Virtuoso® Advanced Analysis Tools User

Guide
Product Version 5.1.41
June 2004
 1999-2004 Cadence Design Systems, Inc. All rights reserved.
Printed in the United States of America.
Cadence Design Systems, Inc., 555 River Oaks Parkway, San Jose, CA 95134, USA
Trademarks: Trademarks and service marks of Cadence Design Systems, Inc. (Cadence) contained in
this document are attributed to Cadence with the appropriate symbol. For queries regarding Cadence’s
trademarks, contact the corporate legal department at the address shown above or call 1-800-862-4522.
All other trademarks are the property of their respective holders.
Restricted Print Permission: This publication is protected by copyright and any unauthorized use of this
publication may violate copyright, trademark, and other laws. Except as specified in this permission
statement, this publication may not be copied, reproduced, modified, published, uploaded, posted,
transmitted, or distributed in any way, without prior written permission from Cadence. This statement grants
you permission to print one (1) hard copy of this publication subject to the following conditions:
1. The publication may be used solely for personal, informational, and noncommercial purposes;
2. The publication may not be modified in any way;
3. Any copy of the publication or portion thereof must include all original copyright, trademark, and other
proprietary notices and this permission statement; and
4. Cadence reserves the right to revoke this authorization at any time, and any such use shall be
discontinued immediately upon written notice from Cadence.
Disclaimer: Information in this publication is subject to change without notice and does not represent a
commitment on the part of Cadence. The information contained herein is the proprietary and confidential
information of Cadence or its licensors, and is supplied subject to, and may be used only by Cadence’s
customer in accordance with, a written agreement between Cadence and its customer. Except as may be
explicitly set forth in such agreement, Cadence does not make, and expressly disclaims, any
representations or warranties as to the completeness, accuracy or usefulness of the information contained
in this document. Cadence does not warrant that use of such information will not infringe any third party
rights, nor does Cadence assume any liability for damages or costs of any kind that may result from use of
such information.
Restricted Rights: Use, duplication, or disclosure by the Government is subject to restrictions as set forth
in FAR52.227-14 and DFAR252.227-7013 et seq. or its successor.
 Virtuoso Advanced Analysis Tools User Guide

Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Typographic and Syntax Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1
Corners Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Getting Started with Corners Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
How Corners Analysis Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Opening and Closing the Virtuoso® Analog Corners Analysis Window . . . . . . . . . . . 10
Getting to Know the Virtuoso® Analog Corners Analysis Window . . . . . . . . . . . . . . . . . 12
Menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Process and Base Directory Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Corner Definitions Pane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Performance Measurements Pane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Split Pane Adjustment Bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Status Display . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Keyboard Navigation and Shortcuts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Running a Corners Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Defining the Process and Corners for an Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Defining Performance Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Controlling the Corners Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Evaluating Corners Analysis Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Text Outputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Graphic Outputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Saving Setup Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Saving Setup Information to the Original Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Saving Setup Information to a Specified File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Saving a Script . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Using Process, Design, and Modeling Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Creating Process and Design Customization Files . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Using a .cdsinit File to Load PCFs and DCFs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

June 2004 3 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

Implementing Modeling Styles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Using the Virtuoso® Analog Corners Analysis Window to Define and Update Processes
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Requirements for Using the Spectre Simulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Working through an Extended Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Folded Cascode Schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Setting Up the Virtuoso® Analog Design Environment Window . . . . . . . . . . . . . . . . 62
Modeling Style . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Process Customization File (PCF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Virtuoso® Analog Corners Analysis Window for Folded Cascode . . . . . . . . . . . . . . . 68
Changing Values in the Virtuoso® Analog Corners Analysis Window . . . . . . . . . . . . 70
Running the Corners Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Evaluating Corners Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

2
Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Getting Started with Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
How Statistical Analysis Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Data Types Generated by the Statistical Analysis Tool . . . . . . . . . . . . . . . . . . . . . . . 76
Opening the Analog Statistical Analysis Window . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Getting to Know the Analog Statistical Analysis Window . . . . . . . . . . . . . . . . . . . . . . . . . 78
Status Display . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Analysis Setup Pane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Outputs Pane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Edit Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Button Bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Running a Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Specifying the Characteristics of a Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . 86
Selecting Signals and Expressions to Analyze . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Defining Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Starting and Stopping the Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Saving Statistical Analysis Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Saving and Restoring a Statistical Analysis Session . . . . . . . . . . . . . . . . . . . . . . . . 101
How the Statistical Analysis Option Uses the Analysis Variation Setting . . . . . . . . . 104

June 2004 4 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

Analyzing Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

Loading Stored Statistical Analysis Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Creating a New mcdata File from Saved Waveform Data . . . . . . . . . . . . . . . . . . . . 108
Filtering Outlying Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Setting Specification Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Generating Plots, Tables, and Reports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Working through an Extended Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Lowpass Filter Schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Model File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Run Analog Simulation to Check Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Specifying the Analysis in the Analog Statistical Analysis Window . . . . . . . . . . . . . 133
Running the Statistical Analysis Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Evaluating Statistical Analysis Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Changing Waveform Expressions at Post-simulation Time . . . . . . . . . . . . . . . . . . . 146
Changing Scalar Expressions at Post-Simulation Time . . . . . . . . . . . . . . . . . . . . . . 147
Appending More Scalar Iterations to Existing Data . . . . . . . . . . . . . . . . . . . . . . . . . 152
Appending Waveforms From Different Statistical Analysis Runs. . . . . . . . . . . . . . . . 154
Performing a Swept Parameter Statistical Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . 155

3
Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Getting Started with Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
How Optimization Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Getting Help . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Opening and Closing the Virtuoso® Analog Circuit Optimization Option Window . . 162
Getting to Know the Virtuoso® Analog Circuit Optimization Option Window . . . . . . . . . 163
Status Display . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
Goals Pane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Variables Pane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Tool Bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Running an Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Defining Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Preparing Design Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Controlling the Optimizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187

June 2004 5 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

Plotting Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

Saving, Changing, and Loading Session Information . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Saving the Session State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Loading a Saved Session State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Saving a Script . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Changing Optimization Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Deleting All Setup Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Working through an Extended Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
Generating the Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
Saving the Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Setting Up and Running the Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201

Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215

June 2004 6 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

Preface

This manual describes how to use the Virtuoso® advanced analysis tools:
■ The Virtuoso® analog statistical analysis option
■ The Virtuoso® analog corners analysis option
■ The Virtuoso® analog optimization analysis option

The guidance here is designed for users who are already familiar with circuit design and
simulation.

Related Documents
The Virtuoso® advanced analysis tools are often used within the Virtuoso® analog design
environment. The following documents give further information.
■ All the analysis tools open from the Virtuoso® Analog Design Environment window. For
information about using that window, see the Virtuoso® Analog Design Environment
User Guide.
■ For information about using the advanced analysis tools in the Open Command
Environment for Analysis (OCEAN) environment, see the OCEAN Reference.
■ For information about Cadence SKILL language procedural interface commands for the
Corners customization files, see the Virtuoso® Analog Design Environment SKILL
Language Reference.

Typographic and Syntax Conventions

The syntax conventions used in this documentation are described below.

literal Words in nonitalic monospaced type indicate text you must type
exactly as it is presented. These words represent command
(function or routine) or option names or system output.

June 2004 7 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Preface

argument... Words in italic monospaced type indicate text that you must
replace with an appropriate argument or other data, such as a
path. The three dots indicate that you can repeat the argument.
Substitute one or more names or values.

italic Words in italics Indicate names of manuals, commands, and

form buttons, form fields, and other features of the user interface
(UI).

June 2004 8 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

1
Corners Analysis

Corners analysis provides a convenient way to measure circuit performance while simulating
a circuit with sets of parameter values that represent the most extreme variations in a
manufacturing process.

With the Virtuoso ® Analog Corners Analysis option, you can compare the results for each
set of parameter values with the range of acceptable performance values. You can ensure the
largest possible yield of circuits at the end of the manufacturing process by also revising the
circuit, so that all the sets of parameters produce acceptable results.

This chapter explains in detail how you can use the corners analysis option to generate
information about the yields from your circuit designs.
■ “Getting Started with Corners Analysis” on page 9
■ “Getting to Know the Virtuoso® Analog Corners Analysis Window” on page 12
■ “Running a Corners Analysis” on page 20
■ “Evaluating Corners Analysis Results” on page 29
■ “Using Process, Design, and Modeling Files” on page 36
■ “Working through an Extended Example” on page 60

Getting Started with Corners Analysis

This section briefly explains the theory behind corners analysis, tells you how to get help and
describes how to open the Virtuoso ® Analog Corners Analysis window.

How Corners Analysis Works

In a theoretical manufacturing process, process variables can have exact values and these
exact values can be used to calculate the yield for the process. However, in a real
manufacturing process, process variables are subject to a manufacturing tolerance—they

June 2004 9 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

fluctuate randomly around their ideal values. The combined random variation for all the
components results in an uncertain yield for the circuit as a whole.

Corners analysis looks at the performance outcomes generated from the most extreme
variations expected in the process, voltage and temperature values (the corners).

With this information, you can determine whether the circuit performance specifications will
be met, even when the random process variations combine in their most unfavorable
patterns.

You can use Corners in the Virtuoso® Analog Design Environment in one of two ways:
1. By using the Corners user interface
2. By using your own process and design customization files

You will see more about these in subsequent sections of this chapter.

Opening and Closing the Virtuoso® Analog Corners Analysis Window

To prepare for a corners analysis,
1. Ensure that the design you use is simulatable with nominal design parameter values.
2. Set up a simulation in the Virtuoso ® Analog Design Environment window to run the
analysis you want to use.
3. Ensure that all design variables in the circuit have an initial value.

June 2004 10 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

4. In the Virtuoso ® Analog Design Environment window, choose Tools – Corners.

If you have defined a set of customization files to be loaded automatically, the Virtuoso ®
Analog Corners Analysis window appears.

To close the Virtuoso ® Analog Corners Analysis window,

➤ Choose File – Close.

June 2004 11 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Getting to Know the Virtuoso® Analog Corners Analysis

Window
The Virtuoso ® Analog Corners Analysis window contains the fields and controls required
to specify the corners and measurements for the analysis you want to run.

Menu

Process field

Base Directory field

Corner Definitions pane

Split Pane Adjustment

Bar

Performance
Measurements pane

Status display

June 2004 12 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Menu
The menu contains the commands needed to prepare for, run and plot the results of a corners
analysis.

For guidance on using the menu selections, see the associated cross-references.

Menu Item For More Information

File
Load “Using the Graphical User Interface to Load PCFs
and DCFs” on page 20
Save Setup “Saving Setup Information to the Original Files” on
page 33
Save Setup As “Saving Setup Information to a Specified File” on
page 33
Save Script “Saving a Script” on page 35
Close “Opening and Closing the Virtuoso® Analog Corners
Analysis Window” on page 10
Edit
Corner Definitions-> “Creating a New Corner” on page 23
Add Corner
Corner Definitions-> “Copying and Modifying an Existing Corner” on
Copy Corner page 23
Corner Definitions-> “Enable Corner” on page 24
Enable Corner
Corner Definitions-> “Disable Corner” on page 24
Disable Corner
Corner Definitions-> “Adding a Row for a New Design Variable” on
Add Variable page 24
Corner Definitions-> “Deleting Corners or Rows” on page 25
Delete Selected

June 2004 13 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Performance “Creating a New Performance Measurement by

Measurements-> Entering It Directly” on page 26, and “Creating a New
Add Measurement Performance Measurement by Using the Calculator”
on page 26

Performance “Deleting a Performance Measurement” on page 27

Measurements->
Delete Measurement
Setup
Add Process “Using the Virtuoso® Analog Corners Analysis
Window to Define a Process” on page 57
Add/Update Model Info “Using the Virtuoso® Analog Corners Analysis
Window to Modify Process Model Information” on
page 58
Simulation
Run “Running and Stopping the Analysis” on page 28
Stop “Running and Stopping the Analysis” on page 28
Tools
Calculator “Creating a New Performance Measurement by
Using the Calculator” on page 26
Get Expression “Creating a New Performance Measurement by
Using the Calculator” on page 26
Plot or Print Outputs “Evaluating Corners Analysis Results” on page 29
Help
Contents Displays the documentation (this user guide)
containing information about the Virtuoso® Analog
Corners Analysis option.

Process and Base Directory Fields

Process refers to the manufacturing process. Therefore, process parameters are parameters
that pertain to the manufacturing process and are variables that help characterize the models
specific to the manufacturing process.

The Base Directory field displays the path that contains the models used in the analysis for
the process being displayed in the process field.

June 2004 14 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The base directory is usually defined by the corAddProcess command in a process

customization file (PCF). You can also define the base directory by choosing Setup – Add
Process or Setup – Add/Update Model Info.

The Process field displays either the name of the current process or None, if no process is
specified. The processes are usually defined in customization files but can also be defined
from the graphical user interface.

If there is no current process, the only active Virtuoso ® Analog Corners Analysis window
menu options are File – Load, File – Close, and Setup – Add Process. These menu
options allow you to either load an existing file that defines a process or to define a new
process.

Corner Definitions Pane

The Corner Definitions pane, located in the upper section of the Virtuoso ® Analog
Corners Analysis window, displays information about the currently defined corners.

The information in this pane is usually loaded from process customization files (PCFs) and
design customization files (DCFs) using paths defined in your .cdsinit file. For details
refer to the section Using a .cdsinit File to Load PCFs and DCFs.
To define or revise corners, you modify the information in this pane.
■ Each column characterizes a corner. You can select a column by clicking on the
corresponding button along the top of the pane.

June 2004 15 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

■ You can also physically move the corner columns by dragging them around in the column
header.
■ You can alter the width of columns by grabbing the separation bar and dragging it one
way or the other. There are certain limits (upper and lower) to how big or small you can
make a column. This also differs by column type in the case of the measurement table.
■ Disabled corners are grayed or fuzzed out in the form, while enabled corners are
displayed in normal text.
■ Each row (or variable) begins with a group name or design variable name, followed by
the values to be used in each of the corners.
■ You can select a variable by clicking on the corresponding button along the left side of
the pane. You can drag a selection of variables by clicking and moving the cursor in the
variable header.
■ Editable data in the Corners table looks like normal text, while uneditable data in the
Corners table has a dark gray background. Data loaded from a PCF is non-editable.
■ Uneditable group/variant entries appear as a text field with the value of the field in text.
Editable group/variant entries appear as a drop-down box.
Note: Temperature is a default variable with default value of 27. From release 5.1.41, this is
an editable field.
The items in the Corner Definitions pane are described in the following table:

Item Description and Usage

Add Corner Click to add a new editable column to the right of the
existing columns.
Add Variable Click to add a new editable variable (row) below the
existing variable (row).
Copy Corner Click to add a new editable column filled with the data from
a highlighted column. This column is added to the right of
the existing columns
Delete Click to delete a highlighted corner or variable (which is
not added by the pcf file).
Note: Variables (rows) and Corners added by the pcf file
cannot be deleted. Variables (rows) and Corners added by
the dcf file or the UI can be deleted.

June 2004 16 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Item Description and Usage

Disable Click to disable a highlighted corner column. This
particular corner will not be analysed after simulation and
the column is grayed out. Once the column is disabled, the
Disable button changes to an Enable button. You can re-
enable the disabled corner by selecting it and clicking on
the Enable button.
Run/ Click Run to run all the corners that are not disabled. Click
Stop Stop to end a running corners analysis.
Click Run to run all the corners in the pane which are not
disabled.
Note: Run turns to Stop once you start a run.

Performance Measurements Pane

The Performance Measurements pane, located in the lower section of the Virtuoso ®
Analog Corners Analysis window, displays information about the currently defined
measurements.

The information in this pane is usually loaded from design customization files (DCFs) using
the loadDcf command in your .cdsinit file. In addition, any outputs defined in the
Virtuoso ® Analog Design Environment window when you first start the corners analysis
option also appear. You can also load measurements from PCF files. You can also use the
Calculator to get an expression.
Note: You can also use the Add Measurement button to add a measurement.

To specify or change the measurements, you modify the information in this pane. You can
select a measurement by clicking in any of the fields in the measurement pane.

June 2004 17 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Note: The Cut, Paste and Copy keys work in the table fields. You can use these keys to
copy a measurement expression into another expression, from within the Corners window.

The items in the Performance Measurements pane are described in the following table:

Item Description and Usage

Measurement column Click to select a field in the Measurement column then
type or edit a name to be used as the label when the
expression is plotted or printed.
Expression column Click to select a field in the Expression column then type
or edit an expression to be evaluated for each corner.
Target column Click to select a field in the Target column and then type
the ideal target value for the measurement. This value is
used when a residual plot is created.
Lower column Left click twice to select a field in the Lower column and
then type the lowest acceptable value for the
measurement. This value is used when a residual plot is
created.
Upper column Left click twice to select a field in the Lower column and
then type the highest acceptable value for the
measurement. This value is used when a residual plot is
created.
Plot checkbox Select a checkbox if you want the output to appear as a
graph.
Print checkbox Select a checkbox if you want the output to appear as text.
Add Measurement button Click to add a new editable row below the existing rows.
Delete Measurement button Click to delete a highlighted row from the Performance
Measurements pane.
Calculator... button Click to open the calculator window.
Note: If the Calculator is not open and if you click on Get
Expression, it will invoke Calculator.
Get Expression button With an Expression field selected, click this button to
retrieve the expression displayed in the calculator buffer.
Note: The existing text in the field is replaced.

June 2004 18 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Split Pane Adjustment Bar

The split pane adjustment bar between the Corner Definitions Pane and the Measurements
Pane can be used to alter the area used by each pane. You can alter the area used by each
pane by dragging the bar upwards or downwards.

Status Display
The status display shows messages in one of three colors, depending on the type of
message.

Red Error Messages

Orange Internal Error Messages
Gray Information Messages

The corners analysis tool also writes messages to the corners log file, corners0.log. The
corners analysis tool puts the log file in the directory where you start the Cadence® software.

Keyboard Navigation and Shortcuts

Listed below are some shortcut keys that can be used for navigation of the form and tables.
These keys can also be used during the row and column selection mode to change the
selected row or column.

Tab Moves through table entries from left to right. Wraps to the next row
and at the end jumps back to the top.
Shift-Tab Reverse of Tab.
Arrow keys Moves as expected, does not wrap around at all.
F2 Opens/Closes a cyclic box.
Page Down Scrolls the table down if there is a vertical scroll bar.
Page Up Scrolls the table up if there is a vertical scroll bar.
Home Moves to the first column in the row.
End Moves to the last column in the row.

June 2004 19 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Running a Corners Analysis

The following sections describe the major steps involved in setting up and running a corners
analysis using PCFs and DCFs and also using the UI.
■ “Defining the Process and Corners for an Analysis” on page 20
■ “Defining Performance Measurements” on page 26
■ “Controlling the Corners Analysis” on page 28

Defining the Process and Corners for an Analysis

To specify the corners for an analysis, you begin by loading a set of predefined corners from
one or more process customization files (PCFs). The loading can occur automatically under
the control of a .cdsinit file or you can load PCFs from the graphical user interface. For
information about using the .cdsinit file to load PCFs and DCFs, see “Using a .cdsinit File
to Load PCFs and DCFs” on page 41. For information about using the .cdsinit file to load
PCFs and DCFs, see “Using a .cdsinit File to Load PCFs and DCFs” on page 41.

To tailor the predefined corners to the specific circuits you are working on, you can also load
one or more files containing changes and additions to the basic set of corners. These files
are called design customization files (DCFs). For information on preparing PCFs and DCFs,
see “Creating Process and Design Customization Files” on page 37.

If, after loading the PCFs and DCFs, you find that more changes are necessary, you can use
the graphical user interface to specify new corners or change any editable existing corners.

You can load multiple sets of corners information into the corners analysis option.
■ If you load a file or files that define more than one process, the processes appear in the
Process cyclic field in the Virtuoso ® Analog Corners Analysis window.
■ If you add more than one file (such as a DCF), that modifies a specific process, the
contents of files that are loaded are added to the contents of the existing files.

You can add a process using the procedure described in “Using the Virtuoso® Analog
Corners Analysis Window to Define a Process” on page 57.

Using the Graphical User Interface to Load PCFs and DCFs

There are various ways to load PCFs and DCFs. One such way is through the .cdsinit file.
The .cdsinit file typically specifies the PCFs and DCFs, so usually when you open the
Virtuoso ® Analog Corners Analysis window, it already contains some corner, variable,

June 2004 20 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

and measurement definitions. However, if there are no predefined corners or if you need to
load a different set, you can use the following steps to load PCFs and DCFs from the
graphical user interface. The third way of loading PCFs and DCFs is through an OCEAN
script.
1. Choose File – Load.
If you have made changes in the Virtuoso ® Analog Corners Analysis window, the
Save Changes? dialog box appears.

2. Click either Save or Save As to save the changes. If you do not want to save the
changes made, but want to load the PCFs and DCFs anyway, click Don’t Save. Click
Cancel Load if you want to retain the existing set of PCFs and DCFs.

June 2004 21 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The Load dialog box appears.

3. Click on the Look In drop down field to go to the specific directory. You can also navigate
using the iconified buttons located next to the Look In field. Placing the pointer on each
of the buttons displays a tooltip that describes the function of the button, as follows:
Button Function
Up One Level Opens the directory one level above the
active directory.
Home Opens the home directory.
Create New Folder Creates a new directory in the active
directory.
List Shows the names of the files in the directory.

June 2004 22 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Button Function
Details Shows additional information about the files
such as the file size and the date last saved.

Note: You can double-click a directory folder to descend into that directory.
4. In the Files of Types field, select a type of file from the given list. The list of files is
automatically updated. Default is All Files(*.*).
5. Select the file that you want to load. The name of the file is reflected in the File Name
field. Click on Load button to load the file. Click on Cancel button if you want to cancel
the operation.

Specifying Additional Corners

If you need to specify additional corners from those loaded in the PCFs and DCFs, you can
create new corners in the graphical user interface. You can either create new corners or copy
existing corners and modify them.

Creating a New Corner

To create a new corner,

1. Choose Edit – Corner Definition – Add Corner. You also click on the Add Corner
button.
The Enter Corner Name form is displayed.
2. Type a name for the new corner.
3. If you do not want to add the corner, click Cancel Add Corner. Click OK if you want to
add the corner. A new column appears at the right side of the Corner Definitions pane of
the Virtuoso ® Analog Corners Analysis window. The new column is named with the
name from Step 2.
4. Edit the rest of the column as desired.

Copying and Modifying an Existing Corner

If one of the existing corners is similar to the corner you want to use, you can copy the existing
corner and change the copy to meet your needs
1. Highlight the column for the corner you want to copy.

June 2004 23 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

2. Choose Edit – Corner Definition – Copy Corner. You can also click on the Copy
Corners button.
The Enter Corner Name form is displayed.
3. Enter a name for the new corner.
4. If you do not want to continue, click Cancel Copy Corner. Click OK if you want to
continue. A new column appears at the right side of the Corner Definitions pane of the
Virtuoso ® Analog Corners Analysis window.
5. Fill in the rest of the column as necessary.

Enable Corner

1. Select a disabled corner.

2. Choose Edit – Corner Definitions – Enable Corner or click Enable. The selected
corner will be enabled.

Disable Corner

1. Select an enabled corner.

2. Chose Edit – Corner Definitions – Disable Corner or click Disable. The selected
corner will be disabled. This particular corner will not be analysed after simulation and
the column is grayed out. Once the column is disabled, the Disable button changes to
an Enable button. You can re-enable the disabled corner by selecting it and clicking on
the Enable button.
Note: You can disable only an editable Variable (row) or column. Variables (rows) and
Corners added by the pcf file are not editable. Variables (rows) and Corners added by
the dcf file or the UI are editable.

Adding New Variables

There are three kinds of variables you can define for a corner: group variables, process
variables and design variables. For information on adding group and process variables, see
“Using the Virtuoso® Analog Corners Analysis Window to Modify Process Model Information”
on page 58. For guidance on adding design variables, see the next section.

Adding a Row for a New Design Variable

To add a new design variable to the existing variables,

June 2004 24 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

1. Choose Edit – Corner Definition – Add Variable. You can also click on the Add
Variable button.
The Enter Variable Name form is displayed.
2. Type a name for the new design variable.
The design variable is added not only to the current process but also to all the other
processes listed in the process cyclic field of the Virtuoso ® Analog Corners Analysis
window.
3. Click OK. If you want to continue. Otherwise, click Cancel Add Variable.
4. (Optional) Select the new variable field in each of the corners, and type the values you
want to use.

Deleting Corners or Rows

You cannot delete corners and variables (rows) added by a PCF. However, if the DCFs load
corners you do not plan to use, you can delete them. You can also delete un-needed rows
added by DCFs or from the Virtuoso ® Analog Corners Analysis window. Deleted corners
and rows disappear from the Corners Definition pane of the Virtuoso ® Analog Corners
Analysis window and their underlying data is erased.Corners added from the Corners UI can
also can be deleted.

Deleting Corners

To delete a corner,
1. Highlight the column for the corner you want to delete.
2. Choose Edit – Corner Definitions – Delete Selected or click Delete.
The highlighted column disappears from the pane.

Deleting Rows

To delete a row,
1. Highlight the row you want to delete.
Note: You can delete only rows defined by a DCF or added by using the Virtuoso ®
Analog Corners Analysis window. You cannot delete rows defined in a PCF.
2. Choose Edit – Corner Definitions – Delete Selected or click Delete.

