inter-satellite link
LEO Low Earth orbit
LP Linear polarization
M2M Machine to machine
MoM Method of moments
PEC Perfect electric conductor
P-POD Poly-picosatellite deployer
RA Reflectarray
RHCP/LHCP Right/left-hand circular polarization
SO Space operations
SLL Sidelobe level
TA Transmitarray
TT&C Telemetry, tracking & command
U Unit
WG Waveguide
Contents
1. General Introduction ................................................................................................................. 1
1.1. The CubeSat Standard .......................................................................................................... 1
1.2. CubeSat Antennas ................................................................................................................. 2
1.3. Objectives and Thesis Organization .................................................................................... 6
1.3.1. Motivation and Objectives ........................................................................................ 6
1.3.2. Thesis Outline ............................................................................................................ 8
1.3.3. Research Projects....................................................................................................... 9

2. Low-Profile Wideband CubeSat Antennas .............................................................................. 11

2.1. Review of CubeSat TT&C Antennas ................................................................................... 12
2.1.1. VHF/UHF Bands ...................................................................................................... 13
2.1.2. S Band ....................................................................................................................... 13
2.1.3. X Band ...................................................................................................................... 15
2.2. Antenna Design Requirements .......................................................................................... 16
2.3. Microstrip-Fed Aperture-Coupled Patch Antenna ........................................................... 17
2.4. Stripline-Fed Aperture-Coupled Patch Antenna .............................................................. 19
2.4.1. Enclosed-Stripline Feeding Network ..................................................................... 21
2.4.2. Antenna Geometry and Performance .................................................................... 21
2.5. Stripline-Cavity Resonance Analysis ................................................................................. 24
2.5.1. Interference Suppression in an Adjacent Frequency Band .................................. 26
2.6. TT&C Antenna ..................................................................................................................... 28
2.6.1. CubeSat Integration................................................................................................. 28
2.6.2. Antenna Prototype Construction ........................................................................... 29
2.6.3. Antenna Performance ............................................................................................. 32
2.7. X-band Antenna .................................................................................................................. 32
2.7.1. Antenna Geometry .................................................................................................. 33
2.7.2. Simulation Results ................................................................................................... 34
2.7.3. 4-Element Antenna Array ....................................................................................... 36
2.8. Summary and Discussion ................................................................................................... 37

3. L-band Antenna System for CubeSats ..................................................................................... 41

3.1. Review of Solar-Panel-Integrated Antennas ..................................................................... 42
3.2. Design Requirements.......................................................................................................... 44
3.3. Single Beam (Stage 1).......................................................................................................... 45
 3.3.1. Chassis Coupling Effects .......................................................................................... 45
3.3.2. Antenna Prototype ................................................................................................... 46
3.4. Two Tilted Beams (Stage 2) ................................................................................................. 49
3.4.1. Sequential Rotation of CP elements ....................................................................... 49
3.4.2. Antenna Array Prototype ......................................................................................... 53
3.5. Four Tilted Beams (Stage 3) ................................................................................................ 55
3.6. Stacked-Patch Antenna in L and S bands .......................................................................... 56
3.6.1. Patch Antenna Elements ......................................................................................... 56
3.6.2. Feeding Networks .................................................................................................... 60
3.6.3. Full Antenna Models ................................................................................................ 61
3.6.4. Antenna Prototype ................................................................................................... 64
3.7. Summary and Discussion ................................................................................................... 66
3.7.1. Antennas Flying in Orbit .......................................................................................... 68

4. Inter-Satellite Links for LEO CubeSats .................................................................................... 69

4.1. Review of ISL Antennas ....................................................................................................... 70
4.2. ISL Communication Scenarios ........................................................................................... 72
4.3. ISL System Requirements ................................................................................................... 74
4.3.1. Frequency ................................................................................................................. 74
4.3.2. Antenna Radiation ................................................................................................... 75
4.3.3. Antenna Size ............................................................................................................. 76
4.4. Reflectarray and Transmitarray Antennas ......................................................................... 78
4.4.1. Antenna Configurations for CubeSat ISL ............................................................... 79
4.4.2. A Unified TA/RA Antenna ....................................................................................... 80
4.5. ISL Antenna Design Requirements .................................................................................... 82
4.5.1. RA Antenna ............................................................................................................... 82
4.5.2. TA Antenna ............................................................................................................... 83
4.6. Summary .............................................................................................................................. 84

5. Reflectarray and Transmitarray Components ........................................................................ 87

5.1. Element Design Approach .................................................................................................. 88
5.1.1. Infinite-Array Approximation ................................................................................. 88
5.1.2. Variable Rotation Technique .................................................................................. 89
5.2. TA Element: CP Aperture-Coupled Patches ...................................................................... 92
5.2.1. Geometry Evolution ................................................................................................. 93
5.2.2. Simulation Results ................................................................................................... 94
5.3. RA Element: Coupled Loops ............................................................................................... 96
5.3.1. Geometry Evolution ................................................................................................. 97
5.3.2. Modal Analysis ......................................................................................................... 99
 5.3.3. Full-Wave Simulation Results ............................................................................... 101
5.4. 3D-Printed Corrugated-Horn Feed Chain ...................................................................... 102
5.4.1. Array Focal Distance.............................................................................................. 103
5.4.2. Axially Corrugated Horn ....................................................................................... 104
5.4.3. Septum Polarizer ................................................................................................... 105
5.4.4. Full Feed Chain ...................................................................................................... 107
5.4.5. Future Integration with a CubeSat ....................................................................... 108
5.5. Summary and Discussion ................................................................................................. 111

6. RA/TA Design Process ............................................................................................................ 113

6.1. Design Process Overview .................................................................................................. 114
6.2. Feeding-Antenna Radiation ............................................................................................. 115
6.2.1. Methodology .......................................................................................................... 115
6.2.2. Calculation of Incident Fields ............................................................................... 117
6.3. Array Synthesis .................................................................................................................. 117
6.3.1. Unit-Cell Incidence Angle ..................................................................................... 117
6.3.2. Assumptions .......................................................................................................... 118
6.3.3. Element Rotations ................................................................................................. 118
6.4. Array Analysis .................................................................................................................... 120
6.4.1. Alternative Approach ............................................................................................ 122
6.5. Aperture Efficiency ............................................................................................................ 122
6.5.1. Discussion .............................................................................................................. 123
6.6. Array Simulations .............................................................................................................. 124
6.7. Prototype Measurements ................................................................................................. 129
6.7.1. Array Prototypes .................................................................................................... 129
6.7.2. RA Measurement Results ...................................................................................... 130
6.7.3. TA Measurement Results ...................................................................................... 134
6.8. Summary and Discussion ................................................................................................. 137

7. Conclusions and Perspectives ............................................................................................... 139

7.1. Summary ............................................................................................................................ 139
7.2. Discussion.......................................................................................................................... 143
7.3. Future Work ....................................................................................................................... 144
7.4. Perspectives ....................................................................................................................... 145

References .................................................................................................................................. 149

Publications ................................................................................................................................ 155

Curriculum Vitae ........................................................................................................................ 157

1. General Introduction

1.1. The CubeSat Standard

CubeSat is a standard for modular nanosatellites, defined by the CubeSat Design Specification
(CDS), also known as the CubeSat standard [1]. A CubeSat is composed of multiple volume units
(U) of 100 x 100 x 113.5 mm3. The CubeSat size varies from 0.25U to 27U, and some common form
factors are shown in Figure 1.1 [2]. Strictly speaking, a nanosatellite is defined in terms of mass
(not volume) and implies any spacecraft with mass between 1 kg and 10 kg [3]. The two definitions
are sometimes misleading since CubeSats in general can have a mass from less than 1 kg to 30 kg,
but a majority falls within the nanosatellite mass range. The two terms are even sometimes
mistakenly used as synonyms, due to the popularity of the CubeSats in the recent years.

The CubeSat was invented in 1999, by Professor Jordi Puig-Suari of California Polytechnic State
University and Professor Bob Twiggs of Stanford University, as an academic project that allowed
graduate students to design and operate a small spacecraft [4]. The 100-mm cube was deemed
sufficient to accommodate basic satellite components – a communication payload, solar panels
and a battery. The first CubeSat was launched in 2003, and over 1000 CubeSats are deployed to
orbit to date, with over 80 destroyed in failures of launch vehicles [2]. The majority of CubeSats
were developed for academic and educational purposes until 2013, when the CubeSat number
starts growing rapidly, a majority being for commercial applications. The development cost of
early CubeSats was reduced by using commercial off-the-shelf (COTS) components for onboard
systems, and conducting only the most important space-eligibility tests. New CubeSats
increasingly use high-performance custom-made modules, which include communication,
propulsion, antennas, etc., specifically developed for certain missions [2].

CubeSats typically serve purposes such as Earth observation, IoT communication, amateur radio,
etc [3], [5]. They present a cost-efficient opportunity to validate a new technology or even perform
biological experiments in space [6]. Several countries used CubeSats as their first national
satellite. Two 6U CubeSats were successfully used in 2018 during a mission to Mars, with more
interplanetary CubeSats planned [7].
1. General Introduction

Figure 1.1. CubeSat models with different common form factors.

The main motivation for the development of any nanosatellite is the cost reduction of the orbital
deployment. CubeSats are typically launched to Low Earth Orbit (LEO) as a secondary payload of
larger missions or from the International Space Station [3]. A CubeSat dispenser (or deployer)
system, called the Poly-PicoSatellite Orbital Deployer (P-POD), was developed simultaneously
with the CubeSat standard [1]. The system is an aluminum compartment that accommodates a
3U CubeSat (or multiple smaller form-factor satellites) with a hatch and a spring mechanism.
Following the hatch opening, the system deploys nanosatellites by simply ejecting them into orbit
using a loaded spring, located at the bottom of the compartment. Most modern dispensers
operate with a similar principle. The dispenser strictly limits the maximum feature dimension on
the spacecraft surface. Therefore, any deployable components (solar panels, antennas,
magnetometers) must be stowed within this volume during launch, with a typical profile margin
from 6.5 mm to 9 mm.

The total number of launched and announced CubeSats is presented in Figure 1.2, versus their
form factor. The first CubeSats were typically 1U. It was shown over time that a 3U CubeSat offers
a good compromise between the cost and the available volume for onboard components, making
it the most popular form factor to date. As the CubeSats in the recent years start to perform
functions that are more complex, the 6U size is also gaining in popularity.

1.2. CubeSat Antennas

Early CubeSats typically communicated in VHF, UHF and S bands due to good propagation
characteristics at those frequencies and the availability of COTS components and modules. Since
then, the CubeSat communication and other RF functions (e.g. RADAR) are increasingly
performed at higher frequencies reaching Ka band, with some demonstration missions for optical
communications [8]. The number of reported CubeSat communication modules versus the
operating frequency is shown in Figure 1.3.

2
 1.2. CubeSat Antennas

Figure 1.2. Launched and announced CubeSats versus the form factor [2].

Figure 1.3. Number of reported CubeSat communication modules versus the operating
frequency band [2].

A known tradeoff is considered when selecting the communication frequency of a CubeSat

mission – lower frequencies generally offer a narrow bandwidth/low data rate and low path losses,
and vice versa. The CubeSat antenna dimensions, complexity and the achievable performance
also greatly depend on the frequency choice. Thus, the antenna type is a very important factor to
be taken into account during the mission design. Antenna dimensions are proportional to the
wavelength at the operating frequency. The antenna size normalized to the free-space
wavelength, also known as the electrical size, often determines what kind of radiation can be

3
1. General Introduction

expected. Typically, electrically smaller antennas exhibit a wide beamwidth and low gain
values, whereas electrically large antennas 5 radiate high-gain narrow or shaped beams.

The electrical size of an entire 1U CubeSat platform is approximately 0.05 to 12 from VHF to
Ka band, where CubeSats communication is performed. This suggests that the antennas, used
across this vast spectrum, will have very different geometries to conform to the CubeSat chassis
and achieve the specified performance. Typical antenna geometries, along with several exotic
examples, are presented in Figure 1.4 versus the operating frequency. The 1U, 2U and 3U CubeSat
dimensions are denoted in the figure for comparison.

Figure 1.4. Examples of CubeSat antennas throughout the frequency spectrum. (Top, left to
right) Deployable dipoles [9], MarCO reflectarray and planar UHF antenna [7], patch-antenna
array [10], mesh reflector [11], one-meter reflectarray [12], ISARA [13]. (Bottom, left to
right) GOMX-3 turnstile and helix [14], COTS GNSS patch, CP patch [15], isoflux patch with
chokes and parasitic dipoles [16], patch-antenna arrays [17], bull’s eye antenna [18].

4
 1.2. CubeSat Antennas

The figure should be considered in combination with the number of communication modules in
Figure 1.3 to get an insight into the representation of certain antenna geometries. It is by no means
exhaustive, and more examples from the literature are presented throughout this thesis and
discussed in detail at the beginning of every chapter. Colored markers in the figure represent the
antennas developed in the scope of the thesis, where the colors correspond to the labels in
Figure 1.7.

A general trend is evident in Figure 1.4, with relatively small antennas in L and S bands, and an
increase in antenna size at the lower and higher frequencies. However, the size increase is driven
by different mechanisms at the two frequency extremes. Low-frequency antennas (dipole,
monopole, turnstile) require large dimensions simply to achieve a resonance with a good
radiation efficiency. On the other hand, the large antenna apertures for the high-frequency
antennas (array, reflector, reflectarray, etc.) compensate large path losses with a high gain or
provide narrow beams for applications like RADAR [11], [12].

The CubeSat in-orbit orientation with respect to Earth (attitude) is determined and maintained
by the attitude determination and control system (ADCS) [19]. The ADCS relies on sensors such
as sun sensors, star trackers, gyroscopes, etc., and actuators such as thrusters, momentum wheels,
moment gyros, etc. The process can be assisted by a purely passive attitude control using the
gravity gradient or the Earth’s magnetic field, depending on the specific CubeSat shape [20]. Early
1U CubeSats did not include an ADCS and they required near-omnidirectional antennas to
sustain the communication link, regardless of the orbital attitude. Modern, miniature types of
ADCS that are suitable for CubeSats achieve three-axis stabilization up to within 1°. Accurately
maintaining the CubeSat orientation is crucial for the utilization of directive antennas or arrays.
A comparison between the two antenna-ADCS scenarios is illustrated in Figure 1.5.

Figure 1.5. A comparison of two CubeSat communication scenarios – with or without ADCS –
and examples of associated antenna geometries.

5
1. General Introduction

1.3. Objectives and Thesis Organization

1.3.1. Motivation and Objectives
The research in this thesis was done in collaboration with Astrocast – a Swiss start-up company
that builds and operates an IoT CubeSat constellation in LEO [21]. The company was founded in
2014 by the team that successfully developed and launched the SwissCube, EPFL’s 1U research
CubeSat, which holds the world record for the longest continuous operation of a CubeSat (over 10
years at the writing time of this thesis) [22].

The aim of the Astrocast network of CubeSats is to provide bidirectional, cost-efficient

communication to remote Earth areas in L band. The network nodes are 3U CubeSats, having
their long axis aligned with the orbital velocity vector, with a large face oriented towards Earth at
all times. Although this attitude is less advantageous in terms of passive stabilization with the
gravity gradient, it allows more planar antennas to be placed on the CubeSat surface facing Earth.
Each satellite is equipped with an accurate ADCS and a small propulsion module for maintaining
the attitude and the orbital velocity over time. The CubeSats inside the network are uniformly
distributed over several polar orbital planes. An inter-satellite link (ISL) is established by each
satellite with two adjacent satellites in the same orbital plane [23]. More details about the
configuration of the constellation can be found in Chapter 4.

The data is sent through the network using different frequencies, protocols and data rates [24].
The system is organized as shown in Figure 1.6. The CubeSats provide low-data-rate bidirectional
M2M L-band links to a large number of user terminals within their coverage area. Each CubeSat
radiates several independent beams through dedicated M2M transceivers, to increase the total
system capacity. The aggregated user data is then sent through a high-data-rate K-band ISL
among CubeSats within the same orbital plane. Once the data reaches the satellite, which has a
connection established with the ground station, the data is downlinked to the Earth using a high-
capacity X-band channel.

A separate Telemetry, Tracking and Command (TT&C) system, operating in S band, constantly
monitors and performs orbital station-keeping of the CubeSat constellation. The satellites are also
equipped with GNSS receivers, for accurate positioning.

Each of the described functions requires specialized antennas onboard every CubeSat platform.
The design specifications for individual antennas are drastically different, based on the
requirements of the corresponding functions. Circular polarization (CP) is used in all the
mentioned links. Other RF specifications include the operating frequency, radiation beamwidth,
maximum gain, impedance matching, and frequency bandwidths in terms of the return loss, gain
and axial ratio (AR).

6
 1.3. Objectives and Thesis Organization

Figure 1.6. The overview of the CubeSat network (constellation).

Figure 1.7. A sketch of the antennas presented in this thesis, placed on a 3U CubeSat.

The goal of this thesis is to propose novel antenna geometries, which satisfy the mentioned RF
specifications under the strict volume constraints of the CubeSat standard, primarily the thickness
limitation. Depending on the operating frequency, the size constraints have a different impact on
the antenna design in terms of the electrical size. The antenna placement on the exterior of the
satellite may in no case violate the thickness specified by the used dispenser system. Two types of
designs are considered, depending on the function – reliable and robust low-profile antennas on
the CubeSat surface for the critical functions such as TT&C, and deployable antenna arrays in
cases where additional gain is required.

7
1. General Introduction

1.3.2. Thesis Outline

Each chapter of this thesis opens with a short literature review of the CubeSat antennas having the
same, or closely related, function and geometry as the designs presented in the chapter itself. The
3U CubeSat platform, considered for all the antenna designs in this thesis, is shown in Figure 1.7.
The figure indicates the locations of the antenna design within the thesis. The operating frequency
and the size of the antennas is also indicated in Figure 1.4 with the same colors as in Figure 1.7.
The thesis is organized as follows:

Chapter 2 presents several low-profile wideband CP stripline-fed aperture-coupled patch

antennas in S and X bands. A shielded asymmetric-stripline structure enables a wideband
element matching in spite of the CubeSat profile limitations at S-band frequencies [C1]. A detailed
study of the electric field inside the shielded-stripline cavity helped to explain the underlying
mechanisms for wideband operation, outlining a set of design rules for this type of antennas [J1].
The same approach is demonstrated on X-band aperture-coupled patch antennas and arrays,
having simpler feeding networks. It is further shown how the cavity resonances can be used to
suppress interference in an adjacent frequency band. The chapter proposes a novel integration
method of the aperture-coupled patch antennas into the CubeSat metallic structure.

Chapter 3 describes a patch-antenna system for CubeSat duplex IoT communication in L band
and proposes two- and four-element array configurations, where the elements are placed on the
metallic backside of deployed solar panels. The antenna system radiates several independent
beams to increase the IoT system capacity. The high-permittivity substrate, used for antenna
miniaturization, causes a strong coupling with the asymmetric 3U ground plane. The resulting
distortion of the radiation pattern demonstrated the influence of the antenna placement on the
CubeSat [C2]. The chapter finally presents a dual-band dual-fed stacked-patch antenna in L and
S bands, as means of optimizing the limited area on the CubeSat surface [C3]. Several feeding
networks are compared, showing that the choice of the feeding method does not significantly
affect the CP-gain bandwidth of the stacked-patch antenna.

Chapter 4 illustrates how a specific flying formation of the CubeSat network can be used to
simplify the conventional complex high-gain ISL antennas. Design requirements are discussed
for the considered CubeSat LEO constellation, in terms of the beam-scanning angle and the gain,
versus the total number of CubeSats and the antenna aperture efficiency. The chapter for the first
time proposes deployable fixed-beam reflectarray (RA) and transmitarray (TA) antennas for ISL
applications in K band. Several RA and TA configurations are presented for 3U CubeSats.

Chapter 5 presents the elements of the RA and TA antennas from Chapter 4 – two novel RA and
TA element geometries and a 3D-printed CP axially corrugated feed horn antenna. The
foundations of the element-rotation technique and the unit-cell simulation environment are
presented first. The coupled-loops RA element is then designed starting from the conventional
split loop, significantly improving the CP reflection bandwidth while not compromising the

8
 1.3. Objectives and Thesis Organization

complexity [J2]. Furthermore, the slot shape of the previously reported aperture-coupled stacked-
patches TA element is modified, which allowed reducing the number of metallic layers from five
to three [J3]. Both proposed elements exhibit an excellent CP bandwidth with a very low profile,
which is one of the main prerequisites for CubeSat ISL applications. A CP feed chain is designed
as the feeding element of the RA and TA antennas, in which the axial corrugations improve the
off-axis cross-polarization, and a septum polarizer enables a dual-CP operation. The
measurement results of the horn prototype, fabricated using additive manufacturing in
aluminum, confirm the predicted performance [C5].

Chapter 6 opens with a description of the array synthesis and analysis procedures, based on the
simulation results of RA and TA elements and the feed horn presented in Chapter 5. Both
antennas exhibit a wideband high-purity CP radiation due to the element-rotation technique
used to achieve the phase shifting over the array surface. Instead of a known analytical formula, a
simple technique is used for the generation of the array geometry, eliminating the phase errors for
larger elevation incidence angles. The feed’s incident radiation on the array is combined with the
response of every unit cell to obtain the scattered fields on the array, which are then used to obtain
the far fields of the full array. The calculation results of the two 900-element RA and TA antennas
show 1-dB CP-gain bandwidths of ~6% with total antenna efficiency values exceeding 60%. The
antenna prototypes will be evaluated in January 2020. The measurement results of two smaller-
scale prototypes confirm the wideband CP performance of the RA and TA elements, in spite of the
sub-optimal aperture efficiency.

Chapter 7 concludes the thesis, outlines the main results and discusses the future trends and
challenges of CubeSat antenna design.

1.3.3. Research Projects

The work performed in the scope of this thesis was mainly funded by two Swiss Innovation Agency
(Innosuisse, former CTI) projects, with Microwave and Antenna Group (MAG) as the research
partner and Astrocast as the implementation partner. The aim of the first project, titled
CubeSatCom, was the development of low-profile wideband patch antennas in S and X bands for
TT&C and potentially for data downlink. The results produced during this project are presented
in Chapter 2.

The second, more encompassing project was organized in the same manner and resulted in the
development of the M2M antennas in L band and a case study on the K-band high-gain ISL
antennas. Chapters 3 through 6 present the main results of this work. Additional work was
performed in the scope of this project to help design several L- and S-band antennas for the first
two precursor missions and the first orbital plane of CubeSats developed by Astrocast. This work
is considered less interesting from a research point of view and it is therefore not included in this
thesis.

9
1. General Introduction

The author of this thesis, together with the thesis supervisor, co-supervised several semester and
bachelor student projects listed below. The work for most projects was performed in coordination
with Astrocast and tackled several issues related to different parts of the CubeSat constellation.
The results of the student projects were not directly included in this thesis, although they often
helped the author of this thesis to get a better understanding of the problematics or served as the
starting point for future work. The students’ devoted work is greatly appreciated and it resulted in
a successful project in all cases. The project reports can be found in the corresponding references.
The student projects are:

[SP1] X-band Antenna for CubeSat Satellite – Joana Maria Llull Coll – 2016.

[SP2] L-band Antenna for CubeSats – Manuel Vonlanthen – 2017.

[SP3] Terminal Antenna for Space Applications in L band – Marina Ramos Cuevas – 2018.

[SP4] Isoflux Patch Antenna in X band – Carlos Megías Núñez – 2019.

[SP5] Patch Antenna Design for the EPFL Rocket Team – Alexandre Devienne – 2019.

For the duration of the doctoral studies, the author of the thesis was involved with several other
research projects, not directly related to the thesis topic. The results of the work are not included
here, since they are out of the scope of this thesis. Nevertheless, they served the author to get an
insight into each of the corresponding domains of antenna research. The topics include:

 SATCOM arrays – a project to design mechanically steered high-gain waveguide arrays in

Ku and Ka bands, to be fabricated using additive manufacturing techniques
 Implantable antennas – an initial study and a design of a cranial implantable antenna, to
be used as a part of a potential system for wireless extraction of brain signals [C6], [C7]
 THz sensors – design of dipole antennas and coplanar-waveguide baluns for VO2 sensors
at THz frequencies [C4]

10
2. Low-Profile Wideband CubeSat
Antennas

CubeSats require telemetry, tracking and control (TT&C) antennas to receive telecommand
signals from the ground stations and to downlink telemetry signals. As these signals are crucial for
the satellite operation, the TT&C antennas need to be reliable. In addition, the antennas need to
achieve a full spherical coverage to provide a link with ground stations regardless of the CubeSat’s
orientation, especially in the early stages following deployment.

A thorough review of existing designs and technologies is performed in the first part of this
chapter, showing a variety of different and innovative designs, for the frequency range spanning
from VHF to X band. A general trend is observed, where the antenna solutions exhibit an increase
in performance and complexity with the increase in frequency: from simple deployable
monopoles in UHF band to isoflux patch antennas in X band. The S-band frequencies seem to
show a large potential for compact and at the same time performant designs, suitable for
CubeSats. A number of solutions was proposed in the literature to tackle this trade-off for
conventional small satellites, usually lacking one or several aspects required for CubeSat
applications.

The goal of this chapter is to present a systematic design process and a detailed analysis of several
low-profile wideband antennas for CubeSats in S and X bands. An aperture-coupled patch
antenna is selected for this application, due to its wideband performance. The disadvantages of
an initial microstrip-fed antenna are addressed and the design is modified for the CubeSat
environment by introducing an asymmetric-stripline feeding network. After carefully
investigating the effects of the field distribution inside the stripline, a significant performance
improvement is obtained by a metallic enclosure of the stripline feed. The conditions and
limitations under which this improvement is possible are studied in detail. The resulting cavity
resonances can be harmful if they appear within the operating bandwidth. It is shown how the
same resonances can be used to suppress the radiation in an arbitrary adjacent frequency band.
The chapter further proposes an elegant method of integrating this type of antennas into the
2. Low-Profile Wideband CubeSat Antennas

CubeSat chassis, reducing the antenna profile on the exterior of the spacecraft. The efficient
technique for performance improvement is finally implemented on patch elements and arrays in
X band, demonstrating its potential for antennas having simple feeding networks.

2.1. Review of CubeSat TT&C Antennas

The most commonly used orbital deployment system for CubeSats is the Poly-PicoSatellite Orbital
Deployer (P-POD), initially developed for the first CubeSats at Cal Poly [25]. The system is
essentially an aluminum compartment that fits a 3U CubeSat (or multiple smaller form-factor
satellites) with a hatch and a spring mechanism. Following the hatch opening, the system deploys
nanosatellites by simply ejecting them into orbit using a loaded spring, located at the bottom of
the compartment. The satellite orientation with respect to the Earth (i.e. satellite attitude) is
undefined immediately after the ejection and ADCS is activated at this stage. The purpose of the
ADCS system is to determine the current satellite attitude and rotate the spacecraft to a favorable
orientation using a set of sensors and a, respectively [26].

The TT&C systems are the first to establish the communication with the ground stations after the
orbital deployment. It is crucial for this link to be reliable, as it supports the attitude control
system, and therefore, all subsequent satellite operations depend on the success of this stage [26].
As it is also the case with many other satellite systems, the key system component is the antenna.
The fundamental goal of any TT&C antenna, or an antenna system, is to be able to communicate
with the ground stations, regardless of the satellite attitude. This is typically achieved through one
of the two approaches:

 Omnidirectional coverage by a single low-directivity antenna, or

 Sectorial coverage by a set of multiple, more directive, antennas placed on several
spacecraft faces [8].

The choice of a suitable antenna configuration depends mostly on the TT&C operating frequency.
As a rule of thumb, the lower-frequency systems use a single antenna, whereas several antennas
are used at higher frequencies, as their reduced size allows for multiple antennas onboard a
CubeSat. The antenna designs are compared in the following sections and sorted with respect to
the operating frequency. The designs include standard geometries, as well as several exotic,
specialized solutions. It should be noted that not all presented antenna solutions perform
exclusively TT&C functions, but are mentioned here since they accurately represent their
respective antenna classes, very often seen on CubeSats. Other than the examples presented in
this thesis, there exists a large and continuously growing number of publications which introduce
innovative solutions for modern CubeSat missions.

12
 2.1. Review of CubeSat TT&C Antennas

2.1.1. VHF/UHF Bands

The frequencies of VHF and UHF bands have been the most popular choice for CubeSat
communication since the standard’s development [2]. For most early CubeSats, it was also the
only means of communication with the ground, and therefore included the telemetry and
telecommand signals. The CubeSat mission designers have traditionally selected these bands for
practical reasons: the good characteristics of the propagation channel and the availability of COTS
components operating at these frequencies. However, the free-space wavelengths at the VHF and
UHF radio bands, as defined by the IEEE [27], fall in the range from 300 mm to several meters. This
makes the design of an efficient radiator a challenging task, having in mind the 1U-CubeSat size
constraint of 100 mm.

To overcome this challenge, the deployable wire-based antenna structure is usually chosen [3].
Namely, the wire antenna is wrapped around the CubeSat or wound inside the spacecraft chassis
during the launch, conforming to the standard maximal dimensions. After the placement in orbit,
the antenna structure is deployed using a release mechanism, into its final, functional form. The
antenna geometries are typically in the form of monopole, dipole, turnstile, and helix antennas.

Among the first batch of CubeSats, launched in 2003, was the XI-IV satellite from the University of
Tokyo [28]. The CubeSat demonstrated the space use of a COTS camera and sent the data at
437 MHz (UHF). A linearly polarized dipole antenna is seen in Figure 2.1(a), as a wound wire at
the exterior of the CubeSat. To further simplify the design and reduce the cost, many early
CubeSats use an actual metallic tape measure as the antenna, due to its flexibility. One example
is the PhoneSat built by NASA Ames Research Center, shown in Figure 2.1(b) [29]. PhoneSat uses
the unmodified COTS smartphone and Arduino platforms, and communicates at 437 MHz (UHF).
The tape-measure monopole antenna is stowed inside the chassis and automatically deploys after
the CubeSat ejection. Two circularly polarized deployable antenna solutions are shown in
Figure 2.1(c) – a high-gain helix antenna and a turnstile antenna, mounted on a GOMX-3 CubeSat
from GOMSpace [14].

VHF/UHF-dual-band antennas have also been investigated. In [30], the first and third harmonics
of a monopole antenna were excited at 130 and 390 MHz to obtain saddle-shaped and
hemispherical patterns, respectively. The physical structure of a passive attitude-stabilization
system of a 3U CubeSat was used as an antenna in [20]. A single-port dual-band performance was
enabled by physical band-stop filters positioned on the metallic strips. A highly-miniaturized
planar UHF antenna is presented in [31].

2.1.2. S Band
The S-band frequencies include the Space Operation (SO) bands (2.025 – 2.11 GHz uplink,
2.22 – 2.29 GHz downlink), and the Amateur-Satellite Service band (2.4 – 2.45 GHz, the so-called

13
2. Low-Profile Wideband CubeSat Antennas

(a) (b)

(c)
Figure 2.1. Deployable VHF antennas for CubeSats. (a) Stowed dipole antenna (left) and the
interior of the 1U XI-IV CubeSat (right) [28]. (b) Monopole tape-measure antenna on the 1U
PhoneSat, with a GPS patch antenna visible [29]. (c) High-gain helix antenna (left) and a
turnstile antenna (right) on a 3U GOMX-3 CubeSat [14].

13-cm band), allocated by the International Telecommunication Union (ITU) [24]. The SO bands
are widely used for TT&C purposes and the amateur band serves for the CubeSat data downlink.

Various types of planar antennas are popular in this frequency range due to their low profile and
reliability, as they require no deployment [3], [32], [33]. An inherent property of conventional
patch antennas is their narrow bandwidth, and a majority of CubeSat patch antennas operates at
a single frequency. Having a multiband or a wideband single-port antenna is an advantage, as the
total number of antennas required for duplex communication is reduced. If a wide impedance
bandwidth is achieved, it is another challenge to obtain good polarization properties across the
same bandwidth, typically in terms of AR for CP antennas. Other common antenna types for small
satellites include helix antennas and patch-excited-cup antennas [5]. These antenna geometries
are not appropriate for CubeSats in S band without significant modifications, as their profile is
significant compared to the CubeSat body.

14
 2.1. Review of CubeSat TT&C Antennas

Due to their low profile, traditional patch antennas are inherently suitable for any CubeSat
operating in S-band. Very often, however, the conventional designs must be adapted for the
particular needs of a specific CubeSat mission. An example is the patch antenna system in [34],
where four LP patch antennas are sequentially excited to achieve CP. The antenna system is built
with a gap at its center to accommodate a camera lens, as seen in Figure 2.2(a). A dual-CP S-band
patch antenna, fed by a branchline coupler, is designed for CubeSats and presented in [35].
Capacitive coupling and a matching network based on chip inductors provide a dual-band
performance, both in impedance and AR.

A variety of other, low-profile antenna geometries are proposed for CubeSats in S band. A CP
crossed-dipole antenna for CubeSats, operating in the SO bands, is presented in [36]. This
antenna type normally requires a metallic reflector at a distance of λ0/4 to achieve a unidirectional
radiation pattern. In this case, an artificial magnetic conductor is placed close to the dipole
elements resulting in a significant profile reduction, while preserving a wideband performance.
The antenna prototype is presented in Figure 2.2(b). A miniaturized slot antenna system for
micro/nano-satellites in S band, shown in Figure 2.2(c), is introduced in [37]. The radiating slots
are highly integrated with the metallic surface and backed by a cavity inside the satellite.

2.1.3. X Band
The TT&C communication in X band is interleaved with the payload data in the Space Research
and Earth Exploration Services bands (7.145 – 7.235 GHz uplink, 8.025 – 8.5 GHz downlink), as
allocated by the European Cooperation for Space Standardization (ECSS) [38]. The increase in
frequency simultaneously allows a larger data rate and a greater freedom in antenna design, as
the wavelengths in this band are in the order of 40 mm. Two examples that demonstrate how this
freedom can be used are isoflux patch antennas shown in Figure 2.3 [16].

(a) (b) (c)

Figure 2.2. S-band CubeSat antennas. (a) Patch antenna array around a central camera
lens [34]. (b) Crossed-dipole antenna over an artificial magnetic conductor [36].
(c) Miniaturized slot antenna system [37].

15
2. Low-Profile Wideband CubeSat Antennas

(a) (b)
Figure 2.3. X-band isoflux CubeSat antennas. (a) Patch antenna with 12 parasitic dipoles over a
corrugated ground plane [16]. (b) A miniaturized helix antenna [39].

