Issue link: https://iconnect007.uberflip.com/i/1524164
52 PCB007 MAGAZINE I JULY 2024 the spot size, these applications are now cov- ered by one machine from Schmoll with a sim- ilar capital investment to a single traditional UV/YAG, with a much smaller heat-affected zone. MBSE and Recipe Development Over the years, the pace of recipe modeling capability improvement has been very slow in the PCB fab industry. Inner layer scaling and dielectric prediction are relatively mature nowadays, but copper patterning tolerances and metal/mask coating thicknesses remain a challenge. In this case, the challenge lies in the sensor quality deficiencies in building the models, as well as a low level of control over the processes. Recently, multiple fabs have implemented 3D interferometric and confocal microscopy technologies which are capable of profiling surface features very effectively. At the same time, the use of specialized throw- ing power test coupons to optimize and mon- itor health of lines for plating through-holes has created the feedback sense loop the mar- ket has been missing. By designing in these new sensing techniques and incorporating the feedback into a basic machine learning sys- tem, while also adopting a galvanic Cu process with onboard complex controls, it is very pos- sible to exclude most trial-and-error test pan- els from fab sites. In addition, despite being very complex sys- tems, systems engineering documentation is entirely lacking in most PCB fab operations. As a result, change simulation and impact across operations are almost never understood. Recently, we have developed a multi-domain interface model in lifecycle modeling lan- guage (LML) with a custom microelectronics grammar which allows for documentation and Monte Carlo yield/cycle time simulation of the whole fab process/product interaction in PCB fab, thus allowing parameterization and simu- lation of new processes and products prior to release to avoid test panels. On complex prod- ucts, this new method is a game-changer. Figure 1: Comparison of typical process utilization.