GENESYS

=EMPOWER=

© Copyright 1986-2000

Eagleware Corporation635 Pinnacle CourtNorcross, GA 30071 USA

Phone: +1 (678) 291-0995FAX: +1 (678) 291-0971E-mail: eagleware@eagleware.comInternet: http://www.eagleware.com

PLEASE NOTE: This address is our new 15,000 squarefoot facility. Please update your records to this new ad-dress.

Printed 10/2000Printed in the USA

TABLE OF CONTENTS

Introduction . . . . . . . . . . . . . . . . . . . viiGeneral Information . . . . . . . . . . . . . . . . . . . . viiNote For Users Adding =EMPOWER= . . . . . . . . . . viiiNote For Users New To Genesys . . . . . . . . . . . . . viii=EMPOWER= Historical Background . . . . . . . . . . ixFeature Overview . . . . . . . . . . . . . . . . . . . . . . x=EMPOWER= Documentation . . . . . . . . . . . . . . xi

Chapter 1: Starting . . . . . . . . . . . . . . . . 1Creating a New Layout . . . . . . . . . . . . . . . . . . . 1Creating a Layout Without a Schematic . . . . . . . . . . 2Drawing the Layout . . . . . . . . . . . . . . . . . . . . . 8Viewing =EMPOWER= Results . . . . . . . . . . . . . . 12Creating a Layout From an Existing Schematic . . . . . 16

Chapter 2: =EMPOWER= Basics . . . . . . . . . 232-d Simulators . . . . . . . . . . . . . . . . . . . . . . . 233-d Simulators . . . . . . . . . . . . . . . . . . . . . . . 232 1/2-d Simulators . . . . . . . . . . . . . . . . . . . . . 23Basic Geometry . . . . . . . . . . . . . . . . . . . . . . 24The Grid . . . . . . . . . . . . . . . . . . . . . . . . . . 28Viaholes and Z-directed Ports . . . . . . . . . . . . . . . 33Emports . . . . . . . . . . . . . . . . . . . . . . . . . . 34=EMPOWER= Options Dialog Box . . . . . . . . . . . . 34

Chapter 3: Tips . . . . . . . . . . . . . . . . . . . 35Cell Size . . . . . . . . . . . . . . . . . . . . . . . . . . 36Maximum Critical Frequency . . . . . . . . . . . . . . . 36Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . 37Thinning Out . . . . . . . . . . . . . . . . . . . . . . . . 38Wall & Cover Spacing . . . . . . . . . . . . . . . . . . . 39Cover Type . . . . . . . . . . . . . . . . . . . . . . . . . 39Lossy Analysis . . . . . . . . . . . . . . . . . . . . . . . 40Viewer Data . . . . . . . . . . . . . . . . . . . . . . . . 40Slot-type Structure . . . . . . . . . . . . . . . . . . . . . 41Preferred Box Cell Count . . . . . . . . . . . . . . . . . 42Thick Metal . . . . . . . . . . . . . . . . . . . . . . . . . 44

Chapter 4: External Ports . . . . . . . . . . . . . 45Placing External Ports . . . . . . . . . . . . . . . . . . . 45Emport Options . . . . . . . . . . . . . . . . . . . . . . 46

Deembedding . . . . . . . . . . . . . . . . . . . . . . . . 49Multimode Ports . . . . . . . . . . . . . . . . . . . . . . 52Generalized S-Parameters . . . . . . . . . . . . . . . . . 55

Chapter 5: Decomposition . . . . . . . . . . . . 57Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Spiral Inductor Example . . . . . . . . . . . . . . . . . . 59Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Port Numbering . . . . . . . . . . . . . . . . . . . . . . 65

Chapter 6: Lumped Elements and Internal Ports 69Placing Internal Ports . . . . . . . . . . . . . . . . . . . 69Manually Adding Lumped Elements . . . . . . . . . . . 71Automatic Port Placement . . . . . . . . . . . . . . . . . 72Planar (X- And Y-directed) Ports . . . . . . . . . . . . . . 74Resonance . . . . . . . . . . . . . . . . . . . . . . . . . . 76

Chapter 7: Viewer . . . . . . . . . . . . . . . . . 77Starting the Viewer . . . . . . . . . . . . . . . . . . . . 77Viewer Interface . . . . . . . . . . . . . . . . . . . . . . 78Viewer Examples . . . . . . . . . . . . . . . . . . . . . . 87Multimode Viewer Data . . . . . . . . . . . . . . . . . . 93Via Hole Viewer Example . . . . . . . . . . . . . . . . . 95Viewer Theory . . . . . . . . . . . . . . . . . . . . . . . 97

Chapter 8: Box Modes . . . . . . . . . . . . . . 101Homogeneous Rectangular Cavity . . . . . . . . . . . 101Higher Order Box Modes . . . . . . . . . . . . . . . . . 102Partial Dielectric Loading . . . . . . . . . . . . . . . . 103Signal Metal Effects . . . . . . . . . . . . . . . . . . . 104Top Cover . . . . . . . . . . . . . . . . . . . . . . . . . 104Cavity Absorber . . . . . . . . . . . . . . . . . . . . . 105

