Issue link: https://iconnect007.uberflip.com/i/253291
76 SMT Magazine • February 2014 ple thermal profile is shown in Figure 1. Yet, this high-speed board fails at high speed at the time of system functional level testing in the system. In the course of the investigation, it must be determined whether or not the material used is straight FR-4, a hybrid, Rogers or polyimide and how it may be affecting the board fabrication process. Also, when it comes to impedance, you have to look at the coupons run at the fabrication level and the tolerances the OEM provided to the fab shop. Is the fab shop looking at imped- ance tolerances of 15–20% or so? Or are they running extremely tight tolerances of 5%, +/-3, or something in that range? If they are running a very tight tolerance, changing even the core of the pre-preg or the laminate is going to adversely affect the im- pedance calculations. Although there might be some materials that are very similar in charac- teristics and thermal profiles, they might tweak the stack-up just enough to throw off the calcu- lation at a point where the final product starts failing at the higher speed levels. Is it the components? Certain questions dealing with ICs and other components on that questionable board also need to be asked and carefully probed. Let's look at the specifications and how they come about. Was the device or component in ques- ZUlKI'S PCb NUggETS Figure 1: Thermal profile showing the actual peak temperature of U1 at 241.51°C. This profile is for a lead-free assembly and the manufacturer's suggested peak temperature range is 230–250°C. uNCOVerING aSSeMbLy PrObLeMS OF HIGH-SPeeD PCbS continues