June 2004 25 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The highlighted row disappears from the pane.

Defining Performance Measurements

For convenience, measurements are often specified in design customization files (DCFs).
Measurements defined in this way are displayed in the Performance Measurements pane,
where you can examine them. In addition, any outputs defined in the Virtuoso ® Analog
Design Environment window when you first start the corners analysis option are also
displayed.

If the existing measurements do not meet your needs, you can add new measurements or
make and modify copies of the existing measurements. If you have no plans to use a
measurement, you can delete it. You can add or change performance measurements either
before or after you run the analysis. You can also specify measurements through a PCF file.

Creating a New Performance Measurement by Entering It Directly

To create a new performance measurement by entering it directly,

1. Choose Edit – Performance Measurements – Add Measurements or click Add
Measurement.
The Enter Measurement Name form is displayed.
2. Enter a name for the new Measurement.
3. Click Cancel Add Measurement if you do not want to continue. Click OK, if you want
to continue. A new row will appear in the Corner Performance measurement pane.
4. Type the expression in the Expression field.
5. (Optional) Type the Target, Lower and Upper values for the new measurement. You
need to specify these values only if you plan to use this performance measurement in a
residual plot. A residual plot allows you to easily see whether a scalar measurement falls
within the specified boundaries for all of your corners using a histogram like bar plot.
Note: Target is a target value for a scalar measurement.Upper is the acceptable upper
boundary for a scalar measurement. Lower is an acceptable lower boundary for a scalar
measurement. If Target, Upper and Lower bound are set for a waveform, they will not
be used at all. These options are only used for the residual plots.

Creating a New Performance Measurement by Using the Calculator

To create a new performance measurement using the calculator,

June 2004 26 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

1. Choose Edit – Performance Measurements – Add Measurement or click Add

Measurement.
The Enter Measurement Name form is displayed.
2. Type a name for the new measurement.
3. Click Cancel Add Measurement if you do not want to continue. Click OK, if you want
to continue. A new row will appear in the Corner Performance measurement pane.
4. Choose Tools ->Calculator or click Calculator to open the calculator window.
5. Build the measurement expression in the calculator.
For information on using the calculator, see the Waveform Calculator User Guide.
6. In the Virtuoso ® Analog Corners Analysis window, select the Expression field for the
new measurement.
Note: If the expression field is selected, then only the Get Expression button will be
highlighted.
7. Choose Tools – Get Expression, or click Get Expression to retrieve the expression
from the calculator and place it in the selected Expression field.
8. (Optional) Type the Target, Lower and Upper values for the new measurement. You
need to specify these values only if you plan to use this performance measurement in a
residual plot.

Deleting a Performance Measurement

You can delete any measurement displayed in the Virtuoso ® Analog Corners Analysis
window. A deleted measurement disappears from the Performance Measurements pane, and
the data underlying it is erased. If you delete a measurement and then save the setup, the
deleted measurement is not included in the saved setup.

To delete a measurement,
1. Highlight the row for the measurement you want to delete.
2. Choose Edit – Performance measurement – Delete Measurement, or click Delete
Measurement.
The highlighted measurement disappears from the pane.

June 2004 27 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Controlling the Corners Analysis

You are ready to run the analysis after defining the corners and specifying the performance
measurements you need. To do this,
❑ Disable the corners that you do not want to run.
❑ Select the output format for the measurements to determine what type of
measurement output you want, if any. You can use the Plot and Print checkboxes
to specify whether you require a graphic or text output.
To choose the output formats for each measurement, click the Plot and Print
checkboxes on the right side of the Performance Measurements pane.

Then, run the analysis.

Running and Stopping the Analysis

To run the analysis,

➤ Choose Simulation – Run or click Run.

To stop an analysis running on a single machine,

1. Choose Simulation – Stop or click Stop.

To stop a simulation running distributed, use Job Monitor. For more information about Job
Monitor, refer to the Virtuoso Analog Distributed Processing Option User Guide.
Note: The Run button automatically changes between Run/Stop depending on whether a
Corners process is currently running or not.

June 2004 28 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Evaluating Corners Analysis Results

When the analysis finishes, the corners analysis option plots or lists the results according to
whether you chose text or graphic outputs in the Performance Measurements pane.
Note: If you run a distributed simulation, the results do not plot or list automatically.

If you want a different set of outputs from those you chose before running the analysis, you
can make new choices in the Performance Measurements pane and then choose Tools –
Plot or Print Outputs from the menu). In response, the corners analysis option evaluates
the selected measurements and displays new lists or plots.

Text Outputs
For a scalar measurement, a text output looks like this.

Each column in this window displays the value of a scalar measurement for each of the
corners. In this example, bandwidth varies from a low of 130.7 K under the slowslow corner
conditions, to a high of 318.7 K under the fastfast corner conditions.

June 2004 29 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

For a waveform measurement, a text output looks like this.

The first column in this window lists the data points for an analysis. Each subsequent column
lists data for a particular corner. In this example, at a frequency of 758.6 Hz, the phase for the
slowslow corner is -336.1 m and for the fastfast corner is -137.6 m.

Graphic Outputs
There are two kinds of graphic output, a residual plot for scalar data and a family-of-curves
plot for waveform data. A residual plot allows you to easily see whether a scalar measurement

June 2004 30 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

falls within the specified boundaries for all of your corners using a histogram like bar plot. A
residual plot looks like this.

The preceding residual plot, has a specified value of 200 for Target and shows that four of
the corners produce values within specifications. One corner produces a value that does not
lie within the lower boundary of the acceptable range. This result implies that yield for the
manufactured circuit will be less than 100 percent if the circuit is produced in its current form.
For greater yield, the circuit designer might want to redesign the circuit so it performs
acceptably for all the corners.

June 2004 31 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The set of curves for all the corners, looks like this.

The preceding plot shows how the phase varies as a function of frequency for each one of
the corners.

Saving Setup Information

The corners option setup consists of all the information in the Virtuoso ® Analog Corners
Analysis window, including the corner definitions and performance measurements. With
menu selections in the File entry, you can save the setup back to the original files, save the
setup to a specified file, and load a saved setup.

June 2004 32 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Saving Setup Information to the Original Files

To save the current setup back to the files from which it was loaded,
➤ Choose File – Save Setup.

If the PCFs and DCFs are not writable, the corners analysis option reports an error, and
changes and additions made in the Virtuoso ® Analog Corners Analysis window are not
saved.

If the PCFs and DCFs are writable, the corners analysis option saves changes back to those
files, overwriting the original contents of those files. As a result, any comments you might
have in the PCFs or DCFs are overwritten and lost. The corners analysis option saves any
additions to the DCF loaded last, if possible or to a newly created file called
NewEntries.dcf. This implies that if you add any Corners, Variables or Measurements,
they are added to the last loaded DCF. The NewEntries.dcf file is created when no PCF
and no DCF are loaded.

To avoid overwriting comments and to have the changed setup saved in a single easy-to-
understand location, use File -> Save Setup As, described in the following section.

Saving Setup Information to a Specified File

To save the current setup in a file you specify,
1. Choose File – Save Setup As.

June 2004 33 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The Save Setup As form is displayed.

2. Click on the Look In drop down field to go to the specific directory. You can also navigate
using the iconified buttons located next to the Look In field. Placing the pointer on each
of the buttons displays a tooltip that describes the function of the button, as follows:
Button Function
Up One Level Opens the folder one level above the active
folder.
Home Opens the home directory.
Create New Folder Creates a new folder in the active folder.
List Shows the names of the files in the directory.
Details Shows additional information about the files
such as the file size and the date last saved.

Note: You can double-click a directory folder to descend into that directory.

June 2004 34 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

3. In the Files of Types field, select a type of file from the given list. The list of files is
automatically updated. Default is All Files(*.*).
Select the file where the setup information is to be saved. The name of the file is reflected
in the File Name field.
4. Click Save. Click on the Cancel button if you want to cancel the operation.
Note: If you double-click on a selected file, the information will be directly saved to the
file.
All of the existing corners option information, whether loaded from PCFs, DCFs or
through the corners option graphical user interface, is saved to the file you specify. If
necessary, you can then cut and paste the lines into other PCFs and DCFs.

Saving a Script
The Open Command Environment for Analysis (OCEAN) lets you set up, simulate, and
analyze circuit data. OCEAN is a text-based process you can run from a UNIX shell or from
the Command Interpreter Window (CIW). You can type OCEAN commands in an interactive
session, or you can create scripts containing your commands and load those scripts into
OCEAN.

You can use the Virtuoso ® Analog Corners Analysis window to set up the analysis you
need, and save the setup procedure in an OCEAN script. You can then edit the script to add
simulation or postprocessing commands as needed.

For more information about OCEAN commands and scripts, see the OCEAN Reference.

To create a script and save it,

➤ Choose File – Save Script.

June 2004 35 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The Save Script As form appears so you can specify a file for the script.

Using Process, Design, and Modeling Files

There are usually three different kinds of files associated with setting up a corners analysis.
■ Process customization files (PCFs) define processes, groups, variants, and corners
shared by an entire organization. PCFs are usually created by a process engineer or
process group.
■ Design customization files (DCFs) contain definitions used for a particular design or for
several designs within a design group. DCFs are usually created by designers, who use
the DCFs to add design-specific information to the general information provided in PCFs.
■ Modeling files specify the model parameter values to be used for components during a
corners analysis. These files are usually created by a process engineer or process
group.

June 2004 36 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Day-to-day use of the corners analysis option typically does not involve changing a PCF or
DCF. However, if you are involved in writing or changing these kinds of files, read the following
sections for guidance.

Creating Process and Design Customization Files

The process customization files (PCFs) and design customization files (DCFs) contain
Cadence SKILL language commands that define the basic corners and measurements to be
used during analysis. The following sections illustrate how you can use the commands to
develop the set of definitions you need.

You can use any of the corners option SKILL language PI commands in either the PCFs or
DCFs. However, the commands used to define the process, the corners, and the corner
variables are customarily placed in the PCF. The commands used to specify design variables
and measurements, because they are design specific, are usually placed in the DCF.
Commands Normally in a PCF Commands Normally in a DCF
corAddProcess corAddDesignVar
corSetModelFile corSetDesignVarVal
corAddCorner corSetCornerRunTempVal
corAddGroupAndVariantChoices corAddMeas
corAddModelFileAndSectionChoices corSetMeasExpression
corSetCornerModelFileSection corSetMeasLower
corAddProcessVar corSetMeasUpper
corSetProcessVarVal corSetMeasTarget
corSetCornerGroupVariant corSetMeasGraphicalOn
corSetCornerNomTempVal corSetMeasTextualOn

Commands Used in Both

corSetCornerVarVal
corCopyCorner

The corSetModelFile command can be used only with the single model library style.

For more information, including the formal syntax for the commands, see the Virtuoso®
Analog Design Environment SKILL Language Reference.

To debug PCFs and DCFs, consider using OCEAN. The feedback the corners analysis option
provides is limited, but OCEAN provides more detailed feedback that makes it easier to find
and correct errors. For examples of OCEAN scripts that illustrate using PCFs and DCFs, see
the following directory in your installation hierarchy:

June 2004 37 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

your_install_dir/tools/dfII/samples/artist/corners

Example: Preparing a Process Customization File

The process customization file (PCF) adds the name of a new process to the corners analysis
option graphical user interface and defines the basic set of corners. For example, the
following PCF adds the process name P50u, specifies the modeling style as
singleModelLib, and defines three corners: slowslow, nominal, and fastfast.
; Example PCF file for the process P50u.
corAddProcess( "P50u" "~/processes" ’singleModelLib )
corSetModelFile("P50u" "P50uModelFile.scs")
; Prepare to add a process variable to each corner.
corAddProcessVar( "P50u" "EdgeEffect" )
; Now add the corners, specifying the values and choices for each.
corAddCorner( "P50u" "fastfast" )
corSetCornerVarVal( "P50u" "fastfast" "EdgeEffect" "1.18" )
corAddCorner( "P50u" "slowslow" )
corSetCornerVarVal( "P50u" "slowslow" "EdgeEffect" "1.12" )
corAddCorner( "P50u" "nominal" )
corSetCornerVarVal( "P50u" "nominal" "EdgeEffect" "1.15" )

The modeling values for the fastest, typical, and slowest variants are not defined in
the PCF. Instead, they are defined in the modeling file. For example, assume the
P50uModelFile.scs referred to by the P50u PCF contains the following statements.
.LIB slowest
.model npn2 npn tf=120n
.model nmosR nmos tox=120n
.ENDL slowest
.LIB typical
.model npn2 npn tf=100n
.model nmosR nmos tox=100n
.ENDL typical
.LIB fastest
.model npn2 npn tf=80n
.model nmosR nmos tox=80n
.ENDL fastest

June 2004 38 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Loading P50u PCF, which refers to the P50uModelFile.scs, produces the following
arrangement in the Virtuoso ® Analog Corners Analysis window.

Note: temp is always added with 27 being the default value for all corners.

Example: Preparing a Design Customization File

The DCF adds design-specific variables and measurements to the corners analysis option
graphical user interface that is specified in general by information in a PCF. For example, the
following DCF adds a design variable, sets the run temperature, and adds information to the
Performance Measurements pane:
corAddDesignVar( "vss" )
corSetDesignVarVal( "vss" "" )
corSetCornerVarVal( "P50u" "fastfast" "vss" "70" )
corSetCornerVarVal( "P50u" "slowslow" "vss" "50" )
corSetCornerVarVal( "P50u" "nominal" "vss" "60" )
corSetCornerRunTempVal("P50u" "slowslow" -35)
; You must add the measurement before you define it.
corAddMeas( "bandwidth" )
corSetMeasExpression( "bandwidth" "bandwidth(VF('/vout') 3 'low')" )
corSetMeasLower("bandwidth" "8Mhz")
corSetMeasUpper("bandwidth" "12Mhz")
corSetMeasTarget("bandwidth" "10Mhz")

June 2004 39 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Loading this DCF along with the P50u PCF described in the previous section changes both
panes in the Virtuoso ® Analog Corners Analysis window. The Corner Definitions pane
looks like this.

The Performance Measurements pane looks like this.

June 2004 40 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Using a .cdsinit File to Load PCFs and DCFs

A convenient way to load process and design customization files is to use your .cdsinit
file. You can set up your files in the following ways.
To load both PCFs and DCFs explicitly in To load DCFs explicitly and have them
the .cdsinit file load the PCFs they need
➤ Make sure your .cdsinit file loads all 1. Add load statements to each DCF for the
the necessary PCFs and DCFs. PCF files it uses. That way, when you
load the DCF file, it loads the PCF files
For example, this .cdsinit file loads automatically.
several PCFs and DCFs.
For example, this fragment of the
loadPcf "process1.pcf"
loadPcf "process2.pcf" myanalog35u.dcf file loads the
loadDcf "cellPhone23.dcf analog35u.pcf file.
loadDcf "opamp47.dcf ; This is the myanalog35u.dcf
loadPcf("mypath/
analog35u.pcf")
2. Set up your .cdsinit file so it loads the
DCF. For example, this .cdsinit file
fragment loads the myanalog35u.dcf
file (which then loads the
analog35u.pcf).
loadDcf("/mnt4/radhikak/
tools/
dfII/src/corners/
myanalog35u.dcf")

Whichever way you choose to load your files, you must make sure PCFs and DCFs refer only
to definitions that have already been loaded. Usually, that means you must load PCFs before
you can define corners or measurements in a DCF.

Implementing Modeling Styles

The Corners analysis option supports five different modeling styles: single model library,
multiple model library, single numeric, multiple numeric and multiple parametric.

Starting with IC 5.1.41 release, some changes have been made to the Corners modeling
styles. The single numeric style is the default. For third-party OASIS corner integration, the
single model library and single numeric styles are also supported. To use and define corner
processes, you need to have a valid tool session created before loading PCF files.

June 2004 41 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Cadence recommends the single model library or multiple model library styles for users
running the Cadence ® Spectre ® Circuit Simulator. For users running the SpectreS
simulator, Cadence recommends the multiple numeric modeling style. The remaining two
modeling styles, single numeric and multiple parametric, should be used with caution.

The following sections illustrate the file structures used by these styles and give examples of
PCFs tailored to each style. For detailed information, see the sections listed below.
■ “Single Model Library Style”
■ “Multiple Model Library Style” on page 48
■ “Single Numeric Style” on page 51
■ “Multiple Numeric Style” on page 52
■ “Multiple Parametric Modeling” on page 54

Single Model Library Style

Cadence recommends this easy-to-read style for use in corners analysis. With this
approach,
■ All models for all corners are located in a single model file
■ The model file is located in the base directory
■ The model file can have any name.
You can type the name in the Virtuoso ® Analog Corners Analysis window or use the
corSetModelFile procedure to specify the name in a PCF or DCF.

June 2004 42 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The following table illustrates the single model library style with an example path, file, and file
contents. If you prefer, you can also use the .LIB syntax for this modeling style. The .LIB
syntax is an hspice modelling syntax that is supported in Spectre.
Single Model Library Style (Native Spectre)

Path Filename File Contents

./CORNERS/fab6/ mylibfile.scs library processA
section slowslow
model npn2 npn tf=120n
model npn9 npn tf=320n
model nmosR nmos tox=120n
model nmos8 nmos tox=320n
endsection

section nom
model npn2 npn tf=100n
model npn9 npn tf=300n
model nmosR nmos tox=100n
model nmos8 nmos tox=300n
endsection

section fastfast
model npn2 npn tf=80n
model npn9 npn tf=380n
model nmosR nmos tox=80n
model nmos8 nmos tox=380n
endsection
endlibrary

The following code illustrates how you can refer to this modeling structure in a PCF.
corAddProcess("fab6" "./CORNERS/fab6/" ’singleModelLib)
corSetModelFile("fab6" "mylibfile.scs")
corAddProcessVar("fab6" "vdc")
corAddCorner("fab6" "slowslow"
?runTemp 20
?nomTemp 27
?vars ’( ("vdc" 2) )
)

June 2004 43 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The Virtuoso ® Analog Corners Analysis window produced by this PCF looks like this.

Changes for Setting New Groups or Variants in the singleModelLib Style

Starting with the IC 5.1.41 release, you can specify the same section in a single model library
style with several corners associated with your design. These changes require that you define
the corSetCornerGroupVariant() function as follows:
corSetCornerGroupVariant( <process_Name> <corner_Name> <group_Name> <variant> )

where:

process_Name The existing process you want to use, for example, fab6.

corner_Name The variant name, the sections in your model file.

group_Name Any name with which you want to group your variants.

variant The name of the variant you want to specify with different
corners.

The following code illustrates how you can refer to same section or variant within several
corners or groups using the single model library style in a PCF.
corAddProcess("singleModelLib" "./CORNERS/singleModelLib" ’singleModelLib)

corSetModelFile("singleModelLib" "singleModelLib.scs")

June 2004 44 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

corAddModelFileAndSectionChoices("singleModelLib" "groupname" ’( "slowslow"

"slowfast" "typtyp" "fastslow" "fastfast") )

corAddProcessVar("singleModelLib" "vdd")

corAddProcessVar("singleModelLib" "vss")

corAddDesignVar( "Cload" )

corAddCorner( "singleModelLib" "cor1" )

corSetCornerNomTempVal( "singleModelLib" "cor1" 27 )

corSetCornerRunTempVal( "singleModelLib" "cor1" 125 )

corSetCornerVarVal( "singleModelLib" "cor1" "Cload" "260f" )

corSetCornerVarVal( "singleModelLib" "cor1" "vss" "-2.7" )

corSetCornerVarVal( "singleModelLib" "cor1" "vdd" "2.7" )

corSetCornerGroupVariant( "singleModelLib" "cor1" "groupname" "slowslow" )

corAddCorner( "singleModelLib" "cor2" )

corSetCornerNomTempVal( "singleModelLib" "cor2" "27" )

corSetCornerRunTempVal( "singleModelLib" "cor2" 27 )

corSetCornerVarVal( "singleModelLib" "cor2" "Cload" "200f" )

corSetCornerVarVal( "singleModelLib" "cor2" "vss" "-3" )

corSetCornerVarVal( "singleModelLib" "cor2" "vdd" "3" )

corSetCornerGroupVariant( "singleModelLib" "cor2" "groupname" "slowslow" )

corAddCorner( "singleModelLib" "cor3" )

corSetCornerNomTempVal( "singleModelLib" "cor3" 27 )

corSetCornerRunTempVal( "singleModelLib" "cor3" 27 )

corSetCornerVarVal( "singleModelLib" "cor3" "Cload" "200f" )

corSetCornerVarVal( "singleModelLib" "cor3" "vss" "-3" )

June 2004 45 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

corSetCornerVarVal( "singleModelLib" "cor3" "vdd" "3" )

corSetCornerGroupVariant( "singleModelLib" "cor3" "groupname" "fastfast" )

corAddCorner( "singleModelLib" "cor4" )

corSetCornerNomTempVal( "singleModelLib" "cor4" "27" )

corSetCornerRunTempVal( "singleModelLib" "cor4" 27 )

corSetCornerVarVal( "singleModelLib" "cor4" "Cload" "200f" )

corSetCornerVarVal( "singleModelLib" "cor4" "vss" "-3" )

corSetCornerVarVal( "singleModelLib" "cor4" "vdd" "3" )

corSetCornerGroupVariant( "singleModelLib" "cor4" "groupname" "fastfast" )

corAddCorner( "singleModelLib" "cor5" )

corSetCornerNomTempVal( "singleModelLib" "cor5" 27 )

corSetCornerRunTempVal( "singleModelLib" "cor5" -55 )

corSetCornerVarVal( "singleModelLib" "cor5" "Cload" "160f" )

corSetCornerVarVal( "singleModelLib" "cor5" "vss" "-3.3" )

corSetCornerVarVal( "singleModelLib" "cor5" "vdd" "3.3" )

corSetCornerGroupVariant( "singleModelLib" "cor5" "groupname" "slowslow" )

corAddCorner( "singleModelLib" "cor6" )

corSetCornerNomTempVal( "singleModelLib" "cor6" 27 )

corSetCornerRunTempVal( "singleModelLib" "cor6" -55 )

corSetCornerVarVal( "singleModelLib" "cor6" "Cload" "190f" )

corSetCornerVarVal( "singleModelLib" "cor6" "vss" "-3.0" )

corSetCornerVarVal( "singleModelLib" "cor6" "vdd" "3.0" )

corSetCornerGroupVariant( "singleModelLib" "cor6" "groupname" "slowfast" )

corAddMeas( "DCgain" )

June 2004 46 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

corSetMeasExpression( "DCgain" "ymax(db20(VF('/vout')))" )

corSetMeasTarget( "DCgain" 60 )

corSetMeasEnabled( "DCgain" t )

corSetMeasGraphicalOn( "DCgain" t )

corSetMeasTextualOn( "DCgain" nil )

corAddMeas( "bandwidth" )

corSetMeasExpression( "bandwidth" "bandwidth(VF('/vout') 3 'low')" )

corSetMeasEnabled( "bandwidth" t )

corSetMeasGraphicalOn( "bandwidth" t )

corSetMeasTextualOn( "bandwidth" nil )

corAddMeas( "gain" )

corSetMeasExpression( "gain" "dB20(VF('/vout'))" )

corSetMeasEnabled( "gain" t )

corSetMeasGraphicalOn( "gain" t )

corSetMeasTextualOn( "gain" nil )

corAddMeas( "phase" )

corSetMeasExpression( "phase" "phase(VF('/vout'))" )

corSetMeasEnabled( "phase" t )

corSetMeasGraphicalOn( "phase" t )

corSetMeasTextualOn( "phase" nil )

June 2004 47 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The Virtuoso ® Analog Corners Analysis window produced by this PCF looks like this.

Multiple Model Library Style

This style uses multiple library files, which must be specified by using the
corAddModelFileAndSectionChoices and corAddCorner commands in a PCF or
DCF. In other ways, this style is the same as the single model library style. For example, the
models might be located in the following files:
./CORNERS/fab6/path1/npn.scs
./CORNERS/fab6/path3/nmos.scs

June 2004 48 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The following table illustrates the multiple model library style. If you prefer, you can also use
the .LIB syntax for this modeling style.
Multiple Model Library Style

Path Filename File Contents

./CORNERS/fab6/ npn.scs library npn
path1/ section slow
model npn2 bjt tf=120n
model npn8 bjt tf=80n
endsection

section nom
model npn2 bjt tf=100n
model npn8 bjt tf=60n
endsection

section fast
model npn2 bjt tf=80n
model npn8 bjt tf=50n
endsection
endlibrary
./CORNERS/fab6/ nmos.scs library nmos
path3/ section slow
model nmosR mos3 tox=120n
model nmos2 mos3 tox=140n
endsection

section nom
model nmosR mos3 tox=100n
model nmos2 mos3 tox=115n
endsection

section fast
model nmosR mos3 tox=80n
model nmos2 mos3 tox=90n
endsection
endlibrary

The following code illustrates how you can refer to this multiple model library structure in a
PCF.
corAddProcess("fab6" "./CORNERS/fab6/" 'multipleModelLib)
corAddModelFileAndSectionChoices("fab6" "path1/npn.scs"

June 2004 49 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

'( "slow" "nom" "fast") )

corAddModelFileAndSectionChoices("fab6" "path3/nmos.scs"
'( "slow" "nom" "fast") )
corAddProcessVar("fab6" "vdc")
corAddCorner("fab6" "slowslow"
?sections '( ("path1/npn.scs" "slow")
("path3/nmos.scs" "slow") )
?runTemp 20
?nomTemp -27
?vars '( ("vdc" 2) )
)
corAddCorner("fab6" "nomnom"
?sections '( ("path1/npn.scs" "nom")
("path3/nmos.scs" "nom") )
?runTemp 30
?nomTemp 27
?vars '( ("vdc" 3) )
)
corAddCorner("fab6" "fastfast"
?sections '( ("path1/npn.scs" "fast")
("path3/nmos.scs" "fast") )
?runTemp 40
?nomTemp -27
?vars '( ("vdc" 4) )
)
corAddCorner("fab6" "fastslow"
?sections '( ("path1/npn.scs" "fast")
("path3/nmos.scs" "slow") )
?runTemp 50
?nomTemp -27
?vars '( ("vdc" 4) )
)

The Virtuoso ® Analog Corners Analysis window produced by this PCF looks like this.

