International Business Machines Corporation

Howard L Heck

John Steven Kresge

David Noel Light

Charles Robert Davis

Ajit Kumar Trivedi

Thomas P Duffy

Steven L Hanakovic

John T Kolias

Thomas P Gall

Disclosed is a parallel processor packaging structure and a method for manufacturing the structure. The individual logic and memory elements are on printed circuit cards. These printed circuit boards and cards are, in turn, mounted on or connected to circuitized flexible substrates extending outwardly from a laminate of the circuitized, flexible substrates. Intercommunication is provided through a switch structure that is implemented in the laminate. The printed circuit cards are mounted on or connected to a plurality of circuitized flexible substrates, with one printed circuit card at each end of the circuitized flexible circuit. The circuitized flexible substrates connect the separate printed circuit boards and cards through the central laminate portion. This laminate portion provides XY plane and Z-axis interconnection for inter-processor, inter-memory, inter-processor/memory element, and processor to memory bussing interconnection, and communication. The planar circuitization, as data lines, address lines, and control lines of a logic chip or a memory chip are on the individual printed circuit boards and cards, which are connected through the circuitized flex, and communicate with other layers of flex through Z-axis circuitization (vias and through holes) in the laminate. Lamination of the individual subassemblies is accomplished with a low melting adhesive that is chemical compatible with (bondable to) the per fluorocarbon polymer between the subassemblies in the regions intended to be laminated, and, optionally, a high melting mask that is chemically incompatible with (not bondable to) the per fluorocarbon polymer between the subassemblies in the regions not intended to be laminated. The subassembly stack is heated to selectively effect adhesion and lamination in areas thereof intended to be laminated while avoiding lamination in areas not intended to be laminated.

Method of fabricating a flex laminate package

Disclosed is a parallel processor packaging structure and a method for manufacturing the structure. The individual logic and memory elements are on printed circuit cards. These printed circuit boards and cards are, in turn, mounted on or connected to circuitized flexible substrates extending outwardly from a laminate of the circuitized, flexible substrates. Intercommunication is provided through a switch structure that is implemented in the laminate. The circuitized flexible substrates connect the separate printed circuit boards and cards through the central laminate portion. This laminate portion provides XY plane and Z-axis interconnection for inter-processor, inter-memory, inter-processor/memory element, and processor to memory bussing interconnection, and communication. Lamination of the individual subassemblies is accomplished with a low melting adhesive that is chemical compatible with (bondable to) the perfluorocarbon polymer between the subassemblies in the regions intended to be laminated, and, optionally, a high melting mask that is chemically incompatible with (not bondable to) the perfluorocarbon polymer between the subassemblies in the regions not intended to be laminated. The subassembly stack is heated to selectively effect adhesion and lamination in areas thereof intended to be laminated while avoiding lamination in areas not intended to be laminated.

Disclosed is a parallel processor packaging structure and a method for manufacturing the structure. The individual logic and memory elements are on printed circuit cards. These printed circuit boards and cards are, in turn, mounted on or connected to circuitized flexible substrates extending outwardly from a laminate of the circuitized, flexible substrates. Intercommunication is provided through a switch structure that is implemented in the laminate. The printed circuit cards are mounted on or connected to a plurality of circuitized flexible substrates, with one printed circuit card at each end of the circuitized flexible circuit. The circuitized flexible substrates connect the separate printed circuit boards and cards through the central laminate portion. This laminate portion provides XY plane and Z-axis interconnection for inter-processor, inter-memory, inter-processor/memory element, and processor to memory bussing interconnection, and communication. The planar circuitization, as data lines, address lines, and control lines of a logic chip or a memory chip are on the individual printed circuit boards and cards, which are connected through the circuitized flex, and communicate with other layers of flex through Z-axis circuitization (vias and through holes) in the laminate. Lamination of the individual subassemblies is accomplished with a low melting adhesive that is chemical compatible with (bondable to) the perfluorocarbon polymer between the subassemblies in the regions intended to be laminated, and, optionally, a high melting mask that is chemically incompatible with (not bondable to) the perfluorocarbon polymer between the subassemblies in the regions not intended to be laminated. The subassembly stack is heated to selectively effect adhesion and lamination in areas thereof intended to be laminated while avoiding lamination in areas not intended to be laminated.

Flex laminate package for a parallel processor

Disclosed is a parallel processor packaging structure and a method for manufacturing the structure. The individual logic and memory elements are on printed circuit cards. These printed circuit boards and cards are, in turn, mounted on or connected to circuitized flexible substrates extending outwardly from a laminate of the circuitized, flexible substrates. Intercommunication is provided through a switch structure that is implemented in the laminate. The printed circuit cards are mounted on or connected to a plurality of circuitized flexible substrates, with one printed circuit card at each end of the circuitized flexible circuit. The circuitized flexible substrates connect the separate printed circuit boards and cards through the central laminate portion. This laminate portion provides XY plane and Z-axis interconnection for inter-processor, inter-memory, inter-processor/memory element, and processor to memory bussing interconnection, and communication. The planar circuitization, as data lines, address lines, and control lines of a logic chip or a memory chip are on the individual printed circuit boards and cards, which are connected through the circuitized flex, and communicate with other layers of flex through Z-axis circuitization (vias and through holes) in the laminate.

Parallel processor and method of fabrication

A two-stage Monte Carlo method of tolerancing components of an assembly is provided. Statistical measures of component features are not time invariant, but change over a production run. That is, the mean value of component feature measures and the standard distribution of the component feature measures about the time dependent mean are not invariant over a production run, but shift with time and throughput. According to the invention, these 'shifted' or 'adjusted' parameters are utilized in a Monte Carlo simulation to determine discrete values for the individual points of each output distribution, x(i), y(i), z(i). The individual points of the output distributions, x(i), y(i), z(i), are combined in a second Monte Carlo simulation step for individual assembly final fit F(x(i), y(i), z(i)). The statistics of the individual assembly final fits are then compared to manufacturing specifications.

Growing community of inventors

Endcott, NY, United States of America

Howard L Heck