Multichip module analyzer

• US 5,675,521 A
• Filed: 05/02/1995
• Issued: 10/07/1997
• Est. Priority Date: 02/11/1994
• Status: Expired due to Term

• Current Assignee: United States Air Force
• Original Assignee: United States Of America As Represented By The Secretary Of The Air Force
• Inventors: Grosse, Ian R., Richards, Dale W., Katragadda, Prasanna, Holzhauer, Douglas J., Corkill, Daniel D.
• Primary Examiner(s): Trammell, James P.
• Assistant Examiner(s): Assouad, Patrick J.
• Application Number:
  US08/434,426
• Time in Patent Office: 889 Days
• Field of Search: 364/578, 395/500, 395/53, 395/600, 395/10, 395/50, 395/51, 395/54, 395/60
• US Class Current: 716/136
• CPC Class Codes: G06F 2119/08 Thermal analysis or thermal optimisation
• CPC Class Codes: G06F 30/23 using finite element methods [FEM] or finite difference methods [FDM]

Abstract

The disclosure describes a method for performing thermal reliability analysis of electronic devices such as multichip modules. The method supports the reliabilty of multichip technology during the design phase by integrating traditional thermal analysis techniques, such as Finite Element Analysis with artificial intelligence techniques. Specifically, the use of object oriented programming, blackboard architecture and knowledge sources (based on expert systems) allow the computer to perform lower level reasoning associated with the development of the finite element mesh. The use of software, called Intelligent Multichip Module Analyzer, results in a great reduction in the amount of time required to model and to perform thermal analysis of multichip modules. This allows the analysis to be integrated with the design process so that reliability assessment can be accomplished when it can best affect the final design.

Claims

1. An intelligent thermal analyzer method for analyzing the thermal characteristics of an electronic device, said method comprising:

inputting a graphic model of the electronics device into an object template of a generic blackboard, said model comprising four basic components of said device, including a package, substrate, interconnect and chip;

providing a plurality of knowledge sources, said knowledge sources including a symmetry knowledge source, model source and an extrusion source, each of said knowledge sources having an output to said generic blackboard;

activating the symmetry knowledge source to provide an input to said generic blackboard to generate new descriptions of said model representing a plurality of parts of the model to be analyzed, said parts each constituting a subset of said model;

providing an output from said model source to said blackboard, to generate a two dimensional geometric model of the subsets of said device;

applying an output from said extrusion source to form three dimensional geometric models of said subsets of said device;

then combining said subsets to create a finite element analysis code; and

displaying said finite element analysis code.

2. The method of claim 1 wherein said extrusion source chooses the two dimensional area for each object sequentially, and then, based on embedded knowledge, determines the height of each object and the mesh density in the vertical direction.

3. The method of claim 1 wherein said extrusion source contains data which includes characteristics of the device such as thickness and probable flaws.

4. An intelligent thermal analyzer method for analyzing the thermal characteristics of an electronic device, such as a multichip module, said method comprising:

determining possible failure mechanisms of said electronics device;

developing a finite element model in accordance with the following steps;

inputting a graphic model of the electronics device into an object template of a generic blackboard, said model comprising four basic components of said device, including a package, substrate, interconnect and chip;

providing a plurality of knowledge sources, said knowledge sources including a symmetry knowledge source, model source and an extrusion source, each of said knowledge sources having an output to said blackboard;

activating the symmetry knowledge source to provide an input to said generic blackboard to generate new descriptions of said model representing a plurality of parts of the model to be analyzed, said parts each constituting a subset of said model;

providing an output from said model source to said blackboard, to generate a two dimensional geometric model of the subsets of said device;

applying an output from said extrusion source to form three dimensional geometric models of said subsets of said device;

then combining said subsets to create a finite element analysis code; and

displaying said finite element analysis code.

5. An intelligent thermal analyzer method for analyzing the thermal characteristics of an electronic device, such as a multichip module, said method comprising the steps of:

a. defining said multichip modules in terms of its geometry, material composition, heat generation/power consumption, ambient conditions and any boundary conditions that may be present (Input model source

     50);

b. creating abstract data objects to represent the various component of the multichip module, the material involved and areas of heat generation (from the computer input 44 and an object data base (42);

c. assigning to said data objects (in the input model knowledge source

     50) the appropriate values, as defined in step a;

d. adjusting the size and location of the components to reduce the complexity of the geometry (in the Knowledge Source

     52), and update the object data base (42);

e. inferring additional data, and annotating database 42 accordingly (in the Knowledge Source

