1. 電磁撿測實驗
所謂非破壞性檢測乃是以電磁，超音波等方法來檢查物質的瑕疵或測量產品的品質，這些方法不會破壞受檢測的物體，故稱之為非破壞性檢測

2. 外加磁場、渦電流、與渦電流產生的磁場

3. 雙D激發線圈

4. Conventional Eddy-current NDE
Exciter / Receiver Electronics
Exciter / Receiver Coil
Magnetic core
Scan
Phase shifts
Conducting specimen
Let’s first see the basic concepts of the eddy-current flaw detection. The conducting specimen with a discontinuity is the object to be test. The exciter coil wrapping the magnetic core can generate an alternating magnetic field in mT range, and this alternating magnetic field can induce eddy-currents in the conducting specimen. When the exciter coil comes across a discontinuity, the signal on the monitor screen can show the change in the coil impedance. Typically the sensitivity of the system can detect the defect signal in the T range.
Discontinuity
The eddy current in the specimen is induced by the excitation coil and the electronics.
The magnetic field generated by the eddy current is then detected by the receiver coil and the electronics.
Defect signal ~ T
Excitation field ~ mT

5. 以雙D型激發線及磁場sensor，做掃描的實驗。

8. Disadvantage of conventional eddy-current NDE
• Poor resolution at low frequencies  Not suitable for detecting deep flaws.
• Large in sensor size  Limits the spatial resolution;  Interaction between the sensor and specimen
is not negligible.
The method has some limitation in the flaw detection. Fist of all is the limit poor resolution when the excitation frequency is low. This limits the method to the detection of the surface flaw. It is not possible to use it for the deep lying flaws. The second challenge is the size of the sensor limits the spatial resolution. Moreover, the interaction between the sensing probe and the specimen affects the defect signal because of the size of the probe. The third challenge is the frequency dependent transfer function dV/dB and the hysteretic behavior of the magnetic core.
These characteristics result in complexity in the quantitative analysis of the defect signals.

9. High-sensitivity Eddy-current NDE
NDE Electronics
Gradiometer
Scan
Excitation Coil
Defect signal
Conducting specimen
Now, let’s turn to the eddy-current probes with a SQUID gradiometer. The major difference between the SQUID eddy-current probe and the traditional eddy-current probe is the SQUID senor, which doesn’t need a magnetic core to amplify the excitation field. The typical magnitude of the excitation field is several miro-tesla, and the resultant defect signal is typically several nT down to pT.
Discontinuity
Defect signal ~ nT
Excitation field ~ T
Defect signal ~ T
Excitation field ~ mT
Conventional

10. Merits of SQUID-based Eddy-current NDE
• High sensitivity in sensing low frequency magnetic
fields
 higher than 1 pT in an unshielded environment
• Large dynamic range at low frequency
 from T to pT
• Small size of the pick-up coil (< 1 cm x 1 cm)
 We can have better spatial resolution.
• Frequency independent transfer function dV/dB.
The some of the problems I just mentioned may be solved by using the SQUID. The merits of the SQUID includes the high sensitivity in sensing low frequency magnetic field, the large dynamic range, and the small size of the pick-up coil, and the broad bandwidth and frequency independent field-to-voltage transfer function dV/dB within the bandwidth limit. Moreover, the dV/dB of the SQUID is very linear for any strength of magnetic field.

11. • Design of three NDE Systems
Single Magnetometer Z
Double-D excitation coil
Axial Electronic Gradiometer
Reference sensor
Sensing
Sensor X
Planar Gradiometer
µ-metal shield
Signal-shift compensation X X
Here shows three kinds of the SQUID-based eddy-current probe. The design rule for the SQUID arrangement is that it can survive in the noisy environment. And the rule for the excitation coil is that it may not induce a strong magnetic field in the SQUID. To achieve these two features, we designed the magnetometer NDE probe with a small µ-metal shield and a double-D coil. Another shown in the middle is an axial electronic gradiometer NDE probe with a double-D coil. The one on the right is a planar gradiometer NDE probe with a circular excitation coil. For the case with a sensing magnetometer, a differential exciter coil is preferred. While for the planar gradiometer, a circular coil is suitable.
Both cases are investigated experimentally in this work. Z Z
Circular excitation coil
Double-D excitation coil

