Improving the reflow process with SPC: continual data analysis of a reflow process offers increased product quality and reduced manufacturing cost.

A good statistical process control (SPC) program for the reflow process should be set up to monitor oven stability and catch any process drift before it can affect product quality. The SPC program enables the operator to easily gather data to monitor and verify the process. Using the analytical tool described in the previous articles, data can be gathered every time a change or a break occurs in the production flow. The analytical tool verifies the oven's performance parameters and helps build a database of control charts that contain the most common causes of variability. These charts show if the oven is stable or if its performance is shifting, thereby allowing the operator to catch and correct any out-of-control situation at its onset.

Once data are downloaded into the oven computer, a data page appears on the computer screen that lists all oven parameters. The data page compares each parameter against established performance criteria such as the upper specification limit (USL) and the lower specification limit (LSL). If the parameters are within the pre-determined USL and LSL, all data will appear in black. The assumption is that the oven is stable and ready to run the product.

However, if a parameter exceeds the USL, the data will appear in red. If a parameter is below the LSL, the data will appear in blue. If the operator sees red or blue data on the screen, then a shift has occurred in the oven environment and corrective action may be needed or, at least, the data must be analyzed to determine the impact the process shift will have on the product's quality.

Setting Up Reflow SPC

The analytical tool is used within the SPC program in two ways. First, it can validate the oven's optimal working parameters as determined by a proven good baseline: the oven profile. Second, it can replace the thermocoupled test boards to reevaluate the oven's performance against any established criteria: the product profile.

Before any SPC program is started, the oven must be stable. A product profiler is used to establish a known good oven profile: the optimum oven temperature setpoints. Next, the analytical tool gathers and stores the oven performance parameters for that profile in a database.

This approach is also done for each new product profile. A separate folder can be made for each oven and every product run, or a single database can be created, with data subsequently filtered down by specific product, solder paste type, process condition or other criteria.

The planning and creation of the process monitoring system are determined by the workmanship and quality standards at each factory as well as the users' computer savvy. The easiest way to set up data files, with the least amount of confusion, is to create one database for each machine and each product being run, plus a baseline, if applicable.

Although ovens are fairly uniform and consistent from model to model and serial number to serial number, some variability does exist. For multiple ovens in a factory, each oven should be listed separately in its own database.

Product Profiling vs. Oven Profiling

The difference between a product profiling tool and the analytical tool can be a little confusing. In most cases, the product profiler should not be used to gather and download board data directly into the SPC program because, if the process is not qualified and stable, all data gathered from that process will be useless.

For example, an out-of-control oven can still be tweaked to conform to a specific product profile. However, the original problem still remains, so the data from that product profile and subsequent profiles will not be true representations of the oven's actual working parameters. The product profiler and analytical tool are both needed to create ideal product thermal processing, to streamline the reflow process and also build a strong SPC program.

The product profiler allows the user to develop and collect separate profiles for each product. However, it does not store the profiles in a database that can instantaneously analyze the data to indicate base machine performance.

In contrast, the analytical tool collects machine-specific information that allows the user to monitor machine performance. The analytical tool can be considered a partner to the product profiler. Once the recipe (product profile) is developed, the analytical tool is run through the oven to generate its optimal profile. This optimal profile is permanently stored and becomes the oven baseline profile for that product.

A printed circuit card attached to a product profiler can only experience about five to seven thermal excursions before the base board material starts to deteriorate, introducing variability into the process. In contrast, the analytical tool can withstand multiple data runs for oven quantification without deterioration. It also eliminates the time associated with instrumenting new "sacrificial" boards with thermocouples for repeat product profiling.

Reflow analytical tool data are like any statistical data; the more data collected, the more control exists over the process. Therefore, the more frequently the analytical tool is used, the more control the operator has over process integrity. A statistically significant population of known good runs is needed to get a baseline quantification of oven data because every process has some variability or natural fluctuation associated with it. A statistically significant population of data will map the naturally occurring highs and lows of the process so these points can be considered when creating control charts, comparing other incoming data from products or re-optimizing oven performance and process control.

Once that population of known good runs is collected for oven optimization, the next step is establishing the upper and lower specification limits for individual products. Specification limits are different from control limits. Control limits are data driven; mapping the upper and lower performance capabilities of the base machine. Specification limits are the customer's defined processing parameter limits for the individual board or product.

When the population of known good runs is available, a nominal product processing tolerance value can be calculated; a user-defined, plus/minus value that will still yield an acceptable quality product. After these specification limits are established, ensuring the integrity of the reflow environment becomes extremely easy.

X-bar and r-bar charts are used when a sample size of 10 or less data collection points can be taken every few hours. When only a single sample is taken over long periods of time, the x-bar moving range chart is typically used.

The process sensors on the analytical tool record the conveyor speed, process deltas and product dwell times for each oven zone and automatically add these measurements to the stored database to create the appropriate control charts. The software automatically puts all data into the selected control chart. Each run is mapped within the control limits and compared to past data. Data tabulation and analysis are straightforward and can be easily viewed by the user.