     54);

f. further reducing the magnitude of the problem by identifying occurrences of symmetry (in Knowledge Source

     56), so that only a subsection of the model will need to be analyzed;

g. creating a set of 2D regions representing the physical space of the module (or its subsection) being analyzed (in the knowledge source 58 and the FASTQ plus

     68), and storing the results in a file (69) and by annotating the appropriate data back to the object database 42;

h. meshing the model by subdividing the regions defined by the existing regions into 2D finite elements, taking into account the need for numerical resolution and the need for numerical accuracy (using knowledge source 58 and the FASTQ

     68) and then storing the resulting 2D meshes in both a file (69), and by annotating appropriate data back to the object database 42;

i. extruding each of the 2D meshes into a single 3D mesh in the third dimension as defined by the input from a knowledge source 60 and a Gen3D (70) and storing the resulting 3D meshes in both a specific file (71) and by annotating the appropriate data back to the object database (42);

j. combining all of the 3D meshes into a single 3D mesh, (using knowledge source 64 and GEN3D

     72) and then storing the resulting 3D mesh in both a specific file (71) and annotating the appropriate data back to the object database (42);

k. determining the resulting temperatures and strains at node points through the multichip model, (using a finite element solver (62) and a FEECAP

     72), along with the accuracy of the values, and the storing the resulting values in both a specific file (73) and annotating appropriate data back to the object database (42);

l. evaluating the sufficiency of accuracy of these results in the knowledge source 64. and if insufficient, going to step 8, and repeating steps 8 to 12; and

m. based on known failure modes from Knowledge Source 66, and material behavior, assigning a value to the reliability of the proposed design.

6. In an intelligent thermal analyzer, having the following tools and other means to provide the related capability:

a blackboard framework/Object oriented Database (42);

a 2D Modeler/Mesh Generator (68);

a 3D Mesh Extruder (70);

a finite element Solver (72);

means for identifying the geometry and other characteristics of the module being analyzed (50);

means for determining the existing of symmetry in a 2D drawing (56);

means for combining several, separate 3D models into a single 3D model (62); and

means for inferring the thermal reliability of a multichip module from its temperature distribution (66);

a method for analyzing the thermal characteristics of an electronic device, such as a multichip module, said method comprising the steps of;

a. defining said multichip module in terms of its geometry, material composition, heat generation/power consumption, ambient conditions and any boundary conditions that may be present (Input model source

     50);

b. creating abstract data objects to represent the various component of the multichip module, the material involved and areas of heat generation (from the computer input 44 and an object data base (42);

c. assigning to said data objects (in the input model knowledge source

     50) the appropriate values, as defined in step 1;

d. adjusting the size and location of the components to reduce the complexity of the geometry (in the Knowledge Source

     52), and updating the object data base (42);

e. inferring additional data, and annotating database 42 accordingly (in the Knowledge Source

     54);

f. further reducing the magnitude of the problem by identifying occurrences of symmetry (in Knowledge Source

     56), so that only a subsection of the model will need to be analyzed;

g. creating a set of 2D regions representing the physical space of the module (or its subsection) being analyzed (in the knowledge source 58 and the FASTQ plus (68), and storing the results in a file (69) and by annotating the appropriate data back to the object database (42);

h. meshing the model by subdividing the regions defined by the existing regions into 2D finite elements, taking into account the need for numerical resolution and the need for numerical accuracy (using knowledge source 58 and the FASTQ

     68) and then storing the resulting 2D meshes in both a file (69), and by annotating appropriate data back to the object database 42;

i. extruding each of the 2D meshes into a single 3D mesh in the third dimension as defined by the input from a knowledge source 60 and a Gen3D (70) and storing the resulting 3D meshes in both a specific file (71) and by annotating the appropriate data back to the object database (42);

j. combining all of the 3D meshes into a single 3D mesh, (using knowledge source 64 and GEN3D

     72) and then storing the resulting 3D mesh in both a specific file (71) and annotating the appropriate data back to the object database (42);

k. determining the resulting temperatures and strains at node points through the multichip model, (using a finite element solver (62) and a FEECAP

     72), along with the accuracy of the values, and the storing the resulting values in both a specific file (73) and annotating appropriate data back to the object database (42);

l. evaluating the sufficiency of accuracy of these results in the knowledge source 64 and if insufficient, going to step 8, and repeating steps 8 to 12; and

m. based on known failure modes from Knowledge Source 66, and material behavior, assigning a value to the reliability of the proposed design.