Research Paper

The analysis and design of multi-layer microstrip moisture

sensor for rice grain
Fariba Jafari*, Kaida Khalid, W.M. Daud W. Yusoff, Jumiah Hassan
Microwave Laboratory, Dept. of Physics, Faculty of Science, University Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia
a r t i c l e i n f o
Article history:
Received 25 August 2009
Received in revised form
25 January 2010
Accepted 12 April 2010
Published online 23 May 2010
The analysis and optimal design of a multi-layered microstrip sensor for measuring
moisture content of rice grain are described. The microstrip sensing structure consists of
three layers: substrate, protective layer and semi-innite grain medium. The effective
dielectric constant, characteristic impedance and attenuation or insertion loss of this
structure are calculated at various moisture contents with respect to the thickness of the
protective layer thickness. The sensor is operated at 9 GHz and is suitable for a broad range
of moisture contents (MC) ranging from 10% to 30% (wet basis). This study investigated the
optimal thickness of the protective layer for suitable sensitivity, the thickness of grain
medium which can be consider as semi-innite thickness, the effect of kernel density and
loading method. The experimental results for the attenuation of the signal at various
moisture contents agree reasonably with the theoretical prediction however for better
accuracy, the compactness of grain medium should be observed. The effect of air inside the
medium and compactness of the material under the test has been studied and the errors of
0.16 dB have been calculated for the measured reected attenuation of the sensor. The
overall accuracy of the sensor is about 1.5% moisture content as compared to oven drying
method.
2010 IAgrE. Published by Elsevier Ltd. All rights reserved.
1. Introduction
Performing fast, accurate moisture content measurements is
of great importance in agriculture, manufacturing, process-
ing, storing and trading of most products and raw materials.
Moisture content (MC) of granular and particulate materials is
generally the most important characteristic used for quality
assessment and process control (Kaatze, 2005; Kraszewski,
1996). The moisture level of grain is also very important
during harvesting, trading, during transportation and storage.
As dene by standards, long term storage is safe for MC below
13% (free from microbial degradation). However over-drying
the grain can decrease its nutritional and reproductive values
and contribute to increased breakage during handling
(Mohamed, Roman, Nelson, Kraszewski, & Barwicz, 1999).
In the late sixties, microwave techniques emerged as
apromising alternativefor rapidandnon-destructive sensing of
moisture. However, the expected widespread commercial use
never materialised because of cost, calibration and accuracy
concerns. In recent years, the cost of microwave components
has dropped drastically because of a surge in demand from the
wireless telecommunications sector (Trabelsi & Nelson, 2007).
Microwave techniques have attracted a lot of interest inthe
agricultural industry especially for measuring MC in grains.
* Corresponding author.
E-mail addresses: fariba3117@gmail.com (F. Jafari), kaida@fsas.upm.edu.my (K. Khalid), wmdaud@fsas.upm.edu.my (W.M.D.W.
Yusoff), jumiah@fsas.upm.edu.my (J. Hassan).
Avai l abl e at www. sci encedi r ect . com
j our nal homepage: www. el sevi er . com/ l ocat e/ i ssn/ 15375110
b i o s ys t e ms e ng i ne e r i ng 1 0 6 ( 2 0 1 0 ) 3 2 4 e3 3 1
1537-5110/$ e see front matter 2010 IAgrE. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.biosystemseng.2010.04.005
Thesuccess of microwavesensingis relatedtothe polar nature
of the water molecule and its high dielectric constant in
comparison to that of the host material (dry matter) (Trabelsi,
Nelson, & Lewis, 2009). It is based on the fact that, at micro-
wave frequency, the permittivity of water is much higher than
that of most dry substances. The stripline sensor operating at
microwave frequency offers distinct advantages over the other
methods. In such measurements, the materials under test are
brought into contact with the free face of the stripline and
insertion loss of the material under test is calibrated as
a function of its moisture content (Kent & Kohler, 1984).
Microwave sensors have been applied for moisture
measurement of grain by various researchers: microwave
microstrip ring resonator (Abegaonkar, Karekar, & Aiyer,
1999), horn/lens antennae (Trabelsi & Nelson, 2003) for
