WO1999036881A1

WO1999036881A1 - Method and apparatus for three-dimensional inspection of electronic components

Info

Publication number: WO1999036881A1
Application number: PCT/US1999/000714
Authority: WO
Inventors: Elwin M. Beaty; David P. Mork
Original assignee: Beaty Elwin M
Priority date: 1998-01-16
Filing date: 1999-01-13
Publication date: 1999-07-22
Also published as: US20050189657A1; JP2002509259A; US7508974B2; US6064756A; US6064757A; EP1046127A1; KR20010040339A; US6072898A; CN1287643A; EP1046127A4; US6862365B1

Abstract

A calibration and part inspection method and apparatus for the inspection of ball grid array, BGA, devices (70). Two cameras (10, 15) image a precision pattern mask (22) with dot patterns deposited on transparent reticle (20). The precision pattern mask (22) is used for calibration of the system. A light source and overhead light reflective diffuser (5) provide illumination. A first camera (10) images the reticle precision pattern mask (22) from directly below. An additional mirror (32, 36) or prism (30, 34) located below the bottom plane of the reticle (20) reflects the reticle pattern mask (22) from a side view, through prisms (30, 34) or reflective surfaces (32, 36, 38), into a second camera (15) and a second additional mirror (32, 36, 38) or prism (30, 34) located below the bottom plane of the reticle (20) reflects the opposite side view of the reticle pattern mask (22) through prisms (30, 34) or mirrors (32, 36, 38) into a second camera (15).

Description

METHOD AND APPARATUS FOR THREE DIMENSIONAL INSPECTION OF ELECTRONIC COMPONENTS NOTICE RE COPYRIGHT

A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent disclosure, as it appears in the Patent and Trademark Office patent files or records, but otherwise reserves all copyright rights whatsoever.

FIELD OF THE INVENTION This invention relates to a method and apparatus for three dimensional inspection and, more particularly, a method and apparatus for three dimensional inspection of solder balls on ball grid arrays and solder bumps on wafer and die, and to a calibration method.

BACKGROUND OF THE INVENTION Prior art three dimensional inspection systems have involved laser range finding technology, moire interferometry, structured light patterns or two cameras. The laser range finding method directs a focused laser beam onto the Ball Grid Array, BGA, and detects the reflected beam with a sensor. Elements of the BGA are determined in the X, Y and Z dimensions utilizing a triangulation method. This method requires a large number of measurement samples to determine the dimensions of the BGA resulting in longer inspection times. This method also suffers from specular reflections from the smooth surfaces of the solder balls resulting in erroneous data.

Moire interferometry utilizes the interference of light waves generated by a diffraction grating to produce a pattern of dark contours on the surface of the BGA. These contours are of known distance in the Z dimension from the diffraction grating. By counting the number of contours from one point on the BGA to another point on the BGA, the distance in the Z dimension between the two points can be determined. This method suffers from the problem of low contrast contour lines resulting in missed counting of the number of contours and resulting in erroneous data. This method also suffers from the contour lines merging at surfaces with steep slopes, such as the sides of the balls on the BGA, resulting in an incorrect count of the number of contours and resulting in erroneous data.

Structured light systems project precise bands of light onto the part to be inspected. The deviation of the light band from a straight line is proportional to the distance from a reference surface. The light bands are moved across the part, or alternately the part is moved with respect to the light bands, and successive images are acquired. The maximum deviation of the light band indicates the maximum height of a ball. This method suffers from specular reflections due to the highly focused nature of the light bands resulting in erroneous data. This method further suffers from increased inspection times due to the number of images required.

Two camera systems utilize one camera to view the BGA device in the normal direction to determine X and Y dimensions and the second camera to view the far edges of the balls from an angle. The two images are combined to determine the apparent height of each ball in the Z dimension utilizing a triangulation method. This method suffers from the need for a higher angle of view of the ball from the second camera resulting in looking at a point significantly below the top of the ball for BGA' s having fine pitch. This method also suffers from limited depth of focus for the second camera limiting the size of BGA's that can be inspected. This system can only inspect BGA's and not other device types such as gullwing and J lead devices.

The prior art does not provide two separate and opposite side views permitting larger BGA' s to be inspected or nonlinear optics to enhance the separation between adjacent ball images in the side perspective view.

It is therefore a motivation of the invention to improve the accuracy of the measurements, the speed of the measurements, the ability to measure all sizes and pitches of BGA' s and to measure other devices including gullwing and J lead parts in a single system.

SUMMARY OF THE INVENTION The invention provides a calibration and part inspection method and apparatus for the inspection of BGA devices. The invention includes two cameras to image a precision pattern mask with dot patterns deposited on a transparent reticle to be inspected and provides information needed for calibration. A light source and overhead light reflective diffuser provide illumination that enhances the outline of the ball grid array. A first camera images the reticle precision pattern mask from directly below. An additional mirror or prism located below the bottom plane of the reticle reflects the reticle pattern mask from a side view, through prisms or reflective surfaces, into a second camera. A second additional mirror or prism located below the bottom plane of the reticle reflects the opposite side view of the reticle pattern mask through prisms or mirrors into a second camera. By imaging more than one dot pattern, the missing state values of the system can be resolved using a trigonometric solution. The reticle with the pattern mask is removed after calibration and a BGA to be inspected is placed with the balls facing downward, in such a manner as to be imaged by the two cameras . The scene of the part can thus be triangulated and the dimensions of the BGA are determined.

The system optics are designed to focus images for all perspectives without the need for an additional focusing element. The optics of the side views may incorporate a nonlinear element to stretch the image in one direction to increase the apparent spacing between adjacent ball images allowing a lower angle of view and inspection of BGA's with closely spaced balls.

The invention provides an apparatus for inspecting a ball grid array, wherein the apparatus is calibrated using a precision pattern mask with dot patterns deposited on a calibration transparent reticle. The apparatus for inspecting a ball grid array comprises a means for mounting the ball grid array and a means for illuminating the ball grid array to provide an outline of the ball grid array. A first camera is positioned to image the ball grid array to provide a first image of the ball grid array. A first means for light reflection is positioned to reflect the ball grid array through a second means for light reflection into a second camera, wherein the second camera provides a second image of the ball grid array. A third means for light reflection is positioned to reflect an opposite side view of the ball grid array into a fourth means for light reflection and into the second camera as part of the second image of the ball grid array. A means for image processing, such as a computer, microprocessor or digital signal processor, processes the first image and second image of the ball grid array to inspect the ball grid array. The invention also provides a method for three dimensional inspection of a lead on a part mounted on a reticle. The method comprises the steps of: locating a first camera to receive an image of the lead; transmitting an image of the lead to a first frame grabber; providing fixed optical elements to obtain two side perspective views of the lead; locating a second camera to receive an image of the two side perspective views of the lead; transmitting the two side perspective views of the lead to a second frame grabber; operating a processor to send a command to the first frame grabber and second frame grabber to acquire images of pixel values from the first camera and the second camera; and processing the pixel values with the processor to obtain three dimensional data about the lead.

