GB2256713A - Eddy current flaw size detecting probe - Google Patents

Abstract

The probe, for generating a signal indicative of a quantitative measurement of the size of a surface defect, comprises a magnetic field generator 10 for inducing currents in the surface and a perturbation current sensor arrangement 20, 40 positioned relative to the field to give an output providing said quantitative measure. The probe comprises a solid cylindrical former 50 on which is wound the perturbation current sensing coils which are rectangular 20a or figure of eight shaped 40a and have axes radial to the former. The field generator coil 10 has an axis aligned with the axis of the former and is constructed to produce a magnetic field which produces a spatially varied magnetic field in a material under test. <IMAGE>

Description

PROBE This application relates to a probe for detection of flaws in materials, and to apparatus including such a probe. In particular, this invention relates to a probe for detection of cracks in structures; particularly, but not exclusively, for detection of cracks in submerged structures such as oil rigs, or in pipes.

One known non-destructive technique for testing the presence of cracks is described in British Patent GB 2012965B (US 4266185), in which a pair of spaced contacting electrodes are engaged with an electrically conductive surface, and an alternating current is passed through the surface between the electrodes. A further pair of spaced apart pick up electrodes are used to measure a potential difference between two points on the surface. When a crack lies between the two points, the potential difference is higher since the current flows down one side of the crack and up the other; in either case, the distance traversed by the current is greater than the shortest distance between the two points. The voltage drop between the two points therefore provides a measure of the crack dimensions, as well as an indication of the presence of a crack.

A different type.of probe is disclosed in GB 2225115A and corresponding US 5019777. In this proposal, there is provided a probe head which comprises a number of different sensors. The probe head comprises a solenoid coil which is fed with an alternating current, and a number of sensor coils. The solenoid coil induces eddy currents in the surface of a material to be tested, when brought into close proximity thereto.

The sensor coils fall into two types. A first type comprises a coil wound round an axis parallel to the solenoid coil; these are referred to as "absolute' or "lateral" coils. The eddy currents flowing in the surface of the material induce a current in the lateral coils. The system of excitation (solenoid) coil, surface to be measured and lateral coil thus act as, respectively, the primary winding, core and secondary winding of a transformer and the magnitude of the voltage induced across the ends of the lateral coil varies with the resistance of the surface, and hence with the physical dimensions of any crack present. The output of such a coil can therefore provide a quantitative indicator of the depth of the crack in the surface. It may also be employed to measure the "lift-off" or separation between the probe and the surface.However, the large dimensions of the lateral coil make it insensitive to the exact position of the crack, and consequently it is unsuitable on its own for determining the crack location and exact crack length.

For this purpose, a second type of sensor is employed comprising an elongate coil provided approximately coplanar with the surface to be tested, and consequently having an axis normal thereto (and to the axis of the solenoid). This coil is referred to as a "current perturbation coil". The current perturbation coil runs substantially the whole length of the solenoid, with its long axis aligned therewith, and is consequently symmetrically mounted about the centre, lengthwise, of the solenoid.

As will be discussed in greater detail below, the currents flowing in the surface of the material under test induce a current and voltage signal in a coil lying (or having a component lying) in a plane parallel to the surface. In the symmetrically mounted current perturbation coil of prior art, when a symmetrical surface (either including or not including a flaw or crack) lies underneath the whole length of the magnetic field generated by the probe, the net signal output from the current perturbation coil is zero. However, if an end of a crack or flaw is present within the magnetic field of the probe, the current perturbation coil produces an output signal.

Thus, the current perturbation coil produces a signal when either end of a crack or flaw is encountered, but not otherwise.

As will be discussed below, we have shown that the currents induced in the surface of the material rotate about centres underlying the ends of the solenoid, in opposite senses. At the centre of the solenoid, where the two contra-rotating currents meet, the current density is higher than that under the ends of the solenoid and beyond. There is thus a current gradient, having a maximum under the centre of the solenoid falling off towards the ends.

If a crack is present in the surface beneath the solenoid, as discussed above the current will journey down one side of the crack and up the other, and the associated material resistance leads to a voltage drop across the sides of the crack. However, we have realised that since the magnetic field, and hence current, varies along the length of the solenoid, where the crack is aligned with the solenoid axis the voltage drop caused will vary along the length of the crack, rising to a maximum under the centre of the solenoid. This voltage variation along the crack leads to local perturbation currents.

The perturbation coil used in the prior art is arranged symmetrically about the centre of the solenoid, so that the rise in potential difference along the crack underlying one half of the coil is matched by the fall in potential difference along the crack underlying the other half of the coil so that the generated perturbation currents are symmetrical and no net voltage across the coil is thereby generated, except when one end of the crack passes under the sensing coil. When this occurs, the voltage rise in one part of the crack is no longer compensated by a corresponding voltage fall along the other, so that there is a net potential difference along the part of the crack underlying the perturbation coil, and hence a net perturbation current and a voltage is consequently generated in the perturbation coil.

Thus, in use, as the probe is moved over a surface, when one end of a crack is reached a pulse output voltage is produced by the perturbation coil, marking the start of the crack. As the probe moves along the crack, a signal indicating the crack depth is derived from the lateral coil. When the other end of the crack is reached, a further pulse is generated by the perturbation coil. By logging the outputs of the two coils, the start and end points of the crack, and hence its length, together with its depth may be accurately measured.

We have, however, discovered that an equally accurate probe may be provided by making use of the variation of the induced current along the length of a crack, to provide a direct measure of crack depth as well as a sensitive measure of the crack start and end positions. This is possible because the depth of the crack, and hence the resistance between the two sides of the crack, not only affects the voltage drop across the crack but also the magnitude of the voltage gradient and hence potential difference along its length.

Accordingly, in one aspect, the invention provides a probe comprising means for inducing a current in a material surface, said current being arranged to run across the direction of a crack therein, and to exhibit a different value at different points along the said crack, and means for providing a signal corresponding to the potential difference along said crack caused thereby, and for deriving therefrom an indication of the depth of said crack.

In one embodiment, the means for inducing a current comprises means for inducing a current gradient, and this may be a solenoid coil as discussed above. In such an embodiment, the sensing means may comprise a coil approximately coplanar with the surface, positioned asymmetrically with respect to the centre of the length of the solenoid. An embodiment of the invention provides a coil including a twist, so as to form an "8" closed shape, disposed about the longitudinal centre of the solenoid, so as to be effectively antisymmetrical.

In a different aspect of the invention, we provide means for measuring the depth of the crack by sensing the current perturbation in the surface of a material including the crack. The means for measuring the current perturbation may comprise current sensors positioned in portions of a field exhibiting an asymmetrical gradient, for example either an asymmetrically positioned coil or coils, or a figure of "8" wound coil, or alternatively means for measuring the vertical fall off of perturbation current from the surface (for example, at least two sensors positioned at different vertical distances from the surface, for example a pair of coils, connected in opposition one to the other).

