Statistical analysis of coupled circuit-electromagnetic systems

• US 7,676,351 B2
• Filed: 07/16/2007
• Issued: 03/09/2010
• Est. Priority Date: 07/14/2006
• Status: Active Grant

• Current Assignee: University of Washington
• Original Assignee: University of Washington
• Inventors: Sathanur, Arun V., Chakraborty, Ritochit, Jandhyala, Vikram
• Primary Examiner(s): FREJD, RUSSELL WARREN
• Application Number:
  US11/778,568
• Publication Number:
  US 20080015829A1
• Time in Patent Office: 967 Days
• Field of Search: 703/2, 703/13, 703/14, 716 1- 4
• US Class Current: 703/2
• CPC Class Codes: G06F 2111/08 Probabilistic or stochastic CAD
• CPC Class Codes: G06F 30/367 Design verification, e.g. using simulation, simulation program with integrated circuit emphasis [SPICE], direct methods or relaxation methods

Abstract

A method and system for combining the process variations in circuits and distributed interconnect-based electromagnetic (EM) objects in order to capture a statistical behavior of overall circuit performance parameters. In an exemplary approach, a coupled circuit-EM system is decoupled at the points where the EM objects connect to the circuit portion, and circuit ports are defined at those points. The sources of variation are identified and used to determine Y-parameters for the ports with EM elements and for all EM elements based on the SPICE-like and EM full-wave simulations. A response surface is generated for each variable and is used to extract circuit and EM parameters by generating many random vectors representing combinations of the random variables. These Y-parameters are merged to produce a probability density function (PDF) of one or more performance metrics for the electronic device or system.

Claims

1. A method for determining a probability density function (PDF) of at least one performance metric in regard to an electromagnetic (EM) component of an electronic device or system, comprising the steps of:

(a) providing variability information as an input to a simulation of the EM component, for each randomly varying parameter to be considered in determining the PDF for the at least one performance metric, wherein the variability information for each randomly varying parameter comprises a range and a nominal value for the randomly varying parameter, and wherein the simulation is executed by a processor;

(b) for all of the randomly varying parameters to be considered, using the simulation executed by the processor to generate a response surface, wherein the response surface is defined over a statistical range for the randomly varying parameter, and as a function of a related variable on which the performance metric is dependent;

(c) using a random vector generator executed by the processor to produce probabilities that the randomly varying parameter will be within each of a plurality of different incremental portions of the range provided in step (a), wherein the random vector generator employs a correlation matrix that correlates values of the randomly varying parameter with the related variable;

(d) extracting the related variable from the response surface, based upon the probabilities of the randomly varying parameter; and

(e) determining the PDF for each performance metric evaluated in regard to the EM component, as a function of the probabilities of each randomly varying parameter, relative to the related variable.

2. The method of claim 1, wherein the step of using the simulation to generate the response surface comprises the step of using a field solver to compute the related variable at terminals of each EM component.

3. The method of claim 2, wherein the field solver carries out a plurality of steps, including:

(a) decomposing each EM component into an equivalent set of interior and exterior regions, each interior region having a bounding surface;

(b) computing scattering due to each EM component using equivalent electric and magnetic surface currents; and

(c) solving for the electric and magnetic surface currents using the Method of Moments, enabling the response surface to be produced in regard to the related variable.

4. The method of claim 1, wherein the at least one performance metric further depends on at least one randomly varying process parameter of a circuit comprising the electronic device or system, further comprising the steps of:

(a) providing variability information for each randomly varying process parameter of the circuit, a list of the randomly varying process parameters for the circuit, and a netlist that describes a connectivity of the circuit, as an input to a general purpose circuit simulation program, wherein the variability information for each randomly varying process parameter comprises a range and a nominal value for the randomly varying process parameter;

(b) for each of the randomly varying process parameters of the circuit to be considered, using the general purpose simulation program to generate a response surface, wherein the response surface is defined over a statistical range for the randomly varying process parameter, and as a function of a related circuit variable on which the performance metric is dependent;

(c) using the random vector generator to produce probabilities that each randomly varying process parameter will be within each of a plurality of different incremental portions of the range provided for the randomly varying process parameter, wherein the random vector generator employs a correlation matrix that correlates values of the randomly varying process parameter with the related circuit variable for ports of an EM component;

(d) extracting the related circuit variable from the response surface, based upon the probabilities of the randomly varying parameter;

(e) merging each related circuit variable and each related variable for the EM component; and

(f) determining the PDF of the at least one performance metric for the electronic device or system as a function of each related circuit variable and each related variable that have been merged.

5. The method of claim 4, wherein the general purpose circuit simulation program comprises a Simulation Program for Integrated Circuits Emphasis (SPICE) software program.

6. The method of claim 4, further comprising the step of transforming a large number of possibly correlated related circuit variables into a smaller number of uncorrelated related circuit variables, for the circuit.

7. The method of claim 1, further comprising the step of transforming a large number of possibly correlated related variables into a smaller number of uncorrelated related variables for one or more EM components.

