1 INTRODUCTION

There is growing demand for inexpensive but reli-

able wireless sensors capable of monitoring struc-
tural health parameters. Wireless sensors are an at-
tractive option because they do not require any
connection with the readout unit and can be embed-
ded within the structure. The current state on wire-
less sensing technology offers both active and pas-
sive wireless sensors. Active sensors can provide
relatively long range but require internal batteries.
As batteries require frequent replacement, this type
of sensor has a short life time and does not afford
permanent embedding in a structure. Passive sensors
have reduced transmission distances but do not re-
quire any local power source (Nowak et al. 2006).
They are very advantageous for civil structural
health monitoring where changing batteries or con-
necting wires between the sensor and the interroga-
tor unit is difficult.
Passive wireless sensors consist of a sensing unit
and an interrogation unit. The interrogation unit ge-
nerates the electromagnetic energy for remote inter-
rogation of the sensor. Inductively coupled coil sen-
sors are a class of passive wireless sensor. They
operate by using a resonant coil whose resonant fre-
quency is a function of the measurand. One coil
(sensor coil) is placed in the medium being moni-
tored and the other one is used as an external inter-
rogator coil. The interrogator coil is inductively
coupled with the sensor coil and its purpose is to
track changes in the sensor coils resonant fre-
quency. Coupled coil sensors are found in many ap-
plications e.g., corrosion detection (Andringa et al.
2005), monitoring water content in civil engineering
materials (Ong et al. 2008), strain measurement
(Butler et al. 2002), and moisture measurement
(Hapster et al. 2002). Changes in the sensor coils
resonant frequency are most commonly detected by
measuring the induced change in the impedance of
the interrogator coil. This technique is affected by
the response of the interrogator coil. As the interro-
gation distance increases, the response of the sensor
coil decreases rapidly and only the interrogator coil
self impedance dominates, which is sensitive to the
background environment. Thus, this interrogation
technique requires a small separation distance be-
tween the sensor and the interrogator.
In this paper, we present a new interrogation me-
thod which makes use of time-domain gating to
eliminate the response from the interrogator coil and
the exciting signal. This results in a measurement
dominated by the response of the sensor coil, ena-
bling detection of the sensors resonant frequency
far more precisely at an extended separation dis-
tance. A test coil was first used to demonstrate the
extended separation distance and accuracy of this
technique. This technique was then applied to meas-
urement of relative humidity.
Coupled resonant coil sensors with increased interrogation distance
S. Bhadra, E. Thompson, B. Kordi, G.E. Bridges & D.J. Thomson
Department of Electrical and Computer Engineering, University of Manitoba, Winnipeg, Manitoba, Canada

ABSTRACT: Wireless sensors are being explored for many structural health monitoring applications. Passive
wireless sensors have the advantage of not requiring a power source on the sensor, as the energy is provided
by the interrogation system. One type of passive wireless sensor, the inductively coupled coil sensor, works
by using a resonant coil coupled to a measurand that shifts the sensor coils resonant frequency. This paper
discusses a new interrogation method that improves the accuracy and interrogation range of coupled resonant
coil sensors. The method uses time-domain gating to produce measurements that are dominated by the re-
sponse of the sensor coil and are immune to background interference. Using this method a test sensor coils
resonant peak position was determined within 30 parts per million at a distance of 20 cm. This is much greater
than that achievable with the impedance measurement method. With a relative humidity sensor coil, the me-
thod was able to detect changes in relative humidity less than 2%.
Bridge Maintenance, Safety, Management and Life-Cycle Optimization Frangopol, Sause & Kusko (eds)
2010 Taylor & Francis Group, London, ISBN 978-0-415-87786-2
125

2 THEORETICAL ANALYSIS
2.1 System Operation
A block diagram of the time-domain gating sys-
tem is shown in Fig. 1. The sensor is modeled as a
lumped element resonant circuit and is inductively
coupled to the interrogator coil. During operation,
the interrogator first energizes the sensor through in-
ductive coupling between the interrogator coil and
the sensor coil. This is the transmit mode (with SW1
and SW3 set to drive the interrogator coil). After
enough time has elapsed to allow the induced energy
in the sensor coil to reach equilibrium, the system
switches to the receive mode, where energy stored in
the sensor is coupled back from the sensor to the in-
terrogator. A delay (in this case 0.8 s) is introduced
when the system switches from the transmit mode to
receive mode to eliminate the transient response of
the interrogator coil (this is done using SW2 with
SW3 used to isolate the signal generator in the re-
ceive mode). The received signal is an exponentially
decaying sinusoidal signal that oscillates at the reso-
nant frequency of the sensor. The amplitude of the
received signal depends on the excitation frequency.
The amplitude is the largest when the excitation fre-
quency matches the resonant frequency of the sen-
sor. The rate of decay of the received signal depends
on the Q factor of the sensor. By sweeping the fre-
quency of the exciting signal, the resonant frequency
of the sensor can be determined.



