The invention also provides a method to inspect a ball grid array device comprising the steps of: locating a point on a world plane determined by a bottom view ray passing through a center of a ball on the ball grid array device; locating a side perspective view point on the world plane determined by a side perspective view ray intersecting a ball reference point on the ball and intersecting the bottom view ray at a virtual point where the side perspective view ray intersects the world plane at an angle determined by a reflection of the side perspective view ray off of a back surface of a prism where a value of the angle was determined during a calibration procedure; calculating a distance L_x as a difference between a first world point, defined by an intersection of the bottom view ray with a Z=0 world plane, and a second world point, defined by the intersection of the side perspective view ray and the Z=0 world plane, where a value Z is defined as a distance between a third world point and is related to _x as follows:

tanθ, = — L_*

Z = E-tanθ.

wherein Z is computed based on the angle; computing an offset E as the difference between the virtual point defined by the intersection of the bottom view ray and the side perspective view ray and a crown of a ball at a crown point that is defined by the intersection of the bottom view ray with the crown of the ball, and can be calculated from a knowledge of the angle and ideal dimensions of the ball where a final value of Z for the ball is:

Spina_l - Z - E . The invention also provides a method for finding a location and dimensions of a ball in a ball grid array from a bottom image comprising the steps of : defining a region of interest in the bottom image of an expected position of a ball where a width and a height of the region of interest are large enough to allow for positioning tolerances of the ball grid array for inspection; imaging the ball, wherein the ball is illuminated to allow a spherical shape of the ball to present a donut shaped image, wherein the region of interest includes a perimeter of the ball wherein the bottom image comprises camera pixels of higher grayscale values and where a center of the bottom image comprises camera pixels of lower grayscale value and wherein a remainder of the region of interest comprises camera pixels of lower grayscale values; finding an approximate center of the ball by finding an average position of pixels having pixel values that are greater than a predetermined threshold value; converting the region of lower grayscale pixel values to higher grayscale values using coordinates of the approximate center of the ball; and finding the center of the ball. The invention also provides a method for finding a reference point on a ball in an image of a side perspective view of a ball grid array comprising the steps of: defining a region of interest in the image from an expected position of a ball wherein a width and a height of the region of interest are large enough to allow for positioning tolerances of the ball grid array; imaging the ball, wherein the ball is illuminated to allow a spherical shape of the ball to present a crescent shaped image having camera pixels of higher grayscale values, and wherein a remainder of the region of interest comprises camera pixels of lower grayscale values; computing an approximate center of the crescent shaped image by finding an average position of pixels that are greater than a predetermined threshold value; using coordinates of the approximate center of the crescent; determining a camera pixel as a seed pixel representing a highest edge on a top of the crescent shaped image; and determining a subpixel location of the reference point based on the camera pixel coordinates of the seed pixel that define coordinates of a region of interest for the seed pixel. BRIEF DESCRIPTION OF THE DRAWINGS To illustrate this invention, a preferred embodiment will be described herein with reference to the accompanying drawings .

Figure 1A shows the apparatus of the invention for system calibration.

Figure IB shows an example calibration pattern and example images of the calibration pattern acquired by the system.

Figure 2A shows a flow chart of a method of the invention used for calibration of the bottom view.

Figure 2B shows a flow chart of a method of the invention used for determining the state values, and the X and Y world coordinates, of the bottom view of the system.

Figure 2C shows a flow chart of a method of the invention used for calibration of the side perspective views.

Figure 2D shows a flow chart of a method of the invention used for determining the state values of the side perspective views of the system.

Figure 2E shows the relationship of a side perspective angle to the ratio of the perspective dimensions to the non-perspective dimensions .

Figure 2F shows a bottom view and a side perspective view of precision dots used in the method for determining a side perspective view angle. Figure 3A shows the apparatus of the invention for part inspection.

Figure 3B shows example images of a part acquired by the system.

Figure 4 shows a method of the invention for the three dimensional inspection of balls on a ball grid array .

Figures 5A and 5B together show a flow chart of the three dimensional inspection method of the invention. Figures 6A and 6B show an example ball of a ball grid array and associated geometry used in a method of the invention for determining the Z position of the ball.

Figure 7A shows one example of an image used in the grayscale blob method of the invention.

Figure 7B shows one example of an image used with the method of the invention to perform a subpixel measurement of the ball reference point.

Figure 8A shows a side perspective image of the calibration pattern magnified in one dimension.

Figure 8B shows a side perspective image of the balls on a BGA, magnified in one dimension.

Figure 9 shows an apparatus for presenting a BGA for inspection. Figures 10A and 10B show an example ball of a ball grid array with associated geometry as used with a method of the invention for determining the Z position of a ball using two side perspective views. Figure 11A shows the apparatus of the invention for system calibration, utilizing a single side perspective view.

Figure 11B shows an example calibration pattern and example images of a calibration pattern acquired by the system, utilizing a single side perspective view, of the invention.

Figure 12A shows the apparatus of the invention for ball inspection utilizing a single side perspective view.

Figure 12B shows an example ball grid array and example images of the ball grid array for three dimensional inspection, utilizing a single side perspective view.

Figure 13 shows the apparatus of the invention for the three dimensional inspection of ball grid array devices, gullwing devices and J lead devices. Figure 14 shows the apparatus of the invention for the three dimensional inspection of parts utilizing three cameras.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT In one embodiment of the invention, the method and apparatus disclosed herein is a method and apparatus for calibrating the system by placing a pattern of calibration dots of known spacing and size on the bottom plane of a calibration reticle. From the precision dots the missing state values of the system are determined allowing for three dimensional inspection of balls on ball grid array devices, BGA devices or balls on wafers or balls on die. In one embodiment of the invention the system may also inspect gullwing and J lead devices as well as ball grid arrays .

Refer now to Figure 1A which shows the apparatus of the invention configured with a calibration reticle for use during calibration of the state values of the system. The apparatus obtains what is known as a bottom image 50 of the calibration reticle 20. To take the bottom image 50 the apparatus includes a camera 10 with a lens 11 and calibration reticle 20 with a calibration pattern 22 on the bottom surface. The calibration pattern 22 on the reticle 20 comprises precision dots 24. The camera 10 is located below the central part of the calibration reticle 20 to receive an image 50 described in conjunction with Figure IB. In one embodiment the camera 10 comprises an image sensor. The image sensor may be a charged coupled device array. The camera 10 is connected to a frame grabber board 12 to receive the image 50. The frame grabber board 12 provides an image data output to a processor 13 to perform a two dimensional calibration as described in conjunction with Figure 2A. The processor 13 may store an image in memory 14. The apparatus of the invention obtains an image of a pair of side perspective views and includes using a camera 15 with a lens 16 and a calibration reticle 20. The camera 15 is located to receive an image 60, comprising a pair of side perspective views, described in conjunction with Figure IB. Fixed optical elements 30, 32 and 38 provide a first side perspective view and fixed optical elements 34, 36, 38 for a second side perspective view. The fixed optical elements 30, 32, 34, 36 and 38 may be mirrors or prisms. As will be appreciated by those skilled in the art additional optical elements may be incorporated. The camera 15 is connected to a frame grabber board 17 to receive the image 60. The frame grabber board 17 provides an image data output to a processor 13 to perform a two dimensional inspection as described in conjunction with Figure 2B. The processor 13 may store an image in memory 14. In one embodiment of the invention, the apparatus may contain a nonlinear optical element 39 to magnify the side perspective image 60 in one dimension as shown in Figure 8A. In another embodiment of the invention optical element 38 may be a nonlinear element. The nonlinear optical elements 38 and 39 may be a curved mirror or a lens .