June 2004 50 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Single Numeric Style

This modeling style is provided for backward compatibility. If you plan to run your corners
analysis with the Spectre simulator, Cadence recommends that you convert to a preferred
modeling style.
■ With this style, each corner is located in a separate file. If there are four corners, there
are four model files. All the model files have the same name.
■ Each model file is located in the subdirectory base_directory/corner_name. For
example, if one of the corner names is allfast, then one of the model files is located
in the base_directory/allfast subdirectory.
■ The common model filename can be anything.
❑ If you use the Spectre direct simulator, specify the name by choosing Setup –
Model Libraries from the menu in the Virtuoso ® Analog Design Environment
window, then type the name into the Model Library Setup form.
❑ If you use a socket simulator, choose Setup – Environment to open the
Environment Options form, then type the name into the Include File field.

The following table illustrates the file structure and contents for a model with three corners,
using the single numeric modeling style.
Single Numeric Style

Path Filename File Contents

./CORNERS/fab6/ models .model npn2 npn tf=120n
allslow/ .model npn9 npn tf=320n
.model nmosR nmos tox=120n
.model nmos8 nmos tox=320n
./CORNERS/fab6/ models .model npn2 npn tf=100n
allnom/ .model npn9 npn tf=300n
.model nmosR nmos tox=100n
.model nmos8 nmos tox=300n
./CORNERS/fab6/ models .model npn2 npn tf=80n
allfast/ .model npn9 npn tf=380n
.model nmosR nmos tox=80n
.model nmos8 nmos tox=380n

The following code illustrates how you can refer to this modeling structure in a PCF.
corAddProcess("fab6" "./CORNERS/fab6" ’singleNumeric)
corAddCorner("fab6" "allslow"
?runTemp -55
)

June 2004 51 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

corAddCorner("fab6" "allnom"
?runTemp -27
)
corAddCorner("fab6" "allfast"
?runTemp 55
)

The Virtuoso ® Analog Corners Analysis window produced by this PCF looks like this.

Multiple Numeric Style

This modeling style, which has the following characteristics, is provided for backward
compatibility.
■ With this style, each model is defined in a separate file. All model parameters are defined
with numeric values.
■ Each model file is located in the subdirectory base_directory/group/variant.
For example, if the model includes the group npn and the variant fast, then at least one
of the model files is located in the base_directory/npn/fast subdirectory.
■ Each model file can have any name, which the designer enters on the Edit Object
Properties form in the Virtuoso ® Analog Design Environment.

The following table illustrates the file structure and contents for the multiple numeric style.
Multiple Numerics

Path Filename File Contents

./CORNERS/fab6/ npn2.scs model npn2 bjt tf=120n
npn/slow/
npn9.scs model npn9 bjt tf=320n
./CORNERS/fab6/ npn2.scs model npn2 bjt tf=100n
npn/nom/
npn9.scs model npn9 bjt tf=300n

June 2004 52 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Multiple Numerics, continued

Path Filename File Contents

./CORNERS/fab6/ npn2.scs model npn2 bjt tf=80n
npn/fast/
npn9.scs model npn9 bjt tf=380n
./CORNERS/fab6/ nmosR.scs model nmosR mos3 tox=120
nmos/slow/
nmos8.scs model nmos8 mos3 tox=320n
./CORNERS/fab6/ nmosR.scs model nmosR mos3 tox=100n
nmos/nom/
nmos8.scs model nmos8 mos3 tox=300n
./CORNERS/fab6/ nmosR.scs model nmosR mos3 tox=80n
nmos/fast/
nmos8.scs model nmos8 mos3 tox=380n

The following code illustrates how you can refer to this modeling structure in a PCF.
corAddProcess("fab6" "./CORNERS/fab6" ’multipleNumeric)
corAddGroupAndVariantChoices("fab6" "npn"
’("slow" "nom" "fast")
)
corAddGroupAndVariantChoices("fab6" "nmos"
’("slow" "nom" "fast")
)
corAddCorner("fab6" "slowslow"
?variants ’(
("npn" "slow")
("nmos" "slow")
)
?nomTemp -55
?vars ’( ("Cload" 260f) )
)
corAddCorner("fab6" "slowfast"
?variants ’(
("npn" "slow")
("nmos" "fast")
)
?nomTemp -55
?vars ’( ("Cload" 200f) )
)

June 2004 53 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The Virtuoso ® Analog Corners Analysis window produced by this PCF looks like this.

Using the Multiple Numeric Modeling Style with the Spectre Simulator

When you run a multiple numeric modeling style corners analysis with the Spectre simulator,
ensure that the .cdsenv variable includeStyle is set to t.

Multiple Parametric Modeling

This modeling style, which has the following characteristics, is provided for backward
compatibility.
■ With this style, each model is defined in a separate file. There is a corresponding
parameter file for every model associated with each corner.
■ Model files are located in the subdirectory base_directory/group. For example, if
the model includes the group npn, then the model files associated with that group are
located in the base_directory/npn subdirectory.
■ Each parameter file is located in base_directory/group/variant. For example,
if the model includes the group npn and the variant fast, then at least one of the
parameter files is located in the base_directory/npn/fast subdirectory.
■ Each model file can have any name, which the designer enters on the Edit Properties
form in the Virtuoso ® Analog Design Environment.

June 2004 54 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The following table illustrates the file structure and contents for the multiple parametric
modeling style.
Multiple Parametric Style

Path Filename File Contents

./CORNERS/fab6/ npn2.scs include "npn2.param"
npn model npn2 bjt tf=TF2
npn9.scs include "npn9.param"
model npn9 bjt tf=TF9
./CORNERS/fab6/ npn2.param parameter TF2=120n
npn/slow/
npn9.param parameter TF9=320n
./CORNERS/fab6/ npn2.param parameter TF2=100n
npn/nom/
npn9.param parameter TF9=300n
./CORNERS/fab6/ npn2.param parameter TF2=80n
npn/fast/
npn9.param parameter TF9=380n
./CORNERS/fab6/ nmosR.scs model npn2 mos3 tf=TOXR
nmos
nmos8.scs model npn9 mos3 tf=TOX8
./CORNERS/fab6/ nmosR.param parameter TOXR=120
nmos/slow/
nmos8.param parameter TOX8=320n
./CORNERS/fab6/ nmosR.param parameter TOXR=100n
nmos/nom/
nmos8.param parameter TOX8=300n
./CORNERS/fab6/ nmosR.param parameter TOXR=80n
nmos/fast/
nmos8.param parameter TOX8=380n

The following code illustrates how you can refer to this modeling structure in a PCF.
corAddProcess("fab6" "./CORNERS/fab6" 'multipleParametric)
corAddGroupAndVariantChoices("fab6" "npn2"
'("slow" "nominal" "fast")
)
corAddGroupAndVariantChoices("fab6" "nmos8"
'("slow" "nominal" "fast")
)
corAddGroupAndVariantChoices("fab6" "npn9"
'("slow" "nominal" "fast")
)
corAddGroupAndVariantChoices("fab6" "nmosR"
'("slow" "nominal" "fast")
)

June 2004 55 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

corAddCorner("fab6" "slowslow"
?variants '(
("npn2" "slow")
("nmos8" "slow")
("npn9" "slow")
("nmosR" "slow")
)
?nomTemp -55
)
corAddCorner("fab6" "slowfast"
?variants '(
("npn2" "slow")
("nmos8" "fast")
("npn9" "slow")
("nmosR" "fast")
)
?nomTemp -55
)

The Virtuoso ® Analog Corners Analysis window produced by this PCF looks like this.

Using the Multiple Parametric Modeling Style with the Spectre Simulator

When you run a multiple parametric modeling style corners analysis with the Spectre
simulator, ensure that the .cdsenv variable includeStyle is set to t.

Using the Multiple Parametric Modeling Style with a Socket Simulator

To use a multiple parametric modeling style with a socket simulator, specify the parameter
files in an update.s file. For example, update.s might contain

June 2004 56 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

use npn2.s
use npn9.s
use nmosR.s
use nmos8.s

If you want to be able to override parameter declarations for specific corners, put these use
statements in an init.s file instead of in an update.s file. Because parameters defined in
the corners analysis option are processed after init.s parameters, you can use the corners
analysis option to override the init.s parameters.
Note: Corners is designed such that it does not create a runObjFile in each psf directory
it creates. Therefore, the results browser cannot be used to browse through the directories.
In order to browse individual directories, you can use the Create ROF feature.

Using the Virtuoso® Analog Corners Analysis Window to Define and

Update Processes
As described in “Creating Process and Design Customization Files” on page 37, processes
are often defined outside of the Virtuoso ® Analog Corners Analysis window and then
loaded when they are needed. However, you can also use the corners option graphical user
interface to define, update, and save a process.

Using the Virtuoso® Analog Corners Analysis Window to Define a Process

To define a new process,

1. Choose Setup – Add Process.

June 2004 57 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The Add Process form appears.

2. In the Process Name field, type the name you want to use for the new process.
3. Choose the model style you want to use for the new process.
4. Type the name of the base directory for the model file or files associated with the new
process.
5. If the model style you choose in Step 3 is Single Model Library, type the name of the
associated model file.
6. Type the names of process variables you want to add, separating them with a comma or
white space. (To delete process variables, select the row and click Delete in the
Virtuoso ® Analog Corners Analysis window.)
7. Click OK in the Add Process form to close it.

Using the Virtuoso® Analog Corners Analysis Window to Modify Process Model
Information

To modify an existing process,

1. Choose Setup – Add/Update Model Info.

June 2004 58 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The Add/Update Model Info form appears.

This form has two tabs: Process and Group/Variants.Only certain modeling styles allow the
Groups and Variants. You need to make modifications in all tabs before adding or updating
the process. Otherwise, you will have to get back in and update the process again.
1. Choose the Process that you want to change.
2. Choose the Model Style that you want to use for the changed process.
3. Type the name of the Base Directory for the model file or files associated with the
changed process. The form appears with the process base directory.
4. If the model style you choose in Step 3 is Single Model Library, type the name of the
associated model file.
5. Type the names of process variables you want to add, separating them with commas or
spaces. (To delete process variables, select the row and click Delete in the Virtuoso ®
Analog Corners Analysis window.)
6. Click OK in the Update Process/Model Info form to close it.

To specify the Groups and Variants, click the Groups/Variants tab.

1. Specify a name in the Group Name field.

June 2004 59 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

2. Specify the variants in the Variants field.

3. Click OK to close the form.

Requirements for Using the Spectre Simulator

When you run a single or multiple model library style corners analysis with the Spectre
simulator, ensure that you comply with the following requirements.
■ The value of the .cdsenv variable useAltergroup must be set appropriately. If your
Spectre model supports altergroups, ensure that useAltergroup is set to t. If your
Spectre model does not support altergroups, set useAltergroup to nil.
Regardless of the useAltergroup value, the corners analysis option does not use
altergroups when you run a mixed-signal (SpectreVerilog) or distributed simulation.
■ Every parameter used in a corner must be in the main circuit.

Working through an Extended Example

This section follows a corners session in detail, demonstrating how you might use the corners
analysis option to examine the characteristics of a real circuit. The example describes a
folded cascode circuit and explains how you might arrange the supporting model files. To
follow along, go to
your_install_dir/tools/dfII/samples/artist/corners/artistExample

and start icms. A .cdsinit file and the other files you need to run this example are all
included at that location.

June 2004 60 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Folded Cascode Schematic

The folded cascode used in this example has the following two-part schematic.

June 2004 61 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

This schematic includes several instances of pmos and nmos transistors. Each of the pmos
transistors is nominally identical. Similarly, each of the nmos transistors is nominally identical.
In reality, however, the attributes of each transistor differ slightly from the attributes of each of
the other transistors. In this example, you explore the extent of the variation and the effect the
variation has on the performance of the circuit.

Setting Up the Virtuoso® Analog Design Environment Window

To run this example, first set up the Virtuoso ® Analog Design Environment window.
1. From the CIW, choose Tools – Analog Environment – Simulation.
The Virtuoso ® Analog Design Environment window appears.
2. Choose Setup – Design.
The Choosing Design form appears.

June 2004 62 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

3. Select the RF_lib library and the foldedCascode cell. Click OK.
4. In the Virtuoso ® Analog Design Environment window, choose Session – Load
State.
The Loading State form appears.
5. Choose Corners from the State Name cyclic field. Click OK.
6. In the Virtuoso ® Analog Design Environment window, choose Outputs – To Be
Plotted – Select On Schematic.
The schematic window appears.
7. Click on the net connected to vout in the right side of the plot, then press the Esc key.
There are other outputs defined in the PCF, but this demonstrates how outputs defined
in the Virtuoso ® Analog Design Environment window are incorporated into the
Virtuoso ® Analog Corners Analysis window.

June 2004 63 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The Virtuoso ® Analog Design Environment window looks like this.

Modeling Style
This example uses the multiple model library style with the variants for pmos components
defined in one file of the multipleModelLib directory and the variants for nmos
components defined in another file of that same directory.

For example, the nmos components are defined in the file

CORNERS/multipleModelLib/nmosLib.scs

June 2004 64 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

This file contains

library nmosLib
section nom
include "../nmos/typ/nmos.scs"
endsection
section fast
include "../nmos/fast/nmos.scs"
endsection
section slow
include "../nmos/slow/nmos.scs"
endsection
endlibrary

As implemented in this example, the parameters for the variants are not actually included in
this file, although they could be. This example instead uses include statements to include
the files that contain the actual models.

The ../nmos/typ/nmos.scs file referred to in the nom section, for example, contains
simulator lang=spice
* VTI-derived Level=2 nominal model
.model nmos nmos level=2
+ vto = 0.775
+ tox = 400e-10
+ nsub = 8e+15
+ xj = 0.15U
+ ld = 0.20U
+ u0 = 650
+ ucrit = 0.62e+5
+ uexp = 0.125
+ vmax = 5.1e+4
+ neff = 4.0
+ delta = 1.4
+ rsh = 36
+ cgso = 1.95e-10
+ cgdo = 1.95e-10
+ cj = 195U
+ cjsw = 500P
+ mj = 0.76
+ mjsw = 0.30
+ pb = 0.8

These are the values the simulator uses when you run a corner that has the value of the nmos
variant set to nom. When you run a corner that uses the slow variant for the nmos
components, the simulator uses the values defined in
../nmos/slow/nmos.scs

The contents of the ../nmos/slow/nmos.scs file are the following:

simulator lang=spice

* VTI Level=2 slowN/slowP model

.model nmos nmos level=2
+ vto = 0.9

June 2004 65 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

+ tox = 430e-10
+ nsub = 1.0e+16
+ xj = 0.15U
*+ ld = 0.20U
+ ld = 0.075U
+ u0 = 620
+ ucrit = 0.62e5
+ uexp = 0.125
+ vmax = 5.1e4
+ neff = 4.0
+ delta = 1.4
+ rsh = 38
+ cgso = 2.10e-10
+ cgdo = 2.10e-10
+ cj = 215U
+ cjsw = 540P
+ mj = 0.76
+ mjsw = 0.30
+ pb = 0.8

The other variants for the nmos and pmos components are defined similarly.

Process Customization File (PCF)

This example does not use a design customization file (DCF) because all the necessary
corners and measurements are defined in a single PCF called multipleModelLib.pcf.
Defining the Virtuoso ® Analog Corners Analysis window in a single file simplifies the
example.

The multipleModelLib.pcf contains the following information.

corAddProcess( "multipleModelLib" "./CORNERS/multipleModelLib"

’multipleModelLib )
corAddProcessVar( "multipleModelLib" "vdd" )
corAddProcessVar( "multipleModelLib" "vss" )
corAddDesignVar( "Cload" )
corAddGroupAndVariantChoices( "multipleModelLib" "pmosLib.scs"
’("slow" "nom" "fast") )
corAddGroupAndVariantChoices( "multipleModelLib" "nmosLib.scs"
’("slow" "nom" "fast") )
corAddCorner( "multipleModelLib" "slowslow" )
corSetCornerGroupVariant( "multipleModelLib" "slowslow"
"nmosLib.scs" "slow" )
corSetCornerGroupVariant( "multipleModelLib" "slowslow"
"pmosLib.scs" "slow" )
corSetCornerNomTempVal( "multipleModelLib" "slowslow" 27 )
corSetCornerRunTempVal( "multipleModelLib" "slowslow" 125 )
corSetCornerVarVal( "multipleModelLib" "slowslow" "Cload" "260f" )
corSetCornerVarVal( "multipleModelLib" "slowslow" "vss" "-2.7" )
corSetCornerVarVal( "multipleModelLib" "slowslow" "vdd" "2.7" )
corAddCorner( "multipleModelLib" "fastslow" )
corSetCornerGroupVariant( "multipleModelLib" "fastslow"
"nmosLib.scs" "slow" )
corSetCornerGroupVariant( "multipleModelLib" "fastslow"

June 2004 66 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

"pmosLib.scs" "fast" )
corSetCornerNomTempVal( "multipleModelLib" "fastslow" "27" )
corSetCornerRunTempVal( "multipleModelLib" "fastslow" 27 )
corSetCornerVarVal( "multipleModelLib" "fastslow" "Cload" "200f" )
corSetCornerVarVal( "multipleModelLib" "fastslow" "vss" "-3" )
corSetCornerVarVal( "multipleModelLib" "fastslow" "vdd" "3" )
corAddCorner( "multipleModelLib" "typtyp" )
corSetCornerGroupVariant( "multipleModelLib" "typtyp" "nmosLib.scs"
"nom" )
corSetCornerGroupVariant( "multipleModelLib" "typtyp" "pmosLib.scs"
"nom" )
corSetCornerNomTempVal( "multipleModelLib" "typtyp" 27 )
corSetCornerRunTempVal( "multipleModelLib" "typtyp" 27 )
corSetCornerVarVal( "multipleModelLib" "typtyp" "Cload" "200f" )
corSetCornerVarVal( "multipleModelLib" "typtyp" "vss" "-3" )
corSetCornerVarVal( "multipleModelLib" "typtyp" "vdd" "3" )
corAddCorner( "multipleModelLib" "slowfast" )
corSetCornerGroupVariant( "multipleModelLib" "slowfast"
"nmosLib.scs" "fast" )
corSetCornerGroupVariant( "multipleModelLib" "slowfast"
"pmosLib.scs" "slow" )
corSetCornerNomTempVal( "multipleModelLib" "slowfast" "27" )
corSetCornerRunTempVal( "multipleModelLib" "slowfast" 27 )
corSetCornerVarVal( "multipleModelLib" "slowfast" "Cload" "200f" )
corSetCornerVarVal( "multipleModelLib" "slowfast" "vss" "-3" )
corSetCornerVarVal( "multipleModelLib" "slowfast" "vdd" "3" )
corAddCorner( "multipleModelLib" "fastfast" )
corSetCornerGroupVariant( "multipleModelLib" "fastfast"
"nmosLib.scs" "fast" )
corSetCornerGroupVariant( "multipleModelLib" "fastfast"
"pmosLib.scs" "fast" )
corSetCornerNomTempVal( "multipleModelLib" "fastfast" 27 )
corSetCornerRunTempVal( "multipleModelLib" "fastfast" -55 )
corSetCornerVarVal( "multipleModelLib" "fastfast" "Cload" "160f" )
corSetCornerVarVal( "multipleModelLib" "fastfast" "vss" "-3.3" )
corSetCornerVarVal( "multipleModelLib" "fastfast" "vdd" "3.3" )
corAddMeas( "DCgain" )
corSetMeasExpression( "DCgain" "ymax(db20(VF('/vout')))" )
corSetMeasTarget( "DCgain" 60 )
corSetMeasEnabled( "DCgain" t )
corSetMeasGraphicalOn( "DCgain" t )
corSetMeasTextualOn( "DCgain" nil )
corAddMeas( "bandwidth" )
corSetMeasExpression( "bandwidth" "bandwidth(VF('/vout') 3 'low')" )
corSetMeasEnabled( "bandwidth" t )
corSetMeasGraphicalOn( "bandwidth" t )
corSetMeasTextualOn( "bandwidth" nil )
corAddMeas( "gain" )
corSetMeasExpression( "gain" "dB20(VF('/vout'))" )
corSetMeasEnabled( "gain" t )
corSetMeasGraphicalOn( "gain" t )
corSetMeasTextualOn( "gain" nil )
corAddMeas( "phase" )
corSetMeasExpression( "phase" "phase(VF('/vout'))" )
corSetMeasEnabled( "phase" t )
corSetMeasGraphicalOn( "phase" t )
corSetMeasTextualOn( "phase" nil )

June 2004 67 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

You can load a PCF from the Virtuoso ® Analog Corners Analysis window, but it is often
easier to insert a statement in your .cdsinit file that loads the necessary PCFs
automatically. For example, if you look in the included .cdsinit file, you find the following
statement that loads the multipleModelLib.pcf file:
loadPcf( "~/multipleModelLib.pcf" )

Virtuoso® Analog Corners Analysis Window for Folded Cascode

To open the Virtuoso ® Analog Corners Analysis window,
➤ From the Virtuoso ® Analog Design Environment window, choose Tools – Corners.

The Virtuoso ® Analog Corners Analysis window, in this example, is defined primarily by
the multipleModelLib.pcf. In addition, the following items affect the appearance of the
window.
■ The run temperature variable temp always appears in the Corner Definitions pane of
the window. By default, it has the value 27. For this example, the
multipleModelLib.pcf sets the value of the run temperature explicitly for each
corner using skill function corSetCornerRunTempVal.
■ Any outputs defined in the Virtuoso ® Analog Design Environment window when you
first start the corners analysis option appear in the Performance Measurements pane.
That is why /vout appears as an expression in the Performance Measurements pane
for this example.

June 2004 68 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

When the Virtuoso ® Analog Corners Analysis window opens, the Corner Definitions
pane looks like this. (The slowslow corner, although not visible in this figure, also appears
in the actual Virtuoso ® Analog Corners Analysis window.)

The Performance Measurements pane looks like this.

June 2004 69 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

The measurement that appears in the Performance Measurements pane is defined in the
Outputs pane of the Virtuoso ® Analog Design Environment window and is automatically
copied into the Virtuoso ® Analog Corners Analysis window.

Changing Values in the Virtuoso® Analog Corners Analysis Window

So far, in this example, everything in the Virtuoso ® Analog Corners Analysis window has
been predefined, either by the PCF or because it is defined in the Virtuoso ® Analog Design
Environment window. You can also use the Virtuoso ® Analog Corners Analysis window
to revise and add to the predefined information. For example, this section describes how you
might add a Lower value to a scalar measurement before you run the simulation.

The DCgain measurement produces a scalar value. To facilitate analysis, you want to add a
visual indication of the lowest acceptable value to the graphical output of the corners
simulation. To do that, you need to add the appropriate value to the cells in the Performance
Measurements pane.

To add a Lower value by using the Virtuoso ® Analog Corners Analysis window,
1. Click on the Lower cell for the DCgain measurement.
2. Type the value 55 in the cell.

To add a Upper value by using the Virtuoso® Analog Corners Analysis window,
1. Click on the Upper cell for the DCgain measurement.
2. Type the Value 65 in the cell.

All of the measurements produce graphical outputs if you make no further changes, but it
might be useful to have the textual output too. To add the textual output,
➤ Turn on the Textual button in the Outputs column for each of the measurements.

Running the Corners Simulation

After the corners and measurements are defined, running the corners simulation involves
only a couple of simple steps.
1. Ensure that the corners, measurements, and outputs you want to use are selected.
2. Choose Simulation – Run or click Run.
The simulation runs and the outputs you requested appear in display windows.

June 2004 70 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Evaluating Corners Results

The graphical outputs appear in a waveform window.

The phase and gain measurements appear as a family of waveforms in the two subwindows
at the top, with each waveform for each corner. The scalar values, DCgain and bandwidth,
appear as bar charts. It is hard to pick out detail in this combined plot, but you can choose
Window – Subwindows in the Waveform Window to open a dialog box that allows you to
choose which plots you want to look at in more detail.

June 2004 71 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Evaluating Residual Plots

Look first at the DCgain bar chart.

In this plot, the horizontal line in the middle represents the Target value, 60. The bottom line
represents the Lower value, which you set in the Virtuoso ® Analog Corners Analysis
window. The top line represents the Upper value. All of the corners reach the target value. (In
the actual window, each corner displays in a different color so you can determine which
corner is which.) If DCgain for one or more corners fails to reach the target, you might decide
to use a slightly different manufacturing process or to change your circuit so DCgain is
greater.