Figure 1. Block diagram of time-domain gating system

2.2 Sensor Operation
In the sensor unit, an inductive coil is connected
in parallel with a sensing element. The sensing ele-
ment is capacitive and changes its capacitance, C
P
,
with the change of measurand. The coil inductance,
L
S
, is not affected by the measurand. The sensor
forms a resonant circuit, which can be modeled as in
Fig. 2, and includes the series resistance, R
S
, of the
coil and the shunt conductance, G
P
, of the capacitive
sensor element.












Figure 2. Sensor coils equivalent circuit

Assuming the losses are small, the resonant fre-
quency, f
0
, of the sensor is approximated by
P S
C L
f
t
t
e
2
1
2
0
0
= = . (1)
Once energized, and after removing the excitation,
the response from the sensor circuit decays exponen-
tially with a time constant, . The quality factor, Q,
of the circuit can be derived from this time constant
as
2
0
t e
= Q . (2)
2.3 Circuit Model
Figures 3a and 3b show equivalent circuits of the
system in transmit mode and receive mode, respec-
tively. In these circuit models, the mutual capaci-
tance between the coils is very small and is ne-
glected. We also assume the coils self-resistances
are small. The interrogator coil is designed so that
the self-resonant frequency is much higher than that
of the sensor.














Figure 3a. Transmit mode equivalent circuit

V
G
R
1
L
1 L
2
Z
2
I
1
I
2
Interrogator Sensor
M
+
-
+
-
V
1
V
2
Z
in
L
S
R
S

C
P
G
P
126














Figure 3b. Receive mode equivalent circuit

The interrogator source, V
G
, produces a time-
varying current, I
1
, which passes through interroga-
tor coil, L
1
. This current generates a time-varying
magnetic flux which links the loop in the sensor coil,
L
2
. By Lenzs law, a resulting current, I
2
is induced
in the sensor loop given by
1 1
2 2
2
I
Z
M j
I
L j Z
M j
I
S
e
e
e
=
+

= , (3)
where is the source frequency, Z
s
= Z
2
+jL
2
is the
total sensor impedance, Z
2
=R
S
+(1/ (jC
P
+G
P
)), and
M is the mutual coupling between the sensor coil
and the interrogator coil. The voltage, V
1
, at the in-
terrogator coil is given by
1
2 2
1 1 1
I
Z
M
I L j V
S
e
e + = . (4)
The current through the interrogator coil, I
1
, is given
by
G
R S
S
G
S
V
M Z Z
Z
V
Z
M
L j R
I
2 2 2 2
1 1
1
1
e e
e
+
=
+ +
= , (5)
where Z
R
= R
1
+jL
1
is the interrogator impedance.
Further, the impedance seen by the source, Z
in
, is
given from eqn. 5 as
S
in
Z
M
L j Z
2 2
1
e
e + = . (6)
When the interrogator switches to receive mode,
the current induced in the sensor coil oscillates at its
resonant frequency, f
0
, and decays with time con-
stant . Provided 1/ <<
0
(or Q>>1) for the sen-
sor, phasor analysis can be continued. The current
passing through the sensor, I
2
'
, is given by
2
'
2
I I ~ . (7)
Referring to Fig. 3b, the current induced in the in-
terrogator coil, I
1
'
, will be
'
2
0 '
2
1 0 1
0 '
1
0
I
Z
M j
I
L j R
M j
I
R e
e
e
e
=
+

