Sensors and Actuators A 314 (2020) 112348

Contents lists available at ScienceDirect

Sensors and Actuators A: Physical

journal homepage: www.elsevier.com/locate/sna

Design method for a GMR-based eddy current sensor with optimal

sensitivity
N.O. Romero-Arismendi ∗ , J.A. Pérez-Benítez, E. Ramírez-Pacheco, J.H. Espina-Hernández 1

Laboratorio de Evaluación No Destructiva Electromagnética (LENDE), ESIME-SEPI, Instituto Politécnico Nacional, UPALM Edif. Z-4, Zacatenco, México DF
07738, Mexico

a r t i c l e i n f o a b s t r a c t

Article history: Giant magnetoresistance (GMR) sensors have been used in non-destructive testing based on eddy cur-
Received 28 May 2020 rent (NDT-EC). The low-pass filtered GMR-based NDT-EC have been successfully used to estimate the
Received in revised form 17 August 2020 dimensions of artificial near-side cracks in aluminum plates. In this paper, an experimental study has
Accepted 17 September 2020
been carried out in order to optimize the characteristics of the probe to improve its sensitivity. It is
Available online 28 September 2020
proposed a new design method for the low-pass filtered GMR-based eddy current probe. Experimental
measurements were performed along artificial notches with different widths and depths in three alu-
Keywords:
minum plates. The asymmetry of the probe is analyzed in detail, giving that the operating point of the
Non-destructive testing
Probe based-eddy current testing
probe depends strongly on the asymmetry between the GMR end the Coil axis. The influence of the exci-
GMR Sensor tation coil parameters on the average sensitivity of the GMR-based EC probe is established. The outcomes
Magnetics sensors revealed the strong dependence of the probe sensitivity to the inner diameter and filling factor of the
excitation coil. Therefore, the methodology for selecting the inner diameter and the filling factor of the
excitation coil and the GMR-coil asymmetry value are proposed. The average sensitivity of the designed
probe using the proposed methodology is twice as high as the sensitivity of a non-optimized probe.
© 2020 Elsevier B.V. All rights reserved.

1. Introduction variation of the Gradient coils [5], it is possible to detect defects

with depths of 3 mm, 6 mm and 9 mm and classifying the orienta-
Non-destructive testing is a set of methods used to evaluate tion angle of the defects in the range of 20–170◦ . Despite of the fact
the properties and characteristics of materials with the objective that using classic EC probes it is possible to detect near-side defects
to avoid potential failures of parts. The eddy current-based Non- in aluminum plates, the EC sensor coils present limitations in the
destructive testing (NDT-EC) has been one of the most successful excitation frequency, the sensitivity, and the depth of penetration
methods to detect and characterize near-side cracks in paramag- of the eddy currents [6]. Therefore, the use of magnetic sensors
netic metals. It has been shown that the width and depth of the such as giant magnetoresistance (GMR), Hall sensor and Supercon-
near-side cracks and notches in aluminum plates can be estimated ducting Quantum Interference Devices (SQUID) in NDT-EC has been
using NDT-EC techniques [1–5]. The classic EC probe is composed increased with the intention of overcoming the limitations of coils
of coils used for both excitation and detection. Different coils con- sensors. The Hall sensor is the most widely used sensor to measure
figurations have been proposed to improve the estimation of the magnetics fields [7]. Hall sensors are inexpensive and are able to
width, depth and orientation angle of near-side cracks in aluminum measure the magnetic field magnitude [6]. Nevertheless, the sen-
plates. Currently, in order to improve the performance of classic EC sitivity of the Hall sensor is lower with respect to the GMR and coil
probes it has been studied the shape of the sensor coil. Peng et al. sensors. Garcí-a-Martin et al. demonstrated that Hall-based probes
in [2,5] proposed a gradient coils configuration. The gradient coils give lower sensitivity than a coil-based one [8]. The studies of the
configuration consists of a circular arrangement of three coils with use of SQUID sensors on Non-Destructive testing are scarced, the
twelve turns and a 10◦ between them [2]. Using the impedance most representative novel SQUID-based probe was proposed by
Tsuknamoto et al. [9]. The SQUID-based probe is able to detect artifi-
cial cracks with length of 50 mm and width 0.3 mm. Besides that the
∗ Corresponding author. probe requires for its operation a big mechanical structure, SQUID
E-mail address: [email protected] (N.O. Romero-Arismendi). sensor is expensive. The basis for the use of GMR sensors have been
1
Present address: scan Messtechnik GmbH, Vienna, Austria..

