Introduction
A method for extracting information from pulsed eddy current tests is described. A personal computer is used to analyze and identify conditions in the test specimen that affect the eddy current field flowing within it. The computer analyzes the results in real time and presents them on the screen and/or printer. The computer is easily trained to recognize those conditions in the test specimen that are of interest to the operator. It also compensates for liftoff or other nuisance conditions that are not of interest. If conditions are encountered in which the computer has not been trained, it identifies the known condition closest to the unknown condition that has been encountered.

 

A Brief History of the Use of Pulsed Eddy Currents as a Method of Nondestructive Testing
Pulsed eddy currents were first used by Donald Waidelich of the University of Missouri early in the 1950s. They were subsequently used for nondestructive testing in the then fledgling nuclear industry, especially in the testing of fuel elements, pin jackets and heat exchanger tubing for various research and experimental reactors that were under construction at that time. The method was also used in the analysis of experimental fuel element performance after irradiation, with the tests being performed on the highly radioactive fuel elements in hot cells. At that time various papers appeared in the literature (Renken and Sather, 1972; Waidelich and Huang, 1972). Since then, the method has been used for many applications all over the world and the bibliography of all the papers that have been published on the subject is very extensive (AEA National NDT Centre).

 

---

The use of a pulse produces various advantages over the sinusoidal waveform and some disadvantages.

---

Characteristics of Pulsed Eddy Current Testing
Pulsed eddy current testing differs from sinusoidal eddy current testing in that a pulse of current is induced in the metal test specimen instead of the sinusoidally varying current used in single frequency or multiple frequency eddy current testing. The use of a pulse produces various advantages over the sinusoidal waveform and some disadvantages. In theory, a pulse can be analyzed into an infinite train of sinusoidal waveforms that are related harmonically. This means that all the information that could be obtained from a large number of sinusoidal tests at various frequencies is already available to the pulsed eddy
current test equipment. Since the pulse usually has a short duration compared to the duration of the duty cycle, the pulse can be relatively powerful without excessively heating the test probe. On the other hand, wideband electronics are needed to handle the pulse. This wideband circuitry is more susceptible to electronic noise than the narrow band circuitry that can be used by sinusoidal eddy current equipment.

As in all eddy current tests, a current field is induced in the specimen that is to be tested, usually by a current carrying coil called the field coil. The field in the test specimen is detected by a current induced in another coil called the pickup coil. Sometimes a Hall detector is used for this purpose. Often equipment is designed so that one coil acts both as field coil and pickup coil. In all cases, changes in the field in the test specimen caused by physical or conductivity variations in the test specimen are what the eddy current equipment must then process and present as useful information to the operator. In far effect eddy current tests, the pickup coil is purposely located a significant distance from the field coil, but the equipment described here uses a pickup coil very close to the field coil.

The pulsed field that the field coil induces in the test specimen travels relatively slowly into the metal test specimen. We are not dealing with a wave phenomenon here like that which occurs in an ultrasonic test. Rather, the diffusion of the current in the metal is much more analogous to the diffusion of heat into a conducting medium. In fact, the same equations can be used to describe both actions. The metal acts as a dissipative medium, weakening the current field rapidly and changing its shape as a function of time. This makes a rigorous definition of the velocity of the pulse in the specimen difficult. The diffusion of current is much faster than the diffusion of heat in the same medium, however.

 

Equipment Description
The equipment described here uses a separate field coil and pickup coil that are located immediately adjacent to each other. The voltage that drives the field coil as it appears on an oscilloscope is shown in Figure 1a. The current in the field coil has practically the same shape since the impedance of the field coil at all frequencies of interest is very low. In this case, the test probe was of the type intended to test tubes from the inside as is done in tests on the condensers used in steam power plants, or on the steam generators used in pressurized water reactors. The voltage appearing across the pickup is shown in Figure 1b when the test probe is located outside a tube. This voltage is reduced to a minimal level by the use of an auxiliary coil which bucks out the strong direct field from the field coil. The voltage across the pickup when the test probe is inside a tube is shown in Figure 1c. It is essentially the first derivative of the flux outside the tube that is generated by the pulsed current flowing in the metal. In this case, both field coil and pickup consist of a few turns of fine wire coaxial to each other and immediately adjacent. The combined length of the whole assembly is about 1.5 mm (0.06 in.).