wheat grain and corn, prototype grain moisture meter con-
sisted of a dielectric resonator type oscillator specically for
rice grain (Kim, Kim, Lee, & Noh, 2002) and dielectric methods
for shelled peanuts from free space measurement of attenu-
ation and phase shift (Trabelsi et al., 2009).
In this project, a microstrip sensor which can be inserted
directly into the grain container is proposed as shown in Fig. 1
(a). The sensing area is a semi-innite 3-layered microstrip
sensor consisting of substrate, protective layer and semi-
innite thickness grain medium. The wave is transmitted
fromone side and received on the other side as shown in Fig. 1
(a). The cross section of the sensor is shown in Fig. 1(b).
The attenuation of the sensor can be estimated by using
the sensor of Fig. 2(a), since the construction of both sensing
areas are quite similar. This sensor consists of three parts: the
coupling system (input/output) representing the transition
between coaxial and stripline, the 50 U stripline section, and
the semi-innite layer of microstrip line. The sensing area is
a semi-innite 3-layered microstrip and its cross section is
shown in Fig. 2(b).
This paper describes the analysis and design of a micro-
strip sensor for moisture determination of rice grain. In this
analysis the effect of thickness of protective layer, thickness
of rice grain sample, loading methods and density on the
signal attenuation are shown.
Fig. 1 e (a) Front and side views of the sensor. (b) Cross section of sensor structure layers.
Notation
C capacitance (F)
c speed of the light in vacuum (m s
1
)
C
a
line capacitance (F)
d height of sample or wet media (mm)
h substrate thickness (mm)
MC moisture content (%)
m
a
, m
g
, m
w
mass of air, grain and water content (g)
Q total charge (C)
q
1
, q
2
, q
3
dielectric lling fractions (F m
1
)
s thickness of protective layer (mm)
S
21
scattering parameter (Port 1 to Port 2) (dB)
W width of line (mm)
Z
0
characteristic impedance (U)
a dielectric constant (1 m
1
)
b phase constant (rad m
1
)
s conductivity (S m
1
)
u angular frequency (radians s
1
)
3
/
dielectric constant (F m
1
)
3
//
loss factor (F m
1
)
b phase constant (dimensionless)
r density (g cm
3
)
r(x,y) surface charge density (C m
2
)
g(b) Fourier transform of the initially unknown charge
density
3
0
permittivity of vacuum (F m
1
)
3
r
relative permittivity (complex) (dimensionless)
m
0
permeability of vacuum (H m
1
)
v
a
, v
g
,v
w
volume fraction of air, grain and water
(dimensionless)
b i os y s t e ms e ng i ne e r i ng 1 0 6 ( 2 0 1 0 ) 3 2 4 e3 3 1 325
2. Theoretical analysis
2.1. Analysis of the attenuation of the microstrip sensor
Semi-innite three-layered microstrip structure of Fig. 2 can
be analysed as a 4-layered semi-innite microstrip as shown
in Fig. 3, where W is the width of the conducting strip and 3
r1
,
3
r2
, 3
r3
are the permittivities of the dielectric layers with
thickness of h, s, d respectively and the fourth layer is air with
permittivity of 3
r4
. Characteristic impedance and phase
constant are two important transmission parameters
involved in this analysis. The propagation of the electromag-
netic wave in the microstrip is based on quasi-transverse
electromagnetic mode (Collins, 1964, p. 154).
The characteristic impedance and the effective dielectric
constant of the structure can be calculated by solving Lap-
laces equation subjected to the appropriate boundary condi-
tions. Following the previous method of analysis (Yamashita,
1968), the line capacitance per unit length, C can be expressed
as,
1
C
=
1
p3
0
Q
2
_
0
[r(b)[
2
g(b)db (1)
where 3
0
is the free space dielectric constant; Q the total
charge per unit length on the strip conductor; [r(b)[ the Fourier
transform of initial unknown charge density distribution.
With fourth layer being air and the thickness considered as
semi-innite, 3
r4
= 1, and g(b)can be derived as
When medium 3 is semi-innite (d /N), g(b) the Fourier
transformof the initially unknowncharge density distribution
equation (2) can be reduced to
g(b) =
1
b
_
3
r2
cothbs 3
2
r3
_
[3
r3
(3
r2
coshbs 3
r1
cothbh) 3
r2
(3
r2
3
r1
cothbscothbh)[
(3)
Different trial functions maybe assumed for the charge
distribution r(x), all of which will have a minimum charge
along the central axis, increasing with varying degrees of
steepness at the strip edge. One of the well-known normalised
trial functions was tested (Yamashita, 1968) and is given by
r(x) =
_
1 [
2x
W
[
3
when
W
2
_ x _
W
2
0 otherwise
(4)
with the Fourier transform
r(b)
Q
=
8
5
_
sin
_
bW
2
_
bW
2