https://doi.org/10.1016/j.sna.2020.112348
0924-4247/© 2020 Elsevier B.V. All rights reserved.
2 N.O. Romero-Arismendi, J.A. Pérez-Benítez, E. Ramírez-Pacheco et al. / Sensors and Actuators A 314 (2020) 112348

proposed by Yashan et al. [10]. It has been demonstrated in [10]

that using a GMR-Sensor, the sensitivity is independent of the fre-
quency of the excitation magnetic field. Yasman et al. [10] showed
that, using the impedance of GMR-Sensors, it is possible to improve
the performance of the measurement than using the impedance of a
coil. GMR sensors have small dimensions, high sensitivity and they
are inexpensive [6,11]. Since then, the use of GMR sensors in the
NDT area has triggered the development of highly sensitive probes
[11–17]. Authors such as Postolache et al. [14], Gao et al. [16] and
Hamia et al. [15] demonstrated that using phase information of
the time-varying output voltage of the GMR Sensors it is possible
characterize near-side cracks. However, it is required the imple-
mentation of complex conditioning circuits as Lock-in amplifiers.
In [14] and [15] uses GMR sensor arrays in order to improve the
characterization of cracks in aluminum plates. In 2000 Dogaru and
Smith [11] developed a probe with the GMR sensor together with
an excitation coil in a pancake configuration where both devices
were centered to each other. The conditioning circuit in Dogaru
and Smith proposal included a low-pass filter, helping to remove Fig. 1. Block diagram of the experimental EC-system.
the high-ordered AC components of the signal and taking only a
proportional DC level at the GMR output. This configuration has
with the same filling factor and different inner diameters. Using
several advantages over classic EC probes, SQUID-based probes and
these last two coils, a study of the playing role of the inner diame-
Hall-based probes. By using a proportional DC level, the condition-
ter on the sensitivity of the probe is performed. Different operating
ing circuit is simpler than that used in GMR-based probes that use
points for each probe are compared and the operating point of the
phase information.
probe is defined considering a new proposal by using the sensitiv-
Using the GMR-based configuration proposed in [11], E. Ramírez
ity function of the GMR sensor. The outcomes allowed to propose a
et al. [1] developed an experimental EC system for near-side cracks
methodology for increasing the sensitivity of the GMR-Coil probe.
characterization in aluminum plates by slightly displacing the GMR
The methodology is based on establishing the optimal range of val-
sensor with respect to the center of the excitation coil. The used
ues from which to choose the inner diameter, filling factor and S/A
coil had a filling factor of 0.53 and an inner diameter of 1.66 mm.
value. The paper is organized as follows: Second section describes
It was found that this symmetric/asymmetric (S/A) configuration
the system used for the detection of defects in aluminum plates,
alters the performance of the probe proposed by Dogaru and Smith
also it describes the Symmetry–Asymmetry characterization of the
[11]. In [18] Ramirez et al. revealed that fixing an operating point
GMR-based EC probe. In Section 3 are described the coils used for
by the slight displacement of the GMR sensor with respect to the
the improvement of the average sensitivity of the EC system. Sec-
symmetry axis of the excitation coil varies the probe sensitivity.
tion 4 describes the experimental measurements. In Section 5.1
Espina-Hernandez et al. [19] proposed that the operating point
the operating point of the system and its influence in the crack
should be located on the linear region of the transfer character-
characterization were analyzed. In this section a condition for the
istic of the GMR sensor in order to increase the probe sensitivity.
operating point localization on the linear region of the sensitivity
In [19] an analytical correlation of two parameters (DV and DX)
function of the GMR sensor is proposed. In Section 5.2 the influence
extracted from the GMR voltage output and the cracks parame-
of the inner diameter and the filling factor of the excitation coil in
ters is given. It is demonstrated that there exist a unique (DV, DX)
the GMR-based EC probe sensitivity was analyzed. In Section 5.3,
pair for cracks with a particular width, depth, and orientation [20].
the proposed designed method is presented in detail.
In [20] a method to classify only two cracks characteristics using
the extracted parameters (DV, DX) were proposed. It was shown,
by analyzing the obtained correlations, that there is some influ- 2. Experimental setup
ence of the filling factor (f ) of the excitation coil on the sensitivity
of the system but not further analysis was presented in order to The block diagram for the EC measurement system used in the
define precisely its influence. In order to obtain the highest sen- experiments is shown in Fig. 1. The system is composed of a signal
sitivity, the authors in [19] proposed that the filling factor must generator Agilent 33210a, that supplies a sinusoidal voltage sig-
follow the condition f ≤ 0.5. This proposed condition imposes a nal with a fixed amplitude and frequency of 1.4 Vpp and 10 kHz,
criterion to select the best inner diameter of the excitation coil. respectively. In order to supply the current to the excitation coil of
However, the variation in the probe sensitivity caused by different the GMR-based EC probe, a transconductance amplifier was used as
inner diameters for a single filling factor was not taken into account. current source, following the idea presented in [21]. The transcon-
Nevertheless, the use GMR sensors for NDE applications are still in ductance amplifier has a G = 0.79 A/V and provides a maximum
development, in particular it has been not performed a full char- rms current value of 800 mA, the operating frequency range is from
acterization of the probe sensitivity with respect to these design 20 Hz to 20 kHz. In [18] it is shown that the use of a transconduc-
parameters and a deeper study of the best range of the operating tance amplifier allows a much better stability of the current than
region is required. Therefore, in the present works an analysis of the using the dual audio amplifier of the setup in [1,19]. In order to
influence of the design parameters in the sensitivity of the GMR- analyze the stability of the current supplied by the transconduc-
based probe is presented. The filling factors of the excitation coils tance amplifier, the excitation coil was fed during 25 minutes with
were chosen considering the condition proposed in [19]. The filling the maximum nominal current of the amplifier. The mean value of
factors were selected as follows, one with f < 0.5, a second one with the current supplied by the transconductance amplifier presented
f = 0.53 as the one used in [18,19], and the third value with f > 0.5. a variation below 0.1%. The eddy current probe contains a flat coil,
The experimental results support the established condition for the which provides a time-varying magnetic field, and a GMR sensor
filling factor proposed in [19]. Two coils were built with the same in pancake configuration as it was used previously [1,18,19]. The
inner diameter but different filling factor and two coils were built hand-made circular flat coils have 2 layers of 10 turns each. A GMR
 N.O. Romero-Arismendi, J.A. Pérez-Benítez, E. Ramírez-Pacheco et al. / Sensors and Actuators A 314 (2020) 112348 3