Multiple crossed-dipole elements and deep concentric corrugations yield low AR values and an
isoflux pattern suitable for LEO satellites. Other designs include the miniaturized helix antenna
over a corrugated ground plane [39], various 2 x 2 sequentially-fed patch-antenna arrays [40], [41],
and other [42].

2.2. Antenna Design Requirements

The following sections describe the design procedure for a low-profile wideband antenna in S
band for 3U CubeSat applications. The initial design requirements are summarized in Table 2.1.
The specified frequencies cover the SO and amateur bands, previously mentioned in
Section 2.1.2. The designed antenna should simultaneously provide TT&C links at the SO
frequencies, and data downlink in the amateur band.

A sketch of a 3U CubeSat is shown in Figure 2.4, where the space allocated for the antenna is
indicated in gray. Throughout this thesis, the CubeSat body was modeled as a PEC box for
simplicity. The antenna surface of 100 x 100 mm2 corresponds to a 1U-CubeSat face, making it
also suitable for CubeSats with a smaller form factor, such as 1U or 2U. Considering the low profile
requirement of 0.08 λ0 at the lowest frequency, it is a challenging task to provide AR values smaller
than 3 dB over a 20 %.

It can be argued that the profile requirement does not comply with the CubeSat standard, which
defines the maximum feature thickness on the satellite surface to be 6.5 mm [1]. Considering the
large required bandwidth in this specific scenario, the implementation partner of this project was
prepared to back off the part of the chassis under the antenna until the standard is satisfied.
Furthermore, a novel method of antenna integration was described in detail in Section 2.6.1,
which reduces the antenna profile on the satellite surface.

16
 2.3. Microstrip-Fed Aperture-Coupled Patch Antenna

Table 2.1. S-band CubeSat antenna design specifications.

Property Value
Frequency [GHz] 2 – 2.45 (~20 %)
Gain [dBi] >5
Beamwidth [°] > 50
S11 [dB] < -10
AR @ BW [dB] <3
Dimensions [mm3] 100 x 100 x 12

(a) (b)
Figure 2.4. CubeSat antenna specifications. (a) Volume allocated for the S-band antenna on a
3U CubeSat. (b) Frequency requirements.

2.3. Microstrip-Fed Aperture-Coupled Patch Antenna

The low profile requirement naturally led to the choice of a patch antenna for this application.
The aperture-coupling feeding method was selected, due to the superior bandwidth performance
of that geometry, compared to other methods of patch-antenna excitation [43], [29], [30]. A
microstrip-fed aperture-coupled patch antenna was initially designed in order to test the
feasibility of covering the required bandwidth [C1]. A conventional LP microstrip-fed element was
first tuned for the desired frequency band. A square metallic patch was suspended above the thin
rectangular slot, under which a microstrip feeding line was placed perpendicularly. The circular
polarization is then achieved using a crossed slot and a feeding network based on T-junction
power dividers, quarter-wavelength transformers and phase-delay lines [C1]. The final antenna
geometry is shown in Figure 2.6(a). The design was carefully analyzed in several intermediate
models, presented in Figure 2.5, gradually increasing the feeding network complexity.

The thickness of a microstrip patch antenna and the dielectric permittivity of the substrate are key
factors that influence the antenna bandwidth. A low-permittivity Rohacell foam (εr=1.08,
tan δ=0.0002) is used as the patch support, as a means of maximizing the bandwidth. For support
thicknesses roughly 0.1 λ0 and larger, it is not difficult to obtain a wideband performance of this

17
2. Low-Profile Wideband CubeSat Antennas

(a) (b) (c)

Figure 2.5. Evolution of the microstrip-fed aperture-coupled patch antenna. (a) – (b) LP models.
(c) 4-port CP model.

(a) (b)
Figure 2.6. Microstrip-fed CP patch antenna. (a) Final antenna geometry. (b) Antenna
prototype. Antenna dimensions (in mm): Lg=100, Lp=48, Ls=35, Ws=2, Llam4=23.64, Wlam4=2.44,
Wf=1.5, stub=9.5, hFR4=0.2, hfoam=10, hsub=0.508.

geometry. The support was then made as thin as possible, while preserving the wideband
matching. The radiating patch was printed on a thin layer of FR4 (εr=4.5, tan δ=0.02) and Rogers
RT Duroid 5870 (εr=2.33, tan δ=0.0012) was used for the feeding network and the ground-plane
slots.

It is important to distinguish the return loss of the radiating element from the return loss of the
entire antenna, including the feeding network. An example is a 4-element patch antenna array
from [34] and [10], where the same radiating elements were used with two different feeding
networks. This resulted in drastically different S11 parameters, seen from the antenna port, while

18
 2.4. Stripline-Fed Aperture-Coupled Patch Antenna

the element return loss remained unchanged. The matter should be investigated properly to avoid
attributing some aspects of the full-antenna S11 response to the radiating element.

Therefore, it was necessary to evaluate the return loss of the radiating patch itself for the given CP
excitation scheme. An intermediate 4-port model, seen in Figure 2.5(c), was created for that
purpose. The port excitation mimics the feeding network performance, and the active S-
parameters are calculated using:

, 1. .4 (2.1)

where Smn are the conventional S-parameters, and an is the incident power wave at port n. The
active-S parameters give the power reflected at each port, with all ports excited in a desired way.
Only active-S1 parameter is presented in the simulated results, since the remaining ones show a
similar trend.

The simulated and measured performance of different antenna models from Figure 2.5 and
Figure 2.6 is presented in Figure 2.7. The full antenna satisfies the impedance matching
requirement across the entire specified bandwidth. The low S11 values at the antenna port are a
consequence of multiple reflections inside the feeding network. The effect is illustrated in
Figure 2.8. Due to the specific feeding-network topology, the signals reflected at the patch arrive
with a relative phase delay of 180° at the T-junction. This corresponds to an odd mode at the
output ports of the divider and it is not supported due to symmetricity. Therefore, the signals suffer
full reflection and ultimately arrive to the antenna input with an undesired phase, degrading the
cross-polarization. This effect is mostly pronounced at frequencies of poor antenna matching,
which is evident from the similarity of the active-S1 and AR curves. The AR deterioration is not
seen in other similar publications, since the available thickness in those cases is much larger,
improving the element matching [C1]. In our scenario, the poor antenna matching at the band
edges stems from the limited patch-support thickness.

The overall results show that this type of antenna geometry is an eligible candidate for CubeSat
wideband applications. However, the design must be further modified to address its drawbacks
and adjust it for our application, primarily improving the polarization characteristics over a wide
bandwidth.

2.4. Stripline-Fed Aperture-Coupled Patch Antenna

Beside the AR, another disadvantage of the microstrip-fed antenna described in the previous
section is a poor front-to-back ratio (FTBR) seen in Figure 2.7(d). The back lobe is caused by the
spurious back-radiation of the slot, and it prevents a simple antenna placement onto a CubeSat
surface. Furthermore, the microstrip-fed antenna is very sensitive to a proximity of a metallic
surface below the feeding network [C1].

19
2. Low-Profile Wideband CubeSat Antennas

(a) (b)

(c) (d)
Figure 2.7. Simulation and measurement results of the microstrip-fed patch antenna [C1]. (a) S
parameters corresponding to structures from figures 2.5 and 2.6. (b) Broadside gain and AR.
(c) Radiation patterns.

Figure 2.8. Incident and reflected signal paths and their relative phase values.

20
 2.4. Stripline-Fed Aperture-Coupled Patch Antenna

Starting from the initial geometry, the feeding network was realized in stripline technology to
ensure a unidirectional pattern and isolate the feeding network from the satellite structure. The
spurious back-radiated power in the microstrip-fed model manifests here as a strong electric field
in the stripline. This effect greatly deteriorates the antenna performance, due to the reduction of
the power accepted and radiated by the patch, predominantly at the edges of the band. Under the
thickness constraints, a combination of methods is required to achieve a sufficiently high coupling
from the feed lines to the radiating element, which will provide acceptable return losses. In
scenarios where a larger antenna thickness is available, only a subset of these modifications would
be sufficient for a satisfactory performance [C1].

2.4.1. Enclosed-Stripline Feeding Network

As a first step, an asymmetric-stripline feeding network is designed. In this structure, the
electromagnetic fields are concentrated in the thinner, higher-permittivity dielectric, increasing
the power delivered to the patch [46]. A second radiating patch is added to improve the coupling
at the higher frequencies, without increasing the total antenna thickness.

Mode suppression techniques [47]–[49], e.g. shorting pins, were applied and studied in a large
number of cases to understand their influence the antenna operation. The electric field is
calculated inside the stripline of a 4-port antenna model, for different pin configurations.
Considering the strong electric field in the stripline, none of the configurations was beneficial for
the antenna operation in the required band. As seen, for instance, in Figure 2.9(a), where the pins
are placed close to the coupling slots, the electric field values drop rapidly towards the slot edges.
This non-uniform illumination of the slots reduces the coupling to the patches and consequently
yields poor active-S-parameter values in the band of interest. In general, any shorting
configuration that perturbs the electric field in the stripline, directly below the slot, also
deteriorates the S-parameters. This effect is evident in Figure 2.9(d). The stripline is shorted at the
corners, leading to a characteristic field pattern with a minimum at the center – below the slot.

Instead of suppressing the electric field in the feeding stripline, it is possible to utilize an
advantageous field distribution to boost the coupling. It was observed that the electric field
distribution directly below the slot determines the amount of coupling to the patch. The stripline
structure is then enclosed with metallic walls on all sides. The electric field inside a fully shielded
stripline, as shown in Figure 2.9(e), is favorable, since the large field strength under the slot boosts
the coupling to the patch above. This was a key modification that significantly improved the
wideband antenna performance.

2.4.2. Antenna Geometry and Performance

The antenna geometry is shown in Figure 2.10(a). The feeding network is modified by replacing
the simple T-junction dividers and quarter-wavelength transformers with Wilkinson power

21
2. Low-Profile Wideband CubeSat Antennas

dividers. Unlike the T-junction dividers, Wilkinson dividers possess a high isolation of the output
ports, which prevents the signals reflected from the patch to travel to the opposite slot and pollute
the cross-polarization.

A prototype of the aperture-coupled stripline-fed stacked-patch antenna is shown in Figure 2.11.

The asymmetric stripline is composed of a Rogers RT/Duroid 5870 (εr=2.33, tanδ=0.0012,
h=0.508 mm), on which the lines and coupling slots are etched, and a low-permittivity Rohacell
foam (εr=1.08, tanδ=0.0002, h=2 mm). The same foam material is used as the patch support,
similarly as in the microstrip-fed antenna. The radiating patches are printed on a thin layer of FR4
(εr=4.4, tanδ=0.02, h=0.1 mm). In the first approximation, the metallic wall surrounding the
stripline is modelled as a metallic pin wall, with pin separation of 15.6 mm (0.11 λg, where λg is
the wavelength inside the stripline at 2 GHz). As the metallization is not feasible on the foam
material, shorting pins are soldered manually to the two ground planes. After the first S11
measurements were done, the stripline with pins was additionally enclosed with a copper tape on
all sides. The individual antenna layers are attached with nylon screws.
Y [mm]

(a) (b) (c)

(d) (e) (f)

Figure 2.9. The complex magnitude of the electric field inside the feeding stripline for different
pin and shielding configurations at 2.225 GHz, and the corresponding active-S1 parameters.
(a) Densely placed shorting pins, spaced by 12.5 mm. (b) Shorting pins placed around the
coupling slots, spaced by 8 mm. (c) Open stripline. (d) Stripline plates shorted on the corners.
(e) Fully shielded stripline. (f) Corresponding active-S1 parameters vs. frequency. The white
plusses and the hatched regions mark the pin and shielding locations, respectively.

22
 2.4. Stripline-Fed Aperture-Coupled Patch Antenna

Figure 2.10. Stripline-fed stacked-patch antenna geometry. Antenna dimensions (in mm):
Lg=100, Ltop=49, Lbot=45.6, Ls=34.7, Ws=2, LWilk=24.75, WWilk=0.75, Wf=1.49, stub=11.8, dpin=15.6,
hFR4=0.1, htop=5, hbot=4, hsub1=0.508, hsub2=2.

The simulation and measurement results of the antenna are shown in Figure 2.12. Simulated
radiation efficiency of the antenna is from 76% to 90.5% in the frequency band 2 – 2.45 GHz. The
highest efficiency values correspond to 2.1 and 2.4 GHz, which is coherent with the antenna
active-S1 parameter. The antenna back lobe is effectively removed, as the measured FTBR is
higher than 30 dB. It would not be adequate to specify a total impedance bandwidth of the
antenna, as the measured S11 is mainly produced by the feeding network. Instead, it can be stated
that the antenna gain is larger than 5.9 dBi (3 dB less than the maximum gain) over a 27%
bandwidth, and the AR bandwidth of the antenna (AR<3 dB) is 32%.

The simulated performance of the feeding network itself is shown in Figure 2.13. By comparing
the return loss at the input port (S55) with the return loss in Figure 2.12(a), it can be concluded that
the antenna input matching is mostly determined by the feeding network. The phase difference
between signals incident on any two adjacent ports is in the range from 82° to 99° along the whole
band.

23
2. Low-Profile Wideband CubeSat Antennas

(a) (b)

(c) (d)
Figure 2.11. Stripline-fed stacked-patch antenna prototype. (a) Individual layers – the numbers
denote the order of assembly. (b) A detail of the SMA connector before enclosing the stripline
with a copper tape. (c) Antenna prototype (back view) with the soldered shorting pins and
copper tape visible. (d) Antenna prototype (front view) inside the anechoic chamber.

2.5. Stripline-Cavity Resonance Analysis

A rectangular cavity is formed by enclosing the stripline feeding structure, which exhibits cavity
resonances at certain frequencies. The resonances have a narrowband character, but are harmful
for the antenna operation and it is crucial that they do not appear within the operating
bandwidth [J1]. To illustrate this effect, the instantaneous electric field is simulated in the
enclosed-stripline cavity, at the resonance frequency and away from it, and shown in Figure 2.15.
Away from the resonance, the electric-field vector has an opposite direction at the opposite sides
of the slots. This difference creates an electric field along the slot itself, which strengthens the
coupling. The particular field distribution shown in the figure is a consequence of the feeding
required for a CP radiation. On the contrary, the electric-field vector at the resonant frequency has
a uniform direction across the entire cavity.

24
 2.5. Stripline-Cavity Resonance Analysis

(a) (b)

(c)
Figure 2.12. Simulation and measurement results of the stripline-fed stacked-patch antenna.
(a) S parameters. (b) Broadside gain and AR versus frequency. (c) Radiation patterns.

(a) (b)
Figure 2.13. Simulated S parameters of the feeding network. (a) Magnitude. (b) Phase. Inset:
Port numbering of the feeding-network.

25
2. Low-Profile Wideband CubeSat Antennas

The presented antenna has cavity resonances seen at the measured S11 curves of Figure 2.12(a).
The model with the pins has a lower resonance (1.84 GHz) compared to the model with both pins
and the copper tape (1.94 GHz), as the fringing fields increase the effective cavity size in the former
case.

A good practice is to allow a safety margin between the operating band and the cavity resonance.
If the cavity size is preserved, a higher permittivity substrate can be used for the feeding network.
This will increase the electrical size of the cavity, further reducing the resonant frequencies. These
resonances are not always visible in the simulation results due to the selection of the interpolating
frequency-sweep technique used in the EM software.

The electric field in the stripline cavity preserves the same distribution at frequencies above and
below the first resonance. This feature enables an equally good antenna operation on either side
of the resonant frequency. Therefore, the cavity size can be reduced, which will shift the resonant
frequency above the operating band. However, the cavity dimensions can be changed to a certain
extent. Figure 2.14(c) shows the active-S parameters for different cavity sizes. A cavity too large
will have a field distribution with low values at the slot location, as seen in Figure 2.14(a).

The consequence is a poor antenna matching at the higher edge of the frequency band, where this
effect initially appears. On the other side, the proximity of the walls to the coupling slot limits the
minimal cavity size. The electric field magnitude at the locations of slot edges is reduced in that
case, as seen in Figure 2.14(b). This effect is dominantly present at lower frequencies, where the
coupling to the patch is consequently deteriorated.

2.5.1. Interference Suppression in an Adjacent Frequency Band

A cavity size of 80 mm exhibits a resonance at 2.42 GHz, which can be seen Figure 2.14(c). At the
resonant frequency, the electric field distribution in the feeding stripline is dominated by the
cavity resonance. The electric field vector has the same direction at all locations in the stripline,
as seen in Figure 2.15. This electric field polarity is analogous to the non-radiating cavity
resonances in cavity-backed slot antennas, which is thoroughly studied in [50]. In our case, since
the slots are used for coupling to the patches, the field polarity at the resonance inhibits this
coupling. The energy is partly dissipated through dielectric and conductive losses in the cavity,
and partly reflected in the feeding network and dissipated in the resistors. All the combined effects
lead to a reduction of the antenna gain at the cavity resonance.

If the cavity size is properly selected, the resonance can be placed at an arbitrary adjacent
frequency, leading to a reduced amount of received interference from that frequency band. The
effect of gain reduction is narrowband and does not affect the antenna performance in the
operating band. This feature is demonstrated in the design presented in the next section.

26
 2.5. Stripline-Cavity Resonance Analysis

65 30

30 15
Y [mm]

0 0

-30 -15

-65 -30
-65 -30 0 30 65 -30 -15 0 15 30
X [mm] X [mm]
(a) (b)

(c)
Figure 2.14. The magnitude of the complex electric field vector in the shielded stripline at
2.225 GHz, for the cavity size of (a) 130 mm and (b) 60 mm, and (c) the corresponding active-S1
parameters. Slot locations are marked with solid white lines.
Y [mm]

Y [mm]

(a) (b)
Figure 2.15. Instantaneous electric field magnitude inside the stripline. (a) 2.225 GHz (center
frequency). (b) 1.94 GHz (cavity resonance). Dominant electric-field vector directions are
marked in black.

27
2. Low-Profile Wideband CubeSat Antennas

Both the field distribution and the resonances of the cavity must be taken into account when
selecting the cavity dimension. The range of viable dimensions depends mainly on the operating
bandwidth of the antenna – the larger the bandwidth, the less frequencies are available for the
suppressed band. This range can be modified by selecting a different combination of stripline
dielectrics. However, it is recommended that the ratio of their dielectric permittivities (2.15) and
the ratio of thicknesses (4) remain the same or larger. Otherwise, it will be difficult to achieve a
sufficient coupling from the feed lines to the patch. If both criteria are fulfilled, the antenna will
benefit from the field-distribution effects, without suffering from adverse resonances.

2.6. TT&C Antenna

The stripline-fed wideband antenna was modified to work as a TT&C antenna, at the ECSS
frequencies in S-band [J1]. The frequency band of interest for this design is from 2 to 2.3 GHz,
while the other design requirements remain unchanged. A small stripline cavity is selected and
the resonance frequency is shifted above the operating band. The total antenna dimensions are
then also reduced, which allows a greater flexibility of the antenna placement on the CubeSat. It
was seen that a satisfactory performance could be achieved with one radiating patch only. In
addition, the total thickness of the antenna is slightly reduced. The thickness can be reduced
further if necessary, while accepting a lower radiation efficiency, and thus, a lower antenna gain
at extreme frequencies.

The selected design has a cavity dimension of 78 mm, yielding a resonant frequency of
approximately 2.45 GHz. That frequency corresponds to the amateur-satellite service band, which
is very popular for CubeSat communications. If the antenna were to operate at the ECSS
frequencies only, the signals in the amateur band would present a significant interference and the
suppression of the received signal from that band would be an advantage.

The following sections present the antenna geometry, prototype construction, as well as
simulation and measurement results. Section 2.6.1 proposes an elegant method of integrating this
type of antennas into the CubeSat metallic chassis, which reduces the antenna extrusion on the
satellite surface.

2.6.1. CubeSat Integration

As it was discussed in Section 2.2, the antenna would significantly violate the total dimension
requirements, if it were to be simply placed on the CubeSat face. A very practical approach is to
integrate the antenna into the CubeSat by placing the feeding stripline inside the satellite chassis,
while keeping the patches on the exterior. The concept is illustrated in Figure 2.16. As a first
approximation, it is assumed that the CubeSat chassis is a hollow metallic box with a thickness of
1.2 mm. In this configuration, the satellite face on which the antenna is mounted acts as a ground

28
 2.6. TT&C Antenna

plane with the coupling slots, and the adjacent faces are used as vertical cavity walls. The second
stripline ground completely isolates the feeding network from the satellite interior. As the
coupling slots are cut in the metallic face of the satellite, they now have a substantial thickness
(1.2 mm) compared to the PCB metallization in the previous prototype (17 μm). The influence of
the increased slot thickness can be partially neutralized by enlarging the slot width, or modifying
the slot shape. If the desired cavity size does not match the dimensions of the satellite face, such
as the TT&C antenna presented here, the antenna structure can be shifted towards one corner of
the CubeSat and the remaining cavity walls can be modeled as pin walls. This halves the required
number of pins, compared to the case where the antenna is positioned on the center of the face.
The cavity’s resonant frequency is then a function of pin-cage dimensions and can be altered,
while keeping in mind the potential problems outlined in Section 2.5.1. Furthermore, the
remaining part of the face remains available for other features on the satellite’s surface, or antenna
filters, diplexer, etc.

2.6.2. Antenna Prototype Construction

The TT&C antenna prototype was built using the same materials as in the previous model, and
integrated in a 3U CubeSat mock-up during the S11 and radiation measurements. The geometry of
the TT&C antenna, mounted on a CubeSat, is shown in Figure 2.17. The antenna was not
significantly modified compared to the previous design, safe for the fact that the power dividers
are tuned to the frequency of 2.15 GHz. A low-cost CubeSat mock-up was built using 1.2-mm-
thick metallized panels of FR4 epoxy. The panel edges are ridged and interlocked for a better
structural stability. The crossed-slot was cut in one of the long faces, where the antenna was to be
mounted. The individual antenna layers, before assembly, are shown in Figure 2.18(a). After the
structure was assembled, the edges and the slot were covered with a copper tape to electrically
connect the individual panels and provide conductivity on the slot’s vertical walls. The feeding

Figure 2.16. Aperture-coupled patch antenna on the CubeSat surface (top) and the same
antenna integrated into the CubeSat chassis (bottom).

29
2. Low-Profile Wideband CubeSat Antennas

network was placed inside the CubeSat, against one of the corners. It was then attached to the
structure with ten metallic screws that formed the remaining two vertical walls of the stripline
cavity. The screws connect the two ground planes electrically, but also provide a mechanically
rigid structure. The foam and the patch were placed above the slot and fixed to the structure with
nylon screws.

The CubeSat mock-up was designed in such a way that the antenna can be placed on both the
large and the small CubeSat face. This is simply done by exchanging the two panels seen on the
left side of Figure 2.18(c), marked with letters ‘L’ and ‘S’. Figure 2.18(d) and (e) show the assembly
of the entire structure, with the antenna mounted on a large and small CubeSat face, respectively.
The SMA connector is placed vertically in both cases, at the inside of the CubeSat. In this
configuration, the measured return losses of this coaxial-to-stripline transition are smaller than -
22 dB in the frequency range of interest.

Figure 2.17. TT&C patch antenna integrated into a 1U CubeSat chassis. Antenna dimensions (in
mm): Lc=78, Lfoam=75, Lp=50, Ls=39, Ws=3, dpin=13.2, hfoam=8, tCS=1.2. The remaining parameters
have the same values as in Figure 2.10.

30
 2.6. TT&C Antenna

(a) (b)

(c) (d)

(e) (f)
Figure 2.18. TT&C antenna prototype. (a) Individual antenna layers before assembly, bottom
view. Numbers indicate the stacking order. (b) Close-up view of the coupling slot covered with
conductive tape. (c) Panels of the CubeSat mock-up before assembly. Panels marked with ‘top’
and ‘side’ were interchanged during measurements. (d) Assembled patch antenna (back view).
(e) Antenna mounted on a large CubeSat face. (f) Antenna mounted on a small CubeSat face.

31
2. Low-Profile Wideband CubeSat Antennas

2.6.3. Antenna Performance

The first set of measurements was performed with the antenna mounted on a large CubeSat face.
The antenna was afterwards moved to the small face, and all the measurements were repeated.
Figure 2.19 compares the simulation and measurement results of the antenna matching and
radiation characteristics. The beam tilt of 5° was observed for the antenna on a large CubeSat face,
caused by the ground-plane size increase. The small beam tilt does not pose a problem, since the
main lobe is a smooth function of the radiation angle. There was not a significant change in the
measurement results when the antenna was moved to the small CubeSat face. Therefore, the
presented results correspond to the antenna mounted on the large CubeSat face only.

There is a larger discrepancy between the measured and simulated responses, with respect to the
wideband antenna prototype. Most notably, the measured results show a shift of the cavity mode
towards the higher frequencies, as it appears at 2.55 GHz. The resonance effect also exhibits a
more-wideband character compared to the simulations. Several aspects of the low-cost mock-up
construction can be used to explain this difference: the copper tape used to provide the slot
conductivity, feed-slot-patch misalignment, material thickness tolerance, etc. Despite this, the
radiation patterns are in agreement with the simulations and the gain reduction at the cavity
resonance frequency is ~4 dBi, as seen in Figure 2.19. This reduction is not larger since the main
driving mechanism of the antenna is not the cavity mode, but the stripline feeding network.

The measured S11 parameter, shown in Figure 2.19, is below -15 dB at both ECSS bands. The
antenna beamwidth is 74° and 67° at 2 and 2.3 GHz, respectively, and the AR values are below 3 dB
across the entire radiation beamwidth, as seen in. If larger gain values are required at the edges of
the band, the single patch can be replaced with a stacked-patch structure, similar to the one in the
wideband antenna model.

2.7. X-band Antenna

In previous sections, it was shown how an enclosed-stripline feeding network can improve the
element matching and the overall performance of limited-profile aperture-coupled patch
antennas. This was demonstrated on several designs in S band, with rather complex feeding
networks. The purpose of this section is to show how this new technique can be used to boost the
performance of other antenna types, such as aperture-coupled antenna geometries with much
simpler feeding methods. A patch antenna for data downlink and telemetry in X band, described
in this section, represents this principle. The frequencies allocated for the purpose are the Earth
Exploration-Satellite service and Space Research frequency bands (8.025 – 8.5 GHz) [38].

32
 2.7. X-band Antenna

(a) (b)

(c) (d)
Figure 2.19. Simulation and measurement results for the TT&C antenna mounted on a long face
of a 3U CubeSat mock-up. (a) S11 parameter vs. frequency, for φ=0°. Inset: antenna prototype
with visible metallic screws that form the cavity. (b) Broadside gain and AR vs. frequency.
(c) Radiation patterns in two orthogonal planes, at 2 and 2.3 GHz. (d) AR vs. angle, at 2 and 2.3
GHz.

2.7.1. Antenna Geometry

Since the early development of circularly polarized microstrip patch antennas, it was evident that
the bandwidth strongly depends on the antenna thickness and substrate permittivity, but also the
feeding technique. There was an effort over time to improve their inherently narrow impedance
and AR bandwidths, without compromising the simplicity of the feeding method. One proposed
solution was a single-feed CP patch antenna, coupled to a microstrip feeding line through a
symmetrical crossed slot in the ground plane [51]. The bandwidth was further optimized by
adding a second patch and introducing asymmetries in the coupling slot and the first patch [52]–
[55].

This geometry was a starting point for our design, which is shown in Figure 2.20. The antenna is
fed with a 50-ohm line realized in asymmetric stripline technology. The feeding structure and
materials are identical to ones used in the S-band antenna. The patch has a square shape, and the
unequal length of the orthogonal coupling slots introduces the 90° phase shift, required for CP

33
2. Low-Profile Wideband CubeSat Antennas

Figure 2.20. X-band aperture-coupled patch antenna geometry.

radiation. As the wavelength in the desired band is ~36 mm, the limitations for the antenna profile
are less stringent, and the total thickness of the structure is less than 0.15 λ0 at 8 GHz.

The initial antenna designs were performed using an open (unshielded) asymmetrical stripline in
the feeding part. A prototype of the design was built and evaluated. The measured impedance
bandwidth was larger than 20%, but the cross-polarization levels were higher than specified. A
detailed study of the first prototype, including the simulation and measurement results, can be
found in [56].

2.7.2. Simulation Results

After the successful performance improvement in S-band antennas by the stripline enclosure, the
same technique was applied here. A PEC wall was added around the stripline in the simulation
model, and the antenna was slightly retuned. To get a better understanding of how this
modification influences the antenna operation, three designs were compared:

 Microstrip feeding
 Open stripline feeding
 Enclosed stripline feeding

Apart from the feeding-network technology, all the design parameters remain identical in the
three models. Their performance is compared in Figure 2.21. Microstrip model is matched at the
entire frequency band, exhibits low AR values, but has a back lobe due to the radiation of the open
slot. Adding a second ground plane in the open-stripline model improves the FTBR, but
deteriorates the AR. An enclosed-stripline model takes the best from both worlds, having at the
same time a similar polarization to the microstrip model and a unidirectional radiation pattern.

34
 2.7. X-band Antenna

S 11 parameter [dB]

[dB]
(a) (b)
=0° =0° =0°

-10 dB -10 dB -10 dB

-20 dB -20 dB -20 dB

-30 dB -30 dB -30 dB

270° 90° 270° 90° 270° 90°

Open stripline Enclosed stripline

Microstrip

180° 180° 180°

(c)
Figure 2.21. X-band patch antenna – the performance of its three variations. (a) S11 parameter.
(b) Broadside gain and axial ratio vs. frequency. (c) Radiation patterns.

A cavity resonance is evident in the enclosed-stripline model at 7.2 GHz, indicating that the
antenna operates above the first resonance. Following the analysis in Section 2.5, the cavity
dimension must be carefully selected in order to have a beneficial field distribution in the
stripline, across the entire operating bandwidth. To give an example of an improperly sized cavity,
the stripline dimension was increased and the field was calculated inside the stripline in the
original and larger models. The plots in Figure 2.22 show the field distribution for the two cases.
For the correct cavity size, electric field is focused below the slot. The field distribution of the larger
cavity gradually changes over frequency, until the second higher-order resonant mode at 8.8 GHz.
The coupling to the radiating patch is consequently inhibited and a full reflection occurs at the
resonant frequency. The first resonant mode of the larger cavity is lower than the displayed range
of S parameters. It should also be noted that the plotted fields are normalized for clarity and that
the maximum field strength at the resonant frequency is an order of magnitude higher.

35
2. Low-Profile Wideband CubeSat Antennas

Lg = 27 mm, f = 8.5 GHz

0

10 Resonances
-5

S11 parameter [dB]

5
-10
Y [mm]

0
-15
-5

-20 L = 27 mm
g
-10 L = 35 mm
g
-25
-10 -5 0 5 10 7 7.5 8 8.5 9 9.5
X [mm] Frequency [GHz]

Lg = 35 mm, f = 8.5 GHz Lg = 35 mm, f = 8.8 GHz

15 15

10 10

5 5
Y [mm]

Y [mm]

0 0

-5 -5

-10 -10

-15 -15

-15 -10 -5 0 5 10 15 -15 -10 -5 0 5 10 15

X [mm] X [mm]

Figure 2.22. (a,c,d) Electric field distribution inside the feeding stripline of the X-band antenna
and (b) the corresponding S parameters.

2.7.3. 4-Element Antenna Array

The small antenna dimension allows for an array of X-band patch antennas on a CubeSat, and the
designs in the literature usually come in the form of 2 × 2 arrays. Although the radiation
beamwidth is reduced in an array, and consequently the duration of the link with the ground
station, shorter high-data-rate bursts are possible due to an increased antenna gain.

A 4-element array was constructed using the previously described element. The sequential
rotation technique was used as a means of improving the AR [57]. Various types of corporate and
serial feeding networks were designed for the four-element array. Finally, a serial FN with a
minimal line length is presented in this chapter. The characteristic impedance of the feeding
sections were modified to obtain matching of the entire structure [58], [59]. As there is no unique
solution to this problem, the authors of [58] used a global optimization algorithm to solve for
characteristic impedances. The result of this optimization process was directly implemented here
and the line dimension were tuned for our specific stripline technology. The geometry of the full
array is shown in Figure 2.23. The four stripline sections are isolated with vertical PEC walls,

36
 2.8. Summary and Discussion

except for a small opening for feeding lines. Similarly as in[60], the openings are small enough not
to perturb the field distribution below the slots, and large enough to allow the TEM mode of the
stripline to pass through. Instead of PEC walls, the stripline sections/cavities can be built with
columns of shorting pins in an antenna prototype, as in the S-band prototype.

The simulation results in Figure 2.24 show that the AR bandwidth of the array easily covers the
required frequencies. The results imply that a smaller patch element, with a higher permittivity
dielectric, can be used inside such an array, with the performance still satisfying the design
specifications.

2.8. Summary and Discussion

Several wideband low-profile antenna designs in S and X bands are presented in this chapter. The
antennas perform the TT&C as well as data downlink functions, in the case of the first S-band
antenna. A TT&C antenna and the system as a whole are identified as the crucial satellite
component, upon which every subsequent operation depends. For this reason, no deployable
structure is used for the antenna geometry, with the high reliability in mind. Considering the
electrical size of a CubeSat at S-band frequencies, it can be seen that the available space on the
spacecraft surface is strictly limited. Having a single-port multi-band antenna efficiently
optimizes the usage of the sparse real estate onboard, simplifies the antenna placement, and
eliminates the cross coupling between several antenna elements.

Figure 2.23. X-band series-fed aperture-coupled patch antenna array. The line sections have a
90° phase delay and the characteristic impedance values are taken from [58].

37
2. Low-Profile Wideband CubeSat Antennas
S 11 parameter [dB]

[dB]
(a) (b)

(c)
Figure 2.24. X-band antenna array performance. (a) S11 parameter. (b) Broadside gain and axial
ratio vs. frequency. (c) Radiation pattern.

Therefore, the S-band antennas are designed to cover several (up to three) frequency bands
simultaneously. The CubeSat antennas in S band are traditionally simple, robust, and
narrowband, in line with the design complexity of early systems. As new technologies become
available regularly, and CubeSat capabilities improve, there will be an increased demand for high-
performance wideband antennas.

In the antenna design process, special attention is paid to distinguish the S11 parameter of the
antenna element and that of the entire antenna, including the feeding network. The difference is
otherwise well known in the antenna community, but often not clearly explained in publications
on CubeSat antennas. Therefore, it is emphasized in this chapter.

