Printed Circuit Board Fault Inspection Based on Eddy Current Testing Using Planar Coil Sensor

This paper presents a printed circuit board (PCB) fault inspection method using eddy current testing generated from Helmholtz coils with a planar array-coil sensor to locate and inspect short and open faults on uniformly spaced interconnect single layer PCBs. The differences between the induced voltages from fault-free boards and faulty boards will be recorded in tables and translated into contour plots. The experimental results showed that in the presence of a short fault, the differences between the induced voltages from fault-free and faulty boards are highly negative. However, in the presence of an open fault, the differences between the induced voltages from fault free and faulty boards are highly positive. These highly positive or negative induced voltages can be translated into high density color regions on contour plots. The potential fault positions can be located by observing the color regions of the contour plots with respect to each element of the matrix sensor.


Introduction
Testing is one of the important stages in the production of electronic products to ensure a fully functioning system.The reliability of PCBs affects the functionality of the devices on which they are mounted.The causes of electronic circuit failure are many, but they can be lumped into two possible categories: shorts and opens (Gizopoulos, 2006).To avoid components from getting damaged and further problems upon power up, the PCB must pass through fault inspection testing.Many testing techniques have been researched and developed for PCB (Benedek, Krammer, Janoczki, & Jakab, 2013;Guifang, Hongcui, Zhenyu, & Yen-Wei, 2012).However, they are either time consuming or not efficient (Enke & Dagli, 1997).Visual methods are based on visual recognition processes which are limited to the inspection of the visible faults (Hara, Doi, Karasaki, & Iida, 1988).Other methods such as reflected light method, automatic optic inspection, and image analysis have limitations when an image surface becomes dark and the density of the circuit is high (Hara, Akiyama, & Karasaki, 1983;Shu-an & Fenglin, 2011).
When a conductive material is placed close to an alternating magnetic field, there will be an oscillating electrical current induced in the conductive material due to Faraday's law of induction (Ulaby, 2005).These induced eddy currents will produce a secondary alternating magnetic field which is in the opposite direction to the original alternating magnetic field (Raj, Jayakumar, & Thavasimuthu, 2007).The presence of any discontinuity or defect in the material will disturb the eddy current flow; hence, the secondary magnetic field will be disturbed.This disturbance can be detected as a voltage across a pick-up sensor which can be used to detect abnormality in the form of open and short faults on PCB traces.
Eddy current testing is a well-known method of non-destructive evaluation (NDE) that is applied to evaluate material flaws without changing or altering the material under test (Mukhopadhyay, 2005).Along with a variety of methods that include ultrasonic testing, dye penetrants and X-ray, eddy current testing is also commonly used for detecting fatigue cracks in conductive materials such as aircraft carriers and jet engines (Dogaru, C. Smith, Schneider, & S. Smith, 2004;Grimberg, Udpa, Udpa, & Savin, 2005;Uesaka et al., 1995).Recently, high frequency eddy current testing has been developed to detect micro defects on micro conductors of bare PCBs with various types of pick up sensors (Chomsuwan, Yamada, & Iwahara, 2007a, 2007b;Chomsuwan, Yamada, Iwahara, Wakiwaka, & Shoji, 2005;Kacprzak, Taniguchi, Nakamura, Yamada, & Iwahara, 2001;Yamada et al., 2004;Yamada, Chomsuwan, Hagino, Tian, & Iwahara, 2005;Yamada, Chomsuwan, & Iwahara, 2006;Yamada, Nakamura, Iwahara, Taniguchi, & Wakiwaka, 2003).In this paper, a low frequency eddy current testing probe structure is proposed which consists of Helmholtz coils exciter and a pick-up sensor made of 5 turn coils in planar array.The induced voltage of each array element is measured by Agilent multimeter.The resultant patterns are recorded in tabular form.The results from the fault free lines will then be used as the reference or ground truth values.Any voltage values which are high or low compared to the reference values will be defined as a potential fault on the PCB interconnect line.