Fig. 4. Coil set used to analyze the sensitivity due to variation of the filling factor
and the inner diameter.

Table 1
Fig. 2. Representation of the Coil-GMR asymmetry value (ı) in the GMR-based EC Physical characteristics of the used Fx-Dy coil set, where “x” is the decimal value of
probe. the filling factor and “y” the approximate value in millimeters of the inner diameter.

Coil Nominal filling Nominal inner Nominal outer

factor diameter (mm) diameter (mm)

F7-D1 0.7 1.1 6.2

F5-D2 0.54 2.2 7.3
F5-D4 0.54 3.9 14.6
F4-D4 0.39 3.9 9

Fig. 3. System used to adjust the GMR-based EC probe asymmetry with the GMR
holder and the decentering screw.

sensor of the type AA-0002 from Non-Volatile Electronics was used

as it was proposed in [1]. The input–output characteristic curve
of the sensor was measured for a DC voltage of 5.15 V. Once the
input-output characteristic curve was experimentally obtained, it
was derived numerically to obtain the sensitivity function of the
GMR sensor.
The GMR-Coil S/A value (ı) is defined as the position difference
between the vertical symmetry axes of the coil and the GMR sensor.
Fig. 2 shows the cross-section of the probe. In Fig. 2 the symmetry
axes of the coil and the sensor are shown, as well as the GMR-Coil
asymmetry (ı). In order to set the ı value of the analyzed GMR-
based EC probe, a custom-made holder with an asymmetry setting Fig. 5. Tangential magnetic field inside the excitation coil for different distances
system was developed. The asymmetry setting system shown in between the GMR-based EC probe and the near-side cracks.
Fig. 3 is composed of a millimeter screw and a GMR holder. Eq. (1)
was used to adjust the Coil-GMR asymmetry (ı) value:
coils. In Table 1 the main physical characteristics of the coils are
1 shown.
ı= ∗˛ (1) The set of the studied cracks is composed of three aluminum
360◦
plates with different crack width (w) of 1.3, 1.0 and 0.6 mm. Each
where ˛ is the rotation angle of the screw expressed in degrees. studied plate has 4 defects with depths (d) of 4, 2, 1.5 and 0.5 mm.
Three ı values were set ı0 = 0 mm, ı1 = 0.5 mm and ı2 = 0.75 mm.
The signal acquisition system is composed of an adjustable gain
4. Methods
instrumentation amplifier and a low-pass filter. The characteris-
tics of these elements are explained in detail in [19]. A custom
Ten measurements were performed for each crack belonging to
LabVIEW program registers the filtered value of the GMR output
the crack set using different probes configuration. The measure-
voltage using a data acquisition card (DAQ-PCI610).
ments were performed along the x axis. Fig. 5 shows the tangential
magnetic field inside the excitation coil as it approaches a near-
3. Materials side crack. It can be observed in Fig. 5 that the tangential magnetic
field inside the coil is distorted as the probe gets closer to the
For the present study four probes were used, and different coils near-side crack. Using the measurement system of Fig. 1, a pro-
for each probe were designed and built. These coils allow us to portional DC level at the output of the GMR sensor is obtained,
analyze the system response to variations in the inner diameter which corresponds to the distorted tangential magnetic field of the
and the filling factor of the excitation coil. The F5-D2 coil, with inspected defect. The resulting output voltage during the surface
f = 0.5 and similar physical characteristics as the one in [18,19], scan gives a profile as shown in Fig. 5. From this profile, the DV and
was taken as reference. In the present work, filling factors from 0.4 DX parameters are extracted [19]. The DV parameter is defined as
to 0.7 were selected. The coils F5-D2 and F5-D4 were built with the the difference between the highest and lowest peaks, and the DX
same filling factor but different inner diameter in order to analyze parameter is defined as the difference in position of the occurrence
this last parameter independently of the filling factor. It is worth of those peaks (Fig. 6).
noting that the filling factor includes the inner diameter. Therefore, Probes F7-D1, F5-D4 and F4-D4 were setted with ı0 , ı1 and
a separate analysis must be performed. Fig. 4 shows a picture of the ı2 values. The results obtained from the probes F7-D1, F5-D2 and
4 N.O. Romero-Arismendi, J.A. Pérez-Benítez, E. Ramírez-Pacheco et al. / Sensors and Actuators A 314 (2020) 112348

