sensors

Article
Design and Implementation of C-Band Large-Power Planar
Butler Matrix in SRS
Jinfeng Li 1 , Liping Yan 1 , Changjun Liu 1, * , He Bai 2 and Wanzhao Cui 2, *

1 The School of Electronics and Information, Sichuan University, Chengdu 610064, China
2 National Key Laboratory of Science and Technology on Space Microwave, China Academy of Space
Technology, Xi’an 710100, China
* Correspondence: [email protected] (C.L.); [email protected] (W.C.)

Abstract: In satellite remote sensing (SRS), there is a demand for large-power microwave components.
A Butler matrix is essential to a transmitting antenna array in SRS. This article illustrates the electrical
and mechanical design, simulation, and test results of a large-power planar beamforming network
for SRS at C-band. It is a 4 × 4 Butler matrix based on square coaxial lines. Short-ended stubs
are used in the Butler matrix to broaden its bandwidth by 10%, support inner conductors, and
enhance heat transfer in vacuum environments. The simulation results are consistent with the
measured results. The reflection coefficient is less than −18 dB, and the isolation is more than
23 dB from 3.8 GHz to 4.2 GHz. The insertion losses are less than 0.6 dB, and the phase errors are
better than ±6◦ . The measured peak microwave power of the proposed Butler matrix is 9 kW. Its
size is 440 × 400 × 40 mm3 . The proposed Butler matrix beamforming network can be applied to
SRS systems.

Keywords: Butler matrix; multipactor; power threshold; short-ended stubs; square coaxial line

1. Introduction
SRS needs to detect ground objects in space and is characterized by long distance
and multiple targets, which requires large transmitting microwave power and multi-beam-
transmitting antennas. Therefore, studying the method of realizing multi-beam antenna
Citation: Li, J.; Yan, L.; Liu, C.; Bai, H.; array and exploring a large-power transmission line structure is necessary.
Cui, W. Design and Implementation The beamforming networks are an essential part of an SRS transmitting antenna array.
of C-Band Large-Power Planar Butler There are many beamforming networks reported [1], like the Rotman lens, Blass matrix,
Matrix in SRS. Sensors 2024, 24, 2132. Nolan matrix [2], and Butler matrix [3–5]. Compared with these passive beamforming
https://doi.org/10.3390/s24072132 networks, the Butler matrix is the most popular choice, given its more straightforward and
Academic Editor: Stefano Savazzi
cost-effective structure and wider bandwidth [6]. In addition, the Butler matrix has the
advantages of reliability and reciprocity; therefore, it is widely used in several applications,
Received: 18 February 2024 such as SRS, Internet of Things (IoT), Wi-Fi, base stations, satellite communications, and
Revised: 19 March 2024 automotive radars [3].
Accepted: 20 March 2024 The study of the Butler matrix [7–9] involves many aspects, such as the perspective of
Published: 27 March 2024
the frequency band, 77 GHz [3], Ka-band [10], K-band [11], X-band [12], C-band [13], and
S-band [13]. From the perspective of large-power design [14], a specific computer-aided
method for a large-power Butler matrix design is given [15]. Various realization techniques
Copyright: © 2024 by the authors.
and associated transmission line technologies have been reported to fabricate Butler ma-
Licensee MDPI, Basel, Switzerland. trices. These include microstrip lines [16], waveguides [14], and square coaxial lines [17],
This article is an open access article to list a few examples. Substrate-integrated waveguide (SIW) Butler matrix has been de-
distributed under the terms and veloped as a compromise between the low loss and good high-frequency performances
conditions of the Creative Commons of rectangular waveguide types, as well as the integrability of planar transmission line
Attribution (CC BY) license (https:// types. However, the most challenging problem of the Butler matrix design is the presence
creativecommons.org/licenses/by/ of path crossings, which may lead to large dimensions when directly using SIW [7]. The
4.0/). fabrication of the substrate-integrated waveguides is complex and expensive [3]. In some

Sensors 2024, 24, 2132. https://doi.org/10.3390/s24072132 https://www.mdpi.com/journal/sensors

Sensors 2024, 24, 2132 2 of 17

contexts, it is preferred to microstrip technology for its higher immunity to interference,

no air radiation, and lower dispersion. However, realizing coaxial RF components with
variable line impedances is generally inconvenient with traditional manufacturing tech-
nologies [9]. The multipactor mechanism of introducing dielectrics between metal plates
as dielectric-filled waveguides is analyzed. However, breakdown and thermal heating
are potential challenges of using dielectric materials in a high-power radio frequency (RF)
environment [18]. There is no size advantage at all over highly integrated transmission lines
such as microstrip lines and substrate-integrated waveguides, but coaxials and waveguides
sacrifice size to achieve kW levels of power, while other transmission lines reach only mW
to W levels of power.
In [12], a miniaturized large-power beamforming network using an 8 × 8 Butler
matrix based on waveguides was successfully designed and tested for an X-band synthetic
aperture (SAR) system. The insertion loss and isolation of the unit are better than 0.4 dB
and 25 dB, respectively, over 600 MHz bandwidth at 9.6 GHz. The measured average and
peak microwave power are 500 W and 6 kW, respectively. The whole unit was machined
with a modified non-planar structure with a mechanical size of 478 × 362 × 264 mm3 .
In [16], a PCB-based beamforming network was designed for Ka-band applications.
Due to the low power capacity of the PCB structure, the measured power value was only
0.01 kW. Finally, by combining multiple phase shifters through a new design approach, a
millimeter-wave beamforming network is introduced to achieve beam-steering antenna
arrays. The beam steering is performed with a single microfluidic actuator and a circuit
model is introduced to facilitate the design of the beamforming network. The design
example presented is for a four-element antenna array operating at 28 GHz, exhibiting
continuous beam steering capability within ±30◦ when its SMP is actuated within its
−100 to 100 µm displacement range. Unlike beam-steering antennas using beamforming
ICs, they do not require active amplification to compensate for high loss and have a reduced
hardware complexity associated with control bits and bias lines. Compared to previously
investigated microfluids-actuated beam-steering antennas, it offers the advantage of not
relying on switched beam techniques and bulky lenses.
In [17], a square coaxial line beam former was designed. Simulation results demon-
strated that the square coaxial line could reach a power capacity of 8 kW in C-band.
Impedance matching was achieved by creating a groove discontinuity in the inner con-
ductor. A cone was provided for smooth impedance transformation for required broad
banding. The measured performance of the beam former was obtained by feeding it to
the feed layer of the electromagnetically coupled multilayer patch antenna. Equal and
unequal power dividers were designed with a proper impedance transformer by changing
dimensions of the inner conductor but keeping same outer conductor dimensions. Smooth
linear taper was provided at the impedance transformer junction. High power calculations
were also presented for square coaxial lines to handle several kW of power.
In this work, a large-power planar Butler matrix is implemented in C-band. Compared
to prior works, this work advances the wide bandwidth design and square coaxial line
for easy machining. In this paper, the aim of designing a 4 × 4 square coaxial Butler
matrix is to divide the overall structure, using four steps to complete it. Firstly, the key
component, the orthogonal coupler, is designed. The design is mainly concerned with
the isolation, i.e., the coupling between input ports, to ensure that the Butler matrix
performance from deterioration. Then, the cross junction is optimized with the design of
the coupler cascade, followed by the design of the phase shifter and the cross junction. With
tunning of the cross junction, the output phase balance is optimized, which also ultimately
determines the phase stability of the Butler matrix. Finally, according to the schematic
diagram, performance indexes, such as isolation, output amplitude flatness, phase flatness,
and others, are optimize with the comprehensive consideration of the inhibition of the
multipactor. It finally results in a well-working Butler matrix.
Sensors 2024, 24, 2132 33 of
Sensors 2024, 24, x FOR PEER REVIEW of 17
18

2. Principle and Design Method

2. Principle and Design Method
2.1. Principle
2.1. Principle
A Butler matrix is a beamforming network [19]. Figure 1 shows the schematic of a
4 × 4AButler
Butlermatrix
matrix designed
is a beamforming network
in this article. [19].
The Figure
input 1 shows
signal the schematic
at input ports P1 to of P4
a 4is×
4 Butler matrix designed in this article. The input signal at input ports P1 to
redistributed in the Butler matrix through four 3 dB hybrid couplers, two crossovers, twoP4 is redistrib-
uted
45 in theshifters,
◦ phase Butler matrix
and two through
0◦ phase four 3 dB hybrid
shifters. couplers,
The Butler two
matrix crossovers,
delivers signalstwo 45°output
to its phase
shifters, and two 0° phase shifters. The Butler matrix delivers signals to
ports with equal magnitudes and a phase increment between them of 45 , −135 , 135 its output
◦ ports
◦ with◦,
equal magnitudes
◦ and a phase increment between them of 45°, −135°, 135°,
and −45 depending on which input port is excited. The phase difference between output and −45° de-
pending
ports on which
is fixed wheninput port isport
any input excited. TheThe
is fed. phase difference
wideband between output
performance ports is
of a Butler fixed
matrix
when
is any due
mainly input
toport is fed. The wideband
the wideband performance
design of its components,of ae.g.,
Butler matrix is
couplers, mainly due
crossovers, andto
the wideband
phase shifters.design of its components, e.g., couplers, crossovers, and phase shifters.