In the above aspects, the invention provides a probe which is less sensitive to variations in "lift-off" or separation from the surface under measurement than the "absolute" coil of the prior art, and furthermore may be provided in a more compact assembly since it may be substantially coplanar with the surface under measurement. Some embodiments appear also to exhibit less sensitivity to electrical noise.

A further problem of the prior art is that the measurement of distance or position of a flaw or crack is based upon a time measurement and an assumption of constant probe speed. This assumption is often wrong. Accordingly, in a further aspect, the invention provides a probe array for measurement of surface flaw comprising means for generating a magnetic field or fields extending in at least one dimension and a plurality of sensors at spaced apart positions within the magnetic field or fields, and means for reading said sensors. In one embodiment, the sensors are read in sequence, so as to generate an electrically scanned signal representing the surface in a scan direction.The field generating means may comprise a plurality of separately energisable field generators, for example; preferably, the sensors are perturbation current sensors according to the above embodiments, for example sensors mounted asymmetrically within the magnetic field generated by each generator.

According to a further aspect of the invention, there is provided a probe comprising a magnetic field generator and an induced current sensor, in which the arrangement is such that the probe may be switched to operate in a plurality of different modes, in which the sensor generates signals which are differently related to the presence and/or dimensions of a flaw.

For example, the arrangement may comprise a single sensor and a pair (or more) of magnetic field generators, the arrangement being such that energising different generators or combinations of generators changes the relative position of the sensor to the magnetic field generated; for example, to change the position from lying symmetrically within the magnetic field to lying asymmetrically therewithin. In this way, a single sensor can function either as a conventional "current perturbation" type sensor responsive only to the ends of cracks or as a sensor giving a measurement of crack depth; but these modes of operation are of course not exhaustive. This aspect of the invention is conveniently employed in a probe array according to a preceding embodiment of the invention.

Other aspects and preferred embodiments of the invention are as described or claimed hereafter.

The invention will now be illustrated, by way of example only, with reference to the accompanying drawings in which: Fig. la shows schematically a perspective view of a crack in a plate; Fig. Ib shows schematically a cross-sectional view through the plate across the line of the crack with surface current and voltages shown; Fig. 2 shows schematically a crack detection, measurement and logging system; Fig. 3 shows schematically a known probe forming part of such a system; Fig. 4 shows schematically the method of use of the known probe in examining weld defects; Fig. 5 shows schematically output signals generated by the probe of Fig. 4 in use as it passes over a crack; Fig. 6 shows schematically the elements of the probe according to one embodiment of the invention; Fig. 7 shows schematically output signals produced by the probe of Fig. 6 in a form corresponding to those of Fig. 5;; Fig. 8 shows schematically the alternating magnetic field produced by the probe of Fig. 6 or Fig. 3; Fig. 9a shows the voltage induced in a coil passing through the field of Fig. 8; Fig. 9b shows the integral of the induced voltage of Fig. 9a, corresponding to the magnitude of the magnetic field of Fig. 8; Fig. 10a is an elevation showing the probe of Fig. 6 aligned with the plate of Fig. la; Fig. 10b is a corresponding plan view, omitting the probe, showing the circulating currents induced in the surface by the magnetic field; Fig. 11 is a corresponding graph of the intensity of a current along a line in the plate lying beneath the axis of the probe; Fig. 12 corresponds to Fig. 10a, and shows the probe of Fig. 6 overlying a crack within a plate; Fig. 13a shows schematically the current flowing across the crack along its length beneath the probe;; Fig. 13b shows the corresponding voltage drop across the crack, for varying crack depth; Fig. 13c shows the gradient of the voltage drop of Fig. 13b along the length of the crack, for varying crack depth; Fig. 14 shows the general arrangement of a first embodiment of the invention; and Figs. 15a and 15b are respectively front and end elevations showing this embodiment in more detail; and Fig. l5c shows the application of this embodiment in use; Fig. 16 shows schematically a driving circuit for connection to this embodiment; Fig. 17 shows schematically the general arrangement of a second embodiment of the invention; and Figs. 18a and 18b are front and end elevations respectively showing this embodiment in greater detail; Figs. l9a and l9b are respectively a sectional end elevation and a semi-sectional front elevation showing this embodiment in yet greater detail; and Fig. 20 shows one application of the embodiment of Fig. 19; and Fig. 21 shows schematically a second application of the embodiment of Fig. 19; Figs. 22a and 22b shows schematically sectional views showing alternative connection states of a third embodiment of the invention; and Figs. 23a and 23b show corresponding outputs of that embodiment in response to a detected defect; Fig. 24 shows a schematic section elevation of a fourth embodiment of the invention; Figs. 25a and 25b are respectively front and side elevations of an alternative embodiment to that of Fig. 24; and Fig. 26 shows a detail of Fig. 25; Figs. 27a and 27b are respectively front and end elevations of a fifth embodiment of the invention; Figs. 28a-i shows schematically various alternative geometries according to the invention and the prior art; Figs. 29a-k show corresponding output signals generated in response to traversing a crack; and Figs. 30a-c illustrate a method of signal processing for use with probes according to the invention.

Eddy Current Crack Depth Measurement Referring to Fig. la, a surface breaking crack is shown in a plate. The crack acts as a stress concentrator, and the presence of a crack therefore increases the stress in the uncracked part of the plate (or other structure). A common source of cracks in metal structures is at welded joins, where the welding process can embrittle the welded parts. When a stress greater than a critical level, which is dictated by the dimensions of the crack (generally its depth) is applied to a cracked part, the crack propagates rapidly and failure occurs. Iron and steel structures such as ships and oil rigs, destined for immersion in relatively cold water at which they are brittle, can thus fail catastrophically at welds.

Other typical welded or other metal structures in which failure due to cracking is a problem are, for example, pressure vessels for chemical process plants and pipelines (particularly pressurised pipes).

Non-destructive techniques of crack detection are preferred, for obvious reasons. One such method employs currents induced in the surface of the material.

Referring to Fig. lb, which shows a section through the crack in the plate in Fig. la, in AC crack detection, a current is generated in the surface of the plate (e.g. by inducing eddy currents, as in GB 2225115). Away from the ends of the crack, the current generally flows down one side of the crack and up the other, so that to traverse the distance between the two sides of the crack the current travels the distance 2D (where D is the crack depth). In portions of the plate in which there is no crack, of course, the current need travel only a much smaller distance directly over the surface. The potential drop between points at either side of the crack is, by Ohms law, the current multiplied by the resistance of the current path, which is the surface resistivity R of the plate multiplied by the length of the path. Thus, the potential difference across the crack is 2DRI.