characteristic of the GMR sensor as proposed in [19]. Figs. 7b, 8 b

and 9 b show the corresponding operating points located in the
sensitivity function of the GMR sensor.
Figs. 7b, 8 b and 9 b show that the operating points for the sym-
metry configuration case (ı0 ) are at the lowest sensitivity values.
The corresponding operating points of the F5-D2 and F4-D4 probes
for the conditions ı1 , and ı2 are in the linear region of the sensitivity
function of the GMR sensor. Next, the analysis of the importance of
the operating point and its relation to the sensitivity curve is given.
Fig. 10 presents the DV parameter as a function of the crack
depth for the three considered ı values using the probes F7-D1,
F5-D2 and F4-D4 with their respective standard deviations. It is
important to emphasize that the obtained standard deviations of
the developed measurements, which are lower than 2 ␮V, guar-
antee the repeatability of the obtained results. It is observed in
Fig. 10 that the magnetic field variations due to the surface cracks
are smaller for the symmetric configuration. In case of the probes
F5-F2 and F7-D1 the sensor is not able to measure the magnetic
field variations corresponding to the variations of the trajectory of
Fig. 6. GMR-voltage output profile obtained from the inspection using the low-pass the eddy current. Therefore, the probe with the symmetric configu-
filtered GMR-probe. ration is not sensitive enough to detect the smaller depth of defects
and it should be not used in the GMR-Coil Probe.
F4-D4 were used to analyze the influence of the asymmetric config- Fig. 11 shows the DV values for the defects with nominal widths
uration and the filling factor on the GMR-based EC probe sensitivity. w1 = 0.6 mm, w2 = 1 mm and w3 = 1.3 mm using the F7-D2 probe
Besides, the results obtained from probes F5-D4, F5-D2 and F4-D4 for the asymmetry ı2 . It could be deduced from Fig. 7 that even
helped to analyze the influence of the inner diameter variations on though the operating point of the F7-D2 probe in this case is in the
the probe sensitivity. linear region of the GMR characteristic function, it is outside the
linear region of the GMR sensitivity function. For this configuration,
in which the operating point is not in the linear region of the GMR
5. Results and discussion
sensitivity function, (the circle in Fig. 11) there is an overlap for
the responses for the w = 0.6 mm and w = 1.0 mm cracks families.
5.1. Influence of the symmetry–asymmetry in the GMR-based EC
Therefore, this is a clear experimental evidence that the operating
probe sensitivity
point should be set in the linear region of the sensitivity function
instead of the transfer characteristic of the sensor as proposed in
In order to estimate the region for the operating point that gives
[19].
the highest sensitivity, the probes F4-D4, F5-D2 and F7-D1 were
On the other hand, in order to study the influence of the inner
selected. The DV/d factor was analyzed for the conditions ı0 , ı1 ,
diameter parameter on the operating point, the spatial distribu-
and ı2 . Figs. 7a, 8 a and 9 a show the operating points of the probes,
tion of eddy current as a function of the inner diameter is analyzed.
located in the transfer characteristic of the GMR sensor. Figs. 7a,
A simulation of the induced eddy currents was performed using
8 a and 9 a reveal that all the corresponding operating points for
the simulation program FEMM. Fig. 12a presents the results of the
the condition ı0 is located in the minimum value of the transfer
simulation of the eddy current density induced by three coils with
characteristic of the GMR sensor. It can be noticed that, only the
inner diameters of 1.09 mm. 2.1 mm and 3.9 mm. The outcomes
corresponding operating points of the F7-D1 probe for the con-
in Fig. 12a shows that the spatial distribution of the eddy currents
ditions ı1 , and ı2 are located in the linear region of the transfer

Fig. 7. Operating points of the GMR sensor using (a) the transfer characteristic of the GMR sensor and (b) the sensitivity function of the GMR sensor using probe F7-D1. The
corresponding linear regions are shown in green.
 N.O. Romero-Arismendi, J.A. Pérez-Benítez, E. Ramírez-Pacheco et al. / Sensors and Actuators A 314 (2020) 112348 5

Fig. 8. Operating points of the GMR sensor using (a) the transfer characteristic of the GMR sensor and (b) the sensitivity function the GMR sensor using probe F5-D2. The
corresponding linear regions are shown in green.