12
5
_
bW
2
_
2
cos
_
bW
2
_
2
sin
_
bW
2
_
bW
2

sin
2
_
bW
4
_
_
bW
4
_
2
_
(5)
The effective dielectric constant (Khalid & Abbas, 1996,
chap. 16; Khalid, Maclean, Razaz, & Webb, 1988) can be
calculated from the total capacitance as:
3
eff
=
C
C
a
(6)
where C and C
a
are the line capacitance per unit length for the
actual and the air-spaced dielectric layers, respectively; the
characteristic impedance of the structure is given by:
Z
0
=
1
c

C
a
C
_ (7)
where c is the speed of light in vacuum.
The variation of characteristic impedance Z
0
, and effective
dielectric constant 3
eff
versus substrate thickness W/h for
various thicknesses of RT-Duroid substrate s/h as a protective
Fig. 2 e (a) Microstrip sensor. (b) Cross section of microstrip
sensing region.
g(b) =
1
b
_
3
r3
cothbd 3
r2
cothbs 3
r2
3
r3
cothbscothbd 3
2
r3

_
3
r1
3
r2
3
r3
cothbscothbdcothbh 3
r1
3
r3
cothbhcothbd 3
r2
3
r3
cothbscothbd 3
r1
3
r2
cothbhcothbs 3
2
r2
3
r3
cothbd
3
2
r3
3
r2
cothbs 3
2
r3
3
r1
cothbh 3
2
r2

(2)
Fig. 3 e Semi-innite double covered microstrip line.
b i o s ys t e ms e ng i ne e r i ng 1 0 6 ( 2 0 1 0 ) 3 2 4 e3 3 1 326
layer with 3
r1
= 2.2 j 0.002, for semi-innite 3-layered
microstrip for a frequency of 9 GHz is shown in Fig. 4.
The signal attenuation can be calculated with a simplied
model of Fig. 2 as shown in Fig. 5 with the transition region
between coaxial cable stripline connector, stripline section
and microstrip sensing area.
The sensor structure can be represented by Fig. 5 and
calculationof S-parameters for this structure has already been
presented (Warner, 1977, pp. 7e8).
In order to calculate the attenuation of the sensor, S
21
, the
complex propagation constant for the microstrip section, g
m
and stripline section, g
s
must rst be established which can be
written as
g
m
= (a
m
jb
m
) and g
s
= (a
s
jb
s
) (8)
where a
m
is the attenuation constant and b
m
is the phase
constant for sensing area, while a
s
is the attenuation constant
and b
s
is the phase constant for stripline section.
In the case of microstrip when the dielectric is non-
homogeneous, the general expression for dielectric loss a
m
(in
dB m
1
) can be given as (Welch & Prat, 1966, pp. 100e101;
Gupta, Garg, & Bahl, 1979, pp. 68e69):
a
m
=
27:3s
eff
l
0
u3
0

3
eff
_ (9)
The effective conductivity and effective dielectric constant
of the three layer semi-innite microstrip structure can be
written as:
s
eff
= q
1
s
1
q
2
s
2
q
3
s
3

_
1 q
1
q
2
q
3
_
s
4
(10)
3
eff
= q
1
3
r1
q
2
3
r2
q
3
3
r3

_
1 q
1
q
2
q
3
_
3
r4
(11)
where q
1
, q
2
and q
3
are the dielectric lling fractions of layer1,
layer2 and layer3 respectively, which can be obtained by
numerical root nding method.
q
1
=
_
3
eff
3
r234
_
(3
r1
3
r234
)
(12)
q
2
=
__
3
eff
3
r34
_
q
1
(3
r1
3
r34
)
_
(3
r2
3
r34
)
(13)
q
3
=
_
1 q
1
q
2
_
(3
r34
3
r4
)
(3
r3
3
r4
)
(14)
Therefore, the nal attenuation of the whole structure a
m
is
given by:
a
m
=
27:3
c