= . (8)
In the receive mode, the voltage developed at the in-
terrogator coil, V
1
'
, can be expressed by combining
eqns. 4, 7 and 8 as
1
1
2
0 '
1 1
'
1
0
I
Z Z
R M
I R V
R S e
e e
= = . (9)
Combining eqns. 5 and 9 then yields the received
signal (at frequency
0
)
( )
G
R R S
V
Z M Z Z
R M
V
0
2 2
1
2
0 '
1
e
e
e e
+
= . (10)
In order to determine the resonant frequency,
0
,
the source frequency, , is swept near
0
and the
maximum received signal, V
1
'
, is obtained using a
peak fitting algorithm.
The signal from the sensor in the receive mode
always oscillates at the resonant frequency,
0
, of
the sensor, but the lock-in amplifier in Fig. 1 is
locked to the exciting signals frequency, . Thus,
the detected signal, S, as a function of frequency is
maximized when =
0.
However, the response does
not directly provide the Q of the sensor. To obtain
the Q, the received signal in the time-domain is ob-
served and fitted with a function of the form Ae
-t/

sin(2f
0
+ ) with unknown parameters A, , f
0
, and
. From and f
0
, Q is determined using eqn. 2.
From eqn. 10, it can be seen that for a fixed excit-
ing frequency the signal from the sensor
(Z
S
Z
R
>>
2
M
2
) can be expressed as
2 '
1
M V . (11)
The mutual coupling, M, can be expressed in terms
of separation distance between the centers of two
coils, R, (Tesche et al. 1997) as
3
1
R
M . (12)
This indicates that the path loss of the transmit-
receive coupled coil system is proportional to
V
1
'
/V
G
1/R
6
and is thus expected to decrease rapidly
with distance.
3 EXPERIMENT
3.1 System Description
A programmable signal generator (Stanford Re-
search Systems DS345) was used to provide a swept
frequency signal to the interrogator. High isolation
SPDT switches (ZASW-2-50DR+) were used for the
switches. A pulse generator (Quantum Composers
Model 9618) provided the required controlling sig-
nals for the switches. The switches enabled the sys-
R
1
L
1
L
2
Z
2
Interrogator Sensor
M
'
1
I
'
2
I
+ +
- -
'
1
V
'
2
V
127

tem to switch between the transmit and receive
mode. An RF lock-in amplifier (Stanford Research
Systems SR844) was used as the detector. The re-
ceived signal could also be observed with an oscillo-
scope. A computer program, developed in Labview,
controlled the swept frequency of the signal genera-
tor, and recorded the output from the lock-in-
amplifier through the General Purpose Interface Bus.
Thus, by sweeping the frequency and measuring the
response, S, the resonant peak of the sensor was de-
termined using a quadratic curve-fitting algorithm
(Robinson & Clegg 2005).
3.2 Interrogator and Sensors
The interrogator coil was constructed of insulated
wire of 1.2 mm diameter and contains 5 turns of 5.1
cm diameter, producing measured values L
1
=2.36
H and a self-resonant frequency, f
res
=26.76 MHz.
Two sensor coils were used in our experiments.
One sensor coil was made of wire of 0.55 mm di-
ameter and contains 19 turns of 8.4 cm diameter,
producing L
2
=66.19 H and R
S
= 12.48 . This coil
was connected in parallel with a fixed capacitor of
C
P
=33.03 pF and G
P
=244.76 nS. This was used as
the test sensor coil with a theoretical f
0
=3.4038
MHz. A second sensor coil was made of wire of 1.2
mm diameter and contains 20 turns of 5.1 cm diame-
ter, producing L
2
=25.30 H and R
S
= 1.86 . A ca-
pacitive humidity sensor HCH-1000 with moisture
sensitive capacitance C
P
and G
P
=282.2 S was con-
nected in parallel with the coil. This was used as a
relative humidity sensor. In all tests the sensor coil
center was aligned concentrically with that of the in-
terrogator coil to get the best response. The separa-
tion distance, R, was measured between the centers
of the two coils.
4 RESULTS
4.1 Test Sensor Coil
Using the experimental setup shown in Fig. 1, the
received signal at different separation distances was
measured for the fixed frequency test sensor. Fig. 4
shows the time-domain signal, v
r
(t), from the sensor
in receive mode for a distance R= 10cm and source
frequency =3.403 MHz. This exponentially decaying
signal was observed with an oscilloscope. An expo-
nentially decaying sinusoid (Ae
-t/
sin(2f
0
+ )) was
fitted to this received signal where, A=0.0299V,
=1.12x10
-5
s, f
0
=3.4038x10
6
Hz, and =3.46 rad.
This yields a Q=120. The response spectrum, S, as
measured with the lock-in amplifier for different se-
paration distances is plotted in Fig. 5.




















Figure 4. Signal from the sensor in receive mode
























Figure 5. Received signal spectrum for different separation dis-
tances

From the peak of the response spectrum, the reso-
nant frequency was determined. The bandwidth of
the system was 41.67 Hz. For this bandwidth, a sig-
nal-to-noise ratio at resonance of 48.5 dB can be
achieved for a distance of 20 cm. Table 1 lists the
resonant frequency for different separation dis-
tances. The results demonstrate 30 parts per million
accuracy at 20 cm.









