ABSTRACT

A novel NDE system with an ultrasound sensitive multi-element array has been demonstrated and is currently in product development. The tool is capable of imaging internal defects (e.g. voids, delamination, corrosion) and provide real-time, depth sensitive C-scan information. The system uses a patented sensor array of ultrasound sensitive elements (128 x 128) that generates images in real time. This low cost, portable tool requires no mechanical scanning to acquire C-scan images and can be used for inspection of subsurface corrosion and other flaws. Since the system requires no mechanical scanning, it is be implemented as a handheld probe, not requiring targets to be submerged in a fluid at all. The probe can be used for spot inspection by an operator, or integrated into manufacturing processes for immediate production control over large areas. The application of the system for composites will enable characterization both while they are forming, as well as for field testing of in-service materials. The system operates at thirty frames a second allowing rapid movement over larger areas. This work is supported in part by the Navy SBIR program.
KEYWORDS: NDE, Imaging, Real-Time, Ultrasound, Corrosion, Composites, On-line, Manufacturing

1. TECHNOLOGY OVERVIEW

We are reporting on the capability of our novel ultrasonic imaging camera system to rapidly characterize various materials. The ultrasound system is capable of imaging entire areas of internal targets at TV frame rates. This contrasts to conventional C-scan type systems which generate images only by moving a point by point sensor. The result is a tool which can provide real-time, large area imagery of subsurface faults.

It is important to perform characterizations of materials rapidly. More specifically, it would be highly desirable to:

1. Evaluate large areas of metals and composite materials quickly
2. Give immediate, user friendly imagery to an operator

The insertion of NDE technologies into material development with the ability to provide immediate production control feedback, and in-field structural health monitoring becomes more readily achievable. The basis for the technology is a patented ultrasound sensitive integrated circuit which reads out two-dimensional ultrasound data into a standard TV output, enabling ultrasound C-scanning in real time.

Although ultrasonic C-scanning was originally developed for inspection of homogeneous materials, the method has been now been applied to inspection for detection of corrosion, delaminations, porosity and inclusions, and to monitor the initiation and progression of damage resulting from applied mechanical loads and other environmental factors. The ultrasound imager builds on that capability by allowing these inspections to be made much more rapidly. The sort of problems encountered that we intend to monitor include real time imaging of:

1. Corrosion
2. Storage Tanks
3. Piping
4. Nuclear vessels
5. Composite materials - orientation of fibers
6. Composite materials - fiber matrix/interface conditions
7. Composite materials - state of cure
8. Composite materials - Interlaminar cracks
9. Composite materials - volume fraction

2. NEED FOR IMPROVED C-SCANNING TECHNOLOGY

For the assessment of small imperfections, high resolution systems are required. Furthermore, measurements need to be made over large areas such as of an aircraft body. This requires considerable time to conduct these measurements and also places a severe burden on human inspectors to evaluate what they observe without being overcome by boredom and a lack of concentration. Automated systems are needed to assist the inspectors in the performance of their tasks. In order to establish improved screening and detection, a system which provides detailed visualization of a 3-D volume of materials is needed. This technology bridges the large gap between taking point by point scans of the materials of interest and evaluations that can be done quickly during the manufacturing process or after field use.

3. APPLICATIONS

Figure 1: Real-time Ultrasound Imaging Implementation

The installation of a real time C-scanning tool can be for any usage of C-scanning technology today, only thousands of times faster, and more easily implemented. Overall application benefits of such a tool include:
1. Better determination of economic impacts of aging systems
2. Easier implementation through low weight, portable probe
3. Better process control
4. Quicker in-situ inspection
5. Less defective products produced
6. Less operator intervention

Specific applications that we have addressed are as follows:

1. Real time corrosion imaging of pipes, aircraft, ships, storage tanks
2. On-line, large area aerospace composite production process control
3. Real-time, in-service composite aircraft inspection
4. Real-time, in-service piping inpsetion
5. Real-time, in-service nuclear vessel inspection
6. Real-time automotive composite manufacturing inspection
7. Real-Time semiconductor package inspection

The implementation of the system for fault detection on aircraft using a hand held probe generating 'slices' inside targets is shown in Figure 1.

