Improving Flaw Detection through Integration of a Novel Eddy Current Probe with Fluxgate Magnetic Sensor

The objective of this study is to improve the detection capability of eddy current testing by developing a novel eddy current probe. The developed probe is composed of six excitation coils made at different areas and suitably aligned with a single-axis sensitive DRV425EVM fluxgate sensor. The distribution of magnetic field intensity and eddy current density of the developed probe were calculated using ANSYS 2019R3 (Maxwell-3D) simulation software. The simulation results revealed that the new probe design enhances the detection capability and the spatial resolution compared to the conventional design that uses a double-D coil. Experiments were carried out using the double-D coil as reference, under the same input parameter to detect flaws with various sizes and locations. The experimental results validated the simulation results and showed that the sensitivity and spatial resolution of the developed sensor are much better than those obtained by the conventional probe.


INTRODUCTION
Eddy current testing (ECT) has become one of the widely preferred electromagnetic NDT methods that is used mainly in nuclear, aerospace, power, and petrochemical industries to examine metallic plates, sheets, tubes, pipes, rods, and bars for the detection and sizing of cracks, corrosion, and delamination of carbonbased composite layers during the production process, or in service [1,2]. Eddy current testing is applicable for both buried and surface-breaking flaws. Eddy current is an electric current that is induced in a conductor by changing the magnetic boundary field. It is usually generated by the flow of alternating electric current in a coil placed near an electrically conductive part. Due to the induced current on the part, a time-dependent secondary magnetic field generates an opposing magnetic field [3]. When the eddy current encounters an obstacle, such as crack, the surrounding current becomes distorted. Thus, the net magnetic field at the crack location increases.
To retrofit eddy current testing as a crack detection comparator, a probe with two alternating current ex-citing coils, commonly made in symmetrical double D shape, is placed on the surface of the workpiece, so that the two coils lie side-by-side. Either a magnetic sensor or a receiving coil is located right at the top center of the D-shaped coils. If no defect is found on both sides of the coil, the vectoral net magnetic fields cancel each other. However, when a defect is found on one side of these coils, a difference in the magnetic field arises. Eventually, the sensor receives these differences as an output signal.
The eddy current depth of penetration has a major role in enhancing the detection capability, especially for subsurface flaws. However, aside from excitation frequency, the depth mainly depends on the electrical conductivity and magnetic permeability of the workpiece material. The receiving sensors and configuration of the excitation coil are also paramount factors, which have been considered by researchers to empower the detection capability of different crack sizes at different distances below the surface [4,5].
Over the past few years, sensitive magnetic field sensors have been implemented with remarkable suc-cess in eddy current testing [6]. SQUID, Fluxgate, and Hall sensors are the most commonly applicable magnetic sensors for eddy current testing. SQUID magnetic sensors are widely accepted, because of their linearity, excellent sensitivity, and high dynamic range capability [7].
Due to the barely optimized factors that influence eddy current and the limited alternative commercial magnetic sensors that are available, relatively few studies have been carried out. Some of these studies have been conducted using a SQUID sensor with a double-D excitation coil, as in [5,6]. Bettaieb et al. have also attempted to couple the Hall sensor with a doublerectangular printed circuit board [8]. Principio and Kreutzbruck conducted experiments with a double-D excitation coil and fluxgate sensor [9,10].
However, the use of SQUID technology sensors involves high cost and complex handling, related to the cooling mechanism [6,10,11]. Furthermore, having the same number of turns in the double-D excitation coil disperses the net field towards and away from the receiving sensor, and thus substantially limits the spatial resolution and defect detectability.
The objective of this study is to improve the detection capability of surface and sub-surface flaws. The paper introduces an innovative design of a probe that assimilates the excitation coil and fluxgate sensor. The proposed design uses a DRV425EVM commercial fluxgate magnetic field sensor with single-axis sensitivity. This sensor has high linearity, high sensitivity, and high dynamic range, with a noise of 1.5 nT/√Hz. Unlike SQUID sensors, it works in a wide range of extended industrial temperature, -40 to +125°C, with offset drift ±5 nT/°C [12]. The DRV425EVM sensor is also ergonomically convenient to incorporate with excitation coil in a probe.
The proposed design enables the generation of high-intensity eddy current, and conveys the net magnetic field towards the sensor to amplify the net magnetic field difference between the adjacent symmetrical coils, and therefore improves the detectability of different flaw sizes and shapes.

