Advertisement

As electronic systems become more complex and integrated into nearly all the devices and networks we interact with today, the importance of systems reliability — and the costs of systems failure — are greater than ever before. If a system fails, it can not only cost companies millions of dollars in repair and warranty costs, it can be fatal. Take, for example, the AirAsia flight that crashed in December 2014, killing all of its 162 passengers and crew. The official report concluded that the sequence of events leading to the crash began with a tiny crack, caused by thermal stress and fatigue, in a soldered electrical connection on a printed circuit board (PCB)1.

Thermal stress is a leading cause of PCB damage, so it is a constant concern in PCB design. It is caused by the difference in the coefficient of thermal expansion (CTE) for the copper and the epoxy resin typically used in a PCB. As the temperature changes from hot to cold in repeated cycles, the solder balls that provide contact between the PCB and the integrated circuit (IC) package are stretched and squeezed. This can cause separation from the IC package or the PCB, leading to disastrous electrical failure.

In the past, it was not practical to accurately simulate the amount of thermal deformation of a proposed board design and accurately identify areas affected by thermal stress. Engineers had a choice between two techniques, a low fidelity “lumped” and a high fidelity “explicit” approach.

Fig 1: IcePak simulation using the “High Fidelity” approach
Fig 1: IcePak simulation using the “High Fidelity” approach
Fig 2: IcePak simulation using the “High Fidelity” approach

With the “lumped” approach, you reduce the complexity of the model by building an approximation of the board’s individual components and giving all the layers of the PCB the same material properties. Since the multitude of traces and vias are not discretized the mesh count is much lower, solver time is reduced. However, the results are not very accurate and you probably won’t be able to pinpoint large heat gradients and areas of stress concentration.

WIth the “explicit” approach, you discretize the entire board with all the thousands of traces, vias, and other interfaces. While this approach may yield a highly accurate solutions, the time to prepare the mesh and compute the solution is often impractical, particularly in early design iterations.

Display of a Metal Fraction
Fig 3: Display of Metal Fraction

The Efficient “Metal Trace Mapping” Technique
ANSYS has overcome the deficiencies of the previous two approaches with a new multiphysics methodology that simplifies PCB geometry while accurately representing its material properties at any point. This technique, called “Trace Metal Mapping”, efficiently simulates board performance under thermal loading. With this technique, a hexahedron mesh is initially created which discretizes each layer of the PCB, ignoring the interfaces between the thousands of metal traces and silicon. Next the ECAD information is imported and the material fractions for each element are calculated and mapped onto the mesh. The result is a fast, easily defined mesh and mapping that takes only minutes to set up. Using ANSYS Icepak, the temperature fields on and around the board are then determined using CFD (Computational Fluid Dynamics). You could then apply these temperature fields to the model as input conditions for a coupled thermal-stress analysis in ANSYS® Mechanical™. Just as importantly, you can create the model and run the analysis on a single simulation platform, which greatly simplifies the entire process.

Fig 4: Thermal Flow Simulation in IcePak
Fig 4: Thermal Flow Simulation in IcePak

Computing DC Solution
The first step is to use ANSYS SIwave to compute DC currents and voltages throughout the PCB. SIwave then calculates the current density throughout the board which in turn determines Joule heating (the process by which heat is produced due to electrical resistance when an electrical current flows through a conductor). Joule heating has increasingly become important as a source of thermal loading in PCBs as board sizes are reduced and power consumption remains steady or rises.

Performing Thermal Simulation New integration between SIwave and IcePak
With the release of ANSYS 17.0, the SIwave and Icepak bidirectional workflow has been integrated and automated, making it much easier to determine Joule heating. This capability imports the board, trace map and current density predictions, and sets the thermal boundary conditions for Icepak. Icepak uses the trace map to calculate the orthotropic thermal conductivity of the PCB. This is important because much of the heat generated on the board is dissipated via convection or radiation from the board itself and the material makeup of the board has a directional nature. Icepak solves fluid flow equations and includes all modes of heat transfer — conduction, convection and radiation — to compute temperatures at every point in the solution domain. The macro then exports the resulting temperatures from Icepak back into SIwave, which updates the electrical properties for the DC solution based on the temperature field. SIwave then recalculates the DC field and exports it to Icepak. This iterative process continues until the power dissipation and temperature results have converged.

Advertisement


SHARE YOUR THOUGHTS & COMMENTS

Please enter your comment!
Please enter your name here