Issue link: https://iconnect007.uberflip.com/i/1265351
84 SMT007 MAGAZINE I JULY 2020 changes seem small, but collectively, they have substantially changed the assembly. To me, that means the original set of building blocks may or may not be working the same as they did ini- tially. They may still be working to the point that the product functions as it leaves the build- ing, but go back and take a look at the pile of "no trouble found" returns and figure out the root cause from there. Many times, you'll discover that while the materials may not have changed, the way they are being used has. This is why it's a great idea to take a look at qualifying new sup- pliers with updated acceptance criteria. For instance, many suppliers have been contrac- tually obligated to test cleanliness using the IPC standard for cleanliness that was found in J-STD-001 but has since been removed. Since the test has been determined to be inad- equate, it's time to develop your own testing plan specific to your product. That test may still include ROSE testing and the acceptance criteria of 1.56 µg of sodium chloride equiva- lence, but that can't be determined until you have a larger data set that includes SIR testing for objective evidence. IPC-WP-019 has all the information on this important (and some say drastic) change and what is recommended going forward. At a min- imum, you should ask for ion chromatography testing for initial acceptance and then possibly on an annual basis as well. For all the interim shipments, you should be able to accept ROSE testing on materials like bare boards; but again, only if that is backed up with IC data. It is also a good idea to ask for similar data from your component suppliers. Many com- ponent assembly processes use the same plat- ing processes as a bare board and can leave the same type of detrimental residues. When a component has plating residue, it can migrate up the leads into the body of the part and cor- rode wire bonds, among other failures. Pack- age-on-package (PoP) components can leave all the same residues as a standard assembly process, so think of them more as tiny assem- blies instead of components. Speaking of assemblies, let's discuss what you should be asking for on those. When you look at the witches' brew of chemistries used to get us to this point, it is a miracle they haven't simply melted into a pile of electronic goo. I assume that is how the electronic pio- neers came up with the formulas that we use today; if it melted, back it off 10%, and try again. Again, historically, cleanliness per the IPC was covered with the simple, quick, and inex- pensive ROSE test. Per IPC WP-019, "This test methodology was originally developed in the 1970s; it was never intended to be used as a cleanliness test, nor as a test for product acceptability, it was only intended to be used as a process control method." Unfortunately, this test has been used for product acceptance for many years, and that's not a good thing. I won't go into all the reasons why, but there is a mountain of studies out there for review. For years, our lab recommended that cus- tomers build at least 20 assemblies, then test half with ROSE and half with ion chromatog- raphy. This gives you a rough correlation to figure out what your ROSE number needs to be—whether that is 1.56 or 156.00—as it is all based on the way the specific ROSE tes- ter reads your product. Many large companies have gone the extra step of not only doing full board extractions for the IC test but also look- ing at specific areas of the PCBA to determine how clean it is. Global extractions with either IC or ROSE will normalize out pockets of contamination across the full surface area of the assembly and most often results in a passing grade. A full board has never failed in the field; it's always a specific component or soldering process. Many times, you'll discover that while the materials may not have changed, the way they are being used has.