Reading time ( words)
I've been writing on thermal management for LED applications for a few years now, seemingly on an endless quest to find the next best thing for LED PCBs. It's been kind of like Indiana Jones and his search for the Holy Grail, except I don't have cool bullwhip skills. To date, we've focused on calculating thermal management needs as well as explore other alternatives to MCPCBs to achieve thermal management, such as standard plated through holes in FR4 material. I'd say at this point the industry has reached a saturation point when it comes to knowledge on how to dissipate heat (of course you never know what’s around the corner).
I think now it's time to move on to a topic that we've overlooked when it comes to PCBs for LEDs: reliability! I think the reason we've overlooked reliability is because it is traditionally associated with via life under thermal cycling conditions. Since most LED PCBs are single sided, this really hasn't been an issue. However, if you look above the waterline, there is another weak link—the solder joint. Some OEMs in the industry have been performing studies on the life of their LED products. While the bulbs themselves are said to have useful lives in excess of 30+ years, they are finding out that the actual LED assemblies can fail in as little as 5–6 years. Initial analysis is pointing to the significant X and Y axes CTE differences between the solder joint, copper circuitry layer, thermally conductive dielectric, and the aluminum. The net result of the CTE differences is a shear effect being created that can eventually disrupt the solder joint, which results in operational failure.
Depending upon how accurate this information is, it could mean the start of a whole new approach to PCBs for LED applications. Below is an abstract of a white paper written by Thomas Tarter from Package Science Services, which performed initial testing on carbon fiber and graphite based materials provided by Stablcor Technology Inc. The carbon fiber constraining cores (CFCC) materials evaluated are carbon-fiber and/or graphite reinforced epoxy cores to aid in heat dissipation, rigidity, weight reduction, and CTE control. These cores can be used independently or in conjunction with current MCPCBs to produce functionally improved heat dissipation while reducing the CTE mismatch currently present on LED assemblies.
Abstract: Introduction and Model Parameters (by Thomas Tarter)
Thermal performance for PCB structures are investigated in the form of steady-state finite element models of various stack-ups of commonly used materials for LED applications. The models show the effect of materials used in the stack up including FR-4, aluminum, copper, graphite and CFCC. The goal of the study is to compare relative thermal behavior of typical boards modified with the enhanced core materials.
The materials are inserted into standard PCB stack-up configurations as an added or replaced layer. Models are solved for maximum temperature on a 25 mm x 25 mm coupon with a 2 mm x 2 mm-square heat source. The stack up resembles substrates known as "metal-clad" where the dielectric and topside copper are laminated directly onto a metal substrate. In addition, FR4 boards are used as a worst-case comparison.
Variables used in the study include material properties and layer thickness. The primary variables are top side copper thickness/weight, dielectric thickness and base material thickness. Table 1 lists the ranges for geometry and material properties. The heat source is simulated as a planar load, directly on the surface of the top-layer copper. One watt is applied over a 2mm x 2 mm square area in the center of the coupon. The models are solved in natural convection with an ambient temperature of 30°C.
Editor's Note: This article originally appeared in the March issue of SMT Magazine.
