Optimizing printer-based solder paste inspection: a 2-D inspection system combined with SPC can improve PCB quality and increase profitability.

Today, the device mix on any given printed circuit board (PCB) can include any combination of ultra small 0201s, extremely fine pitch quad flat packs (QFPs), delicate ceramic ball grid arrays (BGAs) plus large connectors and power supply devices. Modern PCB design utilizes every millimeter of real estate, creating high board densities that were unimaginable a decade ago.

At the sane time that many components are substantially shrinking in size and pitch, PCB physical dimensions are moving to opposite ends of the spectrum; some boards are extremely small while others are 27 to 30 in. and larger. Both types of board configuration produce specific processing challenges.

Solder paste stencil printing is one of the first process steps to feel the pressure. The mixed solder volumes needed to create a board's overall pattern of high and low mass deposits, combined with the ever-increasing requirement for tighter cycle times, are a true printing challenge.

Remaining competitive in the electronics manufacturing industry demands always aspiring to the goal of zero defect production. The tolerance level for defects is smaller than the space between the leads of a fine-pitch QFE Two ways to support this elusive goal are through integrated inspection and statistical process control (SPC). Printer-based, in-line two-dimensional (2-D) solder paste inspection technology (2-D inspection) can be optimized to meet some of the toughest stencil printing production challenges.

Printer-Based Inspection

To reasonably grasp true production fluctuations, inspection tools must be combined with active statistical process control (SPC) features. A good inspection system must be able to feed data back to the process engineers, enabling them to adjust the process. The data must be available in real-time so action can be taken to correct a situation before it becomes a true problem resulting in defective products.

A 2-D inspection system measures the amount of paste covering the target pad and compares it against the required coverage (Figure 1). Using this data, the process engineer can then appropriately adjust the process or allow the printer to automatically initiate corrective action to optimize yields. To not affect production flow, both inspection and data output must happen on the fly. Full off-line or end-of-process inspection systems, although valuable for other reasons, cannot deliver immediate results because they are simply not at the data source in relation to the stencil printing process step.

[FIGURE 1 OMITTED]

When integrated directly into the stencil printer, a 2-D inspection system is at the immediate data source, but it usually cannot check the entire board due to cycle time limitations. Instead, the system can check only portions of the board. However, if properly optimized using the latest onboard inspection tools, software and some strategic planning, the process engineer can effectively eliminate print defects and increase overall yield faster using in-line 2-D inspection. Accuracy will be as reliable as end-of-process inspection systems, but at a fraction of their cost.

Approximately 90 percent or more of all print defects and trends can be identified by verifying the amount of paste covering the target pad. The 2-D inspection method identifies the lack of paste being deposited on the pad resulting in unacceptable solder joints.

The most effective type of 2-D inspection incorporates a gray scale comparison technique to determine the percentage of paste that is covering the target PCB pad. This technique compares the uncovered area of a printed pad against the saved area of the bare pad, determined at the time the device is first programmed. With this data, the equation in Figure 2 can be used to determine the percentage of paste coverage.

[FIGURE 2 OMITTED]

Optimizing the Process

Finding and correcting paste application problems are best done as they occur. With some exceptions, in-line inspection is not designed to inspect every single pad on every single board. So, the inspection system must be programmed to take advantage of the key devices and areas to use as data models. Inspection of these selected devices will transfer over to all devices on all boards being processed in any given run.

To learn how to develop the best data models, you must understand how to check an assembly for the type of components that fit into these categories. Also, you must fully understand the flexibility of the tools available in the inspection software to best determine how to apply these data models in the most effective manner for any stencil printing application.

Device Prioritization in Data Modeling

Single out the devices on any given assembly that will cause the most trouble for printing. Use these devices as the representative components to test. If these problem generators are within print specifications, you can assume that the simpler devices on the board should also be within the required parameters. With some aggressive cycle times, not enough time will be available to inspect a sample of all device types. Accordingly, the following fist has the items most likely to experience solder coverage defects:

* Pinpoint difficult devices. Determine which devices will be the most difficult to print. Examples include chip-scale packages (CSPs) and fine-pitch QFPs where the very small paste volumes are typically very difficult to inspect and measure.

* Find solder thieves. Identify the devices that have large components or pads on the left and right side of the board. These large components will require more solder to be deposited and, therefore, will generally show insufficient solder coverage well before other critical areas become a problem.

* Pattern the inspection. Program the components to be checked as a spread across the entire board. This approach will help detect regional issues such as board support weaknesses or areas of high paste consumption. Also, inspect components in several different quadrants of the board such as left, right, top, bottom and center.

* Identify hardest to check devices. Identify the devices that may be the hardest to check later in the process (during component placement or soldering) such as BGAs and micro-BGAs where seeing the actual pad connections without x-ray inspection is virtually impossible, and, even with that, quality is hard to determine.