12. Quantitative NDE

13. NDE System with a single magnetometer eddy-current probe
Electronics
Output
Signal
Generator
An additional signal generator for signal-shift compensation is required.
Imbalance Compensation
Clock
Signal
Generator
Lock-in
Amplifier
Shielded Magnetometer
Data Acquisition
Reference
Excitation
The NDE system with an shielded magnetometer is shown here. The signal generator is used for the excitation source, and the lock-in amplifier is used to analyze the phase and the magnitude of the gradiometer output.
The additional signal generator here is to make compensation for the additional signal shift caused by the mis-alignment of the magnetic shielding.
Sample
Computer
X-Y stage
RS-232
Stepping motor
controller

14. NDE System with an axial electronic gradiometer eddy-current probe
Gradiometer output
Electronics
With two-channel electronics and one difference electronics.
A - B
Electronics B
Signal
Generator
Lock-in
Amplifier
Axial Gradiometer
Data Acquisition
Reference A
Excitation
The NDE system with an axial gradiometer is shown here. The function generator is used for the excitation source, and the lock-in amplifier is used to analyze the phase and the magnitude of the gradiometer output.
As the motion of the SQUID gradiometer can result in excess noise, the gradiometer is kept stationary while the sample is moving in this system. Note that there are two SQUID operation electronics for the sensor SQUID A and the reference SQUID B respectively, and a difference electronics to perform the electronic subtraction.
Sample
Computer
X-Y stage
RS-232
X-Y stage
Stepping motor
controller

15. B A A-B t Electronic (Axial) Gradiometer signal to be detected
Reference
Reference
SQUID
Sensor A
A-B
Sensor
Sensor
Let’s take a look on the electronic noise cancellation technique. The frequency independence of dV/dB I just mentioned actually permits the technique called electronic noise cancellation. If here is the source of a magnetic signal like this, and the sensor SQUID senses the signal. But It goes without saying that the SQUID must also feel the magnetic noise in the environment, and the output of the SQUID must be something like this, instead of the wave form of the signal source. To get rid of the environmental noise, an easy way is to use another SQUID for the reference of the noise, and the two SQUIDs now becomes an electronic Gradiometer. Then the output of the gradiometer should be similar to the signal source. t
signal to be detected

16. NDE system with a planar gradiometer Eddy-Current Probe
Gradiometer output
Electronics
With a single-channel electronics
Signal
Generator
Lock-in
Amplifier
Planar Gradiometer
Data Acquisition
Reference
Circular excitation coil
Excitation
This is the NDE system with a planar gradiometer. The planar gradiometer was also stationary while the sample is movable to perform the scan. Note that only a single channel SQUID electronic box is necessary in this system. The other portion of this system is identical to the previous NDE system with an axial gradiometer. As the planar gradiometer probe uses the circular excitation coil, it is more easy to compare the experiment with the numerical results.
Sample
Computer
X-Y stage
RS-232
X-Y stage
Stepping motor
controller

17. Integrated Planar Gradiometer
A-B
Loop B
Loop A
signal to be detected
Here shows the integrated planar SQUID gradiometer. This SQUID has two symmetry superconducting pickup loop connected in parallel on the same chip. So if the signal source is put under the SQUID, one can only see the difference signal of the loop A and loop B. The integrated SQUID gradiometer is more stable in unshielded environment, but the electronic SQUID gradiometer is usually more sensitive than the integrated SQUID gradiometer.
 Integrated Gradiometer is more stable, but
 Electronic Gradiometer is usually more sensitive.

18. • Depth analysis for hidden cracks
Usually , NDE only detects the existence of the crack .
Here we try to locate the depth of the cracks.
Now, let’s see the analysis of the defect signal obtained with the SQUID gradiometer eddy-current probe.