45°phase 0°phase
shifter shifter
P1 P5
Hybrid Hybrid P6
P2
crossover crossover
P3
Hybrid Hybrid P7
P4 P8

45°phase 0°phase
shifter shifter
Figure 1. Schematic of a 4 ×
× 44Butler
Butlermatrix.
matrix.

2.2. Square Coaxial Line

2.2. Square Coaxial Line
2.2.1. Design of Square Coaxial Line
2.2.1. Design of Square Coaxial Line
Butler matrices are designed using various transmission lines, such as waveguides,
Butler lines,
microstrip matricesandare designed
coaxial linesusing variousthe
[20]. When transmission
operating band lines, is
such as waveguides,
in the C-band, the
microstrip lines, and coaxial lines [20]. When the operating
rectangular waveguide size of BJ40 is significantly larger than the square coaxialband is in the C-band, the
structure
rectangular
of our design. waveguide
The coaxialsizeline
of BJ40 is significantly
has the advantage larger
that thethan the is
signal square coaxial
shielded structure
by the outer
of our design. The coaxial line has the advantage that the signal
conductor. Therefore, the coupling to the adjacent component can be avoided. A suppres-is shielded by the outer
conductor. Therefore, the coupling to the adjacent component can
sion multipactor concept is considered to use the TEM mode in the coaxial line. A squarebe avoided. A suppres-
sion multipactor
coaxial concept
line is easier is considered
to machine to use the
than a cylinder TEM line
coaxial mode in in the coaxial
a Butler line. A square
matrix.
coaxial line is easier to machine than a cylinder coaxial line in a Butler
Two conditions are met in the design: (1) characteristic impedance should be 50 Ω to matrix.
match Two
the conditions are metterminals
input and output in the design:
of the(1) characteristic
system; and (2) lowimpedance should
attenuation. Thebe 50 Ω of
design to
amatch
beamthe input
former and output
begins terminalsthe
with evaluating of the system;
inner and diameter
and outer (2) low attenuation. Thedistance
to achieve the design
of a beamthem
between formerfor begins
handling with evaluating
higher the inner
microwave power and outerThe
levels. diameter to achieve
characteristic the dis-
impedance
depends on the ratio of outer and inner conductor diameters. For a 50 Ω square coaxial,
tance between them for handling higher microwave power levels. The characteristic im-
pedance depends on the ratio of outer and inner conductor diameters.
the inner conductor used in the square coaxial line is 4.8 mm, and the outer conductor is For a 50 Ω square
coaxial,
12 mm. the
Theinner conductor
cross-section of used in the
a square square
coaxial coaxial
line line is
is shown in4.8 mm,2.and
Figure Thethewhite
outeriscon-
the
ductor is and
vacuum, 12 mm. the The
orangecross-section
is aluminum. of a square coaxial line is shown in Figure 2. The white
is the vacuum, and the orange is aluminum.
Sensors 2024,
Sensors 2024, 24,
24, 2132
x FOR PEER REVIEW 4 of 18
4 of 17

d d1

ε0εr d1 I d
d

ε0εr Id
Od

O
d line cross-section structure.
Figure 2. Square coaxial

Its characteristic impedance [21] can be approximately calculated by

Figure 2. Square coaxial line cross-section structure.
structure.
136.7  O 
Z0 = lg  0.9259 d  (1)
Its characteristic impedance [21] can be εr 
approximately
be approximately I d calculated

calculated by
by
136.7  Odd 
 
Its attenuation constant [21] can
Z 136.7
= be approximately
lg 0.9259 O calculated by
Z0 = 0 √ lg 0.9259 I   (1)
εrεr  Idd 
59.37 Rs  I d  1
Its attenuation
Its attenuation constant
constant [21]
[21] can
can be
beαcapproximately
=
approximately 1 + calculated
 by (2)
η0 Z0  Ocalculatedd  Od
by
59.37Rs  I d 1
where Od and Id are the outerαcand α59.37R
=
c inner 1 + Id  1
s conductor (2)
= η0 Z0 1+ Od  Od diameters, respectively, and(2)εr is the
η0 Z0 Od Od
relative dielectric constant of the vacuum. Od is 12 mm, and Id is 4.8 mm, as presented in
where Odwhich
Figure and Id are thetoouter
Od/Idand inner conductor diameters, respectively, and εr is the
where 2,
Od and Idleads
are the outer is 2.5.
and inner conductor diameters, respectively, and εr is the
relative dielectric constant of the vacuum. Od is 12 mm, and Id is 4.8 mm, as presented in
relative dielectric constant of the vacuum. Od is 12 mm, and Id is 4.8 mm, as presented in
Figure
2.2.2. 2, which leads to Od/Id isSimulation
Electrical 2.5.
Figure 2, which Performance
leads to Od /Id is 2.5.
2.2.2.Use
2.2.2.
the 3DPerformance
Electrical electromagnetic
Electrical Performance Simulation
Simulation
simulation software CST Studio Suite 2020 to simulate
the square
Use the coaxial
the 3D line. The
3Delectromagnetic length
electromagneticsimulation is 1000 mm,
simulation andCST
software the material
Studio is aluminum.
Suite 2020 The electric
to simulate
Use software CST Studio Suite 2020 to simulate the
field
square distribution
the square
coaxial line.is
coaxial entirely
line.
The The symmetrical.
length
length is 1000
is 1000 mm,mm,andThe
andhighest
the electric
the material
material field intensity
is aluminum.
is aluminum. isfield
located on
The electric
The electric
the edge
field
distribution of the
distribution inner conductor,
is entirely
is entirely and
symmetrical.
symmetrical. thehighest
The peak
The electric
highest field
electric
electric is about
fieldfield 3500
intensity
intensity isV/m, asthe
located
is located on shown in
on
Figure
edge of3.the
the edge The
of the insertion
inner
inner loss and
conductor,
conductor, is and
less than
thethe 0.01
peak
peak dBfield
electric
electric within
field
is is aabout
given
about frequency
3500
3500 V/m,as
V/m, range,inand the
as shown
shown
return
Figureloss3. The insertionthan
is greater loss 65
is less
dB, than
as shown
0.01 dBinwithin
Figurea4. given frequency range, and the
than 65
return loss is greater than 65 dB,
dB, as
as shown
shown in
in Figure
Figure 4.
4.

Figure 3. E-field distribution of square coaxial line.

Figure 3. E-field distribution of square coaxial line.
Figure 3. E-field distribution of square coaxial line.
Sensors
Sensors2024,
2024,24,
24,xxFOR
FORPEER
PEERREVIEW
REVIEW 55of
of18
18