From a knowledge of the material of the plate and of the magnitude of the current, the crack depth D is thus readily derived by measuring the potential difference across the crack.

It is desirable to test welded structures at periodic intervals, as new cracks can be generated by impacts, and cracks can grow under fatigue, static fatigue or stress corrosion. However, testing every welded joint in a complex structure such as, for example, an oil rig is a major undertaking and since submerged parts are dark, and divers are hampered by diving suits and may be unable to remain submerged for extended periods of time, it is particularly important in such applications to use simple, robust and reliable means for crack detection and crack depth measurement.

GENERAL DESCRIPTION OF MEASUREMENT SYSTEM Referring to Fig. 2, the general elements of the crack detection and measurement system provided in GB 2225115A (US 5019777), which are applicable also to embodiments of the present invention, will now be discussed.

A crack measurement and detection system for use, for example, in measuring defects in oil rig structures comprises a probe 1 of a size and shape to be hand holdable by a diver, connected via a flexible cable 2 to a probe driving unit 3 typically located above water level. The cable 2 includes signal lines 2a which carry the crack detection/measurement signals from the probe, and power supply lines 2b which carry an AC energising current from an AC power supply 4 located in the driving unit 3 to the probe 1.

The signal from the signalling lines 2a may be plotted, logged or monitored in any way, but it is convenient to provide a data analysis unit 5 which typically comprises a digital processor such as a personal computer and an analogue-to-digital converter (not shown) for converting the analogue signal from the signalling lines 2a into a form to be analysed. A plotter 6 may be connected, if desired, to the analysis unit to produce permanent output traces.

In portable applications, the AC power supply 4 may be co-located in a unit with the analyser 5.

Referring to Fig. 3, the probe unit 1 of GB 2225115 is shown in greater detail. It comprises a plurality of coils each comprising a plurality of windings of conductive metal wire, wound typically on to a former (not shown) and encapsulated in polymeric resin so as to be sealed against the ingress of water or atmospheres. A first coil 10 comprises an elongate solenoid energised from the AC power lines 2b, for example at 40kHz using 125mA. The plane in which the windings are wound is normal to the long axis of the probe 1. The coil is formed on a multifaceted prism former, having a convex lower face. Each facet of the lower face therefore comprises an elongate rectangle running lengthways along the probe 1.

Disposed on several of the facets are flat rectangular coils 20a, 20b, 20c, running substantially the length of each facet and being symmetrically positioned within the length of the facet. The plane of the windings of each of the facet coils 20a-20c is thus normal to that of the windings of the solenoid excitation coil 10. The ends of the coils 20a-20c are connected to signalling lines 2a; one end of each may be connected to a common line.

Also provided are several coils 30 ("absolute" or "lateral" coils) the planes of the windings of which lie parallel to that of the excitation solenoid coil 10. For clarity, only one such coil is shown. The ends of these coils, as with those of the coils 20a-20c, are connected to signalling lines 2a.

Referring to Fig. 4, in use, a human operator grasps the probe 1 and moves the probe at a substantially constant speed (so far as he can judge) over a surface to be scanned for cracks. The facets of the probe lower face are designed to cooperate approximately with the curvature of a weld between two plates as shown in Fig. 4, since it is at such welds that cracks often develop. The facets will therefore generally lie roughly parallel to the surface of the weld. The excitation coil 10 generates eddy currents in the surface of the weld area, and the sensor coils 20, 30 generate respecive output signals which are periodically read by the analyser 5, from the output of which the occurrence and dimensions of cracks can be determined. Periodically, the diver may indicate a position along the weld to enable the positions of the cracks to be accurately monitored; for example, via his voice communication channel.

Referring to Fig. 5, the output signals plotted by a plotter 6 (derived as discussed below) corresponding to the magnitude of the AC signal outputs of the coils 20 and 30 are shown as the probe 1 is moved over a crack; because of the linear probe speed, the time axis also represents distance. The diagrams are not to scale. Referring to Fig. 5a, it will be seen that the output of the coil 30 rises gently from a normal (null) value to a higher value which is related to the depth of the crack, while the crack lies beneath the probe 1.Referring to Fig. 5b, the output of the coil 20, on the other hand, provides a peak when either end of the crack is under the probe 1 but lies at the same (null) level when the probe is over the middle of a crack as when it is not; the peaks at the ends of the crack are in opposite senses, and the sense of a peak depends upon the direction in which the probe is moved.

The magnitude of the signal from the sensors 20, 30 or 40 is affected not only by the crack depth but also by the separation between the probe 1 and the surface to be measured (known as the probe "lift-off").

Variations in the depth of a coating on the surface, or protruberances on the surface, may cause the lift-off to vary as the probe 1 is manipulated over the surface by an operator. Accordingly, it may be desirable to provide a measure of the lift-off during operation and to correct the signals output from the sensors using the measurement.

In the prior art, the lateral or absolute coil 30 functions, as noted above, somewhat like the secondary winding of a transformer and is correspondingly coupled to the excitation coil 10 via the conductive surface which acts as the core. The separation from the surface varies the inductance exhibited, and consequently changes the phase angle between the signals supplied to the excitation coil 10 and that derived from the absolute coil 30. Accordingly, the lift-off or separation can be measured by measuring the phase angle or the imaginary part of the AC output signal (referenced against the AC excitation signal).

It is found that the same technique can be employed to derive a measure of lift-off from the output of the coil 40 positioned according to the invention. The analyser 5 is accordingly arranged to derive a measure of the lift-off from the probe output signal.

SIGNAL PROCESSING Accordingly, the analyser 5 may for example be arranged to analyse the probe outputs by deriving real and imaginary parts of the (notional) AC impedance obtained by referencing the signals received from the signal lines 2a against the energising signals transmitted on the power lines 2b. The signal is then processed by applying a predetermined rotation in the complex impedance plane, and the real part of the rotated signal is plotted as a direct measure of crack depth. Referring to Fig. 31, the degree of rotation is derived, in an initial phase, by measuring the signal on the signal line 2b at a first instant (C) when the operator positions the probe distant from the surface to be measured, and at a second instant (A) when the probe is proximate to the surface to be measured.

The signals at the two instants (A) and (C) define the extremes of lift-off, and are used to calibrate the analyser 5 to attempt to ensure that in the output signal the lift-off affects only the imaginary component and not the real component of the output signal when converted or processed by the analyser 5.

