Issue link: https://iconnect007.uberflip.com/i/860275
30 The PCB Design Magazine • August 2017 ment. The assembly supplier was re-directing PCB orders away from the present supplier in hopes of doing business with a bare board sup- plier who knew the difference between white vs. pink. It started getting personal. Stakehold- ers were reacting rather than responding. To make a long story short, after several weeks, a wise manager stepped in and began asking questions. He asked if anyone were con- ducting any root cause analysis on this issue. He described the use of pareto charts and fishbone diagrams In a very constructive way. By simply asking questions, he helped the department to realize that they were merely spit-balling con- spiracy theories and accusations. The depart- ment had been quick to find fault and place blame rather than getting to the real root cause of the problem. In a few more weeks, amazingly, this talented manager had changed the culture of the department enough so that we could fo- cus on finding the root cause of our fancy pink PCB assemblies. To our surprise, the root cause was something none of us could have expected. Here's how it went down. First, our engineering customers had to come up with some acceptability criteria to de- fine a range of white. We follow the examples of the IPC acceptability specs IPC-A-600 and IPC- A-610, and we photographed a range of accept- ability. A white photo was designated the "tar- get condition." Then, slightly tinted condition photos were noted as "acceptable" and lastly, a pink photo was included and la- belled a "non-conforming" ex- ample. This was a critical step in defining the problem. Second, we sought out all of the process stakeholders and brought them together to collect facts and to understand their materials and processes. What we learned by communicating with all of the stakeholders was monumental. No one was conspiring to accept PCBs by buying them off in incoming inspection. No one was inanely applying pink solder resist to the PCB. After asking all the right people every possible question, we learned that a bare board supplier sought to overachieve with regard to our reflec- tive "white" requirement. They ordered a new reflective white solder resist from their suppli- er. The new product was as white as one could imagine. It was applied to the PCB and covered with a beautifully, opaque, silky white finish. The PCBs were shipped to the assembly suppli- ers and were received, passing the "white" in- spection point on the fabrication drawing with ease. But surprisingly, after the bare PCB's gold contact lands were coated with RoHS solder paste and run into the reflow oven environ- ment, the white solder resist reacted to the temperatures and chemistry combination and turned the solder resist to pink! Then, the pink- ish board assemblies moved on to inspection. There was no solder resist color requirement listed on the assembly inspection document. There was no reason to reject the PCB assem- blies, so they were shipped. Nobody saw how the problem developed until being informed by using some elementary RCA processes. In the end, it turned out that the solder resist supplier had been experimenting with some chemical elements in the formula to increase opacity and improve reflectivity. During a perfect storm of compounds, chemistry and temperatures, voi- là—pink PCBs. How much money was wasted? How much time was wasted? Who's at fault here? Hopefully, we've learned that stuff hap- pens. It's how effectively we address it that mat- ters, so it won't happen again. What else can PCB designers learn from RCA? FAULT-FINDING: IT'S ALL ABOUT PREVENTION, NOT BLAME