Thomas Deckmyn Waveguide Cavity-Backed Antenna Array Design of a 60 GHz Hybrid Mode Substrate Integrated Academic year 2014-2015 Faculty of Engineering and Architecture Chairman: Prof. dr. ir. Daniël De Zutter Department of Information Technology Master of Science in Electrical Engineering Master's dissertation submitted in order to obtain the academic degree of Counsellors: Ir. Martijn Huynen, Ir. Gert-Jan Stockman, Ir. Sam Agneessens Supervisors: Prof. dr. ir. Dries Vande Ginste, Prof. dr. ir. Johan Bauwelinck
70
Embed
Design of a 60 GHz Hybrid Mode Substrate Integrated ...lib.ugent.be/fulltxt/RUG01/002/224/391/RUG01-002224391_2015_0001... · Design of a 60 GHz Hybrid Mode Substrate Integrated Waveguide
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Thomas Deckmyn
Waveguide Cavity-Backed Antenna ArrayDesign of a 60 GHz Hybrid Mode Substrate Integrated
Academic year 2014-2015Faculty of Engineering and ArchitectureChairman: Prof. dr. ir. Daniël De ZutterDepartment of Information Technology
Master of Science in Electrical EngineeringMaster's dissertation submitted in order to obtain the academic degree of
Counsellors: Ir. Martijn Huynen, Ir. Gert-Jan Stockman, Ir. Sam AgneessensSupervisors: Prof. dr. ir. Dries Vande Ginste, Prof. dr. ir. Johan Bauwelinck
Thomas Deckmyn
Waveguide Cavity-Backed Antenna ArrayDesign of a 60 GHz Hybrid Mode Substrate Integrated
Academic year 2014-2015Faculty of Engineering and ArchitectureChairman: Prof. dr. ir. Daniël De ZutterDepartment of Information Technology
Master of Science in Electrical EngineeringMaster's dissertation submitted in order to obtain the academic degree of
Counsellors: Ir. Martijn Huynen, Ir. Gert-Jan Stockman, Ir. Sam AgneessensSupervisors: Prof. dr. ir. Dries Vande Ginste, Prof. dr. ir. Johan Bauwelinck
Preface
First and foremost, I would like to express my gratitude towards prof. dr. ir. Dries Vande
Ginste and prof. dr. ir. Johan Bauwelinck for providing me with the opportunity to carry out
this research at the Electromagnetics Group and the INTEC design group of the Department
of Information Technology. I thank Prof. Vande Ginste for sharing his expertise and giving
splendid advice, and for his never-ending positivism and support throughout the entire year. I
am very grateful as well for the insights and encouragements Prof. Bauwelinck offered.
Two people I would like to thank very profoundly are ir. Martijn Huynen and
ir. Gert-Jan Stockman. Their excellent guidance and exhaustive feedback were indispensable up
until the very last minute. Thanks to both of you, you helped pave the way towards a gratifying
conclusion of five years of studies.
Special thanks goes out to ir. Sam Agneessens for the assistance with the design of the hybrid
mode SIW antenna array. His expertise and proficient counseling enabled me to successfully
complete the design.
Ir. Niels Lambrecht deserves special thanks as well, for offering theoretical and practical
advice on more than one front. Furthermore, I appreciate the helping hand that was extended
to me by everyone at the research group, whenever I needed it.
My fellow thesis students Bob Mertens, Alexander Vindelinckx, Kristof Baes, Jorn Marievoet,
Piet Merckx and Enrico Massoni also deserve an acknowledgement. Thank you for the positive
vibe in the thesis room and the pleasant days (and evenings) I have spent there with you all.
I would like to thank my parents for their support throughout the entire period of my studies.
They have always believed in me and supported every choice I made. Without them, I wouldn’t
be where I am today. Lastly, but definitely not least, a special thank you is in order for Bieke
Keysabyl, for providing moral support when it was needed most.
Thomas Deckmyn, May 2015
Admission to Loan
The author gives permission to make this master’s dissertation available for consultation and
to copy parts of this master’s dissertation for personal use. In the case of any other use, the
limitations of the copyright have to be respected, in particular with regard to the obligation to
state expressly the source when quoting results from this master dissertation.
Thomas Deckmyn, May 2015
Design of a 60 GHz Hybrid Mode SubstrateIntegrated Waveguide Cavity-Backed
Antenna Arrayby
Thomas DECKMYN
Master’s Dissertation submitted to obtain the academic degree of
Master of Science in Electrical Engineering
Academic 2014–2015
Promoters: Prof. dr. ir. Dries VANDE GINSTE, Prof. dr. ir. Johan BAUWELINCK
Supervisors: Ir. Martijn HUYNEN, Ir. Gert-Jan STOCKMAN, Ir. Sam AGNEESSENS
Faculty of Engineering and Architecture
Ghent University
Department of Information Technology
Chairman: Prof. dr. ir. Daniel DE ZUTTER
Summary
The goal of this master’s dissertation is to develop a highly compact and integratable antenna
array that operates in the 60 GHz band, whilst maintaining compatibility with standard printed
circuit board manufacturing processes. Two different antenna technologies, i.e., microstrip patch
and Substrate Integrated Waveguide (SIW), are thoroughly analyzed through extensive simu-
lation procedures and measurements. This leads to the formulation of a founded opinion that
SIW is the best suited technology to leverage in the array configuration. Moreover, to en-
hance the impedance bandwidth of the inherently band limited cavity-backed SIW antennas,
a technique based on the excitation of hybrid modes is exploited. The integration aspect of
this dissertation is fortified by selecting a 50 µm flexible substrate material for the design of the
hybrid mode SIW antenna and the array. Although the substrate material is extremely thin, a
fractional impedance bandwidth of 3.6% is achieved. An antenna gain and directivity of 7.2 dBi
and 12.0 dBi, respectively, are obtained by constructing a four-element Uniform Linear Array
Design of a 60 GHz Hybrid Mode Cavity-BackedSubstrate Integrated Waveguide Antenna Array
Thomas Deckmyn
Supervisors: prof. dr. ir. D. Vande Ginste, prof. dr. ir. J. Bauwelinck, ir. G.-J. Stockman, ir. M. Huynenand dr. ir. S. Agneessens
Abstract— The goal of this master’s dissertation is to develop a highlycompact and integratable antenna array that operates in the 60 GHz band,whilst maintaining compatibility with standard printed circuit board man-ufacturing processes. Two different antenna technologies, i.e., microstrippatch and Substrate Integrated Waveguide (SIW), are thoroughly analyzedthrough extensive simulation procedures and measurements. This leads tothe formulation of a founded opinion that SIW is the best suited technologyto leverage in the array configuration. Moreover, to enhance the impedancebandwidth of the inherently band limited cavity-backed SIW antennas, atechnique based on the excitation of hybrid modes is exploited. The inte-gration aspect of this dissertation is fortified by selecting a 50 µm flexiblesubstrate material for the design of the hybrid mode SIW antenna and thearray. A simulated fractional impedance bandwidth of 3.6% is achieved. Asimulated antenna gain and directivity of 7.2 dBi and 12.0 dBi, respectively,are obtained by constructing a four-element Uniform Linear Array (ULA).