Fig. 9. Operating points of the GMR sensor using (a) the transfer characteristic of the GMR sensor and (b) the sensitivity function of the GMR sensor using probe F4-D4. The
corresponding linear regions are shown in green.

depends on the inner diameter. Therefore, it is important to con- to select a ı value which maintains the operating point within the
sider the inner diameter to improve the probe sensitivity. Fig. 12b established high sensitivity operating region.
presents the normalized induced current. It can be observed that
the spatial distribution of the eddy current is similar for the tree 0.1 < ı/Rin < 0.5 (2)
coils. The ın normalized value (ın = ı/Rin ) is proposed in order to
stablish an optimized ı value. The advantage of using a ın is that it 5.2. Influence of the dimensions of the excitation coil in the
considers the influence of the inner diameter. GMR-based EC probe sensitivity
Fig. 13 shows the ın for each asymmetric configuration of the F7-
D1, F5-D2 and F4-D4 probes located in the GMR sensitivity function. In order to describe the influence of the excitation coil in the
It can be observed in Fig. 13 that those ın values smaller than 0.5 are sensitivity of the GMR-based EC probe, the average sensitivity is
outside the linear region of the GMR sensitivity function. Therefore, considered, and it is defined as the increase rate of the DV parameter
to ensure that the operating point is within the established linear proportional to d. Fig. 14 shows the relation of the DV parameter to
region of the GMR sensitivity function, the ın should be smaller the crack depth for the w = 0.6 mm cracks family using the probes
than 0.5. The outcomes reveal that the GMR sensor could go to F4-D4, F5-D2 and F7-D1 with asymmetry ı1 and their respective
saturation by fixing the ın > 0.5. The ın should be also higher than average sensitivity values, correlation coefficient (R) and the resid-
0.1 to avoid the lowest sensitivity points of the GMR sensor. The ual standard error (RSMD). The average sensitivity (m) in Fig. 14
restrictions that is previously shown allow proposing the Eq. (2) depends on the filling factor and this dependence is observed for
all studied crack’s width families. The results show that the average
6 N.O. Romero-Arismendi, J.A. Pérez-Benítez, E. Ramírez-Pacheco et al. / Sensors and Actuators A 314 (2020) 112348

Fig. 10. Relation of the DV parameter to crack depth for width cracks w = 1.0 mm using the probes: (a) F7-D1, (b) F5-D2, (c) F4-D4.

sensitivity of the F4-D4 probe is approximately twice the sensitivity the obtained outcomes are similar for the complete set of studied
of the F5-D2 probe. According to these results, a coil with a lower widths. So far, the improvement of the sensitivity of the eddy cur-
filling factor should be used in the GMR-based EC probe in order to rent probe due to the GMR-Coil asymmetry, the inner diameter and
get a higher sensitivity. the filling factor has been analyzed. However, some considerations
Probes F5-D4 and F4-D4, which have coils with the same inner must be taken for the design of the coil to guarantee the maximum
diameter but with different filling factors, were used to analyze the average sensitivity.
influence of the filling factor on the average sensitivity. The ı value
was adjusted so that the operating point was set approximate 65 5.3. Design method for the optimization of the GMR-based EC
mV/mT in each probe. Fig. 15 shows the respective average sensi- probe
tivities for the probes F5-D4 and F4-D4. It can be seen in Fig. 15 that
despite of having the same inner diameter and operating point, the The probe is considered small, therefore an outer diameter
sensitivity of the probe F4-D4 is higher than that of the probe F5-D4. Dout ≤ 10 mm is proposed, and two layers maximum with NL turns
Therefore, this outcome demonstrates that the filling factor must per layer. The filling factor (f ) of the excitation coil is defined as
be smaller than 0.5 in order to have a higher sensitivity. Probes F5- [19]:
D4 and F5-D2, which are composed of coils with the same filling
factor, were used to observe the influence of the inner diameter on Dout − Din
f = (3)
the average sensitivity. Fig. 16 shows the DV parameter as a func- Dout + Din
tion of crack depth and their respective average sensitivities for the
probes F5-D4 and F5-D2. The average sensitivity is the same for all where Din is the inner diameter of the coil expressed in mm. Con-
studied crack’s width families. sidering the geometry of the flat excitation coil used here, Dout
It can be seen in Fig. 16 that the sensitivity of the probe with is:
bigger inner diameter coil is higher. According to these results the
Dout = Din + 2 · NL · C (4)
inner diameter of the coil should be as large as possible taking into
account the holder size restriction. It is important to emphasize that where C is the wire gauge in mm.
 N.O. Romero-Arismendi, J.A. Pérez-Benítez, E. Ramírez-Pacheco et al. / Sensors and Actuators A 314 (2020) 112348 7