FLUXGATE SENSOR (DRV425EVM)
The fluxgate sensor is a well-known type of commercial magnetic sensor that measures a magnetic field or a magnetic flux, and produces a voltage output out V that is proportional to the measured field [13]. The operation of the sensor is based on two solenoid coils: an excitation or a drive coil, and a pickup coil. An alternating current through the excitation coil cre-ates a magnetic field that induces a current in the pickup coil. In a magnetically neutral field, the input current to the excitation coil and the output current from the pick-up coil are identical and cancel each other. If there is a difference between them, a detectable output current is created, due to the presence of an ambient magnetic field.
The sensor measures the absolute strength of the surrounding magnetic field by using the change in the magnetic field. Following the SQUID sensor, the fluxgate sensor has the best capability to detect the magnetic field with a high-accuracy sensing range of ±2 mT and a measurement bandwidth of up to 47 kHz. It is less sensitive to temperature changes and can operate over the extended industrial temperature range of -40 to +125°C [12]. Figure 1 shows that the DRV425EVM fluxgate sensor type is designed for single-axis magnetic field-sensing applications and enables electrically-isolated, high-sensitivity, and precise dc-and ac-field measurements [12].

Working Principle
In eddy current testing, there are two magnetic fields present that are superposed. The first is generated when an alternating current (I) is applied to an excitation coil, and produces the primary magnetic field (Bprim), and the second is the reaction field that is associated with the induced eddy currents in the inspected specimen (B snd ) [14].
The presence of a flaw in metal specimens can be detected by measuring the change in B snd , as the sensor is moved from an unflawed region to one contain-  ing a flaw. The presence of a flaw is exhibited by a sudden peak output voltage, due to a relatively stronger magnetic field.
This eddy current working principle can be further extended to play the role of a comparator. Basically, two symmetrical horizontal coils are laid adjacent to each other in parallel to the work surface. When a flaw appears on one side of the coil, a magnetic field difference occurs. This field difference indicates a defect on the work plate. The most common type of magnetic field comparator coils is the "double-D" excitation coil (Fig. 2). It consists of two D-shaped windings with their straight sections carrying current in the same direction and parallel to each other. When the probe containing the excitation coil and sensor is passed through on a work plate with no flaw, the magnetic field due to the coils counterbalance each other and no signal is detected as in Fig. 2a. However, when only one of these coils encounters a flawed surface, the secondary magnetic field weakens. Consequently, the net magnetic field becomes stronger than the adjacent symmetrical D shaped coil, as shown in Fig. 2b.
Due to these phenomena, the fluxgate sensor that is located right above the center of these coils detects the field differences. Likewise, in this study, a new probe was designed to improve the detectability of defects. It also has the same working principle as the double-D coil. However, the new probe mainly plays a major role in the excitation coil design and its interaction with the receiving magnetic field sensor.
The excitation coils, when they are wound, should ideally be symmetrical; however, it is difficult to make two coils that are perfectly symmetric. Therefore, du-ring the winding of the coils, care was taken to keep the winding coil as symmetrical as best could be achieved. The remaining minor field imbalance could be adjusted by the sensor location adjustment. As mentioned in the introduction section, many researchers use the double-D type of excitation coil. However, due to the fact of having the same number of turns, it is less effective in improving spatial resolution and detection capability.
In this study, the design is realized by applying six coil winding areas in a probe. The parallel middle coils merge in one coil, in contrast to the double-D coils. This makes to have a greater number of turns than for the outer sides of the coils. This realization helps to broaden the detection range from the thin surface to the deep subsurface flaws.

Features and Configuration of the Designed Probe
The proposed probe mainly consists of two units: the excitation coil and magnetic field sensor. The symmetrical excitation coil is wound from two copper wires, each having 0.1 mm diameter (38 AWG). Figure 3 shows that it contains pairs of three coils, with a different number of turns, objectively oriented in three distinct cross-sectional areas of the probe (coil 1, coil 2, and coil 3). All these three coils are connected in series, and the direction of current in these coils is labelled with " " and " ".
Coil 1 is oriented parallel to the surface of the work plate. It has an inside size of 6 × 20 mm with 150 number of turns. Its shape helps to optimize the number of turns of the horizontal side of coils in the middle and on both ends. This also helps to increase the excitation current flow around the central probe, which ultimately produces the stronger secondary magnetic field. This process amplifies the field difference between the adjacent coils. Furthermore, discretely optimizing the number of turns of the horizontal coil enhances the spatial resolution of the neighboring flaw by controlling the secondary magnetic field towards the desired direction.
A 140 turns coil 2 creates a strong eddy current field towards the center, which helps to guide a valuable magnetic field on the way to the receiving sensor.
Coil 3 contains 60 turns. It helps to counteract part of the induced magnetic field, to not move away from the middle area where the receiving sensor is located, and to direct the magnetic field towards the comparison plane, as shown in Fig. 4. Figure 5 shows the probe that was produced using 3D printing machine from polylactic acid (PLA) fila- ment for fulfilling the dimensional requirement of the excitation coils and the receiving field sensor.

