sensors Article

A Steel Ball Surface Quality Inspection Method Based on a Circumferential Eddy Current Array Sensor Huayu Zhang 1 , Fengqin Xie 2, *, Maoyong Cao 3 and Mingming Zhong 1 1 2 3

*

College of Mechanical and Electronic Engineering, Shandong University of Science and Technology, Qingdao 266590, China; [email protected] (H.Z.); [email protected] (M.Z.) College of Transportation, Shandong University of Science and Technology, Qingdao 266590, China College of Electrical Engineering and Automation, Shandong University of Science and Technology, Qingdao 266590, China; [email protected] Correspondence: [email protected]; Tel.: +86-136-0648-9180

Received: 17 April 2017; Accepted: 26 June 2017; Published: 1 July 2017

Abstract: To efficiently inspect surface defects on steel ball bearings, a new method based on a circumferential eddy current array (CECA) sensor was proposed here. The best probe configuration, in terms of the coil quality factor (Q-factor), magnetic field intensity, and induced eddy current density on the surface of a sample steel ball, was determined using 3-, 4-, 5-, and 6-coil probes, for analysis and comparison. The optimal lift-off from the measured steel ball, the number of probe coils, and the frequency of excitation current suitable for steel ball inspection were obtained. Using the resulting CECA sensor to inspect 46,126 steel balls showed a miss rate of ~0.02%. The sensor was inspected for surface defects as small as 0.05 mm in width and 0.1 mm in depth. Keywords: steel ball; circumferential eddy current array sensor; defects

1. Introduction Surface defects on steel ball bearings seriously affect the stability and service life of the bearing itself. To rapidly and accurately inspect and identify defective steel balls is fundamental to the development of the bearing industry and its application fields [1]. The Aviko series automatic inspector (Sorting Solutions Ltd., Bílina, Czech Republic) mainly uses a combination of vibration, photoelectric, and eddy current sensors to inspect the surface and subsurface qualities of steel balls [2]. A double-probe eddy current sensor has been used to scan and inspect steel ball surfaces, with two probes symmetrically arranged at 180 degrees on the testing mechanism outside of the steel ball [1]. An instrument composed of several electromagnets and a ring-shaped capacitive sensor has been designed, yielding an instrument capable of inspecting surface defects 0.05 mm in diameter and 0.01 mm in depth on a 4 mm diameter steel ball bearing [3]. A visual inspection technique has been employed, with visual features providing a rapid means of inspection of single balls [4]. Considering the above methods, the setup with a double-probe eddy current sensor includes an unfolding wheel, as the core component of the equipment, that is easily damaged and requires regular replacement, resulting in high maintenance costs [1]. In the multi-sensor Aviko system, the meridian unfolding mechanism of the steel ball surface is set up to reduce the complexity of the unfolding wheel, but the unfolding mechanism is still complex. The system, which employs a ring-shaped capacitive sensor, needs to operate under lapping oil, despite possessing a simple unfolding mechanism [3]. The specular nature of the bearing surface makes it difficult to inspect subsurface defects of steel balls [4]. In this paper, a new method for the quality inspection of steel ball surfaces was proposed based on a circumferential eddy current array (CECA) sensor. The method simplified the unfolding mechanism, reduced processing costs, and improved the equipment’s service life. The CECA sensor was composed Sensors 2017, 17, 1536; doi:10.3390/s17071536

www.mdpi.com/journal/sensors

Sensors 2017, 17, 1536 Sensors 2017, 17, 1536

2 of 11

2 of 11

In this paper, a new method for the quality inspection of steel ball surfaces was proposed based on a circumferential eddy current array (CECA) sensor. The method simplified the unfolding of a number of probe coils in series, which were uniformly arranged on the latitudinal semicircle mechanism, reduced processing costs, and improved the equipment’s service life. The CECA sensor of was the external support of the inspected steel ball (Figure 1). When the inspected steel ball was on composed of a number of probe coils in series, which were uniformly arranged on the a one-dimensional meridional movement driven frictional steel wheel, steel ball surfacethe was latitudinal semicircle of the external support of by theainspected ballthe (Figure 1). When completely scanned and inspected by the CECA sensor. To inspect the ball surface completely, it was inspected steel ball was on a one-dimensional meridional movement driven by a frictional wheel, necessary to consider the cooperation of the CECA sensor and the unfolding mechanism of the steel the steel ball surface was completely scanned and inspected by the CECA sensor. To inspect the ball ball. The following points concerning configuration of the probe coils weresensor discussed: surface completely, it was necessarythe tobest consider the cooperation of the CECA and the unfolding mechanism of the steel ball. The following points concerning the best configuration of the (1)probe Thecoils theoretical basis of the probe operation. were discussed: (2) The appropriate number and arrangement forms of the probes. (1) The theoretical basis of the probe operation. (3)(2) The factors (Q-factors) of four kinds of probe (3-, 4-, 5-, and 6-coil probes), and the Thequality appropriate number and arrangement forms of thecoil probes. frequency of coil excitation currents. (3) The quality factors (Q-factors) of four kinds of probe coil (3-, 4-, 5-, and 6-coil probes), and the (4) The distributions of the eddy current density and magnetic field intensity on the ball surface frequency of coil excitation currents. by four kinds ofeddy eddycurrent currentdensity array coils under thefield same direction current, (4) induced The distributions of the and magnetic intensity onexcitation the ball surface and determination of theofoptimal configuration of under CECAthe sensors. induced by four kinds eddy current array coils same direction excitation current, and determination of the optimal configuration of CECA sensors.