June 2004 72 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

Evaluating Family-of-Curve Plots

Now consider the family-of-curves plot for the gain measurement.

The fastfast corner produces the highest gain and the slowslow corner produces the
lowest gain throughout the frequency range. You need to determine whether these possible
outcomes are acceptable in your application.

June 2004 73 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Corners Analysis

June 2004 74 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

2
Statistical Analysis

Statistical analysis is a powerful method for estimating parametric yields. The sections in this
chapter explain how you can use the Analog Statistical Analysis option to generate
information about the performance characteristics of the circuits you design.
■ “Getting Started with Statistical Analysis” on page 75
■ “Getting to Know the Analog Statistical Analysis Window” on page 78
■ “Running a Statistical Analysis” on page 85
■ “Analyzing Results” on page 106
■ “Working through an Extended Example” on page 128

Getting Started with Statistical Analysis

This section briefly explains the theory behind statistical analysis, tells you how to get help
and describes how to open the Analog Statistical Analysis window.

How Statistical Analysis Works

The manufacturing variations in components affect the production yield of any design that
includes them. Statistical analysis allows you to study this relationship in detail.

To prepare for a statistical analysis, you create a design that includes devices or device
models that are assigned statistically varying parameter values. The shape of each statistical
distribution represents the manufacturing tolerances on a device. During the analysis, the
statistical analysis option performs multiple simulations, with each simulation using different
parameter values for the devices based upon the assigned statistical distributions.

When the simulations finish, you can use the data analysis features of the statistical analysis
option to examine how manufacturing tolerances affect the overall production yield of your
design. If necessary, you can then switch to different components or change the design to
improve the yield.

June 2004 75 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Data Types Generated by the Statistical Analysis Tool

The Statistical Analysis tool creates two types of output data:

Scalar Data

For each iteration during a statistical analysis, the simulator evaluates explicit expressions
that reduce to a single scalar number. These numbers are stored in a file which will ultimately
be used for data analysis by the user at post-simulation.

The simulator evaluates these scalar expressions during runtime so as to reduce the amount
of generated psf data. For each successive iteration analysis, the simulator typically deletes
the psf data from the previous iteration.

However, it is possible to keep all of the psf data from all of the iterations (Spectre only; see
next data type).

The majority of the statistical analysis tool UI is focused on processing and displaying scalar
data.

For the Spectre® simulator, the process parameters declared in a statistics block in the netlist
are also included in the resulting scalar data file.

Psf Data

This is the same kind of psf data that the simulator typically generates. Usually, this data only
includes the last iteration.

However, for the Spectre ® simulator, the user has the additional option to save the psf data
for all of the iterations. From this data the user can either plot waveforms or regenerate new
scalar data files. Since this data is psf, the user will have to evaluate expressions against this
data to created these waveforms and statistical data. Specific waveform expressions are not
processed during a statistical analysis, only scalar expressions are.

Opening the Analog Statistical Analysis Window

To run the statistical analysis option, you must use a simulator that supports statistical
simulation. In addition, the model and device descriptions of the components that you want
to use in the statistical simulations must have statistical values.

To start the statistical analysis option within the Virtuoso® analog design environment,

June 2004 76 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

1. Set up your simulation normally, choosing an appropriate simulator.

2. Choose Tools – Monte Carlo.

June 2004 77 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Getting to Know the Analog Statistical Analysis Window

The Analog Statistical Analysis window contains the fields and controls required to specify
the statistical analysis that you want to run.

Status display
Menu

Analysis Setup
pane

Outputs pane

Edit fields

Button bar

June 2004 78 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Status Display
The status display shows messages that indicate what the statistical analysis option is doing.
The messages include the following:
■ Simulate
■ Ready
■ Plotting Results
■ Simulate Distributed

During a simulation, the status display also shows which iteration is running and how many
iterations are left to run.
Note: This feature only works outside of distributed processing mode.

Menu
The menu contains the commands needed to prepare for, run and analyze the results of a
statistical analysis.

For guidance on using the menu choices, see the associated cross references:
Statistical Analysis Menu Choices

Menu Item For More Information

Session
Save State “Saving the Session State” on page 101
Load State “Loading a Saved Session State” on page 102
Save Script “Saving the Script” on page 102
Quit “Closing the Analog Statistical Analysis
Window” on page 104
Outputs
Retrieve Outputs “Selecting Signals and Expressions to Analyze”
on page 88

June 2004 79 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Statistical Analysis Menu Choices, continued

Menu Item For More Information

Save All “Saving All Voltages or Currents” on page 95
Simulation
Check Expressions “Checking the Output Expressions” on page 96
Define Correlations “Defining Correlations” on page 98
Create Input Files “Creating Input Files for a Socket Simulator” on
page 101
Run “Starting and Stopping the Analysis” on page 99
Stop “Starting and Stopping the Analysis” on page 99
Output Log “Viewing the Output Log” on page 103
Results
Filter “Filtering Outlying Data” on page 108
Specification Limits “Setting Specification Limits” on page 111
Print
Iteration vs. Value “Printing Iteration versus Value Tables” on
page 114
Correlation “Printing Correlation Tables” on page 116
Plot
Histogram “Plotting Histograms” on page 117
Curves “Plotting Families of Curves” on page 119
Scatterplot “Plotting Scatter Plots” on page 120
Yield
Simple “Obtaining Reports on Simple Yields” on
page 123
Conditional “Obtaining Reports on Conditional Yields” on
page 126
Multiconditional “Obtaining Reports on Multiconditional Yields”
on page 125
Select “Analyzing Results” on page 106

June 2004 80 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Statistical Analysis Menu Choices, continued

Menu Item For More Information

Save “Saving Statistical Analysis Results” on
page 100
Evaluate Expressions “Creating a New mcdata File from Saved
Waveform Data” on page 108
Help
Contents “Opening the Analog Statistical Analysis
Window” on page 76
About Analog Statistical “Opening the Analog Statistical Analysis
Analysis Window” on page 76

Analysis Setup Pane

The fields and selections in the Analysis Setup pane specify the characteristics of the
statistical analysis to be run by the simulator.

June 2004 81 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

For a brief description of the items in the Analysis Setup pane, see the following table. For
more detailed information, see “Specifying the Characteristics of a Statistical Analysis” on
page 86.
Field or Selection Description and Usage
Number of Runs Specify how many simulations to run for this statistical
analysis.
Starting Run # Specify the starting run number.
Analysis Variation Select the type of statistical variation to be used.
Swept Parameter If desired, select temperature or a design variable to
sweep.
Append to Previous Scalar Enable this button to append scalar output data to
Data previously saved scalar data. This feature is not supported
in the distributed processing mode. For the Spectre®
simulator, all of the pertinent UI fields are checked for
compatibility with the existing scalar data set prior to
allowing a Monte Carlo run.
Save Data Between Runs to Enable this button to save the raw output data (the
Allow Family Plots parameter storage format [psf] files) for all the statistical
analysis iterations.
Turn this button off if you want the raw output data to be
deleted before each iteration. In which case, only the psf
data for the last iteration will ultimately be saved. The
ability to append to previous psf data is currently not
supported. This button only appears if you use the
Spectre® simulator.

June 2004 82 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Outputs Pane
The Analog Statistical Analysis window Outputs pane initially lists the expressions and
signals defined in the Outputs pane of the Virtuoso® Analog Design Environment
window.

The columns in the Outputs pane are described in the following table.
Column Description and Usage
Name A field showing the existing name for the expression or signal.
For expressions that evaluate to a scalar, the final name of the
statistical data set will be of the form name_varval. Where
name is the name on this pane and varval is the particular
swept variable value (if none, defaults to temperature).
Expression/Signal A field showing either an expression or a signal name.

June 2004 83 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Column Description and Usage

Data Type A field indicating what type of data will result from an
evaluation of the expression. This field is typically auto set by
the UI, but the user can manually set it as well. The possible
values of this field are:
waveform The expression either evaluates to a waveform or
is a signal. These expressions will not be sent to the simulator
for runtime evaluation, and can only be evaluated against
resulting psf data.
Note: The user scan manually set/override any
expression type to be waveform whenever they do not
want a particular expression sent to the simulator.
scalar The expression evaluates to a scalar value. These
expressions will be sent to the simulator for runtime
evaluation.
unknown There is no currently loaded psf data that can be
used to determine the type of the expression. These
expressions will be sent to the simulator for runtime
evaluation. If there are problems detected with this expression
at runtime, either Spectre will abort or error flag values will be
inserted into the generated data.
Note: The equivalent edit field uses a blank to indicate
unknown.
Autoplot A field that indicates whether to automatically plot the pane
entry after all the statistical analysis simulations finish. If the
element is an expression, it is evaluated against the psf data
and plotted. If the element is a signal, then an individual plot is
created for each primary analysis type found in the psf data.

Edit Fields
These fields, located beneath the Outputs pane, are used to add output signals and to add
or modify output expressions.

June 2004 84 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

For more information, see “Working Directly with Expressions and Signals” on page 89.

Button Bar
The buttons at the bottom of the Analog Statistical Analysis window operate on the Outputs
pane and the edit fields.
Button Description and Usage
Add Click to add a signal or expression defined in the edit fields to
the list of signals and expressions.
Delete Click to delete a signal or expression that is highlighted in the
Outputs pane.
Change Click to replace a signal or expression that is highlighted in
the Outputs pane with whatever is in the edit fields.
Clear Click to clear the edit fields and remove any highlighting in the
Outputs pane.
Calculator Click to open the calculator so that you can build a new signal
or expression in the calculator display buffer.
Get Expression Click to fill the Expression/Signal field with the signal or
expression that is currently in the calculator display buffer.

Running a Statistical Analysis

The major steps involved in setting up and running a statistical analysis are described in the
following sections:
■ “Specifying the Characteristics of a Statistical Analysis” on page 86
■ “Selecting Signals and Expressions to Analyze” on page 88
■ “Defining Correlations” on page 98
■ “Starting and Stopping the Analysis” on page 99

June 2004 85 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

■ “Saving and Restoring a Statistical Analysis Session” on page 101

Specifying the Characteristics of a Statistical Analysis

You specify how a statistical analysis proceeds by filling out the fields in the top pane of the
Analog Statistical Analysis window.

1. Specify the Number of Runs for this statistical analysis.

2. Specify the Starting Run #.
By default, this value is 1. However, if you want to collect the results from several sets of
analyses via the Append to Previous Scalar Data boolean, each subsequent set
should not have any run numbers that overlap previous runs numbers. For example, if
your first analysis has a Starting Run # of 1 and the Number of Runs is 100 then the
Starting Run # for the second analysis needs to be at least 101.
3. Choose the type of Analysis Variation.
The available choices depend on the simulator that you are using, but the default choices
include
❑ Process Only
❑ Mismatch Only

June 2004 86 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

❑ Process Variation and Mismatch

Which choice is most appropriate for your analysis depends on whether you want the
statistically valued parameters to vary independently or to track each other. In general,
the parameters of devices on the same die track each other closely and for purposes of
simulation you might want them to track exactly. In a board-level design however, the
parameters of different devices are likely to vary independently. For more information,
see “How the Statistical Analysis Option Uses the Analysis Variation Setting” on
page 104.
4. If desired, choose a parameter to sweep in an inner loop.
The parameter can be either Temperature or one of the design variables. Choose
None, which is the default, if you do not want to sweep a parameter.
5. Select the Append to Previous Scalar Data button if you want to append the scalar
output data from the current analysis to previously saved scalar data.
For example, to add another 100 runs to an existing set of 100 runs, select this button
and, as discussed in Step 2, set the starting Run # to at least 101.
By default, scalar data is saved in the monteCarlo/mcdata file located at the same level
as the psf directory. If you do not select the Append to Previous Scalar Data button,
new scalar data from the current analysis replaces any existing data in that file.
(Spectre simulator only)
This UI field will only be active when valid scalar results are currently selected/loaded.
If the currently loaded scalar results are not from the current copied into the current ADE
run directory when the next monte carlo simulation is run. As a result any previous
existing scalar results in the run directory will be erased.
Before allowing a simulation in this mode, the UI first checks that the current UI
configuration (form field settings) are compatible with the currently loaded results. This
checking includes:
The run numbers do not overlap the run numbers in the existing data.
The Swept Parameter declaration (including values) is the same as in the existing
data.
The scalar expressions in the output pane are the same as in the existing data.
After a simulation in append mode, the UI checks the new data to insure that the
statistical parameters swept are the same as in the pre-appended data. If any
discrepancies are found, the UI blocks reading/loading in the data.

June 2004 87 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

6. (Spectre simulator only) Select the Save Data Between Runs to Allow Family Plots
button if you want to be able to plot graphs showing the variation of entire waveforms or
if you want to evaluate expressions after the analysis finishes.
Be aware that with this option enabled, the amount of data saved during an analysis can
be very large. To reduce disk storage requirements, avoid saving all voltages and
currents. Instead, select only the specific nodes and terminals referenced by your output
expressions.
The statistical analysis option calculates and saves the results of scalar expressions after
every run, whether Save Data Between Runs to Allow Family Plots is selected or
not.

Selecting Signals and Expressions to Analyze

The Analog Statistical Analysis window Outputs pane initially contains any expressions and
signals defined in the Outputs pane of the Virtuoso® Analog Design Environment window.
You can also retrieve these expressions and signals at any time by choosing Outputs –
Retrieve Outputs. These choices and values from the environment window are often the
most useful ones for a statistical analysis, but, if you want, you can change them. You can
add, modify, or delete expressions and add or delete signals by
■ Using the Direct Plot form, Add To Outputs capability
■ Typing them in directly
■ Using the Calculator

June 2004 88 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Working Directly with Expressions and Signals

You can use the Outputs pane, edit fields, and button bar to add or delete signals and to add,
delete or change expressions.

Adding a Signal or Expression

To add a signal or expression by typing it in:

1. If the edit fields are not empty, click Clear.
2. In the edit fields, type a name for the new expression and the expression itself.
3. Set the Autoplot cyclic field, which is located at the right side of the edit fields, to yes if
you want the new signal or expression to be plotted after all the simulations finish.
4. Click Add.
The new signal or expression is added to the list of signals and expressions.

June 2004 89 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Deleting a Signal or Expression

To delete a signal or expression:

1. Highlight the signal or expression in the Outputs pane.
2. Click Delete.

Changing an Expression

To change an expression:
1. Highlight the expression in the Outputs pane.
The expression appears in the edit fields.
2. Working in the edit fields, make any necessary changes to the name, expression, and
autoplot values.
3. Click Change.

Using the Calculator to Build Expressions

To avoid typing in an expression and signal names, you can build an expression in the
calculator and then import the finished expression into the Analog Statistical Analysis
window.
1. Click Calculator to open the calculator window.
2. Use the buttons and commands in the calculator window to build the expression you
need. Leave the completed expression in the display buffer of the calculator.
3. In the Analog Statistical Analysis window, click Get Expression.

The expression appears in the Outputs pane.

Data Access Function Types

In order to create effective monte carlo expressions, knowledge of the two basic types of data
access functions is critical:

Analysis Alias/Type Dependent Functions:

This type searches for a particular analysis alias or type. The alias or type to be sought
after is hardcoded according to the function name, there is no specific type argument on

June 2004 90 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

these functions. For example, the VF() function will look for a voltage (V) signal in the
first found occurrence of ac data (Frequency), whereas the IT() function will look for a
current (I) in transient data (Time).
Here are examples of VF() expressions that can be declared:
mag(VF("/net2") 1e6) => waveform
value(mag(VF("/net2") 1e6)) => scalar

Most of these functions can be found on the calculator tool.

These wildcard functions access data according to the following flow:

a. First look for a specific aliased data name in the current results. For example, the
VF() function would first look for psf data aliased to the name ’ac. If found, then this
data is retrieved.
For more information on alias names, see the section titled Data Name Aliasing.
b. If (1) does not find an alias name match, then the function reverts to a data type
wildcard search operation. Instead of looking for a name match, the function will now
search for the first occurrence of any psf data with a type that matches the one
related to the function name. For example, the VF() function will now look for a
voltage (V) signal in the first found occurrence of ac type data.

Analysis Name Dependent Functions:

This type searches for a particular analysis name (or alias). These functions require
that a psf data name (or its alias) be specified. The benefit to these functions is that
there is no ambiguity as to which data set will be retrieved. Here are examples of ac
data expressions that can be declared:
mag(v("/net2" ?result "ac") 1e6) => waveform
value(mag(getData("/net2" ?result ’ac) 1e6)) => scalar

All of the Ocean data access functions are name dependent. The most commonly
used functions are getData(), v() and i().

Data Name Aliasing

During the course of using the Statistical Analysis tool, the spectre simulator can create up
to nine distinct data names for each analysis type. The name of data being accessed depends
both on the UI configuration for the statistical analysis run and on when the data is being
accessed (by spectre at runtime or by the UI for post processing/plotting).

As a result, a method to declare a single data name that was the equivalent to the nine above
mentioned data name formats was added. Now, when data is selected/loaded, the data

June 2004 91 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

access code automatically aliases any found version of the nine possible data name formats
to the base name equivalent.

For any currently selected/loaded data the user can see what alias names have been
assigned by typing the following command into the ciw:
setof(ana results() symbolp(ana))

An example response is:

(dc ac output dcOp sp_noise model tranOp tran
dcOpInfo instance noise sp variables)

To see the actual psf names generated, enter:

setof(ana foreach(mapcar (na a) results(?noAlias t) results()
list(a na)) symbolp(car(ana)))

An example response is:

(dc "mc1_dc-montecarlo")
(ac "mc1_ac-montecarlo")
(output "mc1_outputParameter-montecarlo")
(dcOp "mc1_dcOp-montecarlo")
(sp_noise "mc1_sp_noise-montecarlo")
(model "mc1_modelParameter-montecarlo")
(tranOp "mc1_finalTimeOP-montecarlo")
(tran "mc1_tran-montecarlo")
(dcOpInfo "mc1_dcOpInfo-montecarlo")
(instance "mc1_element-montecarlo")
(noise "mc1_noise-montecarlo")
(sp "mc1_sp-montecarlo")
(variables "variables")

In this example, we know that the "mc1_ac-montecarlo" psf data will be accessed for
either of the following functions:
VF("/net2")
getData("/net2" ?result ’ac)

So long as all functions which are analysis name dependent use the proper alias name, the
proper data will be retrieved throughout all phases of using the Statistical Analysis tool.

Note that this aliasing is tightly tied to the analysis names that the ADE netlister generates. If
the user tries to run a standalone netlist with different names, the aliasing may not work
properly. The same goes for declaring analyses via an include file. The Statistical Analysis UI
is only intended for use directly within the ADE product.

June 2004 92 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Results Dir Arguments in Expressions

Many of the data access functions have an optional results directory argument. This
argument must not be set for any expression being used in monte carlo. For more information,
see the Creating Monte Carlo Compatible Expressions section.

Creating Monte Carlo Compatible Expressions

Most of the calculator buttons produce valid expressions for use in the monte carlo flow (this
does not include the results browser button or tool).

Most of the results browser generated data access expressions must be hand edited in order
to conform to the expression syntax requirements of the monte carlo flow.

Most of the ADE->Results->Direct_Plot forms produce the proper expression syntax (via
the Add to Outputs utility). These forms typically produce ocean expressions, but some still
produce old ael style functions.

Before any function/expression will work properly in the Monte Carlo tool, the following Monte
Carlo Expression Syntax Rules should be followed:
1. Whenever an analysis name argument is required, like for Ocean functions, always use
the proper alias name.
2. Never include the current run directory in an expression. When using the results browser
to capture a data access expression in the calculator, always manually delete the run
directory component of the expression.

For example, the results browser may declare something like this in the calculator:
v( "/net4" ?result "ac-ac" ?resultsDir
"/hm/test/simulation/ampTest/spectre/schematic" )

This expression must be manually altered to look like:

v( "/net4" ?result ’ac)

Where ’ac is the standard ac analysis alias name.

For more information on alias names, see the section titled Data Name Aliasing.
The most common symptoms of invalid expressions are:
❑ All of the generated scalar data points for an expression is the same nominal value.
❑ A waveform plot for a waveform expression only plots a single waveform and there
is no iterations parametric information.

June 2004 93 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

❑ All of the scalar data points for an expression is either of the two error flag values (-
1.11111e36 and -2.22222e36).
❑ Spectre aborts via the nominal run expression error checker.

Optimizing Monte Carlo Expressions

Some expressions can require significant evaluation times for the Simulation-
>Check_Expressions and Results->Evaluate_Expressions capabilities of the UI. The bigger
the family of psf data created by spectre, the greater the evaluation times of these two utilities.
The analyses that are mostly affected are tran, sp and the rf ones.

The spectre simulator does not experience this slowdown in performance because it only
operates on a single iteration at any one time. It never processes the entire psf family of
iterations data.

Depending on the scope of psf data generated, the user can get better performance out of an
expression if they simply re-arrange it.

For example, the following syntax:

value(exprA num)/value(exprB num)

is faster than using the simpler expression syntax of:

value(exprA/exprB num)

Some functions with complicated internal calculations compound the amount of time required
for evaluation. A prime example is the s-parameter yp() Ocean function. This function
requires performing internal matrix manipulations. As the data sets that yp() operates on grow
linearly, the matrix evaluation time grows exponentially. As a result, it is best to reduce the
data going into the yp() function as opposed to reducing the data coming out of it. For
example, although the following expression is simplistic:
value(yp(sp(1 1 ?result ’sp) sp(1 2 ?result ’sp)
sp(2 1 ?result ’sp) sp(2 2 ?result ’sp)) 1M)

The user will get much better performance if they re-arrange the expression to be:
yp(value(sp(1 1 ?result ’sp) 1M) value(sp(1 2 ?result ’sp) 1M)
value(sp(2 1 ?result ’sp) 1M) value(sp(2 2 ?result ’sp) 1M))

Note: The YP() ael function is doomed to be slow because it does not allow the user to
reduce the dimension of the input s-parameters.

Therefore, it depends on how far the customer wants to go to optimize their expressions. For
transient and s-parameter data, It is highly recommend that they optimize as much as
absolutely possible.

June 2004 94 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Saving All Voltages or Currents

To save all of the node voltages and terminal currents for later use, perform the following,
1. In the Analog Statistical Analysis window, choose Outputs – Save All.

The options displayed in the form depend on the simulator you use.
2. Select appropriate voltages, currents, or both and click OK.
To reduce the amount of disk space required, select the currents and voltages that
appear in your output expressions and consider selecting individual nets or nodes. For
more information about the Save Options form, see the Virtuoso ® Analog Design
Environment User Guide.
Note: Be aware of the following information about cdsSpice simulations:
❑ In cdsSpice simulations, data for terminals in lower-level schematics is not saved
when you use the Save Options form to save all currents. You must explicitly select

June 2004 95 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

each terminal with the Outputs – To Be Saved – Select On Schematic

command from the menu in the Virtuoso® Analog Design Environment window.
❑ In cdsSpice simulations that include a noise analysis, the system turns the Select
all node voltages and Select all terminal currents options off. If you later
deactivate the noise analysis, the system reactivates the options.
3. Select the kinds of information that you want the analog design environment to print.
Turning off the printing improves the performance of a statistical analysis.

Checking the Output Expressions

After all the desired output expressions are declared, the user should execute the following
utility prior to starting the statistical analysis run:
➤ Simulation -> Check Expressions

This capability is used to both check and classify each expression element in the Outputs
Pane. Each expression is evaluated against the currently selected/loaded psf data. This data
can either be from a previous statistical analysis or from a straight Virtuoso® Analog Design
Environment simulation. The result of each evaluation is assigned to the Data Type field on
the Outputs Pane. The three possible outcomes are:

scalar the expression evaluated to a scalar.

waveform the expression evaluated to at least a waveform.

ERROR the expression failed to evaluate.

If no psf data is currently selected/loaded, then a message is produced.

Saving Signals Used in Output Expressions

For any expression entered into the output pane, it is the users responsibility to ensure that
all the pertinent schematic signals contained in all the expressions will be written to the psf
data. The psf data is always used to calculate the scalar values for each iteration run.

In the event that the user will want to produce waveform plots (see the Plotting Families of
Curves and Changing Waveform Expressions at Post-simulation Time sections) or post
generate the scalar data (see the Creating a New mcdata File from Saved Waveform Data
section) then it is a good practice to save any psf schematic signals that will be needed on a
post-simulation basis.

June 2004 96 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

There are two basic ways to declare which schematic signals should be written to psf by the
simulator:
1. Use the ADE->Outputs->To_Be_Saved->Select_On_Schematic menu capability.
This approach allows the user to click directly on the schematic signals they desire.
Note: This approach is best suited when the user wants to limit the amount of disc space
used for psf data.
2. Use the Monte Carlo Outputs->Save_All form. This is the same form as the ADE one.

The options displayed in the form depend on the simulator you use.

June 2004 97 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Defining Correlations
Often circuit components are correlated. You can model this behavior by defining correlations
in the model files that you use. As described in this section, you can also use the statistical
analysis option to define correlations.
1. To define correlations, choose Simulation – Define Correlations.
The Statistical Device Correlations form appears.