Chapter 9: Examples and Benchmarks . . . . 107

Chapter 10: Batch Runs/Console Mode . . . . 109Batch Runs Within Genesys . . . . . . . . . . . . . . . 109The Console Window . . . . . . . . . . . . . . . . . . . 110The Command Line . . . . . . . . . . . . . . . . . . . 111Ascii Text Files . . . . . . . . . . . . . . . . . . . . . . 111Batch Files . . . . . . . . . . . . . . . . . . . . . . . . 112Batch Files For Multiple Runs . . . . . . . . . . . . . . 112=EMPOWER= Command Line Options . . . . . . . . . 114General Options . . . . . . . . . . . . . . . . . . . . . 115

iv Table of Contents

Processing Algorithm Options . . . . . . . . . . . . . . 119Deembedding Options . . . . . . . . . . . . . . . . . . 121Viewer Data File Options . . . . . . . . . . . . . . . . 122Listing File Options . . . . . . . . . . . . . . . . . . . 123Debug Options . . . . . . . . . . . . . . . . . . . . . . 125

Chapter 11: TPL File Format . . . . . . . . . . 127Tpl File Format Overview . . . . . . . . . . . . . . . . 128Emline/line Analysis Mode . . . . . . . . . . . . . . . 134Discontinuity Analysis . . . . . . . . . . . . . . . . . . 135Simple Line Analysis Example . . . . . . . . . . . . . 178Simple Discontinuity Analysis Example . . . . . . . . 185

Appendix A: Theory . . . . . . . . . . . . . . . 191Electromagnetic Problem Formulation . . . . . . . . . 193Method Of Lines . . . . . . . . . . . . . . . . . . . . . 196Mapping On the Grid . . . . . . . . . . . . . . . . . . 197Grid Green’s Function . . . . . . . . . . . . . . . . . . 200Informational Multiport . . . . . . . . . . . . . . . . . 202Numerical Acceleration Procedures . . . . . . . . . . . 204De-embedding Algorithm . . . . . . . . . . . . . . . . 206

Appendix B: File Descriptions . . . . . . . . . . 209Where Are These Files? . . . . . . . . . . . . . . . . . 209Text Files vs. Binary Files . . . . . . . . . . . . . . . . 211File Extensions . . . . . . . . . . . . . . . . . . . . . . 212.BAT (Batch) Files . . . . . . . . . . . . . . . . . . . . 212.EMV (Viewer) Files . . . . . . . . . . . . . . . . . . . 213.l1, .l2, ... .ln (Line Data) Files . . . . . . . . . . . . . . 213.LST (Listing) Fies . . . . . . . . . . . . . . . . . . . . 214Qchk Section . . . . . . . . . . . . . . . . . . . . . . . 215Package Structure . . . . . . . . . . . . . . . . . . . . 215Memory Sections . . . . . . . . . . . . . . . . . . . . . 215Map Of Terminals . . . . . . . . . . . . . . . . . . . . 215Sdtc Section . . . . . . . . . . . . . . . . . . . . . . . 216Line Analysis Mode Results . . . . . . . . . . . . . . . 216S-Matrix Tables . . . . . . . . . . . . . . . . . . . . . 217.PLX (Current/Viewer Data) Files . . . . . . . . . . . 217.R1, .R2, ... Rn (Port Impedance) Files . . . . . . . . . 218.RGF (Line Data) Files . . . . . . . . . . . . . . . . . 218.RX (Frequency vs. Impedance) Files . . . . . . . . . . 219.WSP (Workspace) Files . . . . . . . . . . . . . . . . . 219.SS (S Parameter) Files . . . . . . . . . . . . . . . . . 219.TPL (Topology) Files . . . . . . . . . . . . . . . . . . 220

v

.Y (Y-Parameter) Files . . . . . . . . . . . . . . . . . . 220~SS, ~RG, etc. (Backup) Files . . . . . . . . . . . . . . 220

Appendix C: References . . . . . . . . . . . . . 221General Background . . . . . . . . . . . . . . . . . . . 221The Method Of Lines . . . . . . . . . . . . . . . . . . . 223Richardson’s Extrapolation . . . . . . . . . . . . . . . 224Symmetry Processing . . . . . . . . . . . . . . . . . . 224=EMPOWER= Engine Theory and Algorithms . . . . . 225Test Examples and Comparisons . . . . . . . . . . . . 227

Appendix D: =EMPOWER= Messages . . . . . 229

vi Table of Contents

Introduction

T hank You! Eagleware is proud of a tradition of high-performance, high-quality engineering software.Any suggestions or problems you have are important

to us, so please tell us about your experience with ourproducts. To contact Eagleware, please call or write to:

Eagleware Corporation635 Pinnacle CourtNorcross, GA 30071USA

Tel : +1 (678) 291-0995Fax : +1 (678) 291-0971E-Mail: eagleware@eagleware.comInternet: http://www.eagleware.com

NOTE: This address is our new 15,000 ft2 (1200 m2)facil-ity. Please update your records to this new address.

GENERAL INFORMATION

This manual describes Eagleware’s Planar 3D electromag-netic simulator =EMPOWER=. For installation/startinginformation on all GENESYS products including =EM-POWER=, please refer to the Installation guide or to theGetting Started guide.

