ULTRASOUND METHOD AND APPARATUS

Abstract

A method of calibrating an ultrasound probe having a coupling element for engaging the surface of an object to be inspected, in which the ultrasound probe and a calibration artefact are provided on a positioning apparatus having at least one axis about which the relative orientation of the ultrasound probe and calibration artefact can be changed, the method including, in any suitable order: i) for a plurality of different relative orientations between the ultrasound probe and the calibration artefact about the at least one axis, measuring the signal received by the ultrasound probe; and ii) from the measurements, determining at least one calibration parameter which is indicative of at least one axis of optimum signal of the ultrasound probe, and recording the at least one calibration parameter for subsequent use.

Claims

1. A method of calibrating an ultrasound probe having a coupling element for engaging the surface of an object to be inspected, in which the ultrasound probe and a calibration artefact are provided on a positioning apparatus having at least one axis about which the relative orientation of the ultrasound probe and calibration artefact can be changed, the method comprising, in any suitable order: i) for a plurality of different relative orientations between the ultrasound probe and the calibration artefact about the at least one axis, measuring the signal received by the ultrasound probe; and ii) from said measurements, determining at least one calibration parameter which is indicative of at least one axis of optimum signal of the ultrasound probe, and recording the at least one calibration parameter for subsequent use.

2. A method as claimed in claim 1, in which i) comprises changing the relative orientation of the ultrasound probe and the calibration artefact whilst the ultrasound probe's coupling element remains engaged with the calibration artefact.

3. A method as claimed in claim 1, in which the position of the ultrasound probe's coupling element on the surface of the calibration artefact is the substantially the same for each of the plurality of different relative orientations.

4. A method as claimed in claim 1, in which the positioning apparatus comprises at least two axes about which the relative orientation of the ultrasound probe and calibration artefact can be changed, and in which i) comprises: for a plurality of different relative orientations between the ultrasound probe and the calibration artefact about at least two of the at least two axes, measuring the signal received by the ultrasound probe.

5. A method as claimed in claim 1, comprising determining a map of the signal received by the ultrasound probe with respect to the relative orientation between the ultrasound probe and the calibration artefact.

6. A method as claimed in claim 1, in which the at least one calibration parameter is indicative of the ultrasound probe's longitudinal wave axis.

7. A method as claimed in claim 1, in which the coupling element is a deformable coupling element.

8. A method as claimed in claim 1, in which determining the at least one calibration parameter comprises using known information about the orientation and geometry of the calibrated artefact.

9. A method as claimed in claim 1, further comprising for a given point on an object to be inspected, determining a desired inspection axis and using the at least one calibration parameter to determine a desired relative orientation of the ultrasound probe and object such that the axis of optimum signal and desired inspection axis are arranged in accordance with predetermined criteria.

10. A method as claimed in claim 9, comprising bringing the ultrasound probe coupling element into engagement with said point on the object, in accordance with the desired relative orientation of the ultrasound probe and object, and subsequently changing the relative orientation of the probe and object about at least one axis to determine if a more optimum signal can be found at a different orientation.

11. A method as claimed in claim 1, in which the positioning apparatus comprises a coordinate positioning apparatus.

12. A method as claimed in claim 1, in which the ultrasound probe is mounted on an articulated member, comprising the at least one axis.

13. An inspection apparatus comprising: an ultrasound probe; and a record which comprises at least one parameter indicative of at least one axis of optimum signal of the ultrasound probe; in which the inspection apparatus is configured to orient the ultrasound probe using the record.

14. An inspection apparatus as claimed in claim 13, in which the at least one parameter is indicative of the ultrasound probe's longitudinal wave axis.

15. A method of inspecting an object with an ultrasound probe mounted on an inspection apparatus, the method comprising: taking a parameter indicative of at least one axis of optimum signal of the ultrasound probe; determining how to relatively orient the ultrasound probe and object based on the parameter; and controlling the motion of the inspection apparatus so as to bring the ultrasound probe into engagement with the object, in accordance with the determined orientation.

16. A method of inspecting an object with an ultrasound probe mounted on an inspection apparatus, the method comprising, for a single point of engagement between the ultrasound probe and object, obtaining at least one ultrasound measurement at each of a plurality of different orientations between the ultrasound probe and object, so as to obtain information relating to ultrasound measurements obtained at different orientations between the ultrasound probe and object.

17. A method as claimed in claim 16, in which the information comprises an array of information relating to ultrasound measurements.

18. A method as claimed in claim 16, in which the method comprises obtaining multiple ultrasound measurements at each of a plurality of different orientations between the ultrasound probe and object.

19. A method as claimed in claim 16, in which the method further comprises analysing the information to determine information about the object.

20. A method as claimed in claim 19, in which the method comprises analysing the information to determine information about the backwall of the object and/or to determine information about the material of the object.

21. An apparatus comprising: an ultrasound probe mounted on a positioning apparatus; and a controller, in which the controller is configured to control the apparatus in accordance with the method of claim 1.