Figure 1 — Oscilloscope traces that are observed when a test probe designed to test tubes from the inside is used: (a) voltage across the field coil; (b) probe outside the tube; (c) probe inside the tube. The dotted line is obtained when the wall thickness is 1.78 mm (0.07 in.). The vertical scale is 0.2 V per division. The horizontal scale is 20 ms per division. The tube has an inner diameter of 12.7 mm (0.5 in.) and a wall thickness of 0.89 mm (0.04 in.).

Time Sampling
Information theory teaches that a pulse may be reproduced to an arbitrarily high accuracy if its amplitude is known at a sufficient number of points in time during the duration of the pulse cycle. This means that if one can measure the pulse amplitude at various points along the pulse, one can obtain essentially all the information that the pulse may be carrying provided that the amplitudes at the various sampling points can be interpreted. The basic approach used with this equipment is to obtain the amplitudes of the field coil pulse at a sufficient number of sampling points and program a personal computer to interpret the results. The amplitude of the pickup pulse at various sampling points is obtained by the use of sampling pulses. These are high amplitude, very narrow pulses of 0.05 ms that are superimposed on the pickup pulse at various points. The combined amplitudes of sampling pulse and pickup pulse at a particular time are then measured by the electronic circuitry and presented as a filtered analog voltage. This equipment samples at five different points and so it produces five outputs which may be called sample voltages.

 

Processing of the Sample Voltages
The five analog voltages must first be converted to digital form before they can be used by the computer. The conversion is accomplished by a multiplexing analog to digital converter which samples each of the five output signals in turn and converts each into a digital number with a resolution of 8 bits. The sampling is very rapid compared to the pulse repetition rate, so in effect the computer always has available the latest information developed by the driving pulse. The five digital signals are fed to the computer via the parallel port. It is the fastest port on most personal computers.

 

Interpretation of the Signal Voltages
The set of five signal voltages this equipment produces, which contain information derived from the voltage pulse developed across the pickup, may be regarded as a number set, or as a pattern in a five dimensional space. The later concept was used in developing the program that interprets the patterns. To the extent that the patterns are located in different locations in the pattern space the computer is able to distinguish between them. To make this concept clearer, it is helpful to closely inspect Table 1. Table 1 shows the values of the five signals developed by various conditions in the test specimen, in this case, a copper tube of 12.8 mm (0.5 in.) inner diameter and 0.89 mm (0.04 in.) wall. The amplitudes are the digital values produced by the pattern recognition program. The times are the sample points of the pulse developed across the pickup coil. The top entry is labeled as the standard. A standard plays no important role in the test. It is merely a location exhibiting no noticeable anomalies where the liftoff compensation can be developed. In this case, liftoff was introduced merely by tilting the loosely fitting test probe into an atypical position within the tube. At this point one of the advantages of the pulsed eddy current method should be pointed out. If the driving pulse has a fast risetime and the equipment bandwidth is sufficient to preserve the steep leading edge of the pickup pulse, the sample point at time = 0 is essentially sensitive only to liftoff and sample conductivity, and is relatively insensitive to the latter. This fortunate circumstance allows the computer program to compensate for liftoff automatically. The reason for these phenomena is that the steep leading edge of the pickup pulse waveform is developed mainly by the high frequency components of the current in the specimen. These components stay very close to the surface and are reflected back to the pickup immediately. In practice, one has only to introduce some reasonable value of liftoff, press a key on the computer and the liftoff compensation is then introduced automatically in subsequent tests. The entries in Table 1 after entry two are all liftoff compensated.