256MBytes of RAM or more are recommended for =EM-POWER=.Other GENESYS modules require only 32MBy-tes.

NOTE FOR USERS ADDING =EMPOWER=

Great news! Not only does =EMPOWER= provide accu-racy enhancement for circuit theory simulation, but itallows you to simulate metal with unusual geometricshapes and arbitrary dielectric layers. In fact, many EMprograms do this. However, with =EMPOWER= you can

• Run existing =LAYOUT= files without redescribing thecircuit!

• Have lumped elements in your circuit, automatically!

• Tune spirals, filters and other objects in GENESYS usingmultimode decomposition!

and much more. See the feature list that follows.

NOTE FOR USERS NEW TO GENESYS

GENESYS is a suite of programs (modules) to aid RF/mi-crowave circuit design. You may purchase the completeGENESYS suite or only certain modules. Modules may becategorized into four groups. The groups and modules are:

SYNTHESIS=FILTER= L-C Filter design=OSCILLATOR= L-C, Xtal, SAW & T/line oscillator design=T/LINE= Relate physical & electrical T/line parameters=MATCH= L-C & T/line matching network & amplifier design=M/FILTER= T/line filter design=A/FILTER= Active filter design=EQUALIZE= Group delay equalizer design

GENESYS Environment=SuperStar= Circuit theory simulation, tuning & optimization=SCHEMAX= Schematic entry for simulator=LAYOUT= PWB layout and artwork=EMPOWER= Single Layer EM simulation=EMPOWER= ML

Multi-Layer EM simulation

PLL DESIGN=PLL= Both synthesis and simulation of PLLs

viii Introduction

What is the function and relationship of these modules?The synthesis modules create (synthesize) initial designsfor a wide variety of circuits. The resulting design may becustomized, tuned, and optimized using the circuit theorysimulator or simulated electromagnetically. Alternatively,you may create a design manually, describe that circuit bya schematic and then simulate your design. =LAYOUT=creates a PWB which is used both to create artwork formanufacture and for a physical description for =EM-POWER= to simulate.

Why use both circuit theory and EM simulation? Circuittheory simulation in GENESYS is amazingly fast andinteractive. No other program at any price approaches thespeed of GENESYS. =EMPOWER= simulations are moreaccurate and do not require the use of specific geometricobjects for which circuit models have been developed. EMsimulation complements rather than replaces circuit the-ory simulation.

The GENESYS suite of tools uses one easy-to-learn inter-face to provide you with unprecedented power, conven-ience and accuracy for the design of RF/microwavecircuits.

=EMPOWER= HISTORICAL BACKGROUND

Most commercial electromagnetic (EM) simulators de-signed for MIC and MMIC work are based on integralequations and the method of moments (MoM). =EM-POWER= is based on the method of lines (MoL). Thistechnique has excellent error convergence properties andsubmits well to code optimization to minimize numericcomplexity.

The root of =EMPOWER= is work which began in 1987 atthe Novosibirsk Electrical Engineering Institute. Thislead to the commercial development of TAMIC in 1991 inMoscow. TAMIC saw commercial use in the Soviet Union

Introduction ix

and elsewhere. In late 1996, Eagleware acquired TAMICand the principle contributor joined Eagleware to beginsignificant improvements. The code was integrated intothe GENESYS environment at release Version 6.5 in 1998.

FEATURE OVERVIEW

=EMPOWER= incorporates many features still not pre-sent in competitive, late generation, EM simulators. Prin-ciple features include:

• Benchmarked accuracy

• Easy to use graphical circuit layout editor

• Complete integration with the GENESYS circuit simulation,synthesis and layout tools

• Multilayer simulations (with =EMPOWER= ML)

• Automatic incorporation of lumped elements

• Automatic detection and solution with symmetry

• Generalized S-parameter support

• Multi-mode support for ports and lines

• Tuning of EM objects in GENESYS using decomposition

• Deembedded or non-deembedded ports

• Viaholes including generated fields

• Any number of dielectric layers

• Dielectric and metal loss

• Includes box modes and package effects

• Slot-mode for slot and coplanar circuits

• Thick metal simulation (with =EMPOWER= ML)

• 32-bit code for Windows 95/98/NT

x Introduction

=EMPOWER= DOCUMENTATION

The =EMPOWER= and Viewer programs and the TPLnetlist format are described in this manual. Introductoryuse of the =LAYOUT= program for graphical circuit edit-ing is also described. Details on the use of =LAYOUT= arein the =LAYOUT= manual. The circuit simulator =Super-Star= which is used to display results and the =SCHE-MAX= program which can be used to begin a problem arealso introduced here. Details concerning these programsare given in the User's Guide.

Also described in the Examples manual are example cir-cuits which have been run in =EMPOWER=. Files forthese examples are provided with GENESYS. Examplesinclude benchmarks, resonators, filters, couplers, attenu-ators, amplifiers, and other circuits.

Introduction xi

Chapter 1

Starting

An =EMPOWER= simulation requires a board layoutdescription. The easiest (and recommended) methodis to use the =LAYOUT= program to create a graphi-

cal representation of the desired layout pattern. Theboard can then be simulated by creating an =EMPOWER=Simulation.

