The Parameters option contains three commands:

• Runtime enables you to vary the number of analysis iterations, convergence values, and relaxation factors.
• Grid enables you to vary the number of x and y nodes, and hence the number of elements into which the package is divided for the analysis.
• Legend enables you to change the temperature grid range.

• Select the Runtime command from the Parameters option header menu. The window in Figure 10 appears.
Figure 10. The Runtime Command  
• Maximum iterations is the first field you must assign a value. The value is the maximum number of iterations CADMP-II will use when attempting to solve for individual node temperatures. CADMP-II can perform 100 to 2,000 iterations (see Step 3).

This field is included because certain designs prove very challenging to the computational algorithms. The physical conditions that might result in such designs include very high total board power ( > 50 watts), large numbers of nodes, or very high power-to-area ratios relative to the overall package dimensions. In such cases, the thermal analysis may not converge before it reaches the maximum number of iterations allowed. If the program does not converge, you should check individual active/passive element powers to be certain that no value was entered incorrectly.
 

• Convergence value, the next field requiring as assigned value, and has a range of .001 to .5.

Based on the finite difference approach for steady-state thermal analysis, the energy generated by a node should equal the difference between the amount of energy going into the node and the amount of energy leaving the node. When a balance is achieved for all of the nodes in the problem, a unique solution has been established. In a computer simulation, it is generally impossible to achieve an exact balance due to round-off error and the accuracy model. Therefore, a value is specified for which the solution of the problem is determined to be acceptable. The convergence value represents this number.

The convergence value should be based on the problem size. CADMP-II calculates the total amount of unresolved energy as it performs each iteration by summing the unresolved energy for each node. When this sum is less than the specified convergence value, CADMP-II terminates the calculations and assumes the currently calculated node temperatures represent the solution to the thermal problem.

• Conductivity Convergence, the next field requiring an assigned value, has a range of 0.1 to 99.9.

 
The primary material property used in a thermal analysis is the thermal conductivity. For many materials, this property is temperature dependent. The initial temperature value used to determine the thermal conductivity for each node is the average boundary temperature used to set up the thermal analysis. The post-analysis nodal temperatures always differ from the average boundary temperatures. Therefore, the initial values of thermal conductivity are no longer accurate.

The conductivity convergence field allows you to specify an allowable percentage deviation for nodal conductivity values between successive runs of the thermal analysis. If the maximum deviation is greater than the user-specified value, you will be asked whether you want to re-run the thermal analysis with the newly calculated nodal conductivities. This process continues until the convergence criteria is met.
 

• Relaxation factor, the final field requiring an assigned value, has a range of 1-1.75.

In a finite difference approach, the temperatures of the individual nodes are recalculated during each iteration. A current node temperature is based on its previous node temperature plus the residual temperature based on the energy balance for the node. If the problem is convergent, then the residual temperature becomes smaller after each iteration. The residual approaches zero as a convergent solution is determined. To speed up the convergence, the residual can be multiplied by a constant. Since the residual goes to zero, the constant is negligible when a solution is achieved. The constant range is limited to between 1-1.75 because if the constant is too large, the algorithm may overshoot the solution and diverge, and if it is too small, the algorithm may undershoot the solution. When a constant is multiplied by the residuals, the problem is said to be relaxed and the constant is called the relaxation factor. CADMP-II uses the relaxation factor when performing the finite difference algorithm.
 

Select the OK button the window to accept your entries. Select the Cancel button to cancel.

Next Page