Description

[0057] Embodiments of the invention will now be described, by way of example only, with reference to the following drawings in which:

[0058] FIG. 1 schematically illustrates the fundamental operating principles of an example ultrasound probe;

[0059] FIG. 2 illustrates the principle of ultrasound thickness measurement;

[0060] FIG. 3 illustrates an ultrasound probe mounted on a coordinate measuring machine (CMM);

[0061] FIG. 4 illustrates the different mechanical parts of an ultrasound probe;

[0062] FIG. 5 illustrates the different electrical modules of an ultrasound probe;

[0063] FIG. 6 illustrates how the difference between assumed and calibrated L-wave axis of an ultrasound probe can affect the propagation of the L-wave in a part;

[0064] FIG. 7 illustrates an example flow chart of a process for calibrating the ultrasound probe can be calibrated;

[0065] FIG. 8 illustrates an orbital motion;

[0066] FIG. 9 shows a set of cones illustrating a set of orbital motions, along with an associated 2D energy bin map;

[0067] FIG. 10 shows the 2D energy bin map of FIG. 9 in isolation;

[0068] FIG. 11a shows a polar coordinate map of the backwall energy distribution for an uncalibrated probe;

[0069] FIG. 11b shows a polar coordinate map of the backwall energy distribution for a calibrated probe, measured about the calibrated L-wave axis which is aligned to the surface normal of the calibration artefact;

[0070] FIGS. 12a to 12d illustrate how the ultrasound probe can be oriented so as to ensure that the L-wave axis is aligned with a target vector;

[0071] FIG. 13a illustrates a part having a planar and parallel backwall, and FIG. 13b illustrates the backwall energy distribution map obtained from a series of orbital scans (taken at a single point); and

[0072] FIG. 14a illustrates a part having a planar but non-parallel backwall, and FIG. 14b illustrates the backwall energy distribution map obtained from a series of orbital scans (taken at a single point).

[0073] The fundamental operating principles of an example ultrasound probe will be explained with reference to FIGS. 1 and 2. FIG. 1 shows an ultrasound probe 2 that comprises an outer body 4. A longitudinal wave (“L-wave”) transducer is provided that comprises a piezoelectric element 6 and a deformable coupling element 8. As will be understood, other commonly employed components within the ultrasound probe, such as the backing layer and matching layer (wave-plate), can form part of the ultrasound probe, but which for simplicity are not shown in FIG. 1. To measure the thickness of an object 10, the probe 2 and/or object 10 are moved so as to engage the deformable coupling element 8 and the front surface of the object 10. When an excitation pulse is applied across the piezoelectric element 6, an ultrasound beam 12 is projected into the object 10. As schematically illustrated, this ultrasound beam 12 is reflected by the back wall of the object, this reflected beam is sensed by the ultrasound probe (e.g. via the piezoelectric element 6). This process can be referred to as pulse-echo measurement, or more colloquially as “pinging”. As will be understood, a thickness measurement estimation can be made using a time-of-flight or time-delay measurement of the projected ultrasound waveform.

[0074] FIG. 2 is an example of the ultrasound waveform received by the transducer's active piezoelectric element 6 in response to a transient high voltage excitation pulse being applied across the piezoelectric element 6. This time-domain waveform, referred to as an “A-scan” plot, is more often a time averaged response from a train of such excitation pulses (e.g. a sequence of N pulses where N may be in the range of 16-32) in order to suppress random uncorrelated electronic noise.

[0075] The initial excitation pulse generated by the piezoelectric element 6 is labelled as the “Tx-Pulse” in FIG. 2. This initial excitation pulse causes the inspection L-wave to propagate into the coupling element 8 (which acts as a delay line) and travel along at the speed of sound in the coupling element material (C.sub.L). The first reflection peak (DL1) received back at the piezoelectric element 6 arises from reflection of sound from the interface between the coupling element 8 and the object 10. This reflection from the interface between the coupling element 8 and the object 10 (i.e. the DL1 echo) can be seen to occur at a time well after the initial transmit pulse (Tx-pulse) has completely receded.

[0076] Although some ultrasound is reflected back at the delay line interface and never enters the object 10, a sufficient proportion of ultrasound does transmit into the object 10 as a measurable inspection pulse from which subsequent thickness measurements can be made. The speed of sound in the deformable coupling element 8 can be low compared to the speed of sound within the object 10, and so especially if the object is relatively thin, multiple reflections between the back wall of the part 10 can occur before a second reflection peak (DL2) from the delay line/coupling element 8 is registered at the transducer. These backwall reflections thus provide the pulses BW1, BW2, and BW3 that can also be seen in the “A-scan” plot of FIG. 2. The time window observed within the A-scan between the first and second delay line reflection peak (DL1 and DL2) is thus the probe's primary measurement window.

[0077] The thickness of the object 10 may be calculated from A-scan data of the type shown in FIG. 2 in several ways. In practice, such thickness measurements typically involve one of three modes of operation from which the time delay information can be extracted from the measured A-scan. These different modes are typically termed Mode-1, Mode-2 and Mode-3 respectively. In Mode-1 gauging, the time delay measurement is made between the excitation pulse (t=0) and the first back-wall reflection or primary echo from the object 10. In Mode-2 gauging, the time delay measurement is made between an interface echo representing the near surface of the test part and the first back wall reflection. In Mode-3 gauging, the time delay measurement is made between two or more successive back wall reflections. Mode-3 is most effective where clean high SNR multiple back wall echoes can be observed, suggesting it is most practical in low attenuation high acoustic impedance parts such as fine-grain metals, glass or ceramics. Mode-3 also has the advantage that it does not rely on the absolute time of arrival of back wall or delay line reflections, thus negating the effects of variability in the coupling and delay lines of different coupling modules. Mode-3 also allows parts to be measured that comprise outer coating layers. Any suitable mode (e.g. Mode-1, Mode-2 or Mode-3) could be used as necessary. As will be understood, the below described calibration technique extracts signal information from the measurement window, but does not necessarily involve performing a thickness measurement (e.g. a calculation of thickness). Although the calibration can typically extract information from the primary measurement window (e.g. between DL1 and DL2), the method can also apply when signal/echo information is extracted from subsequent measurement windows (e.g. between DL2 and a third delay line reflection peak “DL3”—not shown in FIG. 2) or from any combination of measurement windows.