Sensors 2024,
2024, 24,
24, 2132
x FOR PEER REVIEW 00
Sensors 55 of
of 17
18

--20
20
|S|S1111| |

Amplitude(dB)
Amplitude(dB)
0
|S|S2121| |
--40
40
−20
--60
60
|S11|

Amplitude(dB)
|S21|
−40
--80
80
−60
--100
100
00 22 44 66 88 10
10
−80 Frequency(GHz)
Frequency(GHz)
Figure
Figure4.4.Simulation
Simulationresults
resultsof
ofsquare
squarecoaxial
coaxialline
lineS-parameter.
S-parameter.
−100
0 2 4 6 8 10
2.2.3.
2.2.3.Multipactor
MultipactorSimulation
Simulation
Frequency(GHz)
Figure
Figure 55shows
shows
4. Simulation the
thesquare
resultssquare coaxial
coaxial
of square line’s
line’s
coaxial linemultipactor
multipactor [22]
[22]simulation
simulationmodel. model.TheTheinitial
initial
Figure line S-parameter.
S-parameter.
particle
particlecloud
cloudisislocated
locatedin inthe
themiddle
middleof ofthe
thesquare
squarecoaxial
coaxialline,
line,thetheinitial
initialnumber
numberof ofpar-
par-
2.2.3.
ticles Multipactor
ticlesis
is1000,
1000,the Simulation
thesimulation
2.2.3. Multipactor Simulationtime
simulation timeisis30
30ns,
ns,and
andthethecenter
centerfrequency
frequencyisis44GHz. GHz.The Thesquare
square
coaxial
coaxial line
line5multipactor
Figure multipactor threshold
shows the squarethreshold simulation
simulation
coaxial result
resultisisabout
line’s multipactor [22] 9500
about 9500W Win
simulation inFigure
model.6.6.
Figure The initial
Figure 5 shows the square coaxial line’s multipactor [22] simulation model. The initial
When
When
particle other
cloud other conditions
conditions
is located are
in the guaranteed
aremiddle
guaranteed
of thenot
not to
tochange,
squarechange,
coaxialthe frequency
theline,
frequency offset
offset
the initial will
willalso
number also
of
particle cloud is located in the middle of the square coaxial line, the initial number of par-
change
changethe
particles the multipactor
multipactor
is 1000, threshold
threshold
the simulation timepower.
power. The
is 30 ns,The
andthreshold
threshold
the centerpower
power isisrelated
frequency related to
toff××The
is 4 GHz. d,d,where
squareff
where
ticles is 1000, the simulation time is 30 ns, and the center frequency is 4 GHz. The square
is
isthe
thefrequency
coaxial frequency and
andddisisthreshold
line multipactor the
thedistance
distance between
simulation two
twometals.
betweenresult metals.
is about 9500 W in Figure 6.
coaxial line multipactor threshold simulation result is about 9500 W in Figure 6.
When other conditions are guaranteed not to change, the frequency offset will also
change the multipactor threshold power. The threshold power is related to f × d, where f
is the frequency and d is the distance between two metals.

Figure
Figure
Figure5. Simulation
5.5.Simulation model
Simulationmodel of
modelof square
ofsquare coaxial
squarecoaxial line
coaxialline multipactor.
linemultipactor.
multipactor.

9,500W
9,500W 2000
2000
particles(number)
of particles(number)

9,750W
9,750W
Figure 5. Simulation model of square coaxial line multipactor.
10,250W
10,250W
1500
1500 10,750W
10,750W
2000 9,500W
Number of particles(number)

1000
9,750W
1000
10,250W
Number of

1500 10,750W
Number

500
500

1000
00
00 55 10
10 15
15 20
20 25
25 30
30
500 Time(ns)
Time(ns)

Figure
Figure6.6.Simulation
Simulationresults
resultsof
ofsquare
squarecoaxial
coaxialline multipactorthreshold.
linemultipactor threshold.
0
0
When 5
other 10
conditions 15 are guaranteed
20 25 not30to change, the frequency offset will also
Time(ns)
change the multipactor threshold power. The threshold power is related to f × d, where f
Figure
is 6. Simulation
the frequency andresults of square
d is the coaxial
distance line multipactor
between two metals.threshold.
 Sensors 2024, 24, x FOR PEER REVIEW 6 of 18
Sensors 2024, 24, 2132 6 of 17

2.3.2.3. Design
Design andand Analysis
Analysis
2.3.1.
2.3.1. Coupler
Coupler Analysis
Analysis
The most
The central
most component
central componentof theofButler matrixmatrix
the Butler is the quadrature coupler, also
is the quadrature known
coupler, also
as known
the branchline hybrid network.
as the branchline The loaded
hybrid network. Theshort-ended stubs coupler
loaded short-ended stubs was designed
coupler was de-
using square
signed usingcoaxial
squarelines. The lines.
coaxial branchThecoupler’s scatteringscattering
branch coupler’s parameters are
parameters are

0  0j j 1 0 

1 0
1  j1 0j 00 0 11 
[S] = − S√ = − 2 1 00 0 j j 
 (3)(3)
2 1 0 
0 10 1j j 0 0

Its scattering matrix has symmetry, and the values at the diagonal are all zero to char-
Its scattering matrix has symmetry, and the values at the diagonal are all zero to
acterize that all four ports are matched. The design of the orthogonal coupler focuses on
characterize that all four ports are matched. The design of the orthogonal coupler focuses
the following parameters: input reflection coefficient, isolation, transmission coefficients,
on the following parameters: input reflection coefficient, isolation, transmission coefficients,
output
output ports
ports phase
phase difference,
difference, bandwidth.
bandwidth.
When cascaded in the Butler
When cascaded in the Butler matrix, matrix,thethe branch
branch coupler
coupler is not
is not sufficiently
sufficiently flatflat
in in
itsits
in-band
in-band output
output amplitude
amplitude to narrow
to narrow the available
the available bandwidth.bandwidth. Short-ended
Short-ended stubs are stubs
loadedare
loaded at the four ports of the branch coupler to extend the bandwidth
at the four ports of the branch coupler to extend the bandwidth of the branch coupler and of the branch cou-
pler and improve
improve isolation. isolation.
Short-ended
Short-ended stubs
stubs [23]
[23] areare used
used in in Butler
Butler matrix
matrix components
components to to broaden
broaden thethe band-
band-
width by 10%. Meanwhile, they help heat dissipation and prevent
width by 10%. Meanwhile, they help heat dissipation and prevent device burnout. Four device burnout. Four
pairs
pairs of of short-ended
short-ended stubs
stubs areare placed
placed to to support
support thethe inner
inner conductors,
conductors, broadening
broadening thethe
bandwidth [24–26] and not affecting the output magnitude. The
bandwidth [24–26] and not affecting the output magnitude. The couplers are optimized couplers are optimized
forfor impedance
impedance matching.The
matching. The simulation
simulation model
model isis shown
shownininFigure
Figure7.7.Ls Ls
= 21.7 mm,mm,
= 21.7 L_off
= 19 mm. The simulation results are shown in
L_off = 19 mm. The simulation results are shown in Figure 8. Figure 8.

Zs
1 2 Ls

1 L_off 2

4 3

4 3 Short-ended
stubs

(a) (b)
Figure
Figure 7. Simulation
7. Simulation model
model of the
of the coupler
coupler (a)(a) without
without short-ended
short-ended stubs;
stubs; (b)(b) with
with short-ended
short-ended stubs.
stubs.

2.3.2. Crossover Analysis

The crossover is an important connecting unit in the Butler matrix. It is also a symmet-
rical four-port device and operates in such a state that when one of the ports is input, the
diagonally opposite output port is at equal amplitude, and the remaining two ports are
isolated. The crossover scattering parameters are
 
0 1 0 1
1 0 1 0
[S] = 
0
 (4)
1 0 1
1 0 1 0
 Sensors 2024, 24, x FOR PEER REVIEW 7 of 18
Sensors 2024, 24, 2132 7 of 17

0
0

without short-ended stubs with short-ended stubs

−10
−10
Amplitude(dB)

Amplitude(dB)
−20
−20
|S11| |S11|
|S21| |S21|
−30 |S31| −30 |S31|
|S41| |S41|
−40
3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0 −40
3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0
Frequency(GHz) Frequency(GHz)

(a) (b)
Figure8.8.Simulation
Figure Simulationresults
resultsof
ofcoupler
couplerS-parameters
S-parameters(a)
(a)without
withoutshort-ended
short-endedstubs;
stubs;(b)
(b)with
withshort-
short-
ended stubs.
ended stubs.

2.3.2. Crossover
Two unloadedAnalysis
short-ended-stub hybrid couplers can be cascaded to create a Butler
matrixThecrossover. Direct
crossover is ansynthesis
importantusing a broadband
connecting coupler
unit in would
the Butler significantly
matrix. increase
It is also a sym-
the
metrical four-port device and operates in such a state that when one of the ports is input
overall size of the Butler matrix. The signal transmission path shows that when input,
port
the 1diagonally
is fed, output port 4output
opposite has equal
portamplitude
is at equaloutput, and both
amplitude, and output port 2 and
the remaining twoinput
ports
port 3 are isolated
are isolated. states modeled
The crossover in Figure
scattering 9a. The are
parameters scattering parameter simulation results
are shown in Figure 9b; it can be seen that the isolation is more than 20 dB in the frequency
Sensors 2024, 24, x FOR PEER REVIEW
range of 3.5–4.5 GHz, the insertion loss is less 0than
1 0.05  and the curve shows symmetry
0 1dB, 8 of 18
1 0 1 0 
S  = 
centered at 4.0 GHz. The input VSWR is lessthan 1.2 in the  frequency range of 3.8–4.2 GHz. (4)
0 1 0 1
 
0 1 0 1 0 
Two unloaded short-ended-stub hybrid couplers can be cascaded to create a Butler
2 4 −10
matrix crossover. Direct synthesis using a broadband coupler would significantly increase
the overall size of the Butler matrix. The signal transmission path shows that when input
Amplitude(dB)

−20
port 1 is fed, output port 4 has equal amplitude output, and both output port 2 and input
port 3 are isolated states modeled in Figure 9a. The scattering parameter simulation results
are shown in Figure 9b; it can be−30
seen that the isolation is more than 20 dB in the frequency
range of 3.5–4.5 GHz, the insertion loss is less than 0.05 dB, and the curve shows|Ssymmetry
11|
−40
centered at 4.0 GHz. The input VSWR is less than 1.2 in the frequency range of| 3.8–4.2
|S21
GHz. |S31|
−50
|S41|
1 3 −60
3.0 3.5 4.0 4.5 5.0
Frequency(GHz)
(a) (b)
Figure9.9. Simulation
Figure Simulation results
resultsof
ofthe
thecrossover
crossover(a)
(a)E-field
E-fielddistribution;
distribution;(b)
(b)S-parameters.
S-parameters.