Fig. 11. Relation of the DV parameter to crack depth for the complete crack set using Fig. 13. Location of the operating points corresponding to ın relations on the sen-
the F7-D1 probes with ı2 . sitivity function of the GMR sensor for all symmetric-asymmetric configurations of
the studied probes.

The filling factor f of a flat coil indicates the filliness of it, then For the limiting case f = 1/2 imposed by expression (6) and
1 ≤ f ≤ 0. Previous experimental results [18,19] demonstrated that using Eq. (7), Din should fulfill:
the parameter DV, follows a relationship of the type:
Din ≥ NL · C (8)
DV = k(1 − f d ) (5)
Expression (8) allows selecting a combination of the number of
where k is a proportionality constant depending on the defect’s turns and the wire gauge for the construction of the excitation coil,
width (wn ) and d is the defect’s depth. It was demonstrated previ- keeping in mind the maximum working current. It is important to
ously that the filling factor should be small in order to increase the note that the inner diameter of the coil should be as large as possible
sensitivity. Therefore, the restriction imposed by Eq. (6) is proposed taking into account the holder size restriction. Considering that the
in order to have a good sensitivity for the detection of defects using excitation coil has two layers at most and with each layer having
the GMR-based EC probe. a number of turns less than 12, the wire gauge should be selected
among those which can stand the higher maximum current. It is
f ≤ 1/2 (6) worth noting that the magnetic field intensity produced by a coil
depends on the number of turns and the excitation current.
Taking into consideration Eqs. (3) and (4), one can expressed Din Besides of the physical characteristics of the excitation coil, it
as: is important to select an appropriate asymmetry value in order
to increase the sensitivity of the probe. It was shown previously
1−f that the operating point must be located in the linear region of the
Din = NL · C · (7)
f GMR sensitivity function. Therefore, the ı value should be selected

Fig. 12. (a) Induced eddy current density in an aluminum plate by three coils with inner diameters of 1.09 mm, 2.1 mm 3.9 mm. (b) Normalized induced eddy current density
in an aluminum plate by three coils with inner diameters of 1.09 mm, 2.1 mm 3.9 mm.
8 N.O. Romero-Arismendi, J.A. Pérez-Benítez, E. Ramírez-Pacheco et al. / Sensors and Actuators A 314 (2020) 112348

Fig. 14. Relation of the DV parameter to crack depth for the w = 0.6 mm cracks Fig. 16. Relation of the DV parameter as a function of crack depth for width family
family using F4-D4, F5-D2 and F7-D1 probes with ı = 0.5 mm. w = 1.00 mm and their respective average sensitivities (m) for probes with the same
filling factor and different inner diameter.

should be located in the linear region of the sensitivity function of

the GMR sensor, not in the transfer characteristic of the GMR sen-
sor as in previous studies [19]. It has been shown that the inner
diameter and the filling factor are the two main characteristics of
the excitation coil for the improvement of the sensitivity of the
asymmetric GMR-based EC probe. It has been demonstrated that
the average sensitivity increases as the filling factor decreases. In
order to improve the GMR-based EC probe sensitivity, the excita-
tion coil should have the maximum possible inner diameter for a
selected filling factor.
The design parameters for a higher sensitivity asymmetric low-
filtered GMR-based EC probe has been proposed in this paper.
The filling factor of the excitation coil for a high sensitivity GMR-
based EC probe should be less than 0.5. The inner diameter of the
excitation coil must meet that Din ≥ NL · C. The Coil-GMR asymme-
try should be selected among those which fulfill 0 < ı/Rin < 0.5.
Using these proposed parameters, the probe sensitivity increases.
However, the probe response should be analyzed for defects not
Fig. 15. Relation of the DV parameter as a function of crack depth for width family
belonging to the studied set.
w = 1.00 mm and their respective average sensitivities (m) for probes with different
filling factor and same inner diameter.