An alternative method is to construct a text netlist (TPLfile) description of the board layout and launch =EM-POWER= from a command prompt. For a description ofthe TPL file format, see Chapter 11.

This chapter describes how to use the =LAYOUT= pro-gram to construct a board layout and obtain an =EM-POWER= simulation. GENESYS is then used to displayand compare the linear simulation with the =EM-POWER= data.

CREATING A NEW LAYOUT

A board layout can be created one of two ways:

• By starting without a schematic

• By starting from an existing schematic

The first method starts in the GENESYS Environment bycreating a layout without an associated schematic. Thelayout is created by drawing lines and placing footprintsin the =LAYOUT= editor.

The second method begins in with a schematic and createsa board layout based on the schematic objects. Thismethod is normally used when a linear simulation (usingGENESYS) has been performed on a schematic and an=EMPOWER= simulation is desired, or when any lumpedelements are needed in the =EMPOWER= Simulation. Inaddition to the schematic objects, any desired =LAYOUT=objects can be added to the board before simulation. Forexample, linear simulation would normally not includeground pours, power supply rails, and lumped elementpads. However, these are included in the =EMPOWER=run, allowing inspection of their effects.

CREATING A LAYOUT WITHOUT A SCHEMATIC

The complete file from this example is LayoutOnly.WSP

This example demonstrates the following topics:

• Creating a layout without a schematic

• Choosing grid spacings

• Choosing the box size

A microstrip stub notch filter with a transmission zero at9.5 GHz is to be simulated. The filter has the followingspecifications:

• 15 mil RT/Duroid substrate (εr=2.2, tan δ=9×10-4)

• Copper metallization

• 50 Ω terminations

• The stub line should be 70 Ω and 90° at 9.5 GHz

The series lines and the stub dimensions were calculatedusing =T/LINE=, and were rounded to the nearest 5 milincrement. The final line dimensions are shown in Figure1-1.

2 Starting

Note: Before beginning this example, you should besure your Workspace Window is visible. Select Work-space Window from the View menu if necessary.

To begin, select New from the GENESYS File menu.Since we do not need a schematic for this circuit, we willdelete the schematic: In the workspace window, Right-click on “Sch1 (Schematic)” and Select Delete This De-sign". Next, we will create a layout. Right-click on“Designs” in the workspace window and select Add Lay-out from the =LAYOUT= menu. Enter “Stub” for thelayout name. The “Create New Layout” dialog appears.

The tabs and prompts on this dialog are described in detailin Chapter 2.

Note: For all dialog boxes, be sure that your screenlooks exactly like the boxes shown in the figures.

Figure 1-1 Dimensions for the stub notch filter.

Starting 3

Box Settings

Note: In =EMPOWER=, the layout’s box dimensions areused to define the bounding box.

The box dimensions are shown in Figure 1-2. Box Widthwas chosen as 425, the width of the filter since there aretwo 200 mil lines and a stub width of 25 mils. The filterheight is 275 mils, including the stub length and seriesline width. The box height was chosen as 600 mils to giveplenty of spacing on either side of the filter. This mini-mizes wall interference in the filter’s frequency response.

Grid Settings

The =EMPOWER= grid settings for this example areshown in the upper right in Figure 1-2. =EMPOWER=simulation time is greatly reduced if dimensions are cho-sen so that metal lies exactly on as large a grid size as

Figure 1-2 Dimensions tab showing =LAYOUT= settings for theexample filter.

4 Starting

possible. The grid width and height settings for this filterwere chosen as 12.5, since the filter dimensions (425×275)are exactly divisible by this value.

General Layers

The general layer settings for this example are shown inFigure 1-3.

Only three layers had to be defined for this filter:

• Top Metal

• Substrate

• Bottom Metal

These are the only layers that are needed to simulate themicrostrip filter. For a general layout, more layers areoften included for purposes only. For example, defining asilk screen or mask layer would not affect simulation sincenone of the filter metal is placed on those layers.

Figure 1-3 General Layers tab showing the =LAYOUT= layersused in the example filter.

Starting 5

Since the bottom of the box will be used as a ground plane,the bottom metal layer defined in Figure 1-3 is not neces-sary. However, since it is often necessary for real-worldboards, it is defined here.

Figure 1-4 General Layers tab showing the =EMPOWER= layersused in the example filter. (Scroll the table to see the second page.)

6 Starting

=EMPOWER= Layers

The =EMPOWER= layers for this example are shown inFigure 1-4. The =EMPOWER= layers are automaticallyselected from the available general layers (see the pre-vious section). They are chosen from the available metaland substrate layers, and can be enabled or disabled for=EMPOWER= simulation.

Since Air layers above and below a substrate are so com-mon, a special option has been given here to add them. Formore information on the individual layer options, seeChapter 2.

Notice that BOT METAL and Air Below are not enabledin Figure 1-4. This places the box bottom at the lowersubstrate boundary so that it acts as a ground plane.

Figure 1-5 =LAYOUT= editor before creating the example filter.

Starting 7

Note: In almost all cases where a completely solidground plane is used, you should use the top or bottomcover to simulate it. This is much more efficient thanusing an extra metal layer.

Click OK. The =LAYOUT= editor appears. The screenshould look like Figure 1-5.

