Reading time ( words)
One AOI solution constitutes a radical departure from existing vision technology in that it senses and measures the Z-axis height of features, in addition to the X/Y 2D image plane, and combines them. Integrated 3D and 2D AOI improves defect detection and mitigates escapes and false calls.
By Kwangill Koh, Ph.D., Koh Young Technology
The basic operation of all automated optical inspection (AOI) systems, to date, has essentially been the same 2D imaging, primarily grayscale. This forms the basis for the system’s analysis and defect determination. For example, flagging or identifying lifted leads (resulting in opens) is determined by examination of a pad’s fillet area. This is not reliable, however, with smaller, fine pitch devices, where there may be less area available for the machine to inspect and measure. Similarly, lifted components and lifted pads are not easily identified due to the lack of ability to measure a given object or feature’s height. This weakness often results in escapes, or undetected defects, that impact yields, exacerbate rework, and determinately affect the manufacturing process and finished product reliability.
Optimizing a manufacturing process for highest yields first requires that the AOI machine find all soldering defects. For AOI, the number of defects found is not the problem; it is the number not found that creates issues. Detected defects can then be addressed and solved at the appropriate step in the process. The defects that get past the AOI machine are the source of ongoing process headaches and compromised product reliability. Because defect detection is in many cases doubtful, without clear yes/no on a defect, AOI systems generate many false calls. Those false calls require skilled operator intervention, adding to delays and costs, and impacting long-term product reliability.
2D vs. 3D AOI
2D inspection is like viewing a landscape through only one eye. There is no depth perception. Newly developed 3D AOI technology allows Z-axis measurement of components, leads, and other assembled PCB topography features, clearly identifying non-conforming items, and eliminating this basic weakness of 2D imaging. 3D measurement immediately shows if one lead or one side of a component is higher than the others, i.e., not planar to the surface of the board. This is especially effective for detecting lifted leads. 3D AOI also allows significantly more accurate shape measurement of fillets, which may be critical in many applications. As a result, many types of defects that have been problematical for 2D AOI are now readily flagged using 3D AOI. These include missing, flipped, skewed, lifted, or tombstoned components; overhang; bridging and shorts; opens; poor solder fillets; coplanarity; component dimensioning; markings; and others in addition to lifted objects.
Real 3D measurement of solder joints was not possible until now. Problems associated with the implementation of 3D measurement had included shadowing, measuring range, and erroneous data from warped PCBs. Additionally, any light measuring system is confronted with specular, reference plane shadow, and directional problems that must also be solved before a useful and practical operating system can be put into production.
Figure 1. Full profilometric measurement of an object (or component) is possible using 3D AOI.
One solution constitutes a radical departure from existing AOI vision technology in that it senses and measures the vertical or Z-axis height of features being inspected, in addition to the typical X/Y 2D image plane, and combines both (Figure 1). This added Z-dimension measurement solves many of the vulnerabilities of 2D AOI and makes it possible to detect all soldering defects on an assembly, preventing escapes. 3D AOI technology is basic in concept. The AOI imaging head uses a series of eight phase-shift-Moiré illuminators mounted in a circular configuration to pick up height data from all sides and the top of the device or feature. It also contains multiple circular rows of LEDs positioned at different heights and angles to the part, each ring firing in succession in different colors. The camera acquires the varying images of pattern, color, and angle of each projection onto the part and thus can construct accurate 3D images and measure heights. This configuration enables shadow-free imaging as well as precise position, shape, and height measurement of the feature under inspection.
Lifted Leads and Depth Perception
With 2D inspection, the height of objects cannot be determined without some foreknowledge by the operator making the observation, as in an aerial photo of, for example, a neighborhood. Since the peaked roofs of houses in an aerial photo all look the same from the sky, there is no way to tell, without a side view or element of depth perception, whether or not the buildings are single story or multi-stories in height. A human observer examining such a photo may be able to correctly make such distinctions based on many subtleties of personal judgment, whereas a machine is far less capable. The same is true for lifted leads and components. 2D AOI cannot determine whether or not these components are at the proper height for an object connected to the PCB. 3D measurement immediately shows if one lead or one side of a component is higher than the others. This is especially effective when detecting lifted leads. 2D AOI uses fillet shape to determine the status of leads, and with very small fillets, it is inadequate.
3D Lifted Components
AOI also allows far more accurate shape measurement of fillets, which proves critical in many cases. The example of a lifted component demonstrates how 3D AOI can flag a defect that would otherwise pass through 2D AOI. As Figure 2 illustrates, a lifted component cannot be detected using 2D AOI because 2D systems cannot measure height variations across the surface of a component. A lifted component is a serious defect requiring rework, as well as investigation of its cause in the manufacturing process. 3D AOI not only shows the obvious differences in height across the component – meaning it can flag lifts as a defect, not a false call, and certainly not an escape – it also accurately measures the telltale variation in height (Figure 2).
Similarly, lifted leads constitute a serious defect that 2D AOI has trouble flagging. 2D AOI makes a judgment about whether a lead is lifted by examining the fillet area, for lack of other criteria. Figure 3 shows an example of the area under examination. However, if the fillet area is small, it’s more difficult to inspect accurately in 2D. The photo (3A) shows some leads slightly lifted off the pads, resulting in opens, which would be difficult to detect. In this case, 3D profilometry clearly shows that the leads are lifted and constitute a defect. The Z-axis measurement capability has made much more information available to AOI, enabling it to make an accurate judgment.
Figure 2. A lifted component (A) is not detectable using 2D AOI because the height differential between the sides of the component cannot be measured. 3D height measurement (B) shows that the height is not uniform.
3D AOI Compares Images with CAD Data, IPC 610
The real value of 3D AOI is inherent in how it functions: template comparison based on the IPC 610 Acceptability Standard. This factor removes any question of the inspected object’s acceptability or non-acceptability, and does not require a decision-making intervention on the part of the operator. The operator can be a trained technician who does not need to be well-schooled to make go/no go defect decisions or interventions in the inspection process. Machine calls based on a globally accepted standard make programming much simpler, based on CAD data. Coplanarity issues, component polarity, lifted leads, tombstoning, and other defects are quickly and unmistakably detected and confirmed based on these two sources of authoritative reference.
Figure 3A. Small fillet area subjected to 2D AOI makes lifted leads difficult to detect.
Figure 3B. Full profilometric measurement with 3D measurement clearly makes the defect obvious.
Manufacturing engineers using AOI are not concerned about the defects that their machines find, as much as they are about the escapes that the machine allows. The consequences of trying to push machines beyond these limitations imposed by the inadequacies of 2D imaging data are false calls. The solution is improved data – 3D imaging and measurement – from adding another dimension (Z-axis) to the 2D information already used in AOI. 3D AOI effectively negates all of the shortcomings of traditional 2D AOI and eliminates virtually all escapes. It also prevents false calls caused by shadowing and specular problems that have long been the primary vulnerabilities of 2D AOI.
Figure 3C. Lifted leads are difficult to detect with top-down visual inspection, as leads line up over pads.
Kwangill Koh, Ph.D., CEO, Koh Young Technology, 14F ACE Techno X, 470-5 Gasan-dong, Geumcheon-gu, Seoul, Korea 153-789, may be contacted at +82 26670 5005; email@example.com; www.kohyoung.com.