NOWADAYS, the omnipresent use and rapid evolution ofelectronic devices puts ever rising demands on the hard-
ware engineer of today. The swift development towards higherbitrates, to satisfy the presently unquenchable mobile user, posesnovel challenges in terms of bandwidth. Moreover, higher oper-ating frequencies are explored, which brings about the addedpredicament of high frequency effects. Paired with the vastminiaturization of high speed electronics, the design task at handbecomes ever more complicated.
This master’s dissertation focuses on the design of a band-width enhanced Substrate Integrated Waveguide (SIW) antennaarray, operating in the 60 GHz band. Hybrid modes are excitedinside the cavity and merged in the desired frequency range, no-tably increasing the impedance bandwidth. The need for minia-turization is tackled by designing the antenna array on extremelythin, i.e., 50 µm and 100 µm, flexible substrate materials.
In this abstract, first, the bandwidth enhancement techniquebased on hybrid modes is discussed (Section II). The designof the hybrid mode cavity-backed SIW antenna is considered inSection III and in Section III-B a Uniform Linear Array (ULA)is constructed utilizing the previously designed antenna ele-ments. Measurement results are presented in Section IV. Con-clusions and future research are discussed in Section V.
II. HYBRID MODE EXCITATION
The limited bandwidth of an SIW antenna can be amelioratedby simultaneously exciting two distinct resonances, i.e., hybridmodes, inside the cavity. By merging these hybrid modes withinthe desired frequency range, the impedance bandwidth is signif-icantly enhanced. The two resonances are, in essence, two dif-
(a)
(b)
Fig. 1. Field distribution in the SIW cavity: (a) Dominant E-field distribution oflower hybrid mode in largest half cavity; (b) Dominant E-field distributionof higher hybrid mode in smallest half cavity [1].
ferent combinations of a TE110 and TE120 mode. By offsettingthe slot from the center of the cavity, two half parts with differ-ent dimensions are created. The lower frequency hybrid modeis dominant in the largest half cavity and is a combination ofa strong TE110 and a weak TE120, as depicted by the E-fielddistributions in Figure 1(a). The total E-field is in phase in bothhalf cavities, but radiation can still be evoked due to the high dif-ference in magnitude. The higher hybrid mode is dominant inthe smallest half part and consists of a strong TE120 and a weakTE110, as illustrated in Figure 1(b). Here, the field is out ofphase in both half parts. A large electric field is present accrossthe slot, hence radiation is generated.
III. DESIGN OF THE HYBRID MODE SIW ANTENNA ARRAY
A. Hybrid Mode SIW Antenna
A general configuration for a hybrid mode SIW antenna isdepicted in Figure 2. The cavity dimensions Lc and Wc aredetermined as such that the resonant frequency of the TE110
is 60 GHz. The length of the slot Ls is much larger than λ2 ,
hence it is non-resonant, and the width Ws tunes the impedancematching to some extent. Here, it is opted for dc,u > dc,l; thehigher frequency hybrid mode is dominant in the lower half partof the cavity.
If dc,u increases, the resonating area of the dominant field ofthe lower hybrid mode is enlarged, hence its resonant frequencydecreases. The area of the weak field of the higher frequency hy-brid mode increases as well, but only causes a minor drop in fre-quency. Hence, the lower hybrid mode decreases in frequency,while the higher hybrid mode remains practically unchanged.This causes an enhancement of the bandwidth.
Ls Ws
dc,l
dc,u
Lc
Wc
Fig. 2. General configuration of a hybrid mode SIW antenna.
After several design iterations and ample optimization, thesimulated reflection coefficient for the design on the 50 µm sub-strate is as presented in Figure 3. Two distinct resonances areperceived at 59.16 GHz and 60.38 GHz. The total impedancebandwidth amounts to 2.2 GHz, which corresponds to a frac-tional bandwidth of 3.6% at 60 GHz.
58 59 60 61 62
0
-5
-10
-15
-20
-25
-30
-35
59.16 GHz 60.38 GHz
2.2 GHz
Frequency [GHz]
|S11|[
dB]
Fig. 3. Simulated reflection coefficient of the designed hybrid mode SIW an-tenna.
B. Uniform Linear Array Configuration
An additional benefit of adopting the SIW technology for thedesign of the antenna array, is the high isolation due to the metal-lic via rows [2]. The array elements can be spaced very closelytogether, without inducing high mutual coupling. The UniformLinear Array (ULA) is constructed using four hybrid mode SIWantennas, while reusing the via wall for adjacent elements. Thesimulated reflection coefficient of the SIW antennas in the ULAconfiguration very closely resembles that of a separate element,as in Figure 3. A simulated antenna gain and directivity of7.2 dBi and 12.0 dBi, respectively, is obtained.
IV. MEASUREMENTS
The fabricated hybrid mode SIW antenna array is depictedin Figure 4. As Uniform Thickness Copper Plating (UTCP) ofthe vias is not feasible for such a small quantity of prototypes,grounding of the top layer is achieved by injecting conductivepaste into the holes. A detailed inspection of the array revealsthat residue of the conductive paste has completely filled the in-sets and part of the slot. Attempts have been made to clear thespillage from both. Still, some remains, as illustrated in Fig-ure 5.
Measurements are performed using an N5242A Pro-grammable Network Analyzer (PNA-X) from Keysight Tech-nologies and indicate that a standing wave is present on thefeed network of the array. This implies that the matching atboth ends of the feed structure is poor. The insets procure a50 Ω Grounded Coplanar Waveguide (GCPW) structure that isperfectly matched to the feed line. Taking into account the re-maining spillage of the paste in the slot, it is clear that the di-mensions of the insets have changed, effectively altering theimpedance matching. The electric performance of the paste bywhich the vias are grounded is not precisely known at 60 GHzeither, hence it is plausible that the impedance of the SIW an-tenna elements is changed significantly.
Fig. 5. Detail of fabricated hybrid mode SIW antenna array.
Measurements indicate that the dimensions of the antenna el-ements and 50 Ω microstrip lines are accurate, hence a fabrica-tion inaccuracy is not to blame for the poor matching. The elec-trical parameters of the substrate material are characterized at1 MHz in [3], thus the loss tangent and permittivity could havenotably changed at 60 GHz. This discrepancy is likely to con-
tribute to the deteriorating of the impedance matching. More-over, due to the extremely thin substrate material, the perfor-mance of the press fit of the connector is likely to have dimin-ished; a good connection is not unquestionably ensured.