Drawing the layout

To draw the series line:

1. Select the Rectangle button from the =LAYOUT= tool-bar This is the third button on the bottom toolbar.

2. Click on the left edge of the page border, and drag to-ward the right and down until the status bar showsDX=425 and DY=50.

3. Release the mouse button.

Figure 1-6 Layout showing series line for the example filter.

8 Starting

This is the series transmission line. The screen shouldnow look like Figure 1-6. Don’t worry if the line isn’t atthe exact same position on the page - the layout will becentered later.

To draw the open stub:

1. Select the Rectangle button from the toolbar.

2. Click at the bottom edge of the line just drawn,one grid cell left of the series line’s center.

3. Drag to the right and down until the status barshows DX=25 and DY=225.

4. Release the mouse button.

The screen should now look like Figure 1-7. If the stubline isn’t centered horizontally on the screen, select thestub by clicking on it, and drag it to the proper position.

Figure 1-7 Layout showing nearly completed example filter.

Starting 9

Centering the layout

As a general rule, =EMPOWER= simulation time isgreatly reduced if the circuit to be simulated exhibitssymmetry in any of several planes. Many circuits willexhibit some form of symmetry if they are centered in thepage area.

To center the example filter, choose Select All from theEdit menu. Then choose Center Selected On Page fromthe Layout menu.

Placing =EMPOWER= ports

Before running =EMPOWER=, the filter’s ports must bedesignated. Select the EMPort button ( ) on the =LAY-OUT= toolbar, and click on the center left end of the seriesline. The EM Port Properties dialog appears. Set thedrawing size to 25. This controls how large the ports will

Figure 1-8 Final example filter layout (zoomed view).

10 Starting

be drawn on the =LAYOUT= screen. Note that the defaultport number is 1. Select the OK button.

Next, select the EMPort button on the toolbar again.Click on the center right end of the series line. The EMPort Properties dialog appears. Again, type 25 into theDraw Size box. Note that the default port number is 2.Select the OK button.

The screen should now look like Figure 1-8.

For simulation, =EMPOWER= will take S-Parametersfrom these ports.

Simulating the Circuit

To run =EMPOWER=, you must create a simulation.Right-click on “Simulations/Data” in the WorkspaceWin-dow and choose Add Planar 3D EM Analysis from themenu. Accept “EM1" as the analysis name. This displaysthe =EMPOWER= Options dialog. This dialog is shown in

Figure 1-9 =EMPOWER= setup dialog showing =EMPOWER=settings for the notch filter example.

Starting 11

in Figure 1-9. For a description of the dialog options, seeChapter 4. For now, just set the prompts as shown inFigure 1-9.

We are starting with 3 sample points in the range 8-11GHz. This will place 1 point at 8, 9.5 (the supposedresonance), and 11 GHz.

Click the Recalculate Now button. This launches =EM-POWER= to simulate the layout.

Note: While =EMPOWER= is calculating, a window simi-lar to the one in Figure 1-10 will be shown. This windowshows the current status throughout the calculationmode. For more details on this window, see Chapter 10,Console Mode.

VIEWING EMPOWER RESULTS

After =EMPOWER= simulation of the layout, the datamust be displayed in GENESYS. This is done by creatinga Data Output such as a Rectangular Graph.

To create a rectangular graph in this workspace:

Figure 1-10 Sample =EMPOWER= screen during calculation.

12 Starting

1. Right-click on Outputs in the Workspace Windowand Select Add Rectangular Graph from themenu. Accept the default name “Graph1".

2. Select EM1.Stub for “Default Simulation/Data orEquations”.

3. Enter S21 for the first measurement and S11 forthe second measurement.

This instructs GENESYS to display a window containing=EMPOWER= data S21 and S11 will be displayed at 8000,9500, and 11000 MHz. (For a complete description ofrectangular graphs, see the GENESYS User’s Guide.)

Figure 1-11 shows the GENESYS display after the =EM-POWER= run with 3 sample points.

In this response, the notch frequency appears to occurexactly at 9.5 GHz. Or does it? Let’s add some frequencypoints to the =EMPOWER= simulation.

To re-simulate, adding more points:

Figure 1-11 Notch filter simulation with 3 =EMPOWER=calculation points.

Starting 13

1. Double-click “EM1" under Simulations/Data in theWorkspace Window.

2. Change the “Number of Points:” prompt in the “Elec-tromagnetic Simulation Frequencies” to 11.

3. Click the Recalculate Now button.

4. Close the =EMPOWER= log (click on the X in the up-per-right corner of the window).

This will add to the previous =EMPOWER= simulation sothat we have 11 instead of 3 data points. =EMPOWER=will intelligently recalculate only the additional points.

Figure 1-12 shows the simulation with 11 =EMPOWER=data points. The notch frequency now appears to be at 9.2GHz! Let’s add the full 31 points to the =EMPOWER=simulation to ensure that we get the actual notch fre-quency. Repeat the previous steps to change the numberof =EMPOWER= points to 31 and recalculate.

Figure 1-13 shows the display after the =EMPOWER= runwith 31 points. The response has not changed noticeably

Figure 1-12 Notch filter simulation with 11 =EMPOWER=calculation points.