June 2004 98 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

2. Define the correlations that you need.

To type device names… To select devices in the schematic…
a. Type the full schematic names a. Click Select on Schematic.
into the field near the bottom of
the form. b. In the Composer window, select
the devices that you want to be
b. Type the correlation coefficient correlated.
for those devices into the field to
the right of the names. c. When you finish selecting, press
the Escape key.
c. Click Add.
d. In the Statistical Device
Correlations form, type the
correlation coefficient for the
devices you selected into the
field to the right of the names.

e. Click Add.

3. When you have defined all the correlations, click OK to close the Statistical Device
Correlations form.
The statistical analysis option adds the correlations to the netlist by inserting a new
statistics block after the statistical analysis definition.

Starting and Stopping the Analysis

To start the statistical analysis, choose Simulation – Run.

Normally, the analysis stops when all the iterations are complete. In addition, analyses that
use the Spectre simulator stop if errors are found during the nominal simulation that the
Spectre simulator performs.

If you want to stop the analysis before all the iterations are complete, choose Simulation –
Stop. In response, the statistical analysis option
1. Completes the current iteration
2. Stops the analysis
3. Saves the simulation results
4. Plots the results for each signal or expression that has Autoplot set

June 2004 99 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Saving Statistical Analysis Results

After a statistical analysis, the resulting psf and statistical scalar data can be saved.
1. Choose Results – Save.
The Save Results dialog box appears.

2. Type new information, as necessary.

Consider entering a comment to help you identify the data later.
Be aware that this command copies the entire parameter storage format (PSF) directory
structure, which might be very large, to the new location.
3. Click OK.

June 2004 100 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Saving and Restoring a Statistical Analysis Session

This section explains how to create input files, save and reload session states, save scripts,
and quit from the statistical analysis option. For information on starting the tool, see “Opening
the Analog Statistical Analysis Window” on page 76.

Creating Input Files for a Socket Simulator

If you follow the usual procedure of specifying the simulation and then specifying the
statistical analysis that you want to use, the statistical analysis option creates the necessary
input files automatically. However, if you want to hand edit the input files, the statistical
analysis option provides a way for you to do that.
➤ From the Analog Statistical Analysis window, choose Simulation – Create Input Files.
The tool creates the mcrun.s and mcparam files.

Saving the Session State

A session state consists of all the information in the Analog Statistical Analysis window,
including the Analysis Setup and Outputs information.

To save the current state,

1. Choose Session – Save State.
The Monte Carlo Save form appears.

2. Type the name of the file where you want to save the session state.

June 2004 101 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

3. Click OK.

Loading a Saved Session State

To load a saved state,

1. Choose Session – Load State.
The Monte Carlo Load form appears.

2. Type the name of the file that contains the saved session state.
3. Click OK.
The values from the saved session state appear in the Analog Statistical Analysis
window.

Saving the Script

To save the script that is automatically generated during each session,

1. Choose Session – Save Script.

June 2004 102 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Monte Carlo Save Ocean Script form appears.

2. Type the name of the directory and file where you want the script to be saved.
3. Click OK.
The script is saved in the file. For additional information about using the statistical
analysis option with OCEAN, see the OCEAN Reference.

Viewing the Output Log

To open a window that contains the history of the statistical analysis simulations,

June 2004 103 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

➤ Choose Simulation – Output Log.

The simulator updates the Output Log while the simulation runs, so you might find it useful to
have this window open during the simulation.

Closing the Analog Statistical Analysis Window

To end the session and close the Analog Statistical Analysis window,
➤ Choose Session – Quit.

How the Statistical Analysis Option Uses the Analysis Variation Setting
When you run the statistical analysis option from the Virtuoso® Analog Design Environment
window, you use the Analysis Variation cyclic field to select the kinds of variations to be
used during the analysis. However, the connection between the cyclic field choice and how
your models behave is not automatic—you must define your models so that they respond to
the cyclic field choice.

For the Spectre simulator, you use a statistics block to specify model behavior under the
Analysis Variation cyclic field. See the Analysis Statements chapter of the Spectre
Circuit Simulator Reference for more information. The statistics block must be included in
the netlist before the models. For example, you might use a statistics block with the following
contents:
statistics {
process {

June 2004 104 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

vary PiRho dist=gauss std=250

vary PbRho dist=gauss std=40
vary beta dist=gauss std=20
vary rin1 dist=gauss std=10
vary cin dist=gauss std=20p
vary rin2 dist=gauss std=100
vary cloop dist=gauss std=16p
vary rout1 dist=gauss std=30
vary rout2 dist=gauss std=50
}
mismatch {
vary stat dist=gauss std=.01
}
}

The process and mismatch blocks define the variations for the cyclic field choices.

Before the simulator can use the statistics block, you must include the block in the netlist. For
example, assume that you have a single file called statsLib.scs that has three sections:
parameters, statistics, and models. These sections are arranged so that the
definitions all build without errors.

Then, to include statsLib.scs in the netlist,

1. From the Virtuoso® Analog Design Environment window, choose Setup – Model
Libraries.
The Model Library Setup form appears.

2. Type the complete path to the statsLib.scs file and click Add.
3. Click OK.

June 2004 105 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Analyzing Results
You can use the procedures described in the following sections to analyze a set of statistical
analysis results. By default, these procedures operate on the data from a just-concluded
statistical analysis, but you can also analyze saved data from earlier runs. The saved data
can take two forms: either you can use the data in a stored output file or, if the statistical
analysis session that you want to analyze had the Save Data Between Runs to Allow
Family Plots button turned on, you can create a new mcdata output file from the saved
waveform data.

Loading Stored Statistical Analysis Results

To load a stored set of data,
1. Choose Results – Select

June 2004 106 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Select Results form appears.

2. Select the results that you want to load and click OK.

June 2004 107 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Initialize Statistical Analysis Data form appears.

3. Verify the names of the files containing the statistical analysis data. At runtime, the UI
automatically assigns the name mcparam for the Parameter File and mcdata for the
Data File. As a result, there are few situations where a user should try to stray from these
names. If the names appear to be proper, then click OK. If either of the file names are
blank, then the data is not present and the user should click Cancel.

Creating a New mcdata File from Saved Waveform Data

If you select Save Data Between Runs to Allow Family Plots before running a simulation,
then after the simulation you can use the data stored by that command to evaluate
expressions and create a new mcdata file.
1. Ensure that the Outputs pane contains the expressions and signals that you want to use.
2. Choose Results – Evaluate Expressions.

Filtering Outlying Data

When it cannot evaluate an expression, a socket simulator returns the value 1e36. The
Spectre simulator returns -1.1111e36 when it cannot evaluate an expression and returns --
2.2222e36 when an expression evaluates to a waveform instead of a scalar. Outlying data
points with values like these can have a large and misleading effect on a statistical analysis.
To avoid distortions, you can follow steps like the ones below to filter outlying data points from
your data set.

June 2004 108 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

1. Choose Results – Filter.

The Data Filter form appears.

2. Check that Data Filter is set on.

3. Choose how to compute yield statistics.
❑ Filter By data set ignores all measurements for a point if the value of any of the
measurements for that point is outside the filter limits.
For example, if a point has a value of 1e36 on the bandw_27 measurement shown
in the previous figure, the value for the ymax_27 measurement for that point is also
ignored even if the value falls between the upper and lower ranges defined by the
filter.
❑ Filter By point filters an outlying point only from the specific measurement that
recorded the outlying point.
4. Choose how to set the limits for each parameter.
❑ Set By sigma lets you specify how many standard deviations around the mean
value to include.
❑ Set By limits lets you set absolute upper and lower values. The Upper and Lower
values are included in the range of acceptable values; so to exclude an error value
of 1e36, you need to specify a smaller value, such as 1e35.

June 2004 109 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Turning Off Filtering

To turn off data filtering,

1. Choose Results – Filter.
2. In the Data Filter form, set Data Filter to off and click OK.

Saving and Restoring Filter Settings

Whenever new data is selected/loaded into the Monte Carlo UI, the existing Data Filter form
and its settings are destroyed. A new form is recreated based on the new scalar data
information being read in. As a result, any and all settings on the previous Data Filter form
will be lost. Therefore, it is highly recommended that any declarations made on this form be
immediately saved. Especially if one will be running subsequent simulations of selecting or
loading in results.

To save the data filter settings to a file,

1. Choose Results – Filter.
2. In the Data Filter form, click Save.
The Save Data Filter Values form appears.

3. Type a name for the settings file and click OK.

To restore saved data filter settings,

1. Choose Results – Filter.
2. In the Data Filter form, click Load.

June 2004 110 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Load Data Filter Values form appears.

3. Type the name of the settings file and click OK.

Setting Specification Limits

The specification limits define, for each parameter, the range that is considered to be within
tolerance.

To set the specification limits,

1. Choose Results – Specification Limits.
The Specification Limits form appears.

June 2004 111 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

2. Choose how to set the specification limits for each parameter.

❑ Set By sigma lets you specify how many standard deviations around the mean
value to allow.
❑ Set By limits lets you set absolute upper and lower values.

Saving and Restoring Specification Limits

Whenever new data is selected/loaded into the Monte Carlo UI, the existing Specification
Limits form and its settings are destroyed. A new form is recreated based on the new scalar
data information being read in. As a result, any and all settings on the previous Specification
Limits form will be lost. Therefore, it is highly recommended that any declarations made on
this form be immediately saved. Especially if one will be running subsequent simulations of
selecting/loading in results.

To save and restore the specification limits to a file,

1. Choose Results – Specification Limits.
2. In the Specification Limits form, click Save.
The Save Specification Limits form appears.

3. Type a name for the limits file and click OK.

To restore specification limits,

1. Choose Results – Specification Limits.
2. In the Specification Limits form, click Load.

June 2004 112 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Load Specification Limits form appears.

3. Type the name of the settings file that you want to load and click OK.

Generating Plots, Tables, and Reports

To help analyze your results, you can generate the following plots, tables, and reports for your
input and output parameters.
Plots, Tables, and Reports

Plot, Table, or
Description For More Information
Report
Iteration Versus A table showing the value of a “Printing Iteration versus Value
Value parameter at the end of each Tables” on page 114
iteration.
Correlation A table showing the correlation “Printing Correlation Tables” on
coefficients of each parameter with page 116
each of the other parameters.
Histogram A plot showing the number of runs “Plotting Histograms” on
with scalar parameter values that fall page 117
in each range of values.
Family-of-Curves A plot showing the superimposed “Plotting Families of Curves” on
waveforms for all iterations of a page 119
waveform valued expression.
Scatter Plot A plot depicting the relationship “Plotting Scatter Plots” on
between pairs of parameters. page 120
Simple Yield A report showing the individual and “Obtaining Reports on Simple
total yields for all parameters, given Yields” on page 123
the specification limits.

June 2004 113 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Plots, Tables, and Reports, continued

Plot, Table, or
Description For More Information
Report
Conditional Yield A report showing the effect on the “Obtaining Reports on
rest of the outputs of specification Conditional Yields” on
limits on one parameter. page 126
Multiconditional A report showing individual and total “Obtaining Reports on
Yield yields, given that the selected Multiconditional Yields” on
parameters pass the specification page 125
limits test.

Understanding Generated Data Names

The output expression names revealed when viewing statistical results will be slightly different
than that shown on the Outputs Pane of the UI.

This is because each scalar expression could be evaluated across several swept parameter
values (a UI capability), there would be several distinct data sets created. In order to cope
with having many data sets for a single output expression, the generated parameter name for
each of these data sets is comprised of the original expression name and the value of the
swept parameter. Each name will be of the form:
Name_ParamValue

Where Name is the name assigned on the Outputs pane of the UI and ParamValue is the
particular value of the swept parameter used while that data was generated.

If the Swept Parameter on the UI is set to None, then the value for ParamValue is set to the
temperature value.
Note: In Spectre, The parameters declared in a statistics block of the spectre netlist will also
be written to the statistical results. The naming convention for these parameters is the same
as for expressions, where Name is the name of the statistically varied parameter.

Printing Iteration versus Value Tables

To print a table showing the value of a parameter at each iteration,

1. Choose Results – Print – Iteration versus Value.

June 2004 114 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Iteration Versus Value window appears.

2. Select a parameter, or type a parameter name.

3. Choose the output format:
❑ sorted lists the runs by parameter value
❑ unsorted lists the runs in chronological order

June 2004 115 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

For example, the following figure illustrates a sorted run, with the value and run number
for each measurement listed horizontally across the window.

Printing Correlation Tables

A correlation table shows the correlation coefficients of each parameter with each of the other
parameters. The parameters are sorted from most correlated to least correlated for each
combination of parameters.

To print a correlation table,

1. Choose Results – Print – Correlation Table.
The Correlation Table window appears.

2. Specify a minimum correlation value. Pairs of parameters with correlations lower than
this value do not appear in the table.
3. Click OK.

June 2004 116 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Results Display Window appears.

Each row lists the pair of measurements being considered, the mean and standard
deviation of the first measurement, the mean and standard deviation of the second
measurement, and the number of data points included in the calculation.

Plotting Histograms

You can plot four types of histograms:

■ Standard
■ Pass/fail
■ Cumulative line
■ Cumulative box

To plot a histogram,
1. Choose Results – Plot – Histogram.

June 2004 117 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Histogram form appears.

2. Highlight one or more parameters and click Add.

You can drag or Shift-click to select a group of adjacent parameters or Control-
click to select individual parameters.
3. Type a value from 1 to 50 in the Number of Bins field.
4. (Optional) Click Density Estimator to plot a curve that estimates the distribution
concentration.
5. Click OK.

June 2004 118 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Waveform window appears, showing the distribution of parameter values found
during the statistical analysis run. In this example, the curved line is the density estimator
line.

Plotting Families of Curves

A family-of-curves shows the superimposed waveforms generated during all of the statistical
analysis iterations. This kind of plot illuminates the variability introduced in waveform
variables by process and mismatch variations.

In order to plot such curves, the Save Data Between Runs to Allow Family Plots button
in the Monte Carlo Analysis Setup Pane must be turned on prior to running the simulation.

To plot a family-of-curves,
➤ Choose Results – Plot – Curves.

June 2004 119 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Waveform window opens with the overlapping waveforms. Depending on the
number of iterations included, you might or might not be able to read the legend that
identifies each individual waveform.

Plotting Scatter Plots

A scatter plot shows the relationship between pairs of parameters.

To plot a scatter plot,

1. Choose Results – Plot – Scatterplot.

June 2004 120 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The ScatterPlot form appears.

2. Highlight one parameter in the X-axis list and one in the Y-axis list.
3. Click Add.
4. (Optional) Repeat steps 2 and 3 for other parameter pairs.
5. (Optional) Click Best Fit Line to draw least squares fit lines on each scatter plot.
6. Click OK.

June 2004 121 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Waveform window appears with the values for one parameter of each pair on the
vertical axis and the values for the other parameter on the horizontal axis.

Obtaining Reports on Yields

You can compute simple, conditional, and multiconditional yield statistics.

Note: The yield calculations represent parametric yields only and do not include yield
reduction due to defect density or packaging factors.

June 2004 122 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The table below gives an example of a yield calculation. The next few sections use this table
to illustrate the different kinds of yield statistics. In this table, pass or fail indicates whether
that sample passed or failed the specification limit for that parameter.
Table 2-1 Yield Calculation Table Example

p1 p2 p3

s1 pass pass pass

s2 pass fail pass

s3 fail fail pass

Individual yield is the percentage of samples that meet the current specification limits for
each individual parameter. For example, if parameter X has 100 samples of which 80 fall
within the specification limits, the individual yield is 80%. Using the values in Table 2-1 on
page 123, the individual yields are as follows: p1, 66%; p2, 33%; and p3, 100%.

Total yield is the percentage of samples that meet the current specification limits for all
parameters. In Table 2-1 on page 123, of the three samples only sample s1 falls within the
specification limits for all parameters. The total yield, therefore, is 33%.

Multiconditional yield is the individual yield when only the samples for the specified
parameters that are within their respective specification limits are used in the yield
calculation. If the parameter fails the specification limit test, that sample is removed from the
yield calculation. In the example table, the multiconditional yield for p1 with sample s3
removed is as follows: p1, 100%; p2, 50%; and p3, 100%.

Conditional yield is the same as the multiconditional yield except the individual yields are
calculated for each parameter separately.

Obtaining Reports on Simple Yields

The simple yield report shows the individual and total yields for all parameters given the
specification limits. A summary line shows
■ Total yield
■ Product of the individual yields
■ Total sample size

To print a simple yield report,

1. Choose Results – Yield – Simple.

June 2004 123 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Simple Yield form appears.

2. (Optional) Type a percentage to filter out statistics for samples with high yields.
3. Click OK.
The Results Display Window appears.

In this report, yindv stands for individual yield.

June 2004 124 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Obtaining Reports on Multiconditional Yields

The multiconditional yield report shows the individual and total yields when the parameters
you select pass the specification limits test. The subset of all data sets that meet these
specifications is determined, and the yield is calculated from only this subset.

To print a multiconditional yield report,

1. Choose Results – Yield – Multiconditional.
The Multi Conditional Yield form appears.

2. Select parameters by double-clicking in the list.

3. (Optional) To omit cases where the individual and simple yields are similar, type a
percentage in the Suppress field.
4. Click OK.

June 2004 125 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Results Display Window appears.

In this report, condlyd stands for conditional yield.

Obtaining Reports on Conditional Yields

The conditional yield report is similar to the multiconditional yield report, except that the yield
is calculated for each parameter separately. This allows you to quickly view the effects of
conditional yield for several parameters in a single command.

To print the conditional yield report,

1. Choose Yield – Conditional.

June 2004 126 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Conditional Yield form appears.

2. Select parameters by double-clicking in the list.

3. (Optional) To omit cases where the individual and simple yields are similar, type a
percentage in the Suppress Printout for Delta Yields Less Than field.
4. Click OK.
The Results Display Window appears.

June 2004 127 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Working through an Extended Example

This section follows a statistical analysis session in detail, demonstrating how you might use
the Analog Statistical Analysis option to examine the characteristics of a lowpass filter. The
sections explain how to arrange supporting files for the Spectre simulator, run the analysis
and analyze the results.

Lowpass Filter Schematic

The circuit used in this example allows low-frequency signals to pass through and attenuates
high-frequency signals. The capacitor values control the attenuation of the circuit, while the
resistor values control the voltage gain of the signals that pass through the circuit.

The lowpass filter has the following top-level schematic.

June 2004 128 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Descending into the amplifier shows that it has the following schematic.

This amplifier schematic includes several instances of npn and pnp transistors. Each of the
npn transistors is nominally identical. Similarly, each of the pnp transistors is nominally
identical. In reality, however, the attributes of each transistor differ randomly from the
attributes of each of the other transistors. In the following sections of this example, you
explore the effect that random variation in transistors has on circuit performance.

To follow along with this example, go to a working directory and use a command like the
following to copy all the contents of the monteCarlo directory into the working directory.

tar -cvhf - -C <install_dir>/tools/dfII/samples/artist monteCarlo |

tar -xvf -

Then go to the working directory you created, start icms, and continue with the following
steps.
1. In the CIW, choose Tools – Analog Environment – Simulation to open the Virtuoso ®
Analog Design Environment window.
2. In the Virtuoso ® Analog Design Environment window, choose Setup – Design.
When the Choosing Design form appears, select the aExamples library and the
lowpass cell. Click OK.

June 2004 129 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

3. In the Virtuoso ® Analog Design Environment window, choose Session – Load

State to open the Loading State form.
4. In the Loading State form, select monte as the State Name that you want to load. Click
OK.

At the end of this series of steps, the Virtuoso® Analog Design Environment window
looks like this.

June 2004 130 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Model File
This example uses a model file called spectreLib.scs, which contains all the parameters,
statistics, and modeling information that are required.
library monteLib
section param
simulator lang=spectre
parameters PiRho=2500 PbRho=200 npnbeta=145.5 pnpbeta=200
parameters rin1=1000 rin2=5000 rout1=1000 rout2=3000
parameters cin=1.7e-08 cloop=1e-09
parameters mmstat=1 initstat=1
function Rpb(l,w)=(PbRho*l/w)
function Rpi(l,w)=(PiRho*l/w)
endsection param
section stats
simulator lang=spectre
statistics {
process {
vary PiRho dist=gauss std=350
vary PbRho dist=gauss std=50
vary npnbeta dist=lnorm std=.9
vary pnpbeta dist=lnorm std=1.1
vary Rin dist=gauss std=150
vary cin dist=gauss std=20p
vary rin2 dist=gauss std=100
vary Cfb dist=gauss std=.58n
vary rout1 dist=gauss std=30
vary rout2 dist=gauss std=50
}
mismatch {
vary PiRho dist=gauss std=19
vary PbRho dist=gauss std=3.75
vary npnbeta dist=gauss std=4
vary pnpbeta dist=gauss std=6
}
}
endsection stats
section models
simulator lang=spectre
inline subckt npn (C B E S)
parameters brvbe=.6
model mynpn bjt type=npn is=5.771e-17 bf=npnbeta nf=0.9895 vaf=201.6
+ ikf=0.01573 ise=8.976e-18 ne=1.179 br=3.204 nr=0.9944
+ var=27.03 ikr=0.0003047 isc=1.505e-13 nc=1.912 rb=8.706
+ irb=0.001509 rbm=5.833 re=111.8 rc=54.97 xtb=1.5 eg=1.11
+ xti=3 cje=1.983e-12 vje=0.4818 mje=0.2486 tf=0.33e-9
+ xtf=4.359 itf=0.01753 ptf=176.2 cjc=1.749e-12 vjc=0.5989
+ mjc=0.3349 xcjc=0.5 tr=400e-9 cjs=1e-12 vjs=0.75
+ mjs=0.33 fc=0.5 bvbe=brvbe bvce=1
npn (C B E S) mynpn
ends npn

June 2004 131 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

inline subckt pnp (C B E S)

model mypnp bjt type=pnp
+is=1.2e-16 bf=pnpbeta nf=1.00 vaf=26.00
+ikf=70e-06 ise=1.1e-15 ne=2.00 br=13
+nr=1.00 var=10.00 ikr=100e-06 isc=7.0e-15
+nc=2.50 rb=100
+re=15 rc=150 cje=33e-15 vje=740e-03
+mje=330e-03 tf=2.50e-09 xtf=1.00
+itf=2.00e-03 ptf=5.0 cjc=130e-15 vjc=690e-03
+mjc=440.00e-03 xcjc=500.00e-03 tr=5.00e-09 cjs=200e-15
+vjs=590e-03 mjs=440.00e-03 xtb=780e-03 eg=1.200
+xti=1.80 kf=1.60e-15 af=1.00 fc=850.00e-03
pnp (C B E S) mypnp
ends pnp
endsection models
endlibrary monteLib

Notice the lines in the models section of the model file that define the mynpn model.

parameters brvbe=.6
model mynpn bjt type=npn is=5.771e-17 bf=npnbeta nf=0.9895 vaf=201.6

In particular, notice how the bf parameter is defined as npnbeta. The npnbeta value varies
randomly according to the distributions specified in the statistics block. Consequently,
the value of the bf model parameter also varies. So that mismatch, which is also specified
for this parameter, is effective, the model is defined within an inline subckt block. This allows
each instance of the model to have a slight perturbation.

The statistics block defines how parameters vary during the analysis. In this case, each
parameter has either a Gaussian or a log-normal distribution with a deviation specified by the
std parameter. All the parameters vary when process variation is specified and four of them
vary when mismatch is specified.

Run Analog Simulation to Check Setup

Run a simulation via the ADE->Simulation->Run banner element.

This part of this example is used to verify that the ADE setup is correct. Note that the db20
and phase outputs created plots and that the bandwidth and ymax outputs evaluated to
numbers.

In addition, the psf data created by this run will be used later for expression checking on the
Analog Statistical Analysis UI.

June 2004 132 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Specifying the Analysis in the Analog Statistical Analysis Window

At this point in the example, you are ready to use the statistical analysis option.
1. To open the Analog Statistical Analysis window, choose Tools – Monte Carlo. The
outputs defined in the Virtuoso® Analog Design Environment window appear
automatically in the Outputs pane of the Statistical Analysis window.
2. In the Analog Statistical Analysis window, choose Session – Load State. The Monte
Carlo Load dialog appears. Change the Directory entry to ./.
3. Ensure that the File to be loaded is .mcState.il. The Monte Carlo Load dialog should
look like this.

4. Click OK.

June 2004 133 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

5. Make sure the Analysis Setup pane is set correctly. It should look like this:

June 2004 134 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Analysis Setup form is set to perform 100 iterations of the circuit, using both process and
mismatch variations. Turning on Save Data Between Runs to Allow Family Plots makes
it possible to plot families of curves for the two waveform expressions, phase and db20.
Note: The state loaded contained the identical Outputs Pane expressions as was initially
copied over from the ADE Outputs section. Had the state contained different expressions,
then they would have been merged with any existing expressions. The original bandwidth
output name was changed to bandw_27 by the state. However, the actual expression
formulation remained unchanged.
Any unwanted Outputs pane expressions can be deleted.

Checking Expressions Prior to Simulation

Whenever expressions have been added, loaded or changed, it is a good practice to check
them prior to submitting a statistical analysis simulation.

In this example, we will use the ADE simulation data created in a previous step. Invoke
Simulation->Check_Expressions utility of the Analog Statistical Analysis window.

The expression checking utility will automatically verify and assign the proper Data Type field
to each expression. For any expression with an error, the Data Type field will be set to
"ERROR". In this example, there should be no errors.