Reflow SPC in the Real World

In general, an analytical tool program should collect data at regular intervals. Initially, analytical tool runs through the oven would be quite frequent, but, as a statistically significant population of known good runs is collected and the normal process variation mapped, the frequency could be reduced to a few times per shift or whenever a new product profile parameter is introduced.

Review and adjustment of the process control parameters would be done on a set basis once the base machine variation is established. Changes in the process stability would help dictate the review and adjustment frequency.

Tables 1 and 2 may be used to develop a workable SPC program for any reflow process. Table 1 is an overview of programs that may be used in common manufacturing scenarios. Table 2 can help determine process verification frequency based on the importance of some major items related to reflow production. These tables are not meant to be cookie-cutter solutions, but they offer some starting points to build on.

Finally, a corrective action system must he established that allows any identified process drifts to be documented, formally investigated and addressed where necessary.

Conclusion

A primary reason to link the reflow oven SPC into the overall production line quality control system is to eliminate the reflow process as a source in production problems. The product data collected can also join with other production level data to create full production accountability for every lot of products made. If defects arise later regarding a particular product lot, the SPC oven/product data will identify what paste was used, what profile the product saw and all the existing oven parameters at that time.

If the reflow process was stable, this finding would guide the engineer to check for other possible causes, such as the specific lot of paste purchased, component supplier or other production factor. For example, perhaps the screen printing was off registration just enough to cause a problem not noticeable at the time of product reflow or perhaps components were reworked and became "work hardened," which shortened their life. The analytical tool provides such traceability for the reflow process.

Using continual data analysis will ensure a stable reflow process and reveal where the process is moving as a whole to eliminate issues before they become costly problems. In clinical trials at the factory, ongoing SPC using the analytical tool verified that the reflow oven was very stable. The SPC database provided with the analytical tool's software maintains proof of stable processing that can be presented on demand. This system offers substantial benefits including less rework, lower overall manufacturing cost, increased product quality, higher yield and increased life of the reflow oven.

TABLE 1: Tailoring reflow SPC to facility type.

Facility Environment           Standard Formula

Low Mix/ High Volume           Verify Oven Process
OEM                            Parameters:

Multiple reflow lines          * Start of each shift
Low reflow profile mix         * After each preventive
Low oven maintenance           maintenance (PM) task
Produces same product.         * After each equipment
New product 1-2 times/         acquisition
year                           * After each break (1)

Moderate Mix OEM               Verify Oven Process
                               Parameters:

Few reflow lines               * Start of each shift
High reflow profile mix        * For each profile
2 shifts/day                   changeover
High oven maintenance          * After each PM task
Mix of simple products;        * After each equipment
new products common            acquisition
                               * After each break (1)

High Mix CM/EMS                Verify Oven Process
                               Parameters:

Multiple reflow lines          * Start of each shift
Multiple shifts/24 hours/      * For each profile
7 days per week                changeover
High reflow profile mix        * After each break (1)
High oven maintenance          * After each equipment
High mix of complex            acquisition
board configurations,          * After each PM task
including lead-free solder,
0201s and ceramic
packages.

Low Volume CM/EMS             Verify Oven Process
                               Parameters:

Few reflow lines               * Start of each shift
Single shift                   * For each profile
Low reflow profile mix         changeover
High oven maintenance          * After each break (1)
Products require high          * After each equipment
level of "cradle to            acquisition
graver" QC documentation/      * After each PM task
travelers; medical,
aerospace, consumer
safety products

Universal Features              SPC Benefits

1. Chart process deltas         1. Enables easy maintenance
and times in each               of copy exact
zone.                           manufacturing
2. Visually monitor control     between production
charts for any                  lines.
process drifts.                 2. Verifies that the
3. Review Cpk process           control.
index to determine              3. Provides documentation
present capability of           for reflow
the process to                  process verification
determine number of products    that can follow the
that conform to                 product and can be
specification.                  used for both in-
4. Measure Cp in conjunction    house and customers.
with Cpk to
determine if the
process is shifted and
how much work is
required to center
the process.

(1) This depends on board cost, the process window specifications; see
Table 2 for more details.

TABLE 2: Priority scale.

Value Scale: 0 = Not Important or Infrequent 10 = Extremely Important
or Very Frequent

Process Input                0   1   2   3   4   5   6   7   8   9   10

PCB value
Paste process window
QC requirements
Oven maintenance schedule
Line tech ability
PCB run rates
Changeover frequency
Product profiling
  frequency
Product traceability
Component robustness
PCB complexity

Majority in the 0-3 range = Verify oven process parameters daily and
after maintenance. Analyze data monthly.

Several in the 4-6 range = Verify oven process parameters daily,
before product changeover and after maintenance. Analyze data weekly.

Several in the 7-10 range = Verify oven process parameters at the
start of each shift, after every product changeover, after each break
over 30 minutes and after maintenance. Analyze data at least once each
day.

Karen Walters is the process engineering manager for BTU International, N. Billerica, MA; e-mail: [e-mail]. Karl Fischbeck is part of the corporate management team at ECD Inc., Vestal, NY; e-mail: [e-mail]. Ray Pearce is the technical director for ECD Inc., Milwaukie, OR; email: [e-mail].