The aperture-coupled patch-antenna geometry maximizes the bandwidth in the allocated

volume for the antenna. Beside the bandwidth, another key requirement is the unidirectional
radiation pattern, which allows a seamless antenna placement on the CubeSat structure. The
initial feeding network was modified from microstrip to stripline, removing the back radiation,
but leading to a set of problems related to the field distribution inside the feeding stripline. After

38
 2.8. Summary and Discussion

a series of attempts to suppress the electric field inside the stripline with shorting pins, it was
observed that the magnitude of the electric field below the coupling slot is the driving mechanism
of the antenna. This magnitude is then maximized, by shielding the stripline on all sides, resulting
in wideband impedance, gain and AR performance of the antenna.

The stripline shielding forms a resonant cavity. At the cavity resonance, unlike at other
frequencies, the field distribution inhibits the coupling to the patch and traps the energy inside
the stripline. By properly sizing the cavity, the resonance can be moved outside the operating
bandwidth. The antenna preserves its performance for a large range of cavity dimensions, given
that the resonance is sufficiently far from the working frequencies. A set of rules is presented for a
correct selection of the stripline cavity dimensions that yield a good performance. The harmful
resonance effects can further be used to reduce the radiation at the resonant frequency, thus
suppressing the interference in an arbitrary adjacent band.

A novel method of integration into the CubeSat is proposed for this type of antennas, where the
feeding network is placed at the CubeSat interior, and the patches at the surface are excited
through coupling slots, cut in the metallic chassis of a CubeSat. This approach significantly
reduces the total antenna thickness on the surface of the CubeSat. Although the profile of the
antenna structure is still larger than what the CubeSat standard specifies, it is deemed necessary
for the given set of requirements. The problem can be alleviated by a recess of the CubeSat face
on which the antenna is mounted.

The stripline-shielding method is applied in the case of the S-band antenna with a rather complex
feeding network. In the last section of the chapter, the method is demonstrated in the case of an
X-band patch antenna and, having a single feed line and an asymmetric coupling slot. The
purpose of this section is to show the potential of the described principle for stripline feeds with
simpler geometries. The same effects are observed in the obtained results as in the previous
designs, confirming the presented analysis. A four-element array of sequentially rotated patch
antennas is designed using the described patch-antenna element. The performance of the patch
array makes it suitable for wideband high-speed CubeSat communication.

In the cases of both single elements and antenna arrays, a foam layer in the feeding network
complicates the fabrication procedure, since no metallization is possible on the foam material.
Instead, this layer can be replaced by a low-permittivity substrate, such as Rogers Duroid 5870.
Following the rules outlined in Section 2.5.1, the second substrate of the feeding network must
then be also replaced by a higher-permittivity substrate, for efficient focusing of the electric field
close to the coupling slots. Such modifications require the complete redesign of the feeding
network, which is not performed here.

39
3. L-band Antenna System for
CubeSats

When a CubeSat’s initial attitude is determined, the spacecraft is oriented to the flight position by
the ADCS. The system afterwards continues to monitor and correct the attitude, allowing the
payload (downlink and/or uplink) antennas to be facing the Earth at all times. This capability
enables high-performance CubeSat antenna designs with an increased directivity or even shaped
beams. Planar antennas or antenna arrays are a good candidate, since their narrower beamwidth
is not a limiting factor in this scenario. If, however, a lower communication frequency is selected
(L, S bands), a 3U CubeSat offers a very limited surface for antennas, much less arrays. Therefore,
various miniaturization techniques must be applied during the antenna design in order to fit the
low-frequency antennas on a CubeSat face.

Another alternative is to use the surfaces allocated to the solar cells and integrate the antenna
radiating elements with them. The degree of integration depends on the relative position of the
antenna and solar cell elements: one above the other, in the same level, or on opposing sides of
deployable wings. A larger degree of integration presents an increased difficulty of preserving a
good performance of both functions – radiation and power generation.

A combination of the two approaches is used in this chapter to design an antenna system for full-
duplex communication in L band [C2]. Miniaturized antenna elements are placed on either the
CubeSat body or the Earth-facing side of deployable solar panels. In the first stage, a single patch
antenna per frequency band is designed. The high-permittivity substrate reduces the antenna
size, but also causes the antenna to couple to the CubeSat chassis, distorting the radiation pattern.
The effects of the pattern distortion are considered in detail for an arbitrary antenna placement
on a 3U CubeSat body or deployable solar panels.

Antenna arrays are created in the following stages, using the designed patch antennas as array
elements. The arrays radiate first two and then four tilted CP beams for an increased capacity of
3. L-band Antenna System for CubeSats

the communication system. The sequential rotation of dual-band Tx/Rx LP antenna elements is
used to generate CP radiation while halving the required number of antennas. Alternatively, the
L-band antennas are combined with S-band elements as another way of minimizing the surface
required for antennas on a CubeSat [C3].

The chapter opens with a short literature review on the planar antennas for solar cell integration.
Although the integration level considered in this thesis is rather low, it was deemed necessary to
compare the different approaches here, in order to show the difficulties and efforts associated with
this task.

3.1. Review of Solar-Panel-Integrated Antennas

The CubeSat onboard power is limited and most of today’s CubeSats have a majority of the surface
covered by the solar cells, including the deployable solar-panel wings (see, for example,
Figures 1.4 and 2.1). The power requirements lead to a contest for the precious satellite real estate,
between the solar cells, antennas, sensors, and other features on the spacecraft surface. To
overcome this challenge, several different attempts were made to combine the solar cells and
antennas on the same surface, with an aim of maximizing the efficiency of both individual
components. There are several ways of how antennas and solar cells can coexist on a common
surface, depending on the spacecraft section:

 Satellite body:
o Antennas below solar cells
o Antennas and solar cells at the same level
o Antennas above solar cells
 Deployable wings:
o Antennas on the back side of solar cells (this thesis)

A detailed comparison of the possible integration configurations versus various antenna types
and geometries are given in [61] alongside several demonstrated implementations.

Several examples are shown in Figure 3.1. Three variations of modified patch antennas, placed
above and below the solar cells, are presented in [62] for CubeSat communication in UHF and S
bands. One of the presented designs is a transparent patch antenna, where both the patch and the
ground plane are made of a gridded conductor on a glass substrate. The entire design is to be
placed on top of the solar cell with a minimal impact on the cell performance, due to a large
antenna transparency. The CP performance was enabled using two sequentially rotated dual-
band LP elements. A similar technique was adopted in [63] for the reflectarray elements on top of
solar cells. A tradeoff between the metallic conductivity (copper or transparent conductors) and
the optical transparency was demonstrated.

42
 3.1. Review of Solar-Panel-Integrated Antennas

The described techniques are useful for scenarios where it is necessary to optimize the usage of
the satellite body. Moreover, similarly as in conventional satellites, CubeSats usually have one or
several deployable solar panels as a source of additional power supply. In LEO, the solar cells are
placed on the panel sides facing away from Earth. This leaves the panel surface facing Earth
available – ideal location for Earth-communication antennas. This feature was used in [13] to
create a structure that contains a reflectarray on one face of the deployable panel and solar cells
on the other, with a metallic ground completely isolating the two components.

Combining the two functionalities does not come free, as the total structure thickness must be
minimal for a neat stowage on the CubeSat exterior. The antenna elements suffer the most from
this limitation since the bandwidth of resonant element is reduced. The effect of a narrow
bandwidth is more critical at low frequencies.

(a) (b)

(c) (d)
Figure 3.1. Solar-panel-integrated antennas. (a) Transparent-mesh patch antenna on a glass
substrate [62]. (b) Deployable patch antenna array [64]. Reflectarray cells on the (c) same [63]
and (d) opposite side of the solar cells [13].

43
3. L-band Antenna System for CubeSats

3.2. Design Requirements

L band is not often used for CubeSat communication purposes, mostly radiometry [2]. However,
the CubeSats can profit from the frequencies allocated for Iridium and other LEO satellite
constellations. Such a scenario is considered in this chapter, where a duplex communication is
specified, at the downlink (i.e. Tx, 1.525 – 1.555 GHz) and uplink (i.e. Rx, 1.6265 – 1.66 GHz)
frequencies of the Mobile Satellite bands, allocated for the by the ITU [24]. The specifications
correspond to bandwidths of ~2.2% per frequency band.

The requirements predict the evolution of the antenna system in several stages:

 Stage 1: Single 90° x 90° broadside beam

 Stage 2: Two 45° x 90° beams, tilted away from broadside by ~20°
 Stage 3: Four 45° x 45° beams, tilted away from broadside by ~35°

The purpose of multibeam radiation is to increase the total system capacity by several
independent transceivers and an increased antenna gain per beam. Each beam must be served
by both Rx and Tx frequencies for full duplex communication. The final operating frequencies of
the system within the Tx/Rx bands will be only a subset of the specified range, and can be a subject
of change over time. Therefore, the goal is to cover as much of the specified range as possible. The
antenna gain should be larger than 5 dBi and the polarization should be circular.

The space allocated for the antennas is the 3U-CubeSat body for Stage 1 and the backsides of solar
panels for Stages 2 and 3. Due to the previously outlined reasons, the antenna elements must have
a low profile. The number of solar panels for the considered scenario varies from one to three 3U
panels on each side of the CubeSat. A visual representation of the radiation-stage evolution is
depicted in Figure 3.2. The number of solar panels on each side (two) is arbitrarily selected for this
illustration.

(a) (b) (c)

Figure 3.2. Evolution stages of the radiation-pattern configuration, having (a) one, (b) two, and
(c) four beams.

44
 3.3. Single Beam (Stage 1)

3.3. Single Beam (Stage 1)

A broadside beam is easily generated by a single patch antenna element; the real difficulty here is
to cover both Tx and Rx frequency bands by that same element. It is difficult to achieve the
bandwidth of 8.5% and control the radiation characteristics with a single patch antenna,
considering the strict profile limitation at L-band frequencies. Therefore, it was decided to use
one patch antenna per frequency band (Tx/Rx). Although this configuration requires more space
on the CubeSat exterior, the system design is simplified, as the transmitter and receiver are
connected directly to the corresponding antennas and a diplexer is not needed inside the
CubeSat.

The antenna size is indeed a key requirement, since two conventional L-band patch antennas
require a majority of the 3U CubeSat face, leaving no space for other devices (e.g. S-band
antenna). A high-permittivity dielectric material Rogers TMM10i (εr=9.9, tan δ=0.002, h=5.08 mm)
is used for the antenna substrates, and the antennas are miniaturized through dielectric loading.
The substrate has stable properties in a large temperature range – a desirable feature for space
applications. The dielectric loading significantly increases the quality factor of the antenna,
leading to reduced antenna dimensions, at the cost of a narrow frequency bandwidth. The two
miniaturized patches fit comfortably on a 3U CubeSat face. As the radiator size is reduced, the
beamwidth also becomes larger, consequently. The associated design tradeoffs are summarized
in Table 3.1.

A small electrical size of the antenna causes the radiating element to couple to the satellite
structure – an effect commonly seen in small antennas [65]. The effects related to the patch
antenna placement on small spacecraft were briefly investigated previously [66]. In the following
section, the effects are studied in detail, for the special case of CubeSats.

3.3.1. Chassis Coupling Effects

Throughout this chapter, the ground plane is approximated by a 300 × 100-mm2 PEC rectangle,
having in mind that the exact dimension of the CubeSat chassis is slightly larger. The ground plane
dimensions accurately model both the large CubeSat face and the backsides of solar panels,
whichever of the two is chosen for the final antenna location.

Three LP patch antennas, having different substrate permittivities, were placed asymmetrically
on the 3U ground plane, and the S11 parameters and radiation patterns were calculated. The
radiation patterns are also calculated on a 100 × 100-mm2 (1U) ground plane, for comparison. The
results clearly show a stronger main-lobe degradation for a higher substrate permittivity,
indicating that the excited 3U ground plane is also contributing to the total radiation. The
degradation can be observed for other antenna locations on the ground plane.

45
3. L-band Antenna System for CubeSats

Table 3.1. Design tradeoffs for CubeSat patch antennas, versus the dielectric permittivity of the
antenna substrate.

Relative permittivity (εr) High Low

Dimensions + Small − Medium/Large
Beamwidth + Wide − Medium
Bandwidth − Narrow + Wide
CubeSat structure coupling − Strong + Weak

The effect is predominant for the antenna polarization aligned with the long axis of the ground
plane, as in Figure 3.3(a), and it is negligible for the orthogonal antenna orientation (not shown
here). Nevertheless, a CP antenna will suffer from the same degradation effects.

In another set of models, a single CP antenna was placed on several locations on an identical
ground plane, as in Figure 3.3(c), and the radiation pattern was calculated. It was observed that
the main lobe shape strongly depends on the antenna position, and this effect must be carefully
considered for a given antenna configuration [C2]. Furthermore, if there is a freedom of selecting
the antenna position, it can be selected to exhibit a certain radiation pattern and suit a specific
scenario. For example, the antenna offset of 0 mm yields a beamwidth of 110° and its shape
resembles an isoflux pattern – a very desirable feature for LEO satellites in general. Note that this
pattern shape is generated by nothing else other that a proper antenna placement.

In other words, the antenna coupling to the CubeSat structure is not necessarily a negative effect,
and in some cases, it can be used to our advantage. Some researchers even propose to excite the
CubeSat chassis itself for radiation, using a set of properly distributed small antenna
elements [67]. In any case, the chassis coupling effects should not be neglected when designing
CubeSat antennas, especially if the antennas are electrically small.

3.3.2. Antenna Prototype

As a high-permittivity substrate reduces the antenna bandwidth, a large substrate thickness was
selected to alleviate this reduction. The antenna geometry is a square patch with chamfered
corners for CP operation. Two patch antennas were designed and tuned for Tx and Rx frequencies.
First antenna prototypes were fabricated and placed on a metallized 100 × 300-mm2 FR4 panel for
anechoic chamber measurements [C2]. The antennas are fed using a small MMCX connector with
a lock-snap mechanism and connected to a VNA with an MMCX-to-SMA adapter. The antenna
simulation model and the Rx antenna prototype are shown in Figure 3.4.

46
 3.3. Single Beam (Stage 1)

(a)
TMM10i r=10.1 TMM4 r
=4.5 Duroid 5870 r=2.33
8 8 8

6 6 6

4 4 4

2 2 2

0 0 0

-2 3U -2 3U -2 3U
1U 1U 1U
-4 -4 -4
-90 -60 -30 0 30 60 90 -90 -60 -30 0 30 60 90 -90 -60 -30 0 30 60 90
[°] [°] [°]

(b)

(c)
8

0
3U Doffset = 100 mm
-2 3U Doffset = 50 mm
3U Doffset = 0 mm
-4
1U Doffset = 0 mm

-6
-90 -60 -30 0 30 60 90
[°]
(d)
Figure 3.3. (a) LP patch antenna on a 3U CubeSat face. Dimension L changes according to the
substrate permittivity, and the antenna is always resonant at the same frequency. (b) LP
radiation patterns along the large axis of the ground plane, for various substrate permittivities.
(c) CP patch antenna on a 3U CubeSat face. (d) CP radiation patterns along the large axis of the
ground plane, versus the offset from the center. Radiation patterns of identical antennas, placed
on a 1U CubeSat face, are shown for comparison. Default dimensions (mm): L=27, Wgnd=100,
Lgnd=300, Doffset=75, dchamf=4.25.

47
3. L-band Antenna System for CubeSats

(a)

(b) (c)
Figure 3.4. (a) Tx and Rx patch antennas on a 3U CubeSat face. (b,c) Rx patch antenna
prototype. (b) Bottom, with the MMCX connector visible. (c) Top. Dimensions (mm): LTx=28.7,
LRx=27, Lsub,Tx=Lsub,Rx=45.

After the first set of prototypes was characterized, a frequency shift of ~35 MHz was observed in
the measured S parameters, as compared to the simulations (Figure 3.5(a)). The permittivity was
tuned in the simulation models, and an estimated final value was found to be ~10.15 in L-band.
After the correction, there was still a discrepancy attributed to the used connectors, which were
not included in the simulation model.

Realized gain and AR values of both prototypes are presented in Figure 3.5(b). The gain of the Rx
antenna is lower due to a mismatch introduced by the used connector, as can be seen in the S-
parameter response. The measured and simulated radiation patterns of the Rx antenna only are
compared in Figure 3.5(c). The results of the Tx antenna are omitted here, as that antenna exhibits
a similar performance. The patches exhibit a wide beamwidth of ~90°, conforming to the design
requirements. The radiation results show that the pattern in the φ=0° plane is not influenced by
the antenna offset with respect to the long axis of the satellite face, except for a small beam tilt of
5°. However, there is a more significant pattern distortion in the φ=0° plane as a consequence of a
large ground.

48
 3.4. Two Tilted Beams (Stage 2)

3.4. Two Tilted Beams (Stage 2)

The next considered scenario is having two 45° x 90° radiating beams, tilted away from broadside.
The given configuration can be enabled using 2 × 1-element patch-antenna arrays. However,
since the elements, described in the previous section, are single-band, a separate array must be
assigned per beam and per frequency band. The total required number of antenna elements in
that case is eight. Without the loss of generality, the scenario can be analyzed considering only the
Rx patch antenna.

The patch antenna described in the previous section is used as an array element, and the inter-
element spacing is tuned to provide the required beamwidth in the narrow plane. A two-element
CP array is shown in Figure 3.6(a). The array center coincides with the ground plane center. The
inter-element spacing of 100 mm yields a beamwidth of 45°, and the beam tilt of -20° is enabled
by an appropriate relative phase shift.

3.4.1. Sequential Rotation of CP elements

Sequential rotation of the two elements can be used to improve the axial ratio performance in
both frequency and angular coverage [57], [68]. The technique was implemented in the array
shown in Figure 3.6(a), as the element rotation angles are selected as multiples of 90°. Active-S
parameters were calculated at each port, which take into account both the matching and the cross
coupling between elements. The individual active-S parameters from Figure 3.6(b,c) strongly
depend on the relative rotation angle the cross-coupling value reaches -16 dB for some cases. The
element rotation can be selected to simultaneously minimize the coupling and to keep the active-
S parameters below -10 dB. The individual antenna elements can be further adapted for each
rotation angle, but it was avoided here for simplicity.

Although this technique exhibits its real potential for a larger number of elements in an array, it
has demonstrated a significant AR improvement even for only two elements considered here. The
results of two representative array configurations are presented in Figure 3.6(d). A two-fold
increase in AR bandwidth is obtained by the sequential rotation technique. Maximum gain
reaches 7.5 dBi. The total array radiation was plotted in polar coordinates in Figure 3.7. The
coverage area is indicated by a dashed line in the polar plots. The gain is larger than 3 dBi and the
average AR value is 2.63 dB, along the entire specified angular range. An identical 2-element array
is to be used to cover the opposite half of the hemisphere.

It might seem that the array performance is not as good as it could be expected from the single-
element results. Several factors prevent further improvement:

 Beam scanning – The main lobe of the array is scanned to 20° from broadside, at which
direction the element pattern naturally has a lower gain.

49
3. L-band Antenna System for CubeSats

 Element rotation – Different orientation with respect to the ground plane cause the two
elements to exhibit a slightly different radiation pattern.
 Element location – It is previously shown that the element position on the asymmetric
ground plane is a dominant factor that determines the element radiation pattern.
Although the elements’ offset from the ground center is identical, the individual gain
values are different in the beam-scanning direction.
 Cross-coupling – The radiation pattern of one element further changes in the presence of
the other.

-10
S11 sim.
S11 meas.
S22 sim.
-20
S22 meas.
S21 sim.
S21 meas.
-30

-40
1.48 1.51 1.54 1.57 1.6 1.63 1.66 1.69 1.72
Frequency [GHz]

(a)
8

Gain sim.
Gain meas.
4
AR sim.
AR meas.

0
1.48 1.51 1.54 1.57 1.6 1.63 1.66 1.69 1.72
Frequency [GHz]

(b)
Normalized CP gain [dBi]

Normalized CP gain [dBi]

(c)
Figure 3.5. Simulation and measurement results of Tx and Rx patch antennas. (a) S parameters.
(b) Broadside realized CP gain and AR vs. frequency. (c) Radiation patterns of the Rx antenna,
in two principal planes, at 1.63 GHz.

50
 3.4. Two Tilted Beams (Stage 2)

(a)

(b) (c)

(d)
Figure 3.6. (a) A 2-element Rx patch antenna array on a 3U CubeSat face. (b,c) S parameters and
active-S parameters, and (d) realized CP gain and AR, for the two rotation configurations, versus
frequency. The gain and AR are calculated at φ=90°, θ=‐20°.

51
3. L-band Antenna System for CubeSats

(a) (b)
0

= 0°, = -90°
-5

-10

-15

-20 LHCP 1.625 GHz

RHCP 1.625 GHz
LHCP 1.635 GHz
-25 RHCP 1.635 GHz
LHCP 1.645 GHz
RHCP 1.645 GHz
-30
-90 -60 -30 0 30 60 90
[°]
(c)
Figure 3.7. The radiation performance of a 2-CP-element patch array. Presented are contour
plots in polar coordinates, at 1.63 GHz. (a) Realized CP gain. (b) AR. (c) Realized CP gain in the
plane marked by a dotted line in (a). The rotation configuration is α=0°, β=-90°. Dashed lines
and the cross indicate the coverage area and the beam scanning direction, respectively.

A MATLAB code was developed to visualize the beam footprint on the Earth’s surface. The
maximum half-angle of Earth visibility, seen from the CubeSat (θmax), is a function of the orbit
altitude, and can be obtained from Figure 3.8(a) using the relation:

asin (3.1)

The radiation pattern of the CubeSat antennas can be projected on the Earth surface by:

sin
atan (3.2)
1 cos
where β is the angle between the CubeSat’s projection and the observation point on the Earth
surface. A contour plot, shown in Figure 3.8(c,d), is obtained using the function
sphere3d.m [69].

52
 3.4. Two Tilted Beams (Stage 2)

(a) (b)

(c) (d)
Figure 3.8. (a) The CubeSat’s position above the Earth surface, with associated dimensions
indicated. h=650 km, R=7378 km. (b) A visual representation of the relation (3.2). (c,d) The
radiation pattern of the array from Figure 3.6, projected on the surface of the Earth.

A 650-km orbit was considered for the contour plot. The projection is useful for estimating the
antenna coverage for a given radiation pattern. The results can be combined with the antenna
transmitting power and the propagation losses to obtain the power density on the surface of the
Earth.

3.4.2. Antenna Array Prototype

An array prototype was built using two patch antenna prototypes from the previous measurement
campaign. The resonant frequency of both elements is 1.6 GHz. The MMCX connectors were
replaced with SMA connectors and the measured S11 parameter was below -9 dB for both
antennas. Two dedicated power dividers were built for the array radiation measurements. The
power dividers consist of Wilkinson dividers and phase-delay lines, where the delay-line lengths
are calculated to suit a specific antenna rotation scenario (0° or 90°). The array prototype is shown
in Figure 3.9.

53
3. L-band Antenna System for CubeSats

(a)
Normalized CP Gain [dB]

Normalized CP Gain [dB]

(b) (c)

(d)
Figure 3.9. (a) Patch-antenna array prototype inside the anechoic chamber. (b,c) Measured
radiation patterns, and (d) measured AR at φ=90°, θ=‐20°, versus frequency, for the two rotation
configurations.

54
 3.5. Four Tilted Beams (Stage 3)

The array was evaluated in an anechoic chamber for the two simulated configurations, and the
radiation patterns are presented in Figure 3.9(b,c). The AR improvement of the sequentially
rotated array was confirmed through the experiment, tripling the AR bandwidth of a single
element, which can be seen in the same figure.

3.5. Four Tilted Beams (Stage 3)

Proceeding in the same fashion as until now, a four-element array is formed to radiate a 45° × 45°
tilted beam. If the same technique is applied for the four-element array, as in the previous stage,
the required number of elements would be 4 elements x 4 arrays x 2 frequencies = 32 antennas.
This configuration is not acceptable for a CubeSat, considering the required surface, associated
total weight and feeding-network complexity. Instead, a different approach is adopted.

Sequential rotation technique was initially proposed to generate CP radiation using LP antenna
elements [57]. Here, we can simply modify our CP patch element into a dual-band dual-LP
element. In that design, the two feeding ports excite two independent orthogonal modes in the
patch, at two different frequencies. The mode independence means that it is possible to generate
a distinct beam at the two frequencies, with dedicated feeding networks for each of the
frequencies. The advantage of this method is the reduction of the total number of antenna
elements by a factor of two.

The antenna array is shown in Figure 3.10, indicating the separate feeding locations for the two
modes/frequency bands. The array radiation is calculated for three different scanning azimuth
angles of -90°, -64° and -45°. The beam tilt is achieved with a varying phase of the excitation
signals, on top of the fixed progressive phase in steps of 90°, which follows the physical rotation.
The radiation patterns in polar coordinates are presented in Figure 3.11. The array layout, shown
next to each set of figures, indicates the scanning direction that corresponds to the results in those
figures.

The three beam-scanning scenarios show a comparable performance, with minor differences in
AR levels at the ends of the coverage area. The last configuration (45°) shows a slightly lower
average AR than the remaining two. The scanning angle will also depend on the CubeSat flying
orientation and the network scenario. In any case, the results demonstrate that such array
configuration can be used for a multibeam radiation scenario for LEO CubeSats. If a 4-element
array of sequentially rotated CP patch antennas were scanned in the same direction, the AR
performance would be further improved, at the cost of an increased number of antennas. The
proposed configuration of four LP patch elements is the optimal compromise.

55
3. L-band Antenna System for CubeSats

Figure 3.10. The geometry of a 4-LP-element patch array. Black and white circles mark the
feeding locations of the 1.63-GHz and 1.54-GHz resonances, respectively. Dimensions (mm):
L=27, W=29, D=100.

3.6. Stacked-Patch Antenna in L and S bands

Another way of reducing the surface allocated for the CubeSat antennas is to combine individual
patch elements into a dual-band antenna. The approach is demonstrated in this section through
a stacked-patch antenna operating in L and S bands. The S-band frequencies are the same as the
ones considered in Chapter 2. Each of the two bands contains two uplink/downlink sub-bands
and the frequencies considered in this section are arbitrarily selected throughout this section. The
presented dual-band stacked patch can also be used as an element of an antenna array in L-band,
with the S-band counterpart working as a standalone antenna.

3.6.1. Patch Antenna Elements

Linearly polarized (LP) square patch antennas, shown in Figure 3.12, were initially designed on
several space-eligible dielectric substrates to study their size versus the dielectric permittivity. The
thickness of the individual substrates was fixed at 3.18 mm, the substrate was 4 mm larger than
the patch in each model, and the permittivity was varied from 2.33 to 9.9. The antennas were
placed on a 100 × 100-mm2 ground plane, which corresponds to a single 1U-CubeSat face. The
patches were tuned to 1.54 GHz and 2.05 GHz, and the resulting geometrical parameters of all
variations are given in Table 3.2.

56
 3.6. Stacked-Patch Antenna in L and S bands

(a) (b)

(c) (d)

(e) (f)
Figure 3.11. The radiation performance of a 4-LP-element patch array. Presented are contour
plots in polar coordinates, at 1.54 GHz. (a,c,e) Realized CP gain. (b,d,f) AR. The scanning
azimuth angle is (a,b) -90°, (c,d) -64°, and (e,f) -45°. Dashed lines and the cross indicate the
coverage area and the beam scanning direction, respectively.

In a stacked-patch configuration, where each radiator is fed independently, the difference in

individual patch sizes must be sufficient to increase the isolation. In addition, each of the patches
must be exposed to the free space for a proper radiation. With that in mind, a typically chosen
configuration is the one with the smaller high-frequency element on top [70]. However,
considering the larger required bandwidth in the S band (4.1%), compared to the L band (2.2%),
it was advantageous to use a low-permittivity substrate for the S-band antenna, as it would yield
a larger bandwidth. For this reason, the geometry that was selected (bold cells in Table 3.2) has a

57
3. L-band Antenna System for CubeSats

higher permittivity substrate and a smaller size for the L-band patch. Consequently, it is actually
placed above the higher frequency S-band patch, contrary to the more conventional
configuration [71].

Two antenna geometries were compared. In the first one, each patch is dual-fed with coaxial
probes and a relative phase offset of 90° for the CP excitation. In the second one, the radiating
elements are corner-chamfered single-fed patches. The two antenna geometries are shown in
Figure 3.13(a,c). Circular holes are removed from the bottom patch in the feeding locations of the
top patch. The bottom substrate is Duroid 5870 (S-band patch) and the top substrate is TMM10i
(L-band patch). The S-band patch dimension has a large influence on the L-band resonant
frequency, and the two patches must be tuned simultaneously.

S parameters of the two antenna geometries are shown in Figure 3.13(b,d). The dual-resonance
behavior in the single-fed model is a consequence of the CP excitation method. A rather large
coupling between the two elements (S13 and S12 parameters in the dual- and single-fed models,
respectively) come from the fact that these ports excite resonant modes on the two patch elements
with parallel current directions. As the radiating edges of the two modes are also parallel and close
to each other, the coupling is large. Additional work is required to improve the port isolation,
which was not the goal of this study. The radiation patterns show a larger beamwidth in L band,
due to a higher permittivity substrate and a smaller patch dimension.

Table 3.2. Geometric parameters and performance

LP patch antennas in L and S bands, versus the
substrate permittivity.
Duroid
TMM10i TMM4
Substrate material 5870
εr=9.9 εr=4.7
εr=2.33
Patch [mm] 30 43 60.5
xfeed [mm] 4 6.2 9
L
Bandwidth [MHz] 13.7 23 27
band
Gain [dBi] 5.8 6.5 7.2
3-dB Gain BW [MHz] 54 71 82
Patch [mm] 22.3 32.3 45.3
xfeed [mm] 2.5 4.3 7
S
Bandwidth [MHz] 22 32 43
band
Gain [dBi] 6.3 7 7.8
Figure 3.12. LP patch antenna 3-dB Gain BW [MHz] 68 100 131
geometry.

58
 3.6. Stacked-Patch Antenna in L and S bands

-10

-20

-30 S11
S33 S12
-40 S34 S14
S23 S13
S13
-50
1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2
Frequency [GHz]
(a) (b)
S parameters [dB]

(c) (d)

-10 LHCP f = 1.54 GHz

RHCP f = 1.54 GHz
[dB]

LHCP f = 2.05 GHz

RHCP f = 2.05 GHz
-20

-30
-150 -100 -50 0 50 100 150
Frequency [GHz]
(e)
Figure 3.13. (a) Dual-feed stacked-patch antenna geometry and (b) the corresponding S
parameters. (c) Single-feed stacked-patch antenna geometry and (d) the corresponding S
parameters. (e) Radiation patterns of the dual-feed antenna in (a).

59
3. L-band Antenna System for CubeSats

3.6.2. Feeding Networks

Three stripline feeding networks were designed and evaluated for the described stacked-patch
dual-fed antenna structure. The networks are based on:

 T-junction power divider

 Wilkinson power divider
 Branchline coupler

The three network layouts are presented in Figure 3.14. In the first two, the phase difference is
achieved with 90° phase-delay lines, whereas the branchline coupler has an inherent relative
phase difference of the two output ports. However, it is also the largest, and some line meandering
was required to reduce its total surface. Rogers RT Duroid 5870 was used for both dielectric layers
of the stripline feed, each having a thickness of 0.508 mm. The stripline ground planes were
grounded in several locations and shielded along the edges, to avoid the excitation of parallel-
plate modes inside the frequency bands of interest. The shielding distribution is performed having
in mind the S-band frequencies, where the modes will appear first. Coaxial probes are placed at
the locations of the patch feeds.

Figure 3.14. Models of the stripline feeding networks used for comparison: (a) T-junction,
(b) Wilkinson-divider, and (c) branchline-coupler models. (d) Side view of the feeding
networks. The two ground planes are shielded around the edges in all cases. Additional
shielding is added in the branchline-coupler model.

60
 3.6. Stacked-Patch Antenna in L and S bands

0 -2 100

-10 -3 95
S 11
-20 -4 90
S 23 S 21
-30 -5 S 31 85
| 31- 21|
-40 -6 80
1.8 1.9 2 2.1 2.2 2.3 1.8 1.9 2 2.1 2.2 2.3 1.8 1.9 2 2.1 2.2 2.3

0 -2 100
S 11
-10 -3 95
S 23
-20 -4 90
S 21
-30 -5 S 31 85
| 31- 21|
-40 -6 80
1.8 1.9 2 2.1 2.2 2.3 1.8 1.9 2 2.1 2.2 2.3 1.8 1.9 2 2.1 2.2 2.3

0 -2 100

-10 -3 95

-20 -4 90
S 11 S 21
-30 S 23 -5 S 31 85 | 31- 21|

-40 -6 80
1.8 1.9 2 2.1 2.2 2.3 1.8 1.9 2 2.1 2.2 2.3 1.8 1.9 2 2.1 2.2 2.3
Frequency [GHz] Frequency [GHz] Frequency [GHz]

Figure 3.15. S parameters of the three feeding networks in Figure 3.14: (left) input matching and
the isolation at the output ports, (center) power distribution at the output ports, and
(right) phase difference at the output ports. The presented performance corresponds to the S-
band section of each model.

The feeding network performance, without the patch elements, is compared in Figure 3.15. The
results are presented for the S band, since the bandwidth requirement is larger in this case, and
identical conclusions could be drawn from the L-band results. The delay lines used in the T-
junction and Wilkinson-divider models provide a relative phase difference that varies linearly
with frequency. However, the difference remains within ±2.5° along the required band. On the
other hand, the branchline coupler exhibits a wideband phase difference. The output isolation is
higher than 20 dB except in the T-junction model (6 dB). All three networks are matched at the
input ports.

3.6.3. Full Antenna Models

Three antenna models in Figure 3.16, composed of the patch elements from Figure 3.13 and the
feeding networks from Figure 3.14, were analyzed on a 100 × 100-mm2 ground plane. The
simulation results are presented in Figure 3.17.

The dual feed and the sophisticated feeding networks cannot influence the narrowband nature of
the patch elements. In all analyzed antenna geometries, the radiating elements only accept the
incident power at the close vicinity of their resonant frequency, and the remaining power is
reflected in the feeding network. Therefore, the overall gain bandwidth changes only slightly
among different models and is mostly limited by the narrowband S11 parameter of the dual-fed

61
3. L-band Antenna System for CubeSats

(a) (b) (c)

(d)
Figure 3.16. Antenna geometry of the three models: (a) T-junction, (b) Wilkinson-divider, and
(c) branchline-coupler models. (d) Side view of all models.

patch element. On the other hand, the single-feed antenna elements from Figure 3.13(c) have a
wider impedance bandwidth than the dual-feed models. In this case, the main factor limiting the
gain bandwidth is the inherently low AR of the single-feed antenna geometry. Consequently, all
four versions have comparable realized CP gain bandwidths.