3
eff
_ f
_
q
1
3
r
tand
1
q
2
3
r2
tand
2
q
3
3
r3
tand
3

_
1 q
1
q
2
q
3
_
3
r4
tand
4

(15)
where tand
1
, tand
2
,tand
3
and tand
4
are the loss tangents for
layer1, layer2, layer3 and layer4 respectively.
b
m
can be calculated from
b
m
= b
o

3
eff
_
(16)
where b
o
is the phase constant.
As d/h increases to the semi-innite thickness, a
m
in
equation (15) can be considered as the attenuation for 3-layer
semi-innite microstrip structure after suitable steps are fol-
lowed. a
s
and b
s
can be calculated by TEM analysis and are
given in detail by Gupta, Chandha, and Garg (1981, p. 60).
Fig. 4 e Characteristic impedance and effective dielectric constant versus W/h for semi-innite microstrip: B, s/h [ 0.01;
- s/h [ 0.1; 3, s/h [ 0.3; e, s/h [ 0.5; 9, s/h [ 1.0; ---, s/h [ 2.5.
Transition
region-b 1 2
Z
0
Z
0
Z
m
Sample
region
Stripline
section
Stripline
section
Transition
region-a
Sample
reference
planes
Fig. 5 e Microstrip sensor with sample inserted.
b i os y s t e ms e ng i ne e r i ng 1 0 6 ( 2 0 1 0 ) 3 2 4 e3 3 1 327
2.2. Relationship between dielectric properties of the
sample and its moisture content
In order to nd the relation between effective dielectric
constant of the micrsotrip structure and moisture content, the
dielectric properties of the layers must be known. A grain
medium as third layer of the microstrip structure can be
considered as a mixture of water, grain and air, with dielectric
constant, 3
r3
. Using mixture theory, the relative complex
permittivity of the rice grain medium(Kraszewski, Kulinski, &
Matuszewski, 1976) can be expressed as:

3
+
m
_
= v
w

3
+
w
_
v
a

3
+
a
_
v
g

3
+
g
_
(17)
where v
w
, v
g
, and v
a
are the volume fraction of the water, dried
grain and air respectively, and3
+
w
,3
+
g
and 3
+
a
are the correspond-
ing complex permittivities. The volume fraction of dried rice
kernel v
g
is considered to be constant and estimated in the
range of 0.64e0.71 at 9 GHz which seems to be closed to those
reported earlier for white rice (You & Nelson, 1988), and then
v
w
can be written:
v
w
= 1 v
a
v
g
(18)
Since the volume fraction of rice grain is constant, the
volume fraction of water can be obtained if v
a
is known.
The moisture content of the sample is expressed on the
wet basis as:
MC =
m
w
m
w
m
g
m
a
(19)
where m
w,
m
g,
m
a
are the mass of the water inside grain seeds,
dry rice grain and air in the mixture material respectively. The
equation(19) can be writtenin terms of density r and volume v
as follows:
MC =
r
w
v
w
r
w
v
w
r
g
v
g
r
a
v
a
(20)
By considering the rice sample as a compact space withxed
volume, when we add water gradually to a compact volume of
dried rice sample, it canll up the air space betweenthe seeds;
so this known measured volume of water can be equal to the
volume of free water in dried grain mixture media v
a
. The
experimental value for v
a
is about 0.3 0.06 (dried rice).
By considering the dried grain (v
w
= 0), volume fraction of
grain can be written as
v
g
= 1 v
a
v
w
(21)
The volume fraction of dried rice kernel v
g
is considered to
be constant and estimated in the range of 0.65e0.71 from
above equation.
By substituting equation (18) into (20), for r
w
= 1 (g cm
3
)
the volume fraction of total air replacing free water can be
written as
v
a
=
MC
_
1 v
g
r
g
v
g
_