0 2 4 6 8
-0.04
-0.02
0
0.02
0.04
Time(us)
R
e
c
e
i
v
e
d

S
i
g
n
a
l
(
V
)


3.3 3.35 3.4 3.45 3.5
0
0.5
1
x 10
-3
10cm
3.3 3.35 3.4 3.45 3.5
0
1
2
x 10
-4
15cm
3.3 3.35 3.4 3.45 3.5
0
2
4
x 10
-5
20cm
3.3 3.35 3.4 3.45 3.5
0
0.5
1
x 10
-5
25cm
3.3 3.35 3.4 3.45 3.5
0
2
4
x 10
-6
Frequency(MHz)
S
i
g
n
a
l

(
V
)
30cm
Ae
-t/

128

Table 1. Resonant frequency for different separation distances
______________________________________________
Distance Resonant frequency
______________________________________________
cm MHz
______________________________________________
10 3.4038
15 3.4038
20 3.4037
25 3.4086
30 3.3953
______________________________________________

The received signal versus distance is plotted for a
fixed exciting frequency 3.40496 MHz in Fig. 6. It
can be seen that the response has a 1/R
6
behavior,
showing that the path loss follows the theory estab-
lished in section 2.3.
























Figure 6. Received signal versus distance fitted to a 1/R
6
re-
sponse

In order to compare our interrogation technique
with the impedance measurement method the re-
sponse spectrum was also measured with an imped-
ance analyzer (Agilent 4294A). The impedance ana-
lyzer measures the impedance seen by the source as
given in eqn. 6. The interrogator coil used was iden-
tical to that of the time-domain gating system. The
bandwidth of the system was set to 39.8Hz so that it
was comparable with the time-domain gating meas-
urements. In this case, the resonant frequency was
obtained from the maximum of the real part of the
impedance. To remove the inductance of the coil, a
background subtraction, using the measured imped-
ance of the coil when the sensor was absent, was
implanted prior to measuring the sensor response
(Ong et al. 2008). Fig. 7 shows the spectrum meas-
ured with the impedance analyzer at different inter-
rogation distances. With this method the maximum
interrogation distance obtainable was 15 cm, which
provided a 76.47 dB signal-to-noise ratio at reso-
nance. This interrogation distance was much less
than that of the time-domain gating.



























Figure 7. Real part of the impedance spectrum for different se-
paration distances
4.2 Relative Humidity Sensor Coil
The relative humidity sensor coil described in sec-
tion 3.2 was placed within a plastic humidity cham-
ber. A small piece of foam soaked in water was kept
in the chamber on a petri dish. Dry air was passed
through the chamber using a flow meter (Matheson
Model No. 7640T W/602) to vary the relative hu-
midity. A thermo-anemometer (Alnor Compuflow
Model 8585) was used to continuously monitor the
relative humidity and the temperature inside the
chamber. The air temperature was kept constant at
23.5+0.6C. The setup of the plastic chamber, sensor
and interrogator coil is shown in Fig. 8. The Q of
this sensor was low (Q~13) because the HCH-1000
has high loss due to the conductance, G
P
. Thus a
small separation distance (4.5 cm) was chosen for
the experiment. The relative humidity (%RH) inside
the chamber was varied from 20% to 70%. The re-
sonant frequency was determined at 5% intervals us-
ing the time-domain gating method. The capacitance
of HCH-1000 was obtained from the resonant fre-
quency. The capacitance of HCH-1000 was also
measured directly with the impedance analyzer in
order to compare with the value obtained from the
time-domain gating method. Fig. 9 shows the ca-
10 15 20 25 30
10
-6
10
-5
10
-4
10
-3
10
-2
Separation Distance(cm)
S
i
g
n
a
l

i
n

L
o
g

S
c
a
l
e
(
V
)


1/R
6
Signal
3.3 3.35 3.4 3.45 3.5
0
0.5
1
10cm
3.3 3.35 3.4 3.45 3.5
-0.1
0
0.1
S
i
g
n
a
l
(
o
h
m
)15cm
3.3 3.35 3.4 3.45 3.5
-0.02
0
0.02
Frequency(MHz)
20cm
129

pacitances measured in both approaches are in good
agreement. According to the datasheet the capaci-
tance of HCH-1000 changes linearly with relative
humidity and the sensitivity varies from 0.55
pF/%RH to 0.65 pF/%RH. A linear-curve was fitted
to the capacitance data measured from the resonant
frequency of the sensor. It can be seen that the linear
curve fits well with the measured capacitance val-
ues. The slope of the curve was 0.596 pF/%RH
which was within the datasheet specifications. The
results indicate a deviation of <2% RH from the lin-
ear fit.