Running the Statistical Analysis Simulation

To run the statistical analysis,
➤ From the Analog Statistical Analysis window, choose Simulation – Run.
The simulation runs and the outputs for which Autoplot is set to yes appear in display
windows.

When the simulation is finished and was completely successful, the ciw will produce the
following information:
simulation completed successfully.
...
Monte Carlo Simulation completed successfully...

Although this example should run clean, additional information concerning simulation
problems is useful to know. When any problems are encountered throughout the course of
running a statistical analysis, both the ciw and simulator output log offer valuable information.
The following three scenarios reflect the most common problems encountered:
1. For situations where a scalar expression could not be evaluated at a particular iteration,
the following error information is produced in the ciw:

June 2004 135 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Problems encountered during simulation.

Use the Simulation->Output Log menu for more information.
...
Monte Carlo Simulation completed successfully...

When the spectre output log is reviewed, we would see an error similar to:
**** Run Status for Monte Carlo analysis ‘mc1’ ****
Monte Carlo iteration 1 failed.

The two most common reasons for this type of expression evaluation error are:

a. An analysis pertinent to the expression did not simulate. This is typically due to
convergence problems.

b. The data created is not sufficient to satisfy the expression. For example, the phase
at that iteration did not cross the proper threshold needed by the phaseMargin
function.
Note: In this case, monte carlo scalar data was successfully produced. Any iterations
with errors are assigned error flag values (-1.1111e36 or -2.2222e36).
2. For situations where a scalar expression had syntax errors, the following error
information is produced in the ciw:
Problems encountered during simulation.
Use the Simulation->Output Log menu for more information.
...
Monte Carlo Simulation unsuccessful...

When the spectre output log is reviewed, we would see an error similar to:
Error found by spectre during Monte Carlo analysis ‘mc1’.
designParamVals: Error evaluating ocean expression ‘foo=getData("out")’.
Unsuccessfully evaluated export statements (based on return code).
Analysis ‘mc1’ terminated prematurely due to error.

In this case, no monte carlo scalar data was produced.

3. For situations where spectre could not run at all, the following error information is
produced in the ciw:
Problems encountered during simulation.
Use the Simulation->Output Log menu for more information.
Monte Carlo Simulation unsuccessful...
See simulation output log for more information.

When the spectre output log is reviewed, we would see the error.
For example:

June 2004 136 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Error found by spectre during circuit read-in.

"input.scs" 9: Unable to open input file
‘/example/monteCarlo/models/spectreLib.scs2’.
No such file or directory.
spectre terminated prematurely due to fatal error.

In this case, no monte carlo scalar data was produced.

Evaluating Statistical Analysis Results

These results might contain error flag values that can distort the statistical results (see
problem case 1 in previous section). To check for possible error values, do the following:
1. Choose Results – Print – Iteration vs. Value.
2. In the Iteration Versus Value form, select the bandw_27 parameter, make sure that
sorted is selected for the Output Format, and click Apply.
3. In the Results Display Window, check the values at the beginning and end of the list.
4. In the Iteration Verses Value form, select the ymax_27 parameter and click OK.
5. Repeat step 3 for the ymax_27 parameter values.

Analyzing Scalar Data

There are several ways to look at scalar data. As described in the previous section, you can
simply list the data. You can also plot the data in the form of histograms and calculate yields
for the data.

Printing Correlations Table

To determine which statistically swept parameters had the biggest impact on the output scalar
measurements, perform the following:
1. Choose Results - Print - Correlation...

June 2004 137 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

2. Set the Suppress Printout for Correlations Less Than field to 0.25 and click OK.

3. Observe the data printed to the Results Display Window.

From this data we can see that the bandw_27 expression was mostly influenced by the
Cfb_27 parameter, and the ymax_27 expression was mostly influenced by the
npnbeta parameter. We will use this knowledge further along in this example.

Using Histograms

To begin looking at the bandw_27 and ymax_27 data,

1. Choose Results – Plot – Histogram.
The Histogram form appears.
2. Add bandw_27 and ymax_27 to the Plot column.
3. Turn the Density Estimator button on.
4. Click OK.

June 2004 138 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The histograms appear in the Waveform Window.

The ymax_27 distribution shows that the ymax_27 value is almost exactly the same for every
iteration. That value is unaffected by the variations introduced into the circuit by the statistical
analysis.

Plotting Scatter Plots

Scatter plots are very useful for verifying dependencies between different statistical data sets.
Our findings in the Printing Correlation Tables section of this document will help guide this
example.
1. Choose Results - Plot - Scatterplot...
2. Select the bandw_27 entry in the X-axis listbox, and the Cfb_27 entry in the Y-axis
listbox. Click the Add button.

June 2004 139 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

3. Select the npnbeta_27 entry in the Y-axis listbox. Click the Add button. The Scatter
Plot form should look as follows:

June 2004 140 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

4. Click the OK button. The following two scatterplots should be added to the Waveform
Window:

From these two plots we can see that the bandw_27 expression is directly dependent on the
Cfb_27 parameter by observing a straight slanted line of points in the plot. Whereas the
npnbeta parameter has little effect on bandw_27 because there is little order in the plotted
points.

Analyzing Yields

To analyze the yield for this circuit, you first need to define the specification limits for the scalar
parameters. For this example, assume that you can tolerate only 1 sigma of variation.
1. Choose Results – Specification Limits.

June 2004 141 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Specification Limits form appears.

2. Turn Set By sigma on.

3. Type 1 in the Sigma fields for each of the measurements.
4. Click OK.
5. From the Analog Statistical Analysis window, choose Results – Yield – Simple.
The Simple Yield form appears.
6. Set the value of the Suppress Printout for Yields Greater Than field to 98 percent.
7. Click OK.

June 2004 142 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Results Display Window appears.

These results show that only 64 percent of the iterations produced results where both the
bandw_27and ymax_27 were within specification limits.

Analyzing Waveform Data

Two of the outputs in this example, dB20 and phase, are waveform data. Because this
example was simulated with the Spectre simulator and because Save Data Between Runs
to Allow Family Plots was turned on in the Analog Statistical Analysis window, you can
use family-of-curves plots to examine the data.
1. From the Analog Statistical Analysis window, choose Results – Plot– Curves.

June 2004 143 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Waveform Window appears with the statistical analysis results.

June 2004 144 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The preceding view is not very useful because the detail is too small to see, but if you
turn off the phase subwindow and zoom in on the dB20 curves, you see a plot like this.

This plot shows that in the frequency range of about 400 K to 100 M, the dB20 value is
affected by the statistical analysis variations introduced in the pnp and npn transistors. If
this frequency range is critical, you might need to redesign your circuit so that the
variation is smaller in this range.

June 2004 145 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Changing Waveform Expressions at Post-simulation Time

At post-simulation time, waveform expressions can be created, deleted or altered to produce
new plots, so long as the pertinent psf data has been saved.

Note that it is not necessary to declare any waveform expressions in the Outputs Pane prior
to running a statistical analysis simulation. However, doing so facilitates the waveform
autoplot feature.

Remember, whenever attempting to plot a family of waveforms, make sure the Save Data
Between Runs to Allow Family Plots boolean is on prior to running the statistical analysis.
It is also necessary that all the needed circuit outputs are declared.

The previous plotting example involved zooming into the dB20 curves to see a particular
region. Now lets change the dB20 expression to only include the region of concern by
following these steps:
1. In the Outputs Pane window, select the dB20 entry.
2. Go to the editable expression string field and change the expression to:
sample(dB20(VF("/OUT")) 100K 100M "log" 20)

Click the Change button.

3. Invoke the Results->Plot->Curves capability.

June 2004 146 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Note that the dB20 plot now shows the desired region without having to manually zoom
in:

Changing Scalar Expressions at Post-Simulation Time

At post-simulation time, scalar expressions can be created, deleted or altered to produce new
statistical data, so long as the pertinent psf data has been properly saved. It is required that
the Save Data Between Runs to Allow Family Plots boolean was on at pre-simulation
time. It is also necessary that all the needed circuit outputs were declared.

Prior to running a statistical analysis, so long as the Save Data Between Runs to Allow
Family Plots boolean is on it is not necessary to declare any scalar expressions in the
Outputs Pane. However, the performance of the Analog Statistical Analysis tool is at its
optimum when all scalar expressions are declared and checked prior to simulation time.

June 2004 147 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

In this example, we are going to add a new scalar to the statistical data by doing the following
steps:
1. In the Outputs Pane window, select the clear button.
2. In the name field, type in gn_mrgn.
3. In the expression field, type in:
"gainMargin(VF("/OUT"))"

4. Set the Data Type to blank (i.e. unknown).

5. Click the Add button. The Outputs pane of the UI should now look like the following:

6. Invoke the Simulation->Check_Expressions capability. The Data Type of this

expression should then get automatically set to scalar. If not, then you will need to
correct/change the expression and try again.
7. Invoke the Results->Evaluate_Expressions capability, and click OK on the Evaluate
Expressions sub-form.

As a result of these actions, the new ph_marg statistical data has been created without the
need to re-simulate. Lets investigate this new data:

June 2004 148 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

a. Invoke the Results->Print->Iteration_Verses_Value... capability.

b. On the Iteration Verses Value form, note that the gn_mrgn_27 entry now appears
in the listbox. Select the gn_mrgn_27 entry.

June 2004 149 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

c. Set the Output Format to sorted. and click OK. The following table will be printed
to the Results Display Window:

June 2004 150 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

June 2004 151 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Note that iteration number 59 is an error flag value. If you look at the phase plot created
from a previous section of this example, you will notice that one of the plots does not dip
below -180 degrees. If you put the mouse curser on that single waveform, you can see
the banner of the waveform tool indicating that the waveform is iteration number 59. For
that waveform, the gainMargin() function cannot evaluate properly. The user would have
to use the UI filter capabilities to remove this data point before analyzing this data.

Appending More Scalar Iterations to Existing Data

In this example, we are going to append more statistical iterations onto the existing scalar
data.
Note that this mode will erase the existing psf data for the first 100 iterations. It the user
wishes to save this psf data for later use, then they should use the Results->Save
capability prior to the next simulation. Although the Analog Statistical Analysis UI
does not facilitate appending psf data for waveform plotting, the user can achieve this
operation by using the data access capabilities outside the Analog Statistical Analysis
UI. More on this subject later. For now, lets save the data:

a. Invoke the Results->Save capability.

b. Set the Save As field to first_100_iterations.

c. Click the OK button.

Now lets proceed with appending an additional 100 runs to the existing scalar data. Set
up the UI as follows:

a. Set the Starting Run # to 101.

b. Turn on the Append to Previous Scalar Data boolean.

June 2004 152 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

The Analog Statistical Analysis form should appear as follows:

c. Invoke Simulation->Run.
When the simulation completes, you will notice that the autoplotted histograms indicate
a sample set of 200 ("N = 200"). You will also notice that the autoplotted waveforms are

June 2004 153 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

for iterations 101 to 200 when you drag the curser across the waveforms and look at the
banner info.
If you use the Results->Iteration_Verses_Value capability, you will notice that the
iterations are from 1 to 200 (use the unsorted mode).
Note, at this point, attempting to use the Results->Evaluate_Expressions capability
will cause the first 100 iterations to be purged from the scalar data. This is because the
psf data does not contain the first 100 iterations. Whenever scalar iterations data will be
purged, the UI will inform the user prior to purging the data. The user will be able to abort
creating the new scalar data.

Appending Waveforms From Different Statistical Analysis Runs.

The steps in this section outline how the user can overlay waveform plots from two different
analog statistical analysis psf data sets. We are assuming we are continuing from the
previous section. Perform the following steps:
1. Close all open Waveform Windows.
2. On the Analog Statistical Analysis UI Outputs Pane, select the desired expression row.
3. Go to the editable expression field and highlight the entire expression.
4. Bring up the Calculator and paste the expression into the buffer. Push the erplot button.
5. On the Analog Statistical Analysis UI, invoke the Results->Select capability. Set the
result name to first_100_iterations and click OK. Click OK on the Initialize Statistical
Analysis Data form.
6. On the calculator, push the plot button.

Now, all 200 waveforms should be shown in a single plot.

June 2004 154 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

Performing a Swept Parameter Statistical Analysis.

An example UI setup for performing a Swept Parameter statistical analysis is shown in the
following picture.

June 2004 155 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

June 2004 156 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

This section will not go into showing all the previously demonstrated UI features. Using the
parametric data sets created by a run with the above UI configuration, please return to the
section titled Evaluating Statistical Analysis Results and repeat the same example
through to the end of this document.

June 2004 157 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Statistical Analysis

June 2004 158 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

3
Optimization

Optimization is the process of automatically modifying design variables so that

specifications are achieved. The tool that performs optimization is called the optimizer. Often
the optimizer can take a design that is close to meeting performance specifications and
generate new component values that bring the design into the acceptable performance
range.

You can apply optimization profitably in a wide range of activities.

■ If you use a top-down design approach, you can optimize a circuit block to match the
performance characteristics of an analog HDL module.
■ Using the opposite approach, you can optimize a macro or behavioral module to
describe the behavior of a circuit block. Then, instead of simulating with the circuit block,
you can simulate with the macro or behavioral model, which usually runs much faster.
■ You can optimize model parameters to match measured device data under various
conditions.
■ You can use optimization to address radio frequency (RF) design problems, such as
impedance matching.
■ To increase circuit yield, you can use optimization to achieve better design center values.
■ You can use optimization to match the frequency response of a filter to the specifications
for the filter.
■ You can use optimization to balance design tradeoffs.

The sections in this chapter explain how you can use the optimizer to help achieve your
design goals.
■ “Getting Started with Optimization” on page 160
■ “Getting to Know the Virtuoso® Analog Circuit Optimization Option Window” on
page 163
■ “Running an Optimization” on page 168

June 2004 159 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

■ “Saving, Changing, and Loading Session Information” on page 193

■ “Working through an Extended Example” on page 198

Getting Started with Optimization

This section briefly explains the theory behind optimization, tells you how to get help, and
describes how to open the Virtuoso®™ Analog Optimization Analysis window.

How Optimization Works

A circuit, as originally designed, often fails to meet its specifications. For example, the
bandwidth might be too small or the design might be off center so that the yield is low. You
might be able to improve a marginal design by using components with different values, but
determining the best values to use is often difficult. The optimizer can provide you with
information that can help you choose values that meet your design goals.

To use the optimizer, you specify initial values for a set of design variables. You also specify
the goals you want the circuit to meet. The optimizer first determines how the values of the
goal expressions vary as a function of changes to the design variables. Then the optimizer
changes the design variables in a manner expected to move the values of the expressions in
the direction of the goals. After the change, the optimizer simulates the circuit to check the
outcome. If stopping criteria are not met, the optimizer iterates through the optimization
process.

The following steps describe, in greater detail, the process that the optimizer follows during
an analysis.
1. The optimizer runs a simulation using the initial values you specify for the design
variables.
This step determines the types and initial values of the goal expressions.
2. The optimizer determines which optimization algorithm to use (unless you specify which
to use).
❑ The LSQ (least square) algorithm is best suited for optimizing measured, noisy,
unconstrained data. For example, this algorithm is appropriate for designing a filter
with an output waveform that matches measured frequency response data.
❑ The CFSQP (C version Feasible Sequential Quadratic Programming) algorithm is
suited for a wide variety of optimization problems, including constrained and
unconstrained, minimizing and maximizing, and sequentially related goals. For
example, this algorithm is appropriate for a low noise amplifier design that has many

June 2004 160 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

goals such as maximizing the gain, minimizing the noise, and maintaining a phase
margin greater than 45 degrees.
3. If the CFSQP algorithm is used, the optimizer runs a simulation to determine whether the
initial values are feasible for the given goals. If the initial values are not feasible, the
optimizer computes new values that are feasible.
4. The optimizer determines how sensitive the goal expressions are to each design
variable.
To determine these sensitivities, the optimizer changes each design variable slightly and
then simulates the design again. This technique is called Finite Difference
Perturbation. In this technique, users generally do not need to select which algorithm
to use for their problem. The optimizer decides which algorithm to use based on the type
of optimization to be done. However, this option provides the ability to force the optimizer
to use a particular algorithm.
5. Using the information on sensitivities, the optimizer calculates a new set of values for the
design variables.
6. The optimizer sets the design variables to the new values and simulates the circuit.
If the values of the goal expressions are not better than they were with the previous
design variable values, the optimizer repeats Step 5.
If the values of the goal expressions are better than they were with the previous design
variable values, the new values become the initial values for the next iteration.
7. If the stopping criteria are not yet met, the optimizer begins the next iteration with Step 4.
Optimization stops when either or both of the following stopping criteria are met:
❑ The values of the design variables change very little or not at all
❑ Further changes to the design variables result in no progress toward the goals

Getting Help
For the most extensive information about using the optimizer, continue reading this
document. To open this document online, choose Help – Contents in the Virtuoso® Analog
Circuit Optimization Option window menu.

For information about the precise product name and version for this tool, choose Help –
About Analog Circuit Optimization.

June 2004 161 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Opening and Closing the Virtuoso® Analog Circuit Optimization Option

Window
When your circuit is ready to optimize,
1. Set up a simulation for it in the usual way.
2. In the Virtuoso® Analog Design Environment window, choose Tools – Optimization.

To close the Virtuoso® Analog Circuit Optimization Option window,

➤ Choose Session – Quit.

June 2004 162 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Getting to Know the Virtuoso® Analog Circuit

Optimization Option Window
The Virtuoso® Analog Circuit Optimization Option window contains the primary controls
and tables you need for an optimization session.

Status display
Menu

Goals pane

Variables pane

Tool bar

Status Display
The status display shows the current state of the optimizer. For example, the status display
indicates whether the optimizer is simulating, optimizing, or in some other phase of the
analysis.

June 2004 163 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Menu
The menu contains the commands needed to prepare for, run, and plot the results of an
optimization.

For guidance on using the menu choices, see the cross-references in the following table.
Optimization Menu Selections

Menu Item For More Information

Session
Save State “Saving the Session State” on page 193
Load State “Loading a Saved Session State” on page 194
Save Script “Saving a Script” on page 195
Options “Changing Optimization Options” on page 195
Reset “Deleting All Setup Information” on page 197
Quit “Opening and Closing the Virtuoso® Analog Circuit
Optimization Option Window” on page 162
Goals
Retrieve Outputs “Using Simulation Outputs as Goals” on page 169
Add “Creating a New Goal by Entering It Directly” on
page 170 and “Creating a New Goal by Using the
Waveform Calculator” on page 172
Edit “Editing a Goal” on page 175
Delete “Deleting a Goal” on page 176
Enable “Enabling or Disabling a Goal” on page 176
Disable “Enabling or Disabling a Goal” on page 176

June 2004 164 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Optimization Menu Selections, continued

Menu Item For More Information

Variables
Add/Edit “Adding a Design Variable” on page 184 or “Editing a
Design Variable” on page 185
Delete “Deleting a Design Variable” on page 186
Enable “Enabling or Disabling a Design Variable” on page 187
Disable “Enabling or Disabling a Design Variable” on page 187
Optimizer
Run “Running the Optimizer” on page 187
Step “Running the Optimizer” on page 187
Run n “Running the Optimizer” on page 187
Stop “Stopping the Optimizer” on page 188
Stop Now “Stopping the Optimizer” on page 188
Reset “Deleting Simulation Results” on page 189
Results
Plot History “Plotting Output Data” on page 192
Set Plot Options “Setting the Plotting Options” on page 189
Update Design “Updating Your Design” on page 192
Help
Contents “Getting Help” on page 161
About Analog Circuit “Getting Help” on page 161
Optimization

June 2004 165 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Goals Pane
The Goals pane displays information about the currently defined goals.

To define or revise the goals, you use the menu choices and buttons in the Virtuoso® Analog
Circuit Optimization Option window. For a description of the items in the Goals pane, see the
following table.
Item Description and Usage
Name The name associated with the goal.
Direction One of the following: maximize, minimize, match, >=, or <=.
Target If Direction is match, a value or waveform that the optimizer
attempts to match.
If Direction is maximize or minimize, a value or waveform used to
determine how important the goal is.
If Direction is >=, a value or waveform that is the lower bound.
If Direction is <=, a value or waveform that is the upper bound.
Initial The value of the goal expression as calculated from the initial values
of the design variables.
Prev The value of the goal expression as calculated from the values of the
design variables used in the previous iteration.
Current The value of the goal expression as calculated from the current
values of the design variables.
Enabled Either yes or no. yes indicates that the goal is included in the
current optimization; no, that it is not.

June 2004 166 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Variables Pane
The Variables pane displays information about the current design variables.

To define, revise, or enable the variables, you use the menu choices and buttons in the
Virtuoso® Analog Circuit Optimization Option window. For a description of the items in the
Variables pane, see the following table.
Item Description and Usage
Name The name of the design variable.
Min The minimum value allowed for the variable.
Max The maximum value allowed for the variable.
Initial The initial value of the design variable.
Prev The value of the design variable used in the previous iteration.
Current The value of the design variable used in the current iteration.
Enabled Either yes or no.
yes indicates that the value of the design variable can be changed
in the current optimization; no, that it cannot be.

June 2004 167 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Tool Bar
The tool bar contains buttons that perform the most important optimization tasks. The buttons
are arranged from top to bottom in the order they are typically used.

Add/Edit Goals

Add/Edit Variables

Delete

Run Optimizer

Stop Optimizer

Plot History

Update Design

Running an Optimization
The following sections describe the major steps involved in setting up and running an
optimization.
■ “Defining Goals” on page 169
■ “Preparing Design Variables” on page 184
■ “Controlling the Optimizer” on page 187
■ “Plotting Results” on page 189

You might be able to skip the first two steps by loading a state you saved in an earlier session.
For more information, see “Loading a Saved Session State” on page 194.

June 2004 168 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Defining Goals
Before you can run the optimizer on a circuit, you must specify the goals for the analysis. A
goal consists of
■ An expression whose value can be determined by simulation
■ A direction specifying how the value of the expression is to change during optimization

For example, you might define a goal called Bandwidth(3dB) with the expression
bandwidth(VF("/out") 3 "low")

and specify that the value of the expression is to be maximized during optimization.

The following sections describe how to create, edit, delete, enable, and disable goals.
■ “Using Simulation Outputs as Goals” on page 169
■ “Creating a New Goal by Entering It Directly” on page 170
■ “Creating a New Goal by Using the Waveform Calculator” on page 172
■ “Editing a Goal” on page 175
■ “Deleting a Goal” on page 176
■ “Enabling or Disabling a Goal” on page 176

For information about how the optimizer assigns weights to goals or about creating waveform
objects, look also at
■ “Creating Waveform Objects from a List of Values” on page 176
■ “How the Optimizer Uses Target and Acceptable Values” on page 179

Using Simulation Outputs as Goals

The optimizer allows you to use the simulation outputs defined in the Virtuoso® Analog
Design Environment window as goals. This approach is particularly useful for RF analyses
where expressions are often developed for Direct Plot. You can easily add these Direct Plot
expressions to the Virtuoso® Analog Design Environment window Outputs pane, and from
there you can use the following steps to retrieve the expressions for use as optimizer goals.
1. From the Virtuoso® Analog Circuit Optimization Option window, choose Goals –
Retrieve Outputs.

June 2004 169 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The expressions (but not any signals) defined in the Outputs pane of the Design
Environment window appear in the Goals pane of the Optimization Option window.
Initially, the new goals are not enabled.
Unnamed expressions are given names like G1, G2, G3, and so on.
2. Highlight a goal and choose Goals – Edit.
The Editing Goals window appears.
3. Finish defining the goal by following the instructions in “Editing a Goal” on page 175.
Until you edit a retrieved goal in the Editing Goals window, choosing Goals – Retrieve
Outputs updates the goal to match the existing expression in the Outputs pane of the
Design Environment window. After you edit a retrieved goal, choosing Goals – Retrieve
Outputs has no effect on the goal.

Creating a New Goal by Entering It Directly

To create a new goal by entering it directly,

1. Choose Goals – Add or click Add/Edit Goals.
The Adding Goals form appears.

2. Type a name for the goal.

June 2004 170 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

3. Type a Cadence® SKILL language expression describing the goal.

The expression can be either a scalar expression or a waveform expression.
4. In the Direction cyclic field, indicate how the value of the expression is to change during
optimization.
5. Type an expression in the Target field.
If the expression you enter in Step 3 is a scalar expression, Target must also be a scalar
expression. If the expression you enter in Step 3 is a waveform, Target can be either a
scalar expression or a waveform expression.
6. Specify a value in the Acceptable field.
As described in “How the Optimizer Uses Target and Acceptable Values” on page 179,
the optimizer uses the Acceptable value to determine how important the goal is.
There are two ways to specify the Acceptable value.
❑ You can type an expression in the Acceptable field.
If the expression you enter in Step 3 is a scalar expression, Acceptable must also
be a scalar expression. If the expression you enter in Step 3 is a waveform,
Acceptable can be either a scalar expression or a waveform expression.
A scalar Acceptable expression must meet the following requirements. A waveform
Acceptable expression must meet the following requirements at every point along
the curve.
If you specify
Then the value of the Acceptable expression
Direction as
minimize Must be greater than the value of the Target
expression
maximize Must be less than the value of the Target expression
match Can be any value except the Target expression
In addition, a waveform Acceptable expression must
be everywhere greater than or everywhere less than
the Target expression.
<= Must be greater than the value of the Target
expression
>= Must be less than the value of the Target expression

June 2004 171 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

❑ If % within Target is turned on, you can specify a scalar or waveform percentage
in the Acceptable field. A small percentage indicates that the goal is to be heavily
weighted.
If the expression you enter in Step 3 is a scalar expression, the percentage in the
Acceptable field must also be a scalar expression. If the expression you enter in
Step 3 is a waveform, the percentage in the Acceptable field value can be either a
scalar expression or a waveform expression.
By specifying a scalar percentage, you can ensure that the optimization results are
consistent at both very small and very large values of the Target expression. By
specifying a waveform percentage, you can explicitly specify the importance of each
segment of a waveform goal.
How the optimizer uses the Target and Acceptable expressions depends on the
direction you choose. For details, see “How the Optimizer Uses Target and Acceptable
Values” on page 179.
7. If you want to include the goal in the current optimization, be sure Enabled is on.
8. Click OK.
The new goal is added to the Virtuoso® Analog Circuit Optimization Option window.