_
v
g
1
_
(MC 1 r
a
MC)
(22)
Therefore, byknowing v
g
(dry basis), volume fractionof air for
free water v
a
can be estimated for different moisture content of
the grain samples. Since the volume fraction of rice grain is
constant, the volume fraction of water can be obtained if v
a
is
known in certain moisture content; then v
w
can be written:
v
w
= 1 v
a
v
g
(23)
Fig. 6 shows the attenuation of 4-layered microstrip which
is affected by the thickness of material under the test at
moisture content of 10% and 30%. The microstrip sensor
parameters are given by 3
r1
= 2.2, s/h = 0.3 and W/h = 3.18.
The theoretical value of attenuation for 4-layer microstrip
structure (curve) is getting closer to the attenuation of 3-layer
semi-innite microstrip structure (line), as d/h increases to
avaluegreater than15for 20%MCand30for 40%MC. Therefore,
we can consider the structure of 4-layer as a three layer semi-
innite microstrip for the d/h ratio larger than 30 (d/h _ 30).
3. Methodology
3.1. Materials under test and moisture measurement
The samples with several moisture contents (9e30%) were
prepared for the measurement, using a standard oven
method. The moisture content is determined by drying 10 g
grain samples for 24 h at 135

C (Joseph, 2000).
Themicrostripmoisturesensor was calibratedandtestedon
ve types of rice grains. The ranges of thickness and length of
the grain seeds were 1.53e1.77 mm and 5.23e7.63 mm respec-
tively. The density and shape of the rice grain can affect the
values of the volume fraction of grain, therefore due to the
mixture model theory, the dielectric constant and the loss
factor of rice medium also depends on the bulk density. It has
also been reported that both the dielectric constant and loss
factor of cereal grain and oilseed measured at microwave
frequencies increase linearly with bulk density, and moisture
content at xed temperature (Kraszewski, & Nelson, 1996;
Lawrence, Nelson, &Kraszewski, 1990; Trabelsi &Nelson, 2005;
Boldor, Sanders, & Simunovic, 2004).
0 3 5 8 10 13 15 18 20 23 25 28 30
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
A
t
t
e
n
u
a
t
i
o
n

d
B
(
1
.
7

c
m
)
-
1
d/h
Fig. 6 e Comparison of total attenuation between 4-layer
microstrip and 3-layer semi-innite microstrip sensor: (---),
40% MC in 3-layer; (d), 10% MC in 3-layer; -, 10% MC in
4-layer; :, 40% MC in 4-layer.
b i o s ys t e ms e ng i ne e r i ng 1 0 6 ( 2 0 1 0 ) 3 2 4 e3 3 1 328
The dielectric constant for the rice grain under this study
varies between 2.2 and 2.5 and loss factor is below 1 for dried
ground samples. The average values of physical and dielectric
properties for the rice grain samples, are shown in Table 1.
The measurements of dielectric properties for rice grain
were carried out with 4 mm open-ended coaxial line sensor
(HP-85070A) and a computer controlled Network Analyser
(HP-8720B) with frequencies ranging from 130 MHz to 20 GHz.
This system measured the input reection coefcients of the
sensors at room temperature (22e24

C) and specimen
permittivity was calculated automatically by the software
using an established relationship (Kraszewski, Stuchly, &
Stuchly, 1982). The average value of dielectric properties was
obtained from 20 measurements for each sample.
3.2. Measurement of attenuation
The sensor used for measurement of attenuation consists of
three parts: the coupling system (input/output) representing
the transition between coaxial and stripline, the 50-U stripline
section which has been covered by aluminium plate, and the
semi-innite layer microstrip line, as shown in Fig. 2(a). The
substrate for stripline andmicrostripsectionis RT-Duroid5880
withdielectricconstant 3
r
=2.20.02andthicknessof h=1.575
mm. The width W of stripline was chosen with the ratio W/
h = 1.72 to give the characteristic impedance of 50 U with the
length of l
2
= 1.7 cm, while for the microstrip sensing area, W/
h=3.18, length, l
1
=3.5 cmandthethickness of protectivelayer,
s/h = 0.3. The HP 8720B vector network analyser was used to
measure S
21
of the semi-innite microstrip moisture sensor.
4. Results and discussion
The dielectric constant of rice kernel was found to increase
with the moisture content of grain. The moisture dependence
of the dielectric properties of the rice grain at 9 GHz is illus-
trated in Fig. 7. In Fig. 7, the solid line was predicted from
numerical method by using the mixture equation (17). As
grain mixture is not a homogenous medium, to obtain accu-
rate permittivity data, the calibration should be done several
times on the different surfaces of the material under the test.
Theaveragedielectricconstant andloss factor for adriedrice
kernel were 2.2 and 0.2 respectively and for rice kernel with12%
moisture at 9 GHz were 4.08 and 0.55. These values are compa-
rable withthose of reportedfor white rice (Yagihara et al., 2007).
The variation of total attenuation of microwave signal in dB
(1.7 cm)
1
for the whole structure of sensor versus MC (%)
indicates that the sensitivity and attenuation of the sensor are
drastically affected by the thickness of the covered protective
layer for Basmati rice (Fig. 8). For too thick protective layer
(s/h = 1.0), probably due to the decreased interaction of stray
eld with the sample, the measured attenuation data has
greater deviation from the predictions; another possible
explanation could be not including the effect of radiation loss
Table 1 e Dielectric properties with Density and Bulk Density for Dried Rice Samples, at Operating Frequency 9 GHz.
Type of rice sample Kernel density,
r, g cm
3
Volume fraction
of kernel, v
g
Dielectric
constant, 3
g
/
Dielectric
loss, 3
g
//
Type B (Thai white rice) 0.726 0.65 2.2 0.15
Type M (Glutinous rice) 0.737 0.66 2.3 0.16
Type D (Basmati rice) 0.762 0.68 2.38 0.18
Type A (mixture of Brown rice
and white rice)
0.764 0.70 2.41 0.21
Type C (Indian white rice) 0.772 0.71 2.5 0.25
8 10 12 14 16 18 20 22 24 26 28 30
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
8.0
D
i
e
l
e
c
t
r
i
c

