Induction defectoscope based on uniform eddy ...

Induction defectoscope based on uniform eddy current probe with GMRs Octavian Postolache

Artur Lopes Ribeiro

Helena Geirinhas Ramos

Instituto de Telecomunicações/IST, Lisboa, Portugal e-mail: [email protected]



Abstract— Defect detection in conductive plates represents an important issue. The present work proposes an induction defectoscope that includes a uniform eddy current probe with a rectangular excitation coil and a set of giant magnetoresistance sensors (GMR). The excitation current, the acquisition of the voltages delivered by the GMR and the signal processing of the acquired signal are performed by a real-time control and processing unit based on a TMS320C6713 digital signal processor (DSP). Different tests were carried out regarding the excitation coil position versus crack orientation and also regarding the GMR position inside the coil and the best response concerning the crack detection for a given aluminum plate specimen. Embedded software was developed using a NI LabVIEW DSP module including sinusoidal signal generation, amplitude and phase extraction using a sine-fitting algorithm and GUI for the induction defectoscope. Experimental results with probe characterization and detection of defects were included in the paper. Keywords-uniform eddy current probe, magnetoresistances sensors, non-destructive testing, digital signal processor

of the proposed UECP architecture to assure accurate detection of the defects in comparison with conventional pancake probe with GMR was considered. Additionally, descriptions of the conditioning circuits and of the embedded architecture based on the DSP that assures the excitation control, the acquisition, processing and defect signaling as parts of the proposed induction defectoscope, are included in the paper. Experimental results regarding the UECP tests and defect detection are also provided. UNIFORM EDDY CURRENT PROBE ARCHITECTURE

II.

The uniform eddy current probe (UECP), described on Figure 1, consists of a large wide tangential exciting coil (rectangular profile) and a set of three giant magnetoresistance sensors (GMR1, GMR2, GMR3) whose sensitive axis are perpendicular to the excitation field (Hex). H ex N

I. INTRODUCTION Eddy current testing has the important advantage of noncontact and fast testing of conductive plates. Thus, different eddy current probe (ECP) architectures to detect flaws are reported in the literature [1-4]. An excitation coil and detection coils usually form the eddy current probe. The use of giant magneto resistors as sensing components of ECPs has greatly increased its sensitivity [5]. Some drawbacks of this kind of architecture are the large noise and distortion due to the lift-off effect and to the electromagnetic material characteristics. Uniform eddy current probes are mentioned in literature as a solution that provides higher immunity to the noise associated with the lift-off effect. One uniform eddy current probe architecture is built with a tangential excitation coil having differential coil detection disposed inside the excitation coil [6]. Taking into account the advantages of uniform eddy current probes (UECP) and of the GMR sensor characteristics, this work presents a novel architecture implemented with a set of magnetoresistances (GMRs) positioned inside a rectangular coil that provides a uniform excitation field applied on the plane of the conductive plates to be tested. Based on the measurement of the magnetic field components tangential to the sample, defects such as flaws and holes in the specimen are detected. A practical approach concerning the capabilities

This work was supported in part by Portuguese Science and Technology Foundation - Project PTDC/EEA-ELC/67719/ 2006 and in part by the Instituto de Telecomunicações - Project CLASSE. This support is gratefully acknowledged.

permanent magnet

magnetic sensor sensitive axis S

rectangular excitation

H ex GMR1

conductive plate under test GMR3 GMR2

Figure 1. Uniform eddy current probe based on a rectangular excitation coil and a 3x GMR for defect detection in conductiove plates

The induction defectoscope system based on the use of a wide tangential excitation coil that induces uniform eddy currents presents an interesting solution to detect flaws and other nonuniformities, because due to its geometry it provides higher response immunity when compared with the generally used pancake solution. Due to a better signal to noise ratio (S/N), it

provides an increased accuracy in the defect detection. The geometrical characteristics of the used rectangular coil are 20 mm width, 20 mm height, 40 mm length, number of turns 80, and diameter of wire 0.5 mm. The rectangular coil was electrically characterized using an RLC-meter and the impedance and phase characteristics are presented in Figure 2.

Figure 2. The electrical characteristics of the used rectangular excitation coil

In order to obtain high sensitivity from the uniform eddy current probe a practical approach concerning the excitation coil position and GMR sensitive axis position versus the crack axis was carried out. The coil position corresponding to eddy currents parallel and perpendicular to the crack have been imposed, and the output signals were measured for different amplitudes and frequencies of the excitation current (100≤Iex≤300 mA and 1≤fIex≤10 kHz).Using the xy motion system, different aluminum plate specimens with induced cracks were scanned considering different distances between the excitation coil and the plate. Taking into account the crack depth, special attention was given to the excitation signal frequency taking into account the dependence between the frequency f and the aluminum plate standard penetration depth, δ:

δ=

1

π ⋅ f ⋅σ ⋅ µ

(1)

where µ is the magnetic permeability and σ is the electric conductivity of the plate. For the particular case of the aluminum plate to be tested, Al-2024-T3, µ = µ0 and σ = 1.75×106 S/m. Referring to the eddy current detection the GMR1, GMR2 and GMR3 (Fig. 1) are located inside the rectangular coil and detect the minor changes on the magnetic fields. The used

