Design of a 3D Microwave
Imaging System
Drew Jaworski
Advisor: Dr. Yong Zhou
Fall 2011 – Senior Design I
Why Microwave Imaging?
Electromagnetic Imaging Systems
Vision
X-Ray
Expensive
Microwave
Limited to surfaces
MRI (quantum mechanics)
Ionizing radiation
Infrared (“thermal”)
Nature doesn’t always know best!
Non-ionizing, penetrating, less expensive!
Applications
Medical imaging (cancerous tumors, etc)
Industrial scanning (forging defects, etc)
Project Specifications
Design of a 3-dimensional microwave
imaging system
Vector network analyzer signal analysis
Automated data acquirement and
processing
Biomedical focus, but adaptable for other
imaging applications
Multiplexed antenna array
Project Constraints
Size
Budget
$300 from department + personal funds
FCC Regulations
Entire system less than 1[m]*1[m]*1[m]
Medical device band: 3.1[GHz]–10.6 [GHz]
Many others related to trying to manage
the above constraints
Electromagnetic Overview
Plane-wave approximation
Imaging subject located in far-field of
antenna array, perpendicular to
propagation of waves
Simplifies analysis at expense of
system size
Scattering through media
A result of multiple layers of
diffraction and refraction, in the
case of the complex human body.
Images courtesy of:
http://en.wikipedia.org/wiki/File:Linear.Polarization.Linearly.Polarized.Light_plane.wave.svg
http://commons.wikimedia.org/wiki/File:Huygens_brechung.png
Vector Network Analyzer
Measures a Two Port Network
Returns S-Parameters (Scattering Parameters)
S11 – Return Loss
S21 – Insertion Loss
Parallel antennas connected to VNA ports
Calibrate response
Place object between antennas
Result is how the object affected the
electromagnetic radiation between the two
antennas
Results can be manipulated with software algorithms
to give dielectric properties of the object!
Inverse Scattering Solution
Repeat with multiple antennas (4x4 array, in this case)
Rotate object between antenna arrays
Result is a set of matrixes of scattering parameters for a 32-port
network (for a range of frequencies!)
Can be manipulated to produce a discretized graphical representation of
the dielectric properties in different regions between antenna arrays
Inverse Scattering Problem – Microwave Tomography
We know the forward transmitted radiation (aka Incident Fields)
We have information about the received fields (aka Scattered Fields)
Now we want to know what made them change!
Very complex calculations that are demanding of computing resources
Fortunately, much research has been published that has mathematically
and/or computationally simplified the solution process (relatively)
Automated Data Analysis
Labview
Automate collection of data
Several colleagues have worked out the details
Rotation mechanism – Juan Nava, Miguel Rivera
TTL communication (for multiplexer) – Julio Vasquez
Matlab
Process data
Numerous published algorithms can be
implemented and tested
Frequency Selection
Often limited by hardware technology (switch/antenna bandwidth)
Biomedical focus – human tissues
Estimates vary, best to come up with your own and justify
accordingly
Begin with what spectrum is available
FCC
“Medical Systems: These devices must be operated in the frequency
band 3.1-10.6 GHz. A medical imaging system may be used for a
variety of health applications to “see” inside the body of a person or
animal. Operation must be at the direction of, or under the supervision
of, a licensed health care practitioner.”