To gain more insight into the effects at hand, the characteris-tics of a 50 Ω reference microstrip line are measured. To quan-tify the amount of additional losses introduced by the combina-tion of connector and substrate, the power balance of the lineis calculated, i.e., |S11|2 + |S21|2. For a lossless transmissionline structure, this should equal unity, indicating that all incidentpower is either reflected or transmitted. Calculations reveal thatapproximately 70% of the incident power is dissipated. This isdue to losses in the copper and substrate, as well as to radiation.This confirms that it is indeed the combination of connector is-sues and an insufficiently characterized substrate material thatare to blame for the faulty operation of the antenna system.
V. CONCLUSIONS AND FUTURE RESEARCH
A highly compact and integratable hybrid mode cavity-backed SIW antenna array for the 60 GHz band was succesfullydesigned. Bandwidth enhancement was achieved by exploitingthe excitation of hybrid modes inside the SIW cavity. Elevatedgain and directivity was procured by utilizing a four-elementULA configuration. The integration aspect was pushed to theutmost extent by selecting extremely thin substrates for the de-sign.
The insufficient characterization of the flexible substrate at60 GHz and the non-ideal processing of the minute vias wasdetrimental for the operation of the fabricated prototypes. Fu-ture research on this topic could certainly encompass investigat-ing the characteristics of flexible substrates for antenna design at60 GHz. Even research towards the development of novel flexi-ble substrate materials for use in the Extremely High Frequency(EHF) band could be performed.
In future work, the combination of the hybrid mode SIW arraywith very high speed phase shifters can yield the development ofadaptive antenna systems at 60 GHz. This can have applicationsin, e.g., millimeter wave radar detection systems.
REFERENCES
[1] G. Q. Luo and Z. F. Hu and W. J. Li and X. H. Zhang and L. L. Sun and J.F. Zheng, “Bandwidth-Enhanced Low-Profile Cavity-Backed Slot Antennaby Using Hybrid SIW Cavity Modes”, IEEE Transactions on Antennas andPropagation, vol. 60, no. 4, pp. 1698 - 1704, April 2012.
[2] M. Bozzi and A. Georgiadis and K. Wu, “Review of Substrate IntegratedWaveguide Circuits and Antennas”, IET Microwaves, Antennas and Prop-agation, vol. 5, no. 8, pp. 909 - 920, September 2011.
As stated above, SIWs closely resemble conventional rectangular waveguides, be it in planar
form, but SIW structures also display similar propagation characteristics. The guided modes
are practically equal to the TEn0-modes that propagate in a rectangular waveguide with similar
dimensions, where the dominant mode is TE10. Due to this high similarity, empirical formulas
have been presented that allow one to obtain the effective width weff of a rectangular waveguide
with the same propagation characteristics as the SIW structure. It holds that [9]
weff = w − d2
0.95s, (2.11)
where d and s are the diameter of the vias and the spacing between the vias respectively. The
frequency of the first order guided mode in an SIW structure can be obtained as the frequency
of the TE10 guided mode in a rectangular waveguide of width weff, and hence equals [3]
f10 =c
2√εrweff
(2.12)
=c
2√εr
(w − d2
0.95s
)−1
. (2.13)
The similarity only holds provided that the radiation leakage through the gaps between the
vias is neglegible, i.e., the row of vias needs to approximate a contiguous metallic wall at the
operating frequency. If this is the case, the SIW structure can be treated as being a rectangular
waveguide with width weff as defined in (2.11). Empirical constraints on the dimensions of and
the spacing between the vias such that radiation leakage is neglegible, have been formulated
in [10] as
s ≤ 2d
d ≤ λ10 ,
(2.14)
with λ the wavelength in the substrate material.
2.3.2 Cavity-Backed SIW Antenna
A cavity-backed SIW antenna consists of a cavity resonator constructed in SIW technology. If
the cavity is made leaky by introducing one or more slots, radiation occurs. The cavity is built
up as depicted in Figure 2.5 and is excited using a microstrip feed line. The resonant frequency
of the TEmn0-mode can be found as [11]
fmn0 =c√
2πεr
√(mπ
Weff
)2
+
(nπ
Leff
)2
, (2.15)
where Leff and Weff can be obtained using (2.11), c is the speed of light in vacuum and εr is the
relative permittivity of the substrate material. As this is an empirical expression, tuning will
ANTENNA DESIGN ASPECTS 9
be necessary to obtain the desired operating frequency. Note that the restrictions on the via
spacing and diameter (2.14) need to be fulfilled to ensure proper operation.
LW
Ls WsRgcpw
dgcpw
Figure 2.5: SIW resonant cavity with microstrip line feed.
Once the cavity is constructed at the right resonant frequency, a slot is added to evoke radiation.
The slot can be etched in the ground plane, which will isolate the radiation caused by the feeding
network from the desired radiation coming from the slot, or it can be etched in the top copper
plane, which will drastically lower backward radiation.
The slot is an important tuning element of the cavity-backed SIW antenna. As described in [12],
the width Ws can be tuned to obtain optimal impedance bandwidth, but the effect on impedance
bandwidth is much less than that of the substrate thickness. The length of the slot Ls can be
used to adjust the resonant frequency by some degree, but the resonance is primarily determined
by the cavity size. The length of the slot also has major influence on the radiation efficiency,
which is at its maximum when the slot length is half a wavelength.
To match the edge impedance of the cavity to the 50 Ω microstrip feed line, the parameters
Rgcpw and dgcpw need to be chosen such that a 50 Ω grounded coplanar waveguide (GCPW)
structure is obtained.
2.4 Simulation Software
When designing millimeter wave antenna systems, one wants to predict the real behaviour of the
device as accurately as possible. A good simulation tool that takes into account all high frequency
effects is thus indispensable. For all designs that follow, the Keysight Technologies Advanced
Design System (ADS) simulation software is used. It incorporates a planar 3D electromagnetic
solver Momentum, which uses a frequency-domain Method of Moments (MoM) to simulate
complete board structures whilst taking parasitic effects, losses and coupling into account.
DEEMBEDDING OF CONNECTOR AND FEED LINE STRUCTURE 10
Chapter 3
Deembedding of Connector and Feed
Line Structure
To connect the antenna test structures that will be designed in the next chapter to measurement
equipment or an exterior circuit, a suitable connector will be needed, i.e. one with excellent high
frequency performance. Still, this connector will introduce a certain amount of reflections, due
to impedance mismatch and insertion loss. In this chapter a method is discussed that enables
extracting the S-parameters of the connector from a series of reference measurements. These
extracted characteristics can then be deembedded from the measured characteristics of the
cascade of connector and antenna, eliminating the added losses and reflections.
3.1 Scattering Transfer Parameters
A different approach to represent the characteristics of a two-port network is by means of T-
parameters [13] (Scattering Transfer Parameters). These are an alternative to the well-known
S-parameters with their prime advantage being cascadeability. The T-parameters relate the
amplitudes of the waves a1 and b1 at the input of the two-port to the amplitudes of the waves
a2 and b2 at the output (see Figure 3.1), while the S-parameters relate the amplitudes of the
reflected waves at both input and output, i.e. b1 and b2, to the incident waves a1 and a2. The
T-parameters are hence defined by the relation
(a1
b1
)=
(T11 T12
T21 T22
)·(b2
a2
). (3.1)
The corresponding T-parameters can be calculated from the two-port’s S-parameters using the
DEEMBEDDING OF CONNECTOR AND FEED LINE STRUCTURE 11
following conversion formulas [14].