2.3.3.
2.3.3. Phase
Phase Shifter
Shifter Analysis
Analysis
AA phase [27] with
phase shifter [27] withgood
goodphase
phaseflatness
flatnessininthe
the operational
operational frequency
frequency band
band di-
directly determines
rectly determines thethe accuracy
accuracy of the
of the Butler
Butler matrix
matrix outputoutput phases.
phases. The lengths
The lengths of theof the
phase
phase
shiftersshifters
and theand the interconnecting
interconnecting linesbeen
lines have havetuned
been to tuned tothe
offset offset the change
phase phase change
caused
caused by the stubs. The broadband 45 ◦ phase shifters and 0◦ phase shifters have been
by the stubs. The broadband 45° phase shifters and 0° phase shifters have been realized.
The simulation model of phase shifters and simulation results are shown in Figures 10–
13, respectively.

Output 2 3
 Figure 9. Simulation results of the crossover (a) E-field distribution; (b) S-parameters.

2.3.3. Phase Shifter Analysis

A phase shifter [27] with good phase flatness in the operational frequency band di-
Sensors 2024, 24, 2132 8 of 17
rectly determines the accuracy of the Butler matrix output phases. The lengths of the phase
shifters and the interconnecting lines have been tuned to offset the phase change caused
by the stubs. The broadband 45° phase shifters and 0° phase shifters have been realized.
realized.
The The simulation
simulation model
model of phase of phase
shifters and shifters andresults
simulation simulation resultsinare
are shown shown
Figures 10–in
Figures
13, 10–13, respectively.
respectively.

Output 2

Short-ended stubs

Input 1
Sensors 2024, 24, x FOR PEER REVIEW 9 of 18
Sensors 2024, 24, x FOR PEER REVIEW 9 of 18
Figure
Figure10.
10.Simulation
Simulationmodel 45◦phase
modelofof45° phaseshifter.
shifter.

The wideband 45° phase shifter 90 has good transmission performance at a frequency
0 90
0 of 3.6–4.2 GHz with a return loss greater than 40 dB and an in-band insertion loss less than
45°phase difference
0.1 dB. The output phase difference between them is 45 ± 45皃hase difference
1°, and the phase flatness is
−10 75
- 10 75 shown in Figure 11.
superior. The simulation results are
Phase difference (degree)
Phase difference (degree)
Amplitude(dB)

−20
Amplitude(dB)

60
- 20 60

−30 45
- 30 45
|S |
−40 |S11 |11
30
- 40
|S21|S|21|
30

−50
- 50 15
15

−60
- 60 3.0 3.5 4.0 4.5 5.0
0
0 3.0 3.5 4.0 4.5 5.0
3.0 3.5 Frequency(GHz)
4.0 4.5 5.0 3.0 3.5 4.0
Frequency (GHz) 4.5 5.0
Frequency(GHz) Frequency(GHz)
(a) (b)
(a) (b)
Figure 11. Simulation
Simulation results
results of◦45° phase shifter (a) S-parameters; (b) phase difference.
Figure
Figure 11.11.
Simulation results of of
45°45phase
phase shifter
shifter (a)(a) S-parameters;
S-parameters; (b)(b) phase
phase difference.
difference.
Output 2 3
Output 2

Short-ended stubs
Short-ended stubs

Input 1 4
Input 1
Figure 12. Simulation model of 0° phase shifter.
Figure 12.12.
Figure Simulation model
Simulation of of
model 0° 0phase shifter.
◦ phase shifter.
The wideband 0° phase shifter has good transmission performance at a frequency of
The wideband 0° phase shifter has good transmission performance at a frequency of
3.6–4.2 GHz with a return loss greater than 30 dB and an in-band insertion loss less than
3.6–4.2 GHz with a return loss greater than 30 dB and an in-band insertion loss less than 0.14
0.14 dB. The output phase difference is 0 ± 1.5°. The simulation results are shown in Figure
dB. The output phase difference is 0 ± 1.5°. The simulation results are shown in Figure 13.
13.
45
0 45
0 0° phase difference
30
0°phase difference
30
ncy (degree)

- 10
cy (degree)

−10
ude(dB)

15
de(dB)

15
- 20
−20
 Figure 12. Simulation model of 0° phase shifter.

The wideband 0° phase shifter has good transmission performance at a frequency of

3.6–4.2 GHz with a return loss greater than 30 dB and an in-band insertion loss less than
Sensors 2024, 24, 2132 0.14 dB. The output phase difference is 0 ± 1.5°. The simulation results are shown in Figure
9 of 17
13.

45
0
0°phase difference
30

Phase differency (degree)

−10
Amplitude(dB)

15
−20
0
−30
−15
|S11 |
−40
|S21 | −30

−50
3.0 3.5 4.0 4.5 5.0 −45
3.2 3.6 4.0 4.4 4.8
Frequency(GHz) Frequency(GHz)
(a) (b)
Figure13.
Figure 13.Simulation
Simulationresults
resultsofof00° phaseshifter
◦ phase shifter(a)
(a)S-parameters;
S-parameters;(b)
(b)phase
phasedifference.
difference.

The wideband 45◦ phase shifter has good transmission performance at a frequency of
3.6–4.2 GHz with a return loss greater than 40 dB and an in-band insertion loss less than
0.1 dB. The output phase difference between them is 45 ± 1◦ , and the phase flatness is
superior. The simulation results are shown in Figure 11.
The wideband 0◦ phase shifter has good transmission performance at a frequency
of 3.6–4.2 GHz with a return loss greater than 30 dB and an in-band insertion loss less
than 0.14 dB. The output phase difference is 0 ± 1.5◦ . The simulation results are shown in
Figure 13.

2.3.4. Butler Matrix Design

The Butler matrix was designed as an eight-port device. The input ports are labeled
P1 to P4, and the output P5 to P8. The rest of the details of the design are chamfered edges,
ventilation holes, square coaxial line corners for corner-cutting, square coaxial line corners
for rounding, avoiding low-pressure discharge, etc. These design factors were considered
to achieve high power threshold.
Aluminum alloy is used to build both inner and outer conductors. The material used
in the simulation of Butler matrix is aluminum, and the actual machining is performed
with aluminum alloy in order to machine the material more rigidly. One significant detail
is that each of the eight ports of the Butler matrix leads to a small cylinder in order to
facilitate testing with the L29 connector. The cylinder size is 2 mm in diameter and 4 mm in
height. The proposed Butler matrix [28] uses a planar open-cover structure, as shown in
Figures 14 and 15. The mounting configuration ensures the strength of the structure. It is
440 mm by 400 mm and has a height of 40 mm.
 with aluminum alloy in order to machine the material more rigidly. One significant detail
is that each of the eight ports of the Butler matrix leads to a small cylinder in order to
facilitate testing with the L29 connector. The cylinder size is 2 mm in diameter and 4 mm
in height. The proposed Butler matrix [28] uses a planar open-cover structure, as shown
Sensors 2024, 24, 2132 in Figures 14 and 15. The mounting configuration ensures the strength of the structure.
10 of It
17
is 440 mm by 400 mm and has a height of 40 mm.

4.8mm

4.8mm
Sensors 2024, 24, x FOR PEER REVIEW 11 of 18
Figure
Figure 14.
14. The
The fabricated
fabricated Butler
Butler matrix’s
matrix’s inner
inner conductor.
conductor.

400 mm
440 mm

Figure 15. The

Figure 15. The fabricated
fabricated Butler
Butler matrix’s
matrix’s outer
outer conductor.
conductor.