Accordingly, from the two signals (A) and (C), a predetermined rotation angle in the complex impedance plane is calculated, centred at (A), such as to rotate (C) to lie beneath (A) (i.e. so as to differ from (A) in the imaginary axis only). As shown in Fig. 30a, ideally, the effect of such signal processing to rotate the probe output imaginary plane is such that a variation in lift-off (in the absence of a flaw) causes the signal to traverse the line A-C (i.e.

alters only the imaginary part of the processed signal), whereas a variation in crack depth with constant lift-off causes the signal to traverse the path A-B (i.e. changes only the real part of the signal). The two orthogonal components of the rotated signal are therefore extracted and one is used as a measure of crack depth, whilst the other is used to measure lift-off. As, in the idealised representation of Fig. 30a, the two components are orthogonal, varying lift-off has no effect on the accuracy of the crack depth measurements (or vice versa). The line portion X-C occurs as it is generally not possible to maintain a constant amplitude of defect signal as lift-off becomes extremely large.

Referring to Fig. 30b, employing the output of the absolute coil described in the probe of GB 2225115A, however, it will be seen that even after this processing to reduce the interaction between lift-off and flaw depth is performed, the resulting signal locii diverge from the ideal shown in Fig. 30a and, in particular, the line segment AB is not particularly horizontal and the line segment AC deviates significantly from the vertical, and both are curved.

Further, the range of defect signal (the length of the line A-B) is small.

GENERAL PRINCIPLE OF THE PREFERRED EMBODIMENTS Referring to Fig. 6, in one exemplary embodiment of the invention, a probe 1 comprises an elongate excitation solenoid coil 10 and, disposed at or nearer an end thereof, a flat sensor coil 40 the plane of which is normal to that of the windings of the excitation coil 10.

Referring to Fig. 7, which corresponds to Fig. 5, Fig.

7b shows that the output of the coil 40 includes a non zero level when the probe lies over a crack, when compared to Fig. 5b in which it does not. We have found that varying the position of the coil 40 along the length of the excitation coil 10 varies the magnitude of this offset signal level and also the heights of the two end peaks. When the coil 40 is positioned symmetrically in the middle of the excitation coil 10, a signal similar to that of Fig.

5b can be obtained whereas at other positions, a signal similar to that of Fig. 5a can be obtained. As will be discussed below, other geometries than that of Fig. 6 can be used to acheive this effect.

The reasons for this behaviour will now be discussed.

Fig. 8 shows the magnetic field generated by the excitation coil 10. The field shown corresponds to a coil having a constant spacing between each winding, and a constant cross section along the winding axis, and a former with a relative permeability close to unity (rather than an iron core, for example).

It will seen that the magnitude of the radial magnetic field at the outer surface of the coil varies along the axial length of the coil. The profile of the variation of the field depends upon the geometry of the coil (for example the ratio of the diameter to length of the coil, and generally also the cross-sectional shape).

When such a coil is energised by an alternating current, an alternating field is produced with the probe as shown in Fig. 8 and described above. If the energised coil is brought into close proximity with the surface of an electrically conductive plate, an electrical field conforming to the profile at the magnetic field will be created within the plane of the surface of the plate but with its direction normal to the magnetic field. This electric field will cause current to flow upon the surface of the plate; if the field alternates at a suitable frequency, the current flows are confined to a surface layer of the plate.

For example, at an excitation of 40 kHz, at a suitable energising current level, the current flow in the surface of a mild steel plate is confined to the surface 0.lmm layer.

The induced current will have a profile which corresponds to the field profile generated by the coil. The current profile therefore also follows a gradient along a line in the surface lying under the coil.

Referring to Fig. 9a, this may be experimentally varified by moving a small flat search coil across the surface of a conductive plate beneath such a coil in close proximity thereto; the output current or voltage signal from the coil is shown in Fig. 9a. This current induced in the search coil is effectively the differential (with respect to distance) of the current flowing in the conductive plate, so that the integral (with respect to distance) of the search coil output will show the magnitude of the current field flowing in the conductive plate; as shown in Fig. 9b, the current flowing in the plate is null a substantial distance away from the excitation coil 10, rising through a substantial level at the ends of the coil to a maximum in the centre of the coil, and is symmetrical in magnitude about the centre of the coil.

Referring to Fig. 10a, a side elevation showing the solenoid excitation coil 10 overlying a conductive plate is shown. Referring to Fig. 10b, the circulating current flow contours within the surface of the conductive plate below the coil 10 are shown in plan view. It will be seen that the current flows in two contra-rotating eddies each centred about a point underlying (at least approximately) an end position of the solenoid 10. The maximum current density is, as noted above, under the centre of the solenoid 10.

Referring to Fig. 11, the intensity of current crossing the central line underlying the axis of the solenoid 10 is shown.

Referring to Fig. 12, a view corresponding to Fig. 10 is shown in which the conductive plate (shown in section) includes a relatively long crack of a depth D. As the depth D (as shown) is substantially constant, the resistance of the path down one side of the crack and up the other as shown in Fig. Ib is substantially constant. However, referring to Fig.

11, the current traversing the crack varies in magnitude and direction along the length of the crack, falling to zero well away from the excitation coil 10.

The voltage drop across the sides of the crack therefore also varies along its length, according to Ohms law, as shown in Fig. 13a.

The graduated voltage along the crack will cause local minor current circulations, which are thereby dependent upon the local depth of the crack and the local rate of graduation of the surface currents as shown in Fig. lOb.

As shown in Fig. 13b, the effect of the crack depth D is to vary the resistance in traversing the crack and consequently to increase the difference between highest and lowest values of potential difference across the crack (or, more generally, to vary the voltage gradient along the crack, as shown in Fig.

13c).

Referring further to Fig. 13c, it will be noted that the voltage gradient along the crack is zero at the position underlying the centre of the excitation coil 10, and anti-symmetrical about that position. It will thus be seen that, for this reason, the symmetrically postioned coils 20a-20c of the prior art shown in Fig.

3 will not under these circumstances generate an output signal since the positive and negative voltage gradients give rise to eddy currents which would, separately, induce a signal within the coils 20a-20c but together induce signals which exactly cancel.

The situation described above is modified when an end of the crack lies close to, or under, the coil 10 since beyond the end of the crack the potential difference is zero. The presence of the end of the crack close to the excitation coil 10 truncates one of the lobes of the voltage gradient plot shown in Fig.

13c, and consequently produces an asymmetrical gradient underneath the sensor coil 20, which correspondingly generates a signal. The magnitude of the signal will be largest when the end of the crack underlies the centre of the excitation coil 10. This behaviour corresponds to the signal shown in Fig. 5b.

Some contribution to the signal may also arise from the tendency of current at the end of the crack to flow around the end, rather than beneath the end; so as to perturb the current flow.