Follower wheel Detect cavity

Ball inlet

Compression bar

Support wheel

Friction wheel

Stepper motor

Ball outlet Support frame

Spring

Steel ball

Support frame

(a)

(b)

Support wheel

Follower wheel

Compression bar

Steel ball

Circumferential Eddy Current Arrays

(c)

Figure 1. 1. The unfolding ball. (a) (a)The Themain mainview viewofofthe the unfolding mechanism; Figure The unfoldingmechanism mechanism of of the the steel ball. unfolding mechanism; (b)(b) The cutaway view The top topview viewofofthe theunfolding unfoldingmechanism. mechanism. The cutaway viewofofthe theunfolding unfoldingmechanism; mechanism; (c) The

UnfoldingMechanism Mechanismof ofthe theSteel Steel Ball Ball Surface Surface 2. 2. Unfolding Therequirements requirements for for the the surface surface defect a a The defect inspection inspection ofofsteel steelballs ballswere weremet metbybyusing using one-dimensional unfolding method to realize the full expansion of the steel ball. The unfolding one-dimensional unfolding method to realize the full expansion of the steel ball. The unfolding mechanism of the steel ball is shown in Figure 1, which was mainly comprised of supporting, follower, and friction wheels, and a CECA sensor. The tested steel ball regularly entered into the inspection cavity through the ball inlet from a feeding system.

Sensors 2017, 17, 1536

3 of 11

mechanism of the Sensors 2017, 17, 1536

steel ball is shown in Figure 1, which was mainly comprised of supporting, 3 of 11 follower, and friction wheels, and a CECA sensor. The tested steel ball regularly entered into the inspection cavity through the ball inlet from a feeding system. The and positioning positioningaaball, ball,while whilethe thefollower followerwheel wheel Thesupport supportwheel wheelwas wasused used for for supporting supporting and was used for applying a force on the ball through a compression bar and a spring to prevent balls was used for applying a force on the ball through compression bar and a spring to prevent balls from down,and andthe thefriction friction wheel rotated the steel a stepping frombouncing bouncingup up and and down, wheel rotated the steel ballsballs via a via stepping motor. motor. The CECA sensor collected the the ball’s surface quality information, and then, the computer processed the The CECA sensor collected ball’s surface quality information, and then, the computer processed information and judged whether thethe steel ballball surface possessed defects. the information and judged whether steel surface possessed defects. theinspection inspectionprocess, process, ifif the the photoelectric photoelectric sensor InInthe sensor detected detectedany anyposition positionchanges changesofofthethe compressionbar, bar,this thisindicated indicatedthat that aa ball ball had entered compression entered the theinspection inspectioncavity; cavity;ininthis thiscase, case,the thefriction friction wheel began to rotate and the CECA sensor was activated at the same time. After being completely wheel began to rotate and the CECA sensor was activated at the same time. After being completely inspected,the thesteel steelball ballwas wasmoved moved out out of of the the inspection inspection cavity sorting inspected, cavitythrough throughthe theball balloutlet outlettotothe the sorting system, and the next ball entered. Repeating the above process, ball surface quality inspection was system, and the next ball entered. Repeating the above process, ball surface quality inspection was entirely automated. entirely automated. BasicPrinciple PrincipleofofEddy EddyCurrent Current Testing Testing for for aa Steel Steel Ball Ball 3. 3. Basic Whenexcited excitedbyby alternating current, probe generated alternating magnetic fields When anan alternating current, thethe probe coilcoil generated alternating magnetic fields that that induced eddy currents on the steel ball surface ahead of the coil. During inspection of ball induced eddy currents on the steel ball surface ahead of the coil. During inspection of ball surface surfaceit quality, it was necessary to study eddy distribution current distribution on the ball. quality, was necessary to study the eddythe current on the ball. The model for using the probe coil to inspect a ball surface is shown in Figure 2. If a micro unit The model for using the probe coil to inspect a ball surface is shown in Figure 2. If a micro unit in in a section of coil was taken as dx∙dy, the spiral coil was considered a combination of many micro a section of coil was taken as dx·dy, the spiral coil was considered a combination of many micro units, units, and the magnetic field produced by them was obtained by superposition. The current density and the magnetic field produced by them was obtained by superposition. The current density of the of the micro unit was: micro unit was: N 𝑁𝐼I ∆i∆𝑖== (1) (1) R𝑜o − 𝑅 R𝑖i)ℎ )h ((𝑅

where NNis is the coil the coil, coil,hhisisthe thecoil coilthickness, thickness,RiRisi is the where the coilturns, turns,I Iisisthe thetotal totalcurrent current flow flow through through the the inner coil inner coilradius, radius,and andRRo oisisthe thecoil coil outer outer radius.

Figure 2. The model of eddy current testing. Figure 2. The model of eddy current testing.

According to Biot–Savart law, the magnetic induction intensity produced by the micro unit According to(0, Biot–Savart law, the magnetic(2): induction intensity produced by the micro unit dx·dy dx∙dy at point O 0) is described in Equation at point O (0, 0) is described in Equation2 (2): 𝜇0 𝑖 𝑥 𝜇0 𝑁𝐼 𝑥2 (2) 𝑑𝐵 = = 𝑑𝑥𝑑𝑦 2 2 3/2 2 2(𝑅𝑏µ−N𝑅I𝑎 )ℎ (𝑥 + 𝑦x22)3/2 µ02i (𝑥 + x2𝑦 ) 0 = dxdy (2) dB = 2 ( x2 + y2 )3/2 2( Rb − R a )h ( x2 + y2 )3/2

where µ0 = 4π × 10−7 (N·A−2 ) is the vacuum permeability, x is the radius of the micro unit, y is the distance between the micro unit and measured steel ball (lift-off), and i is the current through the micro unit dx·dy.