19. Experimental Test of Defect Signals
single magnetometer /
double-D excitation coil X
Artificial flaw
Aluminum plate
Scan
Now, let’s compare the results obtained with the magnetometer probe and the planar gradiometer probe.
We have taken the axial gradiometer as a sensing magnetometer because the upper SQUID for noise reference has no effect on the flaw signal. Therefore, the signal measured by the axial gradiometer is similar to the shielded magnetometer.
Another important thing is that the both kinds of probes are only sensitive to a specific flaw orientation.
The probes are most sensitive as the indicated X axis of the probe is across the crack in specimen.

20. diff Bdiff There are shifts in Bi and Bq. shift
Shifts can be removed after taking the differentiation. dx
dBi
diff
Bdiff
dBq
To see how the differential defect signal works, let’s see the 1-D scan with obvious shifts shown here. When the differentiation is applied, the shift is removed because it is spatially independent. The we calculate the differential phase by using the differential signal dB/dx. The result is a bath-tub shaped curve, and the bottom of the tub indicates the locations of the hidden crack. Note that the phase varies slightly around the crack located at X=0.
Flaw center

21. Differential Defect Signal & Differential Phase
in which  Differential Defect Signal  If V(0) =0, 
Since the response of the SQUID signal V is due to the buried flaw, it is clear that the the differential defect signal dV/dx is also due to the existence of the flaw. Therefore, if the response is virtually zero at the location of the flaw, one may expect the phase  of defect signal is equal to phase of the differential defect signal diff, if the higher order terms in the expansion of V(x) are negligible near to location of the flaw.
Namely, Phase  Differential Phase
*H.-E Horng and J.-T Jeng et al., Physica C 367, (2002)

22. We first found that the differential method can be applied to the NDE technique.

23. diff–z relations at x = 0 for various f
z (mm)
Phase-depth relation is approximately linear.
For example, at 400 Hz :
flaw depth z = 0.037× (diff +47º)
diff /dz (mm)

24. diff–z relations for various flaw sizes
 diff –z relation remains the same if flaw size < skin depth*
* For aluminum, skin depth  = 4 mm at ƒ = 400 Hz 2 4 6 8 10
z (mm) - 1 3
diff (º)
1.0 mm
1.5 mm
2.0 mm
Flaw height
ƒ =400 Hz
0.7 mm
Flaw width

25. Depth analysis of hidden cracks with differential methods
NDE system
f = 400 Hz
crack 1
crack 2
scan
To show the feasibility of the flaw depth analysis, we make a 1-D scan over a stack of aluminum sheets with two hidden cracks. In this case the SQUID is still and the sample is in motion.
On the left is the differential phase obtained at 400 Hz. The bath-tub shaped curves are easily identified, which indicate that there are two distinct cracks in the specimen. The differential phase at the location of the crack gives the depth of the crack if we consider the phase-depth relation at 400 Hz.

26. f = 400 Hz , 1.0 cm excitation coil
The spatial resolution of the system is limited to the size of the excitation coil.
f = 400 Hz , 1.0 cm excitation coil
diff

27. Universal curve of ddiff /dz vs
Universal curve of ddiff /dz vs.  (skin depth) for various conductor specimens
To describe the change of the slope of the phase-depth curve with the frequency, the most impressive way is to plot the ddiff /dz vs.  (skin depth) curve.
In this curve, the change of the slope with the skin depth for aluminum and the graphite is unified into a single curve.
The empirical equation shown here is obtained by a bet fit to the data. The equation will applied to most of the non-magnetic conductors for which the conductivities are between those of aluminum and graphite.

28. Conclusion and Outlook
 We have demonstrated quantitatively with the high-TC SQUID NDE.
 We first found that the differential method can be applied to the NDE technique.
 The depth of the flaw can be determined quantitatively.
 Planar gradiometer is promising for NDE system, because it :  is more stable in an unshielded environment

29. Thank you for your attention!