Referring back to Fig. 6, if, as in embodiments of the invention, a coil 40 is positioned so as not to be symmetrically disposed about the centre of the coil 10. If no crack is present, a constant level of eddy current induces a predetermined constant coil output signal. In the situation shown in Fig. 12, where a crack underlies the whole length of the excitation coil 10 and beyond, there will be a net voltage gradient along the surface beneath the sensor coil 40 and a corresponding local flow of current about the crack, and consequently in the presence of the crack a signal of a magnitude corresponding to the crack depth will be generated. If the coil is positioned so that the greater part lies within the ends of the excitation coil 10, a signal of the form shown in Fig.

7b with peaks at the ends of the crack is obtained; otherwise, with an equal or greater part lying outside the coil 10, the peaks substantially disappear and a form like that shown in Fig. 5a or Fig. 5b is obtained.

It is thus seen that the positioning of the coil or coils 40 relative to the magnetic fields generated by the excitation coil 10 is of considerable importance in determining the signal obtained. For deriving a signal which provides a strong indication of the crack depth, preferably the coil 40 is mounted so as to be positioned at a point of steep change of the inducing fields as this is the point at which the most significant crack depth related currents flow along the sides of the crack and are thus most easily measured. By inspection of Fig. 13c, it will be seen that the maxima of field change and of induced voltage lie at the ends of the excitation coil 10. In preferred embodiments, therefore, the coil 40 is positioned to have a substantial portion of its area at these positions.

Fig. 6 shows a single coil 40 positioned asymmetrically within the field generated by the excitation coil 10. As discussed in greater detail below, it would equally be possible to provide a second coil 40 at the other end of the excitation coil 10. If the two are connected directly together, the result is equivalent to the use of a single symmetrical coil as in the prior art. However, if the two are connected together differentially or in opposition the magnitudes of the induced signals are additive thus providing a larger output signal. As discussed below, this principle may be extended to provide a coil wound as a figure of "8", in other words in two contra-rotating lobes, centred at the middle of the excitation coil 10; in this case, the signals induced in the two lobes of the coil do not cancel, as in the coils 20 of the prior art, but are additive.

DESCRIPTION OF PARTICULAR EMBODIMENTS First Embodiment Referring to Fig. 14, as mentioned above, the structure of Fig. 6 may be improved by the provision of a pair of coils 40a, 40b disposed symmetrically about the centre of the long axis of the excitation coil 10, but mutually connected in opposition so as to be wound in opposite senses. The exact longitudinal positions of the coils 40a, 40b are selected, as taught above, to achieve the desired response. Referring to Fig. 15a, the probe may comprise a cylindrical plastic tube former of external diameter 33mm and internal diameter 31mm, by providing thereon an excitation coil 10 comprising 60 turns of 0.2mm diameter enamelled copper wire.A pair of coils connected in opposite senses 40a,40b each generally circular and comprising 50 turns of 0.05mm diameter enamelled copper wire are provided, as shown, partially outside the edge of the excitation coil 10.

A third coil 20 is provided centrally along the length of the excitation coil 10; for convenience, this coil is of the same dimensions and material as coils 40a,40b. As shown in Fig. 15b, the three coils 20, 40a,40b are adhered to the inner surface of the former 50. In use, as shown in Fig. 15c, the probe 1 is held against the surface to be inspected with the sensor coils 20, 40a, 40b adjacent to the surface.

The output of the coils 40a,40b in this embodiment is an AC output signal of a magnitude and phase angle corresponding to the crack depth and lift-off.

Preferably, the signal is processed to effect a complex plane rotation, so as to generate independant lift-off and crack depth signals, similarly to the "absolute coil" of the prior art as discussed above.

The coil 20 provides a generally null signal with peaks when either end of the crack is caused, as with the current perturbation coil 20 of the prior art.

This probe may thus, if desired, be used with an analyser 5 programmed to process the signals known from GB 2225115.

Referring to Fig. 16, also included in the probe in this embodiment is a probe driving circuit 160 comprising terminals for receiving the cable 2 and for connecting one lead from each coil 10, 20, 40a, 40b to a common line thereof, and for connecting the AC power line 2b to the other side of the excitation coil 10.

The other ends of the sensors coil 20, 40a, 40b are connected to three inputs of a multiplexer chip 61 the output of which is connected to the signal line 2a.

The power control lines for the multiplexer chip 61 may likewise be provided within the cable 2. The multiplexer 61 comprises any convenient analogue multiplexer device and is thereby arranged to alternately connect the output of each coil in turn to the signal line 2a for a predetermined period in accordance with a selection signal which may be supplied from the analyser 5. A line driver 62 may be provided to amplify the probe output signal for transmission over a long cable 2.

The probe assembly, including the former 50, coils and circuit 60, are encapsulated in a suitable hermetic seal by, for example, the application of a ceramic loaded epoxy resin, so as to be resistant to hostile environments such as seawater, or whatever the intended application of the probe is to be. The end of the cable 2 is preferably included in the encapsulation.

Although only three coils are shown, in practice a plurality of further sets of three coils could be provided distributed radially in a ring so that the probe can be applied in any orientation to the surface to be inspected; in this case, the multiplexer 61 includes an appropriate number of input channels.

Second Embodiment Referring to Fig. 17, a second embodiment of the invention is illustrated in which corresponding parts are given corresponding labels. In this embodiment, rather than using a single sensor coil asymmetrically disposed (as in Fig. 6) or a pair of symmetrically disposed but anti-symmetrically wound coils as in Fig.

14, a single coil 40 wound as figure of "8" centred on the longitudinal centre of the excitation coil 10 is provided. As shown, in this configuration, the effect of a given current is to induce currents flowing in opposite directions at opposite ends of the coil, and the effects of exciting the two lobes of the coil from currents of equal magnitude but in opposite senses is therefore to produce additive currents in the coil.

Referring to Fig. 18a, in which dimensions are shown in millimetres, in this embodiment the excitation coil 10 comprises 90 turns of 0.2mm diameter enamelled copper wire and the figure of "8" wound coil 40 comprises 50 turns of 0.05mm diameter enamelled copper wires; the dimensions of the former 50 may be as in the first embodiment and the sensor coil 40 comprises, as shown, two subtantially rectangular lobes adhered to the inner surface of the former end 50.

If, as shown, only a single coil 40 is provided there is of course no need for the multiplexer 61, but as discussed in the first embodiment, a plurality of further coils 40may be provided distributed radially around the probe in a ring so that the probe may be presented to a work surface in any orientation; in this case, the multiplexer 61 includes an appropriate number of inputs.

Returning now to Fig. 31c, by comparison with Fig.