Sensors 2017, 17, 1536

4 of 11

Sensors 2017, 17, 1536

4 of 11

The magnetic induction−7intensity produced by the coil at point O is: −2

Where 𝜇0 = 4π × 10 (N·A ) is the vacuum permeability, x is the radius of the micro unit, y  is the distance between the micro unit and measuredqsteel ball (lift-off), and i is thep current through Z 2 + ( t + h )2  2 2 R R + o Ro + Ro + t  the micro unit dx∙dy. µ0 N I o q q BO = dB = − t ln (t + h) ln The magnetic induction 2( Ro − Rintensity 2 is: 2 2 2 at point O i ) h  produced by the coil

Ri +

Ri + ( t + h )

𝑅 + √𝑅 2 + (𝑡 + ℎ)2

𝜇 𝑁𝐼

Ri +

(3)

Ri + t

𝑅 + √𝑅 2 + 𝑡 2

𝑜 𝑜 where N 𝐵=𝑂 k=×∫(𝑑𝐵 Ro −=Ri ) ×0h/πRw 2(𝑡 , R+wℎ) isln the𝑜radius𝑜 of the enameled factor. − 𝑡 ln wire, and k is the filling (3) 2(𝑅𝑜 − 𝑅𝑖 )ℎ 2 2 2 2 Taking N into consideration, Equation (3) can𝑅be modified as: 𝑅𝑖 + √𝑅𝑖 + 𝑡 𝑖 + √𝑅𝑖 + (𝑡 + ℎ)

{

}

q   p  2 R2o + (t + R + )2 enameled Ro +wire,R2oand + t2kis the filling where 𝑁 = 𝑘 × (𝑅𝑜 kµ −𝑅 of hthe 0 I𝑖 ) × ℎ/𝜋𝑅𝑤 , 𝑅𝑤 ois the radius q q BO = (4) − t ln (t + h) ln 2 factor. Taking N into consideration, Equation (3) can be modified as: 2πR w  Ri + R2i + t2  Ri + R2i + (t + h)2 𝑘𝜇(4), 𝑅𝑜 +induction ℎ) 𝑅𝑜 + √𝑅𝑜 + 𝑡 √𝑅𝑜 + (𝑡 +intensity 0 𝐼 the magnetic According to Equation (𝑡 + ℎ) ln 𝐵𝑂 = − 𝑡 lnin the coil axis with four different (4) 2 2𝜋𝑅 𝑤 dimensional parameters carrying 0.5 A current √𝑅𝑖2calculated. √𝑅𝑖2with 𝑅𝑖 +was + (𝑡 + ℎ)2 The five 𝑅𝑖 +coils + 𝑡 2 different parameters { } are listed in Table 1. 2

2

2

2

According to Equation (4), the magnetic induction intensity in the coil axis with four different Table0.5 1. Parameters thecalculated. probe coil. The five coils with different dimensional parameters carrying A current of was parameters are listed in Table 1. Coil Number The(the Outer Diameter R The Inner Diameter Thelift-off) Thickness h (mm) o (mm) the Then, the Bo–t relationship between magnetic inductionRintensity and curves of i (mm) the five 1# different parameter coils 4.0 were plotted (Figure 3). 3.0 2.5 2# 4.0 2.0 2.5 3# 2.5 Table 1. Parameters of the probe 2.0coil. 2.5 4# 2.0 1.0 2.0 Coil Number The Outer Diameter Ro (mm) The Inner Diameter Ri (mm) The Thickness h (mm) 5# 2.0 1.0 1.5

the

1# 4.0 3.0 2# 4.0 2.0 3# 2.5 2.0 Then, the 4# Bo–t (the relationship 2.0 between the magnetic 1.0induction 5# parameter coils were 2.0 five different plotted (Figure 3). 1.0

intensity

2.5 2.5 2.5 and 2.0 1.5

lift-off) curves of

Figure 3. The relationship curves between magnetic induction intensity and lift-off.

Figure 3. The relationship curves between magnetic induction intensity and lift-off.

The magnetic induction intensity was observed to rapidly decrease with increasing distance t (lift-off, Figure induction 3)). For 1# coil, this decrease was 87%toatrapidly t = 0.5 mm, 74% atwith t = 1increasing mm, and 22% at The magnetic intensity was observed decrease distance t t = R o. For 2# coil, the magnetic induction intensity decreased to 84% at t = 0.5 mm, 67% at t = 1 mm, (lift-off, Figure 3)). For 1# coil, this decrease was 87% at t = 0.5 mm, 74% at t = 1 mm, and 22% at t = Ro . and 16% at magnetic t = Ro. The intensity of intensity 3# coil decreased to 78% at t = at 0.5t mm, at 67% t = 1 mm, 22% and For 2# coil, the induction decreased to 84% = 0.558% mm, at t =and 1 mm,

16% at t = Ro . The intensity of 3# coil decreased to 78% at t = 0.5 mm, 58% at t = 1 mm, and 22% at t = Ro . The intensity of 4# coil decreased to 66% at t = 0.5 mm, 41% at t = 1 mm, and 16% at t = Ro . The intensity of 5# coil decreased to 67% at t = 0.5 mm, 41% at t = 1 mm, and 16% at t = Ro . From the