 

Training the Computer
The computer is trained to interpret the patterns by the operator. The test probe is positioned over some tube condition of interest and the description of the condition is typed into the computer. From then on as it runs the pattern recognition program the computer will recognize the condition when it is encountered and the description will appear on the computer screen. It continually analyzes the patterns that are presented at its input and attempts to match them to the patterns stored in the memory. If it cannot achieve a match, it presents the closest available pattern and the distance to it in pattern space. If the match is sufficiently poor, it identifies the pattern as an unknown pattern. Table 1 shows a simple example of what a pattern file looks like. Note that the conditions labeled "headers" are not actual headers in a condenser or steam generator but are simulations achieved by fitting a narrow, close fitting, thick copper sleeve over the outside of the tube.

Table 2 shows the pattern file for another type of specimen. The test probe in this case was a point probe intended for the testing of flat or nearly flat plates. The specimens are intended to simulate aircraft skin. They consist of various aluminum plates of varying thickness. Some of them are sandwiched together and secured by rivets.

Table 1 The set of five signal voltage amplitudes developed by various conditions in a copper tube of 12.8 mm (0.5 in.) inner diameter and 0.89 mm (0.04 in.) wall
Condition in the Tube	Time (µs)
	0	45	95	120	160
Standard	47	49	51	39	39
Liftoff	70	42	79	30	41
Typical tube	47	49	47	40	39
Header, edge one  Header, edge two 47 55 46 2 57	47	21	60	133	42
Header, center	47	36	55	65	51
0.3 mm (0.01 in.) through hole	47	39	51	41	38
Header center, soldered	47	39	51	60	57
1.6 mm (0.06 in.) wide 1.3 mm (0.05 in.) groove on the outside surface	47	39	50	38	34

 

Table 2 The set of five signal voltage amplitudes developed by various conditions in simulated aircraft skin consisting of various aluminum plates of varying thickness, some sandwiched together and secured by rivets
Condition in the Plates	Time (µs)
	0	45	95	120	160
Standard 1.3 mm (0.05 in.) aluminum plate	115	107	124	101	117
Liftoff (0.25 mm [0.01 in.])	105	104	111	93	106
Two 1.6 mm (0.06 in.) plates sandwiched together	115	95	121	117	85
1.3 mm (0.05 in.) and 0.6 mm (0.02 in.) plates sandwiched together	115	107	124	101	117
0.9 mm (0.04 in.) aluminum plate	115	108	121	55	138
Two 0.8 mm (0.03 in.) plates sandwiched together	115	108	128	109	122
Corrosion between 1.6 mm (0.06 in.) plates	115	99	122	123	91
Two 1.6 mm (0.06 in.) plates sandwiched together with crack in the lower plate	115	94	121	125	83
Header center, soldered	47	39	51	60	57
1.6 mm (0.06 in.) wide 1.3 mm (0.05 in.) groove on the outside surface	47	39	50	38	34

 

 

Conclusion
The information developed by the pulsed eddy current method allows the application of automatic pattern recognition as performed by a personal computer. The equipment is relatively simple, stable and adaptable to a wide variety of test probes and specimens.

 

Acknowledgments
The author would like to thank Dean Froerer and Terry Vallon of Boeing, for providing some of the specimens used in the research of this paper.

 

References
AEA National NDT Centre, NDT Abstracts, Oxon, England, AEA.

Renken, C.J. and Allen Sather, "A Pulsed Eddy Current Test System for Hot Cell Use: Manual of Operation," ANL 7973, Argonne, Illinois, Argonne National Laboratory, 1972.

Waidelich, D.L. and S.C. Huang, "The Use of Crossing Points in Pulsed Eddy Current Testing," Materials Evaluation, Vol. 30, No. 1, January 1972, pp. 20-24.

 

* 36017 Highway T, California, MO 65018; e-mail <redbarn@socket.net>.

 

Copyright © 2001 by the American Society for Nondestructive Testing, Inc. All rights reserved.

IRRSP, NDT Handbook, The NDT Technician and www.asnt.org are trademarks of the American Society for Nondestructive Testing, Inc. ACCP, ASNT, Level III Study Guide, Materials Evaluation, Nondestructive Testing Handbook, Research in Nondestructive Evaluation and RNDE are registered trademarks of the American Society for Nondestructive Testing, Inc. ASNT exists to create a safer world by promoting the profession and technologies of nondestructive testing.