31b, it will be seen that the output signal derived from a probe according to this embodiment is substantially closer to the idealised signal shown in Fig. 31a than is the output of the absolute coil of the prior art. Firstly, the line AC is both more linear and more vertical over a substantial portion of its length. Secondly, the line AB is both more linear and more horizontal, and shows a greater range in response to crack depths; it is also more normal to the lift-off locus so that greater separation between the effects of crack depth and lift-off can be achieved. Thus, in this embodiment the accuracy of the probe output is increased since it depends less upon a separation from the surface of the material being tested, and its range is likewise increased.

Figs. 19a and 19b show, respectively, a sectional end elevation, and a semi cut-away front elevation corresponding thereto. Referring to the figures, a probe of the general type provided in the second embodiment comprises a central solid former 50 of, for example, nylon, symmetrically radially disposed in a ring around which are four rectangular current perturbation coils 20a-20d each provided as 50 turns of 0.05mm enamelled copper wire, overlying which are provided four corresponding figure of "8" wound coils 40a-40d each provided as 50 turns of 0.05mm diameter enamelled copper wire.Surrounding and partially overlying the figure of "8" wound coils 40a-40d is a ring shaped excitation coil 10 comprising 100 turns of 0.125mum diameter enamelled copper wire; it will be seen from Fig. 19b that the disposition of the figure of "8" coils 40a-40d relative to that of the excitation coil 10 is such that the excitation coil 10 overlaps approximately half the area of each lobe of the figure of "8". A layer of sealing compound 70 (for example, ceramic loaded epoxy resin) surrounds the probe.

One example of a probe according to this embodiment was constructed to an overall diameter of ll.Omm for use in a tube of 11.659mm diameter, for the detection of defects or changes in internal diameter. The former 50 extends beyond the coils, as shown, in either direction, for mechanical attachment to cables, for example. Whilst the dimensions of the embodiments shown in Figs. 19a and 19b may be scaled to suit other tube diameters, or other applications in general, it is noted that the ability of the probe to detect small defects is affected by the size of the sensor coils 40a-40d, 20a-20d and thus if the probe is significantly scaled up it is preferable to provide a large number of small sensor coils distributed radially around the probe rather than scaling up the dimensions of the illustrated sensor coils.

Where the probe diameter is small, the driver circuit of Fig. 16 is conveniently located distant from the probe.

Referring to Fig. 20, in one particular embodiment, the probe of Fig. 19 (or of Figs. 15-18) is particularly adapted for internal inspection of pipes by providing radially distributed leaf springs, coil springs or other resilient means 80a, 80b to space the probe from the walls of the tube and maintain it approximately centrally within the tube. Attached to the probe is a cable 2. Attached to the other end of the probe via fitting thereon is a line 90 for pulling the probe along the pipe. If the cable 2 is made sufficiently strong, the probe may be pulled back through the pipe by the cable so as to retrieve the probe. The cable 1 may be made rigid, or a flexible but incompressable rod, so that the probe 1 may be pushed through the tube by the cable, without the need for a separate line 90.

In this embodiment, since the probe is maintained approximately centrally by the resilient means 80a, 80b and a plurality of radial lift-off signals can be derived from the coils 20a-20d, the profile of the pipe can be continuously measured in four directions.

Changes in lift-off will indicate either a change in the pipe geometry (caused, for example, by a dent) or the thickness of a corrosion or other layer within the pipe between the mechanical surface of the pipe and an electrically conductive layer therebeneath. The lift-off measurements therefore, in this embodiment, provide valuable information about defects in the pipe as well as enabling the correction of crack detection measurements from the coils 40a-40d.

Referring to Fig. 21, the use of the probe of Fig. 19 (or Figs. 15-18) in inspecting an acute angled weld between two members is illustrated. The probe 1 has provided at one end thereof a rigid handle 110 with the aid of which the probe is pushed down the angle of the weld. The radius of the probe in this embodiment is therefore selected to match the expected radius of the weld. The cable 2 passes through the handle 110.

Third Embodiment From the foregoing, it will be clear that the nature of the signal generated by the coil 40 in response to a crack or other surface defect will depend upon its position and geometry relative to those of the magnetic field generated by the excitation coil 10.

For example, it has been shown that an elongate coil 20 disposed symmetrically along the length of the excitation coil 10 gives rise to a signal exhibiting peaks at detected ends of a crack but having zero amplitude along its length, whereas a coil positioned predominantly outside the solenoid 10 gives rise to a signal having a predetermined level when positioned over a crack, with no peaks.

In this embodiment of the invention, means are provided for electrically switching between different probe geometries, so as to vary the signal output by the coil 40 and thus obtain different types of signal sensitivity from the same coil, either by varying the sensor coil geometry or the magnetic field geometry.

This could be done by providing two sensor coils 40 and switching the manner in which they are connected between direct and reversed, but in the embodiment shown in Figs. 20a and 20b, a single sensor coil 40 is provided, located so as to partially overlap a pair of excitation coils 10a, lOb mounted end to end. A pair of switches llOa, ll0b, ganged together, enable the AC supply to be routed either via one coil (lOb) only or via both coils in series.

Referring to Figs. 23a and 23b, the outputs of the coil 40 when the ganged switches llOa, 110b are in each position are shown in response to movement of the probe over a defect. When, as in Fig. 22a, the two excitation coils l0a, lOb are connected in series and effectively function as a single coil, the response to the defect shown in Fig. 23a is of the prior art form shown in Fig. 5b, whereas when only one coil is connected (lOb), as in Fig. 22b, the response is according to the invention, a constant response with no peaks over the length of the crack.

The probe system according to this embodiment to the invention therefore further comprises means for controlling the switch means llOa, llOb to alternate between the positions, and the analyser 5 is arranged to read the signal cable 2a in synchronism with the switching an excitation coils 10a,10b so as to separate signals generated by the coil 40 and separately log or plot those obtained for each excitation mode over time. Since the dimensions of the excitation coil 10 vary depending upon the position of the switches 110a, 110b, some different processing of the signals derived from the signal line 2a at respective times may be provided.

Fourth Embodiment In the previously described embodiments, the probe is mechanically moved over a surface typically by an operator and consequently measurements of the distance or crack length along the surface are derived on the assumption of constant probes speed of motion. This assumption is optimistic, however, where a human operator is working under difficult conditions and so the accuracy of crack positioning and length measurements cannot be guaranteed. Furthermore, moving the probe at a constant and steady speed is laborious and time-consuming for the operator and calls for skill and patience under, sometimes, difficult working conditions.

Accordingly, in this embodiment of the invention, a plurality of separate sensors are provided in a one or two dimensional array which is laid over an area of a surface to be surveyed for cracks, and means are provided for reading the sensors so as to automatically measure the response along the line or across an area directly, without the need for moving the apparatus. Distance measurements are thus derived directly in terms of distance, rather than from measurements taken over time and converted to distance on an assumption of constant speed.