Creating a New Goal by Using the Waveform Calculator

To create a new goal by using the Waveform Calculator,

1. Choose Goals – Add or click Add/Edit Goals.

June 2004 172 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The Adding Goals form appears.

2. Type a name for the goal.

3. Click Open.
The calculator window appears.
4. Build the goal expression in the Waveform Calculator. For information on using the
Waveform Calculator, see the Waveform Calculator User Guide.
5. In the Virtuoso® Analog Circuit Optimization Option window, click in the Expression
field.
6. Click Get Expression, which retrieves the expression from the Waveform Calculator
and places it in the Expression field.
7. In the Direction cyclic field, indicate how the value of the expression is to change during
optimization.
8. Type an expression in the Target field (or click in the Target field, then click Get
Expression to retrieve an expression from the Waveform Calculator).
If the expression you retrieve in Step 5 is a scalar value, Target must also be a scalar
expression. If the expression you retrieve in Step 5 is a waveform, Target can be either
a scalar expression or a waveform expression.

June 2004 173 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

9. Specify a value for Acceptable. As described in “How the Optimizer Uses Target and
Acceptable Values” on page 179, the optimizer uses the Acceptable value to determine
how important the goal is.
There are two ways to define this value.
❑ First, you can type an expression in the Acceptable field (or click in the Acceptable
field, then click Get Expression to retrieve an expression from the Waveform
Calculator).
If the expression you retrieve in Step 6 is a scalar expression, Acceptable must also
be a scalar expression. If the expression you retrieve is a waveform, Acceptable
can be either a scalar expression or a waveform expression.
A scalar Acceptable expression must meet the following requirements. A waveform
Acceptable expression must meet the following requirements at every point along
the curve.
If you specify
Then the value of the Acceptable expression
Direction as
minimize Must be greater than the value of the Target
expression
maximize Must be less than the value of the Target expression
match Can be any value except the Target expression
In addition, a waveform Acceptable expression must
be everywhere greater than or everywhere less than
the Target expression.
<= Must be greater than the value of the Target
expression
>= Must be less than the value of the Target expression

❑ Second, if % within Target is turned on, you can specify a scalar or waveform
percentage in the Acceptable field. A small percentage indicates that the goal is to
be heavily weighted.
If the expression you retrieve in Step 6 is a scalar expression, the percentage you
enter in the Acceptable Field must also be a scalar expression. If the expression
you retrieve is a waveform, the percentage value can be either a scalar expression
or a waveform expression.
By specifying a scalar percentage, you can ensure that the optimization results are
consistent at both very small and very large values of the Target expression. By

June 2004 174 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

specifying a waveform percentage, you can explicitly specify the importance of each
segment of a waveform goal.
How the optimizer uses the Target and Acceptable values depends on the direction you
choose. For details, see “How the Optimizer Uses Target and Acceptable Values” on
page 179.
10. If you want to include the goal in the current analysis, be sure Enabled is on.
11. Click OK to add the new goal to the Virtuoso® Analog Circuit Optimization Option
window.

Editing a Goal

To edit an existing goal,

1. In the Virtuoso® Analog Circuit Optimization Option window, highlight the goal you want
to edit.
2. Choose Goals – Edit or click Add/Edit Goals.
The Editing Goals form appears.

Except for the title, this form is identical to the Adding Goals form and you use the two
forms in the same way.

June 2004 175 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

3. Make the changes you want to make.

For details, see “Creating a New Goal by Entering It Directly” on page 170 or “Creating
a New Goal by Using the Waveform Calculator” on page 172.
4. Click OK.
The changes are applied to the highlighted goal.

Deleting a Goal

To delete an existing goal,

1. In the Virtuoso® Analog Circuit Optimization Option window, highlight the goal you want
to delete.
2. Choose Goals – Delete or click Delete.
The highlighted goal disappears from the Virtuoso® Analog Circuit Optimization Option
window.

Enabling or Disabling a Goal

For a quick way to enable or disable a goal,

1. Highlight the goal in the Virtuoso® Analog Circuit Optimization Option window.
2. Choose Goals – Enable or Goals – Disable.
The optimization does not include disabled goals.

Creating Waveform Objects from a List of Values

When the expression that defines a goal is a waveform, the Target and Acceptable values
you define can also be waveforms. This section describes how you can create a waveform
object from a list of values stored in a file. For information on some of the other ways you can
create a waveform, see the Waveform Calculator User Guide.

This scenario assumes you have a list of X and Y values stored in a file. The X data values
must be monotonically increasing. For example, you might have a file called mydata
containing the following information:
; The information in column 1 is for the X axis.
; The information in column 2 is for the Y axis.
0.5 5
0.6 5.1

June 2004 176 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

1.0 4.8
1.2 4.5
2.0 4.5
2.5 4.7
3.3 4.9
3.9 5.1
4.3 5.2
5.0 5.4

To convert this data into a waveform object,

1. Open the Waveform Calculator.
2. Click Special Functions.
3. Choose table from the list of functions.
The Table form appears.

4. In the File Name field, type the filename of the file that contains your data.
5. In the Function Name field, type the name you want the waveform object to have.
6. (Optional) Specify which columns contain the X and Y data. This step is not required if
the X data is in column 1 and the Y data is in column 2.
7. (Optional) Type the number of lines to skip in each column before reading the data. Do
not count comment lines, which begin with a semicolon, and blank lines in the number of
lines to skip.
8. Click OK.
The Table form closes and the wave object appears in the calculator buffer. For example,
filling in the Table form as shown above and clicking OK causes
myWaveObject()

June 2004 177 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

to appear in the calculator buffer.

9. (Optional) In the Waveform Calculator, click plot to plot the waveform.
The Waveform Window opens with a plot of the waveform. The myWaveObject()
waveform, for example, looks like this.

June 2004 178 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

You can use the new waveform object in the optimizer wherever waveforms are valid. For
example, to use myWaveObject() as a target, type the name in the Target field.

How the Optimizer Uses Target and Acceptable Values

The optimizer uses a Target value in two ways: as a goal and as an indication of the weight
(importance) of the goal. The following sections describe each use in more detail.

Target Values Used as Goals

The first use, as a goal, is most obvious when you specify a Direction of match, >=, or <=.
In these cases, the target is the value the optimizer attempts to reach.
When you specify a The optimizer attempts to make the
Direction of value of the goal expression
match Match the Target value exactly
>= Greater than the Target value
<= Less than the Target value

However, when you specify a Direction of minimize, the optimizer does not stop minimizing
the goal expression when the Target value is reached. In fact, the optimizer makes the goal

June 2004 179 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

expression as small as the optimization stopping criteria allow, even if that means the final
value is much less than the Target. Similarly, when you specify a Direction of maximize, the
optimizer makes the goal expression as large as possible, even if that means the value is
much greater than the Target value.
When you specify The optimizer attempts to make the value of the goal
a Direction of expression
minimize As small as possible, regardless of the Target value
maximize As large as possible, regardless of the Target value

Note: The LSQ algorithm makes no distinction among the directions of match, minimize,
and maximize. In each of these cases, the LSQ algorithm works to match the Target value.
For more information, see the information about the LSQ algorithm in “Changing Optimization
Options” on page 195.

Target Values Used to Assign Weights

When you set a Direction of minimize or maximize, the second use of the Target value is
most obvious. In these two cases, the Target value, together with the Acceptable value, is
used only to assign a weight to the goal. By contrast, in the match, >=, and <= cases, the
Target value is used both to set a goal and assign a weight.

The optimizer assigns greater weight to a goal that is defined with an Acceptable value set
very close to the Target value. Similarly, if the Target and Acceptable values are waveforms,
the optimizer assigns more weight to points where the Target and Acceptable values are
close together, and less weight to points where the Target and Acceptable values are farther
apart.

In formal terms, the weight assigned to a goal depends on the current value of the goal
expression( f ), the Target value ( T ), and the Acceptable value ( A ).

f –T
weight = --------------
A–T

For example, assume you have two goals defined as follows.

Name Direction Target Value Acceptable Value
power <= 50 mW 80 mW
delay <= 50 ns 60 ns

June 2004 180 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

If the current value of the power expression is 90 mW, its weight is

90mW – 50mW 40mW

------------------------------------------ = ------------------ = 1.333
80mW – 50mW 30mW

If the current value of the delay expression is 90 ns, its weight is

90ns – 50ns 40ns

-------------------------------- = ------------- = 4
60ns – 50ns 10ns

Because the delay weight is greater than the power weight, the optimizer assigns greater
importance to reducing delay than it does to reducing power. In a tradeoff between the two,
reducing delay comes out ahead.

The next example illustrates how the optimizer determines weights from waveform Target
and Acceptable values. In this example, you match the output of a filter to a particular
waveform. Parts of the range are critical, so you assign a heavy weight to those sections by
defining very similar Target and Acceptable waveforms. The middle section is not critical,

June 2004 181 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

so you assign less weight to that area by defining Target and Acceptable waveforms that
are farther apart.

Critical area, heavily weighted

Not critical, lightly weighted

Gain

Target waveform

Acceptable waveform

Frequency

The final example illustrates how the optimizer determines an Acceptable waveform from a
waveform Target and a waveform % within Target value. In this example, Direction is

June 2004 182 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

specified as >= and the calculated Acceptable waveform is everywhere less than the Target
waveform.

200 Target waveform

100
Acceptable waveform

10%

Heavily Lightly
5% weighted weighted
% within target waveform

0%

At the left side of the plot, the % within target waveform has the value 1, so the calculated
value of the Acceptable waveform in that region is
100-(0.01 * 100) = 99

At the right side of the plot, the % within target waveform has the value 10, so the calculated
value of the Acceptable waveform in that region is
200-(0.10 * 200) = 180

June 2004 183 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Preparing Design Variables

Before you can run the optimizer on a circuit, you must specify which design variables the
optimizer is allowed to change. The design variables you specify must be simulation
environment variables such as component parameters and device model parameters. Typical
examples include variables for resistor, capacitor, and inductor values or for device widths,
lengths, and areas.

The following sections describe how to add, edit, delete, enable, and disable design variables.
■ “Adding a Design Variable” on page 184
■ “Editing a Design Variable” on page 185
■ “Deleting a Design Variable” on page 186
■ “Enabling or Disabling a Design Variable” on page 187

Adding a Design Variable

To add a design variable to the list of variables in the Virtuoso® Analog Circuit Optimization
Option window,
1. Choose Variables – Add/Edit or click Add/Edit Variables.

June 2004 184 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The Editing Variables form opens.

2. Highlight the variable you want to add.

3. In the Initial Value field, type a value to be used as the starting point for optimization.
4. In the Minimum Value field, type a minimum value. The optimizer never sets the variable
to a value lower than this.
5. In the Maximum Value field, type a maximum value. The optimizer never sets the
variable to a value greater than this.
6. If you want to include the variable in the current analysis, be sure Enabled is on.
7. Click OK.
The variable is added to the Virtuoso® Analog Circuit Optimization Option window.

Editing a Design Variable

To edit one of the design variables listed in the Virtuoso® Analog Circuit Optimization Option
window,

June 2004 185 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

1. Highlight the variable you want to edit.

2. Choose Variables – Add/Edit or click Add/Edit Variables.
The Editing Variables form opens.

3. In the Initial Value field, type a value to be used as the starting point for optimization.
4. In the Minimum Value field, type a minimum value. The optimizer never sets the variable
to a value lower than this.
5. In the Maximum Value field, type a maximum value. The optimizer never sets the
variable to a value greater than this.
6. If you want to include the variable in the current analysis, be sure Enabled is on.
7. Click OK.
The changes are applied.

Deleting a Design Variable

To delete one of the design variables listed in the Virtuoso® Analog Circuit Optimization
Option window,

June 2004 186 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

1. Highlight the variable you want to delete.

2. Choose Variables – Delete or click Delete.

Enabling or Disabling a Design Variable

For a quick way to enable or disable a design variable,

1. Highlight the design variable in the Virtuoso® Analog Circuit Optimization Option
window.
2. Choose Variables – Enable or Variables – Disable.

Controlling the Optimizer

After you define the goals and specify the design variables to use, you are ready to use the
optimizer. The following sections describe how to run and stop the optimizer and how to
delete simulation results you do not want to keep.

Running the Optimizer

To run the optimizer,

1. If the previous run of the optimizer stopped because an error occurred, click Stop
Optimizer to clear the existing state.
2. Choose one of the run commands from the Virtuoso® Analog Circuit Optimization Option
window.
❑ Choose Optimizer – Run or click Run Optimizer to start the optimization from the
beginning and run it until the stopping criteria are met.
❑ Choose Optimizer – Step to start the optimization from the most recent stopping
point, iterate once, and then stop.
If you want to start from the beginning of the optimization, choose Optimizer –
Reset before choosing Optimizer – Step.

June 2004 187 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

❑ Choose Optimizer – Run n to open the Run for Fixed Number of Iterations form.

In the form, set the value of n by moving the slider.

Click OK to start the optimization from the most recent stopping point, using the
most recent variable values, and run at most n iterations before pausing or stopping.
If you want to start from the beginning of the optimization, choose Optimizer –
Reset before choosing Optimizer – Run n.

As each iteration finishes, the optimizer updates the Prev and Current values displayed in
the Virtuoso® Analog Circuit Optimization Option window.

Be aware that if you load or reload a state, the next optimizer run starts at the beginning of
the optimization. Be aware also that if you change the characteristics of a goal or the number
of enabled goals and then resume the optimization, the optimizer performs another initial
simulation.

Stopping the Optimizer

To stop the optimizer,

➤ Choose one of the stop commands from the Virtuoso® Analog Circuit Optimization
Option window.
❑ Choose Optimizer – Stop or click Stop Optimizer to stop the optimizer after the
current iteration.
❑ Choose Optimizer – Stop Now to stop the optimizer immediately without
necessarily completing the current iteration.

You can also click Stop Optimizer to clear the state if an error stops the optimizer before the
normal end of the run.

June 2004 188 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Deleting Simulation Results

To delete all simulation results,

➤ Choose Optimizer – Reset.
Any simulation results that exist are deleted. Goals, design variables, and plotting
options remain unchanged.

Plotting Results
The easiest way to track the progress that the optimizer makes toward the goal is to plot the
data as it becomes available at each iteration. When the optimizer achieves acceptable
results, you can update the design to incorporate the optimized variable values.

The following sections explain how to set the plotting options, how to plot the output data, and
how to update your design with the calculated optimal values.

Setting the Plotting Options

To set the plotting options,

1. In the Virtuoso® Analog Circuit Optimization Option window, choose Results – Set Plot
Options.

June 2004 189 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The Setting Plotting Options form appears.

2. If you want to be able to follow the progress of the optimization while the optimizer is
running, turn on Auto Plot After Each Iteration.
If you do not want to follow the progress during the run, you can turn off this button and
plot the results when the run has finished. For more information, see “Plotting Output
Data” on page 192.
3. Select at least one kind of information to be included in the output data plot.
❑ Turn on Design Variables to produce a plot showing how the design variable
values change during the optimization.
❑ Turn on Scalar Goals to produce a plot showing progress toward a scalar goal.

June 2004 190 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

You can also see how the scalar numbers change by looking in the Virtuoso®
Analog Circuit Optimization Option window at the displayed values for Initial, Prev,
and Current.
❑ Turn on Functional Goals to produce a plot, like the example below, showing
progress toward a waveform goal.

If there are too many or too few waveforms displayed in this plot, type the number
you want in the No. of Functional Iterations to Display field.
4. Set the Waveform Window characteristics to the values that work best for you. To make
the plot easier to read, for example, you might enlarge the font size and increase the size
of the window.
5. Click OK.
If the Waveform Window is open, the window changes to reflect the new option settings.
If the Waveform Window is closed, it opens with a new plot drawn in accordance with the
changed option settings.

June 2004 191 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Plotting Output Data

If Auto Plot After Each Iteration is turned on in the Setting Plotting Options window, the
Waveform Window automatically opens and displays the results of each optimization. If the
results do not appear automatically, you can use the following procedure to plot them when
the optimization ends.
➤ Choose Results – Plot History or click Plot History.
The Waveform Window appears in the format specified by the Setting Plotting Options
window. For more information, see “Setting the Plotting Options” on page 189.

If the plotting options are set so that all the output data is plotted, an output plot might look
like this.

Updating Your Design

To copy the optimized variable values back to your schematic,

1. In the Virtuoso® Analog Circuit Optimization Option window, choose Results – Update
Design or click Update Design.
2. In the Virtuoso® Analog Design Environment window, choose Variables – Copy to
Cellview.
3. In the Virtuoso Schematic Editing window, choose Design – Check and Save.

June 2004 192 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Saving, Changing, and Loading Session Information

With the Session menu pulldowns on the Virtuoso® Analog Circuit Optimization Option
window, you can save the session state, load a saved state, change optimization options, and
clear the window of all information.

Saving the Session State

To save the session state (the goals, variables, and options used in the Virtuoso® Analog
Circuit Optimization Option window),
1. Choose Session – Save State.
The Saving State form appears.

2. In the Save As field, type a name for the state if you do not want to use the default name.
3. Select one of the existing states in the Existing States listbox. As a result, the Save As
field displays the name of the selected state. The substates displayed in the What to
Save section are also enabled or disabled accordingly ( ie, according to the substates in
the selected state). Additionally, you can also enable or disable any of the substates and
click OK to overwrite the existing state. You can also type in a new state name in the
Save As field to save the changes in a new state.
Note: Only the information you save is available for retrieval when you reload the saved
state.

June 2004 193 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The optimizer saves the session state in the directory

~/.artist_states/LibraryName/CellName/.asd_optimization/
StateName
In this directory name, LibraryName and CellName are derived from the circuit you
are optimizing, and StateName is the name you specify in the Saving State form.

Loading a Saved Session State

To load a saved session state,
1. Choose Session – Load State.
The Loading State form appears.

2. From the Library cyclic field, choose the library containing the saved state you want to
load.
3. From the Cell cyclic field, choose the cell containing the saved state you want to load.
4. From the State Name field, choose the state you want to load.

June 2004 194 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

5. Turn on buttons to indicate which information you want to use from the saved state.
6. Click OK.

Saving a Script
The Open Command Environment for Analysis (OCEAN) command language lets you set up,
simulate, and analyze circuit data. OCEAN is a text-based process you can run from a UNIX
shell or from the Command Interpreter Window (CIW). You can type OCEAN commands in
an interactive session, or you can create scripts containing your commands and load those
scripts into OCEAN.

You can use the Corners window to set up the analysis you need, and then save the setup
procedure in a script. You can edit the saved script to add simulation or postprocessing
commands as needed.

For more information about OCEAN commands and scripts, see the OCEAN Reference.

To create a script and save it,

1. Choose Session – Save Script.
The Save Ocean Script to File form appears.
2. In the File Name field, specify the name of a file to contain the script.
3. Click OK.

Changing Optimization Options

Most users do not need to change the default optimization options. However, if you want to
use a specific algorithm or if you want to change the values that control the algorithm, follow
these instructions.
1. Choose Session – Options from the Virtuoso® Analog Circuit Optimization Option
window.

June 2004 195 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The Optimization Options form appears.

2. To force the optimizer to use a particular algorithm, select either LSQ or CFSQP in the
Algorithm Selection field. If you want the optimizer to choose an appropriate algorithm
automatically, select Auto.
The LSQ algorithm is best suited for a pure curve-fitting problem, and Cadence
recommends that you use it only for a problem of that kind.
For the LSQ algorithm, the match, maximize, and minimize directions are all
equivalent. In each of these cases, the LSQ algorithm works to match the specified
Target value. To use the LSQ algorithm for a maximization or minimization problem, you
must specify a Target value that is large enough or small enough that the result reaches
the maximum or minimum before it reaches the Target value.
When the Algorithm Selection cyclic field is set to Auto, the optimizer uses the CFSQP
algorithm in most cases. The optimizer uses the LSQ algorithm only when both of the
following conditions are true.
❑ The Direction for all the enabled goals is match.
❑ Every enabled goal has a waveform Target.
3. Type values for the Optimizer Control Options you want to change.
❑ The Percentage Finite Difference Perturbation value affects how sensitivities
are determined.

June 2004 196 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Be aware that some problems are very sensitive to this value and changing it might
cause the algorithm to perform poorly.
Note: It is recommend that user schoose the default value. For advanced users who
have better knowledge of the effect of the step length, the Finite Difference
Perturbation field provides a way to specify the step length that is appropriate.
Caution should be taken in using this; some problems are very sensitive to the step
length used.
❑ The Relative Design Variable Tolerance value affects the LSQ algorithm
stopping criteria. This value has no effect on the CFSQP algorithm.
For example, specifying a value of 0.05 causes the LSQ algorithm to stop when the
relative change in each design variable is smaller than 5 percent.
❑ The Relative Function Value Tolerance also affects the algorithm stopping
criteria.
For example, specifying a value of 0.05 causes the algorithm to stop when the
relative change in each function value is smaller than 5 percent.
Note: The Relative Design Variable Tolerance and Relative Function Value
Tolerance fields are designed in a way such that users can stop the algorithm by
specifying stopping criteria to be used rather than using the default settings. These fields
are entered as absolute numbers. For example, if a user specifies 0.01 in the Relative
Design Variable Tolerance field, that means if the relative change in the design
variables is smaller than 1 percent, the algorithm would stop. Likewise, if 0.01 is specified
in the Relative Function Value Tolerance field, the algorithm will stop when the
relative change in each function value is smaller than 1 percent.
4. Set Warning Message for Long Simulation.
The optimization tool and the Virtuoso® Analog Design Environment are both locked for
the duration of the initial simulation run of the optimization. During that initial run, you
cannot stop the simulation or monitor the progress. If you want to be warned of a
potentially long simulation, leave Warning Message for Long Simulation turned on,
otherwise, turn it off.
5. Click OK.

Deleting All Setup Information

To delete all the setup information about goals, variables, and plotting options,
➤ Choose Session – Reset.

June 2004 197 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Working through an Extended Example

This section follows an optimization session in detail, demonstrating how you might use the
optimizer to improve a real circuit. The example describes how to optimize a Chebyshev filter
so that its frequency response matches a specified waveform and its noise output is
minimized.

The Chebyshev filter has the following schematic:

Notice the two resistors, R0 and R1. These are the components whose values are optimized
during the session.

To follow along with this example, go to a working directory and use a command like the
following to copy all the contents of the optimization directory into the working directory.
tar -cvhf - -C <install_dir>/tools/dfII/samples/artist optimization
| tar -xvf -

Then go to the optimization directory you created, start icms, and continue with the
following steps.
1. In the CIW, choose Tools – Analog Environment – Simulation.
The Virtuoso® Analog Design Environment window appears.
2. Choose Setup – Design.

June 2004 198 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The Choosing Design form appears.

3. In the Library Name field, choose the filterlib library.
4. In the Cell Name field, choose the chebyshev cell.
5. Click OK.
6. In the Virtuoso® Analog Design Environment window, choose Session – Load State.
The Loading State form appears.
7. In the State Name field, choose state1.
8. Click OK.

The Virtuoso® Analog Design Environment window now looks like this.

June 2004 199 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Generating the Targets

This section explains how to generate two waveforms, which are used as targets for the
optimization described later. If this were an actual optimization session, you would probably
have existing targets and could skip directly to the optimization step described in “Setting Up
and Running the Optimization” on page 201.

You run this initial simulation, which does not involve using the optimizer at all, just as you run
other ordinary simulations.
➤ In the Virtuoso® Analog Design Environment window, choose Simulation – Run.
If the Welcome to Spectre window appears, click OK to close it.

When the simulation finishes, a waveform window appears.

Close this waveform window.

June 2004 200 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Saving the Targets

1. In the Virtuoso® Analog Design Environment window, click Results - Save.
2. In the Save Results window, specify schematic-save in the Save As field. Click OK.
3. In the Virtuoso® Analog Design Environment window, click Results - Select.
4. In the Select Results window, select schematic-save and click OK.
5. In the Virtuoso® Analog Design Environment window, click the Plot Outputs icon.

A waveform window similar to the one generated earlier appears. These two waveforms
become the targets for the optimization session described in the next section.

Setting Up and Running the Optimization

This section describes how to set up and run the optimization for the Chebyshev filter.
1. In the Virtuoso® Analog Design Environment window, choose Tools – Optimization.
The Virtuoso® Analog Circuit Optimization Option window appears.
2. Fill in the Goals and Variables panes with the values that are required for the
optimization.
This step is described in the next section.

