CN109376372B - Method for optimizing postweld coupling efficiency of key position of optical interconnection module - Google Patents

Abstract

The invention discloses a method for optimizing coupling efficiency after welding of key positions of an optical interconnection module, which comprises the steps of firstly designing 32 groups of experimental combinations by using a response surface method, establishing 32 corresponding groups of simulation models according to the 32 groups of experimental parameters, then obtaining a functional relation between the coupling efficiency after welding and key factors of the key positions of the optical interconnection module by using the response surface method, carrying out variance analysis on the obtained functional relation, and determining the effectiveness of a regression equation; optimizing the regression equation by using a genetic algorithm, sequentially executing initial population generation, crossing, variation and evolution reversion operation to obtain an optimal combination which is most beneficial to improving the coupling efficiency of the optical interconnection module after welding at the key position, and finally verifying the optimal combination by establishing a corresponding optical interconnection module finite element analysis model; the method has excellent robustness, is simple in calculation, brings great convenience to later-stage parameter optimization design, and the optimized calculation result is ideal.

Description

A method for optimizing post-soldering coupling efficiency at key positions of optical interconnection modules

technical field

The invention relates to the technical field of microelectronic packaging optical interconnection, in particular to a method for optimizing post-soldering coupling efficiency at key positions of an optical interconnection module based on a response surface method and a genetic algorithm.

Background technique

Optical interconnection technology is one of the new interconnection methods to solve the signal transmission bottleneck problem encountered by electrical interconnection technology in surface mount technology (Surface Mount Technology, SMT) and micro-assembly due to high density and miniaturization. Has a very good development prospects. However, due to the influence of factors such as heat and vibration in the process and working environment during the packaging and assembly process, the alignment position of the optical path will shift, resulting in a decrease in coupling efficiency, which has become a key problem that needs to be solved urgently in the application of this technology. A typical optical interconnection module of the present invention is taken as the research object, and after the optical interconnection module is assembled, the heat and vibration of the working environment where the actual application is located may affect the offset of the module alignment position. Using the finite element method, under the conditions of temperature change and vibration coupling, the offset generated by the alignment position of the optical interconnection module in the three axes and relative angles of the Cartesian coordinate system was analyzed, and the offset data was analyzed The coupling efficiency has been simulated and calculated. In order to improve the coupling efficiency of the optical interconnection module under the joint action of multiple physical fields, the finite element modeling of the optical interconnection module with single-factor changes in the solder joint material and structural geometric parameters has been completed, and carried out Coupling simulation of multi-physics field, analysis of offset at the alignment position and analysis and optimization of coupling efficiency, that is, through ANSYS finite element analysis software simulation analysis of the offset at the alignment position, and then use ZEMAX software to analyze the coupling efficiency After the simulation analysis and the analysis and processing of the analysis results, use the Design-Expert response surface analysis software and Matlab software to analyze the analysis factors on the coupling efficiency of the optical interconnection module by optimizing the response surface algorithm and the genetic algorithm, so as to improve the efficiency of the optical interconnection module. Coupling efficiency of interconnected modules. Genetic algorithm is a global optimization algorithm in computational mathematics, which is very suitable for solving large-scale combinatorial optimization problems. The layout of electronic components belongs to the traveling salesman (TSP) problem in combinatorial optimization. In recent years, some scholars have applied genetic algorithms to research in this field. Therefore, using standard genetic algorithms for optimization can get better results, and it is easy to achieve optimization results .

Contents of the invention

The purpose of the present invention is to overcome the deficiencies of the prior art, and provide a method for optimizing the post-welding coupling efficiency of the key position of the optical interconnection module. The method has excellent robust performance, and the calculation is relatively simple. It is very convenient, and the calculation result after optimization is ideal.