3. Results
3. Results
3.1. Simulation
3.1. Simulation
The optimally designed coupler, crossover junction, 45◦ phase shifter, and 0◦ phase
The
shifter areoptimally designed
applied to coupler,
the Butler matrixcrossover junction,
beamforming 45° phase
network, andshifter, and model
the entire 0° phaseis
shifter are applied to the Butler matrix beamforming network, and the entire
electromagnetically simulated to analyze the performance, e.g., port reflection coefficients, model is
electromagnetically simulated to analyze the performance, e.g., port reflection
input and output port isolations, transmission coefficients, and output port phase differ- coeffi-
cients, Due
ences. input toand output port
the model’s isolations,
symmetry, the transmission coefficients,
scattering parameters andto
related output
input port
port phase
1 and
differences.
input port 4 Due to the model’s
are identical symmetry,
when the theisscattering
input port parameters
fed. Similarly, related
this works forto input
input port
port 2
1 and input port 4 are identical when the input port is fed. Similarly, this works
◦
and input port 3 as well. The phase difference between the output ports is 45 , −135 , 135 , for
◦ input◦
port−
and 2 45
and input port 3 as well. The phase difference between the output ports is 45°, −135°,
◦ , respectively.
135°, and −45°, respectively.
When input port 1 and input port 2 are fed separately, the input reflection coefficient,
transmission coefficient, and isolation of the Butler matrix beamforming network are
shown in Figure 16. In the frequency range of 3.85–4.15 GHz, the input reflection coeffi-
cient of input port 1 is less than −20 dB, the input isolation is greater than 20 dB, and the
transmission coefficient is −6.0 dB ± 0.4 dB. The input reflection coefficient of input port 2
 shifter are applied to the Butler matrix beamforming network, and the entire model is
electromagnetically simulated to analyze the performance, e.g., port reflection coeffi-
cients, input and output port isolations, transmission coefficients, and output port phase
differences. Due to the model’s symmetry, the scattering parameters related to input port
Sensors 2024, 24, 2132 1 and input port 4 are identical when the input port is fed. Similarly, this works for input
11 of 17
port 2 and input port 3 as well. The phase difference between the output ports is 45°, −135°,
135°, and −45°, respectively.
When
When input
input port1 1and
port andinput
inputport
port2 2are
arefed
fedseparately,
separately,the
theinput
inputreflection
reflectioncoefficient,
coefficient,
transmission
transmission coefficient, andand
coefficient, isolation of theofButler
isolation matrixmatrix
the Butler beamforming networknetwork
beamforming are shown are
inshown
Figure in16.Figure
In the frequency
16. In the range of 3.85–4.15
frequency range of GHz, the input
3.85–4.15 reflection
GHz, coefficient
the input of input
reflection coeffi-
port 1 isofless
cient inputthan −20
port dB,
1 is thethan
less input
−20isolation is greater
dB, the input than 20
isolation dB, andthan
is greater the transmission
20 dB, and the
coefficient is − 6.0 dB ± 0.4 dB. The input reflection coefficient of
transmission coefficient is −6.0 dB ± 0.4 dB. The input reflection coefficient of input port 2 isinput
less than
port 2
−is
20less
dB.than
The −20 isolation between input ports is greater than 20 dB, and the
dB. The isolation between input ports is greater than 20 dB, and the trans-transmission
coefficient is −6.0 dBis±−6.0
mission coefficient 0.4 dB.
dB The
± 0.4insertion loss of theloss
dB. The insertion matrix is small.
of the matrixItisshows
small.that the
It shows
performance of the Bulter
that the performance matrix
of the is close
Bulter matrixto is
anclose
idealtomatrix.
an ideal matrix.

0 0

−5 −5
−10 −10
−15 |S11 | −15

Amplitude(dB)
Amplitude(dB)

−20 |S21 | |S12|

−20
−25 |S31 | |S22|
−25 |S32|
−30 |S41 |
|S51 | −30 |S42|
−35
|S52|
|S61 | −35
−40 |S62|
|S71 |
−45 −40 |S72||
|S81 | |S82|
−50 −45
Sensors3.62024, 24,
3.7 x FOR
3.8 PEER
3.9 REVIEW
4.0 4.1 4.2 4.3 4.4 3.4 3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5 4.6 12 of 18
Frequency(GHz) Frequency(GHz)
(a) (b)

Figure
Figure 16. Simulationresults
16. Simulation resultsofofButler
Butler matrix
matrix input
input ports’
ports’ scattering
scattering parameters:
parameters: (a) reflection
(a) reflection coef-
ficient of input
coefficient portport
of input 1 and transmission
1 and coefficients
transmission of input
coefficients portport
of input 1 to other ports;
1 to other (b) reflection
ports; coef-
(b) reflection
ficient of input
coefficient portport
of input 2 and transmission
2 and coefficients
transmission of input
coefficients portport
of input 2 to2other ports.
to other ports.

The
Theisolations
isolationsbetween
betweenthetheoutput
outputports
portsare greater
are than
greater than2020
dBdB
in in
thethe
frequency range
frequency of
range
3.8–4.2 GHz,
of 3.8–4.2 as shown
GHz, as shownin Figure 17. The
in Figure highhigh
17. The isolation between
isolation output
between ports
output guarantees
ports the
guarantees
independence
the independence of each output
of each port,port,
output which willwill
which benefit the the
benefit phased array
phased for for
array beamforming.
beamform-
Meanwhile,
ing. Meanwhile, the reflection coefficients of output port 5 and output port 6 areless
the reflection coefficients of output port 5 and output port 6 are lessthan
than
− 20
−20 dB as
as well. The Bulter
The Bulter matrix’s output port is matched to 50 Ω, i.e., standard antennas
matched to 50 Ω, i.e., standard antennas
connected
connectedtotothetheoutput
output port will
port avoid
will avoidimpedance
impedance mismatch
mismatchandand
prevent largelarge
prevent microwave
micro-
power being reflected
wave power to the Butler
being reflected matrix.matrix.
to the Butler

−5 −5

−10 −10

−15 −15

−20 −20
Amplitude（dB）

Amplitude（dB）

−25 −25

−30 −30

−35 −35

−40 −40

−45 |S55| −45 |S56 |

|S65| |S66 |
−50 −50
|S75| |S76 |
−55 −55
|S85| |S86 |
−60 −60
3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4
Frequency（GHz） Frequency（GHz）
(a) (b)
Figure 17.
Figure 17. Simulation
Simulation results
resultsof
ofButler
Butlermatrix
matrixoutput
outputports
portsscattering parameters:
scattering parameters: (a)(a)
reflection co-
reflection
efficient of output port 5 and isolation of output port 5 to other ports; (b) reflection coefficient
coefficient of output port 5 and isolation of output port 5 to other ports; (b) reflection coefficient ofof
output port 6 and isolation of output port 6 to other ports.
output port 6 and isolation of output port 6 to other ports.

When the output port is fed, the phase difference between the output ports is in the
frequency range of 3.9–4.1 GHz, as shown in Figure 18, and the phase difference between
neighboring output ports is 45 ± 5° when input port 1 is fed. The phase difference between
neighboring output ports is −135 ± 5° when input port 2 is fed. The stabilized phase fre-
quency range is shifted to the lower frequency when input port 3 is fed, and the phase
Sensors 2024, 24, 2132 12 of 17

When the output port is fed, the phase difference between the output ports is in the
frequency range of 3.9–4.1 GHz, as shown in Figure 18, and the phase difference between
neighboring output ports is 45 ± 5◦ when input port 1 is fed. The phase difference between
neighboring output ports is −135 ± 5◦ when input port 2 is fed. The stabilized phase
frequency range is shifted to the lower frequency when input port 3 is fed, and the phase
Sensors 2024, 24, x FOR PEER REVIEW 13 of 18
difference between neighboring output ports is 135 ± 10◦ . The phase difference between
neighboring output ports is −45 ± 6◦ when input port 4 is fed.

75 −90

−105
60

Phase Difference （degree）

Phase Difference （degree）

−120
45

−135

30
−150
Ang(S51-S61)
15 Ang(S52-S62)
Ang(S61-S71) −165
Ang(S62-S72)
Ang(S71-S81)
Ang(S72-S82)
0
3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5 −180
3.4 3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5
Frequency（GHz） Frequency（GHz）
(a) (b)
180 0

−15
165
Phase Difference （degree）
Phase Difference （degree）

−30
150

−45

135
−60

Ang(S53-S63) Ang(S54-S64)
120
Ang(S63-S73) −75
Ang(S64-S74)
Ang(S73-S83) Ang(S74-S84)
105 −90
3.4 3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5 4.6 3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4
Frequency（GHz） Frequency（GHz）
(c) (d)
Figure
Figure 18.
18. Simulation
Simulationresults
resultsof
ofButler
Butlermatrix
matrixphase
phasedifference:
difference:(a)
(a)phase
phasedifference
differenceof
ofinput
inputport
port1;1;
(b)
(b) phase difference of input port 2; (c) phase difference of input port 3; (d) phase difference of input
phase difference of input port 2; (c) phase difference of input port 3; (d) phase difference of input
port 4.
port 4.