[0078] With reference to FIG. 3 there is shown an ultrasound probe 100 mounted on a positioning apparatus 200. The positioning apparatus comprises a movement structure, in this case in the form of a coordinate measuring machine (“CMM”). The CMM 200 comprises a base 202, supporting a frame 204 which in turn holds a carriage 206 which in turn holds a quill 208 (or “Z-column”). Motors (not shown) are provided to move the quill 208 along the three mutually orthogonal axes X, Y and Z (e.g. by moving the frame along the Y axis, and the carriage 206 along the X axis, and the quill 208 along the Z-axis).

[0079] The quill 208 holds an articulated head 210, which could be an indexing head or a continuous head, but preferably is a continuous head (e.g. a REVO® head available from Renishaw plc). As will be understood, a continuous head enables orientation of a device mounted on it at substantially any angle about at least one axis and are often described as providing a near infinite number of angular orientations. Also, if desired, the orientation of the measurement device about an axis of a continuous head can be changed during measurement (e.g. whilst a contact probe is in contact with an object being inspected and acquiring measurement information). In contrast, an indexing head has a discrete number of defined (“indexed”) positions at which the measurement device mounted on it can be locked. With an indexing head, the orientation of the measurement device can be changed, but not during the acquisition of measurement data.

[0080] In this embodiment, the articulated head 210 facilitates rotation of the probe 100 mounted on it, about first and second rotational axes D, E via appropriate bearings and motors (not shown).

[0081] The combination of the two rotational axes (D, E) provided by the articulated head 210 and the three linear (X, Y, Z) axes of translation of the CMM 200 allows the probe 100 to be moved/positioned in five degrees of freedom (two rotational degrees of freedom, and three linear degrees of freedom).

[0082] Further, although not shown, measurement encoders may be provided for measuring the relative positions of the base 202, frame 204, carriage 206, quill 208 and the parts of the articulated head 210 so that the position of the measurement probe 100 relative to a workpiece 10 located on the base 202 can be determined.

[0083] A controller 220 is provided for controlling the operation of the CMM 200, such as controlling the position and orientation of the ultrasound probe within the CMM volume (either manually, e.g. via an input device such as joystick 216, or automatically, e.g. under the control of an inspection program) and for receiving information (e.g. measurement information) from the CMM 200. A display device 218 can be provided for aiding user interaction with the controller 220. The controller 220 could, for example, be a dedicated electronic control system and/or may comprise a personal computer.

[0084] As shown, a probe interface 150 (for facilitating communication with the probe 100) can be provided in the controller 320, for example.

[0085] The present invention is suitable for use with a range of different types of ultrasound probe. WO2016/051147 and WO2016/051148 describe various types of ultrasound probe which the invention can be used with. For the sake of illustration, one particular type of ultrasound probe 100 will now be described in more detail in connection with FIG. 4. The ultrasound probe 100 shown in FIG. 4 is arranged for attachment to a two-axis rotary head. It would, of course, be possible to attach such an ultrasound probe to other positioning/measurement systems, for example a robot arm.

[0086] The ultrasound probe 100 comprises a base module that includes a main body portion 102, which is provided at the proximal end of the probe 100 that attaches to the CMM. The main body 102 can contain electronics required to power the probe and communicate control data and activation commands to the probe (e.g. to schedule ultrasonic measurements). Power and/or control data, including ultrasonic data and thickness measurement results, may be passed through the rotary head communication channels and/or wirelessly. Rather than power being provided via the CMM, power could be provided by a battery located in the main body portion 102, for example.

[0087] An elongate tube 104 (e.g. a rigid carbon-fibre tube) extends from the main body 102, along the probe's axial length. A coupling element 106, in particular a deformable coupling element, is located at the end of the tube 104 distal the main body 102. In the embodiment described, the deformable coupling element 106 comprises a hydrophilic elastomer. Optionally, the deformable coupling element 106 is replaceable (e.g. could be attached via a screw or snap-fit). At least the portion of the coupling element 106 which protrudes from the tube 104 is spherical. As will be understood, the deformable coupling element may engage a wear plate of the transducer in the probe, and the deformable coupling element 106 may act as both the coupling element and a delay line. The deformable coupling element 106 can be soft and elastic so as to easily conform to the surface of the object 10 it engages.

[0088] In use, the ultrasound probe 100 is moved by the CMM 200 so as to bring the deformable coupling element 106 into contact with the object 10 to be inspected, thereby forming an acoustic coupling with the object. The ultrasound probe 100 generates ultrasound waves (for example, via a piezo-electric element) which are imparted into the object 10, and reflected by the back wall of the object 10. The reflected ultrasound waves are sensed by the ultrasound probe, and analysed to determine the distance of the back wall from the front wall of the object 10 (i.e. to determine the thickness of the object 10).

[0089] The ultrasound inspection apparatus may excite and receive ultrasound in any known way. The ultrasound inspection apparatus may operate at a high frequency. For example, the operating frequency may be greater than 5 MHz, greater than 10 MHz or more preferably greater than 15 MHZ. In a preferred embodiment, the operating frequency is around 20 MHz. The transducer, which may comprise a piezoelectric element, preferably excites longitudinal sounds waves (L-waves).

[0090] FIG. 5 schematically illustrates an example embodiment of the analogue and digital electronic modules that can be provided with (e.g. within) the ultrasonic probe 100. As schematically shown, the ultrasound probe comprises an ultrasound transducer 110, for example a pulse-echo ultrasonic transducer that includes a piezo-electric element for transmitting high frequency time-discrete longitudinal waveforms (hereinafter termed “L-waves”), when driven by a train of high-voltage impulsive excitation pulses; e.g. negative going transition (NGT) pulses between 50-150V and of duration 1/2f.