The purpose of a feeding network is to provide a stable relative phase difference at the antenna
ports, over a specified frequency range. Thus, only the AR bandwidth improvement can be
expected if a sophisticated feeding network is used in combination with a narrowband radiating
elements. The three feeding network configurations illustrate the discussed points.

In the case of the T-junction divider, the signals, reflected from the patch element, partially travel
to the opposite antenna feed, contributing to the cross-polarization (see Chapter 2 for a detailed
explanation). For this reason, the AR of the T-junction model is low only at frequencies where the
S11 parameter is also low. In the other two scenarios (Wilkinson, branchline) the power reflected
from the patch is dissipated in the resistors, which can be seen from the low S23 values in
Figure 3.17. Consequently, the two models exhibit a wideband AR performance and impedance
matching at the input ports. In fact, the S14 parameter of the branchline coupler model can be used

62
 3.6. Stacked-Patch Antenna in L and S bands

to estimate the return losses of the patch element itself. The S11 of the T-junction models, with two
resonances, resembles the one of the single-feed model. However, this is because the FN (higher
resonance) and the patch element (lower resonance) were not tuned to exactly the same
frequencies.

L band S band L band S band

8 8 0 0
Gain [dBi], AR [dB]

Gain Gain [dBi], AR [dB]

S parameters [dB]

S parameters [dB]
6 AR 6 -10 -10
T junction

4 4 -20 -20
S 22 S 11
2 2 -30 -30
S 21 S 21

0 0 -40 -40
1.5 1.53 1.56 1.59 1.9 2 2.1 2.2 1.5 1.53 1.56 1.59 1.9 2 2.1 2.2

8 8 0 0
Gain [dBi], AR [dB]

Gain [dBi], AR [dB]

Gain S 22 S 11
S parameters [dB]

S parameters [dB]
6 AR 6 -10 -10
S 21 S 21
Wilkinson

4 4 -20 -20

2 2 -30 -30

0 0 -40 -40
1.5 1.53 1.56 1.59 1.9 2 2.1 2.2 1.5 1.53 1.56 1.59 1.9 2 2.1 2.2

8 8 0 0
Gain [dBi], AR [dB]
Gain [dBi], AR [dB]

Gain
S parameters [dB]

S parameters [dB]

6 AR 6 -10 -10
Branchline

4 4 -20 -20
S 33 S 11
2 2 -30 S 31 -30 S 31
S 32 S 41
0 0 -40 -40
1.5 1.53 1.56 1.59 1.9 2 2.1 2.2 1.5 1.53 1.56 1.59 1.9 2 2.1 2.2

8 8 0 0
Gain [dBi], AR [dB]

Gain [dBi], AR [dB]

Gain
S parameters [dB]

S parameters [dB]

6 AR 6 -10 -10
Single Feed

4 4 -20 -20
S 22 S 11
2 2 -30 -30
S 21 S 21

0 0 -40 -40
1.5 1.53 1.56 1.59 1.9 2 2.1 2.2 1.5 1.53 1.56 1.59 1.9 2 2.1 2.2

Figure 3.17. Complete antenna performance of (top) T-junction, (center-top) Wilkinson-

divider, (center-bottom) branchline-coupler models in Figure 3.16, and (bottom) the single-
feed model in Figure 3.13(c).

63
3. L-band Antenna System for CubeSats

Of the analyzed models, the branchline-coupler model provides the most stable phase difference
over a wide frequency band. However, having in mind the required bandwidths, the phase
difference of the Wilkinson-divider model is sufficient here, even in the more limiting case, in the
S band (4.9%). Considering all criteria, it was selected as the optimal for this application. A
prototype of this antenna was fabricated and the results are presented in the following section.

3.6.4. Antenna Prototype

The antenna prototype was finally tuned to the Tx frequencies of L and S bands. A Wilkinson-
based stripline FN was initially fabricated. Some complications were experienced while attaching
the two stripline layers together, and this FN prototype was discarded. Another FN was realized in
microstrip technology and used during experiments. The ground plane of the inverted microstrip
FN acts as a ground plane for the patch elements above. SMA connectors are soldered vertically
to the microstrip board. The two FNs were excited individually, one at a time, while the other one
was matched with a coaxial 50-Ω load. The antenna prototype is shown in Figure 3.18. The
measurement results of the full antenna, presented in Figure 3.19, confirm the predicted antenna
performance. The main performance aspects are compared in Table 3.3. The input reflection
coefficient corresponds mainly to the S11 parameter of the FN and the impedance bandwidth is
therefore not specified. The broadside AR values are lower than 2 dB across the 3-dB-gain
bandwidth, in all models. The beamwidth difference at two frequency bands is a direct
consequence of the substrate permittivity, and thus, the patch size.

(a) (b)
Figure 3.18. Stacked-patch antenna prototype in the anechoic chamber. (a) Front. (b) Back. The
coaxial matched load is connected to the opposite FN connector.

64
 3.6. Stacked-Patch Antenna in L and S bands

0 0

-10 -10

-20 -20
S 11 sim. S 22 sim.
S 11 meas. S 22 meas.
-30 S 21 sim. -30 S 21 sim.
S 21 meas. S 21 meas.

-40 -40
1.4 1.45 1.5 1.55 1.6 1.65 1.7 2 2.1 2.2 2.3 2.4 2.5
Frequency [GHz] Frequency [GHz]
(a) (b)

(c) (d)

8 8
Gain LHCP sim.
Gain LHCP meas.
6 AR sim. 6
AR meas.

4 4 Gain LHCP sim.

Gain LHCP meas.
AR sim.
AR meas.
2 2

0 0
1.4 1.45 1.5 1.55 1.6 2.1 2.15 2.2 2.25 2.3 2.35 2.4
Frequency [GHz] Frequency [GHz]
(e) (f)
Figure 3.19. Measured and simulated performance of the dual-band stacked-patch antenna.
(a,b) S-parameters. (c,d) Radiation patterns. (e,f) Broadside realized gain and AR versus
frequency. The second resonance in the measured S11 curve of the L-band patch is frequency-
shifted patch-antenna resonance.

65
3. L-band Antenna System for CubeSats

Table 3.3. A comparison of the simulated and measured dual-band antenna performance.

L band S band
Simulation Measurement Simulation Measurement
Maximum gain [dBi] 5 4.2 7.3 7.2
3-dB gain
40 40 200 200
bandwidth [MHz]
Maximal AR [dB] 0.9 2 1.4 1.6

Beamwidth [°] 93 100 78 75

3.7. Summary and Discussion

A patch-antenna system in L band is proposed in this chapter, with a focus on 3U-CubeSat
communication in LEO. The CubeSat’s attitude-control system constantly maintains the satellite
orientation with respect to the Earth. Therefore, the locations envisioned for the antenna
elements include the satellite face and the backsides of deployable solar panels, which are facing
the Earth at all times. Both antenna locations are approximated as 300 × 100-mm2 PEC rectangles
throughout this study. The chapter proposes several antenna configurations, which radiate of one,
two and four independent beams.

Considering the small electrical size of CubeSats at L-band frequencies, dielectric loading with
high-permittivity substrates is used to reduce the patch-antenna dimensions. A small, high-
permittivity antenna exhibits a wide beamwidth, narrow bandwidth, and, most of all, strong
coupling to the asymmetric ground plane, which causes the radiation pattern of the antenna to be
deteriorated. A detailed analysis of the pattern deterioration showed that the antenna placement
is critical for its radiation performance, and that it should not be neglected in the antenna design
phase.

Two patch elements are tuned to the Rx and Tx frequencies in L band, and analyzed on a large
face of a 3U CubeSat. In the next stage, the Rx element is used to create two-element sequentially
rotated arrays and provide a beam tilt of 20° from broadside. The sequential rotation technique
improves the AR over frequency, although the number of elements is rather small. The individual
element pattern, caused by the coupling to the asymmetric ground plane, is shown to be the main
limitation of the array performance. Nevertheless, the array satisfies the design requirements,
covering the half of the coverage area. Another antenna array, and the two arrays corresponding
to the Tx frequencies, yield a total of 8 antenna elements required for this scenario.

66
 3.7. Summary and Discussion

Proceeding in the same manner, a total of 32 patch elements would be necessary for 4
independent tilted beams, radiated by 4-element arrays of single-band elements. Instead, a
Tx/Rx-dual-band dual-feed LP patch element is designed, which halves the required number of
elements. A sequential rotation of LP elements is used to radiate CP beams, and the excitation
phase of each feeding port can be tuned to independently scan the beams at Tx and Rx
frequencies. At least four LP elements are necessary for satisfactory performance in the
sequentially rotated array configuration. Therefore, this method cannot be implemented in the
previous design stage. The total number of antenna elements is compared in Table 3.4.

In both the two- and four-element arrays, the patches are rotated in steps of 90°. Other angular
steps (45°, for example) were not investigated, and could potentially further improve the array
performance.

It can be argued that the described patch elements, in their current form, have a profile that is too
large for a convenient stowage of the deployable wings, if the antennas are placed on their
backside. All the effects studied in the scope of this chapter will be also valid for patch antennas
with a smaller profile, except that the frequency bandwidth will be reduced.

A large number of L-band antennas might occupy a majority of the CubeSat’s Earth-facing
surface, not leaving enough space for other features, e.g. S-band antennas. For that reason, the L-
and S-band elements are integrated in a single dual-band stacked-patch antenna in the last part
of the chapter. To improve the AR, three FN, most commonly seen in the literature, were designed
for each of the bands, and compared with a simple single-feed CP stacked-patch antenna. The FN
based on Wilkinson power dividers is found to be the optimal compromise for CubeSat
applications, in terms of its AR performance, size and complexity. Furthermore, the CP gain
bandwidth of a narrowband patch element cannot be significantly improved, regardless of the FN
selection.

Table 3.4. The total number of elements required for the arrays of the CubeSat antenna system
in L band. Each beam is duplex and CP.

Elements Element Feeds/bands Total

Frequencies Stage Beams
per array polarization per element elements
1. 1 / CP 1 2
2 2. 2 2 CP 1 8
(Tx/Rx) CP 1 32
3. 4 4
LP 2 16

67
3. L-band Antenna System for CubeSats

3.7.1. Antennas Flying in Orbit

Several evaluated prototypes of individual patch-antenna elements and antenna arrays were
fabricated and evaluated. The measurement results of all prototypes confirm the antenna
performance, predicted through EM simulations. These are typically first or second prototype
iterations. Afterwards, several other antenna designs, based on the ones presented in this thesis,
were further optimized, including the tuning to the final operating frequencies. The two Tx and
Rx antennas were launched in December 2018 and April 2019, on two of the Astrocast precursor
satellites. Based on the communication-system tests, the antennas operate nominally. A modified
version of the Rx patch antenna and the dual-band stacked-patch antenna are in the final testing
phases, and will be launched on the first batch of commercial satellites in 2020.

68
4. Inter-Satellite Links for LEO
CubeSats

The CubeSat technology offers a cost-efficient alternative to the conventional large-scale

satellites, for achieving global coverage. The reason CubeSats became commercially attractive
is their low development and deployment cost, which allows them to be efficiently launched to
LEO. The advantages of any LEO satellite are low communication latency and propagation loss.
On the other hand, the coverage area of a LEO satellite is small and the link duration with the
ground is short (10-15 minutes), due to the low altitude and high orbital velocity, respectively.
These inherent drawbacks render the global coverage impossible with a small number of
satellites.

The solution is to use a large group of CubeSats to achieve global coverage. In that case, each
point on the Earth’s surface is in contact with at least one CubeSat at all times. A smaller orbit
altitude would then mean a larger required number of CubeSats. However, the information that
a CubeSat collects, or receives from communication devices, is inaccessible until the first time
the CubeSat makes contact with a ground station and downlinks the gathered data. Given the
fact that dedicated ground stations are sparsely distributed, the time delay can reach several
hours. Such a latency is undesirable for most practical commercial applications, and
unacceptable for communication purposes.

One impractical approach is to increase the number of ground stations, which would reduce the
latency only by a small factor and virtually undo the low-cost nature of the CubeSat system [72].
Another idea is to interconnect the constellation in a network and allow the nodes (CubeSats)
to exchange data among themselves. These connections can be established using inter-satellite
links (ISL). The impact a connected space network can have is stunning, as the order of
magnitude of the latency can be reduced from hours to seconds. Although the system
complexity is unavoidably increased by introducing ISLs, it is the optimal solution to achieve a
global coverage.
4. Inter-Satellite Links for LEO CubeSats

The design requirements for ISL antennas and the associated design difficulties are outlined in
this chapter. An advantage can be taken from the specific CubeSat flying formation to simplify
the conventional complex ISL antenna geometries. It is shown that deployable reflectarray and
transmitarray antennas are good candidates for the ISL within a CubeSat constellation, when
the surface of the CubeSat body is not sufficient for a planar antenna to achieve the specified
gain.

4.1. Review of ISL Antennas

The first known ISL was established between Orbiting Satellite Carrying Amateur Radio
(OSCAR) 6 and 7, two amateur-radio satellites on almost identical circular orbits. When the two
spacecraft were 7000 km apart, OSCAR 6 has received the signal, transmitted from OSCAR 7,
and re-transmitted it, thus directly connecting two amateur Earth stations. In the following
period, two-way communication has been reported between points on Earth as far as Japan and
Finland [73]. The later ISL systems were more advanced, being developed for satellite
formations with well-defined and maintained relative positions. ISL communication exists in S-
band and Ka-band frequencies, as well as in the optical range. An exhaustive study of the past,
current and future ISL antennas is presented in [74]. A survey of ISL communication systems for
small satellites can be found in [75].

Data relay satellites represent the first and the largest group of satellites that utilize ISLs. The
purpose of a relay satellite is to connect two satellites, ground stations, airplanes, etc. that
cannot establish a line-of-sight communication. Most famous examples of such systems in GEO
are the Tracking and Data Relay Satellite System (TDRSS) [76] and the European Data Relay
System (ERDS) [77], which primarily serve to enable cross-links between other LEO satellites
and ground stations. These satellites use ISL frequencies from S to Ka bands to relay data using
directional antennas. A notable non-GEO example is the Iridium constellation for satellite
phone and data, shown in Figure 4.1 [78]. Iridium satellites are distributed in several polar orbits
and have four K-band phased-array ISL antennas – two for ISLs in the same orbital plane, and
one for each of the two adjacent planes. Common feature for all of these satellites is their large
size (antenna diameter up to 2 m) and mass (up to several tons).

A second smaller group of satellites using ISL includes small formation flying missions, almost
all being technology demonstrations. This group is particularly interesting for CubeSat ISLs,
since the described spacecraft is much smaller in mass in volume, compared to the first group.
A good example is the Canadian Advanced Nanospace eXperiment (CanX), which is a dual-
nanosatellite formation-flying demonstration mission [79]. CanX 4 and 5 satellites are two
identical 8U CubeSats (200 × 200 × 200 mm3) which communicate at S-band frequencies. Two
ISL patch antennas are placed on opposite sides of the spacecraft, and each antenna provides
hemispherical coverage.

70
 4.1. Review of ISL Antennas

Figure 4.1. Iridium NEXT satellite [80]. Four phased-array ISL antennas are visible, out of
which two serve for ISLs in the same orbital plane, and one for each of the two adjacent orbital
planes. L-band antenna radiates 48 beams for users on Earth.

Another example is the Network and Operation Demonstration Satellite (NODES) mission,
whose objective was to test the control of connected small-satellite swarms that make
distributed scientific measurements [74]. The two NODES are 1.5U CubeSats that achieve the
ISL through two UHF monopoles.

Except for the mentioned examples, no ISL systems or antennas for CubeSats have been
demonstrated to date. On the other hand, several innovative high-gain CubeSat antennas were
designed for different purposes, such as radar, communication, and even deep space. These
antenna geometries are very promising candidates for CubeSat ISL applications, in scenarios
where the satellite relative orientation and distance are defined. The most interesting
geometries were designed in the Jet Propulsion Laboratory (JPL), among which are several
deployable reflectarrays and a deployable mesh reflector antenna, shown in Figure 4.2. The LP
reflectarray (RA) antennas are composed of square patch elements and operate at X and Ka
bands, with a total deployed surface of approximately 18U and 9U, respectively [7], [13]. A very
interesting feature for ISL applications is the capability of CubeSat RA antennas to tilt their
beams at an arbitrary direction, regardless of the relative position of the deployed panels with
respect to the CubeSat body.

71
4. Inter-Satellite Links for LEO CubeSats

(a) (b) (c)

Figure 4.2. High-gain CubeSat antennas designed at JPL. (a) Integrated Solar Array and
Reflectarray Antenna (ISARA) [13]. (b) Mars Cube One (MarCO) reflectarray antenna [7].
(c) Deployable Ka-band mesh reflector antenna [11].

4.2. ISL Communication Scenarios

Depending on the mission design, the configuration of the satellite constellation can vary, from
flying in clusters, to well-ordered formations. The most common formation types, depicted in
Figure 4.3, are [23]:

 Trailing – every satellite of the group flying in the same orbit with an equal relative
distance
 Cluster – satellites follow individual orbits which allow them to remain with a dense,
tightly spaced arrangement
 Constellation – a set of satellites organized in several orbital planes, that cover the entire
Earth, with sufficient overlap in ground coverage

(a) (b) (c)

Figure 4.3. Common types of satellite flying formations. (a) Trailing. (b) Cluster.
(c) Constellation.

72
 4.2. ISL Communication Scenarios

Depending on the selected flying formation, the ISL antenna design requirements can vary
drastically. In an ideal case, each spacecraft should have a number of dedicated high-gain
scanned-beam antennas for communication with several neighboring satellites. Such solutions
have been implemented in large-scale satellites, such as the Iridium constellation. This can be
afforded since the total mass of every Iridium satellite is over 600 kg. The dimension and weight
limitations of CubeSats, however, require simpler ISL designs in terms of mechanical
requirements and total volume.

In order to conform to the specifications without drastically increasing the complexity and cost,
the antenna design must be suited for the specific flying formation. In cluster formations, as the
spacecraft relative position and orientation is not defined, omnidirectional antennas are
preferred, which are mechanically simpler, at the price of a modest data rate. Trailing and
constellation formations do not necessarily require a scanned beam, if the satellite attitude can
be maintained with precision higher than the antenna beamwidth. These constellations can
consequently profit from cost-efficient high-performance fixed-beam antennas, without the
need for a complex scanning scheme (electrical or mechanical).

In this thesis, a constellation formation is considered, where N CubeSats are distributed along a
single orbital plane, and the entire system is composed of M such orbital planes, for a total of
N x M satellites. It is also considered that at least one CubeSat from each orbital plane is in
contact with a ground station, and there is no need for a link between CubeSats from different
orbital planes. Therefore, the CubeSats of a single orbital plane communicate in a ring network,
where each satellite establishes a link only with two adjacent satellites, as shown in Figure 4.4.
The CubeSats are required to communicate in both directions along the ring. Two scenarios are
possible:

 Single frequency – All CubeSats communicate at a single ISL frequency. Two-way

communication is performed in a form of time-division multiplexing (TDM).
 Dual frequency – Every ISL is two-way and consists of two channels, forward and return,
at separate frequencies.

Figure 4.4. ISL communication scenarios.

73
4. Inter-Satellite Links for LEO CubeSats

A single communication channel simplifies the system design of the first scenario, but limits the
total system capacity, as the communication can be performed only in one direction at a time.
The second scenario offers a continuous two-way communication. However, the dual-band
links complicate the ISL system design, primarily its RF part, as it will be shown in the following
sections.

4.3. ISL System Requirements

The key-enabling component of an ISL system is the antenna. Being a subject of strict
mechanical limitations and EM requirements, the antenna is challenging to design. The
antenna geometry mostly depends on the RF specifications, which are determined by the link
budget calculations for the ISL system. In the case of the considered CubeSat constellation, the
main requirements are a high-gain fixed-beam radiation pattern and the antenna bandwidth
that covers the frequency band(s) of interest. The following section discusses different aspects
of the ISL requirements.

4.3.1. Frequency
An individual CubeSat, which is a part of a constellation with a global coverage, will require
collecting and handling a large amount of data. The data is gathered, for example, by Earth
observation, tracking or communicating with a large number of devices, etc. The ISL is the
bottleneck of the constellation network, since all the collective data is aggregated and sent
through it. Therefore, the capacity requirement of the ISL is several times that of any individual
channels for data uplink/downlink. As an excellent example, the satellites of the Iridium
constellation in Figure 4.1 (not CubeSats, but a valid example) provide voice and data services
in L band and ISL in Ka band [78]. The millimeter-wave ISL naturally provides a high-capacity
connection for the aggregated user data, in real time.

The ISL frequency bands are sparsely allocated below 40 GHz, only in S, K and Ka bands [24].
The bands above 40 GHz require sophisticated RF technology that is not mature enough for
commercial CubeSat applications. The choice of ISL frequencies is therefore limited to either S
or K/Ka bands. The tradeoff between these two bands is presented in Table 4.1. Although the
propagation loss is lower at S band and the antenna-gain requirement is not as stringent as for
the Ka band, the antenna dimensions (for instance, a deployable helix) would still have to be
significantly larger than the Ka-band counterpart. Furthermore, the S-band link capacity could
be insufficient to support the ISL data rate. Thus, K/Ka bands are selected as optimal for the
high-data-rate ISL communications in a CubeSat constellation.

74
 4.3. ISL System Requirements

Table 4.1. Tradeoff between available frequency bands for ISL.

S band K/Ka band

Propagation loss + Small – Large
Antenna size – Very large – Large
Data rate – Low/Medium + High

20 22 24 26 28 30 32 34
Frequency [GHz]
Figure 4.5. A visual representation of the frequencies allocated for ISL applications in K/Ka
bands.

Several individual ISL bands are allocated at K/Ka-band frequencies, and visually represented
in Figure 4.5 [24]. A large variation of the frequency bandwidths (from 1.2% to 8.5%) may require
different antenna solutions, depending on the selected band. Some of the wider bands are
composed of narrower segments, and it is expected that only a subset of frequency segments
will be eventually used. However, the exact center frequency within a band may vary over time,
and the goal is to cover as many frequencies of a given band as possible.

4.3.2. Antenna Radiation

The radiation requirements for the ISL antenna are derived from the CubeSat mission
parameters. The CubeSat configuration related to the described process is depicted in
Figure 4.6.

1. After the orbital altitude is specified, low-frequency user antennas (L band in this case)
are designed with sufficient beamwidth for the required coverage of a single CubeSat.
2. The total number of CubeSats per orbital plane N is determined from the full-coverage
condition in a single plane. The number of orbital planes is found in a similar manner.
3. For a given number of CubeSats, the ISL beam tilt, away from the orbit tangent, is fixed
and is calculated simply by:

75
4. Inter-Satellite Links for LEO CubeSats

1 2
. (4.1)
2 2
For practical LEO applications, the beam-tilt angle is always smaller than 30° and the
associated scanning loss smaller than ~0.6 dB.
4. The inter-satellite distance is obtained by:

2 (4.2)
2 1 cos

where RE and h are the Earth radius and the orbit altitude, respectively. The path loss is
calculated at the ISL frequency.
5. The antenna-gain requirement is found using the link budget calculations (path loss,
data rate, modulation, coding, etc.). The preliminary value of 30 dBi was obtained for
eight satellites per orbit, with a minimum bit rate of 0.5 Mb/s using the QPSK
modulation.
6. The ISL antenna beamwidth is estimated from the antenna gain and compared against
the angular stability of the CubeSat’s ADCS.

The satellite attitude must be maintain by the ADCS, so that the high-gain ISL antennas
accurately point towards the adjacent CubeSats. If the angular precision of this system is smaller
than the ISL antenna beamwidth, the link cannot be constantly maintained and one of the
design decisions must be revisited. Shaped-beam radiation patterns (for example, a flattop
beam) are an interesting option to tackle a poor-performing ADCS, at the cost of a reduced
maximum gain. This antenna functionality is not considered in the scope of this thesis.

4.3.3. Antenna Size

The small wavelength at K/Ka-band frequencies provides a large freedom in the antenna
design. The design process, in this and following chapters, focuses on the case of LEO CubeSats,
and, more specifically, 3U CubeSats having their long axis aligned with the orbital velocity
vector. In other words, the small CubeSat face (1U) is facing the adjacent satellites within the
same orbital plane. This orbital attitude allows several TT&C and payload antennas to be
accommodated on the large CubeSat face and oriented towards the Earth.

A natural approach in this case is to allocate the two small faces for two low-profile ISL antennas.
However, by calculating the achievable directivity with an aperture of that area, having a
realistic antenna efficiency, it can quickly be seen that an antenna, with a total area equal to that
of a 1U face, cannot provide the specified antenna gain.

An ideal, uniformly illuminated antenna aperture provides a maximal antenna directivity. In

practice, the electric field across the aperture is non-uniform, and the loss related to this effect
is described by the aperture efficiency , with typical values between 0.4 and 0.7. A theoretical
antenna directivity, versus the aperture efficiency, is calculated by:

76
 4.3. ISL System Requirements

4
10 log (4.3)

where A is the total aperture area. The aperture efficiency in the previous expression does not
include the dielectric and ohmic losses of the antenna (the antenna efficiency is described in
detail in Chapter 6). The directivity is calculated for different efficiency and area values, and
represented graphically in Figure 4.7. Two points should be noted when inspecting this figure.
Firstly, it only shows theoretical values of directivity, not including any losses related to antenna
mismatch, polarization, dielectric or ohmic losses, etc. Secondly, the directivity is calculated at
broadside, and the results should be multiplied by a cosine factor for a tilted beam. In other
words, the final antenna size must be selected with an additional margin, to account for the
mentioned losses.

Figure 4.6. The geometry of a single orbital plane in a constellation, used to determine the
design requirements of the ISL antenna system.

36

34

32

30

28
Aperture
26 1U
2U
24 3U
4U
22
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Aperture efficiency

Figure 4.7. Theoretical antenna directivity at broadside versus the aperture efficiency, for
different antenna apertures in terms of CubeSat Units (U), at 24.6 GHz.

77
4. Inter-Satellite Links for LEO CubeSats

4.4. Reflectarray and Transmitarray Antennas

It can be seen that an antenna surface larger than 3U is required to achieve the specified gain.
This area is not available on the small CubeSat face, and therefore, the antenna will have to
incorporate some sort of a deployable mechanism to increase its effective area. Regardless of
the geometry, the entire antenna must be stowed within the 3U CubeSat volume during launch
in orbit.

Having any microwave components (amplifiers, feeding networks) on the deployed part of the
antenna would require a microwave transmission line between it and the satellite body, such as
an RF hinge or a flexible coaxial cable. These solutions were discarded early on, due to
associated risks. Therefore, the deployable sections should preferably contain only non-
connected components. Two well-known high-gain candidates are reflector and lens antennas.
Both geometries require a feeding antenna in the focal point of the reflector/lens, which can be
placed on the CubeSat body, while the focusing mechanism is deployed.

The conventional reflector and lens antennas come in various forms, usually being bulky and/or
heavy (dielectric lenses). Several techniques exist for the reduction of their total profile. The
most efficient approach is the design of reflectarray (RA) [81] and transmitarray (TA)
antennas [82], which are essentially planar reflector and lens antennas, respectively. The
operating principle of RA and TA antennas is illustrated in Figure 4.8. A RA antenna creates a
reflection phase gradient over the flat antenna surface, modeling the curvature of the
conventional parabolic reflector surface. A TA antenna uses a similar approach to model the
refraction in dielectric lenses. The phase gradient is created with an array of resonant elements,
typically by changing one or several geometrical parameters of the array element, and using a
gradual change in the reflection/transmission phase around the resonant frequency.

RA and TA antennas have a focal point, at which the feeding antenna is placed. The feed has a
low gain, and it is typically a patch antenna (array) or a small horn antenna. The RA and TA
antennas can be designed for an arbitrary position of the feed and the main-beam direction,
which is, besides their low profile, their biggest advantage.

The individual array elements of RA and TA antennas are standalone – not connected through
a feeding network, as in corporate-fed arrays, for instance. Therefore, the number and
distribution of elements are also arbitrary, as long as they follow a specified distribution lattice.
Consequently, the RA and TA antennas can have any shape and size, necessary to achieve the
required antenna gain. This fact enables an interesting mechanical advantage – the antenna can
be made of several separate panels, over which the array elements are distributed. During the
launch, the panels can be stowed on the CubeSat’s surface, similarly as the solar panels, and
deployed once the CubeSat is in orbit. Some examples are the RA antennas in Figure 4.2.

78
 4.4. Reflectarray and Transmitarray Antennas

Figure 4.8. Operation principle of reflectarray (RA) and transmitarray (TA) antennas.
Reflection and transmission phase gradient of RA and TA flat panels models the focusing
effect of the parabolic reflector and dielectric lens, respectively. Feed position and main beam
direction of RA and TA antennas are arbitrary. (Upper-right) Examples of RA and TA elements
presented in Chapter 5.

4.4.1. Antenna Configurations for CubeSat ISL

Two ISL antennas are required on every CubeSat inside the constellation, for a communication
with the two adjacent satellites in the orbital plane, as illustrated in Figure 4.6. If the antennas
are in the form of a RA and/or TA, there are various scenarios how the arrays and feeds can be
arranged on a 3U CubeSat. Several configurations are conceptualized here, and presented in
Figure 4.9, both for single- and dual-band ISL communication scenarios from Figure 4.4. The
presented configurations are drawn in the plane of the orbit, although other options are
possible. Beside the CubeSat orientation, the choice of the final scenario also depends on the
location of other deployable CubeSat features, primarily solar-panel wings.

It is straightforward to obtain single-frequency configurations from Figure 4.9(a-d) simply by

removing the second frequency in the sketches. If a dual-band performance is required, the RA
and TA panels (and their array elements) must operate independently at two frequencies. The
dual-band element design depends greatly on the frequency spacing and bandwidth of the
individual forward and return ISL channels. This can be achieved with either a single dual-band
or wideband element, or by interleaving two lattices of single-band elements.

79
4. Inter-Satellite Links for LEO CubeSats

It is equally important that the feeding antenna operate at both frequencies. Interleaved single-
band elements provide a benefit of having independent operation at the two frequencies. In this
case, two single-band feeding antennas, simpler to design and construct than a dual-band feed,
can be placed in different locations on the CubeSat body. Each of the two sub-arrays is then
designed for focal point at the location of the corresponding feeder, and the two beams are
radiated in the same direction. Otherwise, a single wideband feeder can be used for a wideband-
element scenario.

Any panel, deployed towards the Earth, will partially obstruct the Earth-facing low-frequency
antennas. An RA antenna can be made transparent at low frequencies for this purpose, if the
ground plane is replaced with a frequency-selective surface (FSS) [83]. This scenario is depicted
in Figure 4.9(c). The FSS behind the RA elements acts as a ground plane for the frequency of RA
operation, and transmits other frequencies, including the L and S frequency bands.

4.4.2. A Unified TA/RA Antenna

The last example of Figure 4.9 is a unified transmitarray/reflectarray antenna, shown in
Figure 4.9(d). This exotic configuration consists of a single panel, which performs the functions
of both the RA and TA at two separate frequencies. The FSS layer provides reflection at the RA
frequency and transmission at the TA frequency. It could potentially be omitted if the properly
designed TA element had the ability to provide a reflection at RA frequencies.

The main difficulty in this configuration is the proper handling of the mutual coupling between
RA, FSS and TA elements. The standalone performance of individual layers is drastically
deteriorated when the layers are placed together. In order to reduce the coupling, the total
thickness of the structure would have to be increased to the point where the antenna is not
suitable for the stowage on a CubeSat exterior. The only solution is to account for the coupling
in the design phase, performing a co-design of the three layers simultaneously and thus
significantly complicating the design process.

One example of this structure, with a significant profile and only for LP operation, is previously
proposed in [84]. A design based on a TA/FSS combination provides the desired function, again,
with a large panel thickness. In the scope of this thesis, many attempts were made to design an
array element, without arriving to a solution that satisfied the design requirements. The most
promising geometry, shown in Figure 4.10, is based on the dual-band TA design from [85].

The proposed TARA concept enables independent scanning of the transmitted and reflected
beams. However, if the satellites are uniformly distributed within a single orbital plane (as is the
case here), and if the array aperture is perpendicular to the orbit, the angles for both scanning
directions are the same. Under these conditions, a newly proposed class of antennas, called
transmit-reflect-array (TRA) antennas, is appealing for ISL applications. The TRA is designed as
a single metallic layer, and it provides a bidirectional CP high gain at the same frequency [86].

80
 4.4. Reflectarray and Transmitarray Antennas

A reconfigurable version of the same antenna incorporates PIN diodes for active beam steering,
simultaneously in the two directions [87].

(a) (b)

(c) (d)
Figure 4.9. Reflectarray and transmitarray configurations for ISL between 3U CubeSats.
(a) Two RA antennas. (b) Two TA antennas. (c) TA and RA antenna. The RA antenna is
potentially backed with a FSS. (d) The TARA concept: unified TA and RA antennas, separated
by a FSS.

Figure 4.10. Transmitarray/reflectarray (TARA) antenna concept. (left) Array layout.

(center) Individual elements. (right) Principle of operation.

81
4. Inter-Satellite Links for LEO CubeSats

4.5. ISL Antenna Design Requirements

It was seen that an antenna aperture larger than 3U is necessary at 24.6 GHz to obtain a gain
larger than 30 dBi with a sufficient margin. RA and TA antennas are selected as excellent
candidates due to their ability to provide the required antenna aperture through several non-
connected passive panels. A design of one RA and one TA antenna is presented in this thesis, for
the frequency band II from Figure 4.5. However, the geometry should be scalable and the
antenna geometries should ideally be able to provide a satisfactory CP performance over any of
the other ISL bands, with more demanding bandwidth requirements.

The design requirements for both RA and TA ISL antennas are presented in The main design
goal of the RA element in the following chapters is to improve the CP bandwidth of the RA
element, without compromising the complexity. The element should be designed on a single
dielectric layer, without metallized vias or SMD components. Among several presented
geometries in Chapter 5, one novel RA element satisfies these requirements.

Table 4.2. Beside the given requirements, the antennas should also have a profile low enough,
so that the combined thickness of all stowed panels does not violate the CubeSat limitation of
6.5 mm for features on the chassis exterior. Ideally, the thickness should not exceed 70-80% of
the limitation.

