Continuous Improvement Approaches for Wave Soldering

Estimated reading time: 14 minutes

By Joe Chu and Tony Huang

If an effective process control system can be implemented with the appropriate preparation and support, most of the problems that arise in the wavesoldering process can be eliminated or at least minimized to an acceptable level.

Wave soldering is a mature technology and remains an efficient mass soldering process, especially for through-hole and Type III SMT assemblies.1 However, wave soldering also has been repudiated for its inconsistent performance and complexity. Wave soldering is complex because of process operating variables such as conveyor speed, preheat temperatures, soldering temperature in the wave, board-wave interaction, flux chemistry, machine maintenance, board design, component solderability and operator training.

Recently, some manufacturers have been trying to eliminate wave soldering to streamline their printed circuit board (PCB) assembly processes because of emerging technologies and "smaller, faster, cheaper" demands.2 If the complexity of wave soldering can be mastered so that repeatable excellent soldering performance can be achieved, some experts believe that it will maintain its niche and adapt itself for new challenges.3,4

Figure 1. Deming's PDSA Cycle.

The goal of wavesoldering improvement is to produce perfect wavesolder joints the first time. Everyone concerned with wavesoldering improvement must realize that touch-up of soldering defects is unnecessary and extremely expensive. Besides, touch-up will not improve the original soldered joints. In fact, the soldered joints will degrade because they will be subjected to another thermal cycle, increasing the intermetallic compound thickness.A wavesolder improvement team was established in the Proto Assembly Center at Adaptec. Instead of trying to find a one-time fix solution, a continuous problem-solving methodology was adopted. The team adopted Deming's Plan, Do, Study and Act (PDSA) Cycle, which is a continuous approach for problem solving.

Because people are the foundation of any improvement activity, the kaizen philosophy of the Japanese is concerned first with the quality of people. If the quality of people is improved, then process improvement follows.5 Formal and proper training needs to be conducted regularly for everyone involved with the wavesoldering process. From the wavesoldering improvement perspective, everything related to the soldering process can be improved. No matter how small the improvement is or what the improvement activity is, it will enhance overall soldering performance.

Continuous Improvement Strategies

The improvement process consists of four major phases: defining the goal, establishing and training the focus team, applying the Deming PDSA cycle, and evaluating the changes because of the effects of improvement. With support from management, the project goal was clearly defined as follows: "Incorporating all the related people in the organization to work together to continuously improve current wavesoldering processes without large capital investment."

Figure 2. Pareto chart for wavesolder defects.

Because people are the foundation of any improvement activity, improvement in "quality of people" will result in process improvement.5 Formal and proper training needs to be conducted regularly for everyone involved with the wavesoldering process. To apply the "quality of people" philosophy to the wavesoldering improvement program, two formal training sections were conducted for the team members.

Figure 3. Wavesolder defect control chart.

The first training section consisted of two full days of technical training conducted by an outside expert. Topics such as the theory of wave solder, solder defect analysis, and wavesolder machine operation and maintenance were covered. The company's senior quality manager covered topics such as problem solving and process improvement techniques in the second training section.

Figure 4. Wavesolder defect cumulative chart.

The team continuously evolved to identify and remove all the controllable root causes for wavesolder defects, to identify the uncontrollable root causes for wavesolder defects and to convert the uncontrollable root causes to controllable ones. The uncontrollable root causes (e.g., PCB design, component selection, equipment limitation, etc.) can become controllable.

Application of Deming's PDSA Cycle

Deming's PDSA Cycle (Figure 1) is applicable for different levels of process improvement. For this application, the plan phase consisted of monitoring the current process, collecting soldering defect data, identifying the root causes and developing improvement plans. In the do phase, each action plan was implemented and the results were evaluated in the study phase to see if the plan was working out as expected. Also in the study phase, further problems or opportunities were examined. In the last phase of the Deming Cycle, the act phase, appropriate actions were taken based on the evaluation results from the study phase to either adopt or abandon the changes.