p
r
o
p
e
r
t
i
e
s
Moisture content (%)
Fig. 7 e Comparison between measured and calculated
dielectric loss (< and ---), dielectric constant (
*
and d) of
Basmati rice grain as a function of moisture content.
0 3 6 9 12 15 18 21 24 27 30
0
1
2
3
4
5
6
7
A
t
t
e
n
u
a
t
i
o
n

d
B
(
1
.
7
c
m
)
-
1
Moisture content (%)
Fig. 8 e Comparison between predicted and measured
values of attenuation versus moisture content (wet basis)
for different values of s/h: k, measured moisture content for
8, s/h [ 1.0; C, s/h [ 0.7; -, s/h [ 0.3;
*
, s/h [ 0.1 and
predicted values as lines : continuous, s/h [ 1.0; dashed,
s/h [ 0.7; dotted, s/h [ 0.3; dash-dot, s/h [ 0.1.
b i os y s t e ms e ng i ne e r i ng 1 0 6 ( 2 0 1 0 ) 3 2 4 e3 3 1 329
in the theoretical model. Therefore it is suggested to use the
hybrid model for future work, to obtain more accurate results.
Based on the data from Table 1, the dependence of the
calculated microwave attenuation on moisture content for
ve types of rice is shown in Fig. 9. The sample with higher
density gives much higher attenuation, therefore type C rice is
highest and type B rice is the lowest. At a particular moisture
content, for example at 25% MC, the variation of density from
0.726 to 0.772 gives a variation in attenuation of about 0.7 dB
which corresponds to 6% MC.
Fig. 10 shows the comparison of the value of moisture
content obtained from the attenuation measurement by
microstrip sensor and standard MC by oven method for Bas-
mati Rice The standard error is about 1.44 MC% with R
2
value
of 0.98.
In addition, we have studied the effect of loading method
on measurement by comparing measurements on slowly-
dropped rice grains with grains dropped suddenly into the
sensing area. To nd out the effect of microwave eld which is
propagated through the line to the rice, we also have
measured the moisture content of rice grains when they have
been arranged with long axis parallel to the line and at right
angles (vertical) to the line. It was found that for a certain
moisture content, the value of attenuation varied by around
0.4 dB for different loading methods; slow drop method gave
the lowest result and when we arranged the rice seeds verti-
cally to the microstrip line, attenuation values by the sensor
were slightly higher. The results for various types of loading at
11%, 22% and 30% MC are shown in Fig. 11.
As the rice mediumis not homogenous so the compactness
of the sample can affect the attenuation measured. Therefore
to nd out the errors, several measurements of attenuation
for certain moisture content sample and from same type of
rice were collected. The effect of compacted and non-com-
pacted material on S
21
measured values is shown in Fig. 12
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
0.0
0.3
0.5
0.8
1.0
1.3
1.5
1.8
2.0
2.3
2.5
2.8
3.0
3.3
3.5
3.8
4.0
4.3
4.5
4.8
5.0
5.3
5.5
A
t
t
e
n
u
a
t
i
o
n
,
d
B
(
1
.
7
c
m
)
-
1
Moisture Content, MC%
Fig. 9 e Variation of the attenuation versus moisture
content for ve different type of rice grain: -, Type D
(Basmati rice);
*
, Type C (Indian white rice); C, Type M
(Glutinous rice); ,, Type B (Thai white rice); B, Type A
(mixture of Brown rice and white rice).
0 5 10 15 20 25 30 35 40 45
0
5
10
15
20
25
30
35
40
45
M
e
a
s
u
r
e
d