The technical scheme that realizes the object of the present invention is:

A method for optimizing post-soldering coupling efficiency at key positions of an optical interconnection module, specifically comprising the following steps:

1) Establish the finite element analysis model of the optical interconnection module;

2) After the finite element analysis model of the optical interconnection module is analyzed by the reflow soldering finite element, the alignment offset at the key position of the optical interconnection module is obtained;

3) Use ZEMAX to calculate and obtain the coupling efficiency of the key position of the optical interconnection module after welding;

4) Establish the influencing factors that affect the coupling efficiency;

5) Establish parameter level values of influencing factors;

6) Use the BOX-Behnken central composite design model to design the required 32 groups of experimental samples;

7) Obtain the functional relationship between influencing factors and coupling efficiency;

8) Analysis of variance is performed on the obtained functional relationship;

9) Establish the correctness of the obtained functional relationship;

10) Randomly generate the initial population;

11) Obtain the current evolution algebra gen and optimal fitness value;

12) Implement cross-operation on populations respectively;

13) Implement mutation operations on the populations respectively;

14) Implement evolutionary reversal on the populations respectively;

15) Calculate the fitness function value of the two populations as a whole, and use the optimal preservation strategy to select the best individual;

16) Re-judgment after the population is updated, if the gen value is less than 50 and the num value is greater than 0, a local catastrophe will be implemented on the population.

In step 1), the model includes three layers of PCBs, solder balls, optical coupling elements and embedded optical fibers. The solder balls are set between two adjacent layers of PCBs, and the optical coupling elements are set at the center of the lower PCB. The embedded optical fiber is set on the lower PCB. The size of the upper PCB is 27×27×1.52mm; the size of the middle PCB is 35×35×1.52mm; the size of the lower PCB is 55×50×1.52mm; the optical coupling element The radius is 0.0625mm and the length is 2.76mm; the radius of the embedded optical fiber is 0.0625mm and the length is ^30mm ; the radius of the welding pad is 0.3mm; The volume of the solder balls is 0.2mm ³ , the height is 0.48mm, and the pitch is 1.5mm.

In step 4), the influencing factors are the height H ₁ of the upper solder joint, the height H ₂ of the lower solder joint, the radius R of the solder pad, the center distance L of the solder joint and the volume V of the solder joint.

In step 5), the number of levels of the parameter level value is 5, and the number of factors is 5.

In step 6), 32 groups of experimental samples are designed using the BOX-Behnken central combination design model, of which 26 groups are analysis factors and 6 groups are zero-point factors, that is, the same parameter level combination is used for experimental error estimation.

In step 10), the initial population, the population size is set to 40.

In step 11), the algebra gen and genetic algebra are set to 50.

The present invention provides a method for optimizing the post-solder coupling efficiency of the key position of the optical interconnection module. The method accurately approximates the functional relationship between the factor and the target value within a certain range through a small number of experiments, and uses a simple expression to show In addition, complex response relationships can be simulated within a certain range through the selection of the regression model, which has excellent robust performance, and the calculation is relatively simple, which brings great convenience to the later parameter optimization design.

Description of drawings

Figure 1 is a basic model diagram of the optical interconnection module;

Figure 2 is the ZEMAX analysis result diagram of the basic model;

Fig. 3 is the mean value change diagram after the regression equation is optimized by the genetic algorithm;

Fig. 4 is the change diagram of the optimal solution after the regression equation is optimized by the genetic algorithm;

Fig. 5 is the geometric image analysis result diagram of the optimal combination ZEMAX.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and embodiments, but the present invention is not limited thereto.

Example:

A method for optimizing post-soldering coupling efficiency at key positions of an optical interconnection module, specifically comprising the following steps:

(1) Establish the basic model of the optical interconnection module. The basic dimensions of the model are shown in Table 1, and the model is shown in Figure 1;

(2) The key position of the optical interconnection module is obtained after the model is analyzed by the reflow soldering finite element: the alignment offsets at the center point A of light emission and the center point B of optical coupling are shown in Table 2;

(3) Using the geometric image analysis function of ZEMAX, the coupling efficiency after soldering at the key position of the optical interconnection module is 87.89%, and the analysis result is shown in Figure 2;

(4) Obtain the influencing factors that affect the coupling efficiency: the height of the upper solder joint, the height of the lower solder joint, the radius of the solder pad, the center distance of the solder joint and the volume of the solder joint; select 5 levels for each factor, and the factor levels are as follows Shown in Table 3;