among the restriction proposed in Eq. (1) which assure that the Credit author statement
corresponding operating point is located in the proposed operating
region. The ı value must be chosen in such a way that the operating N.O. Romero-Arismendi: Conceptualization, Methodology,
point is located at the highest sensitivity value where the dynamic Validation, Formal analysis, Investigation, Data curation, Writing
range of the signal is maintained within the linear region. – Original Draft.
J.A. Pérez-Benítez: Conceptualization, Validation, Recourses,
supervision, Funding acquisition, Project administration, Writing
6. Conclusions
– Review & Editing.
E. Ramírez-Pacheco: Methodology, Conceptualization.
In this paper has been shown that using a symmetric configu-
J.H. Espina-Hernández: Conceptualization, Methodology, Val-
ration the GMR-based EC probe has a poor average sensitivity. The
idation, Formal analysis, Recourses, supervision, Funding acquisi-
crack set could not be measured with probes F7-D1 and F5-D2 using
tion, Project administration, Writing – Review & Editing.
a symmetric configuration. Nevertheless, using an asymmetric con-
figuration all studied cracks have been detected by probes F7-D1
and F5-D2. Therefore, to design a higher sensitivity GMR-based EC
probe, an asymmetric configuration must be implemented. It has Declaration of interests
been shown in this paper that when the operating point is outside
the linear region of the sensitivity function of the GMR sensor, there The authors declare that they have no known competing finan-
is an overlap between crack’s width families responses of w = 1.0 cial interests or personal relationships that could have appeared to
mm and 1.3 mm. Accordingly, It is shown that the operating point influence the work reported in this paper.
 N.O. Romero-Arismendi, J.A. Pérez-Benítez, E. Ramírez-Pacheco et al. / Sensors and Actuators A 314 (2020) 112348 9