T11 = 1
S21
T12 = −S22S21
T21 = S11S21
T22 = −S11S22−S12S21S21
.
(3.2)
As stated above, the main advantage of T-parameters over S-parameters is cascadeability. To
illustrate this, consider a two-port network that consists of 2 two-port networks in a cascade
configuration as depicted in Figure 3.1.
a1
b1
a2
b2
a3
b3
a4
b4
T 1 T 2
Figure 3.1: Cascade of two arbitrary two-port networks.
Taking into account (3.1), it holds for the network of Figure 3.1 that(a1
b1
)=
(T 1
11 T 112
T 121 T 1
22
)·(b2
a2
)(3.3)
and (a3
b3
)=
(T 2
11 T 212
T 221 T 2
22
)·(b4
a4
). (3.4)
It now follows that the scattering transfer matrix for the overall network is defined as the matrix
product of the T-matrices of the cascaded two-port networks [14], i.e.(a1
b1
)=
(T 1
11 T 112
T 121 T 1
22
)·(T 2
11 T 212
T 221 T 2
22
)·(b4
a4
). (3.5)
The previous reasoning can be extended to an arbitrary cascade of N two-port networks, yielding
an expression for the overall scattering transfer matrix T given by
T = T 1 · T 2 · ... · TN . (3.6)
3.2 Deembedding Algorithm Based on Matrix-Pencil Method
As mentioned above, a deembedding algorithm will be used to extract the characteristics of
the connector with simple reference measurements as a starting point. A novel matrix-pencil
two-line method that achieves this was developed in [15], where it was exerted to eliminate the
impact of the connector in performing substrate characterization for wearable antenna systems.
DEEMBEDDING OF CONNECTOR AND FEED LINE STRUCTURE 12
To achieve deembedding, the S-parameters of two microstrip lines of length l1 and l2, with
l2 > l1, as depicted in Figure 3.2, need to be measured. The characteristics of the connectors,
in cascade with a coax-to-microstrip transition, are defined by SA and SB. The piece of ideal
lossy transmission line, with length ∆L = l2 − l1, is defined by SL. The two lines are fully
characterized by their propagation factor γ and characteristic impedance Z0. Prior knowledge
of the impedance Z0 is obsolete as the effects of impedance mismatch are included in the model.
l1
SA SB
l2
SA SBSL
Figure 3.2: Microstrip lines with coax-to-microstrip transitions SA and SB and ideal lossy
transmission line section SL.
Defining the scattering transfer matrix of the short line, with length l1, as T short and the matrix
of the long line, with length l2, as T long one can express both matrices as [15]
T short = TA · TBT long = TA · TL · TB.
(3.7)
Here TA, TB and TL are the T-parameter equivalents of SA, SB and SL respectively, which can
be obtained using the conversion formulas (3.2). If (3.7) is solved, one obtains
T · TA = TA · TL, (3.8)
where T = T long · T−1
short · T long. The S-parameter matrices Slong and Sshort can be obtained
through measurements of both microstrip lines and converted to their T-parameter counterparts
using conversion formulas [13].
The method described in [15] calculates the deembedded complex propagation factor γ of the
piece of transmission line defined by TL based on an eigenvalue equation obtained from (3.8).
Perturbations in the complex propagation constant can yield unphysical results, hence they are
minimized by using the matrix-pencil method as an averaging method. Once the propagation
constant γ is obtained, the scattering transfer parameters TA and TB (corresponding to the coax-
to-microstrip transitions SA and SB respectively) are constructed, hence the characteristics of
the connector are extracted.
It can be seen from Figure 3.2 that the extracted characteristics TA in reality comprise the
characteristics of the connector in cascade with a coax-to-microstrip transition and a piece of
DEEMBEDDING OF CONNECTOR AND FEED LINE STRUCTURE 13
transmission line of length lA = 0.5l1. If a suitable choice is made for the lengths of the reference
microstrip lines, not only the influence of the connector, but of the entire feed line exciting
the microstrip patch antenna in Figure 2.1 can be eliminated as well. This implies that after
measurements of the return loss of the fabricated antennas, the additional losses due to connector
as well as feed line can be removed from the measurements, yielding the characteristics of the
stand-alone antenna. How the characteristics of connector and feedline will be deembedded from
the antenna measurements in a later stage of the design is discussed in Section 3.3.
3.2.1 Illustrative Example
To prove the correctness of the deembedding algorithm presented in Section 3.2, an illustrative
example based on simulations will be given. All simulations are performed using the Keysight
Technologies ADS circuit solver. The algorithm will be used to extract the characteristics of
a connector with T-matrix T conn, which is modeled as a 35 Ω discontinuity in a piece of 50 Ωtransmission line. For this example, a 50 Ω transmission line with a 35 Ω discontinuity at each
side is considered (see Figure 3.3), which corresponds to a transmission line with connectors at
both ends. As discussed in Section 3.2 the S-parameters of two microstrip lines with different
lengths need to be simulated and fed to the deembedding algorithm.
35 Ω TL 50 Ω TL 35 Ω TL
50 ΩVs,1
Port plane
50 Ω
Port plane
Vs,2
Figure 3.3: Piece of 50 Ω transmission line with connector modeled as 35 Ω discontinuity.
To assess the correctness of the algorithm, a second S-parameter simulation is performed solely
on the connector as pictured in Figure 3.4. Now the characteristics of the connector simulated
by the circuit solver can be compared to the result of the deembedding algorithm, as presented
in Figure 3.5. It can be seen that the algorithm yields results that correspond very well to what
is predicted by the simulation.
35 Ω TL
50 ΩVs,1
Port plane
50 Ω
Port plane
Vs,2
Figure 3.4: S-parameter simulation performed on modeled connector.
DEEMBEDDING OF CONNECTOR AND FEED LINE STRUCTURE 14
0 10 20 30 40 50 60
−50
−40
−30
−20
−10
Frequency [GHz]
S11
[dB
]
Figure 3.5: Characteristic of model connector: extracted with deembedding algorithm (solid
line) and simulated with circuit solver (dashed line).
3.3 Characteristics of a Stand Alone Antenna
As the S-parameters of the connector and the piece of feed line are extracted as described in
the previous section, these can now be used to calculate the characteristics of the stand-alone
antenna from the S-parameter measurement of the cascade of connector, feed line and antenna.
As stated, the S-parameter measurements are performed on the system as depicted in Figure
3.6, where a1,2 and b1,2 are incident and reflected power waves respectively.
50 ΩVs
Port plane
a1
b1
a2
b2
Connector
Figure 3.6: Block diagram of measurement to be performed.