[0091] An analogue “pulser” circuit 112 is provided that is capable of generating the repeated trains of high voltage (50-150V) a.c. analogue signals (e.g. NGT pulses). Although the pulser 112 is provided, a more sophisticated digital waveform synthesizer could alternatively be employed to generate frequency or amplitude modulated waveforms to drive the piezo in more attenuating environments. The high voltage pulses generated by the pulser 112 effectively drive the piezoelectric active element 110 within the transducer of the probe to output the required ultrasonic waveforms 150, but without exceeding the maximum voltage for such a thin fragile piezoelectric element. Each pulse activation may be instigated and precisely controlled in time by an enable signal sent to the “pulser” circuitry 112 from an FPGA 114 or equivalent processor. For every activation, a fast T/R switch 116 allows the device to instantaneously switch between the transmit mode and the longer duration receive mode during which time the system acquires and digitally records the acoustic response to the transmitted pulse measured by the reciprocal piezoelectric element 112.

[0092] It might be that the amplitude level of the received signals of interest can vary significantly. Accordingly, to try to deal with this a variable gain amplifier (VGA) 118 could optionally be provided to induce gain across the acquired A-scan response in order to amplify the signal prior to digital acquisition. Moreover, to equalise the variability within each A-scan response due to propagation loss or attenuation with some materials, a form of automatic gain control (AGC) known as distance-amplitude correction (DAC) may also be implemented. The amplified A-scan is then digitized using a suitably wide dynamic range (e.g. 12 bit) analogue-to-digital converter (ADC) 120 where sufficient over-sampling above the Nyquist rate is provided as the sample rate fundamentally effects the temporal resolution of the measurement system and thus the accuracy of the thickness measurement; e.g. a sampling rate of 125 MHz or higher may be suitable for a 20 MHz transducer. The encoded digital waveforms from the ADC 120 may also require band pass filtering using a digital filter, for example a low order FIR with a pass band matching the operating frequency of the transducer. The Tx-Rx electronics are designed so as to minimise all possible sources of electronic noise that may be observed within each individual A-scan. Such uncorrelated noise is most effectively suppressed by averaging across N successive repeated A-scan measurements (i.e. providing a theoretical IN SNR gain). As will be understood, averaging is not necessarily conducted during the below described calibration method.

[0093] The ultrasound transducer 110 (e.g. a piezo-electric element) is preferably (although not necessarily) located toward the distal end of the tube 104 (i.e. at the end near the deformable contact element 106). The other electronics can be located anywhere within the probe, for example in the main body portion 102. Optionally, at least some of the other electronics could be located outside the probe, for example at least partially in the controller 220.

[0094] As will be understood, ultrasound probe can have a mechanical axis 103 as well as an ultrasound axis 105. What the mechanical axis is will depend on the probe design. Typically, the mechanical axis of a probe is defined with respect to the probe's coordinate system. In this case, the probe is mounted on an articulated head, and the probe's coordinate system rotates with the probe as the probe rotates about an axis of the articulated head. In the case of the probe being mounted on an articulated head having two perpendicular axes of rotation, the X-axis of the probe's coordinate system can be aligned with one of the axes of the head (e.g. the E axis—illustrated in FIG. 3) and the Z-axis of the probe's coordinate system can be aligned with the other of the axes (e.g. the D axis—illustrated in FIG. 3). The Y axis completes a right-handed rectangular system, and for an axial probe, the mechanical axis is defined as the Z-axis (e.g. [0 0 1].sup.T) of the probe coordinate system. For a crank-angled version of the probe, the mechanical axis is a specified constant Euler rotation from this Z-axis column vector. The ultrasound axis is defined as the direction in which the ultrasound travels in the probe coordinate system. It is a fixed vector within the probe's coordinate system, and so like the mechanical axis, it rotates with the probe as the probe rotates about an axis of the articulated head.

[0095] As schematically illustrated by FIG. 1, it could be assumed that the L-wave axis is aligned with the mechanical axis 3 of the probe 2, and therefore if the ultrasound probe engages the object with the mechanical axis normal to the front surface of object, then the L-wave axis will be projected into the part normal to the front surface of the object. Accordingly, if the probe's ultrasound axis is defined as a unit column vector v.sub.ua, fixed in probe coordinates, in an ideal situation where the probe's ultrasound axis is aligned with the probe's mechanical axis, v.sub.ua will be [0 0 1].sup.T.

[0096] However, the inventors have identified that it is very difficult to manufacture an ultrasound probe with the L-wave axis being perfectly aligned with the mechanical axis of the probe. Rather, it is more than likely that the L-wave axis will be slightly misaligned with the mechanical axis of the probe (e.g. by up to 2 to 3 degrees). Causes of misalignment include mechanical misalignment between the shaft and/or body and the transducer, and/or acoustic misalignment (e.g. where the ultrasound beam might not be perfectly symmetrical). FIG. 6, schematically illustrates such a misalignment between the assumed L-wave axis (which is assumed to the same as the mechanical axis 103 shown in dotted-line) and calibrated L-wave axis 105 (shown in solid-line).

[0097] The inventors have found that even when a small misalignment exists between the assumed and calibrated L-wave axes, the performance of the ultrasound probe apparatus is compromised. In particular, it has been found that the accuracy of measurements obtained from the ultrasound probe apparatus can be adversely affected by misalignment between the assumed and calibrated L-wave axes. Such misalignment complicates, and diminishes the amplitude of, the backwall echo(es).

