How to optimize the thermal design of printed circuit board layout

Page Introduction

Classification	electronic thermal simulation
Keyword	Thermal simulation, thermal analysis, pcb design, Flotherm, optimization method
Article core	In this article, I would like to show a method that using response surface to optimize the thermal design of printed circuit board layout.

Catalogs

Catalogs	I.Why should the thermal simulation be used to optimize the heat dissipation of PCB board
	II.Initial layout scheme
	III.Design goals and limitations
	IV.Thermal simulation
	V.Response surface
	VI.Conclusion

Text

I.Why should the thermal simulation be used to optimize the heat dissipation of PCB board

In order to meet the increasing requirements of the PCB design, many design engineers feel a lot of pressure. Each new type of design is accompanied by a failure risk in terms of performance and reliability. The biggest problem in the design process is how to take off in the heat dissipation scheme and signal integrity. The high-speed clock speed of the connecting element requires close proximity so as to ensure no signal attenuation. But these elements are also unavoidable to have a lot of heat dissipation, so they should be as far away as possible, thereby helping to reduce their temperature.

This paper describes how to use thermal simulation to optimize the heat dissipation performance of PCB board. The PCB plate is locked in the chassis by a wedge and is forced to air cooling on the radiator fins outside the chassis. Under some harsh environmental conditions, according to local ambient air temperature and heat conduction as the main way of cooling, how to achieve normal junction temperature has become a major problem.

II.Initial layout scheme

Figure 1 initial plane layout and important components and wedge locking devices

Figure 1 shows the original layout of the plane. By external chassis forced air cooling can make the PCB wedge locking device at 35 degrees the temperature of C. The local air temperature is 75 degrees C. Although all the components have thermal dissipation, microprocessors and memory are the main components of the thermal dissipation on the entire PCB board.

III.Design goals and limitations

There are many ways to make thermal design for layout, but they all follow a principle: how to transfer heat to the outdoor environment quickly and conveniently. In this case, we use the FloTHERM software of Mentor Graphics Mechanical Analysis to carry out numerical simulation for two improved methods which help to exclude heat.

First, memory and microprocessors are kept away from each other at different distances. Here we keep the memory position unchanged. There are two benefits of doing this, one that moves the processor's position and reduces its heat impact on memory. In addition, the position of the processor is closer to the cuneiform locking device to obtain lower temperature.

Secondly, the influence of the array heat passing through the memory and the lower part of the microprocessor is calculated. The hot hole is amplified in Figure 2. Thermal pores contribute to heat into the inner metal layer PCB board, especially those who almost covered the whole PCB board power supply layer and stratum, in these layers can rapidly transfer heat to the edge of the wedge locking device. Without these thermal holes, there is a great thermal resistance in the microprocessor and wedge locking devices, which is mainly due to the large thermal resistance of the signal layer at the top of the PCB plate.

Figure 2 memory and microprocessor distance and hot hole array

The latest design of this type of PCB board uses the signal frequency of GHz and the rising time of 100 billionth of a second signal to work. Because this kind of rising time has the same state as wavelength, the attenuation of key signal may be greatly increased. So the distance between the memory and the microprocessor should be as short as possible, and in this case it should not exceed 11mm.

The microprocessor (package TBGA) the maximum rated junction temperature is 100 degrees C. Although component providers provide some data for characterizing thermal performance (for example, thermal resistance between chip nodes and environment), these data can only be applied to some specific situations. The most reliable thermal design method is only for the whole PCB board module for 3D simulation.

IV.Thermal simulation

However, the traditional simulation method is only focused on a single study, providing only a feasible or infeasible conclusion. Excellent numerical simulation should be able to study the effects of design changes on heat dissipation. This will help the design engineer to determine the parameters of the optimized design to achieve the entire design goal.

This can be accomplished by building and simulating a design experiment (DoE). The use of this method first needs to determine the design variables. In this example, these variables are the distance between the memory and the microprocessor, and the array density of the heat through the holes under these components. In all 20 simulation programs based on the maximum Wen Shengcheng different combinations and the corresponding microprocessor the two changes the parameters of a 3D.

Figure 3 shows two extreme DoE designs. The scheme 1 is that there is no thermal hole passing under the component and very close to the microprocessor and memory. 2, there are dense hot holes in four components and the location of the microprocessor is very close to the wedge locking device.

Figure 3 the worst and optimal design results

V.Response surface

The results of the 20 simulation schemes give us an intuitive understanding of the heat dissipation of these schemes, for example, the junction temperature of the optimal and worst scheme shown in Figure 3. However, the "response surface" can be fitted by using the results of 20 simulation solutions to obtain more intuitive and complete 3D results. This response surface fitting is a very advanced curve fitting. It combines the influence of the two design parameters to the temperature (Figure 4) perfectly and gives a visual and clear view of the observation.

Note: the thermal conductivity of the lower part of the component is quantified in the form of thermal conductivity. 0.3 W/mK (FR4 thermal conductivity) indicates that there is no hot hole in the lower part of the component. 10 W/mK indicates a dense thermal overhole array in the lower part of the element. 

Figure 4 the response surface between mild design parameters

Response surface 3D figure 4 shows that between memory and microprocessor distance is, the lower the junction temperature of the processor. In addition, the heat crossing density in a relatively small range has a great influence on the junction temperature. If the distance between memory and microprocessor is not considered, in a range of thermal conductivity of about 0~3 W/mK, a small amount of thermal through-hole density will achieve significant cooling effect. There is only a small amount of benefit to further increase the array density of the heat passing through the element. 

  

A more quantified chart can be obtained by using Figure 5. The variable lines displayed are only part of the figure 4, and the array density of the heat through the holes is a variable line. The distance between the elements and the limit of the maximum junction temperature are expressed in a black straight line. By observing the response curve that allows for the design range, it is obvious that by maximizing the number of thermal holes under the element, the whole design will have some allowance.

Figure 5 response surface areas characterized by design constraints

VI.Conclusion

By using the experimental design (Design of Experiments) function, completed a number of numerical simulation, the simulation results created by response surface 3D map, and the design objective of design variables with the response have an intuitive understanding. This helps to quickly determine the compromise in design and minimize the risk of heat dissipation due to the lack of response relationship between design objectives and design variables.

The example described in this paper describes the effect of the interaction of two independent variables on the heat dissipation. In fact, this simulation method can be applied to any number of design variables. For example, as the distance between the two microprocessors and the wedge locking device is variable, the distance between the memory and the wedge locking device can also be used as the design variable. In reality, the limits of the optimization of simulation research are determined by the design engineer and the available resources and time.