For this system it is known that
(b1
a1
)= T conn
(a2
b2
), (3.9)
where T conn is the T-matrix (containing the two-port Scattering Transfer Parameters) of the
connector, which is fully known as a result of the deembedding algorithm. The system in
DEEMBEDDING OF CONNECTOR AND FEED LINE STRUCTURE 15
Figure 3.6 is a one-port, hence the conducted measurement will yield a reflection coefficient Sm,
with
Sm =b1a1. (3.10)
If one wants to retrieve the return loss of the stand-alone antenna, the S-parameter Sa needs to
be extracted, which is defined as
Sa =a2
b2. (3.11)
Combining (3.9) and (3.11) one obtains
(b1
a1
)= T conn
(Sab2
b2
), (3.12)
and evidently
(Sa
1
)b2 = T
−1
conn
(b1
a1
). (3.13)
If one defines the matrix X as the inverse matrix of T conn, the system of equations
Sab2 = X11b1 +X12a1
b2 = X21b1 +X22a1
(3.14)
becomes apparent and provides an expression for the return loss of the antenna, given by
Sa =X11
b1a1
+X12
X21b1a1
+X22
. (3.15)
Taking into account (3.10) the return loss of the antenna can be expressed as a function of the
measured S-parameter Sm of the system defined in Figure 3.6 and the characteristics of the
connector obtained with the deembedding algorithm. One obtains
Sa =X11Sm +X12
X21Sm +X22, (3.16)
with X = T−1
conn.
It is clear that this method enables one to calculate the characteristics of the stand-alone antenna,
eliminating all added losses and reflections due to the presence of the feed line and connector.
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 16
Chapter 4
Design and Measurement of Antenna
Test Structures
4.1 Substrate and Connector
The substrate selected for the design and fabrication of the antenna test structures is the
RO4350B® Rogers Corporation high speed laminate [16]. It offers superior high frequency
performance and is compatible with standard low-cost circuit fabrication techniques. The main
benefits are the low dielectric loss and the stability of the dielectric constant over a broad fre-
quency range. The latter makes this laminate an ideal material for broadband applications. The
main characteristics are summarized in Table 4.1.
εr 3.48 (10 GHz)
tan δ 0.0037 (10 GHz)
thickness 168 µm
Cu thickness 35 µm
Table 4.1: Characteristics of Rogers RO4350B High Speed Laminate.
The connectors that will be used to measure the antenna test structures are 1.85 mm Sub-Micron
version A (SMA) end launch connector assemblies fabricated by Southwest Microwave [17] (Fig-
ure 4.1), which are suited for applications up to 67 GHz. A press fit is used to ensure good
connection between connector pin and circuit board, so no soldering is needed. To obtain con-
sistent measurements, the amount of force by which the connector is pressed onto the circuit
board, is controlled with a torque wrench. The end launch connectors are rather bulky and
made out of solid metal, so as a precaution not to interfere with the proper operation of the
antennas, the connector needs to be at a reasonable distance. This implies a long microstrip
feed line towards the patches. An optimized connector footprint will be used as suggested in [18]
and depicted in Figure 4.2. This footprint uses a 50 Ω Grounded Coplanar Waveguide (GCPW)
launch structure to improve the performance of the microstrip and the line is slightly tapered
to match the connector pin.
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 17
Figure 4.1: Southwest Microwave 1.85 mm end launch connector.
2.82 mm
280 µm
210 µm
Figure 4.2: Connector footprint with GCPW launch structure.
4.2 Microstrip Patch Antenna
The first step in designing the patch antenna test structures at 15 GHz, 30 GHz and 60 GHz is
using (2.5) to calculate a length L of the patch that yields good radiation efficiencies, with the
desired resonant frequency as a starting point. From this the effective relative permittivity εr,eff
can be determined with (2.6). Once the length L and εr,eff are known, the width extension ∆W
can be obtained using (2.7), where in this case W must be replaced with L. Taking into account
all of the above, the physical width W that will yield the desired resonant frequency for the
TM001-mode is obtained by solving
W =1
2fr√εr,eff√ε0µ0
− 2∆W. (4.1)
The last step is adding insets to obtain good impedance matching to the 50 Ω feed line. The
depth R of the inset, as defined in Figure 2.3, is calculated using (2.10).
After several design iterations, including extensive optimization steps, the final designs for the
patch antenna test structures have dimensions as listed in Table 4.2.
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 18
15 GHz 30 GHz 60 GHz
W [mm] 5.95 2.70 1.50
L [mm] 5.21 2.60 1.28
R [mm] 1.83 0.94 0.43
d [mm] 0.20 0.20 0.16
Lfeed [mm] 10.75 10.75 10.75
Table 4.2: Dimensions of designed microstrip patch antenna test structures.
4.2.1 Simulation Results
All simulation results presented below include substrate and copper losses. The simulated re-
flection coefficientes for the designed antennas are depicted in Figure 4.3. As can be seen, all
designs exhibit a well defined resonance peak at the desired frequency. However, there are dis-
tinct differences between the characteristics of the three antennas. The patch antenna at 60 GHz
(Figure 4.3(c)) exhibits much more insertion loss than the one at 15 GHz (Figure 4.3(c)). The
higher insertion loss is to be expected as the attenuation due to both conductor and substrate
losses increases with frequency [3]. The limited bandwidth, which is an inherent disadvantage
of microstrip patch technology, is clearly visible as well. The antennas at 15 GHz, 30 GHz and
60 GHz have a relative impedance bandwidth of 1.2%, 1.8% and 3.5%, respectively.
The simulated directivity and gain in the E-plane are presented in Figure 4.4. The gain is lower
than the directivity in all three cases, because of the substantial losses at these high frequencies.
Directivity is related to gain as [19]
G(θ, φ) =D(θ, φ)
KL, (4.2)
where KL is a real factor, greater than unity and independent of direction, that represents the
power losses in the materials forming the antenna. The radiation efficiency is defined as the
ratio of the gain to the directivity, hence it is equal to 1KL
. As is clear from Figure 4.4, the
efficiency of the three patches is moderate; there is a notable difference between directivity and
gain. The radiation efficiency for all three designs is approximately 60%.
4.2.2 Corner Analysis
As discussed in Chapter 1, the ongoing desire for higher operating frequencies causes a vast
miniaturization of antenna structures and in turn puts a strain on the fabrication processes at
hand. The etching of the antennas will not be infinitely precise, hence an insight into the effects
of possible fabrication flaws is imperative. Consulting Table 4.2, it is clear that the performance
of the patch antenna at 60 GHz is the most sensitive to fabrication inaccuracies because of its
very small dimensions (order of 1 mm). The minimum feature size of the fabrication process is
200 µm and a worst case scenario of a 10% fabrication error is assumed, i.e., a fault of 20 µm.