[0098] For example, the distance travelled by the most intense highest amplitude L-wave will be greater than if the mechanical and ultrasonic axes were perfectly aligned and it would be a lower amplitude part of the projected L-wavefront that takes the shortest path between the front and back-walls. This will result in lower amplitude and potentially dispersed back-wall echoes from which the thickness measurement is derived using signal processing algorithms that estimate the time-of-arrival or time-difference of arrival of these back-wall echoes and this can significantly degrade measurement accuracy in such a high precision thickness measurement system (e.g. within 10 microns).

[0099] This problem is compounded due to refraction. That is, if the L-wave does not enter the object normal to the surface, then it will be refracted from the normal in accordance with Snells Law, the degree of refraction being dependent on the angle of incidence and the speed of sound within the deformable coupling element and the inspection object.

[0100] The inventors have found that such problems associated with misalignment between the mechanical/assumed axis and the calibrated L-wave axis can be significantly reduced by calibrating the ultrasound probe so as to find the L-wave axis. A method 400 of calibrating the ultrasound probe 100 in accordance with one example embodiment of the present invention will now be described with reference to the flow chart shown in FIG. 7.

[0101] In summary, the calibration process involves loading the ultrasound probe onto a known-geometry calibration artefact 20 (e.g. a metallic object having front and back walls of a known arrangement; such as them being planar and parallel) and performing a plurality of orbital motions around the surface normal of the calibration artefact whilst taking repeated pulse echo measurements (“pinging”) at a high rate to obtain an angular distribution of reflected backwall signal strength information that indicates misalignment between the ultrasound and mechanical axes. In this particular embodiment described, two sets of orbital motions are performed; a first set for coarse measurement, and a second set for finer measurement (as explained in more detail below). Accordingly, in the embodiment described two iterations of the process shown in FIG. 7 take place. However, as will be understood this need not necessarily be the case and only one (or more than two) iterations could take place. Furthermore, as will be understood, searching motions other than orbital motions could be performed.

[0102] Turning briefly to FIG. 8, this schematically illustrates an example path the ultrasound probe 100 might be caused to take during an orbital motion. The extent of the orbital motion has been exaggerated for the sake of illustration. Also, the articulated head 210 has been omitted for ease of illustration. As shown, the ultrasound probe is controlled in a way which ensures that the coupling element 106 remains stationary on a single point the calibration artefact, whilst the main body 102 (and hence the current ultrasound axis) is moved about in a circle about the calibration artefact's surface normal N. Such control could be achieved from knowledge of the length of the ultrasound probe, e.g. from prior calibration, and by achieving a known deflection of the deformable coupling element 106. As shown, a motion causes the tip (i.e. the coupling element) of the ultrasound probe to be rolled about two imaginary perpendicular axes A, B which extend perpendicular to surface normal N at the point of interaction between the ultrasound probe and the calibration artefact 20. This enables data (e.g. pulse-echo or backwall energy information) in more than one dimension (with respect to the probe) to be obtained. Accordingly, information about the L-wave axis can be determined for more than one degree of freedom/dimension with respect to the ultrasound probe. As will be understood, synchronous rotation about the D and E axes (in a back-and-forth manner), as well as movement of the articulated head in the Z and Y axes (so that it is driven in a circular manner) will be required to effect such an orbital motion.

[0103] In the embodiment described, each set of orbital motions effectively forms a set of conical shapes. FIG. 9 shows one set of cones (five) around a vector (shown as a black arrow) which represents the surface normal.

[0104] These two sets of cones essentially can be described as two sets of angles (between the surface normal and each cone). Table 1 below shows example angles that could be used:

TABLE-US-00001 TABLE 1 Angles that define conical shapes Angle Angle Number of Set # range step different angles Angles 1 0° to 1.25° 0.25° 6 0°, 0.25°, 0.50°, 0.75°, 1.00°, 1.25° 2 0° to 0.4°  0.04° 11 0°, 0.04°, 0.08°, 0.12°, 0.16°, 0.20°, 0.24°, 0.28°, 0.32°, 0.36°, 0.40°

[0105] Although not explicitly shown in the flow chart of FIG. 7, as will be understood, the method could comprise obtaining data about the front and back walls of the calibration artefact. For example, the method could comprise measuring the position and orientation of the front wall of the calibration artefact 20, e.g. with a position measurement probe, for example a touch-trigger probe. This so that the calibration artefact's surface normal can be established, and so that the motions/orbits of the probe in the CMM coordinate system can be defined. In FIG. 7, the calibration artefact is referred to as “UPA” and the ultrasound probe is referred to as “RUP”. If it can be assumed that the backwall is planar and parallel to the front wall, then it is not necessary to check this by measurement. If such an assumption cannot be made, then the method can comprise also measuring the backwall of the calibration artefact.

[0106] At the start of an ultrasound axis calibration, at step 402, the ultrasound axis result v.sub.ua is initialised to be [0 0 1].sup.T, and this is loaded into the controller at step 404. Accordingly, on the first iteration, v.sub.ua is set as if there is no misalignment, and (as explained in more detail below) on the second interaction, v.sub.ua is updated from the result of the first iteration to adjust for misalignment found during the first iteration.

[0107] Step 406 comprises determining how to control the CMM, in order to perform the orbital motions. This involves determining the D and E articulated head angles. As will be understood, the steps in an orbital motion can be defined with respect to a vector v.sub.cmm in CMM coordinates, and so the D and E articulated head angles need to be calculated so as to realise this vector (such that v.sub.ua is aligned to v.sub.cmm). These D and E angles are then passed to the controller such that the controller can control the orientation of the ultrasound probe accordingly.