(5) Using the BOX-Behnken central combination design model, there are 32 groups of simulation model level combinations, of which 26 groups are analysis factors and 6 groups are zero-point factors, that is, the same parameter level combination is used for experimental error estimation; 32 groups of parameters The combined results are shown in Table 4;

(6) According to calculus knowledge, any function can be approximated by several polynomial pieces, so in practical problems, regardless of the complexity of the relationship between variables and results, polynomial regression can always be used to analyze and calculate. There are 5 variables and the functional relationship between the variables and the target is non-linear. Combined with the number of experimental samples in Table 4, a second-order polynomial model based on Taylor expansion is selected:

（A）

(A)

、线性项

、线性交叉项

、二次项

。

为线性项系数；

为线性交叉项系数；

为二次项系数；

为随机误差；x为设计变量；Y为耦合效率；n为变量个数。(A) Include a constant term in the formula

, linear term

, linear cross term

, quadratic term

.

is the coefficient of the linear term;

is the linear cross-term coefficient;

is the quadratic coefficient;

is the random error; x is the design variable; Y is the coupling efficiency; n is the number of variables.

(7) Perform quadratic multiple regression fitting on the experimental factor combinations and their results in Table 1 to obtain the coupling efficiency (Y) effect on the height of the upper solder joint (X ₁ ), the lower solder joint height (X ₂ ), and the pad radius ( The quadratic polynomial regression equations of X ₃ ), solder joint center distance (X ₄ ) and solder joint volume (X ₅ ) are:

（B）

(B)

(8) In order to ensure the credibility of the regression equation, variance analysis and model significance verification were carried out on the data in Table 3, and relevant evaluation indicators of the regression equation were obtained. The results are shown in Table 5;

(9) The model "Preb>F" obtained by the response surface analysis is less than 0.0001 (generally less than 0.05 means that the item is significant), indicating that the regression effect of the response surface model is particularly significant; the regression equation coefficient (R-Squared) is 0.9979, indicating that the regression equation The fitting degree is very high; the regression equation adjustment coefficient (Adi R-Squared—represents the fitting accuracy after removing insignificant indicators) is 0.994, which more accurately reflects the high fitting accuracy of the equation; the regression equation prediction coefficient (Pred R- Squared) is 0.9505, indicating that the prediction accuracy of the equation is very high; the signal-to-noise ratio (Adeq Precision) of the equation is 71.319, indicating that the regression equation is less affected by interference factors; the coefficient of variation (CV) of the equation is 0.068, indicating that the test operation is reliable. The above result coefficients all show that formula (B) can highly fit the test results in Table 4, and the regression equation is accurate and credible;

(10) The genetic algorithm is used to optimize the appeal regression equation. The algorithm first randomly determines a set of initial solutions from the definition domain, and then searches for the optimum of the objective function within the domain or the algorithm first randomly determines a set of initial solutions from the definition domain , and then search for the optimal or suboptimal solution of the objective function within the range;

The genetic algorithm optimizes the regression equation, specifically as follows:

Step a: Randomly generate the initial population;

Step b: Obtain the current evolution algebra gen and optimal fitness value;

Step c: Carry out the cross operation on the population respectively;

Step d: Carry out mutation operations on the populations respectively;

Step e: implement evolutionary reversal on the populations respectively;

Step f: Calculate the fitness function value of the population as a whole, and use the optimal preservation strategy to select the best individual;

Step g: re-judgment after the population is updated, if the gen value is less than 50 and the num value is greater than 0, implement a local catastrophe on the population, and then return to step b, otherwise directly return to step b; the maximum genetic generation of the algorithm is set to 50 generations, and the gen value If it exceeds 50, the evolution will be terminated.

(11) Through the MATLAB genetic algorithm toolbox, optimize the parameters with the goal of the highest coupling efficiency; the average value of the problem and the change of the optimal solution are shown in Figure 3 and Figure 4.

(12) According to the value range of the influence factor set in the appeal factor parameter table, the optimal combination is obtained: the height of the upper solder joint is 0.45mm, the height of the lower solder joint is 0.65mm, the radius of the solder pad is 0.43mm, and the center of the solder joint The distance is 1.5mm and the solder joint volume is 0.43mm ³ , and the coupling efficiency value is 98.13%.

