EDDY CURRENT ARRAY PROBE AND METHOD FOR LIFT-OFF COMPENSATION DURING OPERATION WITHOUT KNOWN LIFT REFERENCES

Abstract

The invention provides a method for compensating the sensitivity variations induced by lift-off variations for an eddy current array probe. The invention uses the eddy current array probe coils in two separate ways to produce a first set of detection channels and a second set of lift-off measurement channels without the need to add coils dedicated to the lift-off measurement operation. Another aspect of the invention provides an improved calibration process which combines the detection and lift-off measurement channel calibration on a simple calibration block including a reference defect without the need of a pre-defined lift-off condition.

Claims

1. An Eddy Current (EC) system for detecting one or more flaws in a test object, the system comprising: an EC array probe configured with a sensor arrangement, the sensor arrangement including: a plurality of orthogonal sensors arranged to produce respective orthogonal sensitive areas and configured to induce eddy currents in the test object and to sense and output orthogonal signals representative of one or more flaws in the test object; a plurality of absolute EC sensors arranged to produce respective absolute sensitive areas and configured to output a respective absolute vector length representative of a lift-off distance of the orthogonal sensors relative to the test object; and a processor configured to: acquire a plurality of orthogonal signals from an orthogonal sensor of the plurality of orthogonal sensors and a plurality of absolute vector length signals from an absolute EC sensor of the plurality of absolute EC sensors during a scan conducted on the one or more flaws using the EC array probe; and generate a plurality of lift-off compensated orthogonal channels readings based on the plurality of orthogonal signals and on the plurality of absolute vector length signals.

2. The system of claim 1, wherein the processor is further configured to generate the lift-off compensated orthogonal channel reading based on the orthogonal signal by dividing the orthogonal signal by the absolute vector length signal to generate an intermediate result.

3. The system of claim 2, wherein the processor is further configured to generate the lift-off compensated orthogonal channel reading by multiplying the intermediate result by a corresponding reference absolute vector length.

4. The system of claim 3, wherein the processor is further configured to apply the gain and phase calibration values to the lift-off compensated orthogonal channel reading to yield calibrated orthogonal data.

5. The system of claim 1, further including: a computer memory configured to store a setup table comprising corresponding gain and phase calibration values for each of the orthogonal sensors with a corresponding reference absolute vector length for each of the corresponding absolute EC sensors.

6. The system of claim 5, wherein the corresponding gain and phase calibration values are obtained based on calibration signals acquired when the EC array probe is used to scan a calibration notch during a calibration process.

7. The system of claim 1, wherein a ratio of an orthogonal signal of the plurality of orthogonal signals to a respective absolute vector length signal is independent of the lift-off distance.

8. The system of claim 1, wherein the EC array probe is provided on a printed circuit board comprising overlapping coils, each coil configurable as either a driver coil or a receiver coil.

9. The system of claim 1, wherein each of the plurality of EC sensors has an arrangement of at least one driver coil and at least one receiver coil.

10. The system of claim 1, wherein the plurality of orthogonal sensitive areas extends along a first line.

11. The system of claim 10, wherein the absolute sensitive areas are located not to be in line with test object cracks having a crack line parallel or perpendicular to the first line.

12. The system of claim 10, wherein at least two neighboring absolute sensitive areas are adjacent to each of the orthogonal sensitive areas and at least one the neighboring absolute sensitive areas is located on either side of each one of the orthogonal sensitive areas.

13. A method for detecting one or more flaws in a test object using an Eddy Current (EC) system, the method comprising: providing an EC array probe configured with a sensor arrangement, the sensor arrangement comprising: a plurality of orthogonal sensors arranged to produce respective orthogonal sensitive areas and configured to induce eddy currents in the test object and to sense and output orthogonal signals representative of one or more flaws in the test object; a plurality of absolute EC sensors arranged to produce respective absolute sensitive areas and configured to output a respective absolute vector length representative of a lift-off distance of the orthogonal sensors relative to the test object; acquiring a plurality of orthogonal signals from an orthogonal sensor of the plurality of orthogonal sensors and a plurality of absolute vector length signals from an absolute EC sensor of the plurality of absolute EC sensors during a scan conducted on the one or more flaws using the EC array probe; and generating a plurality of lift-off compensated orthogonal channels readings based on the plurality of orthogonal signals and on the plurality of absolute vector length signals.