14 Starting

since the 11 point simulation, so we must have found thecorrect notch frequency.

For the example filter, the notch occurs at 9.2 GHz insteadof the desired 9.5 GHz. Much of this shift is due torounding the line dimensions to the nearest 5 mils.

Using the Viewer

Once the =EMPOWER= run is completed, the viewer canbe loaded if Generate Viewer Data was selected in the=EMPOWER= options dialog. Generating this data slowsthe =EMPOWER= simulation, so it’s usually only checkedduring last-run simulations.

Right-Click the =EMPOWER= simulation in the Work-space Window and select Run Viewer. Figure 1-14 showsthe viewer screen for this example. A top-down view hasbeen selected, and the notch frequency has been specified.Port #1 is at the left of the image, and port #2 is at theright. The plot is color-coded to the scale given in the lowerleft of the figure. Notice that port #2 is nearly black. This

Figure 1-13 Notch filter simulation with 31 =EMPOWER=calculation points.

Starting 15

indicates that very little energy is being delivered to thatport at 9.2 GHz, as we’d expect.

CREATING A LAYOUT FROM AN EXISTINGSCHEMATIC

The file used in this example is EAGLE\EXAM-PLES\TUNEBP.WSP. This example demonstrates thefollowing topics:

• Creating a layout from an existing schematic

• “Tuning” with =EMPOWER= data

• Using lumped elements with =EMPOWER=

This circuit is a tunable bandpass filter. Operationaltheory is given in Chapter 9.

Figure 1-14 =EMPOWER= viewer showing very little energytransfer at the notch frequency.

16 Starting

In GENESYS, select Open from the File menu. Then,select TUNEBP.WSP from the EAGLE\EXAMPLES di-rectory. Double-Click F2000 under Designs in the Work-space Window to display the schematic for this filter(shown in Figure 1-15).

This is the schematic of a 2nd order microstrip comblinebandpass filter with 50 Ω terminations and transformercoupling on the input and output. The lumped capacitorsare gang-tuned to adjust the resonant frequency of the twocenter lines. Tuning in this manner affects only the centerfrequency, and keeps the passband bandwidth constant.

Double-Click Layout1 under Designs in the WorkspaceWindow to display the layout for this schematic. Thelayout for this example is shown in Figure 1-16.

A 0402 Chip Capacitor footprint has been used for each ofthe lumped capacitors. Whenever a lumped element is

Figure 1-15 Schematic for the tunable filter example.

Starting 17

used for an =EMPOWER= run, GENESYS creates aninternal ports for the element. These ports are placed:

• If “Use Planar Ports for two-port elements” is checked inthe =EMPOWER= properties box, one port is created for2-terminal elements (like resistors or capacitors) which arealigned horizontally or vertically.

• In all other cases, an internal port is used for each terminalof the element. This port is placed at the center of the padfootprint, and =EMPOWER= writes data for each portcreated, whether internal or external.

The “1" and ”2" ports pictured in Figure 1-16 are examplesof external ports. External ports are described in Chapter4. Internal ports are described in Chapter 6.

This is a powerful technique, since real time tuning canbe employed in GENESYS once the =EMPOWER= datafor has been calculated.

Figure 1-16 Layout for the tunable filter example schematicshown in Figure 9.

18 Starting

Simulating The Circuit

Double-click EM1 in the Workspace Window. This dis-plays the =EMPOWER= Options dialog (shown in Figure1-17).

Click the Recalculate Now button. If anything has beenmodified since the last =EMPOWER= run, this launches=EMPOWER= to simulate the layout in Figure 1-16.

Note: =EMPOWER= has been given a lot of intelligenceto determine when it needs to calculate. Clicking Recal-culate Now will not do anything if =EMPOWER= believesit is up to date. To force =EMPOWER= to recalculatefrom scratch, right-click on the electromagnetic simula-tion in the workspace window and select “Delete allinternal files”.

Once =EMPOWER= calculation is completed, GENESYSdisplays the calculated data. Figure 1-18 shows the

Figure 1-17 Options tab for the =EMPOWER= Setup Dialog.

Starting 19

GENESYS graphs after =EMPOWER= simulation. (Dou-ble-click the graph items in the workspace window to openthem and select Tile Vertical from the Window menu toorganize them.)

Lumped Elements

The first example in this chapter required several datapoints to pin down the notch frequency. This second ex-ample only used 4 data points, and produces data veryclose to the =SuperStar= simulation. This is because thecapacitors which load the coupled lines (causing reso-nances at the center frequency) were removed during the=EMPOWER= simulation. This effectively removes theresonances from the simulation range, producing a flatresponse from the open coupled lines. Since a flat re-sponse is well suited for linear interpolation, few datapoints are required in the =EMPOWER= simulation. Inthe =EMPOWER= options dialog, the Co-SimulationSweep box is used to set up a simulation with more pointsafter lumped elements are added.

When GENESYS uses the =EMPOWER= results, it re-places the lumped capacitances, resulting in the bandpassresponse shown in Figure 1-18.