[0108] Whilst the ultrasound probe is carrying out an orbital motion, for each pulse-echo measurement (or “ping”), the A-scan and the D and E angles are recorded at step 408. At step 410, the data recorded from a set of orbital motions is processed, as described in more detail below, so as to determine a new v.sub.ua. Thereafter, at step 412, v.sub.ua will get updated. At step 414 it is determined whether or not to repeat this process. In this case, the process is repeated because so far only the coarse measurements have been performed, and a second set of finer measurements are to be obtained.

[0109] Having two sets of orbital motions with different resolutions means that the ultrasound (e.g. L-wave) axis can be identified quickly and coarsely during the first set of motions, and then a much finer search around the result determined after the first set of motions, can be performed during the second set of motions. By doing so, a balance between calibration time and accuracy is well maintained. However, as will be understood, this does not necessarily have to be the case, and just one set of motions could be performed. As will also be understood, a search motion (e.g. orbital motion) can be interrupted before its completion if the local peak of interest has been found before the defined search path motion has completed.

[0110] As indicated above, the processing of the data at step 410 can happen in parallel while the ultrasound probe is “pinging” and performing the orbital motions.

[0111] However, as will be understood this need not necessarily be the case, and for instance could be done after all of the orbital motions in a set have finished, for example.

[0112] In the embodiment described, for recording and processing purposes, energy bins with respect to the calibration artefact's front surface are defined/used. FIG. 10 schematically illustrates energy bins for the first set of orbital motions. Each spoke is 5° (this value is fixed across all sets of orbital motions) apart from its adjacent spokes, making the total number of spokes 72. Each circle is generated from radius as the sine value of the angles listed in the first row, last column in Table 1. In this embodiment, the energy bin map is a 2D map and the backwall energy for the D and E angles are projected into the 2D map as described below. However, as will be understood, this need not necessarily by the case. For example, a 3D energy bin map could be used.

[0113] At step 408, for each ping, the backwall energy is recorded and allocated to its closest energy bin, the process for which is explained in more detail in the following few paragraphs. In order to reduce susceptibility to noise, each orbital motion can be performed a number of times, with multiple results being stored in each bin and therefore each bin can represent a mean energy value. Also, the number of measurements used to get the result in each bin can be saved, such information being used to calculate the current updated averaged energy. Defining the resolution of the polar coordinate grid allows a discrete energy distribution map to be determined. However, as will be understood, the resolution of this map could be increased/improved after the empirically complied map has been populated using any well-established interpolation methods. It is also noted that noise that could be present within the measured distribution could be reduced or removed via post-process filtering or smoothing, thus potentially improving the reliability or accuracy of the extracted energy peak.

[0114] Extracting the backwall energy for each ping can comprise extracting any suitable signal strength/quality indicator. Accordingly, extracting the backwall energy for each ping can comprise extracting any suitable signal amplitude or integrated signal energy measure from the raw ultrasound A-scan signal measurement. For example, within the time-interval of each A-scan known as the measurement window, one or more multiple time-discrete echoes from the calibration artefact's backwall will be observed. Any signal metric related to the amplitude, the duration, the time-of-arrival or the time-integrated energy of one or any time-gated combination of the present backwall echoes can be used. For example, the peak amplitude of the envelope of the first highest amplitude backwall echo can be extracted using the Hilbert transform or any other suitable demodulation technique, such as square-law detector, or Teager-Kaiser energy operation. This peak could be defined as the chosen energy metric of the A-scan at the D and E angles the ping occurred (which are reported by the controller). Using the recorded D and E angles, a vector V.sub.cmm (which is the vector of the direction that the ultrasound travels in the CMM coordinates) is then calculated, for example by a transformation such as that shown in equation 1.

v.sub.cmm=R×v.sub.ua  (1)

[0115] Where R is a 3 by 3 rotation matrix, and where and v.sub.ua is the current ultrasound axis result, which is a column vector in probe coordinates.

[0116] Currently, the vector v.sub.cmm is a vector in 3D coordinates, and therefore v.sub.cmm is then projected onto a plane (e.g. the calibration artefact's front surface) defined by the calibration artefact's surface normal v.sub.sn. A polar coordinate system on that plane is thereby defined. The pole is defined as the point where the cones intersect the plane. The polar axis v.sub.cmm is defined as the projection of the first v.sub.cmm onto the plane. Each v.sub.cmm's projection point will be allocated to its nearest energy bin. In other words, the V.sub.cmm projection point is the point at which the V.sub.cmm vector intersects the plane. The closest energy bin to that projection point will be identified and the averaged energy in that bin will get updated.

[0117] FIG. 11a schematically illustrates an example back-wall energy distribution for an ultrasound probe 100 after a set of orbital motions has completed. As can be seen, for an uncalibrated probe, the highest averaged energy is not symmetrical, which is representative of the misalignment of the mechanical and L-wave axes. The map of FIG. 11a is to be contrasted with the map of FIG. 11b which has been obtained for a calibrated probe (and for which the orbital motions took place about the calibrated L-wave axis). As will be understood, the calibrated map distribution of FIG. 11b would be obtained from the calibrated probe with the artefact positioned at any location and orientation within the CMM volume. In the embodiment described, the backwall energy distribution comprises a 2D map, defined with respect to the front surface of the calibration artefact. As will be understood, this need not necessarily be the case, and for instance the energy bins could be defined in a 3D space.