Proposed Eddy Current Testing Probe Structure
In this research, a new design of ECT probe is constructed and used in detecting short and open faults on the PCB.The probe consists of a pair of Helmholtz coils, and a planar array-coil sensor.The planar array-coil sensor is fabricated on flame retardant 4 (FR4) board with thickness of 1.6 mm and copper cladding of 32 µm.This sensor is placed at 1.6 mm above the PCB trace and in the middle of a pair of Helmholtz coils.The sensor and the board under test are separated by FR4 substrate which acts as an insulator.The primary magnetic field generated by the pair of Helmholtz coils will cut through the PCB trace and induce eddy currents on the surface of the trace.The established eddy currents will circulate on the surface of the trace which further induces an alternating magnetic field perpendicular to the plane of the traces.The vertical magnetic field will cut through the axis of each element of the array-coil sensor.From the law of induction, an electrical current will flow through a conductor when a magnetic field cuts the conductor.Thus the voltage is induced across each coil.The induced voltages of the faulty and fault free lines are detected by the array-coil sensor.Any changes of eddy current due to a fault will introduce the changes of the induced voltage across the sensor.The detected induced voltage of each coil will be amplified and altered by the band pass filter.The induced voltages from each faulty line are compared to the induced voltage of the fault free lines.The altered output voltages will be recorded and used in fault inspection analysis to differentiate the faulty and non-faulty conductive lines.Figure 1 shows the front view of the proposed ECT probe used in PCB fault inspection.

Helmholtz coils Magnetic field
Figure 1.Front view of the proposed ECT probe

Experimental Setup
Figure 2 shows a block diagram of overall testing system.In this diagram there are a function generator, a pair of Helmholtz coils, devices under test (PCB and sensors), an amplifier circuit, a filter circuit, and a multi-meter.All the circuits are placed inside a magnetic chamber except the function generator and multi-meter.The pair of Helmholtz coils gets the sinusoidal signal supplied by the function generator at frequency of 800 Hz.The primary induced magnetic field from Helmholtz coils cut through the PCB lines which are located in the middle of the pair of Helmholtz coils.The eddy current sensor is placed on top of the line under test.The induced output voltage detected by the sensor will be amplified by the instrumentation amplifier circuit which has a gain of 10 000.The circuit comes after the amplifier is the active band pass filter circuit which is used to filter signals from amplifier's output terminal.The filtered signals will be measured in RMS value in millivolt ranges.
The detail view of the fault inspection system setup is shown in Figure 3.In the box, there are the pair of Helmholtz coils, eddy current sensor, device under test, amplifier circuit, and active band pass filter circuit.Figure 4 shows the inside view of the shielded box which is used to improve precision of the system which is susceptible to the noise of magnetic field.

Helmholtz Coils
A Helmholtz coil is a wire-wound device that produces nearly uniform magnetic field.The magnetic field produced from a pair of Helmholtz coils can be obtained by using the following formula: where 0  = 4π×10 −7 T•m/A (permeability of free space), n is the number of turns in each coil, I is the current flows through each coil, r is the radius of the coil.
Equation ( 2) can be obtained by applying Ohm's law to the current flowing through each coil of a resistor R with voltage supply of V.The AC input voltage signal in millivolt at 800 Hz from function generator is applied to each coil to generate the uniform magnetic field density in the middle of both coils.

Instrumentation Amplifier
In this work the instrumentation amplifier, INA129 from Texas Instrument is selected due to its high voltage gain up to 10 000 times, low power and good accuracy.It is placed right after the sensor in the fault inspection system.It is capable to amplify the sensor's output from microvolt ranges to millivolt ranges which can be read and observed clearly by the multi-meter.Its voltage gain, A v can be set by adjusting the value of R gain by the following formula The maximum gain of this amplifier is 10 000 with R gain equals to 4.94 Ω.The amplifier can be operated on power supplies ranging from ±2.25 V to ±18 V. Within this range the amplifier performance remains in good operation.The power supply voltages of ±15 V are chosen in order to obtain wider input linear range and bandwidth.At the highest gain of 10 000, its operation frequency bandwidth is 1 kHz.The input frequency is chosen to be 800 Hz to make sure the input voltage is within the operation frequency range.