The dominant mode of the patch antenna at 60 GHz is the TM001 and is determined by the
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 19
14 14.5 15 15.5 16
0
-5
-10
-15
-20
-25
170 MHz
Frequency [GHz]
|S11|[
dB
]
(a)
29 29.5 30 30.5 31
0
-5
-10
-15
-20
-25
-30
540 MHz
Frequency [GHz]
|S11|[
dB
]
(b)
58 59 60 61 62
0
-10
-20
-30
2.1 GHz
Frequency [GHz]
|S11|[
dB
]
(c)
Figure 4.3: Simulated reflection coefficient for patch antenna test structures: (a) at 15 GHz; (b)
at 30 GHz and (c) at 60 GHz.
width W (see Table 4.2). Figure 4.5(a) illustrates the effects of the fabrication error on the
resonant frequency of the antenna. Another important feature is the amount of matching, as
it determines the quantity of power that is reflected and thus not radiated. This is mainly
determined by the inset depth R, for which the effects of the fabrication error are depicted
in Figure 4.5(b). It is clear that the most critical parameter is the width of the patch, as it
determines the resonant frequency. An error of 20 µm already causes a notable shift of the
resonance.
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 20
−15
−10
−5
0
5
10 dBi
90o
60o
30o
0o
−30o
−60o
−90o
(a)
−15
−10
−5
0
5
10 dBi
90o
60o
30o
0o
−30o
−60o
−90o
(b)
−15
−10
−5
0
5
10 dBi
90o
60o
30o
0o
−30o
−60o
−90o
(c)
Figure 4.4: Simulated directivity (solid) and gain (dotted) in the E-plane of the antenna: (a) at
15 GHz, (b) at 30 GHz and (c) at 60 GHz.
58 59 60 61 62
0
-10
-20
-30
-40
Frequency [GHz]
|S11|[
dB
]
W − 20 µmW + 20 µm
Final Design
(a)
58 59 60 61 62
0
-10
-20
-30
-40
Frequency [GHz]
|S11|[
dB
]
R− 20 µmR+ 20 µm
Final Design
(b)
Figure 4.5: Effect of a 20 µm fabrication error on the 60 GHz patch: (a) error on W affects the
resonant frequency; (b) error on R affects the matching.
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 21
4.3 Cavity-Backed SIW Antenna
As stated in Section 2.3, the equivalency between rectangular waveguides and SIW structures
only holds when the restrictions on the via diameter and spacing (2.14) are met. Sizing the vias
is an obligatory first step to ensure proper operation of the SIW antenna. These restrictions
are empirical, hence it is good practice to stay well below these limits. This will make sure the
via wall approximates a continuous conductive wall as good as possible. Table 4.3 summarizes
the via diameter and spacing maxima for each frequency, as well as the values that will be used
during the design. The dimensions at 30 GHz and 60 GHz are the minimal dimensions that still
comply with the fabrication design rules.
f [GHz] dmax [mm] smax [mm] d [mm] s [mm]
15 1.05 2.1 0.5 1
30 0.52 1.04 0.2 0.4
60 0.26 0.52 0.2 0.4
Table 4.3: Via diameter and spacing maxima for SIW operation.
Now, one can start dimensioning the actual cavity-backed SIW antenna as depicted in Figure 2.5.
Firstly, the dimensions of the cavity that correspond to the desired resonant frequency need to
be determined. The slot divides the SIW cavity into two half cavities with equal dimensions
and the radiation is caused by a strong TE120 resonance [20]. The SIW cavities of the antenna
test structures are hence designed to support a TE120 resonance at their desired frequencies of
operation.
From (2.15) the effective widths Weff and lengths Leff of the rectangular waveguides that are
equivalent to the SIW cavities can be determined. Together with the dimensions of the via
diameter and spacing from Table 4.3, the physical dimensions W and L of the SIW cavities can
be calculated using (2.11).
Next, the slot is added in the top copper plane to procure radiation. The slot is resonant to
ameliorate the radiation efficiency to the utmost extent [20], hence it has a length Ls approxi-
mately equal to half a wavelength at the desired operating frequency. The width of the slot Ws is
optimized to slightly improve the impedance bandwidth. Ideal impedance matching is acquired
through dimensioning Rgcpw and dgcpw for a 50 Ω Grounded Coplanar Waveguide (GCPW) feed.
After a considerable number of design iterations and ample optimization, the final dimensions
for the cavity-backed SIW antennas at 15 GHz, 30 GHz and 60 GHz are obtained as given in
Table 4.4.
4.3.1 Simulation Results
The simulated reflection coefficients for the cavity-backed SIW antennas are presented in Fig-
ure 4.6. It is clear that all three designs exhibit a well-defined resonance at the desired frequency.
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 22
15 GHz 30 GHz 60 GHz
L [mm] 8.00 4.00 2.00
W [mm] 10.00 4.40 2.40
Ls [mm] 5.82 3.05 1.59
Ws [mm] 0.76 0.30 0.21
Rgcpw [mm] 1.98 1.07 0.62
dgcpw [mm] 0.70 0.32 0.21
Table 4.4: Dimensions of designed cavity-backed SIW antenna test structures.
14 14.5 15 15.5 16
0
-5
-10
-15
-20
-25
-30
-35
-40
130 MHz
Frequency [GHz]
|S11|[
dB
]
(a)
29 29.5 30 30.5 31
0
-5
-10
-15
-20
-25
-30
-35
-40
370 MHz
Frequency [GHz]
|S11|[
dB
]
(b)
58 59 60 61 62
0
-5
-10
-15
-20
-25
-30
-35
-40
1.04 GHz
Frequency [GHz]
|S11|[
dB
]
(c)
Figure 4.6: Simulated reflection coefficient for cavity-backed SIW antenna test structures: (a)
at 15 GHz; (b) at 30 GHz and (c) at 60 GHz.
The impedance bandwidths are 1%, 1.2% and 1.7% for the antennas at 15 GHz, 30 GHz and
60 GHz, respectively. This demonstrates the inherent narrowband behavior of the SIW antenna.
Concerning the insertion loss, it is apparent that the difference between the SIW antennas at
15 GHz and 60 GHz is substantially smaller than was the case for the microstrip patch antenna
treated in Section 4.2. This indicates that high frequency losses are less severe when opting for
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 23
the SIW technology. Comparing Figure 4.3(c) with Figure 4.6(c), one perceives the significant
amount of approximately 7 dB less insertion loss for the SIW antenna. The metallic via wall
prevents radiation from spreading through the lossy substrate material, hence reducing substrate
losses as compared to a microstrip patch antenna.
The E-field inside the cavity of the SIW antenna at 60 GHz is presented in Figure 4.7. It is clear
from the configuration of the E-field lines that the designed cavity is in TE120 resonance, which
complies with the design strategy outlined in Section 4.3. The electric field at the two sides of
the slot has opposite phase and a large magnitude. Because of this, a transverse electric field
exists across the slot and radiation arises.