Figure 2: System User Interface LCD

To employ the device, the inspector prepares the target by spraying water or applying ultrasound gel to the structure. The probe assembly is placed against the target, held by a pistol grip, and ultrasound power insonifies the target by squeezing a trigger. The probe is moved along the structure and subsurface imagery appears on a small LCD screen mounted on the back of the probe. Controls are on the handle facing the user and real-time adjustments can be made to add/subtract contrast, gain, ultrasound power, pulse-echo power, etc. As is shown in Figure 2, if an area of corrosion or other subsurface faults are found, the user moves the intersection of two crosshairs on the LCD over the fault. At those crosshairs, the depth of the fault is displayed numerically on the LCD. A built-in microphone can acquire voice from the user. At the base assembly, the video stream is recorded, along with voice if desired. As the probe is moved, both the image and the A-scan would reflect the current position. Alternatively, the probe could be mounted on a track whick scans along the surface of the aircraft skin and cover an entire area, with high resolution, in minimal time (10 to 15 minutes) and stored on video for later evaluation of areas of interest.

4. SYSTEM DESCRIPTION

To generate C-scan imagery over an area, we have emulated the imaging approaches which are performed in mature visible and infrared technologies. In these methods, images are acquired by using a microelectronic array of pixels which are read out by a multiplexer. We have developed and patented a two-dimensional array which is sensitive to the properties of ultrasound, not light or IR. Similar to those other methods, ultrasound images are read out at 30 frames/second through the use of microelectronic chips. Mature electronic circuitry is used to format the images for presentation to the user, similar to a standard video camcorder. Microelectronics is the key to high resolution and high sensitivity at low unit cost. Furthermore, significant chip development has been done for these other methods and our approach is to draw on that technology.

The system is best described as a 'camcorder for sound'. Similar to a conventional camcorder which is based on CCD technology sensitive to light, Imperium has developed a novel 2D read out IC sensitive to ultrasound, not light. The technique can be used in either reflection (pulse-echo) or transmission. The transmitter is separate from the novel IC receiver. The system is broadband, working over a wide range of frequencies.

A large area, uniform ultrasound beam insonifies the desired target. In the reflection mode, the beam strikes the target and returns back towards the sensor. An acoustic lens is used to collect the resulting beam and focus the information onto the IC array.

Figure 3: ultrasound array

The patented array is made up of a both a new read out integrated circuit multiplexer and a piezoelectric layer. Since the array is only a receiver, we have chosen PVDF as the piezoelectric material since it provides a large frequency response. The piezoelectric layer is bonded with the silicon IC, also done through microelectronic processing. One of our arrays is shown in Figure 3. The square area in the middle is approximately 1 cm square and has 128 x 128 active elements, totaling over 16,000 elements each sensing unique ultrasound pressures in 1/30 second.

The pattern formed strikes the ultrasound sensitive pixel elements (128 x 128) of the array. The voltage generated at each pixel is transferred to the silicon CMOS multiplexer which is designed to read out the individual voltages sequentially, producing a TV like image.

Signal processing provides 'range gating' by controlling the acquisition time of the array to occur when the ultrasound pulse generated by the source transducer returns from a predetermined depth within the target (e.g. 5 mm in depth, See Figure 1). Each of the thirty frames per second can be programmed to return from a different depth, providing 3D information (e.g. 3 mm to 8 mm, with 1 mm increments). Alternatively, at a single depth, the frames could be integrated to provide enhanced signal to noise. The RS170 TV signal is then frame grabbed and coupled to a standard machine vision system which performs acceptance/rejection criteria, image enhancement, false color, image annotation, etc.

Figure 4: Adjustable 'Zoom' Selection

The separation of the receive and transmission functions is not intended to imply that they are to be physically separated from each other. However, the separation into the receive and transmission modes does point out the flexibility inherent in the design of the imager.

Our array is approximately 1 cm on a side (128 elements with 85 micron center to center spacing). If 85 micron resolution is desired, it could be achieved but with 1 cm area coverage. In this case, the target area covered is equal to the area of the array. However, if the required resolution is 1 mm, then by setting the lens position, an area coverage of 128 mm of the target will be achieved. Since the camera operates at 30 frames a second or higher, reasonably large areas would be covered in times much less than required by a point by point scan. Similarly, if higher resolution is required, magnification of the image will result in better resolution, but sacrifice area coverage. The objective is a 'zoom in' or 'zoom out' as the operator desires as shown in Figure 4. The ultimate resolution will be limited by the wavelength of the ultrasound employed.