If the array panel is deployed perpendicularly to the CubeSat’s long axis (and the orbit tangent),
the angle between the main beam and the tangent is at the same time the antenna beam-
scanning angle away from broadside. For practical applications, this angle is smaller than 30°
and the associated scanning loss smaller than ~0.6 dB, as seen in Figure 4.11. Depending on the
selected ISL antenna configuration and the relative position of the feed, the deployment angle
with respect to the CubeSat body can differ from 90° to alleviate the scanning loss.

4.5.1. RA Antenna
Conventional reflector and lens antennas have a theoretically infinite bandwidth, as they rely
on the principles of geometrical optics. The resonant nature of RA and TA antenna elements
limits their bandwidth [81]. Furthermore, maintaining good CP performance over the entire
reflection or transmission bandwidth is even more challenging.

The bandwidth limitation is more severe in RA antennas, compared to TA antennas. Since the
introduction of the RA antennas, many techniques, having different levels of complexity, have
been developed to tackle this limitation [81], [88]. The key step in the RA antenna design is the
design of the RA element itself. In the literature, the RA element bandwidth is often improved at
a cost of an increased structural complexity, typically by introducing several metallic and/or
dielectric layers. Such approaches violate the CubeSat requirements, where a simple, low profile
and robust solution is preferred.

82
 4.5. ISL Antenna Design Requirements

The main design goal of the RA element in the following chapters is to improve the CP
bandwidth of the RA element, without compromising the complexity. The element should be
designed on a single dielectric layer, without metallized vias or SMD components. Among
several presented geometries in Chapter 5, one novel RA element satisfies these requirements.

Table 4.2. ISL antenna design requirements.

Parameter Value
Frequency [GHz] 24.45 – 24.75
Polarization CP
Gain [dBi] > 30
S11 [dB] < -15
AR [dB] <3

Figure 4.11. ISL antenna beam-tilt angle with respect to the orbital tangent (see Figure 4.6)
and the associated scanning loss, versus the number of CubeSats in a single orbital plane.

4.5.2. TA Antenna
The bandwidth limitation is not as critical in conventional TA elements, compared to the RA
antennas. A wideband performance of TA antennas typically comes from their large overall
thickness, which can reach a full wavelength [89]. Needless to say, the resulting panel thickness
would be incompatible with the CubeSat standard. The goal in this case is to minimize the
overall thickness of a TA element (and the antenna) while preserving the CP bandwidth.

Among the TA-element geometries, previously presented in the literature, one proposed
configuration is based on two patch elements on opposite sides of the ground plane, connected
with a metallized via [90]. Although the total element thickness is very low, this geometry yields

83
4. Inter-Satellite Links for LEO CubeSats

a wide CP bandwidth. Promising results are also achieved using two pairs of stacked patches
and EM coupling through a slot in the ground plane [91]. In Chapter 5, a novel TA element is
proposed, which combines the best aspects of both previous designs – an aperture-coupled
geometry (without metallized vias) with a low profile and only two patch elements.

4.6. Summary
Inter-satellite links (ISL) are a key feature for CubeSat constellations that aim to provide global
coverage and a very low communication latency. Conventional ISL antennas for small satellites
are not suitable for CubeSats, due to the size and weight limitations. A general approach of
simplifying the antenna geometry is to adapt the design to a specific CubeSat flying formation.

Among several flying formations, a satellite constellation is considered in this thesis. In a

constellation formation, CubeSats are uniformly distributed over multiple orbital planes, having
identical relative distance and orientation within a single plane. High-gain fixed-beam antennas
are a good cost-efficient candidate for ISLs in this configuration, since the two adjacent satellites
see each other at a same angle, at all times. The CubeSat’s attitude control system must have an
accuracy higher than that of an ISL antenna beamwidth, for the link to be maintained without
interruptions.

An ISL in K/Ka band was selected, among the sparse bands allocated for ISL applications below
40 GHz, due to the large capacity at those frequencies. Each 3U CubeSat of the constellation is
flying with its long axis tangential to the orbit and facing the adjacent CubeSats with the smallest
face. Based on the link-budget calculations, the 1U surface cannot provide sufficient gain for a
required data rate. Therefore, a deployable antenna is necessary, and deployable reflectarray
and transmitarray antennas are found to be optimal for this application. The nature of the RA
and TA antennas allows the reflector/lens to be constructed out of several separate panels. The
panels can then be neatly stowed or wrapped at the CubeSat exterior and no electrical or RF
connection is required with the main satellite body.

Several RA and TA configurations for ISL applications are envisioned and depicted in this
chapter. The configurations differ in the relative position of deployable panel and the feeding
antenna, with respect to the chassis. Although they are adapted to the specific CubeSat attitude
described in this thesis, they can also be suitable for CubeSats having different form factors or
in-orbit attitudes.

The RA and TA antennas consist of an array of resonant elements, and one or several
geometrical parameters are tuned to vary the phase of the reflected/transmitted signal.
Different techniques were previously proposed to improve the bandwidth of the inherently
resonant RA elements, usually compromising the overall simplicity. On the other hand, the
existing TA elements have a good bandwidth performance, but a large total thickness. Chapter 5

84
 4.6. Summary

proposes one novel element for each of the RA and TA CubeSat antennas: a single-layer
wideband RA element, and a low-profile TA element. The chapter also presents the design and
measurement results of an axially corrugated horn antenna used as the feeding element for both
arrays. The presented RA and TA element geometries, designed for the CubeSat ISL
applications, can be equally interesting for other applications and different frequency bands.

Although the RA and TA antennas seem to serve completely opposite purposes, the essence of
their design process is identical, safe for the design of the array element itself. Chapter 6
describes the RA and TA antenna design process in detail, based on the simulation results of
individual array elements and the feeding antenna from Chapter 5. The simulation and
measurement results of several manufactured prototypes are finally presented in Chapter 6.

85
5. Reflectarray and Transmitarray
Components

Both RA and TA antennas belong to a group of phase-shifting surfaces (PSS). Although many
different PSS design techniques and geometries exist, the focus of this thesis is on PSS arrays of
discrete elements. A detailed design process of such geometries was previously presented, for
example, in [81], [82], [88]. Individual array elements within most PSS arrays have a similar
topology, and the phase-shifting property is achieved by changing one or several design
parameters of the array element.

A CP radiation is used in space communications to eliminate the losses related to LP

misalignment. For the same reason, it will be used for the design of RA and TA antennas for
CubeSat ISLs. In the CP case, both RA and TA elements can benefit from a design technique
known as the element-rotation technique [92]. The geometry of elements, designed with this
technique, is identical in the entire array, and the phase shifting is done by a simple rotation
around the element’s vertical axis. The rotation angle of individual elements is then proportional
to the phase of the reflected/transmitted waves, depending on the antenna type (RA or TA). A set
of conditions that an array element and the feeding antenna must satisfy and a detailed theoretical
analysis are presented in Section 5.1.2.

The antenna arrays are intended for operation in one of the ISL frequency bands allocated in
K/Ka-band. The antenna design requirements are given in Table 4.2. Both the array elements,
whether RA or TA, and the feeding antenna should have an excellent CP performance across the
entire frequency band. Ideally, the array elements and the feed should also be scalable to the ISL
K/Ka-bands other than the specified one, where the bandwidth requirement is more stringent.

The simulation environment of an array unit-cell and the associated assumptions are presented
in the beginning of this chapter. Following a theoretical description of the element-rotation
technique, two novel RA and TA elements are proposed for CubeSat ISL applications in K band,
based on coupled microstrip loops and aperture-coupled microstrip patches. As discussed in
Chapter 5, the design goal of the RA antenna element is the improvement of the CP bandwidth of
5. Reflectarray and Transmitarray Components

existing geometries without compromising the array thickness or complexity. In the case of TA
elements, the goal is opposite – a thickness reduction with a preserved aperture efficiency.

Finally, a 3D-printed all-metal corrugated-horn feed chain is designed, as a feed for RA and TA
antennas. The CP operation of the feed, required for the element-rotation technique, is achieved
with a septum polarizer in a square WG. Simulation and measurement results of the horn feed
demonstrate its suitability for CubeSat ISL applications.

5.1. Element Design Approach

The fundamental part of the general design process is the design of the array element itself. The
elements of the array are distributed over a square lattice. A common approach is to assume local
periodicity of elements and analyze the element in an infinite-array environment [81]. In other
words, each element of the array is surrounded by elements typically having only slight
differences in geometry. The approximation is less applicable for elements close to the edge of the
array. However, the amplitudes of the waves, incident at these elements, are also smaller, so that
the introduced error is also reduced. This approximation was shown to give excellent results in
RA and TA antenna design for many different element geometries[81], [82]. The approach is
equally suitable for full-3D and quasi-3D EM solvers. It is favorable to use the quasi-3D solver for
the planar geometries presented here, as it requires significantly less computing resources.

5.1.1. Infinite-Array Approximation

In an infinite-array approximation, the simulation relies on the application of Floquet’s theorem
on infinite periodic structures [81], [93]. It states that the field in the infinite periodic structure can
be expressed as a superposition of travelling waves, called Floquet’s spatial harmonics. In the case
of a 2D periodic structure, the total field in two adjacent cells of the periodic structure differs only
by a propagation factor of a plane wave. The theorem then becomes:

(5.1)

where E is the electric field component, r is the element position vector, R is the lattice vector, and
k is the propagation vector of a plane wave. It is shown that, if the element spacing is chosen to
prevent the appearance of grating lobes in the visible space, only one pair of harmonics is
supported. The two harmonics have identical propagation constants and orthogonal
polarizations, and they can be used for a full characterization of the periodic structure, including
the phase response, losses and cross-polarization.

It directly follows from the previous analysis, that it is sufficient to analyze a single element of the
periodic structure (in this case, the array), called the unit cell. The simulation environment of RA

88
 5.1. Element Design Approach

and TA unit cells is shown in Figure 5.1. The element is surrounded on lateral sides by periodic
symmetry planes. The field amplitudes on opposing symmetry planes within each pair are
imposed by the solver to be equal, allowing for a phase shift. The two independent phase shifts of
the cell determine the propagation vector of a supported plane wave, given by:

cos sin

sin sin (5.2)

cos

where and are angles in the spherical coordinate system, and 2 / . Boundary
conditions at the remaining cell faces, defined as Floquet ports, are used for excitation of the unit
cell. A modal decomposition of the fields at Floquet ports is performed, where the two
fundamental supported modes correspond to incoming LP plane waves with orthogonal
polarizations, referred to as TE and TM modes. The relations between all supported (incident and
scattered) modes is presented in the form of S parameters. Therefore, the order of the resulting S
matrix is four and two in the TA and RA case, respectively. Two Floquet ports are defined for TA
cells in order to calculate the transmission as well as reflection coefficients of the cell. In the case
of a RA, the bottom port is usually replaced with a PEC plane. If the RA is used in a combination
with a FSS, a setup similar to the TA cell should be used.

5.1.2. Variable Rotation Technique

The reflection phase shift of most RA elements is obtained by modifying one or several element’s
geometrical parameters. This way, the resonant frequency of the element is shifted around the
central operating frequency of the RA, simultaneously controlling the phase of the reflected signal.
The phase range that can be achieved with simple, single-resonance elements, such as a
microstrip patch, is smaller than 360°, meaning that some of the required phase values will be
impossible to achieve. Moreover, the size variations between adjacent elements can invalidate the
local periodicity assumption in the RA locations where the phase wrapping occurs [88].

A special technique is applicable for CP RA and TA antennas, referred to as the variable rotation
technique. The following conditions must be satisfied [91], [92]:

 The feeding antenna must radiate CP

 RA case:
o The CP sense must be the same for the incoming and scattered waves
 TA case:
o The CP sense must change while passing through the TA surface
o The reflection is minimal

89
5. Reflectarray and Transmitarray Components

Figure 5.1. The simulation environment of TA and RA antennas in a full-wave 3D EM solver.

If these conditions are fulfilled, the phase can be changed simply by rotating the elements around
the vertical axis. The range of attainable phase values is continuous from 0° to 360°, while the
amplitude of reflection/transmission coefficients and the coupling to adjacent elements remains
nearly constant across this range.

An excellent comparison between two types of TA elements – phase-delay and phase-rotation

elements – is presented in [94]. The theoretical development of the element-rotation technique,
for RA and TA antennas, is presented in [91], [92], [94], [95], and the main points are outlined in
this chapter for the more general case of CP TA antennas. The equations for the RA are derived
afterwards from the TA case.

The S-parameter matrix of the unit cell from Figure 5.1 is defined as:

(5.3)

where and and the matrices of scattered and incident TE/TM modes. Its expanded form
is:

(5.4)

If the element is rotated by an angle , the new matrix can be written as:
cos sin 0 0
sin cos 0 0
, , (5.5)
0 0 cos sin
0 0 sin cos
where is the rotation matrix. As the incident and scattered radiation is always CP, it is
convenient to work with circular unit vectors and define the matrix as:

90
 5.1. Element Design Approach

. (5.6)

The matrix relates incident and scattered CP waves, while also taking into account the
variable rotation angle. It can be obtained by:
1 0 0
1 0 0
, , (5.7)
0 0 1
0 0 1
where is the transformation matrix from Cartesian (x, y) to circular (RHCP, LHCP) unit
vectors. Expression (5.6) can also be written as:

(5.8)

where, for example, is the incident LHCP wave at port 2, is the scattered RHCP wave at
port 1, and is the matrix entry that relates the two mentioned waves. Since the feed antenna is
always placed on the top of the array, all signals that are incident at port 2 are of no importance.
The relevant matrix entries are obtained by substituting (5.5) in (5.7) and comparing with (5.8):
1
2
1
2
1
2 (5.9)
1
2
1
2
1
.
2
It directly follows that, if the element satisfies the previously outlined conditions for the TA case,
the phase shift is then equal to twice the value of the rotation angle and the CP sense of the
scattered wave is of the opposite hand compared to the incident wave [92]. The CP-sense change
is achieved if the phase difference between the two co-polarized LP components is 180°, or
equivalently:

arg arg 180°, (5.10)

By properly selecting the rotation angles across the TA surface, the resulting phase distribution
can collimate the radiated beam and scan it in an arbitrary direction.

91
5. Reflectarray and Transmitarray Components

The response of a RA cell is obtained simply by using the expressions of (5.9) with a subscript ‘11’,
which take into account exclusively port 1. Similar conclusions apply as in the TA case, except that
the reflection of the same CP sense is the desired effect in CP RA antennas.

A notable example in the group of RA elements, which satisfies the described conditions, is the
microstrip split loop [96], [97]. Different stacked and coplanar configurations of split loops were
reported for dual band performance, sometimes in a combination with FSS if transparency was
required at lower frequencies [83], [98]. On the other hand, TA examples of CP elements include
an aperture-coupled stacked patches [91], stacked split loops [85], and various other, not listed
here [99]. The existing elements served as a starting point for the RA and TA element design. The
following sections present novel elements, developed from the existing ones, which significantly
improve the bandwidth or panel thickness of the RA and TA structures, respectively.

5.2. TA Element: CP Aperture-Coupled Patches

The general approach in one group of TA elements is to have two resonant elements at opposite
sides of a ground plane. Several element geometries are illustrated in Figure 5.2. The element
facing the feed accepts the incoming radiation, resonates and transfers the energy to its
counterpart, which in turn resonates and radiates the waves on the opposite side of the TA.

The first TA lens based on this concept is shown in Figure 5.2(a) [100]. Two parallel arrays of
rectangular patches are connected with delay lines and metallized vias, where the variable line
length provides the required phase shift. The design from Figure 5.2(b) eliminates the delay lines
and uses only a metallic pin to connect the receiving LP patch to a transmitting CP patch, in
combination with the element-rotation technique [90].

Another work presented a design shown in Figure 5.2(c) where the patches are aperture-coupled
to stripline delay lines, removing the need for unreliable metallized vias [101]. The geometry was
later improved further, by eliminating the stripline delay lines and directly coupling the patches
through a crossed slot in the common ground plane.

Figure 5.2. A comparison of different methods to achieve phase shifting in TA elements based
on microstrip patches. (a) Delay line and via connection. (b) Via connection, variable rotation.
(c) Stripline delay line and aperture coupling. (d) Aperture coupling of stacked patches, variable
rotation. (e) Aperture coupling of single patches, variable rotation (this thesis).

92
 5.2. TA Element: CP Aperture-Coupled Patches

This structure is proposed for dual-LP [102]and CP operation [91] and the element geometry is
shown in Figure 5.2(d). The transmission phase is varied by changing the patch physical
dimensions in the dual-LP case, and by using the element-rotation technique in the CP case. A
stacked-patch structure is used in both designs, due to difficulties to obtain satisfactory
transmission parameters with only a single patch element.

As discussed in Section 5.1.2, the element-rotation technique requires the phase difference
between the transmitted LP components to be 180°. This difference is achieved in [91] with a
rectangular patch shape. A large aspect ratio of the rectangle separates the two orthogonal
resonances in frequency and increases the phase difference of transmitted components. The
authors stated that it was difficult to obtain the required difference of 180° with a single patch, and
that the stacked patches were used for this purpose. The resulting structure has a wide CP
transmission bandwidth, but also a large profile, as it is composed of four dielectric and five
metallic layers.

The remaining part of this section presents a TA element shown in Figure 5.2(e) having only two
dielectric layers, which achieves the 180° phase difference using a modified shape of the coupling
aperture.

5.2.1. Geometry Evolution

Introducing asymmetries in conventional patch antennas is commonly used to excite CP by
creating a phase difference of 90° between orthogonal resonances. This was done in Section 2.7
by introducing a small asymmetry in the coupling slot of an aperture coupled patch antenna. The
same idea is used here for the described TA element. Since the required phase difference is twice
that of the patch antenna, the slot asymmetry must be larger, in combination with rectangular
instead of square patches. The evolution of the TA-element geometry is shown in Figure 5.3. One
slot in the pair of crossed slots is drastically elongated at first, and then loaded at the ends in the
form of a dog-bone slot [103]. It is shown that a single patch on each side is sufficient to achieve a
phase difference of 180° with this geometry.

Indeed, the transmission polarization-sense change and minimal reflection are found to be
conflicting criteria in the design of the aperture-coupled element. In other words, if the reflection
is minimized at the same frequency for both LP components, the difference in transmission phase
tends to be smaller than the required 180°. A phase difference close to 180° is achieved with a
significant slot and patch asymmetries, while maintaining low values of reflected LP components.
After the initial manual tuning, the element was optimized to minimize both reflected and cross-
polarized transmitted CP components at the design frequency. The dimensions of the optimized
element given in Figure 5.4 present the best compromise.

93
5. Reflectarray and Transmitarray Components

Figure 5.3. Design evolution of aperture-coupled patches as TA antenna elements. Two

identical patch elements are placed symmetrically on the two opposite sides of a ground plane
with a slot. The slots are shown in white on the gray ground plane.

Figure 5.4. The geometry of the TA element based on aperture-coupled patches. Dimensions (in
mm): d=6, h=0.76, Px=3.19, Py=1.87, Sx=2.63, Sy=2.99, Sa=2.28.

5.2.2. Simulation Results

The simulated performance of the optimized element is shown in Figure 5.5. Although the
maxima of the LP transmission components do not appear at the same frequency and the phase
difference is not exactly equal to 180°, the unwanted CP components are all below -17 dB at
24.6 GHz.

When used as a part of an array, the incoming radiation arrives at individual elements at a range
of different incidence angles. This range is increased for TA antennas with a low F/D ratio (TA
focal distance over the total array dimension, as explained in the next section) and offset-fed TA
antennas. For a good array performance, the element must show stable characteristics over the
given angular range. The CP performance is calculated for three elevation incidence angles: 0°,
20°, and 40°, and the simulation results are presented in Figure 5.6. The response shifts to lower

94
 5.2. TA Element: CP Aperture-Coupled Patches

frequencies with the increase of the incidence angle. As a compromise, the element is designed at
a slightly higher frequency to account for this effect. Nevertheless, the CP cross-polarized and
reflected components remain below -15 dB at 24.6 GHz in all cases.

The TA element is designed on two Rogers RO6002 laminates (εr=2.94, tanδ=0.0012, h=0.76 mm).
The total thickness of the structure is ~0.13 λ0 at 24.6 GHz. The initial design was done on Rogers
RT/Duroid 5870 boards, yielding a total thickness of ~0.09 λ0 , but the substrate material was
changed due to material availability. By comparing the two models, a better performance-stability
versus the incident angle was observed in the Duroid 5870 element due to a lower profile. The
same conclusion can be made by comparing this element’s performance at large incidence angles
versus the stacked-patch element from [91].

(a) (b)

(c) (d)
Figure 5.5. Simulation results of the aperture-coupled-patch TA element. (a) Transmission and
reflection S parameters for the two orthogonal LP components – TE and TM. (b) Transmission
and reflection S parameters for the two CP components – LHCP and RHCP. (c) Phase and phase
difference of the LP transmission S parameters. (d) The co-polar transmission magnitude versus
the elevation incidence angle, for different azimuth incidence angles.

95
5. Reflectarray and Transmitarray Components

[dB]

[dB]
S LR

S LR
11

21
[dB]

[dB]
S RR

S RR
11

21

Figure 5.6. The dominant S parameters of the aperture-coupled-patches TA element versus the
elevation incidence angle.

5.3. RA Element: Coupled Loops

RA antennas with elements having variable rotation angles were first designed with square
microstrip patch element and attached delay lines [92], followed by the split-loop element [96].
Split loop remains the most attractive CP RA element to date. Its small conductive surface makes
it convenient to stack or interleave elements with different dimensions, for multiband RA
operation [98], [104]. Furthermore, the circular shape of the element minimizes the variations of
the mutual coupling for various rotation angles of adjacent elements. The split loop is used in
combination with a backing FSS structure to make the RA panel transparent to low
frequencies [83]. This configuration is especially appealing for the RA CubeSat configuration
previously illustrated in Figure 4.9(c).

The design goal of this section is to improve the split-loop element bandwidth, starting from the
existing geometries, for it to be able to cover any ISL frequency band in K/Ka bands. The
simulation setup for the RA unit cell is shown in Figure 5.1(b).

96
 5.3. RA Element: Coupled Loops

5.3.1. Geometry Evolution

RA element geometries and the corresponding simulated S parameters are shown in
Figure 5.7 – 5.9. The starting design for the wideband RA element is the known split-loop from
Figure 5.7(a) [96]. It is a metallic ring above a ground plane, cut in two positions on opposite sides.
In this structure, the phase of the reflection coefficient for vertically polarized waves is 180° larger
compared to the horizontally polarized, which is the key condition for the element-rotation
technique. The CP sense of the incident and scattered waves is thus the same. The underlying
principles of its narrowband CP performance will be explained in the following section.

The bottom-left element in Figure 5.7(a) is presented in [97], where it is claimed that the vertical
line is added to reduce the cross-polarization. The bottom-right element in Figure 5.7(a) is the
result of an attempt to increase the capacitive coupling at the edges of the two circular arcs.
Consequently, the element has a smaller dimension compared to the conventional split loop, for
a given resonant frequency, and a slightly smaller bandwidth. However, elements of this type can
be distributed with a smaller separation, potentially improving the bandwidth of the full array.
This element is novel and it can be considered as an intermediate design step that led to the
development of the final wideband geometry. The performance of both split-loop modifications
is found in the simulations to be very similar to a conventional split loop.

The RA element shown in Figure 5.8 is an evolution from the bottom-left element in Figure 5.7(a),
where the vertical line is extended to an I-shaped line. An interesting property of the element is its
dual-band performance. The angular positions of the gaps determine the ratio of the two central
frequencies. Unfortunately, the element cannot be used simultaneously for both bands, since the
required rotation angles will be different in the two frequencies, even if the main beam is scanned
in the same direction. The same conclusion can be made for the dual-band TA element from [99].
While trying to bring the two frequencies together, an anti-resonance was observed between the
two bands, preventing the two resonances to collapse into a single, wide band. This element
geometry could be of interest for a future study.

Although the attempts to generate a wideband geometry were unsuccessful for the element in
Figure 5.8, the final design yields two collapsed resonances with a potential for a significant
bandwidth increase. The RA element in Figure 5.9 is obtained by continuing the process, used to
generate the capacitive-coupled loop in Figure 5.7(a), until the two pairs of arc edges are
connected. The obtained geometry is composed of two coupled loops. When the element is re-
sized to the design frequency, it is larger than the original split loop. However, the total diameter
is equal to only 0.33 at the design frequency. The element-spacing condition that prevents the
excitation of grating lobes in a RA is given by:

1
(5.11)
1 |sin |
where is the element separation and is either the incidence angle from the feed or the beam-

97
5. Reflectarray and Transmitarray Components

tilt angle, whichever is larger [93]. Therefore, the excitation of grating lobes is eliminated by using
the described element in any feeding or scanning configuration, since the element-separation
limit is larger than 0.5 for all incidence angles.

The full-wave simulation results of the coupled loops are presented in Figure 5.9(b,c). A dual-
resonant behavior is evident from both the CP and LP S parameters. Consequently, the CP-
reflection bandwidth of the coupled loops is significantly improved compared to the traditional
split-loop element. The following Sections 5.3.2 and 5.3.3 provide a detailed modal analysis and
the full-wave simulation results of the coupled-loops element, for a deeper insight into the
wideband CP nature.

(a) (b) (c)

Figure 5.7. Geometry of split-loop RA elements and the full-wave simulated performance.
(a) The original element geometry and two modifications. The three geometries have a similar
single-resonant performance. (b) Magnitude of the CP S parameters. (c) Phase of the LP S
parameters.

(a) (b) (c)

Figure 5.8. Geometry of the I-shaped-loop RA element and the full-wave simulated
performance. (a) The element geometry. (b) Magnitude of the CP S parameters. (c) Phase of the
LP S parameters. Note the expanded frequency axis.

98
 5.3. RA Element: Coupled Loops

(a) (b) (c)

Figure 5.9. Geometry of the coupled-loops RA element and the full-wave simulated
performance. (a) The element geometry. (b) Magnitude of the CP S parameters. (c) Phase of the
LP S parameters.

5.3.2. Modal Analysis

The performance of different elements can be explained by examining the structural resonances
of the geometry, as well as the electric currents on the element when it is illuminated by LP and
CP plane waves. Different solvers are used for the two tasks – the eigenmode solver for the
structural resonances in the absence of any excitation and the modal-driven solver for the full-
wave simulation with plane-wave excitations. The periodic boundary conditions are used in both
cases. The only setup difference is that, in the eigenmode simulations, the Floquet port is replaced
with an open (radiation) boundary. The split loop and the coupled loops are used for comparison.

The first two resonant modes of the two geometries are calculated and the characteristic electric-
current distributions are shown in Figure 5.10. The two split-loop modes can be seen as the even
and odd modes and their eigen-frequencies are the same. On the contrary, the first two modes of
the coupled loops have different eigen-frequencies, with a spacing of ~7.6%. When the two RA
elements are excited by LP plane waves at the central frequency in the full-wave simulations, the
currents on the coupled loops correspond exactly to the two mode distributions. However, only
the first mode is excited on the split loop for both vertical and horizontal polarizations, as the
second mode is not supported. CP plane waves excite the same (first) mode, explaining the
narrowband character of the split-loop element.

The current distribution of the coupled loops in the case of a CP plane-wave excitation is a
combination of the two eigen-mode distributions. Individual modes are dominant at different
frequencies within the operating band, in agreement with the relationship of the mode eigen-
frequencies. The current distribution of the coupled loops is presented in Figure 5.11 at several
frequency points. The distribution is mainly determined by the first mode at lower frequencies,
and vice versa.

99
5. Reflectarray and Transmitarray Components

Split Loop Coupled Loops

Mode 1
Eigenmode

Mode 2

Vertical
polarization
Floquet-port excitation

Horizontal
polarization

Circular
polarization

Figure 5.10. Electric currents on the split-loop and coupled-loops RA elements. The top two
rows show the results of the eigenmode analysis, and the bottom three rows show the results of
a full-wave 3D simulation with a Floquet-port excitation. Periodic boundary conditions are used
in both sets of simulations.

100
 5.3. RA Element: Coupled Loops

5.3.3. Full-Wave Simulation Results

The simulated S parameters of the RA coupled-loops element, with a Floquet-port excitation in
the infinite-array environment, are shown in Figure 5.12. It is observed that the two resonances
separate in frequency, as the incidence angle of the incoming wave increases. Therefore, it is
essential for the element to be designed with closely spaced resonances in the case of broadside
incidence, in order to preserve a low cross polarization at the center of the band for large incidence
angles. On the other hand, if the resonances are designed too closely together, the overall element
bandwidth is reduced. The main parameter that influences the resonance separation is the gap
between the two loops. As seen in Figure 5.12(b), the S parameters are very sensitive to the gap
distance.

The coupled-loops element is modeled using zero-thickness conductor layers during the
simulations, including the conductive and dielectric losses in the materials. This approximation
significantly accelerates the quasi-3D simulation process. However, through prototype
measurements in Chapter 6, it will be shown that the effect of the finite conductor thickness is
important, since it also affects the sensitive coupling between the two loops. Therefore, it is
advised to include the finite conductor thickness in the simulation models, for maximum
accuracy.

The effect of resonance separation is also evident from Figure 5.13, where the co-polar reflection
S parameter is presented for different incidence angles. The magnitude of the co-polar reflection
drops most rapidly in the center of the band, at 24.6 GHz, which is in line with the results from
Figure 5.12(a). Nevertheless, a wideband performance of the coupled loops is preserved across a
wide range of incidence angles, with a low cross polarization across the entire band.

24.2 GHz 24.8 GHz 25.1 GHz

Figure 5.11. Electric-current distributions on the coupled-loop element at different frequencies.

The results are obtained using a full-wave 3D simulation and a Floquet-port excitation.

101
5. Reflectarray and Transmitarray Components

(a) (b)
Figure 5.12. Simulated S parameters of the coupled-loops RA element (Element 5 from
Figure 5.7). (a) Variation of the elevation incidence angle. (b) Variation of the gap between the
two D-shaped loops.

0 0 0

-1 -1 -1

-2 24 GHz -2 24.6 GHz -2 25 GHz

inc
= 0° inc
= 0° inc
= 0°
-3 -3 -3
inc
= 30° inc
= 30° inc
= 30°

inc
= 60° inc
= 60° inc
= 60°
-4 -4 -4
inc
= 90° inc
= 90° inc
= 90°

-5 -5 -5
0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60
inc
[°] inc
[°] inc
[°]
(a) (b) (c)
Figure 5.13. The co-polar reflection magnitude versus the elevation incidence angle, for
different azimuth incidence angles. (a) 24 GHz. (b) 24.6 GHz. (c) 25 GHz.

5.4. 3D-Printed Corrugated-Horn Feed Chain

A feeding antenna is an equally important part of the RA or TA system as the array element. Its
properties must be carefully chosen for the given array geometry. In the case of arrays that use the
element-rotation technique, the feeding antenna must be circularly polarized. Moreover, the CP
quality should be preserved along the entire range of angles, under which the array is visible. A
circular horn antenna is selected for this application, although other geometries can be equally
suitable (a patch antenna array, to name one). The choice was driven by a high operating
frequency and the fact that the horn antenna is an all-metal geometry – an attractive feature for
space applications.

102
 5.4. 3D-Printed Corrugated-Horn Feed Chain

5.4.1. Array Focal Distance

The F/D ratio is the main parameter that determines the properties of the feeding antenna. Some
antenna aspects are outlined in Table 5.1 and illustrated in Figure 5.14, versus the F/D value. The
beamwidth of the feed is inversely proportional to the F/D ratio, meaning that a more directive
antenna should be selected for a larger feed distance [81]. The incidence-angle estimate in the
table is given for center-fed antennas or the antennas where the feed projection is at the array
center. This issue becomes more important for offset-fed arrays (and especially for large offset
values) since the maximum incidence angle is increased and the central area of the array is
illuminated at a larger angle as well. The central array area is also the one with the strongest
illumination, and which plays the largest part in the total RA performance. A larger F/D will reduce
the maximum incidence angle in any case, but will require a more directive feed, which will also
be larger.

For the antenna geometries depicted in Figure 4.9, the array is deployed from the CubeSat surface,
and the feed is located on the large face of the CubeSat, with a distance from the array base.
Therefore, an offset value of 0.5D is adopted in this thesis, as the worst-case scenario, while having
in mind that the actual offset in the practical antenna model will be slightly smaller. Offset-fed
arrays are more prone to exciting grating lobes for element-spacing values larger than 0.5λ0. The
maximum allowed element spacing is calculated using (5.11) for different array sizes and F/D
values, and the results are presented in Table 5.2. The elements used in this thesis are designed
with an element spacing of approximately 0.5λ0 and thus never excite grating lobes.

Table 5.1. Design tradeoffs of RA and TA antennas, for different values of the F/D ratio.

F/D ratio Small Large

Feed beamwidth Wide Narrow
Maximum element
Large Medium
incidence angle
Feed size Small Large
Total RA antenna size Medium Large

Figure 5.14. The impact of the F/D value on the design parameters of the feed and the array: feed
beamwidth BW, maximum/average incidence angle , and feed horn diameter Dh.

103
5. Reflectarray and Transmitarray Components

Table 5.2. The maximum element separation (in λ0) that does not generate grating lobes, versus
the focal distance and the array dimension.
F [mm]
100 150 200 250
D [mm]
100 0.59 0.64 0.69 0.73
150 0.54 0.59 0.62 0.66
200 0.53 0.56 0.59 0.62
250 0.52 0.54 0.56 0.59

Table 5.3. The angle (in °) at which the deployed RA or TA aperture is visible from the feed
position on the surface of the CubeSat.
F [mm]
100 150 200 250
D [mm]
100 45 34 27 22
150 56 45 37 31
200 63 53 45 39
250 68 59 51 45
In the given array configuration on a CubeSat, the angle at which the array is visible from the feed,
given in Table 5.3, is calculated by:

atan . (5.12)

This angle gives an initial estimate of the feed’s 10-dB beamwidth. In the literature, maximum
aperture efficiency of the array is typically obtained if the illumination in the array edge is
approximately 10 dB below the center illumination [81]. A similar rule is adopted for the designed
feed, as explained later on.

5.4.2. Axially Corrugated Horn

A horn antenna is the most common choice for the feeding element of a reflector. An all-metal
horn geometry is an attractive feature for space applications. CP radiation of a horn antenna is
excited with two orthogonal modes in the input waveguide, which should be either circular or
square. The radiation performance then depends on the horn geometry. In general, a larger horn
will have a higher directivity and a lower beamwidth. As described in the previous chapter, one of
the main requirements for the RA and TA feeders is a good off-axis CP performance. The cross
polarization of an ordinary horn (circular or square) deteriorates quickly when moving off axis,
making it unsuitable for this scenario. Several known techniques are used for the improvement of
the cross polarization: corrugated horn (axial or radial corrugations), scalar horn, spline horn,
etc [105].