[0118] After each set of orbital motion is finished, the energy bin which has the highest averaged energy will be identified. For example, i.e. its azimuth φ.sub.max and radius r.sub.max are identified. Then it will be translated from 2D to 3D and the result is a vector in CMM coordinates (V.sub.max), representing the vector that gives the highest energy. Accordingly, once transformed, v.sub.max, which is a vector in CMM coordinates, gives you the maximum energy. This itself then needs to be transformed into probe coordinates. Accordingly, the D and E angles are calculated, such that v.sub.ua is aligned to v.sub.max. Next, based on the D and E angles, the rotation matrix R.sup.T from CMM coordinates to probe coordinates, is calculated. Then the updated ultrasound axis result v.sub.ua_new (in probe coordinates) is calculated such that:

q.sub.ua_newR.sup.T×v.sub.sn  (2)

[0119] Therefore, v.sub.ua's x and y components are now non-zero. In this case the angle between v.sub.ua and [0 0 1].sup.T quantifies how misaligned the probe is, while its x and y components indicate the direction of misalignment. Accordingly v.sub.ua provides a definition of the calibrated axis of the L-wave, in the probe's coordinate system.

[0120] If the ultrasound probe's L-wave is perfectly in alignment with its assumed axis (e.g. the probe's mechanical axis) or if it has already been correctly calibrated (i.e. the current Vua is correct), V.sub.max will be the same as/very close to calibration artefact's surface normal. Otherwise, V.sub.max will deviate from the surface normal and based on these two vector values the current V.sub.ua can be updated so that it becomes a better representative (V.sub.ua new) of the Ultrasound Axis (in probe coordinates).

[0121] Accordingly, in summary, the above process enables the calibrated axis of the L-wave to be determined by changing the relative orientation of the ultrasound probe and the calibration artefact, and analysing the backwall energy received at the different angular orientations. In the above embodiment, the ultrasound probe is moved in a circular, orbital fashion, which is repeated at a plurality of different angles with respect to the surface normal. However, as will be understood, this need not necessarily be the case, and other (for example iterative) processes for obtaining data on the backwall energy for different angular orientations can be used. As will be understood, the search motions (i.e. the way in which the orientation of the probe is controlled) could follow any pattern and does not have to be circular/orbital motions. For instance, the ultrasound probe could be swung in a substantially back and forth manner (with the coupling element stationary on the calibration artefact) to achieve similar measurement data. Also, rather than the data being stored/presented in a 2D map, the data obtained during a search/calibration procedure could be stored/presented in other ways. For example, measurement data could be recorded in a 1D array/plot/time series. Further, optionally, averaging does not necessarily have to be performed. For example, the method could comprise performing an N-iteration search using the maximum in the backwall time series, ending up with a descending “step-pattern” in the time-series on the Nth iteration.

[0122] Also, in alternative embodiments, the ultrasound probe could be retracted from the surface of the calibration artefact, re-oriented and placed back on the surface of the calibration artefact to take another measurement. Further still, although it can be advantageous for the ultrasound probe's coupling element to be stationary/reengage the same point for each measurement, this need not necessarily be the case, so long as the relationship between the front and back wall of the calibration artefact is known and taken into consideration in the processing.

[0123] If desired, interpolation and/or smoothing of energy maps to estimate the required peak energy vector can be used in order to reduce the number of actual measurements taken, thereby improving the speed of the calibration process.

[0124] Knowledge of the calibrated axis of the L-wave of the ultrasound probe can be advantageous for a number of reasons. The calibrated probe will more consistently return optimal backwall echoes and improve signal to noise ratio which inherently improves thickness measurement accuracy. In other words, knowledge of the calibrated axis of the L-wave of the ultrasound probe enables the ultrasound probe to be oriented so as to obtain the cleanest signal. For instance, for a given point on an object to be inspected there can be a desired propagation vector for the L-waves to propagate along within the object, and therefrom a target inspection axis/vector may be determined. The target vector can be determined based on the assumed/known geometry of the part, including the part's material properties and, as explained in more detail below, taking into consideration Snells law so that any expected refraction of the L-wave as it enters the part can be compensated for. Once the target vector has been determined, it is possible, based on knowledge of the relationship between the ultrasound probe's L-wave and mechanical axes from the above mentioned calibration procedure, to calculate the D and E angles for the articulated head which will ensure that the L-waves propagate along the desired propagation vector within the object.

[0125] Taking the example shown in FIG. 12a, for the desired propagation vector for the L-waves to propagate along within the object, a target vector is determined, which in this case is perpendicular to the front wall of the part being inspected. Accordingly, FIG. 12b illustrates how that, based on knowledge of the relationship/misalignment between the ultrasound probe's L-wave and mechanical axes from the above mentioned calibration procedure, it is possible to determine how to orient the ultrasound probe in order to ensure that the L-waves propagate along the desired propagation vector within the object.

[0126] As mentioned above, calculation of the target vector may take into consideration the effects of refraction as the L-wave enters the workpiece. For instance, if as shown in FIG. 12c, the desired propagation vector is not perpendicular to the front wall, then the effect of refraction needs to be taken into consideration when determining the desired target vector. As shown in FIG. 12c, on the basis of Snells law, it is predicted that the ultrasound wave will refract away from the surface normal N as it enters the workpiece 10′. Accordingly, the Target vector is determined so as to take this into consideration. Then, based on knowledge of the L-wave axis from the above mentioned calibration procedure, the mechanical axis of the ultrasound probe is offset from the target vector to ensure that the L-waves propagate along the desired propagation vector within the object. Therefore, the calibrated probe will also allow improved measurement of wedge geometries, where incident probe vectors predicted by Snells law will agree more closely with empirical discovery mode measurements.