GMR sensors (AA002 from NVE) present high sensitivity that is independent of the used frequency. Each one of the used GMR sensors includes four 5 kΩ GMR resistors configured in a Wheatstone bridge as presented in Fig. 1. The magnetic characteristics of the sensor used as eddy current detector are: saturation field Hs=15 Oe, linear operation range between 1.5 Oe and 10.5 Oe, (1 Oe=79.577 A/m) and sensitivity of 4 mV/V⋅Oe. In order to detect the optimal mounting of the GMR sensors to attain a high sensitivity of the probe, different positions of the magnetic sensors axes related to the magnetic field detection and to the position of the crack were verified. Several geometrical scenarios are presented in Fig. 1 where the excitation magnetic field on the coil axis is perpendicular to the sensing axis of the GMR sensors package. In these cases the excitation magnetic field produced by the rectangular coil has no effect on the sensor output. However, the imperfections of the geometrical setup lead to a sensor output voltage different from zero even when the used aluminum plate specimen doesn’t present any flaws. Due to the GMR magnetic field sensor output V-shaped characteristic, presented in Fig. 4, a dc magnetic field generated by a small permanent magnet (AlNiCo type with Hc=51 kAm-1) was applied for biasing. A study concerning the position of magnet versus GMRs and detected eddy current magnetic field level and the distortion values was included in the present work. Referring to the scanning experiences of the aluminum plate specimens the uniform eddy current probe motion was assured using a xy-automated positioning system with high resolution (up to 50 µm) that receives the positioning commands from a laptop PC through an RS-232 interface (Fig.3). The UECP tests were done using a PXI system that was described in [7]. III. UECP CONDITIONING CIRCUITS The UECP conditioning circuit includes two main modules that are expressed by a voltage to current conversion module (V-I conv) and a programmable gain amplifier (PGA) module (Figure 3). The V-I converter is designed to provide an excitation current with appropriate values of amplitude and frequency. It is driven by a sinusoidal signal from the line out of the real-time control processing and communication units (RT-CPCU) based on TMS320C6713 and AIC23 codec. The voltage to current converter is implemented using an OPA2604 operational amplifier and SPP03N60C3 cool MOS power transistor. A set of precision, low power instrumentation amplifiers (INA118 from Burr-Brown) materializes the PGA that is used to amplify the GMR sensors voltage output (VGMRi). The PGA selection is done using a set of resistances and a switching circuit based on DG303 from Maxim. The switching circuit control is performed using one of the digital output lines of (RT-CPU). Filtering capacitors were connected to the GMR sensors and PGA power supplies in order to reduce the noise.

V. PRELIMINARY RESULTS In order to obtain the static and dynamic characteristics of the used GMR (AA002) a solenoid and Fluke5700 calibrator was used. Results concerning the GMR characterization are presented in Figure 4

RT-CPCU Embedded JTAG

USB host interface

DSP

SDRAM

USB FLASH CPLD

AIC23Codec DAC

DIO

Vout[mV]

USB/ RS232 conv RS232

TMS320C6713

Rotra Motion

ADC x

y MUX

UECP

PGA

≈ Iex≈

H[Am-1]

V-I conv

Figure 4. GMR sensor experimental characteristics (H – applied magnetic field intensity, Vout – GMR output voltage) Figure 3. The main blocks of the induction defectoscope based on uniform eddy current probe with GMR and real-time control, processing and comunication unit ( PGA- programable gain amplifier, V-I conv – voltage to current converter, MUX-multiplexer, RT-CPCU- real-time control processing and communication unit)

In order to extract the phase difference between the output voltage of the UECP and the excitation current Iex additionally a 10 Ω sensing resistor (Rs) is included in the system together with an instrumentation amplifier (INA118) whose output is applied to the second channel of the AIC23 codec.

IV.

In order to prove the UECP capability on crack detection an aluminum plate with an induced crack with 10 mm length, 0.5 mm width and 1mm depth was considered. Imposing a sinusoidal excitation current of 200 mA RMS and 5 kHz the obtained inductive images for GMR1 and GMR2 are presented in Figures 5-6. CK1

SPEC

UECP1

GMR1

EXCITATION CONTROL, ACQUISITION AND SIGNAL PROCESSING UNIT

The analog signals associated with the UECP outputs are acquired using the ADC in the AIC23 codec (Fig.3) included in the RT-CPCU board based on TMS320C6713 DSP operating at 225MHz [8]. This board also provides the sinusoidal excitation using the AIC23’ DAC. The maximum sampling frequency of the ADC is 96 kS/s that correspond to the imposed frequencies of the UECP excitation signal. Taking into account the AIC23 Codec DAC characteristics the imposed frequency for the excitation signal was limited for the preliminary tests at 5 kHz. Referring the software the NI LabVIEW DSP module functions [9] were used to perform the analog input, analog output and digital output tasks. The extraction of the acquired signal characteristics as amplitude and phase was done implementing a three parameters sine fitting software module [10]. The defect signaling software module is based on the derivative calculation of the amplitude as a function of the UECP position on the conductive plate. The derivative signal changes are associated with crack detection. Using one digital line of the DSP board DIO the aluminum plate crack detection is signaled.