Figure 5: Highly uniform beam used for imaging

All other approaches to ultrasonic imaging have used focused pulses, whereas our preferred approach requires spatially uniform insonation of the plane of interest. Uniform insonation is a research specialty of Dr. George Harrison and of the University of Maryland who have been teamed with Imperium Inc. in some of the efforts. He has the only published calculations and experimental realizations of such insonation fields. One highly uniform field is shown in Figure 5. For the improved spatial resolution, special transducers operating at center frequencies of 375 Khz, 1 MHz, 2.5 MHz, 5 MHz and 15 MHz were used for imaging. The size and shape of the source transducer was determined from field propagation simulations.

The influence of advanced signal processing and pattern recognition algorithms, and an enhanced hardware for ultrasonic non-destructive testing capabilities has been phenomenal. The recent systems store more data and analyze and image with good efficiency and hence make the whole system automatic and more reliable. Our TV formatted image is immediately adaptable to well developed commercial machine vision systems which perform immediate:

1. Frame grabbing of frames with particular features of interest
2. Conversion of shades of gray to color to bring out features not otherwise detectable, especially at higher frame rates
3. Intensity plots of image features rather than relying on image intensity to detect features
4. Dynamic range modification over different regions to enhance contrast beyond what is observed with uniform gain conditions
5. Edge enhancement
6. Image subtraction to not only look for changes, but also to remove fixed pattern noise
7. Image addition for improvement of signal to noise

5. REAL TIME IMAGES

Once real time images were obtained, a number of features of these images could be investigated.

Figure 6: Without Zoom

Figure 7: With Zoom

A relatively small target, a 7 mm diameter metal washer was imaged. The width of the metal ring is approximately 2 mm. In order to obtain images for inclusion in this report, single frames of the video images were frame grabbed and then stored to disk. Figure 6 shows a framed grabbed image of the washer held by a clamp. By moving the washer to a position closer to the focal plane of the lens and refocusing the image plane at the plane of the array, the image size increased. The enlarged image is shown in Figure 7. This is of course standard practice when operating in the visible and infrared regions of the spectrum. Mechanical C-scanning can provide the same information but at a much slower rate. It is the speed at which the images can be generated that makes this system unique. The images in Figure 6 and Figure 7 below were taken at both 30 frames /second and at 50 frames/second. The real time feature greatly simplifies the focusing adjustment, allowing the operator to adjust the focus and then to readjust the focus if a different magnification is chosen.

Close examination of Figures 6 and 7 shows that the images are made up of a set of small squares or pixels. The smallest squares are generated by the actual pixel separation in the focal plane array. In this array the pixels are 85 microns apart. Knowing the pixel size one can determine the image size by counting the number of pixels across the image.

As a first approximation it is typical to invoke the Rayleigh criterion:

_min = 1.22 / D

Figure 8: Aircraft Rivet with Crack Pointing Up Figure 9: Corroded Aluminum Plate

where (min is minimum angular separation of two points in the target, ( is the wavelength of the ultrasound signal, and D is the diameter of the lens. In fact, the resolution of a scanning system with high enough sensitivity and dynamic range can resolve the depth of the saddle point between the two point images and thereby improve on the achievable resolution.

Figure 8 is an image of an airplane rivet where a crack can clearly be seen emanating straight up. Figure 9 is an aluminum plate with spots of corrosion.

It is important to point out that images of targets moving across the field of view, either by panning the camera or by moving the target, appear more detectable. The motion of details within the image catch the observer's eye and are readily tracked across the image plane.

6. CONCLUSIONS

A new generation of ultrasound NDT imaging equipment based on microelectronic techniques has been demonstrated. This system substitutes microelectronic processing for mechanical scanning and is therefore not only much faster than existing equipment, but inherently a lower cost system. Imperium is actively looking for technologically advancing organizations currently involved in nondestructive testing to determine the application specific product specifications that will meet the needs of the NDT industry.

7. REFERENCES

1. X. Zhang, G.H. Harrison, and E.K. Balcer-Kubiczek, Exposimetry of unfocused pulsed ultrasound. IEEE Trans. UFFC 41:80-83, 1994.
2. G.H. Harrison and E.K. Balcer-Kubiczek, Uniform Pulsed Fields for Ultrasonic Bioeffect Experimentation. Ultrasonics, 29:264-267, 1991.

Dr. Marvin Lasser,
Imperium, Inc., Rockville, Maryland 20850
blasser@po.mctec.com
Dr. George Harrison,
University of Maryland Medical School, Baltimore, Maryland