[0127] As will be understood, in the case of FIG. 12b and also in the case of FIG. 12d, the mechanical axis of the ultrasound probe is offset from the target vector in a way which means that the ultrasound probe's calibrated L-wave axis is aligned with the target vector. In the case of FIG. 12b where there is no refraction, the L-waves propagate within the object along a vector that is aligned with the target vector (as illustrated). In the case of FIG. 12d, the ultrasound probe's calibrated L-wave axis is aligned with the determined target vector, but due refraction (which was taken into consideration when determining the target vector), the L-waves will refract at the interface between the ultrasound probe's tip and the object, and therefore will propagate along the desired propagation vector within the object (which is not aligned with the target vector).

[0128] The inventors have found that even with a calibrated probe, it could be beneficial to search around the target vector to see if an even more optimum signal can be obtained. For example, it might be that the actual workpiece being inspected is not a perfect part, and/or accurately predicting or measuring the effect of refraction could be difficult. Accordingly, the inspection apparatus could be configured so as to load the ultrasound probe onto the part such that the L-wave axis is initially aligned with the target vector, and then configured to search for a stronger backwall signal at different angular orientations. For instance, the apparatus could be controlled to perform a series of orbital motions, in a manner similar to that described above, and to analyse the backwall energies received to determine if there is another set of D and E angles which provide an even stronger backwall signal. With a probe that has calibrated according to the invention, the search can be more targeted than if the probe were not calibrated, thereby providing a speed benefit.

[0129] Further, it is noted that for a calibrated probe, it is possible to conduct discovery-mode measurements that compile useful backwall energy distribution information (e.g. polar coordinate maps) that can be used to independently gauge, classify or recognise the internal part geometry without requiring any thickness measurements to be performed, and this could be done from a single measurement point. For example, discovery mode measurements from an ultrasound axis calibrated probe allows planar and/or parallel geometries to be distinguished from planar wedge or curved geometries. Automated recognition of parallel and wedge geometries can improve measurement accuracy and reliability, as this can allow the system to automatically select an optimal time-delay estimation algorithm from which the thickness is derived. Discovery mode measurement from an ultrasound calibrated probe would also allow the wedge-angle to be estimated from a single measurement point.

[0130] Accordingly, the inventors have also devised a technique for obtaining lots of information about the backwall without actually having to perform thickness measurements/calculations. For example, they have found that with knowledge of the ultrasound probe's L-wave axis, it is possible to obtain and analyse a map of the backwall energy, which can provide a range of information about the backwall, for instance its form, for example its orientation and/or shape. For example, from a single point, a series of measurements at different angular orientations can be performed, (e.g. via orbital scans in a manner similar to that described above), in order to obtain an energy distribution map over an area of the backwall (e.g. polar coordinate map).

[0131] For example, FIG. 13a illustrates a part having a planar and parallel backwall, and FIG. 13b illustrates the backwall energy distribution map obtained from a series of orbital scans (taken at a single point) centred about the part's front-wall surface normal. FIG. 14a illustrates a part having a planar but non-parallel backwall, and FIG. 14b illustrates the backwall energy distribution map obtained from a series of orbital scans (taken at a single point) centred about the part's surface normal. As can be seen, the backwall energy distributions of these are different. As such, the energy maps are characteristic of the part geometry. Accordingly, the backwall energy distributions can be analysed in order to determine the form of the backwall. For instance, a symmetrical and centred energy distribution indicates a planar and parallel backwall.

[0132] Not only this, but as shown in FIG. 14b, it can be seen that the backwall energy map obtained from a calibrated probe performing a series of orbital scans of a part having non-parallel front and back walls has two peaks; a first peak P.sub.1 and a second (smaller) peak P.sub.2. The two energy peaks can be extracted using suitable time-gating within the measurement window of each A-scan. The first peak P.sub.1 relates to the target vector at which the faster L-waves return from the closest position on the planar backwall, and the second wider angle and smaller peak P.sub.2 relates to the target vector at which slower shear waves (S-wave) return from this closest point on the backwall. From such a polar coordinate energy distribution map compiled with a calibrated probe that is characteristic of the inspection part geometry, the angles between the part surface normal and L-wave and S-wave peaks can be used together to allow an independent estimation of the ratio of the L-wave to S-wave sound speeds within the material to be estimated directly (e.g. without any thickness measurement or time-delay estimation being performed).

[0133] As such, further material characterisation information (e.g. elastic properties) about the part may be inferred as the ratio of longitudinal to shear wave sound speeds can be related to Poisson's ratio. Furthermore, if the density of the part material can be estimated, the elastic modulus of the material could also be estimated.

[0134] As will be understood, the above techniques for determining how to orient the probe in accordance with a target vector and for obtaining backwall information could also be used without having to calibrate the probe in accordance with the above described calibration techniques, if the ultrasound axis of the probe is known or can be assumed with confidence (e.g. because the probe has been built to very high tolerances and therefore it is known that the mechanical and ultrasound axes are aligned or have a particular relationship).

[0135] The embodiments described above describe using the ultrasound probe to measure the thickness of a part. As will be understood, ultrasound probes can be used for other purposes, such as detecting/measuring flaws in a part. The same calibration process can be used for calibrating an ultrasound probe which is to be used for flaw detection/measurement.

[0136] Also, the discovery mode measurements on a wedge part of known geometry could be used to estimate the sound speed in the part material (e.g. without having to do backwall time-delay estimation). Such sound speed in the part material could be determined from knowledge of the sound speed in the coupling element (which could be known or measured as described in WO2016/051147 and WO2016/051148, for example).