If the improvement plan worked as expected, the changes would be standardized into related specifications to ensure they would be practiced routinely. However, if the improvement plan did not meet expectations, new improvement plans would be created, leading the Deming PDSA Cycle back to the starting phase. No matter whether the changes were adopted or abandoned, the Deming Cycle would be initiated until all the root causes could be removed.

Application of Improvement Tools

Several problem-solving tools were used extensively during the early phases. Pareto charts were used to graphically represent the impact of wavesolder defects and causes to help the team prioritize and direct its problem-solving efforts (Figure 2).

Cause-and-effect (fishbone) diagrams were also heavily used and were constructed in a brainstorming atmosphere. By representing the chain of causes and effects graphically, the cause-and-effect diagrams helped the team sort out the potential causes for wavesolder defects and organize the causes into defined categories.

Based on the results of the wavesolder monitoring and cause-and-effect analysis for soldering defects, the team developed a comprehensive plan that consisted of various sub-plans. Each sub-plan then adopted Deming's Cycle to carry out the plan.

Soldering performance was measured and plotted into a statistical control chart to monitor the improvements and detect process variations in the study phase of Deming's Cycle. Comparisons are against the measurements previously taken instead of standards or acceptable levels. Figures 3 and 4 show the control charts reviewed and discussed in the team's weekly quality meetings.

Wavesoldering Optimization

Basically, wave soldering consists of three sub-processes: fluxing, preheating and solder application. To optimize the wavesoldering process means to optimize the three sub-processes.

Flux selection. Soldering flux is an essential part of the process. It removes oxide and cleans the metal surface to aid wetting. It also protects the metal surfaces during heating to prevent surface reoxidation. To improve the wavesoldering quality, the first step was to find an alternative flux that would enhance solder wettability and was compatible with the no-clean process. Because of the surface insulation resistance (SIR) contamination level requirements for no-clean flux, the choice of flux was limited to those with lower solid content.

Figure 5. Dummy pad designs for 68 pin mini-SCSI connector.

Initially, nine fluxes six alcohol-based and three water-based from different flux suppliers were selected for evaluation. All the fluxes were applied to the bare board and the boards were run through the wavesolder machine. Five fluxes were eliminated because they either created more flux residuals or more contamination on the board after the process. The remaining four fluxes, which passed visual residual evaluation, were put through SIR and ion chromatography tests. The SIR test was performed for seven days at 85°C and 85 percent humidity per ANSI/J-STD-001A, while the ion chromatography test was done per IPC-TM-650 Method 2.3.28. Of the four fluxes, one flux actually failed the ion chromatography test because it had ionic contamination levels of higher than 1 µg per sq. in. of chloride, bromide and sulfate on its surface.

Figure 6. Universal wave fixture for SMT type III assembly.

The final step of the flux evaluation was a wave defect analysis. A complex board, consisting of headers, connectors, and many bottomside chip capacitors and resistors, was selected to test the remaining three fluxes. The evaluation run was conducted using five boards for each flux. The defects used were solder bridges, excessive solder, non-wetting, solder skips and solder balls. The final selection was based on the tally of the defects. With only three defects, an alcohol-based, 2.5 percent solid-content flux was selected.Flux deposition analysis. The amount of flux applied to a PCB and the uniformity of the deposition is crucial to good solder joints. To ensure that the flux is applied uniformly to the PCB, the fluxer needs to be set up correctly. The wave machine used was equipped with an internal spray fluxer that was controlled by an external stand-alone cabinet housing the pneumatics and electronics. The first approach taken by the team to ensure the spray fluxer was set up correctly was to optimize the settings of the flux control units: flux spray speed, air knife pressure and flux volume.

Figure 7. Selective wave fixture for SMT type II assembly.