http://transition.fcc.gov/Bureaus/Engineering_Technology/Orders/2002/fcc02
048.pdf
Begin with properties of human body
Database of dielectric properties of numerous types of tissue available
from Italian National Research Council site:
http://niremf.ifac.cnr.it/tissprop/
Dielectric Properties Database
Skin (wet and Dry), Muscle, Fat, and Bone
Major constituents most body parts
Wavelength and Attenuation Constant versus Frequency
0.14
Wavelength
[cm] – SkinWet
4
Wavelength [cm] SkinDry
3.5
0.12
Wavelength
[cm] – Muscle
3
2.5
0.08
Highest λ
2
0.06
1.5
Highest alpha
0.04
1
Lowest λ
Lowest alpha
0.02
intersection
intersection
0
0.5
0
Frequency [GHz]
Attenuation Constant [1/cm]
Wavelength in Tissue [m]
0.1
Wavelength
[cm] – Fat
Wavelength
[cm] – BoneCortical
Attenuation
Coefficient [1/cm] –
SkinWet
Attenuation
Coefficient [1/cm] –
SkinDry
Attenuation
Coefficient [1/cm] –
Muscle
Attenuation
Coefficient [1/cm] –
Fat
Attenuation
Coefficient [1/cm] –
BoneCortical
Antenna Array Multiplexer
Julio Vasquez’s RF multiplexer design intended for this
project
Overlapping semesters meant his prototype was not yet
completed and could not be used immediately
Microstrip antenna array with integrated multiplexer
switch hierarchy
Avoids requirement of numerous expensive and tangled
SMA patch cables
Integrates network of SPDT switches into antenna array
4x4 microstrip antenna array
1 SMA connector (patched to VNA)
15 SPDT RF switches (operating up to 8[GHz])
16 microstrip patch antennas
8 TTL-level (5V) control lines
Antenna Array with Multiplexer
RF Layout Guidelines
Line Widths
3.08[mm]
Curves
Ideally smooth curves
radius >= 3*lineWidth
Ground fills
Not completely necessary
50Ω impedance
Relatively noise-free environment
Noise reducing padding around experiment setup
Not feasible for hand-produced prototype
Tapered impedance tranformers
Linear (“triangular”) is best for wideband operation (Pozar)
λ/4 ~ λ used in design (as long as could be reasonably fit)
Multiplexer versus Switch Network
Fully featured DC-12[GHz] multiplexer
$700 ~ $1700
Single SPDT RF switch IC
$1 ~ $3
M/A-COM technology solutions
MASW-007107
Pros
Large variety of models available
Distributed by Mouser
Cons
Small package size (GaAs DIE ~ 4[mm]*4[mm]
MASW-007107
Obtained from IC Datasheet: http://www.macomtech.com/DataSheets/MASW-007107.pdf
MASW-007107 (continued)
Switch Network Hierarchy
UWB Microstrip Antenna
Two-port network theory (one-port input
network, in this case)
S11 measures “return-loss” [dB]
Lower is better, -10[dB] indicates half of the input power is
lost in the network
Return Loss is power radiated from antenna (hopefully)
and other losses.
Bandwidth is measured where S11 crosses the
-10[dB] point
Design is UWB when (BW / Fcenter) >= 25%
UWB Microstrip Antenna
There are many published designs for UWB microstrip antennas
Most use complex ground geometries
(continued)
Usually explain it as something to keep the phase response level across the
useable band
After trying several designs, I began modifying the geometries in an
attempt to find something new
UWB Microstrip Antenna
(continued)
BW = 979[MHz]
Fcenter = 5.595[GHz]
-10[dB] BW => 17.52% (close, but not UWB)
UWB Microstrip Antenna
(continued)
Fractal and/or self-symmetry based designs
Intended to induce multiple resonance frequencies
XY Plot 1
patchAntennaV1
0.00
ANSOFT
Curve Info
dB(S(1,1))
Setup1 : Sw eep
$feedW='1.4mm' $patchHeight='15mm' $patchWidth='20mm'
-10.00
dB(S(1,1))
MY1: -10.0000
-5.00
-15.00
-20.00
-25.00
-30.00
3.50
4.00
4.50
5.00
5.50
Freq [GHz]
6.00
6.50
7.00
7.50
Inspired by: Miniaturized UWB Monopole Microstrip Antenna Design by the Combination of Guisepe Peano and Sierpinksi Carpet Fractals, IEEE AWPL, 2011
Budget (proposed)
Double-sided FR-4 boards (2x)
MASW007107 RF Switches (50x)
$37.50 + shipping <Mouser Electronics>
Commercially Manufactured PCBs
$12.66 + shipping <Parts Express>
$150 <Dorkbot PDX PCB group order>
All other supplies already in possession
Gantt Chart – SD1
9/19/15
Select Frequency Operation
Range
Decide Switching Design/Product
Learn HFSS
Decide Antenna Design
Layout Array PCB Design
Simulate Design Explicitely
Modify PCB Design Accordingly
Produce Protoype Board
9/169/30
10/110/15
10/1610/31
11/111/15
11/1611/30
12/112/15
12/1512/30
Gantt Chart – SD2 (proposed)
1/11/15
Layout Final Design
Send Layout Out for Manufacturing
Test New Boards
Establish Mathematics of System
Program Algorithms
Test System and Collect Data
Prepare for Research Symposium
Final Report and Presentation
1/151/31
2/12/15
2/162/29
3/13/15
3/163/31
4/14/30
5/15/31
Future Work
Finalize UWB antenna candidate design
RF Layout of antenna array
Produce a prototype (using materials on hand)
Export Gerber file and have it manufactured
commercially
Develop mathematics of Imaging System
$1 per square inch (min. 150 square inch order)
Microwave Imaging (2011), Matteo Pastorino
Begin making microwave images!