2005 Articles
An Experimental Methodology for Process Optimization
A 24-panel method for characterizing new soldering processes.
One of the great challenges is anticipating and evaluating the interaction of lead-free materials with other process changes. In this column, we’ll describe an experiment that can be used to analyze the interaction between lead-free solder, new PCB finish techniques and reflow methodologies. Using x-ray inspection to evaluate the experiment results will help ensure an effective transition.
The reflow temperature of most SAC (tin-silver-copper) solders is about 217°C. This (higher) temperature means that the thermal profiles of reflow ovens must be recharacterized to obtain proper heating and cooling rates for components and boards to ensure low defect rates. This is usually a trial-and-error process. These higher temperatures are also incompatible with some chemicals used in many manufacturers’ solder pastes. Compounds break down, causing voids in the finished solder joints. Solder paste selection is one more variable in process characterization.
One common PCB finish is tin-lead HASL (hot-air solder leveling). Lead-free HASL processes are still being developed and evaluated, and HASL has the disadvantage of not achieving the planar pads desirable for fine-pitch components. Therefore, most manufacturers are moving to HASL alternatives such as bare copper with OSP (organic solderability protectants), electroless nickel/immersion gold, electroless tin or immersion silver. These new metallurgies add more variables.
Questions are often raised regarding the use of automated x-ray inspection (AXI) with lead-free solders, since tin-lead solders result in high contrast images in x-ray. Tests show that the approximately 95% tin and 4% silver in SAC solders produce x-ray images with only slightly less contrast than those of tin-lead solder. The images are still easily analyzed by AXI systems.
One effective way to characterize a lead-free soldering process is to run an experiment with a limited number of carefully selected variables. For this sample experiment we use three different reflow profiles, three brands or formulations of solder paste and two different board finishes. This provides 18 (3 x 3 x 2) combinations, or cells, which can be evaluated in a single experiment. This is a fairly large number of combinations for a single experiment, but much can be learned with this type of multivariable experiment. Use the number of variables and conditions for each variable that makes the most sense for your process.
Commonly, attributes are used as a measure of merit for experiments. Attributes have two possibilities: good or bad, defective or acceptable, “Go” or “No-Go.” However, in this case if the measure of merit for each combination is the number of defects generated and detected, a very large number of boards would need to be assembled in the experiment to generate a quantity of defects large enough to be statistically meaningful, since defects typically occur in assembly processes at a rate of a few hundred per million opportunities.
Running a very large number of boards under well-controlled conditions in a production environment is often impractical. It can take several days of production to build enough boards to generate enough defects. During this time, the chance is high of introducing other uncontrolled variables such as machine setup variation and differences between shifts and operators.
Generally, the wetting performance of the solder in the joints is a good indicator of the quality of the process: Better wetting means fewer defects. One method of reducing the number of boards required in the experiment is to use a continuous variable (e.g., solder thickness) as a measure of merit instead of an attribute (defective/acceptable.)
In this experiment, using AXI generates continuous variable data which indicate the wetting performance of each combination of process variables. Measurements such as solder fillet thickness, solder joint fillet length, solder thickness differences between heels and under the leads on gullwing joints, and other measurements on selected components can provide this information. These measures of wetting performance correlate to the performance of the soldering process. Commercially available statistical packages analyze the data and determine the significance of each variable and condition.
An experiment for process characterization during the major process transition to no-clean soldering was successfully run with a total of 24 PCB panels, permitting all the panels in the entire experiment to be built in a few hours instead of over several days. The result was selection of an optimal reflow profile and solder paste formulation, and elimination of other variables of lower order importance. This methodology is valid for the transition to lead-free.
In addition to its role in finding the increased number of defects expected with the transition to lead-free processing, AXI is also an effective tool for characterizing new processes and materials encountered during the transition to lead-free soldering processes. n
Glen Leinbach is technical marketing engineer, and Stig Oresjo is a senior test strategy consultant, both with Agilent (agilent.com); glen_leinbach@agilent.com.