8. The method of claim 4, further comprising the step of determining each point where the EM component connects to the circuit, each such point comprising a port for the EM component.

9. The method of claim 4, wherein the at least one performance metric is determined based upon at least one randomly varying parameter that is non-Gaussian and numerical.

10. The method of claim 4, wherein the step of determining the PDF for each performance metric uses randomly varying parameters and randomly varying process parameters that are Gaussian and either correlated or uncorrelated and is carried out without an intermediate step of determining non-Gaussian PDFs for any randomly varying process parameter or randomly varying parameter.

11. A system for determining a probability density function (PDF) of at least one performance metric in regard to an electromagnetic (EM) component of an electronic device or system, comprising:

(a) a memory for storing machine executable instructions, and variability information for each randomly varying parameter to be considered in determining the PDF for the at least one performance metric, the variability information for each randomly varying parameter comprising a range and a nominal value for the randomly varying parameter;

(b) an output device for presenting the PDF of the at least one performance metric to a user; and

(c) a processor that is coupled to the memory and the output device, the processor executing the machine executable instructions to carry out a plurality of functions, including;

(i) accessing the variability information as an input to a simulation of the EM component;

(ii) for all of the randomly varying parameters to be considered, using the simulation to generate a response surface, wherein the response surface is defined over a statistical range for the randomly varying parameter, and as a function of a related variable on which the performance metric is dependent;

(iii) executing a random vector generator to produce probabilities that the randomly varying parameter will be within each of a plurality of different incremental portions of the range, wherein the random vector generator employs a correlation matrix that correlates values of the randomly varying parameter with the related variable;

(iv) extracting the related variable from the response surface, based upon the probabilities of the randomly varying parameter; and

(v) determining the PDF for each performance metric evaluated in regard to the EM component, as a function of the probabilities of each randomly varying parameter, relative to the related variable.

12. The system of claim 11, further comprising the step of enabling a user to access the PDF for any performance metric that was evaluated, by causing the processor to present the PDF on the output device.

13. The system of claim 11, wherein execution of the machine executable instructions further causes the processor to use a field solver to compute the related variable at terminals of each EM component to generate the response surface.

14. The system of claim 13, wherein execution of the machine executable instructions further causes the processor to:

(a) decompose each EM component into an equivalent set of interior and exterior regions, each interior region having a bounding surface;

(b) compute scattering due to each EM component using equivalent electric and magnetic surface currents; and

(c) solve for the electric and magnetic surface currents using the Method of Moments, enabling the response surface to be produced in regard to the related variable.

15. The system of claim 11, wherein the at least one performance metric further depends on at least one randomly varying process parameter of a circuit comprising the electronic device or system, execution of the machine executable instructions further causing the processor to:

(a) provide input to a general purpose circuit simulation program by accessing variability information for each randomly varying process parameter of the circuit, a list of the randomly varying process parameters for the circuit, and a netlist that describes a connectivity of the circuit that are also stored in the memory, wherein the variability information for each randomly varying process parameter comprises a range and a nominal value for the randomly varying process parameter;

(b) for each of the randomly varying process parameters of the circuit to be considered, using the general purpose simulation program to generate a response surface, wherein the response surface is defined over a statistical range for the randomly varying process parameter, and as a function of a related circuit variable on which the performance metric is dependent;

(c) use the random vector generator to produce probabilities that each randomly varying process parameter will be within each of a plurality of different incremental portions of the range provided for the randomly varying process parameter, wherein the random vector generator employs a correlation matrix that correlates values of the randomly varying process parameter with the related circuit variable for ports of an EM component;

(d) extract the related circuit variable from the response surface, based upon the probabilities of the randomly varying parameter;

(e) merge each related circuit variable and each related variable for the EM component; and

(f) determine the PDF of the at least one performance metric for the electronic device or system as a function of each related circuit variable and each related variable that have been merged.

16. The system of claim 15, wherein the general purpose circuit simulation program comprises a Simulation Program for Integrated Circuits Emphasis (SPICE) software program.

17. The system of claim 15, wherein execution of the machine executable instructions further causes the processor to transform a large number of possibly correlated related circuit variables into a smaller number of uncorrelated related circuit variables, for the circuit.

18. The system of claim 11, wherein execution of the machine executable instructions further causes the processor to transform a large number of possibly correlated related variables into a smaller number of uncorrelated related variables for one or more EM components.

19. The system of claim 15, wherein execution of the machine executable instructions further causes the processor to determine each point where the EM component connects to the circuit, each such point comprising a port for the EM component.

20. The system of claim 15, wherein execution of the machine executable instructions further causes the processor to determine the at least one performance metric based upon at least one randomly varying parameter that is non-Gaussian and numerical.

21. The system of claim 15, wherein execution of the machine executable instructions further causes the processor to determine the PDF for each performance metric using randomly varying parameters and randomly varying process parameters that are Gaussian and either correlated or uncorrelated, and without performing an intermediate determination of non-Gaussian PDFs for any randomly varying process parameter or randomly varying parameter.