Active Bandpass Filter
Active band pass filter is a cascade of a single high pass filter and a single low pass filter with the amplifier circuit in between.In active filter design, the amplification circuit is used to introduce gain and provide isolation between stages of filter.Under filter setup, the RC components of the band pass filter are calculated and its circuit is designed and simulated in PSpice.The band pass filter is designed with the center frequency of 800 Hz, and voltage gain, Av =1.The following formulas are used to calculate the gain, Q-factors, and center frequency of the band pass filter.Figure 5 shows that the band-pass filter at -3 dB has the lower frequency at 420.6 Hz and the higher frequency at 1.572 kHz.The center frequency located at 804.129 Hz has the loss of 4.25 dBm.The multiple feedback band pass filter is designed with the low Q factor to get wide bandwidth in order to pass wide range of the input frequency from the instrumentation amplifier.LM 358N operational amplifier is selected as the main component for the filter prototype.With the low power and wide bandwidth, this amplifier is suitable for multiple feedback active band pass filter design.The active band pass filter circuit configuration is realized as shown in Figure 6.The multiple feedback band pass filter topology allows the designer to be able to adjust the Q factor, and gain based on the values of resistors R 1 and R 2 .Resistor R 3 in this design is used to adjust the center frequency without affecting the bandwidth and gain.However, for the low Q factor the band pass filter can work without the resistor R 3 .From the LM 358N configuration the resistor R 3 can be selected as high input impedance of 10 kΩ and the output capacitor is selected to be 10C equal to 140 nF to obtain a smooth filter output.The parameters for this active band pass filter configuration setup are listed in Table 2

Array-Coil Sensor Characterization
This section focuses on the characterization of each element of the array-coil sensor.The matrix composes of 16 single coil sensors of 5 turns.The array-coil sensor is designed and fabricated on FR4 board.Figure 7 shows the array-coil sensor prototype of matrix 4 by 4. The sensitivities of each element of array-coil sensor are recorded in Table 3.Each row and column from Table 3 represents each element of the matrix sensor.From Table 3, each element's sensitivity in the same row is approximately close to each other.However, the sensitivity of the array-coil sensor drops from the 1 st to the 4 th row.This is due to the arrangement of array-coil sensor position, extra length of connectors or jumpers, and additional magnetizing conductive material on the board.The change in sensitivity may be due to the changes in area and shape of the jumpers which are connected from each element to the amplifier.It can be concluded that the sensitivity of the sensor is easily affected by the additional area created by external conductive wires or jumpers.Therefore, all the component parameters used in the array-coil sensor characterization are fixed in order to obtain optimal PCB fault inspection results.

PCB Inspection Setup
Having modeled and simulated PCB lines and eddy current sensor in Computer Simulation Technology (CST), the PCB design layouts from Advanced System Design (ADS) were sent for fabrication on FR4 board.Three sets of PCB lines have been fabricated for normal, opened and shorted lines.These boards were passed through ECT to inspect the changing induced voltage patterns in the conditions of faulty and fault-free and to locate the potential fault on the lines.Figure 9 and Figure 10 show the PCB normal line prototypes on FR4 boards.The PCB lines have been fabricated in two types of width; 300 µm and 600 µm with different line spacing of 4 mm, 2 mm and 1 mm.Having been discussed in Section 3.5, the reference or the normal boards were assumed to be fault free boards.Firstly, the fault free boards were inspected and the output induced voltages were recorded.Secondly, the single fault inspection was conducted and the output induced voltages were recorded.Thirdly, the two faults at a time inspection was experimented and the output induced voltages were recorded.Lastly, these recorded output voltages were analyzed to generate patterns to differentiate between faulty and fault free boards and to locate the potential fault on the lines.

Ground Truth Reference
The boards in Figure 9 and Figure 10 were used as the reference boards (fault-free boards).The induced voltages from these boards were tested and recorded for several times until encouraging results were obtained before proceeding further for PCB inspection.While the calibration of the testing system was done, the inspection on fault free boards would be conducted at least three times.The induced voltages from each board were recorded in three tables.These tables were used to obtain the table of the average induced voltage values.The average induced voltage values from each board were used as the ground truth data or reference values in order to do fault analysis patterns to differentiate between faulty boards and fault free boards.