Sensors 2017, 17, 1536

5 of 11

Bo–t of 4# 2017, and 17, 5#1536 coils, thickness showed little influence on the coil’s magnetic induction5intensity. Sensors of 11 Therefore, the closer the distance between the ball and probe, the stronger the eddy current intensity, at t =the Ro. The intensity of 4#conditions coil decreased to for 66%the at tsurface = 0.5 mm, 41% at t = 1 mm, and 16% balls. at t = ROn o. and thus more favorable were quality inspection of steel the The intensity of 5# coil decreased to 67% at t = 0.5 mm, 41% at t = 1 mm, and 16% at t = Ro. From the other hand, the probe coil was arranged on the support frame outside the steel ball (Figure 1c) and Bo–t of 4# and 5# coils, thickness showed little influence on the coil’s magnetic induction intensity. the closer the steel ball was to the probe, the more difficult it was for the steel ball to enter the test Therefore, the closer the distance between the ball and probe, the stronger the eddy current chamber. A compromise was adopted in which thethe inspection distance between the steel intensity, and thus the scheme more favorable conditions were for surface quality inspection of steel ball and probe was 0.5 mm, which was not only convenient for introducing a ball, but also ensured balls. On the other hand, the probe coil was arranged on the support frame outside the steel ball inspection of closer the surface quality. (Figureaccuracy 1c) and the the steel ball was to the probe, the more difficult it was for the steel ball to enter the test chamber. A compromise scheme was adopted in which the inspection distance

4. Arrangement and Finite Analyses of the was CECA between the steel ball andElement probe was 0.5 mm, which not Sensor only convenient for introducing a ball, but also ensured inspection accuracy of the surface quality.

4.1. Arrangement of the CECA Sensor Probe Coil

4. Arrangement and Finite Element Analyses of the CECA Sensor The CECA coils needed to cover half of the ball circumference to inspect the steel ball surface completely, with each coil covering the distance of C/2n along the ball’s circumferential direction 4.1. Arrangement of the CECA Sensor Probe Coil (Figure 1c). Consequently, the outer radius Ro of each coil was: The CECA coils needed to cover half of the ball circumference to inspect the steel ball surface completely, with each coil covering the distance C/2n along the ball’s circumferential direction Ro = 2Rsin 2Rsin (C/4nRof) = (π/2n ) (Figure 1c). Consequently, the outer radius 𝑅𝑜 of each coil was:

(5)

where C is the steel ball circumference, R is ⁄the ball radius, ) = 2𝑅𝑠𝑖𝑛(𝜋 ⁄2𝑛) and n is the number of (5)CECA 𝑅𝑜 = 2𝑅𝑠𝑖𝑛(𝐶 4𝑛𝑅steel probe coils. where C is the steel ball circumference, R is the steel ball radius, and n is the number of CECA probe For coils.a steel ball 8 mm in diameter, if the number of probe coils was 2, the angle between the two coils was For 90 degrees and theinouter coil diameter wasofmeasured 5.66 mm. Such a large a steel ball 8 mm diameter, if the number probe coils at was 2, the angle between thecoil twoouter diameter (relative to an and 8 mm the sensor sensitivity and caused inspection coils was 90 degrees the ball) outerreduced coil diameter was measured at 5.66 mm. Such aerrant large coil outer of diameter (relative an 8 mm reduced the 6, sensor sensitivity caused errant inspection of coil the ball surface. If the to number of ball) probe coils was the balls were and completely inspected, but the the ball surface. If the of probe 6, no theneed balls were inspected, the coil the was difficult to mount onnumber the support, so coils therewas was to usecompletely more than six coils.but Therefore, was difficult mountprobe on the support, so there was use more thanthere six coils. proper number oftoCECA coils was concluded to no be need 3 to 6.toThat is to say, wereTherefore, four possibly the proper number of CECA probe coils was concluded to be 3 to 6. That is to say, there were the fourangle useful probe coil arrangements: 3-, 4-, 5-, and 6-coil probes. For this series of probe forms, possibly useful probe coil arrangements: 3-, 4-, 5-, and 6-coil probes. For this series of probe forms, between each coil was 60, 45, 36, and 30 degrees and the outer coil diameter was 4, 3, 2.5 and 2 mm, the angle between each coil was 60, 45, 36, and 30 degrees and the outer coil diameter was 4, 3, 2.5 respectively (Figure 4). and 2 mm, respectively (Figure 4).

4

(b)

c coil 3 oil 4 l2 co i

steel ball

steel ball

(c)

(d)

coil 6

5

coil 1

5

1

il

coil

coil

co

co il 4

Figure 4. The arrangement of the multi-coil probe of the circumferential eddy current array (CECA)

Figure 4. The arrangement of the multi-coil probe of the circumferential eddy current array (CECA) sensor. (a) 3-coil probe arrangement of CECA sensor; (b) 4-coil probe arrangement of CECA sensor; sensor. 3-coil probe arrangement of sensor; CECA (d) sensor; 4-coil probe arrangement of CECA sensor; (c) (a) 5-coil probe arrangement of CECA 6-coil(b) probe arrangement of CECA sensor. (c) 5-coil probe arrangement of CECA sensor; (d) 6-coil probe arrangement of CECA sensor.