Figure IB shows an example image 50 from camera 10 and an example image 60 from camera 15 acquired by the system. The image 50, a bottom view of dot pattern 22, shows dots 52 acquired by camera 10. The dot pattern contains precision dots 24 of known dimensions and spacing. The precision dots 24 are located on the bottom surface of the calibration reticle 20. The image 60 shows two side perspective views of the dot pattern 22. A first side perspective view in image 60 contains images 62 of dots 24 and is obtained by the reflection of the image of the calibration reticle dot pattern 22 off of fixed optical elements 30, 32 and 38 into camera 15. A second side perspective view in image 60 contains images 66 of dots 24 and is obtained by the reflection of the image of the calibration reticle dot pattern 22 off of fixed optical elements 34, 36 and 38 into camera 15.

Optical element 36 is positioned to adjust the optical path length of a second side perspective view to equal the optical path length of a first side perspective view. Those skilled in the art will realize that any number of perspective views can be utilized by the invention. In one embodiment of the invention, ^' the maximum depth of focus of a side perspective view includes an area of the reticle including the center row of dots . This allows for a fixed focus system to inspect larger parts, with one perspective view imaging half of the part and the second perspective view imaging the other half of the part .

Figure 2A shows a flow diagram for the calibration of the bottom view of the system. The method starts in step 101 by providing a transparent reticle 20 having a bottom surface containing a dot pattern 22, comprising precision dots 24 of known dimensions and spacing. The method in step 102 provides a camera 10 located beneath the transparent reticle 20 to receive an image 50. In step 103 the processor 13 sends a command to a frame grabber 12 to acquire an image 50, comprising pixel values from the camera 10. The method then proceeds to step 104 and processes the pixel values with a processor 13. Figure 2B shows a flow diagram for determining the state values of the bottom view of the system. In step 111 the method begins by finding the dots 52 in image 50, corresponding to the calibration dots 24. The processor finds a dimension and position for each dot visible in image 50 in subpixel values using well known grayscale methods and stores these values in memory 14. By comparing these results to known values stored in memory, the processor calculates the missing state values for the bottom calibration in steps 112 and 113. In step 112 the processor 13 calculates the optical distortion of lens 11 and the camera roll angle with respect to the dot pattern 22. Step 113 calculates the pixel width and pixel height by comparing the subpixel data of dots 52 with the known dimensions of the precision dot pattern 22. The pixel aspect ratio is determined from the pixel width and pixel height. In step 114 the processor defines the X and Y world coordinates and the Z=0 plane from the image 50 of the precision dot pattern 22. The processor then stores these results in memory. These results provide conversion factors for use during analysis to convert pixel values to world values . Figure 2C shows a flow diagram for the calibration of the side perspective views of the system. The method starts in step 121 by providing a transparent reticle 20 having a bottom surface containing a dot pattern 22, comprising precision dots 24 of known dimensions and spacing. The method in step 122 provides fixed optical elements 30, 32, 34, 36 and 38 to reflect two perspective images of the precision dot pattern 22 into camera 15. The method in step 123 provides a camera 15 located to receive an image 60. In step 124 the processor 13 sends a command to a frame grabber 12 to acquire an image 60, comprising pixel values from the camera 15. The method then proceeds to step 125 and processes the pixel values with a processor 13. Figure 2D shows a flow diagram for determining the state values of the side perspective views of the system. In step 131 the method begins by finding dots 62 in image 60, corresponding to the calibration dots 24. The processor finds a dimension and position for each dot visible, comprising the group of dots 62, in image 60 for a first side perspective view in subpixel values and stores these values in memory 14. By comparing these results to known values stored in memory, the processor calculates the missing state values for a side perspective view, comprising the group of dots 62, in steps 132 and 133. In step 132 the processor 13 calculates the optical distortion of lens 16 and the camera roll angle with respect to the dot pattern 22. In step 133 the processor 13 calculates the pixel width and pixel height by comparing the subpixel data of dots 62 with the known dimensions of the precision dots 24. The pixel aspect ratio is determined from the pixel width and pixel height. In step 134 the processor defines the X and Y world coordinates and the Z=0 plane from the dots 62 in image 60 of the dot pattern 22. The processor then stores these results in memory. These results provide conversion factors for use during analysis to convert pixel values to world values. In step 135 the method of the invention computes the side view angle. In step 136 the method is repeated for a second side perspective view using the dots 66 in image 60.

Figure 2E shows the relationship of a side perspective angle to the ratio of the perspective dimension to the non-perspective dimension. Ray 171, 172, and 173 defining point 181 is parallel to ray 174, 175 and 176 defining point 182. Point 181 and point 182 lie on a plane 170 parallel to a plane 180. The intersection of ray 175 and ray 176 define point 186. The intersection of ray 176 and ray 172 define point 184. The intersection of ray 173 and ray 172 define point 187. The intersection of ray 174 and ray 172 define point 183. The reflecting plane 179 intersecting plane 180 at an angle D is defined by ray 172 and ray 175 and the law of reflectance. Ray 172 and ray 175 intersect plane 170 at an angle 177. Referring to Figure 2E it can be shown:

tanθ = — r r

Therefore: C = L

D_< D, cosθ =

D, cosθ

Substituting:

D cosθ D tanθ =

D_B D_Bcosθ

D_c

(tanθ) (cosθ) - - - sinθ

>„

Figure 2F shows a bottom view and a side perspective view of precision dots used in the method for determining a side perspective view angle 177 as shown in Figure 2E of the system. A bottom view image 200 comprising precision dots 201, 202 and 203 of known spacing and dimensions from the calibration method described earlier can be used to provide a reference for determination of a side perspective view angle 177. The value D_H and D_B are known from the bottom view calibration. A side perspective view image 210 comprising precision dots 211, 212 and 213, corresponding to bottom view dots 201, 202 and 203 respectively, of known spacing and dimensions D_s and D_h from the calibration method described earlier, can be used to determine the side view perspective angle. The ratio of (D_h / D_H) from the bottom image 200 and the side perspective image 210 can be used in the bottom view to calibrate D_B in the same units as the side perspective view as follows: D_Bcal= D_B(D_h/D_H)

Substituting into the equation for the side perspective view angle 177 described earlier yields:

.

θ = arcsin

Figure 3A shows the apparatus of the invention for a three dimensional inspection of the balls of a ball grid array. The apparatus of the invention includes a part 70 to be inspected. The apparatus further includes a camera 10 with a lens 11, located below the central area of part 70, to receive a bottom image 80, described in conjunction with Figure 3B, of part 70. The camera 10 is connected to a frame grabber board 12 to receive the image 80. The frame grabber board 12 provides an image data output to a processor 13 to perform a two dimensional inspection as described in conjunction with Figure 3A. The processor 13 may store an image in memory 14. The apparatus of the invention obtains an image of a pair of side perspective views with a camera 15 and a lens 16. The camera 15 is located to receive an image 90, comprising a pair of side perspective views, described in conjunction with Figure 3B and utilizing fixed optical elements 30, 32 and 38 for a first side perspective view and fixed optical elements 34, 36 and 38 for a second side perspective view. In one embodiment of the invention, the apparatus may contain a nonlinear optical element 39 to magnify the side perspective image 60 in one dimension as shown in Figure 8B . In another embodiment of the invention optical element 38 may be the nonlinear element. The fixed optical elements 30, 32, 34, 36 and 38 may be mirrors or prisms. As will be appreciated by those skilled in the art additional optical elements may be incorporated without deviating from the spirit and scope of the invention. The camera 15 is connected to a frame grabber board 17 to receive the image 90. The frame grabber board 17 provides an image data output to a processor 13 to calculate the Z position of the balls, described in conjunction with Figure 3B. The processor 13 may store an image in memory 14.