104
 5.4. 3D-Printed Corrugated-Horn Feed Chain

Starting from an ordinary circular horn, designed at 25 GHz, two variations were implemented.

 Scrimp horn – a multimode horn element, consists of a flared section, a single axial
corrugation and the cylindrical section
 Axially-corrugated horn – the same geometry, only with three axial corrugations

The geometry and radiation characteristics of the two horn geometries are compared with a
regular circular horn in Figure 5.15. The diameter of the feeding circular WG is chosen to be
7.8 mm to prevent excitation of higher-order modes in the waveguide. Two orthogonal WG
modes, with the same cutoff frequency, are supported in the specified frequency band. In the
simulations, the waveguide is fed with a wave port, and the two modes are excited with a phase
delay of 90° for CP radiation. The corrugated horn has the largest aperture area, and therefore the
highest directivity. Several corrugations yield a significant improvement in the broadside cross-
polarization, over the other two models. The corrugated horn, with its increased directivity and a
narrower beamwidth, will require a larger F/D value to achieve an optimal aperture efficiency, as
it will be explained in the next chapter. The cross-section of the axially corrugated horn with
optimized dimensions is shown in Figure 5.16.

5.4.3. Septum Polarizer

The practical excitation of CP in horn antennas can be done in a number of ways, including
external feeding networks, meander-line polarizers, etc. One of the simplest methods is to use a
so-called septum polarizer, shown in Figure 5.17 [106]–[108]. The septum polarizer consists of a
thin, ridged, metallic wall, placed in the middle of a square waveguide.
Directivity [dBi]

Directivity [dBi]

Directivity [dBi]

(a) (b) (c)

Figure 5.15. Simulated CP radiation patterns of the three horn antennas. (a) Regular circular
horn. (b) Scrimp horn. (c) Axially corrugated horn. Cross-sections of horn-antenna geometries
are shown above the corresponding radiation patterns.

105
5. Reflectarray and Transmitarray Components

Figure 5.16. Cross-section of the axially corrugated horn antenna. Dimensions (in mm): DWG=9,
Dap=31, t=1, Wr=1.5, Hr=2.5, H=18.7.

Figure 5.17. Septum polarizer in a square WG. Numbers denote the rectangular ports (1,2) and
the common square port (3). Dimensions (in mm): Wout=9.8, Win=7.8, Hin=3.4, t=1.

0 -2 110
S 3:1,3:1 S 1,3:1 S 1,3:2 -S 1,3:1
-10 S 11 -2.5 S 1,3:2 100

-20 -3 90

-30 -3.5 80

-40 -4 70
23 23.5 24 24.5 25 25.5 26 23 23.5 24 24.5 25 25.5 26 23 23.5 24 24.5 25 25.5 26
Frequency [GHz] Frequency [GHz] Frequency [GHz]
(a) (b) (c)
Figure 5.18. Septum-polarizer S parameters. ‘3:1’ in the subscript stands for port 3, mode 1.
(a) Matching at ports 1 and 3. (b) Transmission magnitude from port 1 to the two orthogonal
modes of port 3. (c) Transmission phase difference from port 1 to the two orthogonal modes of
port 3.

The polarizer consists of three ports: a square port that leads to the horn antenna, and two
rectangular ports. If one of the rectangular ports is excited, the resulting wave scattered on the
square port is CP. The excitation of the other port results in the CP of an opposite sense.

During the simulations, the fundamental TE10 modes are excited in the rectangular wave ports 1
and 2, and two fundamental modes, TE10 and TE01, in the square wave port 3. The heights and
lengths of the individual ridged sections of the septum are optimized to give the best CP
performance with a minimum total length. The optimized polarizer performance is presented in
Figure 5.18. The return loss at all ports and rectangular-port isolation are better than 30 dB in the

106
 5.4. 3D-Printed Corrugated-Horn Feed Chain

frequency band of interest. The equal power division to the two orthogonal modes in the square
WG and the transmission-phase difference of 90° ensure the excitation of a high-quality CP
radiation. The polarizers of this kind can be optimized further for a wideband or dual-band
performance.

5.4.4. Full Feed Chain

The 3D simulation model of the full feed-chain geometry is shown in Figure 5.19. The composing
parts of the antenna are:

1. Corrugated Horn – described in Section 5.4.2.

2. Circular to square WG transition – a simple linear taper from a 9-mm circular WG to a
7.8-mm square WG
3. Septum polarizer – described in Section 5.4.3.
4. WR42 transition – an S-shaped WG twist, with a gradual linear taper from a 7.8 x 3.4-mm2
WG to a standard-sized WR42 (10.668 x 4.318 mm2)
5. WR42 flange – a standard flange for mating with the adapters of the measurement system

The fabrication method, selected for the prototyping of the feeding system, is the additive
manufacturing in aluminum, or more precisely Direct Metal Laser Sintering (DMLS) [109]. It
allows the entire feed chain to be built as a single metallic piece. During this procedure, an
aluminum-alloy powder is selectively melted, layer by layer, with a high power-density laser. The
structure is therefore gradually ‘grown’ along one axis. In the case of this antenna, the printing
direction is along the Z-axis. The WR42 transition has a characteristic smooth shape since the
DMLS process cannot handle interior walls perpendicular to the printing direction (these would
appear if right-angle bends were used). The standard WR42 is designed as the input WG since the
WR42-to-coaxial (3.5 mm connector) adapters are available in the measurement facilities.

The pictures of the fabricated prototype are shown in Figure 5.20. During the anechoic-chamber
measurements, two WG-to-coaxial adapters were connected to both ports of the feed chain, and
a 50 Ω termination was additionally connected to the port that is not used for measurements. The
measured radiation pattern and AR of the antenna are compared with simulations in Figure 5.21.
The measured co-polarized pattern is in agreement with the simulated one. There is a larger
discrepancy in the cross-polarized pattern, which is attributed to the fabrication accuracy of the
DMLS method. The most critical part of the feed chain, in terms of fabrication, is the polarizer,
since it is difficult to accurately 3D-print the small geometrical details of the septum. However,
the measured AR is smaller than 2 dB within a beamwidth of ±20°, and the broadside AR is smaller
than 0.5 dB across the specified frequency bandwidth. Moreover, the radiation patterns retain the
shape and levels in this frequency range.

107
5. Reflectarray and Transmitarray Components

Figure 5.19. Simulation 3D model of the feed chain, with denoted individual components.

5.4.5. Future Integration with a CubeSat

The measured performance of the horn antenna prototype makes it suitable for RA and TA feed
applications. A stable AR over frequency is especially important for the operation with the
wideband RA antenna element, described in Section 5.3.

The presented feed-chain geometry is primarily developed for the laboratory evaluation of RA and
TA prototypes. The final antenna prototype will have to be redesigned for an efficient integration
with the CubeSat structure. The main aspects that could be modified are:

 Coaxial-to-WG transition – Instead of the bulky adapters, the coaxial probe can be
directly integrated into the feeding WG immediately before the septum polarizer.
Depending on the polarization used in the system, it could be possible to match one of
the two polarizer’s ports.
 Horn aperture – If the total horn antenna diameter (33 mm) is too large, the size can be
reduced by using only two corrugations. The increased horn beamwidth would then
require a smaller F/D of the antenna system, for an optimal illumination of the array
aperture.
 Septum polarizer in a circular WG – The square-to-circular WG transition can be omitted
by designing the septum polarizer in the circular WG, and gradually changing the cross-
section to the two rectangular output ports of the polarizer.
 Total length optimization – Once one or several of the mentioned modifications are
implemented and individually optimized, the entire structure should be optimized as a
whole with the main goal of minimizing the total length.

108
 5.4. 3D-Printed Corrugated-Horn Feed Chain

(a) (b)

(c) (d)
Figure 5.20. (a) Prototype of the 3D-printed feed chain. (b) Polished mating surfaces of the
WR42 flanges, for ensuring a good contact. (c) Feed chain mounted on a plastic support inside
the anechoic chamber, before adding the absorbers. The WG-to-coaxial adapters and the 50-Ω
load are visible in the picture. (d) Horn antenna detail, after the absorbers were placed.

109
5. Reflectarray and Transmitarray Components

0 5
RHCP sim. AR sim.
LHCP sim. AR meas.
-10 4
RHCP meas.
LHCP meas.
Gain [dBi]

-20 3

-30 2

-40 1

-50 0
-90 -60 -30 0 30 60 90 -90 -60 -30 0 30 60 90
[°] [°]
(a) (b)

2.5 19
AR sim.
AR meas.
2 17

1.5 15

1 13

0.5 11 RHCP sim.

RHCP meas.
0 9
23 23.5 24 24.5 25 25.5 23 23.5 24 24.5 25 25.5
Frequency [GHz] Frequency [GHz]
(c) (d)

-10

-20

-30
S 11 sim.
S 11 meas.
-40 S 21 sim.
S 21 meas.
-50
21 22 23 24 25 26 27
Frequency [GHz]
(e)
Figure 5.21. Simulation and measurement results of the feed chain. (a) CP radiation patterns.
(b) AR versus angle. (c) AR versus frequency. (d) Realized gain vs. frequency. (e) S parameters.

A drastic size reduction could be achieved with these modifications, without a significant loss in
performance. However, the horn aperture remains the bulkiest part of the feed chain. During the
CubeSat launch, it would have to be stowed in a dedicated compartment inside the satellite body.
One envisioned configuration is shown in Figure 5.22. The feed chain is stowed inside the CubeSat

110
 5.5. Summary and Discussion

and covered with the array panel(s). Following the panel deployment in orbit, the horn simply
moves up to the required position by a spring mechanism. The rotation is done around the axis of
the feeding coaxial probe. The feeding structure can be integrated into the rotating hinges.

This type of feed occupies a significant volume compared to planar feeds (patch arrays) that are
stowed on the CubeSat exterior. As such, it is not compatible with most standard CubeSat chassis
and internal components directly below the feeding mechanism. However, the ISL functionality
is, in itself, an exotic feature for CubeSats in today’s standards. Therefore, the feed and array
designs are primarily intended for high-performance CubeSats, for which a custom-built chassis
is typical.

Figure 5.22. The stowage and the deployment mechanism of the proposed feed chain. The side
view shows the cross section along the CubeSat center.

5.5. Summary and Discussion

RA and TA antennas offer an excellent compromise in terms of mechanical complexity and
electrical performance. They are an attractive choice for CubeSat CP ISL communications in K/Ka
bands, when a required antenna aperture is not available on the satellite surface. These antenna
types consist of a planar array of discrete elements and a feeding element, placed in the array focal
point. Different methods are used to achieve the phase shifting of the incident radiation from the
feed, in order to scan the beam radiated by the antenna aperture, in the desired direction. This
chapter focused on the study of individual elements of RA and TA antenna systems.

Most of the elements in the considered arrays are similar in geometry and they can be efficiently
analyzed using the infinite-array approximations. Simulation setups for RA and TA elements,
using periodic boundary conditions and Floquet-harmonic excitations, are presented in
Section 5.1.1.

If the array radiation is CP, a special technique of variable element rotation can be used for the
phase-shifting purpose. A detailed theoretical analysis of the technique is given in Section 5.1.2.
The RA elements that qualify for the element-rotation technique must have the same CP sense of

111
5. Reflectarray and Transmitarray Components

the scattered and incident waves, whereas the CP sense of the wave should change while passing
through an eligible TA element. The CP-sense change is achieved with a phase difference of 180°
between the two LP wave components. The phase shifting is then done with a simple element
rotation around the vertical axis. Two novel elements that satisfy described conditions, one each
for RA and TA antennas, are presented in this chapter for CubeSat ISL applications.

Conventional TA elements achieve the required phase difference with either a significant profile
or unreliable metallized vias. The TA element proposed in Section 5.2 introduces a simple
modification to the slot shape, thus reducing the number of required metallic layers from five to
three. The associated profile reduction and the manufacturing simplicity can be the enabling
factors for CubeSat applications. The element is incorporated in two final TA antenna designs.
The full array performance is simulated using the approach described in Chapter 6, where the
prototype measurements of the TA antenna are also presented.

The novel coupled-loops RA element, presented in Section 5.3, is an evolution from a well-known
microstrip split-loop. By coupling two microstrip loops, a significant bandwidth improvement is
obtained. The frequency spacing of the two resulting resonant modes is a key factor for enabling
a wideband performance. The increase in complexity is minimal, compared to the original
element, as the final geometry remains single-layer without any metallized vias or SMD
components. The simple geometry and a wideband CP performance make the coupled-loops
element a very promising candidate for CubeSat communications, as well as other RA
applications.

The feed antenna should radiate CP across the entire frequency band that the RA and TA elements
cover. An axially corrugated horn antenna is designed for this purpose. The axial corrugations
provide a lower off-axis cross polarization, compared to the conventional circular horn. A septum
polarizer in the common square WG provides the required CP radiation at the horn aperture. The
feed-chain prototype is fabricated for laboratory testing, using the additive manufacturing in
aluminum. The measured standalone performance of the all-metal feed-chain prototype
confirms its eligibility for CubeSat ISL applications as the RA and TA feeding element. The
presented horn antenna is used during the design and measurement phases of several RA and TA
antennas in the Chapter 6.

112
6. RA/TA Design Process

Once the individual components (array elements and the feeding antenna from Chapter 5) have
been designed and characterized, the simulation results are combined to predict the performance
of the entire RA or TA antenna system. The array synthesis and radiation analysis processes are
presented in the beginning of this chapter. Both functions are performed with a custom MATLAB
code, which imports the simulated results from the EM software, generates the array geometry,
and performs the necessary far-field calculations.

Both the arrays and the feeding antenna exhibit CP radiation, which is a main requirement for ISL
communications. The AR of individual array elements and the feed horn, described in the
previous chapter, is low across the entire specified frequency band. Consequently, the variable
rotation technique can be used to generate the required phase shift across the array, for
collimating the array radiation and tilting the main beam.

The variable rotation technique for RA and TA antennas in its standard form uses an analytical
formula to calculate the rotation angles that provide the necessary phase shift across the array
aperture. Depending on the element geometry and the elevation angles of incidence, a phase
error is introduced with this approach. A different, and yet simple, process is used here to
eliminate the phase errors and consequently increase the array gain.

One RA and one TA antenna, with 900 elements each, are designed and analyzed using the
described code. The aperture efficiency of the arrays is calculated as a combination of individual
efficiency factors. The designed array yield a gain of at least 32 dBi, with the aperture efficiency
larger than 60%. The 1-dB CP bandwidths of the two array are comparable, at ~6%.

Two prototype sets of the designed arrays are fabricated for measurements at two different
facilities. The prototype measurements are compared with the simulations at the end of the
chapter. The agreement between the conducted measurements and simulations verifies the
antenna performance and the RA and TA design process.
6. RA/TA Design Process

6.1. Design Process Overview

The entire design process is summarized with a flowchart in Figure 6.1. The program is separated
into two main parts:

 Synthesis and analysis – calculates the element rotation angles and the array radiation at
the central frequency
 Frequency analysis – calculates the radiation parameters over a frequency band for the
obtained element rotations

The first part uses the array unit-cell and the feed simulation results at the center frequency as the
input. The simulation results are then interpolated and sampled, and the scattered fields are
calculated using the procedure described in Section 5.1.2. The results are then used to obtain the
array geometry in the form of the element-rotation matrix. The total array radiation and aperture
efficiency are calculated at the center frequency with the approach described in Sections 6.4 and
6.5, starting from the scattered fields on individual array elements.

Figure 6.1. A flowchart of the RA and TA antenna design and analysis process. Frequency f0
represents the central (design) frequency of the array, and fmin:fmax the analyzed frequency range.

114
 6.2. Feeding-Antenna Radiation

The inputs for the second part of the code are the unit-cell results in the entire frequency range of
interest and the element-rotation matrix, obtained in the first part. The same analysis procedure
is then repeated at every frequency point, for which the unit-cell and the feed responses are
available.

The unit-cell simulations can become time consuming over a frequency band, having in mind the
large number of incidence angles that must be considered. The use of a quasi-3D EM solver for
this part accelerates the process and obtains the wideband unit-cell performance in a shorter
time.

The antenna geometry and coordinate systems used for the arrays considered in this thesis are
presented in Figure 6.2. The focal point is located above one edge of the array, to model the feed
antenna placed on the CubeSat surface and the array deployed perpendicularly to that surface.
The following sections (6.2 – 6.5) describe the individual steps of the array-design process, and
discuss the main points from the CubeSat perspective.

6.2. Feeding-Antenna Radiation

6.2.1. Methodology
The radiated fields from the feeding antenna illuminate the individual array elements with a
different incidence angle, magnitude and relative phase. Determining the incident fields
accurately is necessary for a good array synthesis and the estimation of the array far field. Two
methods, commonly used to evaluate the array illumination, are depicted in Figure 6.3.

The first method involves sampling the far field of the feeding antenna, calculating the distance
from the feed to the individual elements and combining the far field with the path loss to obtain
the magnitude and phase of the incident fields. The method assumes that the array is located far
enough from the feed, so that it is not influenced by the near fields of the feed. Since an FEM code
is used for the 3D simulation, the model of the feeding antenna needs to be enclosed with an
‘airbox’ – a 3D object with radiation/PMC boundaries at the object faces.

The second method calculates the near fields in the exact locations of the individual elements.
The method implies either having a large airbox or using a hybrid FEM-MoM setup (FEM inside
the two regions and MoM at the exterior). The Poynting vector is calculated at the element
locations and adopted as the incidence angle for each element. The electric fields transversal to
this vector are adopted as the incidence fields, from which the magnitude and phase are obtained.
Alternatively, (6.1) can be used for the same purpose.

115
6. RA/TA Design Process

The second method can be considered more accurate, as it calculates exact fields at element
locations. However, if the feed position changes, the entire model must be re-simulated, since the
relative position of the array elements has changed. For the first method, it is only required to re-
sample the existing far-field results, which requires significantly less resources.

The first method is adopted in this thesis. It is found more convenient in the case where several
arrays with different F/D values are used. It gives the freedom to optimize the F/D for a given
feeding antenna, with just one 3D simulation. The array elements considered in this chapter are
located in the far-field zone of the feeding antenna, and thus satisfy the conditions to use this
method.

Figure 6.2. The array geometry and nomenclature used throughout the chapter, with respect to
the deployed-array configuration on a 3U CubeSat (top-left).

Figure 6.3. Two common EM simulation setups for the calculation of the feeding-antenna
radiation. The dashed lines represent the airbox boundaries for the FEM model and the dotted
lines represent the boundaries for the hybrid FEM-MoM regions.

116
 6.3. Array Synthesis

6.2.2. Calculation of Incident Fields

Two Cartesian coordinate systems are defined for the feed and the array elements. The systems
are aligned at first, and then the elements are moved to their final position using geometrical
manipulations (rotation and translation matrices). The element positions are determined in the
feed’s coordinate system and converted to the spherical coordinates. The far field is sampled at
coordinates and (in the feed’s system) and the element distance can be obtained from the
coordinate r. Finally, the path loss is calculated from the distance to the individual elements and
combined with the radiation pattern of the feed to obtain the incident fields on the array.

Examples of the normalized magnitude of the incident electric field is presented in Figure 6.4 for
three values of F/D, in the case of an offset-fed array from Figure 6.2. The axially corrugated horn
from Chapter 5 is used as the feeding element in all cases, pointing towards the array center. As
often suggested in the literature, the goal is to have the array edges illuminated by a field
magnitude approximately 10 dB lower compared to the center illumination, for optimal aperture
efficiency. This condition is fulfilled in Figure 6.4(b). The two sub-optimal scenarios are illustrated
in Figure 6.4(a,c). Having a lower F/D results in a poor utilization of the array aperture, as the feed
radiation is focused in the array center. On the other hand, a large F/D value yields a large power
spillover. The loss mechanisms are further discussed in Section 6.5.

6.3. Array Synthesis

6.3.1. Unit-Cell Incidence Angle
The element-rotation technique states that the phase of the co-polarized reflected or transmitted
signal is equal to twice the element rotation angle. This is, however, strictly valid only for the
broadside wave incidence. The actual phase of the considered RA and TA elements (and many
other) deviates from this rule for large elevation angles, as seen in Figure 6.5.

(a) (b) (c)

Figure 6.4. The normalized magnitude of the electric field incident on the offset-fed array, for
different values of F/D. (a) F/D=0.7. (b) F/D=0.9. (c) F/D=1.1.

117
6. RA/TA Design Process

Figure 6.5. Phase response of the (a) RA and (b) TA elements, versus the azimuth incidence
angle, for different elevation incidence angles.

The effect is more severe for offset-fed RA and TA antennas, as the average elevation angle of
incidence is larger in that case. For an accurate array synthesis, the element rotation and the
resulting phase of every element must be determined for the given elevation angle of incidence.

6.3.2. Assumptions
The RA and TA are typically composed of hundreds of elements, and the angle of incidence is
different for most of the elements. Ideally, the unit cell should be simulated for every combination
of azimuth/elevation angles of incidence and rotation angles. This would not only be a tedious
task, it would also have to be repeated every time a different position of the feed is desired. Instead,
a given unit cell is simulated for a set of uniformly distributed azimuth/elevation angles, and the
results are interpolated. The phase of a certain incidence angle is then obtained by a simple table
lookup in the interpolated results. During this process, an assumption is made that the negative
change of the element-rotation angle is equal to an identical positive change of the azimuth angle
of incidence [91]. The unit cell is then simulated only with the element rotation set to zero.
Although this arrangement is not representative for the majority of the array elements, the
assumption is valid if the inter-element coupling is low, as is the case in this thesis.

6.3.3. Element Rotations

For the array configuration of Figure 6.2, the incidence angles of element n, located at coordinates
, , can be found by:

acos , atan (6.1)

2

118
 6.3. Array Synthesis

,
2
where angles and are defined starting from z and x axes, respectively. To scan the beam in the
, direction, the phase distribution across the array is given by:

sin cos sin sin (6.2)

where is an arbitrary constant. Expressions (5.9) suggest that the element rotation should be
set to:

(6.3)
2
since the phase shift is equal to twice the value of the rotation angle. However, this rule is valid
only for broadside incidence, although it yields more accurate results for center-fed arrays, or
arrays having a large F/D value.

A different approach of element-rotation calculation is adopted here to eliminate the phase errors
that arise for large elevation angles of incidence. The arrays considered in this thesis are offset-
fed, meaning that the angles have a large value on average. As an illustration, if F/D equals
0.8 for the configuration in Figure 6.2, approximately 66% of elements will have the angle
larger than 30°.

The array synthesis is described here for RA antennas. The procedure for the n-th element of the
array is as follows:

1. Calculate the azimuth angle of incidence using (6.1).

2. Determine the current phase of for the given azimuth angle .
3. Determine the required phase in that array location with (6.2) and the associated
required azimuth angle.
4. The element rotation angle is the difference between the required and initial azimuth
angles.

The procedure is graphically represented in Figure 6.6. An equivalent procedure is used for TA
antennas, except that the parameter is used in step 2. The proposed approach does not
notably increase the execution time, compared to the closed-form expression (6.3), even for
several thousand elements within the array. The procedure is repeated for every array element, to
obtain the element-rotation matrix for a given F/D and beam-steering direction.

The element rotations are calculated for three F/D values as an example, similarly as in Figure 6.4,
and shown in Figure 6.7. A lower F/D yields more phase wraps. Unlike the conventional RA
elements, which cannot achieve a full 360° reflection-phase range, the phase wrapping is not an
issue here, and it is hardly visible in the final antenna prototypes.

119
6. RA/TA Design Process

Figure 6.6. A graphical representation of the array-synthesis procedure on the element level.
Numbers in bold mark the order of steps.

(a) (b) (c)

Figure 6.7. Characteristic element-rotation angles across the RA surface, for different values of
F/D. (a) F/D=0.7. (b) F/D=0.9. (c) F/D=1.1.

6.4. Array Analysis

The analysis in Chapter 5 is used to calculate the fields, scattered from the array of rotated
elements. The described procedure is performed for every array element, summing the individual
far-field contributions afterwards.

The unit-cell simulations are done with LP excitations (TE, TM) in the quasi-3D EM software and
the S parameters are converted directly in the software to using expressions (5.9). This is
convenient since the field radiated by the feeding antenna is also CP, and the CP performance of
the unit cells can be quickly inspected within the software. The incident fields on the array ( )

120
 6.4. Array Analysis

are obtained by adding the propagation loss to the horn-antenna gain, sampled in the direction
of every element. The incident fields are then combined with CP S parameters to obtain the
scattered fields ( ) using (5.6), taking into account the exact incidence and rotation angles for
every element. The rotation angles are calculated using the procedure in Section 6.3.2.

The far-field radiation of the array is calculated using the special case of the Field Equivalence
Principle applied to radiation of rectangular apertures. A RA scenario is considered here for
simplicity, although an identical procedure is applied for the TA case. Every array element is
treated as a uniformly illuminated rectangular aperture in an infinite PEC plane and the element
is replaced by the equivalent magnetic currents, from which the radiation is calculated. The
scattered fields are represented in the form of two LP orthogonal components. For an
arbitrary oblique incidence, the scattered fields are first converted to TE and TM components:

1 1 1
(6.4)
√2
and then to Cartesian components:

1
(6.5)

where and can be obtained by replacing , with , in (5.2), and:

sin . (6.6)

Index n is omitted in (6.4)-(6.6) for clarity, and it is implied that the expressions are calculated for
every array element. The spherical components of the radiated fields are then obtained by:

, cos sin (6.7)

2

, cos sin cos . (6.8)

2

sinc sinc (6.9)

2 2
where and are obtained from (5.2) and is the square unit-cell dimension. Expressions
(6.7) and (6.8) give the radiated electric-field components for a single cell n, and the total far field
of the entire array is obtained by summing the contributions from N individual cells:

, , , , , . (6.10)

Finally, the RHCP and LHCP radiated components, directivity and AR are:

1
| ∓ (6.11)
√2

121
6. RA/TA Design Process

2
| 10 log | (6.12)

| | | |
20 log (6.13)
| | | |
all being functions of spherical angles and , where is the free-space wave impedance. The
power radiated by the feed is set to 1 W.

The described analysis procedure is performed at the central frequency, for which the element
rotation angles are calculated. After the element-rotation matrix is obtained, the far-field
calculation procedure is repeated for every frequency point, for which the cell and feed
performance is available, as illustrated in Figure 6.1. During the frequency analysis, a fixed
element-rotation matrix is used. The radiation pattern is calculated at the central frequency, and
the directivity and AR are calculated at the beam-steering direction over the frequency band.

6.4.1. Alternative Approach

As an alternative to the method described in this section, the total array radiation can be
calculated using the phased-array theory. With this approach, the element’s radiation pattern is
evaluated in the 3D EM software, and the far field is obtained using the scattered field of every
element as the array excitation vector. The approach simplifies the far-field calculations, but
requires the calculation of the active element pattern across a hemisphere. The active element
pattern is obtained in the infinite-array approximation by illuminating the unit cell from a set of
directions and recording the realized gain in the wave-incidence direction. This is a lengthy
procedure, as it requires simulating the unit cell many times. However, the simulations for the
magnitude/phase response of the cell can be re-used for this purpose, and the angular range can
be reduced if the element possesses symmetry planes. The method has not been implemented in
this thesis.

6.5. Aperture Efficiency

A uniformly illuminated aperture with a linear phase gradient would have an ideal efficiency of
100% and a maximal theoretical directivity. In a realistic antenna, several factors decrease this
ideal directivity, and their combined contributions are quantified by the total efficiency:

(6.14)

The total efficiency value consists of several terms:

 – the mismatch and ohmic losses associated with the feeding antenna

122
 6.5. Aperture Efficiency

 – taper efficiency; quantifies the effects of the electric-field magnitude across the
aperture deviating from the ideal, uniform distribution:
∑ | |
(6.15)
∑ | |
 – spillover efficiency; the ratio between the power incident on the array and the total
radiated power by the feed:
| | | |
cos (6.16)
2
 – quantifies effects of the electric-field phase across the aperture deviating from the
ideal, linear gradient, required for beam steering to 0°, :

1 cos ∑ (6.17)
4 ∑ | |
 – the ratio between the total power scattered from the array to the total power
incident on the array:
∑ | | | |
(6.18)
∑ | | | |
 – the ratio between the power scattered from the array in the desired polarization to
the total power scattered from the array
∑ | |
(6.19)
∑ | | | |
The expressions (6.16) – (6.19) are obtained by combining the analysis in [82], [91], [105] and
writing the main results in the terminology used here. Depending on the specific RA or TA
scenario and the feed/array polarizations, the waves and should be replaced by the proper
polarization components (RHCP or LHCP) and port indices.

6.5.1. Discussion
The taper and spillover efficiencies are combined into a single parameter called the aperture
efficiency:

(6.20)

This component is the dominant contributor to the total antenna efficiency, and it is imperative
to maximize it by a proper selection of the system geometry and the feeding antenna parameters.
In the literature, the aperture efficiency is typically maximized by using a center-fed circular array.
A circular array aperture is the most suitable due to the circular symmetry of the feed’s radiation
pattern. A center-fed array results in a good taper efficiency and a relatively low average value of
incidence angles, especially in the strongly illuminated array center, which is preferable for many

123
6. RA/TA Design Process

RA or TA elements with poor angular performance. Closed-form expressions for the aperture
efficiency are available for this specific scenario [81].

The array geometry for antennas described in this thesis is constrained by the CubeSat structure,
the deployment mechanism of the RA/TA panels, and the beam-steering direction. Therefore, the
feeding element is always positioned with an offset from the array center. On the other hand, a
rectangular (or square) array aperture is more suitable for the stowage on the CubeSat surface.
From a clearly engineering standpoint, a rectangular aperture that maximizes the available
surface is more advantageous due to an increased gain, although its aperture efficiency is lower.
A detailed study on the efficiency of square reflectarray antennas showed how the aperture
efficiency drops as the function of the feed offset-angle [110].

In our case, it would not be justified to use a circular aperture, simply to artificially boost the array
performance (efficiency). Therefore, the values of aperture efficiencies for the square RA and TA
antenna prototypes, presented in Chapter 7, are lower than what the proposed array elements can
reach. To alleviate the effect, a larger F/D value is selected and the feeding antenna is designed to
have a higher directivity than the conventional feed horns. In this way, the average value of
incidence elevation angles is reduced.

The phase efficiency is calculated by (6.17) having in mind that the considered beam scanning is
only within the plane 0°, (in other words, in the plane containing the focal point and
the CubeSat’s long axis). The ohmic and dielectric losses of the unit cell are described by ,
and the element’s polarization performance by . All EM simulations are performed including
realistic substrate materials. Other factors that contribute to the total efficiency (feed blockage,
specular reflection) can only be quantified by 3D simulations of the entire array [82], which is not
performed due to limited available computational resources.

6.6. Array Simulations

The total array size, for either RA or TA antennas, should be larger than a surface equivalent to
three CubeSat units (3U) to achieve a gain larger than 30 dBi with a realistic efficiency and a small
margin (see Figure 4.4). The arrays composed of 30 x 30 elements (a total of 900) are selected. As
the unit-cell dimension is 6 mm, the total array dimensions are 180 x 180 mm2
(~218 λ02 @ 24.6 GHz) or 3.24 U in CubeSat units.

If the (quasi-)3D simulation results are available, the complete array synthesis and analysis
procedures are efficiently executed on an ordinary PC, for 900-element RA and TA antennas. This,
combined with the re-sampling of the feed radiation, allows us to vary the F/D of the arrays is in
several sequential program runs. An optimal F/D value, in terms of the aperture efficiency (6.20),
is found to be 0.9. The design procedure is again repeated several times for various phase

124
 6.6. Array Simulations

constants . Changing moves the elements having a poor reflection/transmission efficiency

away from the strongly illuminated array center and increases the total gain.

The simulation results of the RA and TA antennas are presented in Figure 6.9 – 6.12, and their
performance is summarized in Tables 6.1 and 6.2. Both considered arrays have an F/D value of
0.9. According to Figure 4.11, the required beam-scanning angle for ISLs is in the range from 10 to
30°. Here, this angle is arbitrarily set to 25° and 20° for the RA and TA antennas, respectively.
Figure 6.8 shows the same result as presented in Figure 6.9, only along one dimension of the RA
aperture.
Angle [°]

Angle [°]

Angle [°]
Magnitude [dB]

Magnitude [dB]

Magnitude [dB]

Figure 6.8. Simulated incident and scattered electric fields, and the S parameters, of a 900-
element RA antenna at 24.6 GHz, for one row of elements along the x axis. The field components
are the same as in Figure 6.9.

Table 6.1. Simulated-performance summary of the RA and TA antennas. Gain and AR values are
calculated at 24.6 GHz in the main-beam direction.

Parameter Limit RA antenna TA antenna

Gain @ , 32 dBi 32.6 dBi
AR @ , 1.4 dB 0.7 dB
SLL -19 dB -19.2 dB
Beam-Steering Angle -25° -20°
Beamwidth 1-dB 2.5° 2.5°
1-dB 6.5% 5.7%
Gain bandwidth
3-dB 8.2% > 8%
AR bandwidth 3-dB 7.4% > 8%

125
6. RA/TA Design Process

Figure 6.9. Simulated incident and scattered electric fields, and the S parameters, of a 900-element
RA antenna at 24.6 GHz. (Top) The angle and (bottom) magnitude of the (left) incident electric
field, (center) the S parameters, and (right) the scattered electric field.
Directivity [dBi]
Directivity [dBi]

AR [dB]

Figure 6.10. Simulated far-field performance of a 900-element RA antenna. (left) CP gain versus
angle, at 24.6 GHz. (top-right) CP gain versus frequency, at the maximum-gain angle. Dashed
lines represent the aperture-efficiency values, in percent. (bottom-right) AR versus frequency, at
the maximum-gain angle.

126
 6.6. Array Simulations

Figure 6.11. Simulated incident and scattered electric fields, and the S parameters, of a 900-
element TA antenna at 24.6 GHz. (Top) The angle and (bottom) magnitude of (left) the incident
electric field, (center) the S parameters, and (right) the scattered electric field.

40 33
70%
LHCP
RHCP 60%
32
30
50%

31
20
DRHCP @ = -20° 40%

30
23.8 24.2 24.6 25
10 Frequency [GHz]

3
AR @ = -20°
0
2

-10
1

-20 0
-90 -60 -30 0 30 60 90 23.8 24.2 24.6 25
[°] Frequency [GHz]

Figure 6.12. Simulated far-field performance of a 900-element TA antenna. (left) CP gain versus
angle, at 24.6 GHz. (top-right) CP gain versus frequency, at the maximum-gain angle. Dashed
lines represent the aperture-efficiency values, in percent. (bottom-right) AR versus frequency, at
the maximum-gain angle.