One particular embodiment is shown in Fig. 24, in which the excitation coil 10 is tapped in the manner shown in Fig. 22 at a plurality of points along its length so as to be divided into a plurality of separately energisable segments. Associated with each segment is a coil 40a-40e provided so as to lie asymmetrically within the magnetic field generated by the respective segment; the coils may also be arranged, as shown in Fig. 22, so as to be symmetrically disposed between a pair of adjacent segments when energised together. A control unit 120 is connected to control the energisation of the segments. In one embodiment, the control unit 120 is arranged to energise each segment in turn, or each alternate segment, and route the signal from the associated coil 40a to the signal line 2a.This is advantageous in that when only one coil is energised at a time, the magnetic fields from different coils of the array cannot affect the current sensed by each sensor coil 40a-40e. However, segments of the excitation coil 10 may be excited together provided they are sufficiently widely separated; for example, every fourth coil may be energised together.

If only one coil at a time is energised, then the resulting signal derived from the signal line 2a corresponds to a stepwise approximation to the signal derived in the above described embodiments where the coil itself is moved; this embodiment corresponds to moving the point of application of the magnetic field rather than moving the probe itself.

Preferably, the control device 120 is arranged to operate also in the mode shown in Fig. 22, so that either a single segment or a pair of segments can be energised at a given time, so as to be able to derive separate output signals corresponding to crack depth and to the occurrence of the ends of the crack as described above, by causing the energised sections of the coil 10 to either overlap symmetrically or overlap asymmetrically with the current perturbation coils 40a-40e.

This arrangement may further be extended to replace the tapped excitation coil 10 of Fig. 24 by a plurality of discrete excitation sources each associated with one of the sensing coils 40a-40e. For example, the discrete excitation sources may be ferrite beads aligned in a two-dimensional array and each energised by an appropriate energising winding.

The array may be made two dimensional to scan an entire surface without mechanical motion. Preferably, in this embodiment, the array is formed on a flexible subtrate so as to be capable of conforming closely to a surface of varying sizes. The same is, naturally, true of the embodiment of Fig. 24; by providing the former on which the windings of the excitation coil 10 are provided as a flexible rod, the array may be fitted to varying workpiece geometries.

Referring to Fig. 25, in which dimensions are shown in millimetres, in another particular embodiment if this form of the invention, a central generally rectangular former 50 shown in Figs. 25a and 25b has provided thereupon four spaced apart excitation windings lOa-lOd (similar materials to the above described examples may be employed) within each of which is provided, on one face, an oval centrally positioned flat coil 20 providing a signal, as discussed above, indicating the detection of the end of the crack, and a figure of "8" wound coil 40 (provided outwardly of the coil 20) having generally rectangular lobes as shown in Fig. 26. The excitation coils l0a-lOd are spaced apart so that their magnetic fields do not significantly overlap.

As described above the coils may be energised sequentially in a scan or a plurality may be energised together; in the former case, the sensors 40a-40d and 20a-20d may be connected in common to a single signalling line 2a, whereas in the latter case their outputs may be multiplexed as described above onto a single line.

Fifth Embodiment Referring to Fig. 27a and b, in an alternative arrangement, a former 50 of the type employed in the first embodiment may be used having wound thereupon an excitation coil 10 comprising 80 turns of 0.2mm diameter enamelled copper wire. Within the former 50 is provided a flat rectangular sensor coil 130a having an axis normal to the excitation coil axis, comprising 50 turns of 0.08mm diameter enamelled copper wire, and spaced axially therefrom is a second (or several) similar coaxial coil(s) (130b) so as to lie at a different height from the workpiece. The two coils are connected in opposition and are located centrally of the length of the excitation coil 10.The output of a single coil would therefore be of the prior art form shown in Fig. 5b. However, since the spacing between the first coil and the workpiece differs from that between the second coil and the workpiece, the current signals are of different magnitudes in the two coils; connecting the coils in opposition (so that induced current flaws in opposite directions) thus gives rise to a net additive output signal which can be used to indicate crack depth. On the other hand, the background output (in the absence of a crack) from the two coils is subtractive. This embodiment thus makes use of the vertical potential difference gradient above the crack. Other types of sensor coil may of course be present additionally.

Alternatives and Modifications From the foregoing, it will be clear that the factor determining the type of signal derived from a sensor responsive to current perturbation in the surface is determined by the position of the sensor(s) in the (non-uniform) excitation field. Some further examples of sensor geometries will be briefly discussed with reference to Figs. 28a-k and corresponding probe output signals (processed as discussed above responsive to detected cracks) are shown in Figs.

29a-k for traversing a crack.

Referring to Fig. 28a, the current perturbation sensing coil 40 is positioned centrally with respect to lengthwise axis of the excitation coil 10, but its axis is rotated through 90 degrees relative to that used in GB 2225115. As shown in Fig. 29a, the signal derived in traversing the crack is similar to that in Fig. 5a with a peak corresponding to either end of the crack and a null signal in between.

Referring to Fig. 28b, if the coil 40 is now slightly displaced from the centre position along the length of the excitation coil 40 as to lie asymmetrically in the magnetic field thereof, the output signal shown in Fig. 29b retains a peak corresponding to each end of the crack but has a non-zero magnitude along the length of the crack.

Referring to Fig. 28c, if the coil 40 is displaced to a position within the end of the excitation coil, the predominant feature of the output signal is substantially constant region along the length of the crack of a depth relating to the crack depth. Peaks corresponding to either end of the crack remain, however.

Referring to Fig. 28d, when the sensor coil 40 is positioned with its area equally distributed around the end of the excitation coil 10, the peaks substantially disappear as shown in Fig. 29d.

Referring to Fig. 28e, if the sensor coil 40 is positioned outside the excitation coil 10, but within its magnetic field, no peaks are present and the edges of the region of constant amplitude corresponding to the crack are of shallower inclination.

Referring to Fig. 28f, for reference, the sensor coil is positioned to correspond to the alignment of the current perturbation coil in the prior art.

Referring to Fig. 28g, the geometry corresponds to that of Fig. 28d except that the long axis of the sensor coil 40 runs parallel to that of the excitation coil 10. The form of the signal of Fig. 29g is seen to be similar to that of Fig. 29d.

In Fig. 28h, the length of the sensor coil 40 of Fig.

28f is increased relative to that of the excitation coil 10 so as to extend beyond the ends of the coil in either direction; as shown in Fig. 29h this does not significantly alter the bipolar peaked nature of the output signal derived from the sensor.

