Issue link: https://iconnect007.uberflip.com/i/750006
November 2016 • The PCB Design Magazine 17 CAT'S DAVID WOLF ON VIA RELIABILITY ANALYSIS a peak temperature of 260ºC and 500 air-to-air thermal cycles from -40ºC to +145ºC. We have collected via reliability data from many of the other standardized PCQR2 test panel designs that exhibit greater than a 5% change in via net resistance after the 6X con- vection reflow assembly simulation. May times those same via nets will not show failures (great- er than 10% change in via net resistance) after 500 air-to-air thermal cycles; this due to the difference in z-axis expansion between +145ºC and +260ºC. Another design trend that is becoming more common for higher layer-count boards, 14 lay- ers and above, is the use of two-high (2+N+2) and three-high (3+N+3 or 3+N+N+3) staggered or stacked microvias on either side of multi- layer sub-composite mechanically drilled vias. The unknown factor with these types of com- posite via structures is whether the stacked or staggered version is more reliable. The PCQR 2 database subscribers are requesting that these composites via structures be added to the next revision of PCQR 2 test panel designs so that via reliability benchmarking data can be collected. One major challenge with forming stacked via structures is registration during the sequential build-up of the via structures. Shaughnessy: What are some of the more common via defects you see? How do blind/buried vias fare as far as defects? Wolf: For through vias and sub-composite mul- tilayer buried or blind via structures, the de- fects typically seen after assembly simulation and air-to-air thermal cycling are barrel cracks, knee cracks and/or inner layer separation. In blind microvia structures, whether staggered or stacked, separation of via hole plating from the target pad is a common defect. Generally, 1-deep (layer 1-2) and 2-deep (skip layer 1-3 with no interconnect on layer 2) blind micro- via structures tend to be more reliable than high aspect ratio through or sub-composite via struc- tures. This is due to less z-axis expansion associ- Figure 3: Through, blind and stacked vias shown after reflow simulation.