3.2.
3.2. Measurement
Measurement
AA vector
vector network
network analyzer
analyzer (VNA,
(VNA, Agilent
Agilent N5230A,
N5230A,Agilent,
Agilent, Santa
Santa Clara,
Clara, CA,
CA, USA)
USA)
was
was applied
applied to
to measure
measure the
the S-parameter.
S-parameter. Figure
Figure 19
19 shows
shows the
the Butler
Butler matrix
matrix S-Parameters
S-Parameters
measurement
measurement system.
system.For
Forthe
thefirst
firstmeasurement,
measurement,two
twoports
portsof ofthe
theButler
Butlermatrix
matrixare
arecon-
con-
nected
nected to
to two
twocables
cables of
ofthe
thevector
vectornetwork
networkanalyzer,
analyzer, and
and the
the rest
rest of
ofthe
theports
portsare
areconnected
connected
to N-matched
to N-matched loads.
loads.
Sensors 2024, 24, x FOR PEER REVIEW 14 of 18
Sensors 2024, 24, 2132 13 of 17
Sensors 2024, 24, x FOR PEER REVIEW 14 of 18

Agilent N5230A
Agilent N5230A

400 mm
400 mm

440 mm
440 mm
40 mm
40 mm

Figure 19. The Butler matrix measurement system.

Figure 19.The
Figure19. TheButler
Butlermatrix
matrixmeasurement system.
measurement system.
The
The test results are
areshown
shownininFigure
Figure20,
20,Figure 2121,
and Figure22,
22,respectively,
respectively,andandare
are Thetest
testresults
consistent results
with the are shown in
simulation Figure
results.
Figure
20,
Figure Figure
20a shows
Figure
21 and
that Figure
the 22, respectively,
reflection coeffi- and
consistent with the simulation results. Figure 20a shows that the reflection coefficients of
are consistent
cients of ports with the simulation results.20b
Figure 20ameasured
shows that the reflection coeffi-
ports are betterare better
than −18 than −18 dB.
dB. Figure 20bFigure presents
presents measured transmissiontransmission
coefficientscoeffi-
when
cients
cients
the of
when
Butler ports are better
the Butler
matrix is fedmatrixthan −18
is fed
at input dB. Figure
portat1input 20b
port 1port
and input presents
and2.input
The portmeasured transmission
2. Thecoeffects
coupling couplingare well coeffi-
coef-
cients when
fects are
equalized,well the Butler
equalized,
around matrix
−7 dBaround
between is3.8
−7 fed
dB at 4.2
input
between
and port
3.8
GHz. and14.2
and input port 2. The coupling coef-
GHz.
fects are well equalized, around −7 dB between 3.8 and 4.2 GHz.
0
−2
|S11| |S55| |S51| |S61| |S71| |S81|
0
-2
|S22||
|S |S|S66| | |S52|
|S51| |S62|
|S61| |S72|
|S71| |S82| |S81|
11 55
|S33|
|S22| |S77|
|S66 | −4 |S52| |S62| |S72| |S82|
−15 |S44| |S88|
|S33 | |S77 |
Amplitude(dB)

−5 - 4
Amplitude (dB)

−18
- 15 |S44| |S88 |
Amplitude(dB)

−6 - 5
Amplitude (dB)

- 18

−30
−7 - 6

−8
- 30
-7

-8
−45
3.8 3.9 4.0 4.1 4.2 −10
Frequency(GHz) 3.8 3.9 4.0 4.1 4.2
Frequency (GHz)
- 45
3.8 3.9
(a)
4.0 4.1 4.2 - 10 (b)
Frequency(GHz)
Figure 20.
20. Measured
Measured results
results of
of Butler
Butler matrix
3.8
matrix S-parameters:
3.9
S-parameters: (a)
(a) ports’
4.0 coefficients;
ports’ reflection
4.1(b) input
reflection coefficients;
4.2
Figure Frequency (GHz) (b) input
ports’ transmission
ports’ transmission coefficient.
coefficient.
(a) (b)
Figure
Figure
Figure 21 a,bshow
21a,b
20. Measured show themeasured
the
results measured
of results
Butler results
matrix ofof
thethe isolations
isolations
S-parameters: between
(a)between
ports’ thethe input
input
reflection andand out- (b) inpu
output
coefficients;
put ports.
ports.
ports’ In In band
the the band
transmission of 3.9–4.1
ofcoefficient.
3.9–4.1 GHz,GHz,
the the isolations
isolations are are close
close to our
to our goal
goal of dB.
of 25 25 dB. Moreo-
Moreover,
ver,measured
the the measured results
results agreeagree
with with the simulations
the simulations well.well.
Figure 21 a,b show the measured results of the isolations between the input and out
put ports. In the band of 3.9–4.1 GHz, the isolations are close to our goal of 25 dB. Moreo
ver, the measured results agree with the simulations well.
 |S21| |S31| |S41|
|S65| |S75| |S85|
|S32| |S42| |S43|
|S76| |S86| |S87|
−15 −15

Amplitude(dB)
Sensors 2024,
Sensors 24, x FOR PEER REVIEW 15 of 18
−20 24, 2132
2024,
Amplitude(dB) −20 14 of 17

−25 −25

0−30 −30
0
|S21| |S31| |S41|
|S65| |S75| |S85|
|S32| |S42| |S43|
−45 |S76| |S86| |S87|
−15−45 −15

Amplitude(dB)
−20 −20
Amplitude(dB)

−25 −25 3.8 3.9 4.0 4.1 4.2

3.8 3.9 4.0 4.1 4.2
Frequency(GHz) −30
Frequency(GHz)
−30
(a) (b)
Figure 21. Port isolation of the Butler matrix (a) input ports; (b) output ports.

−45 −45
Figure 22 shows the measured phase differences for the four input ports. The phase
differences between different output ports when the signal is fed at each input port are
45° ± 6°, −135° ± 5°, 135° ± 6°and −45° ± 6° over the frequency range. Good agreement
3.8 3.9 4.0 4.1
3.8 3.9 4.0
Frequency(GHz)
4.1
between simulations 4.2 measurements
and is found, although the
Frequency(GHz)measurement results4.2are
(a) slightly worse than the simulation. The machining error (b)of the short-ended stubs impacts
the phase shift and will also stress the test connector. The machining parameter error of
Figure 21.21.
Figure
0.4 mm Port isolation
isPort
out of theofrange
isolation thethe
of Butler matrix
ofButler (a)(a)
matrix
machining input ports;
input
accuracy (b)(b)
ports;
and output
output
causes aports.
ports. error of about 2°.
phase

Figure
180 22 shows the measured phase differences for the four input ports. The phase
differences between different output ports when the signal is fed at each input port are
45° ± 135
6°, −135° ± 5°, 135° ± 6°and −45° ± 6° over the frequency range. Good agreement
Phase(S51-S61) Phase(S53-S63)
Phase Difference (degree)

between simulations and measurements is found, although the measurement results are
90 Phase(S61-S71) Phase(S63-S73)
slightly worse than the simulation. The
Phase(S71-S81)
machining error of the short-ended stubs impacts
Phase(S73-S83)
the phase
45 shift and will also stress the test connector. The machining parameter error of
0.4 mm is out of the range of machining accuracy and causes a phase error of about 2°.
0
180
−45
Phase(S52-S62) Phase(S54-S64)
135
−90 Phase(S -S72)
Phase(S51-S62
61)
Phase(S -S74)
Phase(S53-S64
63)
Phase Difference (degree)

Phase(S -S82) Phase(S -S84)