l1 coi

coil

coil

(a)

coil 3

co

3

steel ball

steel ball

il 2

coil

2

1

coil l3 coi

coi l1

coil 2

Sensors 2017, 17, 1536

6 of 11

Sensors 2017, 17, 1536

6 of 11

4.2. The Coil’s Q-Factor 4.2. The Coil’s Q-Factor The Q-factor directly affected the inspection accuracy of the steel ball surface quality. When a coil was close to the steel ball, the power the consumption the coil increased, and thesurface coil’s Q-factor The Q-factor directly affected inspectionofaccuracy of the steel ball quality. reduced. When a The Q-factor, influenced measured clearanceoft (lift-off), was expressed as Q(x) [5]: coil coil’s was close to the steel ball,by thethe power consumption the coil increased, and the coil’s Q-factor reduced. The coil’s Q-factor, influenced by the measured clearance t (lift-off), was expressed as Q(x) Q( x ) = 2π f L( x )/R( x ) (6) [5]:

⁄𝑅(𝑥) (6) 𝑄(𝑥)to=see 2𝜋𝑓𝐿(𝑥) According to Equation (6), it was easy that the Q-factor was determined by the measured clearance t, the frequency of (6), the it coil’s current, the steel ball’s material.byIfthe themeasured model of According to Equation wasexcitation easy to see that theand Q-factor was determined the steel ball was determined, same material andball’s steelmaterial. ball batch, the model Q-factor clearance t, the frequency of theunder coil’sthe excitation current,properties and the steel If the of of the probe coil was mainly determined by the measured clearance and frequency of the coil’s the steel ball was determined, under the same material properties and steel ball batch, the Q-factor excitation current. of the probe coil was mainly determined by the measured clearance and frequency of the coil’s When the distance between the probe coil and steel ball was 0.5 mm, the inductance and excitation current. resistance values were measured using inductance, and resistance (LCR) meter. In and this When the distance between the an probe coil andcapacitance, steel ball was 0.5 mm, the inductance work, an LCR IM3536 (Hioki E.E. Corp., Nagano, capacitance, Japan) was used measure(LCR) the inductance resistance values weremeter measured using an inductance, and to resistance meter. In and resistance. this work, an LCR IM3536 meter (Hioki E.E. Corp., Nagano, Japan) was used to measure the The parameters for multi-coil probes in a CECA sensor under different arrangements are listed in inductance and resistance. TableThe 2. The four kinds sensor setups (3-, in 4-,a5-, and 6-coil probes) all composed of copper wire, parameters forofmulti-coil probes CECA sensor underwere different arrangements are listed 7 S/m [6]. According to Equation (6), with the frequency of the coil’s with a conductivity of 5.998 × 10 in Table 2. The four kinds of sensor setups (3-, 4-, 5-, and 6-coil probes) were all composed of copper excitation variable of from 350× kHz to 1.3 the corresponding was calculated wire, withcurrent a conductivity 5.998 107 S/m [6].MHz, According to Equation Q-factor (6), withvalue the frequency of the and a fitting curve is plotted (Figure 5). coil’s excitation current variable from 350 kHz to 1.3 MHz, the corresponding Q-factor value was calculated and a fitting curve is plotted (Figure 5).

Table 2. The parameters for multi-coil probes in a CECA sensor under different arrangements and parameters of the measured steel ball. Table 2. The parameters for multi-coil probes in a CECA sensor under different arrangements and parameters of the measured steel ball. Parameters 3 Coils 4 Coils 5 Coils 6 Coils Parameters

The outer coil diameter (mm) 4 The outer coil diameter (mm) The inner coil diameter (mm) 1.5 The inner coil diameter (mm) Coil thickness (mm) 2 Coil thickness (mm) Copper wire diameter (mm) 0.06 Copper wire diameter (mm) Turn number of each coil (mm) 320 Turn number of each coil (mm) Steel ball diameter (mm) 8 Steel ball diameter (mm) Steel ball material quality GCr15 Steel ball material quality

3 Coils 4 Coils 3 4 3 1.5 1.5 1.5 2 2 2 0.06 0.06 0.06 260 320 260 8 8 8 GCr15 GCr15 GCr15

5 Coils 6 Coils 2.5 2.5 2 1.5 1.5 1 2 2 1.5 0.06 0.06 0.06 222 222 200 8 8 8 GCr15 GCr15 GCr15

2 1 1.5 0.06 200 8 GCr15

Figure with different different excitation excitation current current frequencies. frequencies. Figure 5. 5. Q-factors Q-factors of of multi-coil multi-coil probes probes with

The Q-factor of a 3-coil probe in a CECA sensor reached a maximum of 20.1 at an excitation frequency of 920 kHz, but at 940 kHz, the Q-factor of 4-, 5-, and 6-coil probes reached the maximum,