A two-level factorial experiment with four factors was conducted to determine the main effects on the flux coverage and their interactions (Table 1). Flux coverage was the response used to determine the uniformity of flux deposition. The method used to determine flux coverage was to attach a chemical-sensitive fax paper to a nonporous plate. As the plate passed over the fluxer, flux was sprayed onto the paper, causing the paper to change color. The percentage of the areas on the paper that changed color were tallied and recorded as flux coverage percentage per board.Based on a standard pareto chart, flux volume and air knife pressure were the most significant factors. The flux coverage increased with the increase of flux speed and volume as well as the decrease of the air knife pressure. Conveyor speed was not a significant factor compared to the other three factors. To determine the settings for these factors, another 24-1 factional factorial experiment with the level changes was conducted. In this experiment, the low-level values for flux speed and flux volume were decreased to 40 psi, and the high-level value for air knife pressure was increased to 30 psi. The experiment included five center-point runs to get additional information about main effects and interactions. The experimental results helped to establish the following optimal fluxer settings:

• Flux speed: 45 psi
• Flux volume: 45 psi
• Air knife pressure: 25 psi.

Optimal profiling. After the wave fluxer settings and flux selection were finalized, the next area for improvement was controlling the wavesolder variables and creating profiles for the board products. The critical variables were conveyor speed, preheat temperature, solder contact dwell time, immersion depth, conveyor parallelism and solder temperature. Because the most direct way to achieve accurate profiling is to acquire data from an assembly passing through the wave machine, it is important to create individual profiles for each PCB. Because the PCBs used in the Proto Assembly Center have similar or consistent characteristics, such as size, thickness, layer count and component placement, they can be divided into family groups to minimize the number of profiles and achieve the same soldering results. Eventually, six different PCB family groups were created based on their characteristics:

• PCB 4.5 x 7", single-sided
• PCB 4.5 x 13", single-sided
• PCB 4.5 x 7", double-sided without selective wave fixture
• PCB 4.5 x 7", double-sided with selective wave fixture
• PCB 4.5 x 13", double-sided without selective wave fixture
• PCB 4.5 x 13", double-sided with selective wave fixture.

One of most difficult tasks during wave profiling was determining the set points of the preheaters.6 The optimal goal for heating a PCB during the preheat process is to raise the topside board temperature to between 200° and 210°F, while keeping the bottomside board temperature at an acceptable level. To minimize the number of runs needed to determine the preheater settings, two 24-1 factional factorial experiments were used. The factors used for both experiments were the same: preheater zone 1, preheater zone 2, preheater zone 3 and conveyor speed. The response used for the first experiment was the topside temperature, while the response used for the second experiment was solder yield. The second experiment was conducted after the variable ranges were determined during the first experiment. Tables 2 and 3 list all the variables, responses and variable ranges for both experiments.

Control of board-wave interaction parameters. After the fluxing process was optimized and the optimal heater and conveyor speed settings were determined, a system to control the solder application process was implemented. Variables such as solder contact dwell time, immersion depth and conveyor parallelism were the key factors that needed to be controlled while the boards passed through the wave. These board-wave interaction parameters were measured and controlled using a wavesolder optimizer.The variables that affect dwell time and immersion depth are conveyor speed, solder pot level, solder pump height, the method in which the PCBs sit on the fingers and whether fixtures are used. Because conveyor speed was determined based on the PCB preheat requirement, it remained unchanged while adjusting the solder contact dwell time and immersion depth. The approach to control dwell time and immersion depth was to fix the solder pot level, and to adjust the solder pump speed based on the predetermined optimal values. The optimal immersion depth and dwell time were determined based on the times that produced the lowest defect rate.

Wavesoldering Innovation

PCB design for wave soldering. Another important item for the team to consider was proper PCB design. Without proper PCB design, it is impossible to reduce defect rates just by controlling the process variables. Proper PCB design for wave soldering should include correct component alignment, footprint design for wave soldering, and sufficient clearances between selective wave fixture openings and adjacent components.