90 Phase(S61-S72
71) Phase(S63-S74
73)
−135 Phase(S71-S81) Phase(S73-S83)
45
−180
3.8 3.9 4.0 4.1 4.2
0
Frequency(GHz)
−45 22.
Figure 22. The
The phase
phase difference
difference of the
the Butler
Butlermatrix.
matrix.
Figure Phaseof
(S -S ) Phase(S -S 52 62 54 64)
−90 Phase(S62-S72) Phase(S64-S74)
Figure
Figure 22 23 shows
shows the the Phase
measured
(S72-S82) phase
experimental differences
[29–31]
Phase -S84) offor
setup
(S74 thethe four inputmeasurement
multipactor ports. The phase[18],
differences
which includes
−135 between different
the data output ports
acquisition when the signal
and processing system.is fed at each of
A portion input
the port are
incident
45 ◦ ± 6◦ , −135◦ ± 5◦ , 135◦ ± 6◦ and −45◦ ± 6◦ over the frequency range. Good agreement
power applied to the device under test is extracted, fed into a coupling circuit, and nulled
−180
between
against simulations and measurements is found, although the measurement results are
3.8 a part of 3.9the extracted 4.0 reflected4.1 power. Power
4.2 detection is performed using aver-
slightly worse than
age power meters (Agilent the simulation. The machining
E14419B), and the data recording error of therateshort-ended
is 1000 timesstubs
perimpacts
second.
Frequency(GHz)
the
A traveling wave tube amplifier (TWTA) is applied to provide large microwaveerror
phase shift and will also stress the test connector. The machining parameter power.of
0.4
Figure mm
22. is
Theout of
phase the range
difference of machining
the Butler accuracy
matrix. and causes a phase error of about 2 ◦.
We adopt nulling of the forward/reverse power detection method to conduct multipactor
Figure 23inshows
experiments the experimental
test samples. [29–31] setup
The test frequency of the
is 4 GHz. multipactor
Using measurement
a dual-directional [18],
coupler,
Figure
which 23 shows
includes the experimental [29–31] setup of the multipactor measurement [18],
the incident andthe data acquisition
reflected microwaveand powersprocessing
are sampled.system. A portion of the incident
which
power includes
appliedthe to thedata acquisition
device under test andisprocessing
extracted, fed system.
into aAcoupling
portion circuit,
of the and
incident
nulled
power
againstapplied
a parttoofthe
thedevice under
extracted test ispower.
reflected extracted,Powerfed detection
into a coupling circuit,using
is performed and nulled
average
against
powera meters
part of (Agilent
the extracted reflected
E14419B), andpower.
the data Power detection
recording rate isis 1000
performed using
times per aver- A
second.
traveling
age wave tube
power meters (Agilentamplifier (TWTA)
E14419B), and the is applied to provide
data recording ratelarge microwave
is 1000 times perpower.
second.We
A adopt nulling
traveling waveoftube the amplifier
forward/reverse
(TWTA)power detection
is applied method
to provide large to microwave
conduct multipactor
power.
experiments in test samples. The test frequency is 4 GHz. Using
We adopt nulling of the forward/reverse power detection method to conduct multipactor a dual-directional coupler,
the incident
experiments inand
test reflected
samples. microwave powersisare
The test frequency sampled.
4 GHz. Using a dual-directional coupler,
the incident and reflected microwave powers are sampled.
 Sensors2024,
Sensors 2024,24,
24,2132
x FOR PEER REVIEW 16 of 18
15 of 17

Figure23.
Figure 23.The
Theexperimental
experimentaltest
testsetup
setupof
ofmultipactor.
multipactor.

After
Afterrealizing
realizingpower-level
power-level zeroing,
zeroing, the
the spectrum
spectrum analyzer
analyzer monitors
monitors thethe zeroed
zeroed sig-
sig-
nal
nal [32] in real time. The multipactor threshold can be tested by pulse modulationsignal
[32] in real time. The multipactor threshold can be tested by pulse modulation signal
with
withaadutydutycycle
cycleofof10%.
10%.Thirty
Thirtyminutes
minutesshould
shouldbebeallotted
allottedfor
foreach
eachtest
testpower
powerlevelleveltill
till
breakdown
breakdown occurs. Nulling signal level of forward/reverse power is less than −60dBm
occurs. Nulling signal level of forward/reverse power is less than − 60 dBm
when
whentest testsamples
sampleswork
workcorrectly.
correctly.Nulling
Nullingsignal
signallevel
level will
will increase
increase as
as soon
soon asas the
the multi-
multi-
pactor takes place.
pactor takes place.
In
Inaddition,
addition,these
thesetest
testsamples
samplesmust
mustbe becleaned
cleanedwith
withalcohol
alcoholbefore
beforethey
theyare
aremounted
mounted
in a vacuum. The temperature of the test sample
in a vacuum. The temperature of the test sample must−be must be monitored during
monitored the
during experiment.
the experi-
The
ment.airThe
pressure in the vacuum
air pressure must bemust
in the vacuum less than 1×
be less 10 16 ×Torr
than 10−6and
Torrremain for more
and remain than
for more
24 h until the experiment begins. Radioactive source Cesium 137 was used
than 24 h until the experiment begins. Radioactive source Cesium 137 was used to produce to produce
sufficient
sufficientelectrons.
electrons.The
Themultipactor
multipactormicrowave
microwavepower powerthreshold
thresholdwaswastested
testedtotobe
be99kW.kW.
The bandwidth of the Butler matrix is significantly better than that of other research
The bandwidth of the Butler matrix is significantly better than that of other research
works listed in Table 1, reaching 10%. According to the product of frequency and distance,
works listed in Table 1, reaching 10%. According to the product of frequency and distance,
the measurement results of multipactor microwave power threshold based on the square
the measurement results of multipactor microwave power threshold based on the square
coaxial line Butler matrix is also good.
coaxial line Butler matrix is also good.
Table 1. Performance with the prior Butler matrix.
Table 1. Performance with the prior Butler matrix.
Refs
Ref f0/GHz f0 /GHz dmin /mm 20 dB-Bandwidth
dmin/mm 20 dB-Bandwidth Configuration
Configuration Power/kW
Power/kW
[16]
[16] 28 28 0.203 0.203 5.7%5.7% 1 1××4 4PCB
PCB 0.01
0.01
[12]
[12] 9.6 9.6 10.16 10.16 6.25%
6.25% 8 8××8 8WR90
WR90 66
[17] 5 2.75 4% 1 × 8 square coaxial line 8
[17]Work
This 5 4 2.75 2.35 4% 10% 1 ×4 8×square
4 squarecoaxial
coaxialline
line 98
This Work 4 2.35 10% 4 × 4 square coaxial line 9
4. Conclusions
4. Conclusions
To solve the problems of large-power transmission and planar structure, it is proposed
To solve
to design the matrix
a Butler problems of large-power
beamforming transmission
network based onand planarcoaxial
a square structure,
line.itSquare
is pro-
posed to
coaxial design
lines a Butler
are easily matrixand
fabricated beamforming
have a highnetwork based
microwave on athreshold
power square coaxial
comparedline.
Square coaxial lines are easily fabricated and have a high microwave power
with circular ones. Shunt short-ended stubs on the branchline coupler are adopted in the threshold
compared
strategy, withcan
which circular
increaseones. Shunt short-ended
the bandwidth stubs on
of a conventional onethe
bybranchline coupler
10%. The whole are
Butler
adopted in the strategy, which can increase the bandwidth of a conventional
matrix with introduced short-ended stubs is simulated and optimized. The matrix with one by 10%.
aThe
400whole
mm by Butler
400 mm matrix with introduced
dimension short-ended
was fabricated stubs is simulated
and measured. and optimized.
The measured results
Theconsistent
are matrix withwitha those
400 mm by simulations.
of the 400 mm dimension was fabricated
The measurement and
results measured.
of the developed The
measured results
large-power are consistent
4 × 4 Butler matrix 9with those
kW are of the simulations. The measurement results
presented.
of the
In developed
follow-up work,large-power
further 4research
× 4 Butler
willmatrix 9 kW areonpresented.
be conducted a compact design and widen-
ing the microwave transmission bandwidth. Moreover, we will coat the inner conductors
Sensors 2024, 24, 2132 16 of 17

with special materials to decrease the electron emission coefficient, and thus further en-
hance the microwave power threshold of the matrix. It could be used for large-power
Butler matrices for SRS applications.