The simulated directivity and gain are presented in Figure 4.8. Again, there is a difference
between the gain and directivity, indicating losses in the materials forming the antenna. The
simulated radiation efficiencies are obtained from calculation of the factor KL as defined in (4.2)
and amount to approximately 35% for all three designs. The simulated gain appears to be
substantially less than for the microstrip patch antennas treated above. Additional simulations
of the SIW antenna with Microwave Studio (MWS) from Computer Simulation Technology
(CST), a full 3D solver based on the Finite Element Method (FEM), yield different results. The
simulated directivity corresponds to the results obtained with Advanced Design System (ADS),
but the simulated gain is notably higher. The simulated radiation efficiency is 75%. As stated
above, there is a large electric field across the slot and the substrate losses are considerably less
than for the patch antenna topology. Hence, a higher radiation efficiency is to be expected. In
literature, similar designs, using the cavity-backed SIW antenna topology, show a high radiation
efficiency as well [12].
0 1 2 3 4 5 6 7 8 9
[kVm ]
Figure 4.7: E-field inside the cavity of the SIW antenna at 60 GHz: TE120 resonance.
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 24
−15
−10
−5
0
5
10 dBi
90o
60o
30o
0o
−30o
−60o
−90o
(a)
−15
−10
−5
0
5
10 dBi
90o
60o
30o
0o
−30o
−60o
−90o
(b)
−15
−10
−5
0
5
10 dBi
90o
60o
30o
0o
−30o
−60o
−90o
(c)
Figure 4.8: Simulated directivity (solid) and gain (dotted) in the E-plane of the SIW antenna:
(a) at 15 GHz; (b) at 30 GHz and (c) at 60 GHz.
4.3.2 Corner Analysis
In a similar fashion as Section 4.2.2, the sensitivity of the parameters of the SIW antenna at
60 GHz are investigated. Once more, the worst case fabrication error is assumed to be 10% of
the minimum feature size allowed by the manufacturing process, i.e., 20 µm. An error on the
dimensions of the cavity, affects the resonant frequency, as depicted in Figure 4.9(a). An error
on the length of the slot Ls causes a slight shift in resonant frequency and an error on the
width Ws affects the impedance matching. This supports the theoretical analysis performed in
Section 2.3.2. It is clear from Figure 4.9 that the most critical parameters are the dimensions
of the cavity. An inaccuracy of 20 µm on the length of the cavity imposes a resonance shift of
approximately 300 MHz. To ensure that the cavity size is exact, the placement of the via wall
needs to be precise. If the row of vias is skewed, the dimensions of the cavity are altered and
the resonance is shifted. Moreover, if the spacing between the vias no longer complies with the
restrictions (2.14) due to inaccuracies, the equivalence with rectangular waveguides is lost and
the theoretical analysis of Section 2.3.2 no longer holds.
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 25
58 59 60 61 62
0
-5
-10
-15
-20
-25
-30
-35
-40
Frequency [GHz]
|S11|[
dB
]
L− 20 µmL+ 20 µm
Final Design
58 59 60 61 62
0
-5
-10
-15
-20
-25
-30
-35
-40
Frequency [GHz]
|S11|[
dB
]
W − 20 µmW + 20 µm
Final Design
(a)
58 59 60 61 62
0
-5
-10
-15
-20
-25
-30
-35
-40
Frequency [GHz]
|S11|[
dB
]
Ls − 20 µmLs + 20 µm
Final Design
58 59 60 61 62
0
-5
-10
-15
-20
-25
-30
-35
-40
Frequency [GHz]
|S11|[
dB
]
Ws − 20 µmWs + 20 µmFinal Design
(b)
Figure 4.9: Effect of a 20 µm fabrication error on 60 GHz SIW antenna: (a) error on dimensions
of the cavity affects the resonant frequency; (b) error on slot length Ls slightly shifts resonant
frequency and error on Ws affects impedance matching.
4.4 Transmission Line Test Structures for Deembedding
As discussed in Chapter 3, the effects of the feed line structure that excites the antenna, and of
the connector used for measurements, can be eliminated by exploiting a deembedding technique.
For this, reference measurements need to be performed on microstrip transmission lines, using
the same measurement connectors (Figure 3.2). It is stated in Section 3.2 that if a suitable
choice is made for the lengths of the lines, the effect of the connector and the entire feed line
can be eliminated; the shortest line needs to have a length equal to twice the length of the
feed line. The length of the feed is equal for all antenna test structures, as is apparent from
Table 4.4. Taking this into account, the reference transmission lines are designed as depicted
in Figure 4.10. The optimized footprint for the connector, as illustrated in Figure 4.2, is used
as well.
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 26
l2 = 51.5 mm
l1 = 2Lfeed = 21.5 mm
Figure 4.10: Designed reference transmission lines for deembedding of connector and feed line
structure.
4.5 Measurements
All S-parameter measurements presented in this section are performed with an N5247A Pro-
grammable Network Analyzer (PNA-X) from Keysight Technologies [21]. All reflection coeffi-
cients depicted below are the characteristics of the stand-alone antennas as discussed in Sec-
tion 3.3, i.e., deembedding of the connector and feed line is already performed. The deembedding
algorithm yields the characteristics of the end launch connector, as presented in Figure 4.11. It
is clear from Figure 4.11(b) that the connector introduces a maximum attenuation of 2 dB at
67 GHz. Looking at the overall trend of the reflection coefficient in Figure 4.11(a), it is perceived
that the matching deteriorates for rising frequencies. The fluctuations of the characteristics of
the connector are a result of the deembedding procedure. At each frequency where one of the
reference microstrip lines corresponds to a multiple of half a wavelength, a singularity occurs.
This also causes an additional ripple in the characteristics of the stand-alone antennas (see
further).
DESIGN AND MEASUREMENT OF ANTENNA TEST STRUCTURES 27
10 20 30 40 50 60
0
-10
-20
-30
Frequency [GHz]
|S11|[
dB
]
(a)
10 20 30 40 50 60
0
-0.5
-1
-1.5
-2
-2.5
Frequency [GHz]
|S21|[
dB
]
(b)
Figure 4.11: Characteristics of the end launch connector as obtained with the deembedding
The fabricated antenna array is presented in more detail in Figure 5.17. As explained above, the
conductive paste is pushed into the holes to ensure a good connection. Doing so, some spillage of
the paste ends up in the slot and the insets. After fabrication, the insets are completely sealed,
i.e., reducing the inset depth to zero, which is detrimental for the impedance matching. The
slots are partially occupied by the conductive material as well, plausibly affecting the proper
operation of the antenna array. As explained in Section 2.3.2, the slot is an important tuning
element for the impedance matching. Attempts have been made to clear both the insets and
the slot of the paste spillage. Still, some residue remains, as depicted in Figure 5.17.
HYBRID MODE SIW ANTENNA ARRAY 46
paste in inset
paste in slot
850 µm
Figure 5.17: Detail of fabricated hybrid mode SIW antenna array.