M
o
i
s
t
u
r
e

C
o
n
t
e
n
t

%
Standard Moisture Content %
Fig. 10 e Comparison between standard (straight line) and
measured moisture content values (C) of Rice Grain.
Slow Drop parallel To Line sudden Drop Vertical To Line
3.0
3.5
4.0
4.5
5.0
5.5
6.0
A
t
t
e
n
u
a
t
i
o
n

d
B
(
1
.
7

c
m
)
-
1
Type of Loading Method
Fig. 11 e The effect of loading method in moisture content
measurement using the microstrip moisture sensor for rice
grain: B, 11% MC;
*
, 22% MC; ,, 30% MC.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
2.8
2.9
3.0
3.1
3.2
3.3
3.4
3.5
3.6
3.7
A
t
t
e
n
u
a
t
i
o
n

d
B
(
1
.
7
c
m
)
-
1
Number of Measurements
Fig. 12 e The effect of compactness, presenting the average
value of measured attenuation for several measurements
of rice with 22% moisture content.
b i o s ys t e ms e ng i ne e r i ng 1 0 6 ( 2 0 1 0 ) 3 2 4 e3 3 1 330
with standard errors of 0.16 dB for the data calibration at
frequency 9.0 GHz.
5. Conclusions
Theoretical and experimental studies have been carried out to
formulate the relationship between geometrical and electrical
parameters of the sensor with the attenuation with respect to
the moisture content in the rice grain. The sensor was fabri-
cated on a laminate RT/Duroid 5880 with W/h = 3.18 and
1.7 cm length and protective layer s/h = 0.3, 3.5 cm length and
working at 9 GHz.
In this study, investigation was made to nd the relation
between the dielectric properties and moisture content of rice
grains. As the grain medium is not homogenous, dielectric
mixture theory has beenderived to solve the problem. To verify
the mixture theory, the theoretical results and empirical results
were comparedandcloseagreement betweenthe results allows
us to determine the complex permittivity of the mixture with
particular moisture content from the relative density and
complex permittivity of the dry material. The sensitivity and
error of the sensor were very much affected by the thickness of
protectivelayer of microstripline, densityof ricegrains, effect of
loading method. For compactness, the error is around 1e2%
while, for the density, the error can go up to 6%.
The problem of compactness and loading effect can be
reduced by using a data acquisition system that can take data
from various positions in the container or sack and take the
average value from that measurement.
A close agreement has been obtained between computed
and experimental results of the insertion loss as a function of
moisture content in rice grain. This agreement suggests that
a realistic optimisation-based design approach with respect to
the effects of the geometrical and electrical parameters of the
sensing area can be developed.
Acknowledgement
We gratefully acknowledge research grant from Ministry of
Science, Technology and Innovation under the IRPA funding
(Project No: 09-02-04-0438-EA 001). Thanks are also due to all
staff of Physics Department, Faculty of Science, University of
Putra Malaysia for their support and generous assistance.
r e f e r e n c e s
Abegaonkar, M. P., Karekar, R. N., & Aiyer, R. C. (1999). A
microwave microstrip ring resonator as a moisture sensor for
biomaterials: applied to wheat grains. Journal of Measurement
Science and Technology, 10, 195e200.
Boldor, D., Sanders, T. H., & Simunovic, J. (2004). Dielectric
properties of in-shell and shelled peanuts at microwave
frequencies. Transactions of the ASAE, 47(4), 1159e1169.
Collins, R. E. (1964). Field theory of guided waves. New York:
McGraw-Hill.
Gupta, K. C., Chandha, R., & Garg, R. (1981). Computer aided-design
of microwave circuit. Dedham, Massachusetts, USA: Artech
House Inc.
Gupta, K. C., Garg, R., & Bahl, I. J. (1979). Microstrip lines and