Declaration of Competing Interest [19] J.H. Espina-Hernández, E. Ramírez-Pacheco, F. Caleyo, J.A. Pérez-Benitez, J.M.
Hallen, Rapid estimation of artificial near-side crack dimensions in
aluminium using a GMR-based eddy current sensor, NDT E Int. 51 (2012)
The authors report no declarations of interest. 94–100, http://dx.doi.org/10.1016/j.ndteint.2012.06.009.
[20] N.O.R. Arismendi, E.R. Pacheco, O.P. Lopez, J.H. Espina-Hernandez, J.A.P.
References Benitez, Classification of artificial near-side cracks in aluminium plates using
a GMR-based eddy current probe, 2018 28th International Conference on
Electronics, Communications and Computers, CONIELECOMP 2018, 2018,
[1] E. Ramírez-Pacheco, J.H. Espina-Hernández, F. Caleyo, J.M. Hallen, Defect January (2018) 31–36, http://dx.doi.org/10.1109/CONIELECOMP.2018.
detection in aluminium with an eddy currents sensor, Proceedings – 2010 8327172.
IEEE Electronics, Robotics and Automotive Mechanics Conference, CERMA [21] O.B. Laug, A wide-band transconductance amplifier for current calibrations,
2010 (2010) 765–770, http://dx.doi.org/10.1109/CERMA.2010.91. IEEE Trans. Instrum. Meas. IM-34 (4) (1985) 639–643, http://dx.doi.org/10.
[2] X. Peng, H. Jun, A new eddy current sensor composed of three circumferential 1109/TIM.1985.4315425.
gradient winding coils, Proceedings of the International Conference on
Sensing Technology, ICST (2013) 912–915, http://dx.doi.org/10.1109/ICSensT.
2013.6727782. Biographies
[3] P. Xu, S. Huang, W. Zhao, A new differential eddy current testing sensor used
for detecting crack extension direction, NDT E Int. 44 (4) (2011) 339–343,
http://dx.doi.org/10.1016/j.ndteint.2011.01.010.
[4] M.X.J. Wrzuszczak, Eddy current flaw detection with neural network Nestor Orlando Romero Arismendi received the degrees
applications, TC7 Symposium on New Developments in the Field of in Electronic Engineering and Electrical Engineering from
Measurement Science 2002 (38) (2002) 161–164, http://dx.doi.org/10.1016/j. the Universidad Nacional de Colombia, Colombia in 2014
measurement.2005.04.004. and 2016 respectively. He received the Master in Electron-
[5] X. Peng, H. Jun, A new eddy current sensor composed of three circumferential ics from Instituto Politécnico Nacional, México, in 2018.
gradient winding coils, Proceedings of the International Conference on He is a PhD student in Communication and Electronics
Sensing Technology, ICST (2013) 912–915, http://dx.doi.org/10.1109/ICSensT. in the same institute. His current research interests are
2013.6727782. related to simulation and application of eddy currents in
[6] J. García-Martín, J. Gómez-Gil, E. Vázquez-Sánchez, Non-destructive Non-Destructive Evaluation of materials.
techniques based on eddy current testing, Sensors 11 (3) (2011) 2525–2565,
http://dx.doi.org/10.3390/s110302525.
[7] N. Novkovsky, Progress and limitations in magnetic field measurements, in: J.
Rasson, T. Delipetrov (Eds.), Geomagnetics for Aeronautical Safety, first ed.,
2005, pp. 201–212. Jose Alberto Benitez Perez graduated in Physics from
[8] J. Garcia-Martin, J. Gomez-Gil, Comparative evaluation of coil and hall probes University of Oriente, Cuba (1997), Master in Physics,
in hole detection and thickness measurement on aluminum plates using eddy University of La Habana, Cuba (2004); PhD in Physics,
current testing, Russ. J. Nondestruct. Test. 49 (8) (2013) 482–491, http://dx. University of Oriente in collaboration with University of
doi.org/10.1134/S1061830913080044. São Paulo, Brazil (2007). Postdoctoral researcher at the
[9] A. Tsukamoto, T. Hato, S. Adachi, Y. Oshikubo, K. Tsukada, K. Tanabe, Mechanics Faculty in the University of São Paulo, Brazil
Development of eddy current testing system using HTS-SQUID on a hand cart (2009), Visiting professor at University of Aveiro, Portugal
for detection of fatigue cracks of steel plate used in expressways, IEEE Trans. (2010). Currently, he is a professor and senior researcher
Appl. Supercond. 28 (4) (2018), http://dx.doi.org/10.1109/TASC.2018. at National Polytechnic Institute, Mexico and member of
2795614. the National System of Researchers. He has 20 years of
[10] A. Yashan, R. Becker, G. Dobmann, Use of GMR-sensors for eddy-current experience on the research about electromagnetic non-
testing, International Workshop on Electromagnetic Nondestructive destructive evaluation methods, and has published more
Evaluation, Budapest (2000) 187–193. than 35 papers in this research area.
[11] T. Dogaru, S.T. Smith, Giant magnetoresistance-based eddy-current sensor,
IEEE Trans. Magn. 37 (5) (2001) 3831–3838. Eduardo Ramirez Pacheco received the Engineering
[12] D. Pasadas, A.L. Ribeiro, H.G. Ramos, T. Rocha, 2D geometry characterization of degree in Communication and Electronics and the Master
cracks from ECT image analysis using planar coils and GMR-sensors, in Electronics from Instituto Politécnico Nacional, Méx-
Conference Record – IEEE Instrumentation and Measurement Technology ico, in 2002 and 2009, respectively. He received his PhD
Conference 2016, July (2016), http://dx.doi.org/10.1109/I2MTC.2016. in Communication and Electronics in the same institute.
7520426. His current research interests are related to eddy currents
[13] O. Postolache, A.L. Ribeiro, H. Ramos, A novel uniform eddy current probe applied to Non-Destructive Evaluation of materials.
with GMR for non destructive testing applications, EUROCON 2011 –
International Conference on Computer as a Tool – Joint with Conftele 2011
(2011) 1–4, http://dx.doi.org/10.1109/EUROCON.2011.5929410.
[14] O. Postolache, A.L. Ribeiro, H.G. Ramos, Uniform eddy current probe based on
GMR sensor array and image processing for NDT, 2012 IEEE I2MTC –
International Instrumentation and Measurement Technology Conference,
Proceedings (2012) 458–463, http://dx.doi.org/10.1109/I2MTC.2012.6229366. José Hiram Espina Hernández received the degrees in
[15] R. Hamia, C. Cordier, S. Saez, C. Dolabdjian, Eddy-current nondestructive Electronic Engineering and Master in Automation from
testing using an improved GMR magnetometer and a single wire as inducer: a Instituto Superior Politécnico José Antonio Echeverría
FEM performance analysis, IEEE Trans. Magn. 46 (10) (2010) 3731–3737, Havana Cuba, in 1993 and 1998, respectively. He received
http://dx.doi.org/10.1109/TMAG.2010.2052827. his PhD in Technical Sciences from the same University
[16] P. Gao, C. Wang, Y. Li, F. Li, Y. Yan, Y. Hu, Defect evaluation using the phase in 2005 after having performed the whole research at
information of an EC-GMR sensor, Conference Record – IEEE Instrumentation the Vienna University of Technology. His current research
and Measurement Technology Conference (2014) 25–29, http://dx.doi.org/10. interests are related to electromagnetic non-destructive
1109/I2MTC.2014.6860516. evaluation of materials and in general applied electromag-
[17] P. Gao, X. Wang, D. Han, Q. Zhang, Eddy current testing for weld defects with netism.
different directions of excitation field of rectangular coil, 2018 4th
International Conference on Control, Automation and Robotics (ICCAR) (2018)
486–491.
[18] E. Ramírez, J.H. Espina, J.A. Pérez, F. Caleyo, J.M. Hallen, Some particularities of
EC crack detection in aluminum using an asymmetrical GMR-coil
configuration, IEEE Lat. Am. Trans. 13 (5) (2015) 1331–1339, http://dx.doi.
org/10.1109/TLA.2015.7111986.