Issue link: https://iconnect007.uberflip.com/i/1288481
114 PCB007 MAGAZINE I SEPTEMBER 2020 where that one component can cost $100,000, then you don't want to build a board and put a couple (or five or six) $100,000 components on a space board only to find out that the ve- hicle instructions are not reliable. It is better to just test this coupon using a lit- tle bit of real estate. Put the same geometries, drill sizes, land sizes, and vias—whether they are stacked or staggered—in designs, build a coupon with planes and signals, and find out before you commit the components to the as- sembly. Now, you can test whether that struc- ture can sustain the reflow you expect to do to the part. You need to test that coupon for accep- tance before you commit the parts. Many people have not tested with reflow tempera- tures, measuring resistance at reflow to find out what is going to happen. With OM test- ing and IPC TM650 2.6.27, you can measure the resistance at reflow and find out whether these boards will go through assembly safe- ly. And you do not want to assemble the se- rial number that coupon came from. As seen in Figure 3, the cost of any failure goes up with time. All of my testing is IPC 2.6.27 for real-world validation for today's demanding PCB design. I'm going to review the current evaluation methods that are required by a PCB fabricator and talk about the different temperature meth- ods. Why did some methods not provide us a true assessment in some cases? And how does OM testing provide a real-world evaluation of your finished product by testing the coupons? In our current evaluation methods, we take A-B coupons and thermal stress them, depend- ing on the temperature. If it's tin-lead to 230°C or 260°C, then we evaluate the cross-sections and look at what we have after we micro- etched a polished coupon. Our shops are required to do monthly per- formance testing. I will talk about that, as well as what happened to a board at electrical test, and why it does not really tell us whether we have weak microvias. I'll also show some re- al-world builds in which the D-coupon testing catches a non-conforming via when the resis- tance caught it, as well as it being non-con- forming to IPC-6012 for internal evaluation. Why did the older methods seem to fail us? When we do a micro-section on A or propa- gated B coupons, we begin by baking. We sol- der float them and then grind them. In the as- polished condition, we look for separations. If we do not identify the separation of a micro- via evaluated at 200x, and if we do not notice anything, we will take a Q-tip with micro-etch and rub the coupon that has just been polished (Figure 4). We see this result in Figure 5, especially the different layers of copper plate in the starting foil. There's a plated layer and layer two. In this example, the flash plated microvia plat- ed shut. Then, we do the final calculating, but all the evidence is now gone, so it's kind of a moot point. We polish and look; if we don't Figure 3: Why are we testing for reliability? Figure 4: Micro-etched cross-sections hide small separations found in the polished condition.