Sensors 2017, 17, 1536

7 of 11

The Q-factor Sensors 2017, 17, 1536 of a 3-coil probe in a CECA sensor reached a maximum of 20.1 at an excitation 7 of 11 frequency of 920 kHz, but at 940 kHz, the Q-factor of 4-, 5-, and 6-coil probes reached the maximum, values of of 2121.5, 2121.5, and and 21.6, 21.6, respectively respectively (Figure (Figure 5). 5). From the above analysis, a coil excitation excitation with values kHz in in aa CECA CECA sensor sensor was was found found to to be be reasonable reasonable for for 4-, 4-, 5-, 5-, and and6-coil 6-coilprobes. probes. current of 940 kHz 5. The Finite Element Analyses in a CECA Sensor Under frequency andand lift-off conditions, the stronger the eddy density Under the thesame sameexcitation excitation frequency lift-off conditions, the stronger thecurrent eddy current and magnetic field intensity on the ball the lower missed rate for ball density and magnetic field intensity on surface, the ball surface, thethe lower the inspection missed inspection rate defects. for ball The optimal coils inofthe CECA sensor was sensor determined by analyzingbythe distribution defects. The number optimal of number coils in the CECA was determined analyzing the of induced eddy currenteddy density and magnetic field intensityfield on steel balls on under coil distribution of induced current density and magnetic intensity steeldifferent balls under arrangements with the finitewith element method. different coil arrangements the finite element method. Based on the above analysis of the probe coil’s Q-factor, Q-factor, the the parameters parameters of the the multi-coils multi-coils used in the CECA sensor and the measured ball are listed in Table 2, 2, with with the the material material of the steel ball being GCr15 (similar in grade to 52,100 in American, IIX15 in Russia, SUJ2 in Japan, En31 in the UK, and 100C6 in permeability of of GCr15 was µμrr ≈ 1.0 and in France). France). The relative magnetic permeability and the the electrical electrical conductivity 4.3478 × 1066S/m S/m[7]. [7]. × 10 The magnetic induction intensity distribution curves curves on a steel ball with different probe coil arrangements under the same direction excitation excitation current current (current, (current, 0.5 0.5 A A and and frequency, 940 kHz) are shown in Figure 6. On the ball surface, for 3- and 4-coil probes, the magnetic induction intensity maximum amplitude was 26 mT, while the 5- and 6-coil probes were 28 and 25 mT, respectively. respectively. The distribution of magnetic induction intensity in 5- and 6-coil probes was observed to be more uniform than that of 3- and 4-coil probes. probes.

Figure 6. 6. Distribution Distributioncurves curvesofofinduction inductionintensities intensities a steel ball with different probe under Figure onon a steel ball with different probe coilscoils under the the same direction excitation current. same direction excitation current.

The eddy current distribution of the CECA sensor on the ball surface showed that the The eddy current distribution of the CECA sensor on the ball surface showed that the maximum maximum eddy current density of the 3- and 4-coil probes was 2.6637 × 107 and 2.4137 × 107 A/m2, eddy current density of the 3- and 4-coil probes was 2.6637 × 107 and 2.4137 7× 107 A/m2 , respectively, respectively, while that of the 5- and the 6-coil probes were 2.3576 × 10 and 1.9662 × 107 A/m2, while that of the 5- and the 6-coil probes were 2.3576 × 107 and 1.9662 × 107 A/m2 , respectively respectively (Figure 7). Although the eddy currents induced by 3- and 4-coil probes were denser (Figure 7). Although the eddy currents induced by 3- and 4-coil probes were denser than that of than that of the 5- and 6-coil probes, the distribution was not uniform. In particular, between the 5- and 6-coil probes, the distribution was not uniform. In particular, between adjacent coils, adjacent coils, the minimum induction eddy current densities of the 3- and 4-coil probes 3were the minimum induction eddy current densities of the 3- and 4-coil probes were 3.288 × 10 and 3.288 × 103 and 1.3287 × 104 A/m2, respectively, which was a disadvantage for quality inspections of steel ball surfaces. The induced eddy current densities of 5- and 6-coil probes were uniform, but the induced eddy current density of the 5-coil probe was larger than that of the 6-coil probe. Therefore, the 5-coil probe was chosen as the optimal coil arrangement in this study.

Sensors 2017, 17, 1536

8 of 11

1.3287 × 104 A/m2 , respectively, which was a disadvantage for quality inspections of steel ball surfaces. The induced eddy current densities of 5- and 6-coil probes were uniform, but the induced eddy current density of the 5-coil probe was larger than that of the 6-coil probe. Therefore, the 5-coil probe was chosen as 17, the1536 optimal coil arrangement in this study. Sensors 2017, 8 of 11

Coil1

Coil1

Coil2 Coil2 Coil3 Coil3 (a)

Coil4 (b)

Coil1

Coil1

Coil2 Coil2 Coil3 Coil3

(c)

Coil4

Coil4

Coil5

Coil5

Coil6 (d)

Figure 7. 7. Eddy distributions of multi-coil probes in a CECA sensorsensor on steelon ball surfaces. Figure Eddy current currentdensity density distributions of multi-coil probes in a CECA steel ball The eddyThe current density distribution on a steel ball by (a)induced 3-coil probe; 4-coilprobe; probe; surfaces. eddy current density distribution onsurface a steel induced ball surface by (a)(b) 3-coil (c) 4-coil 5-coil probe; probe; (c) (d)5-coil 6-coilprobe; probe.(d) 6-coil probe. (b)

Experiments 6.6.Experiments Thefeasibility feasibility of of using a CECA sensor was verified by The sensor to toinspect inspectthe thesurface surfacequality qualityofofa asteel steelball ball was verified performing experiments onon8 8mm lift-off and andthe the by performing experiments mmGCr15 GCr15steel steelballs ballsunder underthe the conditions conditions of 0.5 mm lift-off same excitation current each probe coil coil (current of 0.5of A 0.5 and A current frequency of 940 kHz). same currentdirection directioninin each probe (current and current frequency of A 5-coil sensor was sensor used, inwas which the in coils werethe uniformly distributed on the semicircle the 940 kHz).CECA A 5-coil CECA used, which coils were uniformly distributed onofthe inspectionof chamber, and the chamber, parameters of the coils and steel shown in Table semicircle the inspection and parameters ofball the are coils and steel ball2.are shown in Table 2.