slotlines. Dedlam, Massachusetts, USA: Artech house.
Joseph, St (2000). Moisture measurement e unground grain and
seeds. ASAE Standards 563.
Kaatze, U. (2005). Electromagnetic wave interaction with water
and aqueous solutions. In K. Kupfer (Ed.), Electromagnetic
aquametry (pp. 15e37). Berlin: Springer.
Kent, M., & Kohler, J. (1984). Broadband measurement of
stripline moisture sensors. Journal of Microwave Power, 19(3),
448e453.
Khalid, K. B., & Abbas, Z. (1996). Development of microstrip
sensor for oil palm fruits. In A. Kraszewski (Ed.), Microwave
aquametry (pp. 239e248). New York: IEEE Press.
Khalid, K. B., Maclean, T. S. M., Razaz, M., & Webb, P. W. (June
1988). Analysis and optimal design of microstrip sensor. IEEE
proceedings, 135(3), 188, Pt. H.
Kim, K., Kim, J., Lee, S. S., & Noh, S. H. (2002). Measurement of
grain moisture content using microwave attenuation at 10.5
GHz and moisture density. IEEE Transaction on Instrumentation
and Measurement, 51(1).
Kraszewski, A. (1996). Microwave aquametry. Piscataway, NJ: IEEE.
Kraszewski, A. W., Kulinski, S., & Matuszewski, M. (1976).
Dielectric properties and a model of biphase water suspension
at 9.4 GHz. Journal of Applied Physics, 47, 1275e1277.
Kraszewski, A. W., & Nelson, S. O. (1996). Moisture Content
Determination in Single Kernels and Seeds with Microwave
Resonant Sensors. In A. Kraszewski (Ed.), Microwave Aquametry
Electromagnetic Wave Interaction with Water Containing Materials
(pp. 177e203). IEEE Press.
Kraszewski, A. W., Stuchly, M. A., & Stuchly, S. S. (1982). ANA
calibration method for measurement of dielectric properties.
IEEE Transactions on Instrumentation and Measurement, 32(2),
385e387.
Lawrence, K. C., Nelson, S. O., & Kraszewski, A. W. (1990).
Temperature dependence of the dielectric properties of
wheat. Transactions of the ASAE, 33, 535e540.
Mohamed, B. S., Roman, Z. M., Nelson, S. O., Kraszewski, A. W., &
Barwicz, A. (1999). Calibration of a microwave system for
measuring grain moisture content. IEEE TRANSACTIONS ON
INSTRUMENTATION AND MEASUREMENT, 48(3), 778.
Trabelsi, S., & Nelson, S. O. (2005). Microwave dielectric properties
of cereal grain and oilseed. In ASAE annual international meeting
(Tampa, FL), Paper Number 056165.
Trabelsi, S., & Nelson, S. O. (2007). Unied microwave moisture
sensing technique for grain and seed. Journal of Measurement
Science and Technology, 18, 997e1003.
Trabelsi, S., Nelson, S. O., & Lewis, M. A. (2009). Microwave
nondestructive sensing of moisture content in shelled
peanuts independent of bulk density and with temperature
compensation. Sensing and Instrumentation for Food Quality and
Safety, 3, 114e121.
Trabelsi, S., & Nelson, S. O. (2003). Free-space measurement of
dielectric properties of cereal grain and oilseed at microwave
frequencies. Journal of Measurement Science and Technology, 14,
589e600.
Warner, F. L. (1977). Microwave attenuation measurement. London.
Welch, J. D., & Prat, S. H. J. (1966). Losses in microstrip trans. systems
for integrated microwave circuits. In NEREM, Vol. 8.
Yagihara, S., Oyama, M., Ionue, A., Asano, M., Sudo, S., & Naoki, S.
(2007). Dielectric relaxation measurement and analysis of
restricted water structure in rice kernels. Journal of
Measurement Science and Technology, 18, 985.
Yamashita, E. (1968). Variational method for the analysis of
microstrip-like transmission lines. IEEE Transactions on
Microwave Theory and Techniques-16, 18, 529e535.
You, T.-S., & Nelson, S. O. (1988). Microwave dielectric properties
of rice kernels. Journal of Microwave Power and Electromagnetic
Energy, 23, 150e159.
b i os y s t e ms e ng i ne e r i ng 1 0 6 ( 2 0 1 0 ) 3 2 4 e3 3 1 331