Power Cell Synthesizer Optimizing On-Resistance and Layout Area

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current power cell design methodologies for ICs face challenges in achieving low on-resistance (RDSon) values efficiently, often leading to overdesign, increased costs, and longer design times due to the complexity of optimizing both intrinsic transistor and metal interconnect resistances, which are typically addressed through iterative and resource-intensive processes.

Innovation Solution

The development of a compact analytical method and apparatus that employs a G-function processor to optimize geometric layout parameters, predicting and minimizing RDSon by accurately modeling the effects of parasitic metal resistance and its variability with device layout, allowing for a closed-form solution to find the global minimum in a continuous solution space.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If the gate width is increased to decrease intrinsic transistor resistance, then the on-resistance decreases, but the device area increases

Engineering Contradiction:
Improveon-resistanceVSAvoiddevice area
Core Design Contradiction:
Manufacturing precisionVSArea of stationary object

Solution Approach 1:

The power cell is divided into multiple unit cells, each with its own intrinsic transistor and metal interconnect structure. This segmentation allows the total on-resistance to be distributed across parallel paths, reducing the overall resistance without proportionally increasing the area of each individual unit cell. The segmented approach enables independent optimization of each unit while achieving system-level performance targets.

Inventive Principle:
Principle #1Segmentation

2Manufacturing precision

If the cell area is increased to decrease intrinsic transistor resistance, then the on-resistance decreases, but the metal interconnect resistance increases

Engineering Contradiction:
Improveon-resistanceVSAvoidmetal interconnect resistance
Core Design Contradiction:
Manufacturing precisionVSObject-generated harmful factors

Solution Approach 1:

The metal interconnect structure is optimized with non-uniform geometry, where the interconnect width varies along its length to compensate for resistance effects. The interconnect is wider at regions where current density is higher and narrower where current density is lower. This local quality variation allows the metal interconnect resistance to be minimized while maintaining a compact cell area, directly addressing the contradiction between area reduction and resistance management.

Inventive Principle:
Principle #3Local quality

3Manufacturing precision

If traditional iterative design methods are used to optimize power cell parameters, then design accuracy can be improved, but design time and computational resources increase

Engineering Contradiction:
Improvedesign accuracyVSAvoiddesign time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

Closed-form analytical expressions are derived beforehand that directly calculate the optimal power cell parameters (gate width, interconnect dimensions, unit cell count) based on target on-resistance specifications. These pre-derived formulas eliminate the need for iterative simulation and adjustment during the design process. The designer simply inputs the target RDSon value and device voltage rating, and the analytical model immediately provides the optimized geometric parameters, dramatically reducing design time while maintaining high accuracy.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS8762921B2Apparatus and method for optimized power cell synthesizer
Publication Date: 2014.06.24 GLOBALFOUNDRIES US INC
  • US8762921B2 patent drawing
  • US8762921B2 patent drawing
  • US8762921B2 patent drawing

AI summary

Apparatus for providing semiconductor device with an analysis module to receive device information, a G-function processor producing an ordered relationship representation corresponding to an optimization parameter specification, and a power cell optimizer to produce an optimization parameter from the ordered relationship representation. A method for designing a semiconductor device includes receiving an optimization target specification; receiving an optimization parameter specification corresponding to an optimization parameter; receiving the target parameter; receiving a G-function corresponding to an ordered relationship representation; optimizing the optimization parameter specification as a function of the predetermined G-function; and producing at least one optimized geometric layout parameter (GLP) by the optimizing, wherein the at least one GLP corresponds to an optimized power cell.