Single Fault Inspection
Table 5 presents the difference between the induced voltages detected by each element of the array-coil sensor from the fault free boards and a single short fault boards.The highlighted elements have negative voltage values of higher than 30 mV.They are defined as the high resultant voltage values and a high potential location of a short fault occurrence on the lines which could be covered and observed by that element of the matrix.In the presence of a single short fault, higher induced voltages occur around the defect point compared to the fault free lines.This leads to a strong difference in the negative values in the matrix shown in the above Table 5.The matrices in Table 5 can be represented by the contour plots to clearly display and locate the defect point detected by the elements of the array-coil sensor.Figure 14 illustrates the contour plots of each matrix.The contour plots are obtained from MATLAB software.In Figure 14, the star shape on each contour plot represents the real physical fault location with respect to the matrix element.The negatively resulted induced voltage between the normal lines and short lines are displayed in the color ranges from light blue to dark color regions.From these plots, the array-coil sensor is able to locate the potential position of the single short fault.The fault can be clearly located at the highly negative voltage of the element of row 3, column 2.   From Table 6 it can be seen that the single open defect induced a lower voltage as compared to the normal lines.As a result, the difference of the induced voltage varies from the fault free lines and a single open defect line is highly positive in its values.These highly positive values which are greater than 30 mV are highlighted.The first matrix from PCB line width of 300 µm with the line spacing of 1 mm shows the high positive values located on the middle elements.This means that there is a possible fault occurrence on the line which is covered by the 2nd and 3rd row of the sensor.The location of the single open fault can be located around those highlighted elements.These matrices have been translated in the contour plots to ease the single open fault observation as shown in

Double Fault Inspection
In this inspection, two pairs of shorted lines are experimented on the same board.The array-coil sensor is used to detect the changes of the induced voltages from the boards.Table 7 shows the difference between the induced voltages from the reference lines and the two short faults on the lines.In Table 7, there are many highly negative values detected by the elements of the array-coil sensor compared to the inspection of single fault at a time.The potential fault locations can be found by observing the highlighted negative elements of greater than 30 mV.The contour plots have been used to interpret the matrices in Table 7.
Figure 16 shows the contour plots representing the difference between the induced voltages from the reference lines and the two short fault lines.The red starts represent the real locations of the faults.The highly negative regions have been represented by the color ranging from the light blue to black.Since there are two short faults on each board, the wider regions of the highly negative values in the contour plots can be observed.These regions give the information about the locations of the two short faults which can be detected by the elements of the matrix.These results have shown that two faults at time provide a clear visualization to locate the potential faults on the PCB lines compare to single fault at time.The two star locations are totally covered by the highly negative regions of the contour plots.In the case of two open faults at a time, the difference of induced voltages between fault free and faulty boards are highly positive.The high positive values occur around the potential defect points which are written in bold as shown in Table 8. Figure 17 illustrates these values of the matrices in the form of contour plots.From the contour plots the range of the high positive regions are in the ranges from orange to red to white color.These regions are the potential regions of the defect points.The experiments of PCB fault inspection have shown that the faulty boards induce higher or lower induced voltages compare to the induced voltages from the normal lines.The differences between the voltages from the fault free boards and faulty boards are recorded.The location of the faults can be identified by reading the contour plot color regions.From the observation, the PCB lines with the bigger line spacing of 4 mm produced a high difference between the induced voltages from the fault free boards and from the faulty boards.These differences have provided a clear observation to identify and locate the potential fault regions on the elements of the matrix as well as on the lines.Moreover, from the contour plots, two faults at a time could be noticed by having the wider regions of the highly negative or positive color ranges than a single fault.This is due to the higher number of faults which leads to higher changes of eddy current flow compared to one fault.