14. The method of claim 13, comprising generating the lift-off compensated orthogonal channel reading based on the orthogonal signal by dividing the orthogonal signal by the absolute vector length signal to generate an intermediate result.

15. The method of claim 14, comprising generating the lift-off compensated orthogonal channel reading by multiplying the intermediate result by a corresponding reference absolute vector length.

16. The method of claim 14, comprising applying the gain and phase calibration values to the lift-off compensated orthogonal channel reading to yield calibrated orthogonal data.

17. The method of claim 13, comprising: storing a setup table comprising corresponding gain and phase calibration values for each of the orthogonal sensors with a corresponding reference absolute vector length for each of the corresponding absolute EC sensors.

18. The method of claim 17, comprising obtaining the corresponding gain and phase calibration values based on calibration signals acquired when the EC array probe is used to scan a calibration notch during a calibration process.

19. The method of claim 13, wherein a ratio of an orthogonal signal of the plurality of orthogonal signals to a respective absolute vector length signal is independent of the lift-off distance.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0031] FIG. 1 illustrates a simplified representation of the prior art flat shaped orthogonal eddy current array probe built on a two layer printed circuit board

[0032] FIG. 2 is the four layer printed circuit board extension of the FIG. 1 representation and the illustration of the possible shape of the orthogonal channels

[0033] FIG. 3 illustrates the absolute channels created for the probe structure of FIG. 2

[0034] FIG. 4 shows that cracks affecting the orthogonal channels will not affect the corresponding absolute channel.

[0035] FIG. 5 shows the absolute and orthogonal sensitive area on the four layer probe of FIG. 2.

[0036] FIG. 6 illustrates the scan of a calibration block including a reference notch and a probe lift-off.

[0037] FIG. 7 is the impedance plane results obtain on the orthogonal channels from the scan of FIG. 6 block with four defined lift off condition.

[0038] FIG. 8 is the impedance plane results obtain on the absolute channels from the scan of FIG. 6 block with four defined lift off condition.

[0039] FIG. 9 proposes an amplitude based analysis of the test signals shown on FIG. 7 and FIG. 8.

[0040] FIG. 10 is a flowchart describing the proposed calibration method.

[0041] FIG. 11 is a flowchart describing the proposed signal processing method.

[0042] FIG. 12 shows a hardware configuration for a system of the invention.

DETAILED DESCRIPTION OF THE INVENTION

[0043] Related patent application Ser. No. 12/832,620 describes how to build an ECA probe on a printed circuit board. The contents of patent application Ser. No. 12/832,620 and of Ser. No. 12/847,074 are incorporated by reference herein. The structure presented in application Ser. No. 12/832,620 is disposed on two PCB layers. A simplified representation of such structure is shown on FIG. 1 for a four sensor orthogonal ECA probe 1 including five driver coils (2a to 2e) and five receiver coils (3a to 3e) resulting in four sensitive spots (4a to 4d) with orthogonal sensor response. The coil arrangement (such as 5 and 6) that generates orthogonal sensor response will be referenced to as orthogonal channels in this document.

[0044] As first stated in paragraph [0035] of the mentioned application, it is also possible to use the multi-layer capabilities of the printed circuit boards to increase the resolution of the orthogonal ECA probe. FIG. 2 illustrates a four layer version of the ECA probe 10 built using this principle. The two bottom layers (11a to 11e and 12a to 12e) are connected to driver signals while the two upper layers (13a to 13e and 14a to 14e) are connected to receiver signals in order to generate orthogonal channels such as 16, 17, 18 and 19. The bottom layer coils 11a to 11e operate with the top layer coils 14a to 14e to generate a first set of orthogonal channels while coils 12a to 12e operate with the coils 13a to 13e to generate a second set of orthogonal channels.