Proper component alignment is required to prevent solder skips, uneven solder fillets and solder bridges. For successful wave soldering of some surface mount devices (SMD), normal footprints can be modified to improve the yield. Some of the requirements for modifying the SMD footprint are proper gaps between pads for adhesive dots, additional solder thieving to reduce solder bridges, and minimum clearance between components or pads. For some staggered high-pin-count through-hole connectors, adding dummy thieving pads to the trailing pins can also help prevent solder bridges. As far as wave soldering the PCB using a selective fixture is concerned, the clearance between shielded SMDs and exposed pins is critical to the solder results as well as the durability of the fixture. If the clearance is not sufficient, solder skips or solder bridges may occur.

A test vehicle was designed to evaluate component stagger impact, minimum spacing between SMDs, minimum spacing between SMD and through-hole components, footprint design, and shadow effect during the wavesoldering process. The test board included different footprint designs for 0603, 0805, 1206 and different component clearances. The through-hole headers and connectors were oriented differently for component orientation evaluation. In addition, eight different dummy thieving pad designs were added to a mini-SCSI connector footprint, which usually suffers from solder bridging on the trailing rows during production. Figure 5 shows some of the dummy thieving pad designs on the mini- SCSI connector.

Selective and universal wavesolder fixtures. Most boards are mixed-technology assemblies. A very challenging board layout is to have different types of connectors laid out on adjacent sides of the board at 90° angles. No matter which side of the board is run through the wave, the tailing pins of the connector on the other side are subject to bridging.

Both selective and universal adjustable rotary wave fixtures were designed to resolve this board layout challenge. The universal rotary adjustable wave fixture shown in Figure 6 was designed for SMT type III boards only passive components are mounted on the secondary side of the boards. The selective adjustable rotary wave fixture shown in Figure 7 was designed for SMT type II boards both active and passive SMT components are mounted on the primary and secondary sides. The through-hole components are mounted on the primary side of both SMT type II and III assemblies.

The angle of the rotary fixture can be adjusted to allow the leads to contact undisturbed wave solder. This adjustable rotary fixture helped eliminate bridging and skipping defects on the boards when challenging component layouts occurred because of design constraints.

Conclusion

Overcoming the complexity of wave soldering is intricate and there appears to be no such thing as a one-time fix solution that guarantees repeatable results. By adopting the continuous improvement concepts and tools discussed in this article, the team has effectively produced encouraging results 96 percent of touch-ups were eliminated.

Touch-up is an expensive process7 and it damages the solder joints. Therefore, producing perfect wavesolder joints the first time is not just jargon or a motivational program. Indeed, it is the most cost-effective approach for a manufacturer to use. With the prominent improvement results, a gradual, continuous improvement process is the right approach to attain the goal of "zero wavesolder defects." This project also has proven that the "quality of people" philosophy is applicable.

This article is adapted from a presentation originally given at SMTA International 1999.

REFERENCES

1. R. Prasad, Surface Mount Technology Principles and Practice, Chapman & Hall, p. 534.
2. R. Robertson, "Streamlining PCB Assembly," Circuits Assembly, April 1999, p. 32-35.
3. H. Markstein, "Wave Soldering Maintains its Niche in PCB Assembly," Electronic Packaging & Production, August 1998, p. 42-50.
4. L. Hymes, "Wave Soldering Still Viable," Circuits Assembly, April 1999, p. 28-30.
5. J. Evans and W. Lindsay, "The Management and Control of Quality," Proceedings of NEPCON West, Anaheim, March 1998, p. 233-34.
6. C. Kelly, "Preheat Impact in a No-clean Process," Proceedings of NEPCON West, Anaheim, March 1998, p. 615-28.
7. Howard H. Manko, Soldering Handbook for Printed Circuits and Surface Mount, 2nd Ed. ITP, 1995, p. 496.

JOE CHU and TONY HUANG may be contacted at Adaptec, 691 South Milpitas Blvd., MS #65, Milpitas, CA 95035; (408) 957-1678; Fax: (408) 957-6625; E-mail: jchu@corp.adaptec.com.