Figure 3B shows an example image 80 from camera 10 and an example image 90 from camera 15 acquired by the system. The image 80 shows the bottom view of the balls located on the bottom surface of a part 70. The image 90 shows two side view perspectives of the balls located on part 70. A first side perspective view in image 90 contains images of balls 91 and is obtained by the reflection of the image of the part 70 off of fixed optical elements 30, 32 and 38 into camera 15. A second side perspective view in image 90 contains images of balls 92 and is obtained by the reflection of the image of the part 70 off of fixed optical elements 34, 36 and 38 into camera 15. Optical element 36 is positioned to adjust the optical path length of a second side perspective view to equal the optical path length of a first side perspective view. In one embodiment of the invention, the maximum depth of focus of a side perspective view just includes an area of the part including the center row of balls. This allows for a fixed focus system to inspect larger parts, with one perspective view imaging at least half of the part and the second perspective view imaging at least the other half of the part . Those skilled in the art will realize that any number of perspective views can be utilized by the invention. In another embodiment of the invention, all of the balls are in focus from both side perspective views resulting in two perspective views for each ball. This permits two Z calculations for each ball as shown in conjunction with Figures 10A and 10B. Figure 4 shows a flow diagram for the three dimensional inspection of balls on a ball grid array. The method starts in step 141 by providing a part 70 having balls 71 facing down. The method in step 142 provides a camera 10 located beneath the part 70 to receive an image 80. In step 143 a frame grabber 12 is provided to receive the image 80 from camera 10. In step 144, fixed optical elements are provided for obtaining two side perspective views of the part 70. A first optical path is provided by optical elements 30, 32 and 38. A second optical path is provided by optical elements 34, 36 and 38. A second camera 15 receives an image 90 of two side perspective views in step 145. In step 146 a second frame grabber board 17 is provided to receive the image 90 from camera 15. A processor 13 sends a command to frame grabbers 12 and 17 to acquire images 80 and 90 comprising pixel values from cameras 10 and 15. The method then proceeds to step 147 and processes the pixel values with a processor 13 to obtain three dimensional data about part 70.

The invention contemplates the inspection of parts that have ball shaped leads whether or not packaged as a ball grid array. The invention also contemplates inspection of leads that present a generally curvilinear profile to an image sensor.

Figures 5A and 5B together show a flow chart of the three dimensional inspection method of the invention. The process begins in step 151 by waiting for an inspection signal. When the signal changes state, the system initiates the inspection. The processor 13 sends a command to frame grabber boards 12 and 17 to acquire images 80 and 90 respectively of part 70 having balls 71. In step 152, camera 10 captures an image 80 comprising pixel values and camera 15 captures an image 90 comprising pixel values and the processor stores the images in memory 14. The images comprise information from both a bottom view and two side perspective views as shown in Figure 3B. In step 153, the inspection system sends a signal to a part handler shown in Figure 9 to allow the part handler to move the part out of the inspection area and allows the next part to be moved into the inspection area. The handler may proceed with part placement while the inspection system processes the stored image data.

The inspection system processes the pixel values of the stored image 80 in step 154 to find a rotation, and X placement and Y placement of the part relative to the world X and Y coordinates. The processor determines these placement values finding points on four sides of the body of the part. In step 155, the processor employs a part definition file that contains values for an ideal part.

By using the measurement values from the part definition file and the placement values determined in step 154, the processor calculates an expected position for each ball of the part for the bottom view contained in image 80. The processor employs a search procedure on the image data to locate the balls 81 in image 80. The processor then determines each ball's center location and diameter in pixel values using grayscale blob techniques as described in Figure 7A. The results are stored in memory 14. The processor proceeds in step 156 to calculate an expected position of the center of each ball in both side perspective views in image 90 using the known position of each side view from calibration. The processor employs a subpixel edge detection method described in Figure 7B to locate a reference point on each ball in step 157. The results are stored in memory 14.

Now refer to Figure 5B . In step 158 the processor converts the stored pixel values from steps 154 and 157 into world locations by using pixel values and parameters determined during calibration. The world locations represent physical locations of the balls with respect to the world coordinates defined during calibration.

In step 159 the Z height of each ball is calculated in world coordinates in pixel values. The method proceeds by combining the location of the center of a ball from the bottom view 80 with the reference point of the same ball from a side perspective view in image 90 as described in Figures 6A and 6B . The processor then converts the world values to part values using the calculated part rotation, and X placement and Y placement in step 160 to define part coordinates for the ideal part. The part values represent physical dimensions of the balls such as ball diameter, ball center location in X part and Y part coordinates and ball height in Z world coordinates .

In step 161 these part values are compared to the ideal values defined in the part file to calculate the deviation of each ball center from its ideal location. In one embodiment of the invention the deviation values may include ball diameter in several orientations with respect to the X and Y part coordinates, ball center in the X direction, Y direction and radial direction, ball pitch in the X direction and Y direction and missing and deformed balls. The Z world data can be used to define a seating plane, using well known mathematical formulas, from which the Z dimension of the balls with respect to the seating plane can be calculated. Those skilled in the art will recognize that there are several possible definitions for seating planes from the data that may be used without deviating from the spirit and scope of the invention. In step 162 the results of step 161 are compared to predetermined thresholds with respect to the ideal part as defined in the part file to provide an electronic ball inspection result. In one embodiment the predetermined tolerance values include pass tolerance values and fail tolerance values from industry standards. If the measurement values are less than or equal to the pass tolerance values, the processor assigns a pass result for the part. If the measurement values exceed the fail tolerance values, the processor assigns a fail result for the part. If the measurement values are greater than the pass tolerance values, but less than or not equal to the fail tolerance values, the processor designates the part to be reworked. The processor reports the inspection result for the part in step 163, completing part inspection. The process then returns to step 151 to await the next inspection signal.

Figures 6A and 6B show an example ball of a ball grid array and associated geometry used in a method of the invention for determining the Z position of the ball. The method determines the Z position of a ball with respect to the world coordinates defined during calibration. Using parameters determined from the calibration procedure as shown in Figures 2B and 2D to define a world coordinate system for the bottom view and the two side perspective views, comprising world coordinate plane 250 with world coordinate origin 251 and world coordinate axis X 252, Y 253 and Z 254 shown in Figure 6A, and a pair of images 80 and 90 as shown in Figure 3B, the processor computes a three dimensional location.