Filling in the Goals Pane

In this example, you want to use the optimizer to determine what resistor values will allow you
to match the AC Response waveform while minimizing the noise waveform. The following
sections describe how to specify the goals that correspond to these waveforms.

Specifying the AC Response Goal

To define the goal corresponding to the AC Response, follow these steps.

1. In the Virtuoso® Analog Circuit Optimization Option window, choose Goals – Add.

June 2004 201 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The Adding Goals window appears.

2. Enter a name for the goal in the Name field.

For this example, type magVF.
3. To create the expression, you can use the Waveform Calculator. To open it, click Open.
4. Open the Virtuoso Schematic window by choosing Session – Schematic Window in
the Virtuoso® Analog Design Environment window.
5. In the Waveform Calculator, click vf and then go to the Virtuoso Schematic window and
select the net connected to Vout. Press Esc to end the selection.
6. In the Waveform Calculator, click mag.
The calculator display now contains the value mag(VF("/Vout")).
7. In the Adding Goals window, highlight the Expression field, then click Get Expression
to copy the expression from the calculator.
8. For this example, you want to match the AC Response waveform, so choose match in
the Direction cyclic field.
9. The Target value for this goal is to be the AC Response waveform calculated earlier, as
described in “Generating the Targets” on page 200. To specify the waveform, first click
clst in the calculator to clear the calculator display.

June 2004 202 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

10. In the calculator, click wave, then go to the Waveform Window and click on the AC
Response waveform.
An expression similar to the following appears in the Calculator display:
mag( VF("/Vout" "/old2/lorenp/simulation/chebyshev/spectre/
schematic-save") )
This expression represents the AC Response waveform.
11. In the Adding Goals window, select the Target field, then click Get Expression to copy
the waveform.
12. Type 5 in the Acceptable field and turn on % within Target.
13. Ensure that Enabled is turned on.
The Adding Goals window now looks like this.

14. Click Apply.

The new goal appears in the Virtuoso® Analog Circuit Optimization Option window.

Specifying the Noise Goal

The steps required to define the noise goal are similar to those required for the AC Response
goal.

June 2004 203 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

1. If the Adding Goals window is not open, choose Goals – Add in the Virtuoso® Analog
Circuit Optimization Option window.
2. Type a name for the goal in the Name field.
For this example, type noise.
3. Erase any existing information, then type VN2() in the Expression field.
4. Specify the Target, which for this goal is to be the Noise Response waveform calculated
earlier, as described in “Generating the Targets” on page 200. To specify the waveform,
first open the calculator by clicking Open in the Adding Goals window.
5. Click clst in the calculator to clear the calculator display.
6. In the calculator, click wave, then go to the Waveform Window and click on the Noise
Response waveform.
An expression similar to the following appears in the calculator display:
VN2()
Type the following path in the brackets:
"/old2/lorenp/simulation/chebyshev/spectre/schematic-save"
This expression represents the Noise Response waveform.
7. Return to the Adding Goals window, select the Target field, then click Get Expression
to copy the waveform.

June 2004 204 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

8. Fill in the other fields of the Adding Goals window, as follows.

9. Click OK.. The new goal appears in the Virtuoso® Analog Circuit Optimization Option
window.
10. You can return to the calculator window and close it by choosing Window – Close.

Filling in the Variables Pane

In this example, you want to optimize the values of two resistors: res0 and res1. To prepare
for the optimization, you need to set the initial, minimum, and maximum allowed values.
1. In the Virtuoso® Analog Circuit Optimization Option window, choose Variables – Add/
Edit.
The Editing Variables form appears.

June 2004 205 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

2. Click on res1, and then fill in the other fields as shown.

3. Click Apply.
The information about the res1 variable appears in the Virtuoso® Analog Circuit
Optimization Option window.

June 2004 206 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

4. In the Editing Variables form, click on res0 and then fill in the other fields as shown.

5. Click OK.
The information about the res0 variable appears in the Virtuoso® Analog Circuit
Optimization Option window.

June 2004 207 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Running the Optimization

With the goals and variables defined, the Virtuoso® Analog Circuit Optimization Option
window looks like this.

At this point in the example, you are ready to run the optimization.
➤ In the Virtuoso® Analog Circuit Optimization Option window, choose Optimizer – Run
or click Run Optimizer.
The optimization starts and the status display updates to reflect the current activity.

Looking at the Output

With the default plotting options, a Waveform Window appears soon after the optimization run
begins and updates after each iteration. When this example optimization ends, the Waveform

June 2004 208 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

Window displays the iteration history of the res0 and res1 variables. It also displays the
changing waveforms for the magVF and noise goals.

You can change the information that displays in the Waveform Window. For example, to look
at the variables in more detail, follow these steps.
1. In the Virtuoso® Analog Circuit Optimization Option window, choose Results – Set Plot
Options.

June 2004 209 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

2. In the Setting Plotting Options window, turn off Display History of Scalar Goals and
Display History of Functional Goals, leaving only the Display History of Variables
selected.

3. Click OK.

June 2004 210 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The Waveform Window appears, showing only the variables.

During the 13 iterations of this optimization (iteration 0 shows the initial values), the value
of res1 went from the starting value of 2.00 K to the final value of 304.7. This final value
also appears in the Current column of the Variables pane in the Virtuoso® Analog
Circuit Optimization Option window.

To look at the goals in more detail,

1. In the Setting Plotting Options window, turn on Display History of Functional Goals.
2. Turn off Display History of Variables and Display History of Scalar Goals.
3. Set No. of Functional Iterations to Display to 1.
4. Click OK.

June 2004 211 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The Waveform Window appears or redraws with only the goals showing. For example,
the noise goal looks like this.

The last iteration, I#13, shows that the optimizer was able to lower the noise below the
target throughout the entire frequency range.

The optimized values for the two resistors are displayed in the current column of the
Virtuoso® Analog Circuit Optimization Option window: res1 has an optimized value of 304.7
and res0 has an optimized value of 1.311 K. To use these values in your design, do the
following:
➤ In the Virtuoso® Analog Circuit Optimization Option window, choose Results – Update
Design.

June 2004 212 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

The new values appear in the Design Variables pane of the Virtuoso® Analog Design
Environment window.

June 2004 213 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide
Optimization

June 2004 214 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

Index
Symbols appending scalar output data to saved
data 82
% within Target 172, 174 Autoplot After Each Iteration button
.cdsinit file (optimization) 190, 192
example 68 Autoplot button (Monte Carlo) 89
loading PCFs and DCFs from 15, 20, autoplot field 84
41
B
A Best Fit Line 121
Acceptable values button bar (Monte Carlo) 85
creating waveform objects for 176 buttons
entering an expression for 171, 174 Corners
restrictions on 171, 174 Add Corner 16
setting as percentage of Target 172, Add Measurement 18, 26, 27
174 Add Variable 16
valid 171, 174 Calculator 18, 27
weights, using to assign 180 Copy Corner 16
Add button (Monte Carlo) 85, 89 Delete Measurement 18, 27
Add Corner button (Corners) 16 Delete Row 25
Add Measurement button (Corners) 18, Delete Selected / Delete Corner /
26, 27 Delete Row 16
Add Process dialog box 58 Get Expression 18, 27
Add Variable button (Corners) 16 Plot/Print 29
Add/Edit Goals button (optimization) 170, Run 28
172, 175 Run / Stop 17
Add/Edit Variables button Monte Carlo
(optimization) 184, 186 Add 85, 89
adding Autoplot 89
corners 16 Calculator 85, 90
design variables (optimization) 184 Change 85, 90
variables (Corners) 16, 24 Clear 85
Adding Goals window 202, 203, 205 Delete 85, 90
algorithms Density Estimator 118
automatic selection of 196 Get Expression 85
CFSQP 160 Load 110, 112
choosing manually 196 optimization
LSQ 160 Add/Edit Goals 170, 172, 175
amplifier schematic (Monte Carlo) 129 Add/Edit Variables 184, 186
Analysis Setup pane Autoplot After Each Iteration 190,
description 81 192
Monte Carlo extended example 134 Delete (design variables) 187
Analysis Variation cyclic field (Monte Delete (goals) 176
Carlo) 104 Design Variables 190
analysis variation type, specifying 86 Enabled (goals) 172, 175, 185, 186

June 2004 215 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

Open 173 menu 13

Scalar Goals 190 corners
Special Functions 177 defining, in Virtuoso analog corners
Stop Optimizer 188 analysis window 23
Update Design 192 deleting 25
Waveform Goals 191 specifying 20
corners0.log 19
Correlation Table window (Monte
C Carlo) 116
correlation tables
calculator description 113
building expressions with (Monte printing 116
Carlo) 90 corSetModelFile procedure
building goal expressions with used only with single model library
(optimization) 172 style 37
creating waveforms with using to enter model file name 42
(optimization) 177 cumulative box histograms 117
opening 173 cumulative line histograms 117
Calculator button Current field 166, 167
Corners 18, 27 currents, saving all 95
Monte Carlo 85, 90
.cdsinit file 68
.cdsinit file, using to load PCFs and D
DCFs 15, 20, 41
cdsSpice simulations data
and select all options 96 filter
CFSQP algorithm, data suited for 160 reloading settings 110
Change button (Monte Carlo) saving settings 110
description 85 specifying settings 109
example of use 90 turning off 110
Chebychev filter outlying error values, filtering out 108
description 198 saving between runs 88
Clear button (Monte Carlo) 85 Data Filter form 109
columns, selecting in Corner Definitions DCFs. See design customization files
pane 15 (DCFs)
conditional yield debugging PCFs and DCFs 37
definition 123 Delete button (Monte Carlo) 85, 90
reports, description 114 Delete button (optimization)
reports, printing 126 for design variables 187
Conditional Yield form 127 for goals 176
Copy Corner button 16 Delete Measurement button (Corners) 18,
copying an existing corner 16 27
Corners Delete Row button (Corners) 25
analysis Delete Selected / Delete Corner / Delete
definition 9 Row button (Corners) 16
extended example 60 deleting
overview 9 corners 25
starting 28 design variables 186
stopping 28 goals 176
Corner Definitions pane, description 15 simulation results 189
log file 19 user-defined corners 16

June 2004 216 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

variables or groups 25 Edit – Delete Row 25

Density Estimator button (Monte edit fields
Carlo) 118 clearing 85
Design – Check and Save 192 description 84
design customization files (DCFs) editing
.cdsinit, loading with 41 design variables 185
commands normally placed in 37 goals 175
debugging with OCEAN 37 Editing Variables form 205
loaded after PCFs 41 Enabled button
loading from the graphical user for goals 172, 175, 185, 186
interface 21 Enabled field 166, 167
tailoring a Corners analysis with 20 enabling
use of 36, 39 design variables 187
design variables goals 176
adding to DCF 39 error messages (Corners) 19
deleting all 197 example, extended
deleting specific 186 optimization 198
determining sensitivities of 161 Expression column, in Corners Performance
editing 185 Measurements pane 18
enabling or disabling 187 expressions
examples of 184 adding 85
names, as displayed in Corner adding by typing in 89
Definitions pane 16 changing 85, 90
pane showing values of 167 checking validity 96
plotting 190 creating with the calculator 202
setting maximum value for 185, 186 deleting 85, 90
setting minimum value for 185, 186 entering in Performance Measurements
stopping criteria for 197 pane 18
sweeping 87 getting from the calculator 202
updating schematic, with optimized 192 listed in the Outputs pane 83
Design Variables button 190 names for 83
design, updating with optimized retrieving from calculator (Corners) 18
values 192 retrieving from calculator (Monte
device descriptions, for Monte Carlo 76 Carlo) 85
direction used to specify goals 169
specifying for Chebychev filter using calculator to build 90
example 202 extended examples
specifying for goals 171, 173 Corners analysis 60
Direction field 166 Corners, PCF for 66
disabling Monte Carlo 128
design variable 187
goals 176
disk storage requirements, reducing 88 F
distribution concentration, estimating 118
family of curves
for Monte Carlo extended example 143
E plots, description of 113
plotting 119
Edit – Add Measurement 26, 27 saving data for 82, 88
Edit – Delete Corner 25 feasible initial values, determined by CFSQP
Edit – Delete Measurement 27 algorithm 161

June 2004 217 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

File – Close 11
File – Load 21
G
File – Save Ocean Script 35 GAUSS function, used in Monte Carlo 132
File – Save Setup 33 Get Expression button (Corners) 18, 27
File – Save Setup As 33 Get Expression button (Monte Carlo) 85
Filter By data set 109 goals
Filter By point 109 definition 169
filter, Chebychev 198 deleting all 197
filtering, turning off 110 deleting specific 176
folded cascode, schematic 61 editing 175
formats, choosing for output 28 enabling or disabling 176
forms and windows example of specifying 201
Add Process (Corners) 58 saving 193
Adding Goals 170, 173, 202, 203, 205 specifying direction for 171, 173
Conditional Yield 127 specifying expression for 171, 173
Correlation Table 116 specifying for Chebychev filter
Data Filter 109 example 203
Editing Goals 175 specifying name for 173
Editing Variables 185, 186, 205 Goals – Add 170, 172, 201, 204
Histogram 118 Goals – Delete 176
Iteration Versus Value 115 Goals – Disable 176
Load (Corners) 22 Goals – Edit 175
Load Data Filter Values 111 Goals – Enable 176
Load Specification Limits 113 Goals pane 166
Loading State 194 graphical user interface
Monte Carlo Load 102 Corners 12
Monte Carlo Save Ocean Script 103 Monte Carlo 78
Multiconditional Yield 125 optimization 163
Optimization Options 196 groups
Process/Model Info Setup (Corners) 59 deleting 25
Run for Fixed Number of Iterations 188 name, as displayed by Corners 16
Save Changes? (Corners) 21
Save Data Filter Values 110
Save Ocean Script (Corners) 36, 195
Save Results 100
H
Save Specification Limits 112 help
Saving State 193 for optimization 161
ScatterPlot 121 Histogram form 118
Select Results 107 histograms
Setting Plotting Options 190, 210 description 113
Simple Yield 124 for Monte Carlo extended example 138
Specification Limits 111 plotting 117
Virtuoso analog corners analysis 12
Virtuoso analog optimization
analysis 163, 173
Virtuoso analog statistical analysis 78
I
Virtuoso® analog corners analysis 10 individual yield 123
Virtuoso® analog statistical analysis 78 information messages, as displayed by
Waveform 178, 191, 200 Corners 19
frequency response, matching 198 Initial field 166, 167
Functional Goals button 191

June 2004 218 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

initial values, determining feasible values creating with calculator 26

for 161 deleting 18, 27
input files, creating by hand 101 setting highest acceptable value for 18
iteration history 209 setting lowest acceptable value for 18
iteration versus value tables setting target value for 18
description 113 menu
printing 114 Corners analysis tool 13
Iteration Versus Value window 115 Monte Carlo tool 79
optimizer 164
messages, displayed in Corners status
L bar 19
Min field 167
least squares fit lines, for scatter plots 121 minimize
Load button (Monte Carlo) 110, 112 equivalent to match for LSQ
Load Data Filter Values form 111 algorithm 180
Load dialog box 22 Target used only for weighting when
Load Specification Limits form 113 chosen 179
loading minmax, algorithm for 160
session state 102 Mismatch Only variation type 86
stored outputs 106 model files
Lower column, in Performance connecting to Analysis Variation
Measurements pane 18 cyclic 104
lowpass filter for Monte Carlo 76
description 128 lowpass filter 131
model file 131 model files, purpose 36
schematic 128 modeling styles
LSQ algorithm multiple model library 48
conditions of use for 196 multiple numeric 52
data suited for 160 multiple parametric 54
equivalence of match, minimize, and single model library 42
maximize 180 single numeric 51
supported by Corners analysis tool 41
Monte Carlo analysis
M Analysis Setup pane 81
analyzing results 137
Max field 167 button bar 85
maximize edit fields 84
equivalent to match for LSQ extended example 128
algorithm 180 graphical user interface for 78
Target used only for weighting when menu 79
chosen 180 number of runs, specifying 86
mcdata file, creating from saved waveform Outputs pane 83
data 108 overview 75
mcparam file, creating 101 requirements for running 76
mcrun.s file results, analyzing 106
creating 101 specifying analysis variation type for 86
Measurement column, in Performance specifying characteristics for 81
Measurements pane 18 specifying initial run number for 86
measurements starting 99
adding new row for 18 status display 79
creating by hand 26 stopping 99

June 2004 219 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

Monte Carlo Load form 102 control options, setting 196

Monte Carlo Save Ocean Script form 103 definition 159
Monte Carlo tool running 187
closing 104 stopping 188
starting 76 Optimizer – Reset 189
multiconditional yield Optimizer – Run 187, 208
definition 123 Optimizer – Run n 188
reports, description 114 Optimizer – Step 187
reports, printing 125 Optimizer – Stop 188
Multiconditional Yield form 125 Optimizer – Stop Now 188
multiple model library style options
described 48 for optimization 195
requirements for using 60 for plotting 189
multiple numeric modeling outlying data, filtering 108
described 52 output data, appending to saved data 82
example of 64 output formats
PCF for 53 choosing (Corners) 28
multiple parametric modeling residual plot example (Corners) 31
described 54 text example (Corners) 29
PCF for 55 output information for optimization,
changing 209
output log, viewing 104
N outputs
choosing after the analysis runs 29
Name field (goals) 166 loading 106
Name field (variables) 167 of the Chebychev filter example 208
No. of Functional Iterations to Display 211 saving all 95
noisy data, best optimized with LSQ saving to specific file 100
algorithm 160 Simulation window outputs appear in
number of runs, specifying 82 Corners 68
Outputs – Retrieve Outputs 88
Outputs – Save All 95
O Outputs – To Be Saved – Select On
Schematic 96
OCEAN script, saving 102 Outputs pane 83
Open button 173
opening the Virtuoso® analog optimization
analysis window 201 P
optimization
definition 159 parameter storage format (PSF)
extended example 198 directory, copying 100
help for 161 files, saving between runs 82
output, changing what is displayed parameters
in 210 for Spectre, must be in main circuit 60
overview 160 showing correlations among 116
setting options for 195 sweeping 87
setting up and running 201 pass/fail histograms 117
steps followed during 160 PCFs. See process customization files
using waveform goals for 202 (PCFs)
optimized values, where displayed 212 % within Target 172, 174
optimizer percentage finite difference

June 2004 220 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

perturbation 196 files, saving between runs 82

performance measurements
adding to DCFs 39
creating by hand 26 R
creating with calculator 26
deleting 27 red colored messages 19
pane 17 relative design variable tolerance 197
Plot/Print button 29 relative function value tolerance 197
plotting reloading a session state 194
automatic 84 requirements for running Monte Carlo
histograms 117 tool 76
options, resetting all 197 residual plot 31
options, saving 193 Results – Evaluate Expressions 108
output data 192 Results – Filter 109, 110
progress toward scalar goals 190 Results – Plot – Curves 119, 143
progress toward waveform goals 191 Results – Plot – Histogram 117, 138
setting options for 189 Results – Plot – Scatter Plot 120
setting Waveform Window Results – Plot History 192
characteristics 191 Results – Plotting Options 189
specifying number of waveforms to Results – Print – Correlation Table 116
display 191 Results – Print – Iteration versus
waveforms 178 Value 114
Prev field (for goals) 166 Results – Print – Iteration vs. Value 137
Prev field (for variables) 167 Results – Save 100
process customization files (PCFs) Results – Select 106
.cdsinit, loading with 15, 41 Results – Set Plot Options 209
commands normally placed in 37 Results – Specification Limits 111, 141
creators of 36, 37 Results – Update Design 192, 212
debugging with OCEAN 37 Results – Yield – Multiconditional 125
example 38 Results – Yield – Simple 123, 142
for multiple numeric modeling 53 results, plotting 189
for multiple parametric modeling 55 Run / Stop button (Corners) 17, 28
for single file modeling 43, 49 Run Optimizer button 208
for single numeric modeling 51 run temperature, always appears in Corners
loaded before DCFs 41 window 68
loading from DCF 41 running
loading from the graphical user Monte Carlo analysis 99
interface 21 optimization
used for predefined corners 20 for a fixed number of iterations 188
process field 15 for one iteration 187
Process Only variation type 86 major steps in 168
process variables until stopping criteria are met 187
adding in Corners window 58
Process Variation and Mismatch variation
type 87 S
processes
defining in Corners window 57 Save All command 95
modifying in Corners window 58 Save Changes? dialog box 21
name of 15 Save Data Between Runs to Allow Family
PSF (parameter storage format) Plots button 108
directory, copying 100 Save Data Filter Values form 110

June 2004 221 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

Save Ocean Script form 36, 195 saving to default files 33

Save Results dialog box 100 saving to specified files 33
Save Specification Limits form 112 signals
saving adding 85, 89
all currents 95 changing 85
all node and terminal values 95 deleting 85, 90
all voltages 95 Simple Yield form 124
Monte Carlo session state 101 simple yield reports
optimization session state 193 description 113
Scalar Goals button 190 printing 123
scalar output data Simulation – Check Expressions 96
analyzing 137 Simulation – Create Input Files 101
appending to saved data 87 Simulation – Output Log 104
data type 84 Simulation – Run 28, 99, 135, 200
scatter plots Simulation – Stop 28, 99
description 113 simulations
plotting 120 outputs, saving all 95
ScatterPlot form 121 specifying number to run 82
schematics simulator, choosing 77
for Chebychev filter example 198 single model library style
for folded cascode example 61 described 42
updating with optimized values 192 PCF for 43, 49
Select Results dialog box 107 requirements for using 60
sensitivities single numeric modeling
changing default values for 196 described 51
how determined 161 PCF for 51
Session – Load State (Monte Carlo) 102 SKILL PI commands, using in PCFs and
Session – Load State (optimization) 194 DCFs 37
Session – Options (optimization) 195 sorting Monte Carlo outputs 115
Session – Quit (Monte Carlo) 104 Special Functions button 177
Session – Quit (optimization) 162 specification limits
Session – Reset (optimization) 197 for Monte Carlo extended example 142
Session – Save Script (Monte Carlo) 102 saving 112
Session – Save State (Monte Carlo) 101 set by limits 112
Session – Save State (optimization) 193 Specification Limits form 111
session state Spectre simulator, using with Corners 60
loading (Monte Carlo) 102 standard deviations 109
loading (optimization) 194 standard histogram 117
saving (Monte Carlo) 101 starting
saving (optimization) 193 Corners analysis 28
Set By limits (Data Filter form) 109 Monte Carlo analysis 99
Set By limits (Specification Limits Monte Carlo tool 76
form) 112 optimization 187
Set By sigma (Data Filter form) 109 starting run number, specifying 82, 86
Set By sigma (Specification Limits statistical values, for Monte Carlo 76
form) 112 statistical variation, specifying which to
Setting Plotting Options window 210 run 82
Setup – Add Process 57 statistics block
Setup – Add/Update Model Info 58 example 104
setup information using with Spectre simulator 104
deleting all 197 status display (Corners) 19

June 2004 222 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

status display (Monte Carlo) 79

Stop Optimizer button 188
V
stopping variables
Corners analysis 28 adding (Corners) 24
criteria for optimization 161, 197 deleting (Corners) 25
Monte Carlo analysis 99 display of, by optimizer 211
optimization 188 setting for Chebychev filter
Suppress field example 205
Conditional Yield form 127 Variables – Add/Edit 184, 186, 205
Multiconditional Yield form 125 Variables – Copy to Cellview 192
swept parameter, specifying 82, 87 Variables – Delete 187
Variables – Disable 187
Variables – Enable 187
T Variables pane 167
variants
table function, using in optimization 177 defined in modeling file 38
Target column, in Performance Virtuoso® analog corners analysis window
Measurements pane 18 closing 10
Target values opening 10
creating waveform objects for 176 Virtuoso® analog optimization analysis
field 166 window, closing 162
how used by optimizer 179 window, opening 162
how used to assign weights 180 Virtuoso® analog optimization analysis
relation to Acceptable values 171, 174 window
setting Acceptable values as percentage description 163
of 172, 174 opening 201
used as goal 179 Virtuoso® analog statistical analysis
valid 171, 173 window 78
Temperature, sweeping 87 voltages, saving all 95
text
output example (Corners) 29
tool bar (optimization) 168
Tools – Calculator 27
W
Tools – Corners 11 waveform data
Tools – Get Expression 27 creating mcdata file from 108
Tools – Monte Carlo 133 data type 84
Tools – Optimization 162, 201 waveform window 200
Tools – Plot or Print Outputs 29 waveforms
total yield 123 creating 176
generating 200
plotting 178
U using as an optimization goal 202
weights
unknown data 84 determined by Target and Acceptable
Update Design button 192 together 180
Upper column, in Performance formula for 180
Measurements pane 18 windows. See forms and windows
useAltergroup variable
value to use with Spectre 60

June 2004 223 Product Version 5.1.41

 Virtuoso Advanced Analysis Tools User Guide

X
X and Y values, creating waveforms
from 176

Y
yellow colored messages 19
Yield – Conditional 126
yields
analyzing 141
conditional
description 123
printing report on 126
individual
description 123
printing report of 123
multiconditional
description 123
printing report on 125
simple, printing report on 123
total 123

June 2004 224 Product Version 5.1.41