Friday, May 27, 2016
HOME / CURRENT ISSUE >  Industry Articles > 

Apply Line Scan Technology for Improved AOI
This diagram shows how multiple LEDs aid in illuminating a PCB for a quick, full-surface inspection.

Automated optical inspection (AOI) is a critical part of ensuring high quality in production printed-circuit boards (PCBs). But to ensure quality, many AOI equipment suppliers simply will add an additional downward-looking, field-of-view (FOV) camera, circled by a ring light to help detect any problems with solder quality or other PCB defects.

  • How many cameras are enough? Some AOI systems employ as many as eight cameras.
  • How does an AOI system convert the images from these multiple cameras into usable inspection data? Many AOI systems use a "golden" image from a reference area of a PCB as a model for comparison to an area under study. But problems can occur with differences in color or lighting between the area of study and the reference, or board-to-board or lot-to-lot variations. Will the golden image work anywhere on the PCB, in any orientation? Does a multiple camera system take more time to program and debug? Does increasing the number of cameras increase the number of false failures or false calls?
  • How sensitive is calibration and alignment of an AOI system with multiple cameras?
  • How reliable is an AOI system with additional cameras? Does additional and more sensitive hardware make an AOI system more prone to mechanical and electrical problems in a production environment?
  • Does adding more cameras (and image data to process) impact the inspection speed?
  • Do the additional cameras and supporting hardware add to the price and operating cost of an AOI system?
In conventional AOI systems, the regions for inspection are limited to only where the cameras are positioned.

Line scan technology (LST) may represent a realistic alternative to multiple-camera AOI systems. In an AOI system with LST, a charge-coupled-device (CCD) array sensor rapidly scans the surface of a PCB, capturing its complete image. For an 18 x 20-in. (460 x 550mm) PCB, this requires about 10 seconds. The scanning speed is not affected by the number of components on the PCB, and smaller PCBs require less time to scan. For commercial AOI systems with LST from Saki Corp. (, the lighting system — which the company refers to as its Co-axial Toplight concept — contains more than 3000 light-emitting diodes (LEDs), arranged in multiple banks, for toplight, sidelight, and low-light illumination. During each scan, the LED lighting is modulated several thousand times to produce more than 20 different lighting schemes, to provide flexibility for finding the required lighting contrast for any inspection issue.
The Co-axial Toplight approach helps to differentiate (a) normal solder, (b) cold or missing solder, and (c) lack of solder by the amount and quality of the reflected light.

Conventional AOI systems, whether with or without side cameras, employ a downward-looking, large-area-array CCD camera. The camera is usually surrounded by ring lighting projected at an oblique angle. To inspect a PCB, the camera is positioned over multiple areas of interest, gathering snapshots. Depending on the camera's FOV, the size of the PCB, and its component density, data acquisition can take from a few seconds for small sparsely populated assemblies to several minutes for large, densely populated boards. In comparison, an LST AOI system scans at a constant speed, covering the complete PCB in one single, rapid scan.

Best Light Source
For effective illumination and solder inspection of PCBs with components having different heights, Saki found the best light source to be its Co-axial Toplight approach, projected directly perpendicular to the PCB surface. Off-axis illumination, especially with components having different heights, would create shadows, possibly increasing the number of false calls for the AOI system. The firm also discovered that good solder fillets reflected this overhead lighting much differently than cold solder or no solder, and that the differences in reflections could be used to distinguish between solder that was "good" or "no good." In conjunction with its Co-axial Toplight illumination approach, Saki added telecentric optics, increasing the sensor's depth of field, and removing parallax errors associated with other highly focused optical designs. The approach virtually eliminated shadowing and optical distortion, and made the use of component inspection libraries more effective.
The Co-axial Toplight approach is also helpful in differentiating good leads (top) from bent leads (bottom).

The Co-axial Toplight technique works well for solder fillet inspection, but other inspection tasks require different projection angles and illumination color for the highest contrast between "G" (good) and "NG" (no good) conditions. Tasks such as optical character recognition (OCR) and optical character verification (OCV) may require different projection angles — as well as data from multiple projection angles — to accurately determine that the correct part has been placed. In some cases, the color of the component is used to detect presence/absence or the correct part. Color inspection usually requires another unique illumination setting. At times, the contrast of color and/or reflectivity between component and substrate is subtle and may require a special lighting setting.

Off-Axis Lighting
To augment the top lighting from the Co-axial Toplight approach, Saki added banks of LEDs for off-axis side lighting — its "Sidelight" setting, effective for color inspection — and for low light conditions — its "Lowlight" setting, effective for OCR and OCV inspection. At times, combinations of the different lighting options provided superior results. But rather than add more lighting hardware, the company applied software to generate different lighting schemes in its AOI systems, with more than 20 lighting schemes available and no impact on scanning time.

Conventional AOI systems with FOV or FOV with side cameras inspect only those areas on the X/Y plane of the PCB that can be reached by the cameras. These areas are usually limited to portions of the PCB that are heavily populated, since taking multiple FOV snapshots to cover the whole PCB area increases tact time. Unfortunately, inspecting only heavily populated areas leaves large areas of the PCB that are not inspected, potentially with loose chips, solder balls or other problems going undetected and leading to production problems later on.

One Quick Scan
An AOI system with LST captures an image of the entire PCB in one quick scan. To guard against foreign materials evading detection and passing to the next PCB process, the LST-based AOI systems developed by Saki use an Extra Component Detection (ECD) algorithm to compare 10 known good sample images from a database with the inspected image, looking for irregularities across the entire PCB surface. With such AOI systems, users can select size boundaries to focus only on expected problems, and to minimize false calls.

The use of these algorithms has helped with the inspection of such PCB parameters as solder quality and component leads. For example, for solder fillets properly formed, light projected perpendicular to the PCB will reflect off at an angle, generating a "low brightness" condition, and produce a dark area where the fillet is formed. For a case with no fillet, as with insufficient solder, light will reflect in an equal but opposite direction, as in a mirror, creating a "high brightness" condition. In the software inspection libraries that support the physical inspection, high and low thresholds for brightness can be set for each inspection window. In addition, several windows can be used for one inspection task, and configured with conditional jumps, for faster program and de-bug times, and lower false call rates.

Detecting lifted lead conditions in a PCB's component might be the driving force behind implementation of side view cameras in AOI systems. But as Saki has learned with its LST AOI systems, lifted leads can be accurately detected without the additional hardware, software, and operating expense of side-view cameras.

From a perpendicular, overhead view, a lifted lead will appear to have a different length than properly attached leads. By comparison to benchmarks or references such as a measured length value, or comparing the length of an inspected lead to its neighboring leads, a pass or fail or okay ("OK") or no good ("NG") condition can be registered. Using software algorithms, the acceptable thresholds can be selected by the AOI system user, providing a robust inspection while reducing false calls. Another symptom of a lifted lead is the quality of the reflowed solder on the pad. The reflection of light from the remaining pad of a good solder fillet and that of a pad with no lead attached will be very different, enabling optical inspection.

Simple Accurate Results
The LST approach to AOI system design is simple yet elegant, providing accurate results without the use of multiple cameras and sensors which can add system cost, programming time, and reliability issues. LST-based AOI systems use a combination of algorithms to provide accurate, high-speed inspection capability. In today's 'rough and tumble' manufacturing environments, production tools that are proven to be powerful yet simple are keys to efficiency and economy.

Contact: ASC International, 830 Tower Drive, Suite 200 Medina, MN 55340 763-479-6210 fax: 763-479-6206 Web: or

search login