2-D and 3-D Inspections Catch Solder-Paste Problems

Machines prove better than people at making repetitive quantitative measurements, so automated optical inspection (AOI) now plays a key role in ensuring the quality of assembled boards. Repair of an SMT assembly after solder-paste deposition costs 10 times less than repair after component placement, and 50 times less than after in-circuit test. By identifying problems quickly through the use of automated-inspection systems, test engineers and technicians can correct quickly any solder-paste deposition problems.¹ Catching problems and reducing rework further down the production line saves time and money.

By measuring solder-paste area, height, placement, and volume, you can identify specific types of visible defects and link them with specific corrective actions (Table 1). The table lists the primary type of solder-paste inspection systems you would use to automatically obtain paste-deposit information. A 2-D video camera system can quickly measure characteristics such as paste area and paste position, and a 3-D triangulation system can measure paste characteristics such as height and volume.

In our experience, you must perform at least a 2-D inspection of all solder-paste deposits to help control future solder-joint failures. When production requires more stringent process control, say for fine-pitch PCBs that includes small-volume solder joints for chip-scale packages (CSPs), you must also use 3-D measurements. The small sizes of the contacts on these fine-pitch boards and components require tighter margins of error for solder placement and solder-paste characteristics.

Recent research describes a relationship between solder-joint volume and joint-fatigue failure.² On average, the number of heating-cooling thermal cycles prior to a failure is proportional to V^~2, where V represents the volume in cubic mils. Thus, smaller solder joints are less reliable than larger solder joints, so the reliability of the small joints is more sensitive to small variations in solder volume. The volume of a solder joint can vary tenfold from CSP to larger SMT components, so it is important to accurately measure the volumes for the smallest paste deposits.

Accurate volume measurements require a 3-D inspection system, usually based on a laser, that can measure x, y, and z coordinates. A high-speed production line can outpace a 3-D system, so laser-based systems cannot measure the characteristics of every solder-paste deposit. Instead, they sample a subset of all solder-paste deposits. Manufacturers of 3-D systems offer a variety of technologies that increase scanning rates (Fig. 1), but sampling is still the norm.

Figure 1. Solder-paste inspection systems take several approaches: a) a basic 3-D laser triangulation, b) a high-speed acousto-optic laser deflector that speeds scanning, and c) a hybrid approach that uses 3-D multi-axis scanning and a basic 2-D camera with a ring light.

Laser-Systems Need a Reference

How well laser-based inspection systems work also depends on factors such as how accurately the laser system can locate a reference plane and how transparent the PCB materials are at the laser’s wavelength. By measuring the zero-height reference plane, usually the height of pads and traces, the system can correct for any PCB warping. The reference plane provides the starting “base” for solder-paste volume measurements, and it lets the system make accurate and repeatable 3-D measurements. If the inspection system cannot accurately locate the base, it may miscalculate the volume of solder on a pad. (Fig. 2). To determine a reference plane, the laser can locate a nearby copper pad and use its height as a reference. But the laser’s wavelength can affect the measurement. At near-infrared wavelengths, such as those required by acousto-optic deflectors, laser radiation may penetrate into the PCB materials before reflecting to a sensor, thus yielding an inaccurate measurement (Fig. 3).

To determine a reference plane, a near-infrared laser system must locate a pad near the site of a measurement and use its height as a reference. This action requires trace information not in the paste-screen aperture CAD file. Thus, you’ll have to add this information manually. For small components such as CSPs, the PCB may provide no local copper pads that can serve as a reference.

Figure 2. When measured far from a solder-paste deposit, the zero-height reference plane may vary slightly from the local plane due to board warping. In effect, measurements using a distant reference plane will add to or subtract from the volume of solder paste calculated from the laser measurements. This figure exaggerates the warping to illustrate the effect.

Figure 3. Near-infrared laser radiation can penetrate PCB materials, thus yielding somewhat inaccurate results for zero-height reference planes. Laser radiation closer to the visible spectrum reflects off solder-mask materials.

On the other hand, laser wavelengths closer to the visible spectrum will reflect off the solder mask placed on the PCB and conductive traces. As long as you know the thickness of the solder mask, you can use the mask to calculate the zero-height reference plane at the conductive pads. Some inspection systems use a single mask-thickness value across an entire PCB instead of measuring mask thickness close to each solder deposit. The former approach will add any slight variations in mask thickness to paste volumes.² The hybrid approach of using 2-D inspection of all solder-paste deposits coupled with 3-D laser measurements of smaller volume fine-pitch and CSP components helps manufacturers properly control the paste-deposition process. T&MW

FOOTNOTES

1. Burr, Donald, “Printing Guidelines for BGA and CSP Assemblies”, Proceedings of the Surface Mount International Conference and Exposition, August 23–27, 1998. San Jose, CA, Surface Mount Technology Association, Edina, MN, pp. 417–424.

2. Cleck, Jean-Paul, “Flip-Chip/CSP Assembly Reliability and Solder Volume Effects,” Proceedings of the Surface Mount International Conference and Exposition, August 23–27, 1998. San Jose, CA, Surface Mount Technology Association, Edina, MN, pp. 315–323.

ACKNOWLEDGEMENT

This article was adapted from a paper entitled, “Process Control of Solder Paste Deposition,” by Mark Owen and Jeff Hawthorne; it was first presented at the SMTA International Conference, San Jose, CA, September 1999.

Mark Owen is the advanced product manager for MV Technology. He has held positions at ESI, Tektronix, and the Irish Department of Science and Technology. He has a B.Sc.M.E. from Oregon State, and an MTechAMT from the University of Limerick, Ireland. E-mail: marko@mvt.ie.