Referring to Figs 28i-28k, various connections of the geometry of the embodiment of Fig. 22 are shown. In Fig. 28i, the two coils, lOa,lOb are connected in series, so that the sensor coil lies symmetrically within the joint magnetic field, so that the output signals of Fig. 29a corresponds to that of Fig. 28f.

Referring to Fig. 28j, when one of the coils lOb is not energised the arrangement of the sensor coil 40 corresponds to that of Fig. 28g, so that the output signals shown in Fig. 29j resemble those of Fig.

29g.

Referring to Fig. 28k, if the two excitation coils lOa,l0b are connected back-to-back the resulting output signal is shown in Fig. 29k, in which the peaks have disappeared; this would appear to correspond to the addition, in similar senses, of two waveforms of Figs. 29j displaced in time.

To sum up, the shape of the sensor coil is found to be less significant than the symmetry of its position in the magnetic field. The long axis of the sensor coil may be oriented normal to the long axis of the solenoid coil 10. We have also found that a sensor coil inclined at a skew angle (e.g. 150) to the solenoid coil axis provides an output signal according to the invention if a crack is traversed off the centre line of the solenoid since the perturbations caused by the crack in this case are asymmetrically seen by the sensor coil (the portion of which nearest the crack generates the higher output).

Many other modifications or alternative geometries may be adopted whilst maintaining the spirit and scope of the invention; for instance, different number of turns may be employed, or different wire materials may be employed, or different means for sensing perturbation currents or potential difference gradients may be provided (for example, suitably connected Hall effect sensors), or different sources of magnetic field may be provided. Furthermore, although the invention is particularly suitable for use in difficult or dangerous environment such as oil rigs and pipeline testing, it is equally suitable for use in detection of cracks in, for example, aircraft.

In the light of the foregoing, the invention is limited only by the scope of the accompanying claims as granted.

Claims (28)

CLAIMS:

1. A probe for generating a signal indicative of a quantitative measurement of the size of a surface defect, comprising a magnetic field generator for inducing currents in the surface, and a perturbation current sensor arrangement positioned relative to the field to give an output providing said quantitative measure.

2. A probe according to claim 1, wherein the quantitative measure provides a quantitative measure of the depth of a surface breaking flaw in the surface.

3. A probe according to claim 1 or claim 2, in which the output signal provides a quantitative measure of separation between the probe and the surface.

4. A probe according to any one of claims 1 to 3, in which the magnetic field generator is arranged to generate a magnetic field so as to produce a spatial density variation in the induced current in the surface proximate to the probe.

5. A probe according to claim 4, wherein the sensor arrangement is positioned, arranged or connected asymmetrically with regard to the magnetic field produced by said field generator.

6. A probe according to claim 5, wherein said probe arrangement is disposed so as substantially to overlie regions of maximum current density gradient in said surface.

7. A probe according to any of claims 4 to 6, in which the magnetic field generator is elongate and is such as to generate a pair of current density maxima towards ends thereof, symmetrical about the centre thereof.

8. A probe according to claim 7, wherein the field generator comprises a solenoid.

9. A probe according to claim 7 or 8, comprising a pair of current sensors disposed centro-symmetrically and connected in opposition so that opposed induced signals are additive.

10. A probe according to any preceding claim, wherein the sensor arrangement comprises a coil.

11. A probe according to claim 10, wherein the coil comprises a plurality of windings each comprising a first and a second lobe, the direction of induced current rotation in said first and second lobes being opposite, so that opposing currents induced in said lobes are additive.

12. A probe according to any of claims 1 to 9 or 11 appended thereto, in which the sensor arrangement comprises first and second current sensors positioned to be spaced apart from the surface at different spacings and connected so that corresponding induced signals arising due to surface defects are additive and other induced signals are subtractive.

13. A probe according to any preceding claim, comprising a plurality of sensor arrangements disposed around the magnetic field generator symmetrically so as to permit use of the probe in a plurality of orientations.

14. A method of measuring the thickness of a coating or corrosion layer in a pipe by providing an eddy current sensor generating a signal indicative of separation between the sensor and a conductive surface beneath the coating or corrosion.

15. A probe for measuring the depth of a surface breaking flaw comprising means for inducing a current in the surface, said current running across the direction of the flaw, and exhibiting a different density at different points along the flaw, and sensor means for providing a signal corresponding to the potential difference along the flaw caused thereby, and means for deriving therefrom an indication of the depth of said flaw.

16. A probe comprising means for generating a magnetic field and means for sensing perturbations in eddy currents induced thereby, further comprising means for switching said probe to operate between a plurality of different modes in which the sensor output exhibits different sensitivities.

17. A probe according to claim 16, in which the magnetic field generator comprises a plurality of separately energisable generator sections, and the sensor is disposed so as to lie symmetrically within the field generated by a first generator or combination of generators and to lie asymmetrically within the field generated by a second generator or combination thereof.

18. Surface defect measuring apparatus comprising means for generating a magnetic field extending in at least one direction and a plurality of sensor means for sensing eddy currents induced thereby at a plurality of points along said at least one direction.

19. Apparatus according to claim 18, wherein said magnetic generator comprises a plurality of separately energisable magnetic generators.

20. Apparatus according to claim 19, comprising electric scanning means arranged to read out said sensors in sequence to provide an electronic scan across said surface.

21. An eddy current surface measurement probe device comprising a magnetic field generator (10) and a perturbation current sensor (40), the perturbation current sensor (40) being positioned relative to the field generated by the field generator (10) to give an output which is significantly non-zero at any point along the length of a surface flaw.

22. Use of the edge effect generated by a magnetic field generator (10) to provide an output signal responsive to a surface defect.

23. An eddy current surface flaw sensor comprising an excitation member (10) and a perturbation current sensor (40) positioned non-centrally therein.

24. An eddy current sensor comprising an elongate excitation field generator (10) and a perturbation current coil (40), the windings of which are twisted so as to have portions in which the sensor current includes components flowing in contra-rotating directions.

25. A probe array for measurement of surface flaws comprising an elongate magnetic field generator (10) comprising a plurality of independently energisable portions (10a, lOb, lOc) and a corresponding plurality of perturbation current sensors (40a, 40b, 40c etc), and means for selectively energising one of said segments and reading one of said sensors.

26. An eddy current probe comprising a magnetic field generator (10) and a perturbation current sensor (40), the relationship between the position of the perturbation current sensor (40) and the magnetic field being switchable between at least two modes of operation.

27. A probe according to claim 26, wherein the generator (10) comprises a plurality of independently energisable segments.

28. A probe comprising an excitation coil (1) and a pair of sensing coils (40a, 40b), having their axes parallel or approximately so to the excitation coil (10), located at different positions relative to the excitation coil (10) so as to lie at different distances from a workpiece in use.