Now refer to Figure 6A. The processor locates a point 258 on the world plane 250 determined by a bottom view ray 255 passing through the center 257 of a ball 71 on a part 70. The processor locates a side perspective view point 260 on the world plane 250 determined by a side perspective view ray 256 intersecting a ball reference point 259 on ball 71 and intersecting the bottom view ray 255 at a virtual point 261. Ray 256 intersects the world plane 250 at an angle 262 determined by the reflection of ray 256 off of the back surface 263 of prism 30. The value of angle 262 was determined during the calibration procedure . Now refer to Figure 6B. The distance _x is calculated by the processor as the difference between world point 258, defined by the intersection of ray 255 with the Z=0 world plane 250, and world point 260, defined by the intersection of ray 256 and the Z=0 world plane 250. The value Z is defined as the distance between world point 261 and 258 and is related to _x as follows:

tanθ. = —

Z = .tanθ-

Z can be computed by processor 13 since the angle 262 is known from calibration. The offset E 265 is the difference between the virtual point 261 defined by the intersection of ray 255 and ray 256 and the crown of ball 71 at point 264, defined by the intersection of ray 255 with the crown of ball 71, and can be calculated from the knowledge of the angle 262 and the ideal dimensions of the ball 71. The final value of Z for ball 71 is:

^Fmal = Z - E

Figure 7A shows one example of an image used in the grayscale blob method of the invention. The image processing method finds the location and dimensions of a ball 71 from a bottom image 80. From the expected position of a ball 71, a region of interest in image 80 is defined as (XI, Yl) by (X2,Y2) . The width and height of the region of interest are large enough to allow for positioning tolerances of part 70 for inspection. Due to the design of the lighting for the bottom view, the spherical shape of balls 71 of part 70 present a donut shaped image where the region 281, including the perimeter of the ball 71, comprises camera pixels of higher grayscale values and where the central region 282 comprises camera pixels of lower grayscale values. The remainder 283 of the region of interest 280 comprises camera pixels of lower grayscale values.

In one embodiment of the invention the processor 13 implements image processing functions written in the C programming language .

The C language function " FindBlobCenter" , as described below, is called to find the approximate center of the ball 71 by finding the average position of pixels that are greater than a known threshold value. Using the coordinates of the approximate center of the ball 71, the region 282 of lower grayscale pixel values can be converted to higher grayscale values by calling the C language function "FillBallCenter" , as described below. The exact center of the ball 71 can be found by calling the C language function "FindBallCenter" which also returns an X world and Y world coordinate . The diameter of the ball 71 can be calculated by the C language function, "Radius=sqrt (Area/3.14) " . The area used in the diameter calculation comprises the sum of pixels in region 281 and 282. Figure 7B shows one example of an image used with the method of the invention to perform a subpixel measurement of the ball reference point . The method of the invention finds a reference point on a ball 71 in an image 90 of a side perspective view as shown in Figure 3B. From the expected position of a ball 71, a region of interest 290 in image 80 is defined as (X3, Y3 ) by (X4,Y4). The width and height of the region of interest are large enough to allow for positioning tolerances of part 70 for inspection. Due to the design of the lighting for a side perspective view, the spherical shape of balls 71 of part 70 present a crescent shaped image 291 comprising camera pixels of higher grayscale values and where the remainder 293 of the region of interest 290 comprises camera pixels of lower grayscale values .

The C language function "FindBlobCenter" is called to compute the approximate center of the crescent image 291 by finding the average position of pixels that are greater than a known threshold value. Using the coordinates of the approximate center of the crescent image 291, the C language function "FindCrescentTop" is called to determine the camera pixel, or seed pixel 292 representing the highest edge on the top of the crescent. The camera pixel coordinates of the seed pixel are used as the coordinates of a region of interest for determining the subpixel location of the side perspective ball reference point . One example of grayscale blob analysis and reference point determination implemented in the C language is presented as follows:

/////////////////////////////////////////////////////

// // FindBlobCenter - finds the X,Y center of the pixels that have a value greater than THRESHOLD in the region (xl,yl) to (x2,y2)

/////////////////////////////////////////////////////

// long FindBlobCenter (int xl,int yl,int x2 , int y2 , double* pX, double* pY)

{ int x,y; long Found = 0 , long SumX = 0 long SumY = 0 for (x=xl;X<=x2;X++)

{ for (y=yl;y<=y2;y++)

{ if (Pixel [x] [y] > THRESHOLD)

{

SumX += x; SumY += y;

Found ++;

}

) ' if (Found > 0)

*pX = (double) SumX / (double) Found; *pY = (double) SumY / (double) Found;

} return Found; ///////////////////////////////////////////////////// //

// FillBallCenter - fills the center of the BGA "donut" ///////////////////////////////////////////////////// // void FillBallCenter (double CenterX, double CenterY, double Diameter)

{ int x,y; int xl = (int) (CenterX - Diamter / 4.0) int x2 = (int) (CenterX + Diamter / 4.0) int yl = (int) (CenterY - Diamter / 4.0) int y2 = (int) (CenterY + Diamter / 4.0) for (x=xl;X<=x2 ;X++)

{ for (y=yl,-y<=y2;y++)

{ Pixel [x] [y] = 255;

11111 / II Il^' l 11111111 II ! 11 / 1 ill HI 1111111111111 / 1 III 111 //

// FindBallCenter - finds the X,Y center of the a BGA // ball using the grayscale values

///////////////////////////////////////////////////// // long FindBallCenter (int xl,int yl,int x2 , int y2 , double* pX, double* pY, double* pRadius)

{ int x,y; long Found = 0 long Total = 0 long SumX = 0 long SumY = 0 for (x=xl;x<=x2;x++)

{ for (y=yl;y<=y2;y++)

{ if (Pixel [x] [y] > THRESHOLD) {

SumX += x*Pixel [x] [y] ;

SumY += y*Pixel [x] ϊyϊ ;

Total += Pixel [x] [y] ;

Found ++ ; }

} } if ( Found > 0 !

{

*pX (double) SumX / (double) Total ;

*pY (double) SumY / (double) Total ;

*pRadius sqrt ( (double) Found / 3.14159279);

return Found;

} ///////////////////////////////////////////////////// //

// FindCresentTop - finds the X,Y top position of a BGA cresent

///////////////////////////////////////////////////// // void FindCresentTop (int CenterX, int CenterY, int Diameter, int* pX,int* pY)

{ int x, y, Edge , Max, TopX, TopY; int xl = CenterX - Diamter / 2 int x2 = CenterX + Diamter / 2 int yl = CenterY - Diamter / 2, int y2 = CenterY;

*pY = 9999; for (x=xl ;x<=x2 ;x++)

{

Max = -9999; for (y=yl;y<=y2;y++)

{ Edge = Pixel [x] [y] - Pixel [x] [y-1] ; if (Edge > Max)

{

Max = Edge ;

TopY = y; TopX = x;

}

} if (TopY < *pY)

{ *pX = TopX;

*pY = TopY;

(c) 1997 Scanner Technologies Inc. Figure 8A shows a side perspective image of the calibration pattern magnified in one dimension. Figure 8A shows a side perspective image 300 of a reticle calibration pattern where the space 303 between dot 301 and dot 302 is magnified, increasing the number of lower value grayscale pixels when compared to a non magnified image.

Figure 8B shows a side perspective image of the balls on a BGA, magnified in one dimension. In Figure 8B a side perspective image 310 of two views are shown where the space 313 between ball image 311 and ball image 312 is magnified, increasing the number of lower value grayscale pixels when compared to a non magnified image. The increased number of lower grayscale value pixels allows for the successful application of the subpixel algorithm. In another embodiment of the invention, the method and apparatus disclosed herein is a method and apparatus for calibrating the system by placing a pattern of calibration dots of known spacing and dimensions on the bottom plane of a calibration reticle and for providing for two side perspective views of each ball for the three dimensional inspection of parts. From the precision dots the missing state values of the system are determined allowing for three dimensional inspection of balls on BGA devices or balls on wafers or balls on die. Figure 9 shows an example apparatus for presenting a BGA to the system for inspection. An overhead light reflective diffuser 5 includes a vacuum cup assembly 6. The vacuum cup assembly may attach to a BGA part 70 having balls 71 and suspend the BGA part 70 below the overhead light reflective diffuser 5.