Conclusion
Printed circuit board (PCB) has played an important role as an electronic board to support and electrically connects the electrical components.With the advancement of the electronic circuit design technology, the reliability of the bare PCB on which the electronic components are mounted is very important to ensure a fully functioning system.The PCB inspection methods, such as optical inspection, and magnetic image analysis have been used to inspect the quality of PCBs.However, these methods are not sufficient to fully cover all of the possible faults as the density of the PCB traces become higher.In this work, the feasibility investigation of fault inspection using the magnetic field properties of PCB interconnect is experimented.The non-contact probe which is known as eddy current testing sensor (ECT) is realized.The sensor consists of the array of single coil of 5 turns is used to detect the changes of the induced magnetic field.This work has highlighted the proposed ECT sensor probe using the planar array-coils senor, to detect and locate the potential faults (short and open) on PCBs.
The PCBs have been fabricated and tested for three cases: fault free (reference), short and open fault boards.
The study on PCB fault detection has been conducted in the magnetic shielded box.It has been shown that the short fault induces higher voltage levels around the fault area compared to the fault free lines.It leads to highly negative in difference between fault free and short lines induced voltage values.In contrast to the short fault, the open fault induces lower voltage levels around the fault area compared to the fault free lines.Thus, the difference between fault free and open lines induced voltage is highly positive in values.The potential fault locations can be located and observed by the highly positive or negative region of the contour plots from each matrix sensor.
From the contour plots the PCB lines of width 300 µm and the line spacing of 4 mm provide a better vision of highly positive or negative regions of potential faults.The experimental results have proved the feasibility investigation of short and open faults inspection by exploiting the magnetic field property of PCB interconnects which is detected by the eddy current testing probe.The testing fault patterns have been generated and analyzed.

Figure 2 .
Figure 2. Testing system block diagram R 1 and R 2 are selected to be 10 kΩ and 20 kΩ respectively.The capacitor values can be achieved by assuming nF.The PSpice simulation result of band pass filter is shown in Figure5.

Figure 5 .
Figure 5. Band pass filter output response

Figure 6 .
Figure 6.Active band pass filter configuration

ArrayFigure 9 .
Figure 9. Line width=300 μm and the line to line distance of 4 mm, 2 mm and 1 mm respectively

Figure 11 .
Figure 11.Line width=300 μm and line to line distance of 1 mm, 2 mm and 4 mm respectively

Figure 13 .
Figure 13.PCB fault injection on PCB lines for a single and two faults at a time

Figure 14 .
Figure 14.Contour map of the difference of the induced voltages between the fault free and single short fault boards: the lines of the width of 300 µm: (a) 1 mm, (b) 2 mm and (c) 4 mm line spaces and the lines of the width of 600 µm, (d) 1 mm, (e) 2 mm and (f) 4 mm line spaces

Figure 15 .
Figure15.Contour map of the difference induced voltages between the fault free and single open fault boards: the lines of the width 300 µm, (a) 1 mm, (b) 2 mm and (c) 4 mm line spaces and the lines of the width 600 µm: (d) 1 mm, (e) 2 mm and (f) 4 mm line spaces

Figure 16 .
Figure 16.Contour map of the difference induced voltages between the fault free and two short faults boards: the lines of the width 300 µm: (a) 1 mm, (b) 2 mm and (c) 4 mm line spaces and the lines of the width 600 µm: (d) 1 mm, (e) 2 mm and (f) 4 mm line spaces

Figure 17 .
Figure 17.Contour map of the difference induced voltages between the fault free and two open faults boards: the lines of the width 300 µm: (a) 1 mm, (b) 2 mm and (c) 4 mm line spaces and the lines of the width 600 µm: (d) 1 mm, (e) 2 mm and (f) 4 mm line spaces Table 1 describes the specification of each Helmholtz coil used in this work.

Table 2 .
Active band pass filter parameters

Table 4 .
Table4provides the first, second, and third reference boards testing results and the average induced voltages from the three testing results.Average induced voltages for reference (fault free) boards

Table 5 .
Difference of the induced voltages from the fault free and single short fault boards

Table 6 .
Difference of the induced voltages from the fault free and single open fault boards

Table 7 .
Difference of the induced voltages from fault free and two short faults boards

Table 8 .
Difference of the induced voltages from fault free and two open faults boards