6.1. Inspection System A block diagram of the experimental set-up for CECA sensor measurements, which mainly included a CECA sensor, a data acquisition card of USB4711A (Advantech Co., Ltd., Milpitas, CA, USA), and a computer, is shown in Figure 8. The CECA sensor inspection circuit was composed of an oscillator, inspector, amplifier, and bias adjustment circuit (Figure 9). A capacitance-connecting

Sensors 2017, 17, 1536

9 of 11

6.1. Inspection System A block diagram of the experimental set-up for CECA sensor measurements, which mainly of 11 11 sensor, a data acquisition card of USB4711A (Advantech Co., Ltd., Milpitas,99of CA, USA), and a computer, is shown in Figure 8. The CECA sensor inspection circuit was composed of frequency carrier carrier signal signal by by the the capacitive capacitive three-point three-point oscillation circuit. circuit. The peak peak inspection inspection circuit circuit frequency an oscillator, inspector, amplifier, and bias adjustmentoscillation circuit (Figure 9).The A capacitance-connecting was used to inspect the output signal amplitude. After inspection, the output voltage signal was was to inspect the output signal amplitude. After inspection, thethe output threeused point-type oscillator was adopted as the oscillation circuit, using CECAvoltage sensor signal coils aswas the amplified and biased by the interface circuit, which helped in matching with the input voltage range amplified and biased by the interface circuit, which helped in matching with the input voltage range inductive elements. The sensor’s Q-factor was changed into the amplitude and frequency of the high of the the data data acquisition acquisition card. card. of frequency carrier signal by the capacitive three-point oscillation circuit. The peak inspection circuit The voltage signal of the the measured measured steel steel ball ball was was compared compared with with the the setting setting threshold threshold to to judge judge signal was The usedvoltage to inspect theofoutput signal amplitude. After inspection, the output voltage signal was whether the the steel steel ball ball possessed possessed aa defect(s), defect(s), which which in in turn turn controlled controlled the the actions actions of of the the sorting sorting whether amplified and biased by the interface circuit, which helped in matching with the input voltage range system, such such that that aa defective defective steel steel ball ball was was identified identified and and sorted. sorted. system, of the data acquisition card.

coil coil33 cco oilil iolil22 o 44 cc

Sensors 2017, 2017,a17, 17, 1536 Sensors 1536 included CECA

il 55 ccooil

usb 4711A 4711A data data acquisition acquisition card card usb Excitation Excitation & & detection detection

steel ball ball steel

Computer(Labview) Computer(Labview)

circuit circuit Steel Ball Ball surface surface quality quality Separation Separation System System Steel

ccooiill 1 1

8. Block the experimental experimental set-up set-up for for CECA CECA sensor sensor measurements. measurements. Figure 8. Block diagram diagram of of the measurements. Figure

+Vcc +Vcc LL C1 C1

R2 R2 R1 R1

C4 C4 D1 D1

Uaa U

R3 R3

Q1 Q1

R6 R6

Uaa U R8 R8

C2 C2 C3 C3

R7 R7

R10 R10

output output signal signal

R12 R12

R9 R9 C5 C5

R4 R4 L1 L1

R11 R11

D3 D3

D2 D2

R5 R5

voltage regulator regulator voltage

Three point point Three oscillator circuit circuit oscillator

Detector Detector circuit circuit

Amplifier and and bias bias Amplifier adjustment circuit circuit adjustment

Figure 9. The The CECA sensor sensor inspection circuit. circuit. Figure Figure 9. 9. The CECA CECA sensor inspection inspection circuit.

6.2. Results Results 6.2. The voltage signal of the measured steel ball was compared with the setting threshold to judge Thethe common defects in steel steel balls include include linear, arc,controlled and cross cross the cracks (Figure 10). The width width of The common defects in linear, arc, and cracks (Figure 10). The of whether steel ball possessed a balls defect(s), which in turn actions of the sorting system, a linear crack is ~0.12 mm and depth ~0.5 mm; the width of an arc crack is ~0.05 mm and depth asuch linear is ~0.12steel mmball and depth ~0.5 mm; width of an arc crack is ~0.05 mm and depth thatcrack a defective was identified and the sorted. ~0.1 mm; mm; the the width width of of aa cross cross crack crack is is ~0.06 ~0.06 mm mm and and depth depth ~0.9 ~0.9 mm. mm. ~0.1

Sensors 2017, 17, 1536

10 of 11

6.2. Results The common defects in steel balls include linear, arc, and cross cracks (Figure 10). The width of a linear crack is ~0.12 mm and depth ~0.5 mm; the width of an arc crack is ~0.05 mm and depth ~0.1 mm; Sensors 2017,the 17, width 1536 of a cross crack is ~0.06 mm and depth ~0.9 mm. 10 of 11 Sensors 2017, 17, 1536

10 of 11

(a) (a)

(b) (b)

(c) (c)

Figure Figure 10. 10. Samples Samples of of steel steel ball ball defects. defects. (a) (a) Linear Linear cracks; cracks; (b) (b) Arc Arc cracks; cracks; (c) (c) Cross Cross cracks. cracks. Figure 10. Samples of steel ball defects. (a) Linear cracks; (b) Arc cracks; (c) Cross cracks.

inspection system ball surface surface quality Using Using the the inspection inspection system presented presented in in this this study study to to inspect inspect the the steel steel ball ball surface quality automatically, the testing experiments were completed in three steps. automatically, the testing experiments were completed in three steps. automatically, the testing experiments were completed in three steps. Step 1: 1: Three Three steel steel balls balls with with three three kinds kinds of of standard standard defects defects and and one one steel steel ball ball without without defect defect Step 1: standard defects and one steel ball without defect were placed placed into into the the inspection inspection system. When the the inspection inspection threshold was set to 3.5 V, the the results results were inspection threshold threshold was was set set to to 3.5 3.5 V, were placed into the inspection system. When the results were obvious differences between defective the ball without showed that there were obvious differences between the defective steel balls and the ball without showed that there were obvious differences between the defective steel balls and the ball without defect (Figure (Figure 11). 11). defect defect (Figure 11).