Figures 10A and 10B show an example ball on a ball grid array and associated geometry for use with the method of the invention for determining the Z position of a ball with respect to the world coordinates defined during calibration, using two perspective views for each ball. Using parameters determined from the calibration procedure as shown in Figures 2B and 2D to define a world coordinate system for the bottom view and the two side perspective views, comprising world coordinate plane 700 with world coordinate origin 701 and world coordinate axis X 702, Y 703 and Z 704 shown in Figure 10A and Figure 10B, and a pair of images 80 and 90 as shown in

Figure 3B, the processor computes a three dimensional location.

Now refer to Figure 10A. The processor locates a point 709 on the world plane 700 determined by a bottom view ray 705 passing through the center 708 of a ball 717. The processor locates a first side perspective view point 711 on the world plane 700 determined by a side view ray 706 intersecting a ball reference point 710 on ball 717 and intersecting the bottom view ray 705 at a virtual point 714. Ray 706 intersects the world plane 700 at an angle 715 determined by the reflection of ray 706 off of the back surface of prism 30. The value of angle 715 was determined during the calibration procedure. The processor locates a second side perspective view point 713 on the world plane 700 determined by a side view ray 707 intersecting a ball reference point 712 on ball 717 and intersecting the bottom view ray 705 at a virtual point 718. Ray 707 intersects the world plane 700 at an angle 716 determined by the reflection of ray 707 off of the back surface of prism 34. The value of angle 716 was determined during the calibration procedure.

Now refer to Figure 10B. The distance L_x is calculated by the processor as the distance between world point 709 and world point 711. The distance L₂ is calculated by the processor as the distance between world point 713 and world point 709. The value Z_x is defined as the distance between world point 714 and 709 and is related to L_x as follows:

tan θ, = — ^Zι

Z_j = L_j tan θ_j

The value Z₂ is defined as the distance between world point 718 and 709 and is related to L₂ as follows:

tan θ„ =

Z₂ = I₂ tan θ₂

The average of Z_x and Z₂ are calculated and used as the value for Z of the ball. This method is more repeatable and accurate than methods that use only one perspective view per ball.

In still another embodiment of the invention, the method and apparatus disclosed herein is a method and apparatus for calibrating the system by placing a pattern of calibration dots of known spacing and dimensions on the bottom plane of a calibration reticle and for providing a single side perspective view for the three dimensional inspection of parts. From the precision dots the missing state values of the system are determined allowing for three dimensional inspection of balls on BGA devices or balls on wafers or balls on die.

Figure 11A shows the apparatus of the invention for system calibration, utilizing a single side perspective view. The method and apparatus for calibration of the bottom view is identical to the method and apparatus described earlier in Figure 2A and 2B for the two side perspective views method. The apparatus for an image of a side perspective view includes a camera 15 with a lens 18 and a calibration reticle 20. The camera 15 is located to receive an image 64 of a side perspective view comprising dots 65, described in conjunction with Figure 11B, and utilizing fixed optical elements 40 and 42. The fixed optical element 40 may be a mirror or prism. The fixed optical element 42 is a nonlinear element that magnifies the image in one direction. In another embodiment fixed optical element 40 may be this nonlinear element. As will be appreciated by those skilled in the art additional optical elements may be incorporated. The camera 15 is connected to a frame grabber board 17 to receive the image 64. The frame grabber board 17 provides an image data output to a processor 13 to perform a two dimensional inspection as described in conjunction with Figure 2B. The processor 13 may store an image in memory 14.

Figure 11B shows an example calibration pattern and example images of a calibration pattern acquired by the system, utilizing a single side perspective view, of the invention. Figure 11B shows an example image 50 from camera 10 and an example image 64 from camera 15 acquired by the system. The image 50 showing dots 52 acquired by camera 10 includes a bottom view of the dot pattern 22, containing precision dots 24 of known dimensions and spacing, located on the bottom surface of the calibration reticle 20. The image 64 shows a side perspective view of the dot pattern 22, containing precision dots 24 of known dimensions and spacing, located on the bottom surface of the calibration reticle 20. A side perspective view in image 64 contains images of dots 65 and is obtained by the reflection of the image of the calibration reticle dot pattern 22 off of fixed optical element 40, passing through nonlinear element 42 and into camera 15.

The side perspective calibration is identical to the method shown in Figure 2C except the fixed optical elements may have different properties. The determination of the state values for the side perspective view is identical to the method shown in Figure 2D except the fixed optical elements may be different and there is only one side perspective view. The principles and relationships shown in Figure 2E and Figure 2F apply.

In still another embodiment employing a single side perspective view, the invention does not include the nonlinear element 42.

Figure 12A shows the apparatus of the invention for ball inspection utilizing a single side perspective view. The apparatus of the invention includes a part 70 to be inspected. The apparatus further includes a camera 10 with a lens 11, located below the central area of part 70, to receive a bottom image 80, described in conjunction with Figure 12B, of part 70. The camera 10 is connected to a frame grabber board 12 to receive the image 80. The frame grabber board 12 provides an image data output to a processor 13 to perform a two dimensional inspection as described in conjunction with Figure 12B. The processor 13 may store an image in memory 14. The apparatus for an image of a single side perspective view includes a camera 15 with a lens 18. The camera 15 is located to receive an image 94, comprising a single side perspective view, described in conjunction with Figure 12B and utilizing fixed optical element 40 and nonlinear, fixed optical element 42, to magnify the side perspective view in one dimension. In another embodiment of the invention optical element 40 may be the nonlinear element. The fixed optical element 40 may be a mirror or prism. As will be appreciated by those skilled in the art additional optical elements may be incorporated. The camera 15 is connected to a frame grabber board 17 to receive the image 94. The frame grabber board 17 provides an image data output to a processor 13 to calculate the Z position of the balls, described in conjunction with Figure 12B. The processor 13 may store an image in memory 14. Figure 12B shows an example ball grid array and example images of the ball grid array for three dimensional inspection, utilizing a single side perspective view. Figure 12B shows an example image 80 from camera 10 and an example image 94 from camera 15 acquired by the system. The image 80 shows the bottom view of the balls 71 located on the bottom surface of a part 70. The image 94 shows a side perspective view of the balls 71 located on part 70. The side perspective view in image 94 contains images of balls 95 and is obtained by the reflection of the image of the part 70 off of fixed optical element 40 and passing through the nonlinear fixed element 42 into camera 15.

In an alternate embodiment of the invention, the system can be used to inspect other types of electronic parts in three dimensions, such as gullwing and J lead devices. By utilizing only one camera and adding an additional set of prisms on the reticle 400 these other devices may be inspected. The advantage of being able to inspect different devices with the same system includes savings in cost, and floor space in the factory. Additionally this design allows more flexibility in production planning and resource management . Figure 13 shows the apparatus of the invention for the three dimensional inspection of ball grid array devices, gullwing devices and J lead devices. The apparatus described in Figure 13 allows the inspection of BGA, gullwing and J lead devices all on the same system. The apparatus includes a part 402 to be inspected located over the central area of a transparent reticle 400 with prisms 401 glued to the top surface to receive side perspective views of part 402. A gullwing and J lead inspection device 21 may be integrated into the ball grid array inspection device. One example embodiment of such a gullwing and J lead inspection device is the "UltraVim" scanner from Scanner Technologies of Minnetonka, Minnesota. The apparatus further includes a camera 10A with a lens 11A, located below the central area of part 402 and reticle 400 to receive a bottom view and side perspective views of part 402. The camera 10A is connected to a frame grabber board 12A to receive an image . The frame grabber board 12A provides an image data output to a processor 13A to perform a three dimensional inspection of part 402. The processor 13A may store an image in memory 14A. These components comprise the hardware of the gullwing and J lead inspection device 21 and are shared by the ball grid array inspection device as described herein.