(13) According to the final parameter combination obtained above, establish the corresponding optical interconnection module simulation model, and obtain the key position of the optical interconnection module after reflow soldering finite element analysis: the alignment of the light-emitting center point A and the optical coupling center point B The offset is shown in Table 6; the coupling efficiency calculated by using the geometric image analysis function of ZEMAX is 98.172%, as shown in Figure 5, which is very close to the predicted value of the genetic algorithm, which proves that the genetic algorithm optimizes the key position of the optical interconnection module Effectiveness of coupling efficiency after welding.

Table 1 Basic dimensions of the model

。

.

Table 2 Alignment offset of key points A and B in 2080s

。

.

Table 3 Structural parameter factor level table of optical interconnection module

。

.

Table 4 Combination results of 32 groups of parameters

。

.

Table 5 Response surface analysis results

。

.

Table 6 Alignment offsets of key points A and B of the optimal model

。

.

Claims (3)

1. A method for optimizing the post-welding coupling efficiency of optical interconnection module key positions, is characterized in that, specifically comprises the following steps: 1) Establish the finite element analysis model of the optical interconnection module; 2) After the finite element analysis model of the optical interconnection module is analyzed by the reflow soldering finite element, the alignment offset at the key position of the optical interconnection module is obtained; 3) Use ZEMAX to calculate and obtain the coupling efficiency of the key position of the optical interconnection module after welding; 4) Establish the influencing factors that affect the coupling efficiency; 5) Establish parameter level values of influencing factors; 6) Use the BOX-Behnken central composite design model to design the required 32 groups of experimental samples; 7) Obtain the functional relationship between influencing factors and coupling efficiency, where: Choose a second-order polynomial model based on Taylor expansion

The quadratic polynomial regression equation of the coupling efficiency Y to the height of the upper solder joint X1, the lower solder joint height X2, the pad radius X3, the solder joint center distance X4 and the solder joint volume X5 is:

8) Analysis of variance is performed on the obtained functional relationship; 9) Establish the correctness of the obtained functional relationship; 10) Randomly generate the initial population; 11) Obtain the current evolution algebra gen and optimal fitness value; 12) Implement cross-operation on populations respectively; 13) Implement mutation operations on the populations respectively; 14) Implement evolutionary reversal on the populations respectively; 15) Calculate the fitness function value of the two populations as a whole, and use the optimal preservation strategy to select the best individual; 16) Re-judgment after the population is updated, if the gen value is less than 50 and the num value is greater than 0, a local catastrophe will be implemented on the population; In step 1), the model includes three layers of PCBs, solder balls, optical coupling elements and embedded optical fibers. The solder balls are set between two adjacent layers of PCBs, and the optical coupling elements are set at the center of the lower PCB. The embedded optical fiber is set on the lower PCB. The size of the upper PCB is 27×27×1.52mm; the size of the middle PCB is 35×35×1.52mm; the size of the lower PCB is 55×50×1.52mm; the optical coupling element The radius is 0.0625mm and the length is 2.76mm; the radius of the embedded optical fiber is 0.0625mm and the length is ^30mm ; the radius of the welding pad is 0.3mm; The solder ball volume is 0.2mm ³ , the height is 0.48mm, and the pitch is 1.5mm; In step 4), the influencing factors are the upper solder joint height H ₁ , the lower solder joint height H ₂ , the pad radius R, the solder joint center distance L and the solder joint volume V; In step 5), the number of levels of the parameter level value is 5, and the number of factors is 5; In step 6), 32 groups of experimental samples are designed using the BOX-Behnken central combination design model, of which 26 groups are analysis factors and 6 groups are zero-point factors, that is, the same parameter level combination is used for experimental error estimation. 2 . The method for optimizing post-soldering coupling efficiency at key positions of an optical interconnection module according to claim 1 , wherein, in step 10), the population size of the initial population is set to 40. 3 . 3 . The method for optimizing post-soldering coupling efficiency at key positions of an optical interconnection module according to claim 1 , wherein, in step 11), the algebra gen and the genetic algebra are set to 50. 4 .

Images