Software Tools and Data Models

The best in-line 2-D inspection systems will offer an array of tools to streamline the inspection and data gathering process and allow the engineer to customize inspection patterns and timing to best meet the needs of the production line. These tools include: inspection of portions of devices only; device prioritization; and random device inspection.

* Inspection of portions of devices. If available cycle time is very tight and the board has many critical devices, check only portions of several devices rather than fewer devices. An effective 2-D inspection system must allow for programming at the pad level. This system allows you to check only the corners of the QFPs and BGAs on the board. However, this approach will require a tradeoff-between checking more regions or more full devices.

* Device prioritization. If pulse rate fluctuations are prevalent in the line and if boards are sitting at the end of the printer conveyor waiting to proceed downstream, a device prioritization utility wall help you take advantage of this downtime. To make best use of this idle time, all of the devices may be programmed and a priority level set for each device. Devices with priority levels are inspected in numerical order with two qualifications: always inspect or inspect on every board; or inspect until downline is ready (when time is available because the board is sitting idle waiting to be conveyed downstream).

For example, a device can be given a number between 1 and 5, with the number 1 set to always inspect. Each remaining device is then assigned a number from 2 to 5. The devices with a rating of 2 will be inspected if the board is not being called to move downstream. After these devices are completed, if no call has come for the board, the system will inspect the devices with a rating of 3. This process is repeated until all devices are inspected or the board is called to proceed downstream. This type of utility ensures that as many devices are inspected as time permits.

* Random device inspection. The objective of using this feature is to inspect as many devices as possible over a range of boards, rather than on a single board, to save time when cycle times are tight. This tool allows the system to inspect only a few devices on each board but cover all devices over several boards.

The frequency of inspecting a device can be set to allow for the most difficult devices to be inspected more often than the easier devices. The order of inspection can also be random, allowing the system to choose the order in which the devices are inspected. By programming a few non-critical device patterns, runaway small errors that the process engineer may overlook because they do not fit into any of the most critical device patterns can be caught.

SPC and Process Control

The goal of SPC is to prevent defective product at the earliest production stage possible because every stage creates a more valuable product that will cost far more to rework or scrap. Therefore, if an SPC program is to be effective, it must receive data from the line at the stages when this data is most important. This reason is foremost for in-line print inspection because verifying the results of a freshly printed board is the optimal way to determine that the print process is in control and that acceptable boards are being produced. Also, if a problem exists, it can be corrected at a stage that only requires washing off a few boards to rerun under correct printing conditions.

This SPC data must be made available to anticipate potential problems and direct the process engineer to take corrective action or automatically initiate a stencil cleaning cycle. SPC data collection and reporting provide powerful tools for reducing process variability and increasing throughput. Industry standard control charts measure pertinent aspects of the process. Real-time charts will keep the operator informed at all times, alerting when chart limits are reached or exceeded. Some printers have tools that allow a process engineer to replay the SPC charts for offline review of a freshly printed batch of boards.

A variety of programmable SPC tools will help find, define and eliminate printing production problems. These tools are also highly effective for comparing print performance over time to ensure that any non-critical deviations do not become actual problems. Using these SPC tools is not difficult, but some basic understanding is required of what is typically available to use them effectively.

A model 2-D inspection system must allow user configuration of the SPC software so that the real-time charts will use factory preference parameters as the guidelines to warn or alert the operator of deviations. The system should also be programmable so the data collected will indicate when a deviation from the norm is being caused by the printing system (Figure 3).

[FIGURE 3 OMITTED]

Conclusion

Optimizing the 2-D inspection process, using efficient strategies and taking advantage of all tools available in this program will greatly increase the power of this useful system. It basically turns the printer into a smart machine that is a true engineering partner on the production line. The engineer will have full control of current stencil printing variables with the ability to flag even the slightest process drift. The ability to analyze numerous parameters over time also enables the engineer to make knowledgeable decisions for detailed improvement in specific areas, streamlining the full solder deposition process.

Inspecting a PCB immediately after printing verifies the printing operation itself to characterize the process. The SPC features, both immediate and long-term, will catch and eliminate minor issues before they become real problems. This capability translates into immediate cost savings and brings the total process closer to the elusive goal of zero defect production.

As the board manufacturer takes advantage of the programming flexibility and full SPC capabilities offered by optimized 2D in-line solder paste inspection, the bottom-line profit will increase substantially. When fully implemented, this system will provide a quick return on investment and keep adding value by increasing board quality and yield. The reduction in board defects and virtual elimination of scrap add dollars back to the company profit margin. The increases that can be seen from this investment in overall yield are only limited by the production capability of the facility itself.

John Morini is senior applications engineer, John Cronin is software development engineer and Dennis O'Neal is product marketing manager, all with Speedline MPM, Franklin, MA; e-mail: [e-mail].