[0045] From the probe structure shown on FIG. 2, it is an object of the current invention to teach how to obtain absolute channels for monitoring the lift-off. FIG. 3 illustrates how these channels are built from the structure of probe 10. Some coils, partially overlapping the driver coils, are connected as receivers. For example, pitch-catch sensor configuration 36 uses two fourth layer coils (11b and 11c) as a driver and one second layer coil (13d) as receiver. In another example, pitch-catch configuration 33 uses two third layer coils (12b and 12c) as a driver and one fourth layer coil (11c) as a receiver. As a matter of fact, the same coil can be used as a driver and a receiver through the sequence (as already disclosed in prior art patent U.S. Pat. No. 6,344,739). Using coil combinations similar to 31, 32, 33, 34, 35 and 36 for the whole probe 10, we obtain a set of nine sensitive areas (30a to 30i) with absolute sensor response extending over the whole probe length. The coil arrangements (such as 31 to 36) that generate absolute sensor response will be referred to herein as absolute channels in this disclosure.

[0046] The newly created absolute channels are inherently very sensitive to lift off, because the proximity of the inspected part will directly impact the magnetic field flux in the shared area of the driver and receiver coils (11c and 13d for example) defining the sensitive area (30f for example) of the absolute channel (36 for example). FIG. 4 also demonstrates that the new absolute channel will not be sensitive to a longitudinal 41 or transversal 40 crack to be detected by the orthogonal channel because the absolute channel sensitive area (30f for example) is not in line with longitudinal or transverse crack when this crack is located on the orthogonal channel sensitive area (15d for example). So, the described structure makes it possible to substantially completely decouple the lift-off and crack measurements for this probe.

[0047] As further demonstrated on FIG. 5, each orthogonal channel sensitive area is located exactly in between two absolute channels sensitive areas on the index axis. So, we can use the average value between these two absolute channels to produce an approximation of the lift-off conditions for the corresponding orthogonal channel.

[0048] It must be understood that the selection of coils to be used in the absolute channel construction was made in order to acquire the orthogonal and absolute channels simultaneously and with a pitch-catch type configuration which is naturally more stable than an impedance bridge. For example, orthogonal channel 16 and absolute channel 36 use the same set of two driver coils 11b and 11c. So, these two channels can be acquired simultaneously by the acquisition electronics. This configuration is advantageous because it allows a faster acquisition (through simultaneous operation) and a stable signal, but it is not a mandatory requirement so there will be other possible arrangements respecting the essence of the invention.

[0049] Connecting the driver coils as part of an impedance bridge to build the absolute channels, for example, is another method to obtain a valid set of absolute channels for lift-off monitoring without adding new coils in the probe structure. It is also possible to envision other ECA probe types respecting the scope of this invention. For example, in U.S. Pat. No. 5,371,461 FIG. 3, one could dispose of the compensation coil 52 of U.S. Pat. No. 5,371,461 by connecting driver coil 42 of said patent through an impedance bridge.

[0050] Now that we have described means for building channels for detection (orthogonal channels in the preferred embodiment) and lift-off monitoring (absolute channels made out from a pitch-catch sensor arrangement in the preferred embodiment), we describe how these signals are processed in order to obtain a lift-off compensated eddy current probe array without the use of a lift-off reference.

[0051] As shown on FIG. 6, a reference block 50 comprising a long transversal reference notch 51 is scanned in direction 53 with a given lift-off 52. The probe is first nulled in AIR to generate a reference point for an infinite lift off condition, which will become important later in this discussion. The block 50 is then scanned four times with increasing lift off (in the example; Lift-off A=0 mm; Lift-off B=0.63 mm; Lift-off C=1.27 mm; Lift-off D=1.9 mm) to provide the required background information needed to describe the invention.