Finite Element Method Simulation
Regarding eddy current and related calculations, the complexity of the excitation coils necessitates the use of simulation modeling. Simulation of the con-  The copper coils of the proposed probe and an aluminum alloy (Al-6061) plate having dimensions of 120 × 120 × 15 mm that contains a 30 ×1 × 5 mm defect was modeled. The excitation coil was 0.6 mm far from the work plate to be inspected.
ANSYS Maxwell 3D contains a specially designed module to simulate eddy current distribution in a work plate due to the excitation coil. This module simulates the magnetic field distribution across the boundary areas, and computes the matrices of resistance, inductance, and impedance of excitation coils. It also simulates the defect response with the model-assigned sensor and calculates the average magnetic field distribution at a specified location.
In addition, the software plots the magnetic field distribution along the specified axis with predetermined height in the real and imaginary parts of alternate current magnetic response; however, the real component exhibits better intensity than the imaginary unit. Figure 6 shows the real and imaginary B field distribution plots in both double-D and the proposed probe at the sensor location along the z axis. In this study, to simulate the magnetic field and eddy current distribution throughout the boundary area, including probe and work plate; a variable with an alternating current I in = 100 mA and an excitation frequency 1 kHz was used. Figures 7-9 show that comparative analysis, referencing the conventional double-D coil, was performed with the same input parameter and inflated defect size to show the effect of field differences. Based on the simulation results, the new coil design showed the following strong merits. The simulation was computed under the same input parameter.    worthy to improve the spatial resolution and detection capability in comparison with the double-D coil. The average magnetic field intensity along the z axis in the midplane was more than two times with that of the double-D coil. Figure 7 shows that the new probe significantly improved the average magnetic field intensity at the sensor position. Figures 7 and 9 shows that unlike with the double-D coil, by applying the new coil design, it is possible to convey the secondary magnetic field towards the sensor direction.
Consequently, the new design probe helps to better resolve deeper subsurface defects, and thus, enhances the flaw detection capability.

Experimental Setup
The experimental measurement system consists of three components (Fig. 10): 1) coil and sensor power source (AC and DC), 2) XY probe scanning, 3) digital output data receiving and current measuring.
To excite the probe coil with a variable range of frequencies, a high-performance function generator (GW-Instek-2120) with a frequency range of (1 Hz-20 MHz) was used. Depending on the location of the defect, a wider amplitude voltage source is required for the probe excitation coil, so a BQ3100 power am- plifier was used. A 5V DC power source was used for the fluxgate magnetometer and the current measuring transducer. To perform the measurements on the specimens, an XY-scanning table, with either fully or semi-automatic is essential. A frictionless semi-automated numerical controlled XY-scanning machine with an accuracy of 0.1 µm was used. To hold the specimen and the probe with a scanning machine, a special nonmagnetic fixture was prepared. A TDS7104 digital storage oscilloscope with four-input channel was used to display the output voltage signal from the sensor, excitation voltage, and current. The distance between the probe and the specimen mainly depends on the surface condition. The probe surface was in contact with the specimen in this experiment.

Materials and Experimental Parameters
An aluminum alloy (Al-6061) plate with artificial surface and subsurface flaws of various sizes was prepared. The flaws were induced by using CNC milling and EDM machine. Depending on their sizes and orientations, the flaws were categorized into four subgroups: 1) variable depth surface flaw, 2) variable length surface flaw, 3) variable width and depth thin shallow surface flaw, 4) variable depth under-surface flaw. The distribution of the flaws and the distances between them and the edges of the plate were considered, so that unnecessary distortion and compression of the eddy current flow were avoided. Tables 1 and 2 show the parameters used for the experiment in the above-mentioned categories.