(a) (a)

(b) (b)

(c) (c)

(d) (d)

Figure 11. 11. The Theinspection inspection voltage voltage curve. curve. (a) (a) The The inspection inspection curve curve of of linear linear cracks; cracks; (b) (b) The The inspection inspection Figure Figure 11. The inspection voltage curve. (a) The inspection curve of linear cracks; (b) The inspection curve arc cracks; The inspection curve of cross cracks; (d) The inspection curve of the steel ball ball curve of arc cracks; (c) The inspection curve of cross cracks; (d) The inspection curve of the curve of arc cracks; (c) The inspection curve of cross cracks; (d) The inspection curve of the steel steel ball without defect. without defect. defect. without

Step 2: 2: Twenty Twenty steel steel balls balls possessing possessing the the three three kinds kinds of of standard standard defects defects were were mixed mixed into into Step 980 qualified nondefective steel balls and then automatically sorted. The inspection results indicated qualified nondefective nondefective steel steel balls balls and and then then automatically automatically sorted. sorted. The inspection results indicated 980 qualified 22 unqualified unqualified balls balls and and 978 978 qualified qualified balls. balls. Then, Then, the 22 unqualified balls balls were were reexamined reexamined 22 unqualified balls and 978 qualified balls. the 22 unqualified reexamined Then, the individually using a high-power microscope, and two of these balls were found to have been been individually using a high-power microscope, and two of these balls were found to have mistaken by the system for defective balls. Under the condition that all balls with defects were mistaken by the system for defective balls. Under the condition that all balls with defects were inspected and and identified, identified, aa small small number number of of qualified qualified steel steel balls balls being being mistaken mistaken for for defective defective steel steel inspected balls is acceptable. balls is acceptable. Step 3: 3: This This inspection inspection system system was was used used to to inspect inspect 46,126 46,126 steel steel balls, balls, randomly randomly selected selected from from Step

Sensors 2017, 17, 1536

11 of 11

individually using a high-power microscope, and two of these balls were found to have been mistaken by the system for defective balls. Under the condition that all balls with defects were inspected and identified, a small number of qualified steel balls being mistaken for defective steel balls is acceptable. Step 3: This inspection system was used to inspect 46,126 steel balls, randomly selected from seven batches of different quality steels. The process took 5.7 h and the inspection rate was 8092 balls/h. These 46,126 balls were then individually reexamined by a high power microscope. The results showed that only 10 defective steel balls were missed—a miss rate of 0.02%. 7. Conclusions In this paper, a new method for inspecting steel ball surface defects based on a CECA sensor was proposed. This method decreased the dimension of the steel ball surface spreading mechanism and reduced the complexity of the unfolding wheel. Experimental results showed that the inspection miss rate was 0.02% and the inspection efficiency was 8092 balls/h. It was shown that this testing system effectively inspected and detected steel ball surface defects completely. Acknowledgments: This work is financially supported by the development of science and technology project of Shandong Province (No. 2013YD09008) and Qingdao postdoctoral application research project. Author Contributions: Huayu Zhang conceived and designed of sensor, electromagnetic simulation and wrote the paper; Fengqin Xie designed mechanism and analyzed the data; Maoyong Cao reviewed article; Mingming Zhong performed experiments. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2.

3. 4. 5. 6. 7.

Xie, F.Q.; Xiao, L.J.; Lu, Y.P.; Zhang, H.Y. Design of the eddy current sensor of steel ball nondestructive inspection. Appl. Mechan. Mater. 2012, 271–272, 842–846. [CrossRef] Technical Description and Instructions for Attendance of Automatic Sorting Machine for Defect-Metric Inspection of Bearing Ball Surfaces; Types AVIKO K-06140 E AND AVIKO-1418 E; SOMET Ltd. of Czech: Bílina, Czech Republic, 1985. Kakimoto, A. Inspection of surface defects on steel ball bearings in production process using a capacitive sensora. Measurement 1996, 17, 51–57. [CrossRef] Ng, T.W. Optical inspection of ball bearing defects. Meas. Sci. Technol. 2007, 18, N73–N76. [CrossRef] Fang, J.; Wen, T. A wide linear range eddy current displacement sensor equipped with dual-coil probe applied in the magnetic suspension flywheel. Sensors 2012, 12, 10693–10706. [CrossRef] [PubMed] Wang, H.; Li, W.; Feng, Z. A compact and high-performance eddy-current sensor based on meander-spiral coil. IEEE Trans. Magn. 2015, 51, 1–6. [CrossRef] Editorial Committee of China. Bearing Steel. China Aeronautical Materials Handbook, 2nd ed.; Standards Press of China: Beijing, China, 2002; Volume 1, p. 381. © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).