VUECPN

10

1 0.5 0 0

5

10

20

30

40

x(mm)

0

y(mm)

Figure 5. Variation of normalized UECP voltage output (VUECP) for GMR1 as active magnetic field detector for an induced crack CK1 and for different positions of the UECP

Analyzing the obtained UECP voltage output profiles for two of the GMRs, it can be underlined that both of them provide clear information about the crack existence. However for extraction of additional elements such as crack localization and crack geometrical characteristics the utilization of the voltage output profile that corresponds to GMR2 usage represents a better choice.

At the same time the embedded software performs the peaks localization expressed in motion steps imposed by the used xy motion system. On the present case 1 mm motion step values were used. More elements related to the embedded software, and the GUI of the implemented induction defectoscope based on uniform eddy current probe and DSP will be included in the final form of the paper.

CK1

SPEC

GMR2

UECP

VUECPN GMR2

VI. CONCLUSIONS

1

0.5 10 0 0

5

10 20

y(mm)

x(mm)

30 40

0

Figure 6. Variation of the normalized UECP voltage output (VUECP) for GMR2 as active magnetic field detector for an induced crack CK1 and for different positions of the UECP (Spec – aluminum plate specimen)

The localization of the crack for the particular case when the uniform excitation field is applied perpendicular to the crack is based on derivative calculations. Thus, for one line scanning (fast conductive diagnosis case) the existence of critical points on the derivative points (maximum or minimum) (Figure 7) are detected using embedded software and a “1” logic is delivered through one line of the DIO port connected to the defect signaling LED. VUECP(V)

Single scan

Crack region

y(mm) dVUECP/dy

Crack localization

y(mm)

Figure 7. Single scan UECP voltage output (VUECP) profile and the derivative profile associated with crack localization

A new uniform eddy current probe that uses a rectangular coil and a set of GMRs was implemented and characterized. With this new probe we seek for higher reliability and accuracy with better signal to noise ratios in the experimental data, and overall better performances in comparison with single GMR probes. Appropriate conditioning circuits and real-time control processing and communication unit were designed and implemented. Embedded software processing modules were developed in order to assure amplitude and phase estimation of the acquired signal (using sine-fitting algorithm) and fast detection procedure of the cracks through amplitude or phase variation analysis of the UECP output signals for single line motion. Future work will concern the improvement of the software in order to allow the analysis of the inductive images associated with 2D scans. REFERENCES [1]

Dogaru, T.; Smith, S.T., “Giant magnetoresistance-based eddy-current sensor”, IEEE Transactions on Magnetics, Volume 37, Issue 5, pp. 3831 – 3838, 2001. [2] Postolache, O.; Dias Pereira, J. M.; Ramos, H. G.; Ribeiro, A. L.; "NDT on Aluminum Aircraft Plates based on Eddy Current Sensing and Image Processing", Proc IEEE I2MTC Conf, Victoria, Canada, Vol. I, pp. 1803 - 1808, May, 2008. [3] A. Lopes Ribeiro, H. Geirinhas Ramos, “Inductive Probe for Flaw Detection in non-Magnetic Metallic Plates Using Eddy Currents”, Proc. I2MTC-IEEE International Instrumentation and Measurement Technology Conference, Victoria, Canada, pp.1447-1453, 12-15 May, 2008. [4] T. E. Capobianco, J. D. Splett2and H. K. Iyer, “Eddy current probe sensitivity as a function of coil construction parameters”, Research in Nondestructive Evaluation Journal, Springer New York, Vol. 2, No 3 . pp. 169-186, September 1990 [5] A. Bernieri, G. Betta, G. Rubinacci, F. Villone, “A Measurement System Based on Magnetic Sensors for Nondestructive Testing” IEEE Transactions On Instrumentation And Measurement, Vol. 49, No. 2, pp. 445-459, April 2000 [6] Koyama K., Hoshikawa H., Taniyama N. “Investigation of Eddy Current Testing of Weld Zone by Uniform Eddy Current Probe” , Proceedings of WCNDT 2000, on-line at http://www.ndt.net/article/wcndt00. [7] H. Geirinhas Ramos, A. Lopes Ribeiro, Francisco Alegria, O. Postolache, “Virtual Instrument to Detect Defects in Conductive Materials”, Proc. 7th Conference on Telecomunications, Santa Maria da Feira, Portugal, pp.137-140, 3-5 May, 2009. [8] Spectrum Digital Inc., “TMS320C6713 – Technical reference” on-line at http://c6000.spectrumdigital.com/dsk6713/V2/docs/dsk6713_TechRef.p df [9] National Instruments, “Inside LabVIEW DSP”, on-line at: http://zone.ni.com/devzone/cda/tut/p/id/3501, Sept 2006 [10] M. Fonseca da Silva, A. Serra, “New methods to improve convergence of sine fitting algorithms”, Computer Standards & Interfaces, Volume 25 Issue 1 (March 2003), pp. 23-31.