Figure 8. Setup of the plastic chamber, sensor and interrogator
coil for relative humidity sensing


























Figure 9. Capacitance of HCH-1000 measured directly with an
impedance analyzer and from the resonant frequency measured
by the RH sensor.

From the linear curve fitting, eqn. 14 and eqn. 15
were obtained, which relate the relative humidity
(%RH) to the capacitance, C
P
and resonant fre-
quency, f
0
, of the sensor, respectively.
508 68 . 1 % =
P
C RH . (14)
508
1680
%
2
0
=
f
RH . (15)
where units of C
P
is pF and f
0
is MHz.
5 DISCUSSION AND CONCLUSION
We demonstrated an interrogation technique for
coupled coil resonant sensors employing time-
domain gating. Results show that this technique pro-
vides extended interrogation distance/improved ac-
curacy in the determination of the resonant fre-
quency. For the test coil sensor, the resonant
frequency could be detected up to 20 cm and at this
distance the resonant peak position could be deter-
mined within 30 parts per million. This accuracy is
much higher than passive wireless sensors employ-
ing an impedance measurement technique (e.g., 6
cm interrogation distance with less than 1% meas-
urement error (Ong et al. 2008), 0.6 cm interrogation
distance with peak position within 7740 parts per
million (Butler et al. 2002), 8 cm interrogation dis-
tance with peak position within 1714 parts per mil-
lion (Ong et al. 2002)). The time-domain gating
method was used to determine the resonant fre-
quency of a relative humidity sensor. Results show a
linear relationship between the capacitance of the
sensor and the relative humidity with a 0.596
pF/%RH sensitivity. This sensor was able to detect
changes in %RH with less than 2%. This new inter-
rogation technique can be very useful where an in-
creased interrogation distance is required for civil
structural health monitoring.
6 REFRENCES
Andringa, M.M., Neikrik, D.P., Dickerson, N.P. & Wood, S.L.
2005. Unpowered wireless corrosion sensor for steel rein-
forced concrete. IEEE Sensors 2005: 155-158.
Butler, J.C., Vogliotti, A.J., Verdi, F.W. & Walsh, S.M. 2002.
Wireless, passive, resonant-circuit, inductively coupled, in-
ductive strain sensor. Sensors and Actuators A, 102(1): 61-
66.
Hapster, T.J., Stark, B. & Najafi, K. 2002. A passive wireless
integrated humidity sensor. Sensors and Actuators A, 95(2-
3): 100-107.
Nowak, M., Delome, N., Conseil, F. & Jacquemod, G. 2006. A
novel architecture for remote interrogation of wireless bat-
tery free capacitive sensors. Presented at the 13
th
IEEE In-
ternational Conference on Electronics, Circuits and Sys-
tems.
Ong, J.B., You, Z., Mills-Beale, J., Tan, E.L., Pereles, B.D. &
Ong, K.G. 2008. A wireless, passive embedded sensor for
20 30 40 50 60 70
310
315
320
325
330
335
340
345
Relative Humidity(%)
C
a
p
a
c
i
t
a
n
c
e

(
p
F
)


Direct measurement
Time-domain gating
Linear fit
1.5cm 3cm
19.5cm
Plastic chamber
Interrogator coil
Sensor
Dry air
Moist foam
130

real time monitoring of water content in civil engineering
materials. IEEE Sensors Journal, 8(12): 2053-2058.
Ong, K.G., Puckett, L.G., Sharma, B.V., Loiselle, M., Grimes,
C.A. & Bachas, L.G. 2002. Wireless, passive resonant-
circuit for monitoring food quality. Proceedings of SPIE,
4575: 150-159.
Robinson, M.P. & Clegg, J. 2005. Improved determination of
Q-factor and resonant frequency by a quadratic curve fitting
method. IEEE Transaction on Electromagnetic Compatibil-
ity, 47(2): 399-402.
Tesche, F.M., Ianoz, M.V., & Torbjrn 1997. EMC analysis
methods and computational models. New York: John Wiley
& Sons, Inc.

131