[0052] FIG. 7 shows the impact of lift-off on the reference defect detection amplitude on orthogonal channel 16 impedance plane display. In this case, defect amplitude 55 is obtained with lift-off A, defect amplitude 56 is obtained with lift-Off B, defect amplitude 57 is obtained with lift-off C and defect amplitude 58 is obtained with lift-off D.

[0053] FIG. 8 shows the impact of lift-off on absolute channel 36 impedance plane display. In this case, the total signal amplitude vector 60 results from lift-off A, the total signal amplitude vector 61 results from lift-off B, the total signal amplitude vector 62 results from lift-off C and the total signal amplitude vector 63 results from lift-off D. It is interesting to note that reference notch 51 generates very weak signals on FIG. 7 compared to the strong lift-off signal. For example, with lift A, defect amplitude 64 is orders of magnitude lower than the corresponding total signal amplitude vector 60 resulting from lift A. This is a desirable behavior since we want to use the absolute channels for lift-off monitoring only.

[0054] FIG. 9 shows a graph representing a combined view of the defect amplitude readings 55, 56, 57, 58 on the orthogonal channels and the total signal amplitude vector readings 60, 61, 62 and 63 on the absolute channels relative to the lift-off conditions. As seen in the figure, both data series can be fitted by exponential curves 70 and 71. Moreover, the shape of curves 70 and 71 (which is defined by the exponent) is almost the same (e.sup.0.6322*Lift vs. e.sup.0.6557*Lift in this example). This observation is very important because it means the ratio Ortho_Amplitude(Lift)/Abs_Vector(Lift) is almost independent of the lift. For example: 0.3834*e.sup.0.6557*Lift/2.3521 e.sup.0.6322*Lift=0.163*e.sup.0.0235*Lift . . . which is about 0.2 dB/mm variation compared to about 5.7 dB/mm for the orthogonal channel. This later observation forms the foundation of the signal processing method of the invention. For the following discussion we will approximate Ortho_Amplitude(Lift)/Abs_Vector(Lift) as being a constant, pre-determined value totally independent of lift-off. It must be understood that the use of the same coil set for defect detection and lift-off monitoring contributes to having similar shaped curves, since the shape of the curve is provided by the magnetic coupling between coils and the inspected part. Thus, by dynamically comparing the orthogonal and absolute amplitudes at each measured point (channel), the orthogonal amplitude can be connected for the actual prevailing lift-off during each measurement, without specific knowledge of the lift-off amount, per se.

[0055] We now turn our attention to FIG. 10, which describes how the probe is to be calibrated with a reference notch such as 51 but without a known reference lift-off. We first NULL the probe in AIR, (Step 102) start the acquisition (Step 104), scan the reference notch (Step 106) and define the notch position by manually or automatically indicating where the notch signal begins and ends (Step 108). At this point, the information we have is equivalent to the signals presented in FIG. 8 and FIG. 7 but for a single unknown lift-off. In other words, if notch 51 is scanned in the calibration process with lift-off B, the system should be able to read defect amplitude 56 and total amplitude vector 61 but the actual value of lift-off B will be unknown to both the acquisition system and the user.

[0056] The information available at this point is first used to calibrate the orthogonal channels by applying a calibration GAIN and ROTATION on the raw signal (Step 1010), in order to reach a pre-defined value for the reference defect 51. This pre-defined value (which typically includes both an angular and amplitude target) is common to all orthogonal channels and thus makes it possible to obtain a uniform detection of the reference defect 51 for all orthogonal channels. The calibration GAIN and ROTATION for each orthogonal channel is saved in the setup (Step 1012).