Tuning With =EMPOWER= Data

As stated before, GENESYS creates ports internal to alayout structure containing lumped elements before in-voking =EMPOWER=. During calculation, =EM-POWER= creates s-parameter data with port data for allports, whether internal or external. This allowsGENESYS to tune the lumped elements while still usingthe =EMPOWER= data.

To see an example of tuning:

1. Click inside the C2000 prompt in the Tune Window.

20 Starting

2. Type a new value for the capacitor, or tune usingPage Up/Page Down keys or the spin buttons.

Figure 1-19 shows the GENESYS screen after tuning thecapacitors from 0.55 pF to 1.2 pF. The response shown onthe left in this figure is the =SuperStar= linear simulationresponse. The =EMPOWER= data is combined with thelumped elements in the rightmost response.

The =EMPOWER= viewer would not lend much insight tothis example since the lumped elements are not includedin the =EMPOWER= data. Therefore, no viewer data wasgenerated for the example.

In this chapter, the methods for creating a layout for an=EMPOWER= simulation were illustrated with two ex-

Figure 1-18 GENESYS screen showing the linear and=EMPOWER= simulated data for the combline filter example.

Starting 21

amples. For most cases, it will be easier and more conven-ient to create a layout if a schematic is drawn first.

In some cases (e.g. when using lumped elements), a sche-matic provides functionality not available for a =LAY-OUT= only approach.

Figure 1-19 GENESYS screen showing linear and =EMPOWER=data after tuning the loading capacitors.

22 Starting

Chapter 2

=EMPOWER= Basics

A major part of any electromagnetic simulation is tobreak the problem down into manageable size piecesthat allow an approximation of Maxwell’s equations

to be solved. Electromagnetic simulators traditionally fallinto three major categories: 2-D, 3-D, and 2-1/2-D.

2-D SIMULATORS

2-D simulators can only analyze problems that are infi-nitely continuous in one direction. Ideal transmissionlines and some waveguide problems are practical prob-lems which fall into this category. A 2-D simulator willanalyze a slice of the line(s) and determine propagation,impedance, and coupling values. 2-D simulators are thefastest but most limited type of simulator available.

3-D SIMULATORS

3-D simulators can analyze virtually any type of problemand are ideal for use with non-planar geometries such asa coaxial T-junction, radar target reflections, or other trulythree dimensional problems. 3-D simulators have theadvantage that they can analyze almost any problem, butthey have the disadvantage that they are extremely slow.

2 1/2-D SIMULATORS

2 1/2-D Simulators are simulators designed for mainlyplanar (microstrip, stripline, etc.) circuits. While they

have less flexibility than true 3-D simulators, they aremuch faster and are ideally suited for microstrip, stripline,and other similar geometries. =EMPOWER= is an ad-vanced 2 1/2-D simulator. It can solve planar problems aswell as problems with via holes and other z-directed cur-rents, putting it in a class above true 2 1/2-D simulatorswhich do not allow z-directed currents. In fact, mostpeople would consider =EMPOWER= to be a 3-D simula-tor because it can handle z-directed currents.

BASIC GEOMETRY

All circuits in =EMPOWER= exist in a rectangular box.(shown in Figure 2-1). The Media (substrate) layers eachhave specific dielectric and permittivity constants and losstangents. There must be at least two media layers: Oneabove the metallization layer and one below. For standardmicrostrip, there is a substrate below and air above. Forsuspended microstrip, there are three media layers (twoair and one substrate). For buried microstrip, there arealso three media layers (two substrate and one air).

z

x

y

h

h...

h

2

1

P

εε

ε

1

2

P

TOP_W

BOTTOM_W

SIDEWALLS

µ

µ

µ

1

2

P

a

b

(MEDIA) LAYER

Figure 2-1 Basic =EMPOWER= geometry. These parametersare modified from the Layer and Dimension tabs in the =LAYOUT=File/Preferences dialog box.

24 =EMPOWER= Basics

Figure 2-2 shows two typical =EMPOWER= Layer Tabsetups: one for microstrip and one for stripline (triplate).The =EMPOWER= Layer Tab must be carefully checkedwhen a new problem is created, as it is probably the mostlikely source of errors when setting up an =EMPOWER=run.

Note: Complete reference for =EMPOWER= layers is inthe Dialog Boxes section of the reference manual.

Figure 2-2 Two typical =EMPOWER= Layer Tab setups. The oneon the top is for microstrip, and the one on the bottom is for stripline(Triplate).

=EMPOWER= Basics 25

The =EMPOWER= Layer Tab consists of the followingmain entries:

Top Cover and Bottom Cover - Describes the top andbottom covers (ground planes) of the circuit:

• Lossless: The cover is ideal metal.

• Physical Desc: The cover is lossy. These losses aredescribed by Rho (resistivity relative to copper), Thickness,and Surface Roughness.

• Electrical Desc: The cover is lossy and is described by animpedance or file. See the description below under metalfor more information.

• Semi-Infinite Waveguide: There is no cover, and the circuitis simulated as if the box walls and uppermost substrate/airlayer extend up or down forever (an infinite tube).

• Magnetic Wall: The cover is an ideal magnetic wall. Thissetting is only used in advanced applications.

• =SCHEMAX= substrates: Choosing a =SCHEMAX=substrate causes the cover to get the rho, thickness, androughness parameters from that substrate definition. Werecommend using this setting whenever possible so thatparameters do not need to be duplicated in =SCHEMAX=and =LAYOUT=.