The UltraVim is described in United States patent application Serial Number 08/850,473 entitled THREE DIMENSIONAL INSPECTION SYSTEM by Beaty et al . , filed May 5, 1997 which is incorporated in its entirely by reference thereto.

Refer now to Figure 14. In still another embodiment of the invention, the system may use three cameras to image directly the bottom view and two side perspective views as shown in Figure 14. Figure

14 shows the apparatus of the invention for a three dimensional inspection of the balls of a BGA. The apparatus of the invention includes a part 70, with balls 71 to be inspected. The apparatus further includes a camera 10 with a lens 11, located below the central area of part 70, to receive a bottom image 80, described in conjunction with Figure 12B, of part 70. The camera 10 is connected to a frame grabber board 12 to receive the image 80. The frame grabber board 12 provides an image data output to a processor 13 to perform a two dimensional inspection as described in conjunction with Figure 12B. The processor 13 may store an image in memory 14. The apparatus for an image of a first side perspective view includes a camera 15 with a lens 19. The camera

15 is located to receive an image 94, comprising a single side perspective view, described in conjunction with Figure 12B and utilizing fixed optical element 38, to magnify the side perspective view in one dimension. The camera 15 is connected to a frame grabber board 17 to receive the image 94. The frame grabber board 17 provides an image data output to a processor 13 to calculate the Z position of the balls, described in conjunction with Figure 12B. The processor 13 may store an image in memory 14. The apparatus for an image of a second side perspective view includes a camera 15 with a lens 19. The camera 15 is located to receive an image similar to 94, comprising a single side perspective view, described in conjunction with Figure 12B and utilizing fixed optical element 38, to magnify the side perspective view in one dimension. The camera 15 is connected to a frame grabber board 17 to receive the image similar to 94. The frame grabber board 17 provides an image data output to a processor 13 to calculate the Z position of the balls, described in conjunction with Figure 12B. The processor 13 may store an image in memory 14. In another embodiment, the nonlinear fixed optical element 38 may be missing. In still another embodiment of the invention, only one side perspective view may be utilized.

The invention has been described herein in considerable detail in order to comply with the Patent Statutes and to provide those skilled in the art with the information needed to apply the novel principles and to construct and use such specialized components as are required. However, it is to be understood that the invention can be carried out by specifically different equipment and devices, and that various modifications, both as to the equipment details and operating procedures, can be accomplished without departing from the scope of the invention itself . What is claimed is:

Claims

1. An apparatus for inspecting a ball grid array

(70) , wherein the apparatus is calibrated using a precision pattern mask (22) with dot patterns (24) deposited on a calibration transparent reticle (20) , the apparatus for inspecting a ball grid array (70) comprising: (a) a means for mounting the ball grid array

(70) ; (b) a means for illuminating the ball grid array (70) to provide an outline of the ball grid array;

(c) a first camera (10) positioned to image the ball grid array to provide a first image (80) of the ball grid array (70) ;

(d) a first means for light reflection (34) positioned to reflect the ball grid array through a second means for light reflection (36) into a second camera (15) , wherein the second camera (15) provides a second image

(90) of the ball grid array (70) ;

(e) a third means for light reflection (30) positioned to reflect an opposite side view of the ball grid array into a fourth means for light reflection (32) and into the second camera (15) as part of the second image (90) of the ball grid array (70) ; and

(f) a means for image processing (13) the first image (80) and second image (90) of the ball grid array (70) to inspect the ball grid array (70) .

2. The apparatus of claim 1 wherein the means for illuminating the ball grid array (70) further comprises a light source for illuminating the ball grid array (70) , and an overhead light reflective diffuser (5) provides illumination of the ball grid array from the light source.

3. The apparatus of claim 1 wherein the first means for light reflection further comprises a mirror

(32, 36) .

. The apparatus of claim 1 wherein the first means for light reflection further comprises a prism (30, 34) .

5. The apparatus of claim 1 wherein the first means for light reflection further comprises a curved mirror (38) .

6. The apparatus of claim 1 wherein the second means for light reflection further comprises a mirror (32, 36) .

7. The apparatus of claim 1 wherein the second means for light reflection further comprises a prism (30 , 34) .

8. The apparatus of claim 1 wherein the second means for light reflection further comprises a curved mirror (38) .

. The apparatus of claim 1 wherein the ball grid array (70) further comprises balls on a wafer.

10. The apparatus of claim 1 wherein the ball grid array (70) further comprises balls on a die.

11. The apparatus of claim 1 wherein the first camera (10) is connected to a frame grabber board (12) to receive the first image (80) .

12. The apparatus of claim 11 wherein the frame grabber board (12) provides an image data output to a processor (13) to perform a two dimensional inspection.

13. The apparatus of claim 1 further comprising a nonlinear optical element to magnify the second image in one dimension (39) .

14. The apparatus of claim 1 wherein an optical path length (34, 36, 38) of a second side perspective view is equal to an optical path length (30, 32, 38) of a first side perspective view.

15. The apparatus of claim 1 wherein a maximum depth of focus of a side perspective view allows for a fixed focus system to inspect larger ball grid arrays, with one perspective view (80) imaging one portion of the ball grid array and a second perspective view (90) imaging a second portion of the ball grid array (70) .

16. The apparatus of claim 15 wherein one portion of the ball grid array is at least a first half of the ball grid array and the second portion of the ball grid array is at least a second half of the ball grid array (90) .

17. The apparatus of claim 1 wherein a maximum depth of focus of a side perspective view includes an area of the ball grid array including a center row of balls (90) .

18. The apparatus of claim 1 wherein all of the balls on the ball grid array are in focus from both side perspective views resulting in two perspective views for each ball (90) .

19. The apparatus of claim 1 further comprising a third camera, wherein the first camera (10) enables direct imaging of a bottom view, wherein the first camera is located below a central area of the ball grid array (70) , wherein the third camera (15) is located to receive an image of a single side perspective view and uses fixed optical elements (19, 38) to magnify the single side perspective view in one dimension, wherein the second camera (15) is positioned to image a second side perspective view and uses fixed optical elements (19, 38) to magnify the second side perspective view in one dimension, and wherein the means for image processing (13) calculates a Z position of a ball on the ball grid array from the bottom view, first side perspective view and second side perspective view.

20. The apparatus of claim 1 further comprising a means for inspecting gullwing and J lead devices (402) .

21. A method for three dimensional inspection of a lead on a part mounted on a reticle, the method comprising the steps of :

(a) locating a first camera to receive an image of the lead (142) ;

(b) transmitting an image of the lead to a first frame grabber (143) ; (c) providing fixed optical elements to obtain two side perspective views of the lead

(144) ;

(d) locating a second camera to receive an image of the two side perspective views of the lead (145) ;

(e) transmitting the two side perspective views of the lead to a second frame grabber

(146) ; (f) operating a processor to send a command to the first frame grabber and second frame grabber to acquire images of pixel values from the first camera and the second camera (147) ; and (g) processing the pixel values with the processor to obtain three dimensional data about the lead (147) .