Measurements are performed using the SouthWest Microwave end launch connectors, discussed
in Section 4.1, and an N5247A PNA-X from Keysight Technologies [21]. These revealed that
standing waves occur in the corporate feed network of the array, as depicted in Figure 5.18. A
resonance occurs, approximately, every 5 GHz, which corresponds to a wavelength of circa 3.2 cm
inside the substrate material. The combined length of the microstrip feed line and the corporate
feed network corresponds to approximately half a wavelength at 5 GHz. This implies that a
resonance occurs whenever the length of the feed structure is a multiple of half a wavelength. A
standing wave is formed when two waves of the same frequency propagate in opposite directions,
i.e., when reflections occur at both ends of the feed network. Hence, the impedance of the SIW
antenna has changed significantly and is no longer properly matched to the 50 Ω feed line. This
can be attributed to the discrepancy in the size of the insets and the slot. Another hypothesis is
that the impedance of the SIW antenna has drastically changed due to the non-ideal electrical
properties of the paste at these high frequencies. The standing wave signifies that the impedance
matching at the connector end of the line is not sufficient either. As the dimensions of the 50 Ωmicrostrip are accurate, this indicates that there is an amount of mismatch at the junction
between the connector and the circuit. The thickness of the flexible substrate is substantially
less than in Chapter 4. Although proper operation for any board thickness is assured in [17], it
stands to reason that the press fit of the connector no longer ensures a good connection. Also,
the characteristics of the substrate are not precisely known at 60 GHz; if these have gravely
changed as compared to the values in Table 5.1, that could explain the faulty matching as well.
HYBRID MODE SIW ANTENNA ARRAY 47
40 45 50 55 60 65
0
-5
-10
-15
-20
-25
-30
-35
Frequency [GHz]
|S11|[
dB
]
Figure 5.18: Measured reflection coefficient for the fabricated hybrid mode SIW array on 50 µm
substrate.
To gain insight into the effects at hand, one of the reference microstrip lines for deembedding the
connector and feed line structure is measured. The characteristics are depicted in Figure 5.19.
The reflection coefficient in Figure 5.19(a) demonstrates that the matching of the line to the con-
nector is poor, i.e., above −10 dB on average. From the measured insertion loss in Figure 5.19(b)
it is clear that a considerable amount of attenuation is introduced. As stated above, the width of
the trace is accurate, hence the faulty matching is not caused by a fabrication inaccuracy. This
implies that the additional losses and inferior matching are due to the combination of unknown
substrate and connector characteristics.
To quantify the amount of additional losses that is introduced, the power balance of the line is
calculated. For an ideal, lossless transmission line it holds that [25]
|S11|2 + |S21|2 = 1, (5.5)
which denotes that all incident power at port 1 is either reflected back to the source or trans-
mitted to port 2. For a lossy transmission line this power balance becomes
|S11|2 + |S21|2 + Ploss + Prad = 1, (5.6)
where Ploss is the normalized power that is dissipated due to substrate and copper losses, and
Prad is the normalized power radiated into free space. Figure 5.20 depicts the power balance for
the measured reference microstrip line. It is clear that a substantial amount of power is lost,
70% on average. This indicates that the substrate material is notably more lossy than suggested
by the characteristics in Table 5.1, hence it is insufficiently characterized at 60 GHz.
HYBRID MODE SIW ANTENNA ARRAY 48
50 52 54 56 58 60 62 64 66 68
0
-10
-20
-30
-40
Frequency [GHz]
|S11|[
dB
]
(a)
50 52 54 56 58 60 62 64 66 68
-5
-6
-7
-8
-9
Frequency [GHz]
|S21|[
dB
]
(b)
Figure 5.19: Measured characteristics of the reference microstrip line for deembedding: (a)
reflection coefficient; (b) insertion loss.
50 52 54 56 58 60 62 64 66 680
0.2
0.4
0.6
0.8
1
Frequency [GHz]
|S11|2
+|S
21|2
Figure 5.20: Power balance for the measured reference microstrip line.
CONCLUSION AND FUTURE RESEARCH 49
Chapter 6
Conclusion and Future Research
The goal of this master’s dissertation was to design a highly compact and integratable antenna
array that operates in the 60 GHz band, whilst maintaining compatibility with standard printed
circuit board processing steps. Firstly, antenna test structures, in both microstrip patch and
Substrate Integrated Waveguide (SIW) technology, at different frequencies were developed to
make an assessment of the impact of high frequency effects. This brought forth a comparative
study of both technologies, based on both simulation and measurement results, and enabled the
formulation of a founded opinion that SIW is the most advantageous technology for the design
of the array.
The analysis of the SIW antenna test structures has confirmed the inherent band limited be-
havior of this technology, which is due to the high Q-factor of the resonance and the low profile.
A bandwidth enhancement technique based on hybrid mode excitation was proposed and suc-
cessfully executed. Simulated fractional impedance bandwidths of 3.6% and 4.5% were achieved
at 60 GHz on extremely thin substrates. To push the integration aspect of the dissertation to
the utmost extent, it was opted to design and fabricate the hybrid mode SIW antenna array
on 50 µm and 100 µm flexible substrate material. Moreover, exploiting a Uniform Linear Array
(ULA) configuration with four antenna elements, high gain and directivity were achieved, i.e.,
7.2 dBi and 12.0 dBi, respectively.
The insufficient characterization of the flexible substrate at 60 GHz and the non-ideal processing
of the minute vias were detrimental for the operation of the fabricated prototypes. Future
research on this topic could certainly encompass investigating the characteristics of flexible
substrates for antenna design at 60 GHz, or even research towards the development of novel
flexible substrate materials for use in the Extremely High Frequency (EHF) band.
The fabricaton process aside, future development can certainly cover a beam forming system,
which could be utilized in, for example, systems that provide a Wireless Local Access Network
(WLAN) in the 60 GHz band in office environments. Due to the beam forming, spacial selectivity
is achieved; this enables multiple systems to work side-by-side without causing interference.
Higher bandwidths, and correspondingly higher bit rates, can be offered to the mobile user.
Further research could be performed towards phase shifter circuits operating at 60 GHz as well.
CONCLUSION AND FUTURE RESEARCH 50
Combined with the hybrid mode SIW array and a Digital Signal Processing (DSP) unit, an
adaptive beam steering system could be developed. Adjusting the respective phases of the
excitations of the antennas in the array allows for scanning. This means the direction of maxi-
mum radiation is moved electronically, which could find its application in millimeter wave radar
detection systems.
If the bandwidth of the hybrid mode SIW antenna can be enlarged even further or research is
performed towards Ultra Wideband (UWB) SIW antennas, radar operation based on very short
pulses is possible. Applications could be, for example, an indoor positioning system. Due to
the UWB operation of the antenna, the round trip time of a very short pulse can be measured,
which allows triangulation.
BIBLIOGRAPHY 51
Bibliography
[1] Federal Communications Commision. (2013) Part 15 Rules for Unlicensed
Operation in the 57-64 GHz Band [Online]. http://www.fcc.gov/document/