EXPERIMENTAL RESULTS AND DISCUSSION
As mentioned earlier in the introduction section, the excitation AC frequency and probe coil size have essential variables to convey the process of improving the sensitivity of eddy current nondestructive testing [7]. Based on the skin depth  and the material type of work plate, as in Eq. (1), appropriate frequency values were selected for detecting surface and subsurface flaws. The other factor for enhancing the sensitivity of the probe is the magnitude of alternating current. The magnetic field induced increases proportionally to the excitation AC increment [5]. However, optimal current (Iin = 100 mA) was used in all experiments, in order to work on both coil types for an extended period without overheating the probe skin depth is where  is resistivity of the material, f is frequency,  r is relative permeability, o = 4  10 -7 is permeability constant. Moreover, to show the significant effect of the designed probe on the detectability, comparative plots were performed. The comparison was carried out between the new probe coil and the reference double-D coil under the same input parameter, the same number of turns of the coil, and the same setup. To validate the obtained FEM comparative simulation results, a variable size flaw was prepared for the experimental work. In almost all of the experimental categories performed, the proposed probe considerably improved the detection sensitivity for both surface and subsurface flaws.
To illustrate the sensitivity improvement by the proposed probe in comparison to the double-D coil, experiment 1 was conducted with a variable depth surface defect ( Table 1). The proposed probe offered a significant sensitivity improvement. These experi-  ments were conducted under three selected range frequencies, and at 3 kHz, both coils have shown good results. However, Fig. 11 shows that the proposed probe appreciably improved the sensitivity. Experiment 2 was conducted on a variable length surface flaw by keeping the other flaw dimensions constant with the parameters shown in Table 1. The variable length flaw was also easily detected using the proposed probe, as shown in Fig. 12.
As in experiment 1, the best result in both coil types was obtained at 3 kHz frequency. In this case, the sensitivity is also improved, and the output voltage differences are raised, especially with the defect length increase, as shown in Fig. 12.
To further challenge the sensitivity of the proposed probe, experiment 3 was carried out by applying the parameter shown in Table 2, and using thin and shallow surface cracks with variable width and depth cracks.
In all cases, the sensitivity was considerably improved by the proposed probe, as shown in Fig. 13.
Improving the detectability of subsurface flaws is another major expected target in this study. Experi-   (3), with the double-D coil 100 (4), 300 (5), 1000 Hz (6) (color online). ment 4 was specially designed to assess the detection capability of subsurface flaws. For this experimental category, a hole with a diameter of 1.8 mm was drilled into the side of the plate. Table 2 shows the experimental data used for the comparison.
At 300 Hz frequency, the detectability was substantially improved when the coil with the proposed design was used, as shown in Fig. 14. A good result was also obtained in the detection of subsurface flaws.
To figure out if the new design coil is capable enough at the smallest unit scan area of defect, a surface flaw of 1  1  30 mm 3 size was taken from the previously mentioned sample. During the experiment, 2 kHz frequency and 100 mA coil excitation current were used for result comparison. Figure 15 shows the improved spatial resolution due to new design coil by the data taken from the probe with 1mm interval along and across surface flaw. The magnitude of color intensity is designated by output voltage (mV) from the sensor.
During all experimental setups, careful attention was given to the symmetricity of the field before performing all the experiments. This study implies that the coil design in a probe and its interaction with the magnetic sensor have a major influence on the detection sensitivity in eddy current testing.

CONCLUSIONS
In this study, the eddy current testing capability in the detection of surface and subsurface flaws was enhanced by introducing a novel probe design.
The simulation result showed that the average magnetic field intensity of the proposed probe is more than two times with that of the conventional double-D type probe at the sensor location. Furthermore, the eddy current distribution was appropriate for the detection of the flaws. The simulation results were validated by experiments, and both simulation and experimental results were compared to a testing specimen that uses a double-D coil under the same input parameter and setup.
The designed probe can be applied to the inspection of continuous manufacturing processes and finished products.
The study indicated the potential of the developed design to suit the recently emerging magnetic sensors to retrofit the eddy current testing in a wide range of applications. Even though the considerations to design the proposed probe are valid for different applications, the probe is specially designed to work with the DRV425EVM fluxgate sensor and further work is needed to adapt the probe to other magnetic sensors.

FUNDING
This study was supported by the Ministry of Knowledge Economy through the Regional Innovation Centre Program.