[0057] Simultaneously, we use the information generated in [0046] on the absolute channels to calculate the vector length between AIR and the signal's baseline obtained on the calibration block 50 (Step 1014). A single absolute vector length value (which could in fact be the average between two absolute channels or other absolute channel combinations adapted to the probe and application) is saved in the setup and associated with its corresponding orthogonal channel. For example, in probe 10, if we use absolute channels at position 30a and 30b to compensate the lift-off for the orthogonal channel at position 15a we could average absolute channels at position 30a and 30b and save this pre-determined value in the setup with reference to the channel at position 15a. This value will be referenced here as Absolute_RefLenght(n,Cal_Lift) where n is the orthogonal channel #identifier and Cal_Lift is the lift-off condition present during calibration (Step 1016).

[0058] Now looking at FIG. 11, we will use the information now included in the calibration file and the properties of the Ortho_Amplitude(Lift)/Abs_Vector(Lift) ratio to generate a lift compensated orthogonal channel. The process described on FIG. 11 is applied dynamically (during the acquisition) but could easily be applied in post-processing as well (after the acquisition). The first step of the process is, again, to NULL the probe in AIR (Step 112) in order to have an infinite lift-off reference. After starting the acquisition (Step 114), each new data set corresponding to the impedance plane results (x,y) for one given orthogonal channel at one given scan position is processed separately (Step 116). Such data set will be referenced here as Ortho_raw(n, Lift) where n is the orthogonal channel #identifier and Lift is the lift-off condition at the time of measuring the data set. The first step in processing is to find the absolute channel total vector length, at the current scan position, corresponding to the orthogonal channel currently being processed (Step 118). The relationship between the orthogonal and absolute channel must be the same as previously defined in calibration. This value will be referenced here as Absolute_VLenght(n, Lift) where n is the orthogonal channel #identifier and Lift is the lift-off condition at the time of measuring the data set.

[0059] Ortho_raw(n,Lift) is then processed with the following relationship to generate a lift-off compensated orthogonal channel reading; Ortho_compensated(n, Cal_Lift)=(Ortho_raw(n,Lift)/Absolute_Vlenght(n,Lift))*Absolute_RefLenght(n) (Step 1110). The generated Ortho_compensated(n,Cal_Lift) channel is then relatively independent of the current lift-off but is then dependent on the lift-off present during the system calibration. To remove this dependency and thus provide a completely lift off independent reading, the calibration GAIN and PHASE are applied to Ortho_compensated(n,Cal_Lift) (Step 1112), until all channels are so processed (Steps 1114, 1116 and 1118). As an end result, for a given flaw size, the system should generate a uniform defect signal amplitude no matter which orthogonal channel detects the flaw and without regard to the calibration and inspection lift-off.

[0060] FIG. 12 shows a hardware configuration of a typical system that can implement the foregoing method. The subject EC probe array system comprises a processor 122 or acquisition unit which is operable and controlled through a user interface 124 and which can display test results, commands and the like, display 126. Orthogonal sensors 128, as well as absolute sensors or coils 129 interact, electromagnetically, with the test object 50 to obtain the various signals and to implement the methods described above via software program instructions stored or loaded onto the processor 122, in a manner well known in the art.

[0061] It is important to point out that the described lift compensation method can easily be adapted to operate a multi-frequency inspection. This can be done either by generating absolute and orthogonal channels for each frequency or by using a unique set of absolute channels to compensate the multi-frequency orthogonal channels.

[0062] It is also important to mention that while the figures and description describes an ECA probe with eight orthogonal sensors, the method proposed in this invention is applicable as long as the coil configuration makes it possible to build at least one sensor for defect detection and one sensor for lift-off measurement.

[0063] In the foregoing embodiments, the EC sensors have been described and depicted as being coil windings. However, as will be recognized by one of skill in the art, other types of magnetic field sensors can be used, such as, for example, GMR (Giant Magneto Resistance), AMR (Anisotropic Magneto Resistance), or Hall Effect sensors.

[0064] Although the present invention has been described in relation to particular embodiments thereof, many other variations and modifications and other uses will become apparent to those skilled in the art. It is preferred, therefore, that the present invention be limited not by the specific disclosure herein, but only by the appended claims.