dc analysis of nonlinear circuits - peopleee219a fall 1998 3.1.11 rate of convergence ee219a fall...

15
Prof. A. Richard Newton University of California at Berkeley Page 1 Copyright © 1997, A. Richard Newton EE219A Fall 1998 3.1.1 EE219A: Computer Analysis of Electrical Circuits Outline Lecture 3.1 u DC Solution of Nonlinear Equations EE219A Fall 1998 3.1.2 DC Analysis of Nonlinear Circuits DC Analysis of Nonlinear Circuits

Upload: others

Post on 26-Mar-2021

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 1Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.1

EE219A: Computer Analysis of Electrical CircuitsOutline

Lecture 3.1

u DC Solution of Nonlinear Equations

EE219A Fall 1998 3.1.2

DC Analysis of Nonlinear CircuitsDC Analysis of Nonlinear Circuits

Page 2: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 2Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.3

DC Analysis of Nonlinear CircuitsDC Analysis of Nonlinear Circuits

EE219A Fall 1998 3.1.4

Contraction Mapping TheoremsContraction Mapping Theorems

Page 3: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 3Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.5

Contraction Mapping TheoremsContraction Mapping Theorems

EE219A Fall 1998 3.1.6

Newton’s MethodNewton’s Method

Page 4: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 4Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.7

Newton’s MethodNewton’s Method

EE219A Fall 1998 3.1.8

Newton-Raphson MethodNewton-Raphson Method

Page 5: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 5Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.9

Rate of ConvergenceRate of Convergence

EE219A Fall 1998 3.1.10

Rate of ConvergenceRate of Convergence

Page 6: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 6Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.11

Rate of ConvergenceRate of Convergence

EE219A Fall 1998 3.1.12

Techniques for Reducing the Cost ofNewton-Raphson

Techniques for Reducing the Cost ofNewton-Raphson

u 1. Computation of J(x) by built-in derivatives(SPICE, SPLICE, RELAX)

u 2. Update J(x) every p iterations (SPLICE, RELAX)u 3. Selective Update: If value of a state variable does

not change, do not update f(x) or J(x)s Bypass (SPICE) (for digital circuits, savings up to 50%

here)s Event-driven selective-trace at the state-variable and

device level (SPLICE)

u 1. Computation of J(x) by built-in derivatives(SPICE, SPLICE, RELAX)

u 2. Update J(x) every p iterations (SPLICE, RELAX)u 3. Selective Update: If value of a state variable does

not change, do not update f(x) or J(x)s Bypass (SPICE) (for digital circuits, savings up to 50%

here)s Event-driven selective-trace at the state-variable and

device level (SPLICE)

Page 7: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 7Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.13

Potential ProblemsPotential Problems

EE219A Fall 1998 3.1.14

ImplicationsImplicationsu Device model equations must be continuous with continuous

derivatives and derivative calculation must be accurate derivativeof function (not all models do this - Poor diode models andbreakdown models don’t - be sure models are decent - beware ofuser-supplied models)

u Watch out for floating nodes (If a node becomes disconnected,then J(x) is singular)

u Give good initial guess for x(0)

u Most model computations produce errors in function values andderivatives. Want to have convergence criteria || x(k+1) - x(k) || < εsuch that ε > than model errors.

u Device model equations must be continuous with continuousderivatives and derivative calculation must be accurate derivativeof function (not all models do this - Poor diode models andbreakdown models don’t - be sure models are decent - beware ofuser-supplied models)

u Watch out for floating nodes (If a node becomes disconnected,then J(x) is singular)

u Give good initial guess for x(0)

u Most model computations produce errors in function values andderivatives. Want to have convergence criteria || x(k+1) - x(k) || < εsuch that ε > than model errors.

Page 8: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 8Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.15

Computational AspectsComputational Aspects

EE219A Fall 1998 3.1.16

Computational AspectsComputational Aspects

Page 9: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 9Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.17

Modeling a DiodeModeling a Diode

EE219A Fall 1998 3.1.18

Modeling a DiodeModeling a Diode

Page 10: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 10Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.19

MNA TemplatesMNA Templates

EE219A Fall 1998 3.1.20

MNA TemplatesMNA Templates

Page 11: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 11Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.21

Modeling a MOSFET(MOS Level 1, linear regime)Modeling a MOSFET

(MOS Level 1, linear regime)

d

EE219A Fall 1998 3.1.22

Modeling a MOSFET(MOS Level 1, linear regime)Modeling a MOSFET

(MOS Level 1, linear regime)

Page 12: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 12Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.23

DC Analysis Flow DiagramDC Analysis Flow DiagramFor each state variable in the system

EE219A Fall 1998 3.1.24

Numerical OverflowNumerical Overflow

Page 13: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 13Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.25

NonconvergenceNonconvergence

EE219A Fall 1998 3.1.26

Limiting AlgorithmLimiting Algorithm

Page 14: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 14Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.27

Improving ConvergenceImproving Convergence

EE219A Fall 1998 3.1.28

Optimization-Based ApproachesOptimization-Based Approaches

Page 15: DC Analysis of Nonlinear Circuits - PeopleEE219A Fall 1998 3.1.11 Rate of Convergence EE219A Fall 1998 3.1.12 Techniques for Reducing the Cost of Newton-Newton-Raphson u1. Computation

Prof. A. Richard NewtonUniversity of California at Berkeley

Page 15Copyright © 1997, A. Richard Newton

EE219A Fall 1998 3.1.29

Continuation MethodsContinuation Methods

EE219A Fall 1998 3.1.30

Continuation MethodsContinuation Methods