Author Contributions: Conceptualization, C.L. and W.C.; methodology, C.L. and J.L.; validation,
W.C., H.B. and C.L.; writing, J.L.; supervision, C.L. and L.Y.; project administration, C.L., W.C. and
L.Y. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the National Natural Science Foundation of China (NSFC),
grant numbers U22A2015 and 62071316.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Data are contained within the article.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Vallappil, A.K.; Rahim, M.K.A.; Khawaja, B.A.; Murad, N.A.; Mustapha, M.G. Butler Matrix Based Beamforming Networks
for Phased Array Antenna Systems: A Comprehensive Review and Future Directions for 5G Applications. IEEE Access 2021, 9,
3970–3987. [CrossRef]
2. Li, P.; Ren, H.; Arigong, B. A Symmetric Beam-Phased Array Fed by a Nolen Matrix Using 180◦ Couplers. IEEE Microw. Wirel.
Compon. Lett. 2020, 30, 387–390. [CrossRef]
3. Sakhiya, R.; Chowdhury, S. A Low-Loss, 77 GHz, 8 × 8 Microstrip Butler Matrix on a High-Purity Fused-Silica (HPFS) Glass
Substrate. Sensors 2023, 23, 1418. [CrossRef] [PubMed]
4. Shallah, A.B.; Zubir, F.; Rahim, M.K.A.; Majid, H.A.; Sheikh, U.U.; Murad, N.A.; Yusoff, Z. Recent Developments of Butler Matrix
from Components Design Evolution to System Integration for 5G Beamforming Applications: A Survey. IEEE Access 2022, 10,
88434–88456. [CrossRef]
5. Rao, L.; Pant, M.; Malviya, L.; Parmar, A.; Charhate, S.V. 5G beamforming techniques for the coverage of intended directions in
modern wireless communication: In-depth review. Int. J. Microw. Wirel. Technol. 2021, 13, 1039–1062. [CrossRef]
6. Imani, A.; Bayati, M.S. A novel design of compact broadband 4 × 4 Butler matrix-based beamforming antenna array for C-band
applications. Aeu-Int. J. Electron. Commun. 2021, 138, 153901. [CrossRef]
7. Qiu, L.; Zhu, L.; Ouyang, Z.; Deng, L. Wideband Butler Matrix Based on Dual-Layer HMSIW for Enhanced Miniaturization. IEEE
Microw. Wirel. Compon. Lett. 2022, 32, 25–28. [CrossRef]
8. Nasseri, H.; Bemani, M.; Ghaffarlou, A. A New Method for Arbitrary Amplitude Distribution Generation in 4 × 8 Butler Matrix.
IEEE Microw. Wirel. Compon. Lett. 2020, 30, 249–252. [CrossRef]
9. Palazzi, V.; Cicioni, A.; Alimenti, F.; Mezzanotte, P.; Tentzeris, M.M.; Roselli, L. Compact 3-D-Printed 4 × 4 Butler Matrix Based
on Low-Cost and Curing-Free Additive Manufacturing. IEEE Microw. Wirel. Compon. Lett. 2021, 31, 125–128. [CrossRef]
10. Mohan, M.P.; Cai, H.; Alphones, A.; Karim, M.F. mmWave Zero Order Resonant Antenna with Patch-Like Radiation Fed by a
Butler Matrix for Passive Beamforming. Sensors 2023, 23, 7973. [CrossRef]
11. Jeon, M.; Seo, Y.; Cho, J.; Lee, C.; Jang, J.; Lee, Y.; Kwon, H.; Kahng, S. Investigation on Beam Alignment of a Microstrip-Line
Butler Matrix and an SIW Butler Matrix for 5G Beamforming Antennas through RF-to-RF Wireless Sensing and 64-QAM Tests.
Sensors 2021, 21, 6830. [CrossRef] [PubMed]
12. Paul, T.; Harinath, M.; Garg, S.K.; Aich, S.; Kumar, A.; Trivedi, J.; Kumar, A.; Patel, M.K.; Rao, C.V.N.; Jyoti, R. Miniaturized
High-Power Beam Steering Network Using Novel Nonplanar Waveguide Butler Matrix. IEEE Microw. Wirel. Compon. Lett. 2021,
31, 678–681. [CrossRef]
13. Kim, S.; Choi, D. Analysis of Beamforming Antenna for Practical Indoor Location-Tracking Application. Sensors 2019, 19, 3040.
[CrossRef] [PubMed]
14. Tsao, Y.; Hsu, H.; Desai, A. High power handling GaAs SP4T Switch-based Beam-Switching planar antenna module for 5G
New-Radio FR2 applications. Aeu-Int. J. Electron. Commun. 2023, 159, 154494. [CrossRef]
15. Alessandri, F.; Mongiardo, M.; Sorrentino, R. Computer-aided-design of beam forming networks for modern satellite antennas.
IEEE Trans. Microw. Theory Tech. 1992, 40, 1117–1127. [CrossRef]
16. Sandoval, J.J.M. Mm-Wave Reconfigurable Antenna Arrays, Phase Shifters and Beamforming Networks with Reduced Hardware Complexity
Using Integrated Microfluidic Actuation; Engineering University of South Florida: Tampa, FL, USA, 2022.
17. Pandey, B.K.; Kulshrestha, S.; Dey, R.; Chakrabarty, S.B.; Sharma, S.B. Square Coaxial Beam Forming Network for Multilayer
Microstrip Antenna. In Proceedings of the 2007 Asia Pacific Microwave Conference, Bangkok, Thailand, 11–14 December 2007;
Volumes 1–5, pp. 1226–1229.
18. Li, Y.; Wang, D.; Yu, M.; He, Y.; Cui, W. Experimental Verification of Multipactor Discharge Dynamics between Ferrite Dielectric
and Metal. IEEE Trans. Electron Devices 2018, 65, 4592–4599. [CrossRef]
Sensors 2024, 24, 2132 17 of 17

19. Pezhman, M.M.; Heidari, A.-A.; Ghafoorzadeh-Yazdi, A. A compact 4 × 4 SIW beamforming network for 5G applications.
Aeu-Int. J. Electron. Commun. 2021, 135, 153714. [CrossRef]
20. Guo, Y.J.; Ansari, M.; Fonseca, N.J.G. Circuit Type Multiple Beamforming Networks for Antenna Arrays in 5G and 6G Terrestrial
and Non-Terrestrial Networks. IEEE J. Microw. 2021, 1, 704–722. [CrossRef]
21. Lau, K.H. Loss calculations for rectangular coaxial lines. Proc. Inst. Elect. Eng. Microw. Antennas Propag. 1988, 135, 207–209.
[CrossRef]
22. Wu, C.B.; Sun, Q.F.; Luo, Y.H.; Li, Z.J. Studies of multipactor threshold in square coaxial lines. In Proceedings of the 2000 5th
International Symposium on Antennas, Propagation and Em Theory, Beijing, China, 15–18 August 2000; pp. 422–425.
23. Feng, W.; Duan, X.; Shi, Y.; Zhou, X.Y.; Che, W. Dual-Band Branch-Line Couplers with Short/Open-Ended Stubs. IEEE Trans.
Circuits Syst. II 2020, 67, 2497–2501. [CrossRef]
24. Dong, L.; Lv, Y.; Xiao, H.; Lu, J.; Jiang, Y.; Li, Y. Micro-rectangular Coaxial Structure-based Broadband Millimeter-wave Power
Divider. In Proceedings of the 2020 21st International Conference on Electronic Packaging Technology (Icept) 2020, Guangzhou,
China, 12–15 August 2020.
25. Sa’Ad, B.M.; Rahim, S.K.A.; Peter, T.; Bin Abdul Rani, M.S.; Ausordin, S.F.; Zaidel, D.N.A.; Krishnan, C. Transparent Branch-Line
Coupler Using Micro-Metal Mesh Conductive Film. IEEE Microw. Wirel. Compon. Lett. 2014, 24, 857–859. [CrossRef]
26. Ding, K.; Kishk, A. Wideband Hybrid Coupler with Electrically Switchable Phase-Difference Performance. IEEE Microw. Wirel.
Compon. Lett. 2017, 27, 992–994. [CrossRef]
27. Alnahwi, F.M.; Al-Yasir, Y.I.A.; See, C.H.; Abdullah, A.S.; Abd-Alhameed, R.A. Generalized Concept and MATLAB Code for
Modeling and Analyzing Wideband 90◦ Stub-Loaded Phase Shifters with Simulation and Experimental Verifications. Sensors
2023, 23, 7773. [CrossRef] [PubMed]
28. Prakash, V.; Dahiya, S.; Kumawat, S.; Singh, P. Design of 4 × 4 Butler Matrix and Its Process Modeling Using Petri Nets for Phase
Array Systems. Prog. Electromag. Res. C 2020, 103, 137–153. [CrossRef]
29. Menargues, E.; Sirci, S.; Calleau, A.; Dimitriades, A. High-power tests in Additive Manufacturing Ka-band feeds and feed
components. In Proceedings of the 2023 International Conference Electromagnetics in Advanced Applications (ICEAA), Venice,
Italy, 9–13 October 2023.
30. Bekasiewicz, A.; Koziel, S. Low-Cost Unattended Design of Miniaturized 4 × 4 Butler Matrices with Nonstandard Phase
Differences. Sensors 2021, 21, 851. [CrossRef] [PubMed]
31. García-Patrón, M.; Rodríguez, M.; Ruiz-Cruz, J.A.; Montero, I. Uncertainty Budget in Microwave High-Power Testing. IEEE Trans.
Instrum. Meas. 2023, 72, 1–12. [CrossRef]
32. Monerris-Belda, O.; Marin, R.C.; Jodar, M.R.; Diaz-Caballero, E.; Guillen, C.A.; Petit, J.; Vicente, E.B.; Gimeno, B.; Raboso, D. High
Power RF Discharge Detection Technique Based on the In-Phase and Quadrature Signals. IEEE Trans. Microw. Theory Tech. 2021,
69, 5429–5438. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.