127
6. RA/TA Design Process

Table 6.2. A breakdown of RA and TA antenna efficiencies at 24.6 GHz. The factor is set to
100%.

RA antenna TA antenna
Efficiency
% dB % dB
Taper 87 -0.58 87 -0.59
Spillover 81 -0.91 81 -0.91
Aperture 71 -1.49 71 -1.5
Phase 100 0 99 -0.03
Transmission – – 95 -0.21
Polarization 85 -0.71 98 -0.09
Total 60 -2.21 69 -1.62

In the case of the RA antenna, the phase distribution of the incident field in Figure 6.9 is a
consequence of a spherical wave propagation form the feeding antenna. In combination with the
phase of the reflection coefficient , it yields a linear phase across the antenna aperture,
required to scan the beam in the desired direction. The average magnitude of the parameter
is -0.55 dB, and therefore, the magnitude of the scattered field has a very similar distribution to
the incident one. Concentric circles, visible in the magnitude plot, correspond to the
elements that have a lower reflection efficiency for the combinations of azimuth/elevation
incidence angles at those element locations. As previously discussed, the positions of these lower-
efficiency circles can be varied by changing in the array-synthesis design phase.

It should be noted that the presented S-parameter is calculated at the design frequency of
24.6 GHz, which is between the two resonances of the coupled-loops RA element, as previously
presented in Figure 5.10. This frequency corresponds to the minimum value of the parameter
across the element bandwidth and consequently, a minimum efficiency value of the RA antenna
across the operating bandwidth. Therefore, the RA efficiency is further improved above and below
the design frequency, resulting in a wideband performance. Both the gain and the AR over
frequency follow a similar trend, with slightly better values above and below the design frequency.

For the TA antenna, the according transmission parameter and the transmitted scattered
component are presented in a similar fashion. The average magnitude of the parameter is
- 0.29 dB, and the aperture efficiency reaches a value of 69% with an AR lower than 0.8 dB.
Although the aperture-coupled-patch element has a single-resonance response and a low profile,
a 1-dB CP bandwidth of 5.7% is achieved.

Overall, the simulated performance of the RA and TA arrays is significantly improved over the
similar designs in the literature. The reported 1-dB and 3-dB gain bandwidths are typically 2 – 3%
and 5 – 7% respectively, for the conventional single-layer split-loop RA geometries and other CP

128
 6.7. Prototype Measurements

single-layer RA elements. In contrast, the calculated values for the proposed coupled-loop RA
element are 6.5% and 8.2%. Note especially the improvement of the 1-dB gain bandwidth by a
factor of two. The proposed TA antenna reaches an aperture efficiency of 69% with only three
metallic and two dielectric layers, as compared to a previously reported simulated efficiency of
56%, for a similar array having five metallic layers [91]. The CP aperture efficiency is among the
highest reported for this type of antennas, with the total thickness of only 0.13 1.6 mm .
Despite the low electrical profile, a good gain bandwidth of the array is a consequence of the
wideband performance of the aperture-coupled patch element.

As an additional illustration, the TA antenna, having a circular array geometry, is simulated with
a simple CP square horn as the feeding element, centered on top of the circular array. A total
efficiency of 0.73 (-1.4 dB) is achieved with an F/D value of 0.8 and a beam scanned to broadside,
showing that the efficiency of both arrays can be increased further if a different system geometry
and the feeding antenna parameters are selected.

Most importantly, the total efficiency of the proposed arrays is mainly determined by the aperture
efficiency, which does not take into account the element performance. The efficiency degradation
by the elements themselves is minimal. In other words, the reflection/transmission and
polarization performance of the presented RA and TA element geometries maximizes the
achievable capability of this antenna type, under the constraints of the given feeding scenario.

6.7. Prototype Measurements

6.7.1. Array Prototypes
Several antenna prototypes are fabricated to evaluate the antenna performance. The laminate
Rogers Duroid 5870 ( 2.33, tan 0.0012, 0.508 mm) is used for the RA substrate, and
Rogers RO6002 ( 2.94, tan 0.0012, 0.76 mm) for the two TA substrates. The two TA
boards are bonded together using the Rogers 2929 BondPly ( 2.94, tan 0.003, 51 μm)
adhesive film. The selected bonding layer has the same permittivity as the two substrates, to
eliminate the potential reflections at the dielectric interfaces.

The 30 x 30-element arrays (180 x 180 mm2) are printed on a 200 x 200 mm2 PCBs. A PVC plastic
support seen in Figure 6.13 is designed and fabricated specifically for these arrays, to place the
feed and the array in the correct position during the measurements. Both RA and TA antennas are
measured using this support. The support also allows to translate the feed along the axis
perpendicular to the array aperture, and to rotate the feed around its phase center. Arrays of
different F/D can be measured using this support and the position can be finely tuned during the
measurements.

129
6. RA/TA Design Process

The far-field anechoic chamber, available at the MAG premises, offers a maximum measurement
distance of 2.3 m. The 900-element antennas have the boundary between the radiating near field
and the far field zones at:

2 √2 (6.21)
10.63 m .

Measuring full-sized arrays in the said anechoic chamber is not adequate. Therefore, a smaller
version of each array is designed and fabricated for the initial measurement campaign. The
elements inside the smaller arrays are arranged in a circle with a diameter of 120 mm and the far-
field zone exactly at 2.3 m. The total number of elements is 316.

The same plastic support, designed for the 900-element arrays is also used for the measurements
of the 316-element array. Due to the support constraints, the projection of the feed’s phase center
is outside of the circular array, putting it in a disadvantageous position for achieving a high
aperture efficiency. Therefore, the overall performance for this set of prototypes are lower than in
the case of full-sized arrays.

After the initial far-field measurements in the anechoic chamber, the 900-element square arrays
are measured using the near-field scanner at Viasat premises, located in the EPFL Innovation
Park. The plastic support is modified for a proper mating to the standard flange inside the scanner.

6.7.2. RA Measurement Results

The 316-element prototype of the circular RA antenna, and the corresponding simulation and
measurement results, are shown in Figure 6.14. The horn-antenna rotating mechanism, the
sliding rail and the plastic frame around the array are not considered in the numerical analysis of
the array. The sliding rail presents an obstacle for the RA radiation, and it is the cause of a large
discrepancy between the simulations and measurements for positive angles.

The measured RA radiation agrees with the numerical analysis in the remaining angular range.
The measured SLL in the RHCP radiation pattern are within 2 dB of the predicted values. The
maximum-gain direction is shifted by 0.5°, which is explained by the incorrect estimation of the
substrate permittivity in the unit-cell simulation model. The resulting frequency shift is also
visible in the gain-vs-frequency plot in Figure 6.14(e).

Both the gain and AR measurement results show an improvement in the central part of the band,
between the two coupled-loop resonances, compared to the calculations. The effect stems from
an increased coupling between the two D-shaped loops in the RA prototype, with respect to the
simulation model. As analyzed in Figure 5.10(b), a smaller gap between the loops, and a higher
coupling at the same time, correspond to a smaller separation between the two resonances. The
RA element is simulated with zero-thickness conductors throughout the study, to accelerate the
analysis with the quasi-3D solver. On the other hand, the actual prototype has finite-thickness

130
 6.7. Prototype Measurements

conductors ( ~17 μm ), as it can be noticed in Figure 6.14(a). In the prototype, the coupling is
increased due to the increased lateral surface between the two loops, most importantly in the gap
region. As the element resonances are less spaced in frequency, the gain over frequency is
consequently flatter than predicted, and the AR values lower. For a more accurate prediction of
the RA antenna radiation based on this type of elements, finite-thickness conductors must be
included in the simulation models.

In spite of all the unaccounted factors, the circular RA prototype, although of suboptimal system
geometry, achieved a maximum gain of 26.2 GHz (52% efficiency), 1-dB gain bandwidth of 7.5%
and 3-dB AR bandwidth of 8.1%, validating the results predicted through simulations.

Similar conclusions apply for the 900-element square RA antenna prototype, as for the smaller-
scale circular antenna. The large prototype is shown in Figure 6.15 alongside simulation and
measurement results. As expected, the aperture efficiency and radiation-pattern agreement are
significantly improved. The improvement is due to a more advantageous feeding location, as well
as an increased array size. The antennas of this type are known to perform better as their size
increases, since the infinite-array assumption is valid for a larger number of elements inside the
array. The maximum gain of this antenna at -25° reaches 31.7 dBi for a total efficiency of 60%. The
measured 1-dB CP-gain bandwidth is 6.8% for an array profile of only 0.08 λ at 24.6 GHz. The
measured 3-dB AR bandwidth is 10%.

(a) (b)
Figure 6.13. RA measurement setup inside the anechoic chamber. (a) The array and the feed
horn mounted on the plastic support. (b) Feed horn detail.

131
6. RA/TA Design Process

(a) (b)

Axial Ratio [dB]

Gain [dBi]

(c) (d)
28 5

60% AR sim.
AR meas.
27 4
50%
Axial Ratio [dB]

26 3
Gain [dBi]

40%

25 2

30%

24 1
RHCP sim.
RHCP meas.
23 0
23 23.5 24 24.5 25 25.5 26 23 23.5 24 24.5 25 25.5 26
Frequency [GHz] Frequency [GHz]
(e) (f)
Figure 6.14. The 316-element circular RA antenna prototype. (a) Array detail. (b) Antenna
prototype in the anechoic chamber. (c) Radiation patterns in the plane 0° , at 24.6 GHz.
(d) AR versus angle, in the vicinity of the main lobe. (e) CP gain versus frequency, at 30°.
Gray lines represent the aperture-efficiency values. (f) AR versus frequency, at 30°.

132
 6.7. Prototype Measurements

(a)

(b) (c)
Axial Ratio [dB]
Gain [dBi]

(d) (e)
Figure 6.15. The 900-element square RA antenna prototype. (a) Antenna prototype in the near-
field scanner. (b) Radiation patterns in the plane 0°, at 24.6 GHz. (c) AR versus angle, in the
vicinity of the main lobe. (d) CP gain versus frequency, at 25°. Gray lines represent the
aperture-efficiency values. (e) AR versus frequency, at 25°.

133
6. RA/TA Design Process

6.7.3. TA Measurement Results

The prototype of the circular TA antenna and the corresponding simulation and measurement
results are shown in Figure 6.17. Unlike the RA case, the effect of the plastic support is minimal in
the radiating hemisphere, as the feed and the sliding rails are located in the opposite side of the
array. In this setup, the feeding horn antenna is oriented towards the test antenna in the anechoic-
chamber setup. Therefore, the spillover RHCP radiation from the horn becomes dominant when
it appears behind the TA aperture, as seen from the test antenna. The effect is visible in the
radiation patterns (especially at 40° ), where the cross-polarized RHCP radiation is
significantly increased, with respect to the numerical analysis.

Nevertheless, the calculations accurately predict the TA antenna radiation, as the measured and
simulated main lobes practically overlap, as well as the AR value at that angle. The measured
maximum gain shows a larger discrepancy, compared to the RA case. The measured 1-dB gain
bandwidth is 6.5% and the 3-dB AR bandwidth is larger than the measured frequency range (12%).

The plastic support had to be significantly modified for measurements of the 900-element TA
antenna in the near-field scanner. The differences between two measurement setups are shown
in Figure 6.16. The measurement results of the 900-element prototype are shown in Figure 6.18.

The maximum measured gain of the large square TA antenna reaches 31.7 dBi for an efficiency of
56%. The measured 1-dB gain bandwidth of this prototype is 5.7%, as predicted through
calculations. The AR values are below 0.6 dB throughout the gain bandwidth, and below 1 dB in
the entire measurement range. The maximum-gain difference between the calculations and
measurements is larger than in the RA antenna case. This discrepancy is potentially related to the
assumption made in Section 6.3.2, and the fact that the RA element has a more compact shape,
which reduces the inter-element coupling inside the array. This effect will be further investigated
in the future.

(a) (b)
Figure 6.16. The measurement setups of the TA antenna. (a) Anechoic chamber, 316-element
array. (b) Near-field scanner, 900-element array.

134
 6.7. Prototype Measurements

(a) (b)

(c) (d)

(e) (f)
Figure 6.17. The 316-element circular TA antenna prototype. (a) Array detail. (b) Antenna
prototype in the anechoic chamber. (c) Radiation pattern in the plane 0° , at 24.6 GHz.
(d) AR versus angle, in the vicinity of the main lobe. (e) CP gain versus frequency, at 30°.
Gray lines represent the aperture-efficiency values. (f) AR versus frequency, at 30°.

135
6. RA/TA Design Process

(a)
Normalized Gain [dB]

Axial Ratio [dB]

(b) (c)
33 70% 6
60% AR sim.
32 AR meas.
5
50%
31
40% 4
30
30% 3
29
2
28

27 1
LHCP sim.
LHCP meas.
26 0
23.5 24 24.5 25 25.5 23.5 24 24.5 25 25.5
Frequency [GHz] Frequency [GHz]
(d) (e)
Figure 6.18. The 900-element square TA antenna prototype. (a) Antenna prototype in the near-
field scanner. (b) Radiation patterns in the plane 0°, at 24.6 GHz. (c) AR versus angle, in the
vicinity of the main lobe. (d) CP gain versus frequency, at 20°. Gray lines represent the
aperture-efficiency values. (e) AR versus frequency, at 20°.

136
 6.8. Summary and Discussion

6.8. Summary and Discussion

Following the design of RA and TA elements and the feeding antenna in Chapter 5, this chapter
presents the synthesis and analysis procedures for the complete RA and TA antennas, based on
the designs in Chapter 5.

A custom code is developed for the generation of the array geometry, starting from the
interpolated results of the quasi-3D EM unit-cell simulations and the corrugated-horn feed. The
arrays in this chapter use the variable rotation technique (Section 5.1.2) to convert the incident
spherical phase-fronts to a linear phase of the scattered field across the array aperture, required
for beam steering. Instead of a well-known formula (6.3), a different procedure was implemented
here to obtain the element-rotation matrix, which eliminates the phase errors for any angle of
incidence on the individual elements. The electric field, incident on the individual elements, is
obtained by sampling the interpolated far-field radiation of the feeding antenna and combining it
with the path loss. The method allows for a fast re-calculation of the incident fields, if the array
focal point changes.

The analysis portion of the code calculates the scattered fields at each array element, using the
incident fields and the element S parameters, for a specific incidence angle and the element
rotation obtained in the previous step. Different S parameters are used, depending on the
considered case (RA or TA). The far-field array radiation is calculated using the Field Equivalence
Principle applied on the radiation of rectangular apertures. The scattered fields on each element
are decomposed into two orthogonal components, and the individual contributions are
combined to find the CP performance of the full array.

As a final analysis step, the simulated array efficiency is evaluated as a combination of individual
factors, and the aperture efficiency is found to be the dominant factor. It does not depend on the
array elements, but solely on the feed and the antenna-system parameters. In the case of
deployable CubeSat arrays, these parameters are constrained by the mechanical requirements –
an offset feeding and a rectangular aperture. Under these conditions, the arrays do not reach their
maximum potential, but their performance is maximized from an engineering standpoint.
Nevertheless, the elements themselves introduce minimal losses in terms of phase, polarization
and reflection/transmission.

The calculated 1-dB gain bandwidth of the 900-element RA antenna, based on the coupled-loop
element from Chapter 5, reaches 6.5%. This result is a significant improvement over the reported
bandwidths of simple split-loop elements, originally proposed for RA CP applications. A literature
review reveals that it is somewhat easier to achieve a wideband-gain performance of single-layer
microstrip LP elements, compared to the CP elements, as the microstrip resonators can be
coupled with a larger freedom. Some CP elements report the CP bandwidths reaching 20%. These
elements are either composed of several dielectric and metallic layers, or a use a thick air layer

137
6. RA/TA Design Process

between the RA substrate and the ground plane. The novel RA element proposed in this thesis
yields a unique wide bandwidth while having a profile of only 0.042 λ0 at 24.6 GHz.

The presented aperture-coupled-patch TA element is a modification of the element from [91]. The
number of dielectric layers is reduced from five to three by modifying the slot shape in the ground
plane, thus improving the element performance versus the incidence angle. Although the total
profile is reduced to 0.13 λ0 at 24.6 GHz, the calculated 1-dB gain bandwidth of the TA antenna is
5.7%, while the efficiency reaches a value of 69%. This value is constrained by the offset feeding
and the rectangular aperture. Efficiencies of up to 73% are obtained in the calculations for a
circular array geometry.

The performance of RA and TA antennas is verified through measurements of several prototypes.

Smaller-scale array prototypes are built for compatibility with the available far-field anechoic
chamber. The plastic support, originally developed for the 900-element antennas, is used for the
measurements of all arrays. Although this configuration is suboptimal in terms of aperture
efficiency, the agreement between the measurement and calculation results confirms the
wideband performance of the novel array elements. The large 900-element square arrays are
measured in the near-field scanner. The measurement results show an improvement with respect
to small-scale prototypes, due to a larger array size and a more advantageous feed position. The
RA and TA antennas exhibit 1-dB CP-gain bandwidth of 6.8% and 5.7% and AR values smaller than
1 dB and 0.6 dB, respectively, across the entire gain bandwidth.

A combination of an extremely low profile and wideband performance is highly desired for
CubeSat applications, due to the limited antenna-stowage volume and the simplicity of the
fabrication process. Both presented designs are fabricated with a simple PCB etching process (and
bonding for the TA antenna) and require no metallized vias, SMD component soldering, or a
complicated multi-layer assembly. The array aperture can be constructed of several panels folded
at the satellite exterior, while the total thickness does not violate the CubeSat constraints.

138
7. Conclusions and Perspectives

7.1. Summary
The CubeSat has established itself as the dominant nanosatellite standard, since its conception in
1999. CubeSats have since proven their capabilities for Earth observation, RADAR, and even inter-
planetary missions, in spite of their small size. Their low development and orbital deployment
cost are the main features that attracted satellite developers. In recent years, this increased
interest has caused the fast-growing number of CubeSats for commercial applications to surpass
the number of CubeSats developed for academic purposes, for which the standard was initially
developed.

The volume constraints specified by the standard have encouraged scientists and engineers to
redesign and miniaturize conventional satellite components, for them to fit into the tiny CubeSat
chassis. Among different on-board components, CubeSat antennas are one of the most
constrained, since their physical size is directly proportional to the wavelength at the operating
frequency. This was especially important for early CubeSats, which communicated in the VHF
and UHF frequency bands, where the 1U CubeSat dimension is only /20 and /7, respectively.

The domain of CubeSat antennas has since seen a large and still growing number of innovative
antenna solutions in a wide frequency spectrum, spanning from VHF to V band. Depending on
the application and the design specifications, different methods were used to adapt the
conventional antenna geometries to the CubeSat environment. The main novelty for a majority of
the proposed geometries (primarily the exotic deployable antenna geometries) is actually in their
mechanical design, as their electrical performance is well known.

The research for this thesis is conducted with a final goal to enable different communication
aspects of an IoT constellation of CubeSats in LEO. The thesis presents several new antenna
designs, innovative in terms of both their CubeSat integration, and most importantly, their EM
performance. The proposed antennas operate in frequency bands ranging from L to K bands, and
exhibit a different performance depending on their intended application. The common design
requirement of all presented antennas is to increase or preserve the operating bandwidth under
7. Conclusions and Perspectives

a constrained or reduced available volume, respectively, while the critical limiting factor is the
total antenna thickness. The main achievements per thesis chapter are summarized in the
following paragraphs.

Chapter 2 – Wideband CP stripline-fed aperture-coupled patch antennas in S and X bands are

presented in this chapter. The primary function of the S-band antennas is the satellite TT&C, for
which a highly reliable antenna is necessary. Therefore, a low-profile solution is preferred here
over the deployable geometries. An asymmetric-stripline feeding network enables a
unidirectional radiation and isolates the antenna from the CubeSat structure, but introduces
difficulties with the antenna matching over a wide band, considering the profile limitation.
Adding a metallic shielding around the stripline structure drastically improves the antenna
impedance bandwidth. After a detailed investigation of the electric fields inside the formed
stripline cavity, it was discovered that the strong electric field magnitude below the coupling slot
boosts the coupling to the radiating patches. Design rules for this antenna type are outlined based
on this study. The wideband antenna covers up to three adjacent frequency bands
simultaneously. A unidirectional radiation pattern, a 3-dB gain bandwidth of 27% and an AR
bandwidth of 32% were obtained with a total antenna prototype profile of 0.08 at 2 GHz. In a
modified version of the antenna, the resonances of the stripline cavity are used to reduce the gain
and suppress the interference from an adjacent band. A novel method of integrating this type of
antennas in the CubeSat structure is proposed.

The chapter also emphasizes the importance of evaluating the radiating-element matching in the
case of multi-fed antennas of antenna arrays, and distinguishing it from the matching at the
feeding network input. The S-band antennas incorporate a rather complex feeding network to
enable high-quality CP over the operating bandwidth. As a means of demonstrating that the
shielded stripline technique is equally powerful with antennas having simpler feeding networks,
a CP stripline-fed patch antenna is designed in X band with a single feed line and an asymmetric
coupling slot. The same improvements over the microstrip-fed model are observed as with the S-
band antenna. An array of four sequentially rotated elements is designed with the described patch
element that achieves a gain of 15 dBi a wideband AR.

Chapter 3 – A patch-antenna system in L band is presented in this chapter for IoT 3U-CubeSat
duplex communications in LEO. Several antenna array configurations are proposed, with antenna
elements placed on the backsides of deployed solar panels, facing the Earth. A high-permittivity
dielectric substrate is used to miniaturize the size of CP patch elements until they are suitable for
placement on a CubeSat. A separate antenna element is designed for each of the communication
directions (uplink/downlink) due to a reduced impedance bandwidth. The small electrical size of
the antennas resulted in a large coupling to the asymmetric 3U ground plane, causing a distortion
of the radiation pattern. It is shown that the placement on the 3U CubeSat is critical for the
radiation performance of this type of antennas. Several independent radiating beams, scanned
away from broadside, are envisioned to increase the capacity of the IoT system. Four two-element

140
 7.1. Summary

sequentially rotated arrays are required to provide two duplex beams using the described CP
patch antenna elements. Similarly, 32 CP antennas would be necessary for four independent
beams. Instead, arrays of four sequentially rotated dual-band LP patch antennas are used for the
same purpose, reducing the number of antennas to 16.

As another method of optimizing the available real estate on a CubeSat, an L-band patch antenna
is combined with an S-band element in a stacked-patch configuration to obtain a dual-band dual-
fed CP antenna. The L-band element is actually smaller than the higher-frequency S-band
element, due to the heavy dielectric loading, and consequently located on top of the stacked
structure. Three feeding configurations are compared, and the network based on a Wilkinson
divider is found to be optimal for CubeSats in terms of size, complexity and AR. It is further
demonstrated that the choice of a feeding network does not significantly influence the CP gain
bandwidth of the antenna, even compared with a simple single-fed CP antenna.

Chapter 4 – This chapter proposes several configurations for high-gain inter-satellite-link (ISL)
antennas in K band, which enable a low-latency high-data-rate global coverage with a CubeSat
constellation. The ISL for small satellites was previously reported in literature only in low
frequencies, such as S band, and this thesis discusses the K-band CubeSat ISL for the first time.
The conventional K-band ISL antennas are either too large, heavy and/or too complex for
CubeSats. Therefore, advantage is taken from the specific flying scenario to design simpler and
robust fixed-beam antennas. In the considered constellation, the CubeSats are uniformly
distributed over distinct orbital planes and the individual satellites are connected with ISLs to
adjacent neighbors within one plane. Based on the number of satellites per orbit, the antenna
requirements in terms of CP gain and beam-steering angle are calculated and presented in this
chapter. The antenna gain is calculated versus the aperture efficiency for different aperture areas,
and it is found that the smaller face of a 3U CubeSat does not provide sufficient gain for a required
ISL data rate at K-band frequencies. Among the candidates for deployable antennas, reflectarray
(RA) and transmitarray (TA) antennas offer an excellent tradeoff between the performance,
complexity and the stowage volume. Several RA and TA ISL configurations are envisioned and
illustrated for 3U CubeSats in LEO.

Chapter 5 – This chapter presents the individual elements of the CubeSat ISL antenna system
described in Chapter 4 – one novel CP array element for each of the RA and TA antennas, and a
3D-printed CP axially-corrugated horn antenna. The CP specification allows the phase shifting
across the RA and TA antennas to be achieved using the element-rotation technique, where the
phase shift of the reflected or transmitted waves is proportional to the element rotation angle. A
unit-cell infinite-array simulation environment and the theoretical developments are presented
for the element-rotation technique. From the results, a set of conditions is obtained, which an
element must satisfy in order to qualify for this technique. Two new TA and RA elements, based
on the conventional CP elements, are designed to satisfy these conditions. The shape of the
coupling slot is modified for a TA element based on aperture-coupled patches, reducing the

141
7. Conclusions and Perspectives

required number of metallic layers from five to three and preserving a high-quality CP
performance. A novel coupled-loops RA element is designed as a modification of a known split-
loop. The dual-resonant nature of the proposed element doubles the CP bandwidth, while
preserving the simplicity and the low profile of a single-layer design. The presented modal
analysis of the element provides a deeper insight into the mechanisms behind the wideband CP
operation. An axially corrugated CP horn is designed as the feeding element for RA and TA
antennas. The axial corrugations improve the off-axis AR of the horn, and a septum polarizer
generates two CP radiation modes of the horn, depending which input port is excited. A feed-
chain prototype is fabricated for laboratory testing using additive manufacturing in aluminum
(DLMS). The presented prototype measurement results validate the simulations and confirm the
feed eligibility for ISL CubeSat applications.

Chapter 6 – The synthesis and analysis, as well as prototype measurements of RA and TA antenna
geometries is presented in this chapter. The simulated performance of the RA, TA elements, and
the feeding horn from Chapter 5 is first used to generate the array geometry. The radiation pattern
of the feed is combined with the path loss to obtain the incident fields on the array – a process that
allows a fast re-calculation of incident fields for different feed positions. Instead of a known
analytical formula suggested by the analysis in Chapter 5, the element rotation angles are
determined with a simple and fast procedure, which eliminates the phase errors for large
elevation incidence angles. The array geometry is first generated at the design frequency, and the
array far-field radiation is then calculated at several frequency points. To achieve this, the
scattered field on all elements is obtained by combining the incident fields and the unit-cell
responses for different incidence angles. The CP far field of entire arrays is finally obtained using
the Field Equivalence Principle applied on the radiation of rectangular apertures.

The achievable aperture efficiency for this type of antennas is limited by the specific CubeSat ISL
scenario. The offset rectangular RA and TA antennas conform to the CubeSat structure during
stowage and maximize the performance from an engineering standpoint, although their aperture
efficiency is reduced. A study of the overall antenna efficiency shows that the main factor for both
antennas is indeed the aperture efficiency, which only takes into account the array geometry. In
other words, the proposed RA and TA elements introduce minimal losses in terms of phase,
polarization and reflection/transmission and maximize the antenna capabilities under the given
constraints. The calculations for two 900-element RA and TA antennas show a 1-dB CP gain
bandwidth of 6.5% and 5.7% with total efficiency of 60% and 69%, respectively. This achievement
is especially important in the case of RA antennas, since the bandwidth is significantly improved
with respect to the conventional elements, without any loss of simplicity. Two smaller-scale
prototypes are fabricated for anechoic chamber measurements, whereas the prototypes of 900-
element arrays were measured in a near-field scanner. The presented measurement results
confirm the high-quality CP performance of the designed elements, in spite of a sub-optimal

142
 7.2. Discussion

aperture efficiency. The CP performance, along with the low profile and simplicity of the proposed
antennas, are the main enabling factors for ISL CubeSat applications.

7.2. Discussion
A shielded asymmetric stripline is used in Chapter 2 to generate a unidirectional radiation pattern
while preserving a good impedance matching under a low profile. The stripline cavity formed by
the shielding exhibits resonant modes and it is essential to keep the modes outside of the
operating bandwidth. The narrowband nature of these modes, and the fact that the field
distribution inside the cavity is similar below and above the resonance, allow for a certain freedom
in selecting a suitable cavity dimension. Electric-field analysis helps to understand the causes of
mismatch in case of an improperly sized stripline cavity. Very often, the required dimension will
be different from the CubeSat face, and a row of metallic screws can be used to create a cavity of
arbitrary dimensions or shape. In addition, the feeding network can be placed inside the metallic
CubeSat chassis, reducing the antenna protrusion on the surface. This integration method
requires a custom-made CubeSat chassis with slots cut for the coupling aperture between the
feeding network and the radiating patches.

The S-band low-permittivity patch antenna in Chapter 2 and the L-band high-permittivity patch
antenna in Chapter 3 are studied on a large face of a 3U CubeSat (modeled as a 300 x 100 mm2
PEC). The effect of the substrate permittivity is seen by comparing their radiation performance.
The miniaturized high-permittivity antennas couple strongly to the asymmetric ground plane and
their radiation pattern is consequently deteriorated. The pattern distortion is negligible in the S-
band case due to a low substrate permittivity, despite the extreme offset on the 3U CubeSat face.
In conclusion, the antenna placement on a CubeSat is critical in case of small antennas, and the
satellite structure, or at least parts of it, must be considered in the design phase. Furthermore, the
proper antenna placement can also be used to generate isoflux-like patterns, which is
demonstrated in Chapter 3.

Distinguishing the matching of the multi-fed antenna elements from the matching at the feeding-
network input is crucial for an accurate antenna design. This fact is well known in the antenna
domain, but often omitted in the CubeSat literature. Therefore, it is emphasized in Chapters 2 and
3. In the beginning of the design phase, it is recommended to create several intermediate antenna
models in the absence of the feeding network, and evaluate the active S parameters. Afterwards,
a suitable feeding network can be selected. The comparison of three feeding-network geometries
in Chapter 3, with varying performance and complexity, can be of assistance in this step.

RA and TA antennas for CubeSat ISL in LEO, as well as the IoT arrays in L band, are proposed
specifically for the 3U CubeSats, flying with their long axis aligned with the orbit. In the considered
scenario, the deployable flat reflectors and lenses offer an excellent compromise between the

143
7. Conclusions and Perspectives

complexity and the performance. However, this might not be the case for similar CubeSat
constellations, with satellites having different form factors (6U and more) or flight attitudes. As an
example, a patch antenna array, covering a large face of a 3U CubeSat, could achieve a gain
comparable to the presented deployable antennas. Such scenarios are not considered here, but
should definitely be taken into account if the design specifications allow.

TA antennas are for the first time proposed for CubeSats in this thesis, to the best knowledge of
the thesis author. TA antennas are less sensitive to the array flatness, compared to RA
antennas [81]. The symmetrical dielectric and metallic structure of the proposed TA element can
help to minimize the bending of the array panel with drastic changes in temperature in LEO.

The presented RA and TA elements – the coupled loops and the aperture-coupled patches – have
demonstrated their excellent CP performance, taking into account their simplicity and low profile.
These elements are equally suitable for other applications (terrestrial, GEO, etc.) in different
frequency bands.

7.3. Future Work

Both the feeding network and the patch elements of wideband antennas in S band can be
redesigned on substrates with a higher permittivity. This would reduce the total antenna
dimensions and improve the beamwidth required for TT&C applications. Although the
impedance bandwidth will be reduced by this approach, a stacked-patch configuration should be
sufficient to cover the two TT&C frequency bands.

The L-band patch antennas are too thick in their present form, for an efficient stowage of one or
several solar panels, on which the antennas are placed. Furthermore, a feeding-network substrate
required for arrays will additionally increase the total stowed profile. The antennas should be
redesigned on a thinner substrate. In combination with the sequential rotation, the arrays should
still be able to cover the entire Tx or Rx band, in terms of both impedance matching and AR.

The combined TA/RA antenna panel is a promising concept that could result in a single
deployable panel for an independent two-way dual-band ISL communication. The initial study
showed that the response of the RA, FSS and TA layers changes when they are assembled, due to
a large coupling between individual elements. The approach to this problem would be either to
analyze the entire structure simultaneously or to design individual elements that exhibit less
cross-coupling.

The measurements of the 900-element arrays showed a larger difference between the calculated
and measured maximum gain of the TA antenna, with respect to the RA antenna. The reasons for
this difference will be investigated in the future.

144
 7.4. Perspectives

7.4. Perspectives
A very low substrate permittivity (1.08) in Chapter 2 is used to achieve a CP bandwidth larger than
20% in the wideband patch antenna. However, the TT&C antenna bandwidth criteria are less
demanding (~12%). A higher feeding network and patch substrates can be selected in this case, in
order to reduce the antenna size (primarily profile), increase the beamwidth and simplify the
fabrication and assembly processes. This would require a complete redesign of the feeding
network and the radiating element. Therefore, it is not performed in this thesis, but should be used
if demanded by the volume constraints.

The presented patch antennas in Chapter 3 cover the entire allocated downlink and uplink
bandwidths. The antenna thickness is increased to alleviate the narrowband matching, caused by
a high substrate permittivity. The antennas are not immediately suitable for placement on the
backsides of solar panels, as the total thickness of stowed panels would violate the CubeSat
constraints. Once the final operating frequency is known within the bands, a thinner patch having
a narrower bandwidth can be tuned to that frequency. The conclusions of Chapter 3 would then
also apply for such patches, although these are not studied in this thesis. Otherwise, a different
substrate/thickness combination can achieve the same bandwidth performance, if the allocated
space allows.

It has been over one year since the first Astrocast CubeSat is launched with several L-band
antennas on-board. During this time, the satellite communication systems are tested and the
hardware, including the antennas, performs as predicted. This result has encouraged the use of
more complex dual-fed L-band and L/S-dual-band stacked-patch antennas for the first batch of
commercial satellites, which is to be launched during 2020. The high-gain deployable arrays in K
band are currently being evaluated by the mechanical engineers, for use in the future satellite
versions.

The total count of launched CubeSats in Figure 7.1 shows a growth explosion in the past several
years [2]. This rapid increase is driven by the increased interest of the commercial sector for
CubeSat platforms. One reason for this is that the CubeSats’ cost-efficient nature opens many new
opportunities in the space industry, which would otherwise be inaccessible with conventional
satellite technologies. Several companies have either deployed or announced LEO constellations
incorporating tens or hundreds of CubeSats, with indications that the progress will continue in
the future. The trends show that recent CubeSats have a larger form factor (6U and more) to
accommodate specialized components and modules that provide new communication functions.
The future developments in this field will likely include a higher level of component integration
and customization, which will also include the antennas.

145
7. Conclusions and Perspectives

Figure 7.1. The total number of nanosatellites and CubeSats over time [2].