22. The method of claim 21 wherein the lead is a curved surface lead.

23. The method of claim 21 wherein the lead is a ball (71) .

24. The method of claim 21 wherein the part is a ball grid array (70) .

25. The method of claim 21 wherein the step of processing the pixel values with the processor (144) further comprises the step of sending a signal to a part handler to allow the part handler to move the part out of an inspection area and allow a next part to be moved into the inspection area (153) .

26. The method of claim 21 wherein the processor processes the pixel values to find a rotation, an X placement value and a Y placement value of the part relative to world X and Y coordinates by finding points on four sides of the part

(154) .

27. The method of claim 26 further comprising the steps of :

(a) using a part definition file that contains measurement values for an ideal part (155) ;

(b) calculating an expected position for each lead of the part for a bottom view using the measurement values from the part definition file and the X placement value and Y placement value (155) .

28. The method of claim 21 further comprising the step of using a search procedure (Fig. 7A) on the image data to locate the lead (155) .

29. The method of claim 21 further comprising the step of determining a lead center location and a lead diameter in pixels and storing the lead center location and lead diameter in memory

(155) .

30. The method of claim 29 further comprising the step calculating an expected position of a center of each lead in both side perspective views in the image using a known position of each side view from calibration (156) .

31. The method of claim 21 further comprising the step using a subpixel edge detection method (Fig. 7B) to locate a reference point on each lead (157) .

32. The method of claim 27 further comprising the step of converting the pixel values into world locations by using pixel values and parameters determined during calibration wherein the world locations represent physical locations of the lead with respect to world coordinates defined during calibration (15) .

33. The method of claim 32 wherein a Z height of each lead is calculated in world coordinates in pixel values by combining a location of a center of a lead from a bottom view with a reference point of the same lead from a side perspective view (159) .

34. The method of claim 33 further comprising the step of converting the world values to part values using the rotation, the X placement value and the Y placement value to define part coordinates for the ideal part where the part values represent physical dimensions of the lead including lead diameter, lead center location in X part and Y part coordinates and lead height in Z world coordinates (160) .

35. The method of claim 34 further comprising the step of comparing ideal values defined in the part file to calculate deviation values that represent a deviation of the center of the lead from its ideal location (161) .

36. The method of claim 35 wherein the deviation values may include lead diameter in several orientations with respect to the X placement value and Y placement value, lead center in the X direction, Y direction and radial direction, lead pitch in the X direction and Y direction and missing and deformed leads, further comprising the step of calculating the Z dimension of the lead with respect to the seating plane based on the Z world data (161) .

37. The method of claim 35 further comprising the step of comparing the deviation values to predetermined tolerance values with respect to an ideal part as defined in the part definition file to provide a lead inspection result (162) .

38. The method of claim 37 wherein the predetermined tolerance values include pass tolerance values and fail tolerance values from industry standards (162) .

39. The method of claim 38 further comprising the steps of:

(a) determining if the deviation values are less than or equal to the pass tolerance values, wherein the processor assigns a pass result for the part (162) ; (b) determining if the deviation values exceed the fail tolerance values, wherein the processor assigns a fail result for the part (162) ; and (c) determining if the deviation values are greater than the pass tolerance values, but less than or equal to the fail tolerance values, wherein the processor designates the part to be reworked (162) .

40. The method of claim 38 further comprising the step of reporting a part inspection result for the part (163) .

41. A method to inspect a ball grid array device comprising the steps of:

(a) locating a point (258) on a world plane (250) determined by a bottom view ray (255) passing through a center (257) of a ball (71) on the ball grid array device (70) ;

(b) locating a side perspective view point

(260) on the world plane (250) determined by a side perspective view ray (256) intersecting a ball reference point (259) on the ball (71) and intersecting the bottom view ray (255) at a virtual point

(261) where the side perspective view ray (256) intersects the world plane (250) at an angle (262) determined by a reflection of the side perspective view ray (256) off of a back surface (263) of a prism (30) where a value of the angle (262) was determined during a calibration procedure;

(c) calculating a distance L_x as a difference between a first world point (258) , defined by an intersection of the bottom view ray (255) with a Z=0 world plane (250) , and a second world point (260) , defined by the intersection of the side perspective view ray (256) and the Z=0 world plane (250) , where a value Z is defined as a distance between a third world point (261) and first world point (258) and is related to h as follows tan╬╕, = ΓÇö ¹

L_j tan╬╕,

wherein Z is computed based on the angle; (d) computing an offset E (265) as the difference between the virtual point (261) defined by the intersection of the bottom view ray (255) and the side perspective view ray (256) and a crown of a ball (71) at a crown point (264) that is defined by the intersection of the bottom view ray (255) with the crown of the ball (71) , and can be calculated from a knowledge of the angle (262) and ideal dimensions of the ball (71) where a final value of Z for the ball (71) is:

Spi_nal = Z - E .

42. A method for finding a location and dimensions of a ball in a ball grid array from a bottom image, the method comprising the steps of:

(a) defining a region of interest (280) in the bottom image of an expected position of a ball (71) where a width and a height of the region of interest are large enough to allow for positioning tolerances of the ball grid array (70) for inspection;

(b) imaging the ball (71) , wherein the ball (71) is illuminated to allow a spherical shape of the ball (71) to present a donut shaped image, wherein the region of interest (280) includes a perimeter (281) of the ball (71) wherein the bottom image comprises camera pixels of higher grayscale values and where a center (282) of the bottom image comprises camera pixels of lower grayscale value and wherein a remainder of the region of interest (283) comprises camera pixels of lower grayscale values;

(c) finding an approximate center of the ball (71) by finding an average position of pixels having pixel values that are greater than a predetermined threshold value;

(d) converting the region of lower grayscale pixel values to higher grayscale values using coordinates of the approximate center of the ball (71) ; and

(e) finding the center of the ball (71) .

43. The method of claim 42 further comprising the step of determining a diameter of the ball (71) .

44. The method of claim 43 wherein the diameter of the ball (71) is equal to (2 * sqrt ( Area/3.14)) wherein the Area is equal to a sum of pixels in the region.

45. A method for finding a reference point on a ball

(71) in an image (90) of a side perspective view of a ball grid array, the method comprising the steps of :

(a) defining a region of interest (290) in the image (80) from an expected position of a ball wherein a width and a height of the region of interest are large enough to allow for positioning tolerances of the ball grid array (70) ; (b) imaging the ball, wherein the ball is illuminated to allow a spherical shape of the ball to present a crescent shaped image

(291) having camera pixels of higher grayscale values, and wherein a remainder

(293) of the region of interest (290) comprises camera pixels of lower grayscale values;

(c) computing an approximate center of the crescent shaped image (291) by finding an average position of pixels that are greater than a predetermined threshold value;

(d) using coordinates of the approximate center of the crescent (291) ; (e) determining a camera pixel as a seed pixel

(292) representing a highest edge on a top of the crescent shaped image (291) ; and

(f) determining a subpixel location of the reference point based on the camera pixel coordinates of the seed pixel (292) that define coordinates of a region of interest (290) for the seed pixel.

46. The method of claim 45 wherein in a side perspective image (90) of a reticle calibration pattern (22) , space between dot images is magnified, increasing a number of higher value grayscale pixels when compared to a nonmagnified image .