Besides CubeSats, there is an even more drastic increase in the number of planned satellites in
the small-satellite class (<500 kg). Several companies have announced and started the
deployment of LEO constellations spanning from hundreds to even tens of thousands of small
satellites (SpaceX, OneWeb, Kepler). Some questions were recently raised regarding the
sustainability of such mega-constellations, considering the long lifetime of space debris (non-
operational satellites), especially at high LEO altitudes above 1000 km. Therefore, it is imperative
to incorporate a deorbiting functionality on every LEO satellite, to preserve a clean and safe space
environment in orbit. There is also an effort to use dedicated spacecraft that track and deorbit
non-responsive satellites from orbit.

The growth of the CubeSat and small-satellite industry will provide new exciting challenges and
opportunities for antenna design. As the CubeSats continue to expand their communication
functions, there will be an increased demand for innovative, performant antenna designs that
conform to the nanosatellite paradigm. Modern CubeSat antennas will most likely achieve this by
revisiting conventional antenna geometries and adapting them to the CubeSat environment, in
combination with new fabrication and integration techniques.

146
References
[1] CalPoly, “Cubesat design specification,” 2009.

[2] E. Kulu, “Nanosats Database,” 2019. [Online]. Available: https://www.nanosats.eu/database. [Accessed:

24-Jul-2019].

[3] C. Kakoyiannis and P. Constantinou, “Electrically Small Microstrip Antennas Targeting Miniaturized
Satellites: the CubeSat Paradigm,” Vienna, Austria: InTech, pp. 273–316, 2011, doi: 10.5772/14947.

[4] S. Clark, “A chat with Bob Twiggs, father of the CubeSat,” 2014. .

[5] S. Gao et al., “Antennas for small satellites,” 2008 Loughbrgh. Antennas Propag. Conf. LAPC, pp. 66–69,
2008, doi: 10.1109/LAPC.2008.4516867.

[6] NASA, “Mission Overview: GeneSat-1,” 2007. [Online]. Available:

https://www.nasa.gov/centers/ames/missions/2007/genesat1.html. [Accessed: 21-Dec-2019].

[7] R. E. Hodges, N. Chahat, D. J. Hoppe, and J. D. Vacchione, “A Deployable High-Gain Antenna Bound for
Mars: Developing a new folded-panel reflectarray for the first CubeSat mission to Mars,” IEEE Antennas
Propag. Mag., vol. 59, no. 2, pp. 39–49, 2017, doi: 10.1109/MAP.2017.2655561.

[8] S. Gao, Y. Rahmat-Samii, R. E. Hodges, and X. X. Yang, “Advanced Antennas for Small Satellites,” Proc.
IEEE, vol. 106, no. 3, pp. 391–403, 2018, doi: 10.1109/JPROC.2018.2804664.

[9] ISIS, “CubeSat Antenna System for 1U/3U.” [Online]. Available:

https://www.isispace.nl/product/cubesat-antenna-system-1u-3u/. [Accessed: 21-Dec-2019].

[10] E. Pittella et al., “Feature article: Reconfigurable S-band patch antenna system for cubesat satellites,” IEEE
Aerosp. Electron. Syst. Mag., vol. 31, no. 5, pp. 6–13, 2016, doi: 10.1109/MAES.2016.150153.

[11] N. Chahat, R. Hodges, J. Sauder, M. Thomson, and Y. Rahmat-Samii, “CubeSat Deployable Ka-Band Mesh
Reflector Antenna Development for Earth Science Missions,” 2016 IEEE Antennas Propag. Soc. Int. Symp.
APSURSI 2016 - Proc., vol. 64, no. 6, pp. 1531–1532, 2016, doi: 10.1109/APS.2016.7696472.

[12] N. Chahat, G. Agnes, J. Sauder, and T. Cwik, “One meter deployable reflectarray antenna for earth science
radars,” 2017 IEEE Antennas Propag. Soc. Int. Symp. Proc., vol. 2017-Janua, pp. 245–246, 2017, doi:
10.1109/APUSNCURSINRSM.2017.8072165.

[13] R. E. Hodges, M. J. Radway, A. Toorian, D. J. Hoppe, B. Shah, and A. E. Kalman, “ISARA - Integrated Solar
Array and Reflectarray CubeSat deployable Ka-band antenna,” IEEE Antennas Propag. Soc. AP-S Int.
Symp., vol. 2015-Octob, pp. 2141–2142, 2015, doi: 10.1109/APS.2015.7305460.

[14] A. Tatomirescu, G. F. Pedersen, J. Christiansen, and D. Gerhardt, “Antenna system for nano-satelite
mission GOMX-3,” Proc. 2016 6th IEEE-APS Top. Conf. Antennas Propag. Wirel. Commun. IEEE APWC
2016, pp. 282–285, 2016, doi: 10.1109/APWC.2016.7738178.

[15] EnduroSat, “S-band Antenna ISM.” [Online]. Available: https://www.endurosat.com/cubesat-store/all-

cubesat-modules/s-band-patch-antenna/. [Accessed: 21-Dec-2019].

[16] J. Fouany et al., “New Concept of Telemetry X-Band Circularly Polarized Antenna Payload for CubeSat,”
IEEE Antennas Wirel. Propag. Lett., vol. 16, pp. 2987–2991, 2017, doi: 10.1109/LAWP.2017.2756565.

[17] EnduroSat, “X-band Patch Antenna Array.” [Online]. Available: https://www.endurosat.com/cubesat-

store/all-cubesat-modules/x-band-4x4-patch-array/. [Accessed: 21-Dec-2019].

[18] C. J. Vourch and T. D. Drysdale, “V-band ‘bull’s eye’ antenna for cubeSat applications,” IEEE Antennas

147
 Wirel. Propag. Lett., vol. 13, pp. 1092–1095, 2014, doi: 10.1109/LAWP.2014.2327852.

[19] X. Xia et al., “NanoSats/CubeSats ADCS survey,” Proc. 29th Chinese Control Decis. Conf. CCDC 2017, pp.
5151–5158, 2017, doi: 10.1109/CCDC.2017.7979410.

[20] R. Lehmensiek, “Using the physical structure of a passive aerodynamic attitude- stabilization system as a
multi-frequency antenna on nano-satellites,” IEEE Antennas Propag. Mag., vol. 54, no. 3, pp. 39–49, 2012,
doi: 10.1109/MAP.2012.6293948.

[21] “Astrocast: The first Swiss CubeSat IoT Constellation.” .

[22] Y. Voumard, L. Etienne, and F. George, “Space Center EPFL: SwissCube Live Tracking,” 2019. .

[23] R. Radhakrishnan, W. W. Edmonson, F. Afghah, R. M. Rodriguez-Osorio, F. Pinto, and S. C. Burleigh,

“Survey of Inter-Satellite Communication for Small Satellite Systems: Physical Layer to Network Layer
View,” IEEE Commun. Surv. Tutorials, vol. 18, no. 4, pp. 2442–2473, 2016, doi:
10.1109/COMST.2016.2564990.

[24] International Telecommunication Union, “Radio Regulations Articles,” ITU Radio Regul., vol. IV, no.
Radio stations and systems, p. 435, 2012, doi: 10.1016/B978-0-12-394583-9.00020-X.

[25] W. Lan, R. Munakata, R. Nugent, and D. Pignatelli, “Poly Picosatellite Orbital Deployer Mk III Rev. E User
Guide,” 2014.

[26] S. Gao, “Antennas for Modern Small Satellites,” IEEE Antennas Propag. Mag., vol. 51, no. 4, 2009.

[27] Handbook of Frequency Allocations and Spectrum Protection for Scientific Uses: Second Edition.
Washington, DC: The National Academies Press., 2015.

[28] Y. Tsuda et al., “University of Tokyo’s CubeSat Project - It’s Educaitonal and Technological Significance,”
in 15th Annual AIAA/USU Conference on Small Satellites, 2001.

[29] A. Guillen et al., “PhoneSat in-flight experience results,” 2014.

[30] E. S. Moghaddam, N. Aboutorablad, S. Amiri, S. Nikmehr, and P. Rezaei, “Design and Analysis of a
Dualband Antenna for Small LEO Satellite Applications,” in 3rd International Conference on
Computational Electromagnetics and Its Applications, 2004, pp. 228–231.

[31] M. Samsuzzaman, M. T. Islam, S. Kibria, and M. Cho, “BIRDS-1 CubeSat Constellation Using Compact
UHF Patch Antenna,” IEEE Access, vol. 6, pp. 54282–54294, 2018, doi: 10.1109/ACCESS.2018.2871209.

[32] A. Awaludin, J. Tetuko, S. Sumantyo, S. Gao, C. E. Santosa, and M. Z. Baharuddin, “Wideband Circularly
Polarized Triangular – Ring Slot Antenna for GAIA – I Microsatellite,” Eur. Conf. Antennas Propag., pp.
2283–2286, 2017.

[33] F. E. Tubbal, R. Raad, K.-W. Chin, and B. Butters, “S-band shorted patch antenna for inter pico satellite
communications,” Telecommun. Syst. Serv. Appl. (TSSA), 2014 8th Int. Conf., pp. 1–4, 2014, doi:
10.1109/TSSA.2014.7065938.

[34] A. Nascetti, E. Pittella, P. Teo, and S. Pisa, “High-Gain S-band Patch Antenna System for Earth-
Observation CubeSat Satellites,” IEEE Antennas Wirel. Propag. Lett., vol. 14, pp. 434–437, 2015, doi:
10.1109/LAWP.2014.2366791.

[35] A. Tatomirescu and G. Pedersen, “Compact S Band Antenna for CubeSat,” 2018 12th Int. Conf. Commun.
COMM 2018 - Proc., pp. 231–234, 2018, doi: 10.1109/ICComm.2018.8430138.

[36] A. Bellion, K. Elis, and S. De Gaetano, “New compact S-band antenna for Nanosatellite TeleMetry and
TeleCommand applications - EyeSat program,” 2016 10th Eur. Conf. Antennas Propagation, EuCAP 2016,
pp. 6–10, 2016, doi: 10.1109/EuCAP.2016.7481875.

[37] J. Padilla, G. Rosati, A. Ivanov, F. Bongard, S. Vaccaro, and J. Mosig, “Multi-functional miniaturized slot

148
 antenna system for small satellites,” Antennas Propag. (EuCAP), 2011 Proc. Fifth Eur. Conf., no. 1, pp.
2170–2174, 2011.

[38] ESA Requirements and Standards Division, “Standard ECSS-E-ST-50-05C Rev. 2,” 2004.

[39] R. Manrique, G. Le Fur, N. Adnet, L. Duchesne, J. M. Baracco, and K. Elis, “Telemetry X-band antenna
payload for nano-satellites,” 2017 11th Eur. Conf. Antennas Propagation, EUCAP 2017, vol. 1, pp. 549–552,
2017, doi: 10.23919/EuCAP.2017.7928249.

[40] R. Lehmensiek, “Design of a wideband circularly polarized 2 x 2 array with shorted annular patches at X-
band on a CubeSat,” in ISAP2017, 2017, pp. 6–7.

[41] K. Elis, T. Maleszka, and A. Bellion, “TT & C and Payload Telemetry Antennas for Nanosatellite – Eye-Sat
Nanosatellite Program,” in European Conference on Antennas and Propagation (EUCAP), 2017, pp. 537–
541.

[42] E. Arnaud, L. Duchesne, K. Elis, J. Fouany, T. Monediere, and M. Thevenot, “X-band compact choke horn
antenna with circular polarization and isoflux pattern for nanosatellite applications,” Int. J. Microw. Wirel.
Technol., vol. 8, no. 3, pp. 651–659, 2019, doi: 10.1017/S1759078715000677.

[43] J. F. Zurcher, “The ssfip: a global concept for high-performance broadband planar antennas,” Electron.
Lett., vol. 24, no. 23, pp. 1433–1435, 1988, doi: 10.1049/el:19880979.

[44] D. M. Pozar, “Microstrip antenna aperture-coupled to a microstripline,” Electron. Lett., vol. 21, no. 2, p.
49, 1985, doi: 10.1049/el:19850034.

[45] S. D. Targonski, R. B. Waterhouse, and D. M. Pozar, “Design of wide-band aperture-stacked patch

microstrip antennas,” IEEE Trans. Antennas Propag., vol. 46, no. 9, pp. 1245–1251, 1998, doi:
10.1109/8.719966.
[46] J.-F. Zürcher and F. E. Gardiol, Broadband Patch Antennas. Artech House, 1995.

[47] A. Bhattacharyya, O. Fordham, and Y. Liu, “Analysis of stripline-fed slot-coupled patch antennas with vias
for parallel-plate mode suppression,” IEEE Trans. Antennas Propag., vol. 46, no. 4, pp. 538–545, 1998, doi:
10.1109/8.664118.

[48] P. Brachat and J. M. Baracco, “Dual-polarization slot-coupled printed antennas fed by stripline,” IEEE
Trans. Antennas Propag., vol. 43, no. 7, pp. 738–742, 1995, doi: 10.1109/8.391150.

[49] J. Y. Lee and S. N. Hwang, “Parallel-plate mode suppression technique for stripline-fed microstrip
antennas,” Trans. Korean Inst. Electr. Eng., vol. 57, no. 6, pp. 982–984, 2008, doi: 10.1049/el.

[50] D. Sievenpiper, H. Hsu, and R. M. Riley, “Low-Profile Cavity-Backed Crossed-Slot Antenna With a Single-
Probe Feed Designed for 2.34-GHz Satellite Radio Applications,” IEEE Trans. Antennas Propag., vol. 52,
no. 3, pp. 873–879, 2004.

[51] T. Vlasits, E. Korolkiewicz, A. Sambell, and B. Robinson, “Performance of a cross-aperture coupled single
feed circularly polarised patch antenna,” Electron. Lett., vol. 32, no. 7, pp. 3–4, 1996.

[52] C. B. Ravipati and L. Shafai, “A Wide Bandwidth Circularly Polarized Microstrip Antenna Using a Single
Feed,” in IEEE Antennas and Propagation Society International Symposium, 1999, pp. 244–247.

[53] B. Lee and T. Yang, “A Novel Wide Bandwidth CP Stacked Microstip Patch Antenna,” in IEEE Antennas
and Propagation Society International Symposium. Transmitting Waves of Progress to the Next
Millennium, 2000, pp. 1004–1007.

[54] K. Chung and A. M. Sanagavarapu, “Gain and bandwidth enhancement of a 2.4 GHz singly-fed cross-
aperture coupled patch antenna,” in IEEE Antennas and Propagation Society International Symposium,
2002, no. June 2014, doi: 10.1109/APS.2002.1016334.

[55] R. Glogowski, “Planar Antennas for Ka-Band Space Applications,” vol. 6098, 2013.

149
[56] J. M. Ll. Coll, “X-Band Antenna for Cubesat Satellite,” EPFL, 2017.

[57] T. Teshirogi, M. Tanaka, and W. Chujo, “Wideband Circularly Polarized Array Antenna with Sequential
Rotations and Phase Shift of Elements,” in Proceedings of ISAP, 1985, pp. 117–120.

[58] H. Evans, P. Gale, B. Aljibouri, E. Korolkciwicz, and A. Sambell, “Application of simulated annealing to
design of serial feed sequentially rotated 2 2 antenna array,” IET Electron. Lett., vol. 36, no. 24, pp. 1987–
1988, 2000, doi: 10.1049/el:20001407.

[59] H. Evans and A. Sambell, “ROTATED PATCH ANTENNA ARRAY WITH A SERIES FEED,” vol. 40, no. 4,
pp. 292–294, 2004, doi: 10.1002/mop.11356.

[60] P. Park and R. Elliott, “Design of collinear longitudinal slot arrays fed by boxed stripline,” 1980 Antennas
Propag. Soc. Int. Symp., vol. 18, no. 1, pp. 135–140, 1980, doi: 10.1109/APS.1980.1148243.

[61] S. Vaccaro, “SOLANT: Study and Development of Planar Antennas Integrating Solar Cells,” EPFL, 2003.

[62] X. Liu et al., “Transparent and Nontransparent Microstrip Antennas on a CubeSat: Novel low-profile
antennas for CubeSats improve mission reliability,” IEEE Antennas Propag. Mag., vol. 59, no. 2, pp. 59–
68, 2017, doi: 10.1109/MAP.2017.2655529.

[63] P. Dreyer, M. Morales-Masis, S. Nicolay, C. Ballif, and J. Perruisseau-Carrier, “Copper and transparent-
conductor reflectarray elements on thin-film solar cell panels,” IEEE Trans. Antennas Propag., vol. 62, no.
7, pp. 3813–3818, 2014, doi: 10.1109/TAP.2014.2316539.

[64] M. P. Magalhaes, M. V. T. Heckler, J. C. M. Mota, A. S. B. Sombra, and E. C. Moreira, “Design and analysis
of microstrip antenna arrays for meteorological nano-satellites for UHF uplink,” 2014 Int. Telecommun.
Symp. ITS 2014 - Proc., 2014, doi: 10.1109/ITS.2014.6947985.

[65] A. K. Skrivervik, “Design Strategies and basic antenna structures,” in Handbook on small antennas,
Universitat Politecnica de Catalunya, 2012, pp. 61–152.

[66] K. Muchalski et al., “Optimizing TT&C antenna placement on minisatellites,” 15th Int. Conf. Microwaves,
Radar Wirel. Commun. (IEEE Cat. No.04EX824), vol. 2, 2004, doi: 10.1109/MIKON.2004.1357073.

[67] G. Marrocco, L. Mattioni, A. Potenza, F. Milani, A. Giacomini, and M. Sabbadini, “Distributed multi-
function antenna system for micro- and nano-satellites,” Antennas Propag. (EuCAP), 2010 Proc. Fourth
Eur. Conf., pp. 1–4, 2010.

[68] P. S. Hall, “Application of sequential feeding to wide bandwidth, circularly polarised microstrip patch
arrays,” IEE Proc. H Microwaves, Antennas Propag., vol. 136, no. 5, p. 390, 1989, doi: 10.1049/ip-h-
2.1989.0070.

[69] J. M. de Freitas, “SPHERE3D: A Matlab Function to Plot 3-Dimensional Data on a Spherical Surface,” 2005.
[Online]. Available: https://ch.mathworks.com/matlabcentral/fileexchange/8585-3d-plot-on-a-
spherical-surface.

[70] D. M. Pozar and D. H. Schaubert, Eds., Microstrip Antennas: The Analysis and Design of Microstrip
Antennas and Arrays. New York, NY, United States: IEEE Press, 1995.

[71] Y. Yao, S. Liao, J. Wang, and K. Xue, “A New Patch Antenna Designed for CubeSat: Dual feed, L/S dual-
band stacked, and circularly polarized,” IEEE Antennas Propag. Mag., vol. 58, no. 3, pp. 16–21, 2016.

[72] F. J. Dietrich, “Globalstar satellite cellular communication system: Design and status,” Wescon Conf. Rec.,
pp. 180–186, 1997, doi: 10.1109/wescon.1997.632335.

[73] R. SOIFER and P. I. KLEIN, “Intersatellite Communication Using the AMSAT-OSCAR 6 and AMSAT-
OSCAR 7 Radio Amateur Satellite,” no. October, pp. 1974–1975, 1975.

[74] A. Tiainen, “Inter-Satellite Link Antennas,” 2017.

150
[75] P. Muri and J. McNair, “A survey of communication sub-systems for intersatellite linked systems and
cubesat missions,” J. Commun., vol. 7, no. 4, pp. 290–308, 2012, doi: 10.4304/jcm.7.4.290-308.

[76] H. B. Poza, “TDRSS Telecommunications Payload: An Overview,” IEEE Trans. Aerosp. Electron. Syst., vol.
AES-15, no. 3, pp. 414–429, 1979, doi: 10.1109/TAES.1979.308836.

[77] H. Hauschildt, S. Mezzasoma, H. L. Moeller, M. Witting, and J. Herrmann, “European data relay system
goes global,” 2017 IEEE Int. Conf. Sp. Opt. Syst. Appl. ICSOS 2017, pp. 15–18, 2017, doi:
10.1109/ICSOS.2017.8357204.

[78] O. P. Gupta, “Iridium: A Global Communication Network,” p. 7, 2007.

[79] G. Bonin, N. H. Roth, S. Armitage, J. Newman, B. Risi, and R. E. Zee, “CanX–4 and CanX–5 Precision
Formation Flight: Mission Accomplished!,” in 29th Annual AIAA/USU Conference on Small Satellites,
2015, doi: 10.1016/j.jcrysgro.2014.02.053.

[80] “Iridium NEXT Overview,” 2018. [Online]. Available: https://aerospacedefenseforum.org/wp-

content/uploads/2018/04/Iridium-18-03-23-Slides.pdf.
[81] J. Huang and J. A. Encinar, Reflectarray Antennas. Wiley Interscience, 2008.

[82] L. Dussopt, “Transmitarray Antennas,” in Aperture Antennas for Millimeter and Sub-Millimeter Wave
Applications, A. Boriskin and R. Sauleau, Eds. Springer, 2018, pp. 191–220.

[83] T. Smith, U. Gothelf, O. S. Kim, and O. Breinbjerg, “An FSS-backed 20/30 GHz circularly polarized
reflectarray for a shared aperture L- and Ka-band satellite communication antenna,” IEEE Trans.
Antennas Propag., vol. 62, no. 2, pp. 661–668, 2014, doi: 10.1109/TAP.2013.2292692.

[84] F. Xue and H. Wang, “Design of a Novel Ku / X-Band Reflectarray / Transmit-Array Antenna with
Frequency Selective Surface,” Prog. Electromagn. Res. C, vol. 73, no. March, pp. 17–25, 2017.

[85] H. Hasani, J. S. Silva, J. R. Mosig, and M. Garcia-Vigueras, “Dual-band 20/30 GHz circularly polarized
transmitarray for SOTM applications,” 2016 10th Eur. Conf. Antennas Propagation, EuCAP 2016, pp. 5–7,
2016, doi: 10.1109/EuCAP.2016.7481869.

[86] F. Yang, R. Deng, S. Xu, and M. Li, “Design and Experiment of a Near-Zero-Thickness High-Gain
Transmit-Reflect-Array Antenna Using Anisotropic Metasurface,” IEEE Trans. Antennas Propag., vol. 66,
no. 6, pp. 2853–2861, 2018, doi: 10.1109/TAP.2018.2820320.

[87] M. Wang, S. Xu, F. Yang, and M. Li, “A 1-Bit Bidirectional Reconfigurable Transmit-Reflect-Array Using a
Single-Layer Slot Element With PIN Diodes,” IEEE Trans. Antennas Propag., vol. 67, no. 9, pp. 6205–6210,
2019, doi: 10.1109/tap.2019.2925925.

[88] D. M. Pozar, S. D. Targonski, and H. D. Syrigos, “Design of millimeter wave microstrip reflectarrays,” IEEE
Trans. Antennas Propag., vol. 45, no. 2, pp. 287–296, 1997, doi: 10.1109/8.560348.

[89] A. H. Abdelrahman, A. Z. Elsherbeni, and F. Yang, “Transmission phase limit of multilayer frequency-
selective surfaces for transmitarray designs,” IEEE Trans. Antennas Propag., vol. 62, no. 2, pp. 690–697,
2014, doi: 10.1109/TAP.2013.2289313.

[90] L. Di Palma, A. Clemente, L. Dussopt, R. Sauleau, P. Potier, and P. Pouliguen, “Circularly Polarized
Transmitarray With Sequential Rotation in Ka-Band,” IEEE Trans. Antennas Propag., vol. 63, no. 11, pp.
5118–5124, 2015.

[91] R. H. Phillion and M. Okoniewski, “Lenses for circular polarization using planar arrays of rotated passive
elements,” IEEE Trans. Antennas Propag., vol. 59, no. 4, pp. 1217–1227, 2011, doi:
10.1109/TAP.2011.2109694.

[92] J. Huang and R. J. Pogorzelski, “A ka-band microstrip reflectarray with elements having variable rotation
angles,” IEEE Trans. Antennas Propag., vol. 46, no. 5, pp. 650–656, 1998, doi: 10.1109/8.668907.

151
[93] J. Perruisseau-Carrier, “MICROWAVE PERIODIC STRUCTURES BASED ON
MICROELECTROMECHANICAL SYSTEMS ( MEMS ) AND MICROMACHINING TECHNIQUES,” vol.
3969, 2007.

[94] P. Naseri, S. A. Matos, J. R. Costa, and C. A. Fernandes, “Phase-Delay Versus Phase-Rotation Cells for
Circular Polarization Transmit Arrays-Application to Satellite Ka-Band Beam Steering,” IEEE Trans.
Antennas Propag., vol. 66, no. 3, pp. 1236–1247, 2018, doi: 10.1109/TAP.2017.2787540.

[95] S. V. Hum, “Reflectarrays,” in Aperture Antennas for Millimeter and Sub-Millimeter Wave Applications, A.
Boriskin and R. Sauleau, Eds. Springer, 2018, pp. 143–190.

[96] C. Han and K. Chang, “Ka-band reflectarray using ring elements,” Electron. Lett., vol. 39, no. March, pp.
491–493, 2003.

[97] B. Strassner, C. Han, and K. Chang, “Circularly polarized reflectarray with microstrip ring elements having
variable rotation angles,” IEEE Trans. Antennas Propag., vol. 52, no. 4, pp. 1122–1125, 2004, doi:
10.1109/TAP.2004.825635.

[98] C. Han, C. Rodenbeck, J. Huang, and K. Chang, “A C/Ka dual frequency dual layer circularly polarized
reflectarray antenna with microstrip ring elements,” IEEE Trans. Antennas Propag., vol. 52, no. 11, pp.
2871–2876, 2004, doi: 10.1109/TAP.2004.835144.

[99] P. Naseri, R. Mirzavand, and P. Mousavi, “Dual-Band Circularly Polarized Transmit-Array Unit- Cell At X
and K Bands,” 2016 10th Eur. Conf. Antennas Propag., pp. 1–4, doi: 10.1109/EuCAP.2016.7481440.

[100] D. T. McGrath, “Planar Three-Dimensional Constrained Lenses,” IEEE Trans. Antennas Propag., vol. 34,
no. 1, pp. 46–50, 1986, doi: 10.1109/TAP.1986.1143726.

[101] D. M. Pozar, “Flat lens antenna concept using aperture coupled microstrip lpatches,” Electron. Lett., vol.
32, no. 23, pp. 2109–2111, 1996.

[102] E. G. Plaza, G. León, S. Loredo, and F. Las-Heras, “Dual polarized transmitarray lens,” 8th Eur. Conf.
Antennas Propagation, EuCAP 2014, no. EuCAP, pp. 2305–2308, 2014, doi: 10.1109/EuCAP.2014.6902276.

[103] V. Rathi, G. Kumar, and K. P. Ray, “Improved coupling for aperture coupled microstrip antennas,” IEEE
Trans. Antennas Propag., vol. 44, no. 8, pp. 1196–1198, 1996, doi: 10.1109/8.511831.

[104] T. G. Smith, N. Vesterdal, U. Gothelf, O. S. Kim, and O. Breinbjerg, “An Offset-Fed 20 / 30 GHz Dual-Band
Circularly Polarized Reflectarray Antenna,” 2013 7th Eur. Conf. Antennas Propag., pp. 2959–2962, 2013.

[105] T. A. Milligan, Modern Antenna Design. John Wiley & Sons, Inc., 2005.

[106] M. H. Cheng and G. N. Tsandoulas, “A Wide-Band Square-Waveguide Array Polarizer,” IEEE Trans.
Antennas Propag., pp. 389–391, 1973.

[107] J. Esteban and J. M. Rebollar, “Field Theory C.A.D. of Septum OMT-Polarizers,” in IEEE Antennas and
Propagation Society International Symposium 1992 Digest, 1992, pp. 2146–2149.

[108] A. A. Kirilenko, D. Y. Kulik, L. A. Rud, V. I. Tkachenko, and N. Herscovici, “Electromagnetic modeling and
design of dual-band septum polarizers,” Appl. Comput. Electromagn. Soc. J., vol. 21, no. 2, pp. 155–162,
2006.

[109] T.-H. Chio, G.-L. Huang, and S.-G. Zhou, “Application of Direct Metal Laser Sintering to Waveguide-
Based Passive Microwave Components, Antennas, and Antenna Arrays,” Proc. IEEE, vol. 105, no. 4, pp.
632–644, 2016, doi: 10.1109/jproc.2016.2617870.

[110] J. Nourinia, C. Ghobadi, B. Mohammadi, and F. Alizadeh, “Aperture Efficiency Study of Square Reflect
Array Antennas,” Wirel. Eng. Technol., no. 9, pp. 66–78, 2018, doi: 10.4236/wet.2018.93006.

152
Publications
[J1] M. J. Veljovic and A. K. Skrivervik, “Aperture-Coupled Low-Profile Wideband Patch
Antennas for CubeSat,” IEEE Transactions on Antennas and Propagation, vol. 67, no. 5,
pp. 3439–3444, 2019.
[J2] M. J. Veljovic and A. K. Skrivervik, “Wideband Single-Layer Circularly Polarized
Reflectarray Antenna for CubeSat Inter-Satellite Links,” submitted to IEEE Antennas and
Wireless Propagation Letters.
[J3] M. J. Veljovic and A. K. Skrivervik, “Low-Profile Aperture-Coupled-Patch Transmitarray
Antenna for CubeSat Inter-Satellite Links,” submitted to IEEE Antennas and Wireless
Propagation Letters.
[C1] M. J. Veljovic and A. K. Skrivervik, “Wideband cavity-backed CubeSat antenna in S band,”
12th Eur. Conf. Antennas Propagation, EuCAP 2018, London, UK, 2018.
[C2] M. J. Veljovic and A. K. Skrivervik, “Patch Antenna System for CubeSats in L band,” 13th
Eur. Conf. Antennas Propagation, EuCAP 2019, Krakow, Poland, 2019.
[C3] M. J. Veljovic and A. K. Skrivervik, “A Comparison of Three Feeding Networks for CubeSat
L/S-dual-band Stacked-Patch Antennas,” 13th Eur. Conf. Antennas Propagation, EuCAP
2019, Krakow, Poland, 2019.
[C4] F. Qaderi, A. Muller, A. Krammer, M. J. Veljovic, Z. Ollmann, M. Hayati, A. Skrivervik, A.
Schueler, T. Feurer, A. Ionescu,, “Millimeter-wave-triggering of insulator-to-metal
transition in Vanadium dioxide,” 2019 44th Int. Conf. Infrared, Millimeter, Terahertz
Waves, Paris, France, 2019.
[C5] M. J. Veljovic and A. K. Skrivervik, “Circularly Polarized Axially Corrugated Feed Horn for
CubeSat Reflectarray Applications,” in 14th European Conference on Antennas and
Propagation, EuCAP 2020, Copenhagen, Denmark, 2020.
[C6] A. J. M. Montes, I. V. Triviño, M. Bosiljevac, M. J. Veljovic, Z. Sipus, and A. K. Skrivervik,
“Antenna for a Cranial Implant: Simulation Issues and Design Strategies,” 14th European
Conference on Antennas and Propagation, EuCAP 2020, Copenhagen, Denmark, 2020.
[C7] A. K. Skrivervik, A. J. M. Montes, I. V. Triviño, M. Bosiljevac, M. J. Veljovic, and Z. Sipus,
“Antenna design for a cranial implant,” in 2020 International Workshop on Antenna
Technology (iWAT), Bucharest, Romania, 2020.
[W1] A. K. Skrivervik, M. J. Veljovic, “CubeSat Antennas,” a tutorial given at the ICECOM
Satellite Workshop, 23rd International Conference on Applied Electromagnetics and
Communications, ICECOM 2019, Dubrovnik, Croatia, 2019.

153
Curriculum Vitae

EDUCATION
2015 – present PhD EPFL, Lausanne
Microwave and Antenna Group (former LEMA)
 PhD thesis: “Antennas for CubeSat Communication”
Thesis advisor: Prof. Anja Skrivervik
 Research on cost-efficient antennas for satellite communications:
CubeSat antennas, SATCOM waveguide arrays
 Antenna design for Astrocast: the first Swiss nanosatellite network
 Low-profile, multi-layer antenna design, frequencies from L to Ka band
 PCB antenna manufacturing and assembly, nanosatellite antenna
integration, VNA and anechoic chamber measurements

2010 – 2015 Master & Bachelor School of Electrical Engineering, Belgrade

Department of Telecommunications and Information Technology
Module Microwave Engineering
 Master thesis: “Electromagnetic wave propagation in the vicinity of
humans”
 Average grade: 10/10 (master), 9.41/10 (bachelor)
 Undergraduate laboratory assistant (2011-2014): Conducting student
laboratory exercises in electrical engineering fundamentals
 Student project: Microwave amplifier design, fabrication and testing at
2.45 GHz

EXPERIENCE
2015 – present Doctoral assistant EPFL, Lausanne
Teaching: antennas and radiation course
Conducting laboratory experiments: laboratory on microwaves
Supervision of five semester and bachelor student projects

Mar – Sep 2015 Internship Institute for Microwaves and Electronics, Belgrade
3-D design and prototype measurements of microwave couplers and baluns

Sep – Oct 2013 Internship WIPL-D, Belgrade

Software design company for 3-D electromagnetic simulation
 Modeling of car shells
 Optimization of waveguide filters

155
 PROFESSIONAL COURSES
Sep 11-15, 2017 Frequency Domain Techniques University of Florence, Italy
 Methodology and limitations of computational techniques for
scattering and antenna problems

Jul 4-8, 2016 Millimeter-Wave Antennas IETR, Rennes, France

 Antenna structures and specific technologies for industrial applications
at millimeter-wave frequencies

Mar 14-18, 2016 Antennas for Space Applications European Space Agency
 Design requirements and solutions for space applications
 An overview of space antennas – ground and space segment

SELECTED SCHOLARSHIPS
2015 – 2016 Doctoral Fellowship
EPFL Electrical Engineering Doctoral School
 First-year fellowship given to a single PhD student per year

2015 Undergraduate/Pre-graduate Scholarship

IEEE Microwave Theory and Techniques Society (MTT-S)
 A scholarship for the master thesis research project

SKILLS
Software Ansys HFSS and Designer, CST, WIPL-D
AWR Microwave Office, Altium Designer
MATLAB

Language English: Proficient

French: Basic
Russian: Basic
Serbian: Native

PERSONAL ACTIVITIES
2011 – present Rowing
Belgrade University Rowing Club, Rowing Club Lausanne
 Competing at regional and international regattas in an 8-seat boat
 Assistant of the coach

2003 – present Clarinet

Music School in Belgrade, Folklore Ensemble Orchestra
 Performing with the orchestra at local concerts
 Touring at international music festivals

156