Air Above and Air Below - The presence of air at the topof the box (as in microstrip) or the bottom of the box (as insuspended microstrip) is so common that special entrieshave been provided for these cases. Checking the box toturn these layers on is the equivalent of adding a substratelayer with Er=1, Ur=1, and Height (in units specified inthe Dimensions tab) as specified.

CAUTION: When setting up a new circuit, be sure tocheck the height of the air above, as it is often the onlyparameter on this tab which must be changed, and istherefore easily forgotten.

26 =EMPOWER= Basics

Metal Layers - In =LAYOUT=, multiple METAL layers(e.g., copper and resistive film) are automatically con-verted to one =EMPOWER= signal layer if no media layeris in between the metal layers.

All metal layers from the General Layer Tab are alsoshown in the =EMPOWER= Layer tab. These layers areused for metal and other conductive material such asresistive film. The following types are available:

• Lossless: The layer is ideal metal.

• Physical Desc: The layer is lossy. These losses aredescribed by Rho (resistivity relative to copper), Thickness,and Surface Roughness.

• Electrical Desc: The layer is lossy and is described by animpedance or file. This type is commonly used for resistivefilms and superconductors. If the entry in this box is anumber, it specifies the impedance of the material in ohmsper square. If the entry in this box is a filename, it specifiesthe name of a one-port data file which contains impedancedata versus frequency. This data file will beinterpolated/extrapolated as necessary. See theReference manual for a description of one-port data files.

• =SCHEMAX= substrates: Choosing a =SCHEMAX=substrate causes the layer to get the rho, thickness, androughness parameters from that substrate definition. Werecommend using this setting whenever possible so thatparameters do not need to be duplicated in =SCHEMAX=and =LAYOUT=.

CAUTION: Thickness is only used for calculation oflosses. It is not otherwise used, and all strips arecalculated as if they are infinitely thin.

Substrate/Media Layers - All substrate layers from theGeneral Layer Tab are also shown in the =EMPOWER=Layer tab. These layers are used for substrate and othercontinuous materials such as absorbers inside the top

=EMPOWER= Basics 27

cover. An unlimited number of substrate/media layers canbe used. The following types are available:

• Physical Desc: The layer is lossy. These losses aredescribed by Height (in units specified in the Dimensionstab), Er (relative dielectric constant), Ur (relative permittivityconstant, normally 1), and Tand (Loss Tangent).

• =SCHEMAX= substrates: Choosing a =SCHEMAX=substrate causes the cover to get the height, Er, Ur, andTand parameters from that substrate definition. Werecommend using this setting whenever possible so thatparameters do not need to be duplicated in =SCHEMAX=and =LAYOUT=.

CAUTION: For true stripline (triplate), be sure to checkthe “Use 1/2 Height” checkbox if you are using a sub-strate from =SCHEMAX=. This forces =EMPOWER= touse 1/2 of the =SCHEMAX= substrate height for eachsubstrate (above and below) so that the total height forboth media layers is correct.

In addition to the metallization and substrate layers,viaholes and other z-directed currents can be used. Thesecurrents can go from the metallization layer through onemedia/air layer to either the top or bottom walls.

Besides conductive materials, ports are placed on themetal layers and in z-directed positions.

THE GRID

All conductive surfaces and ports must be on a grid. Thisgrid is composed of regular rectangular cells. Figure 2-3shows an example of mapping a microstrip bend to thegrid. The left half of the figure shows the circuit as itappears in =LAYOUT=. The right half of the circuit showsa part of the =EMPOWER= listing file. Each of the plussigns (“+”) in the listing file represents an intersection oftwo grid lines as shown on the layout. Lines connecting

28 =EMPOWER= Basics

plus signs represent metal. Numbers represent port loca-tions. Notice that the ports map onto the grid in place ofmetal, so the ports go between the end of the line andground (the wall), so each port has a ground reference aswould be expected.

=EMPOWER= will move all surfaces to the nearest gridcell before analyzing a circuit. =EMPOWER= maps thestructure onto the borders of the cell, not onto the spaceinside the cell. Figure 2-4 shows a slightly more complexexample which does not exactly fit the grid. There arethree important things to notice in this figure: 1) The stubline going up is about 2 1/2 cells wide, but is approximatedby =EMPOWER= as being 2 cells wide. 2) The chamferedcorner is approximated by a “stairstep.” 3) The viahole

Figure 2-3 An example of mapping a microstrip bend to the grid.Notice that even the ports are mapped on the grid (shown asnumbers).

0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 515+−−−−−−−−−−−−−−−−−−−−−−−−−−−−−+

| |14| + + + + + + + + + + + + + + |

| |13| + + + + + + + + + + + + + + |

| |12| + + + + + + + + + + + + + + |

| |11| + + + + + + + + + + + + + + |

| |10| + + + + + + + + + + + + + + |

| |9|1+-+-+-+-+-+-+-+-+ + + + + + |

| | | | | | | | | | |8|1+-+-+-+-+-+-+-+-+ + + + + + |

| | | | | | | | | | |7|1+-+-+-+-+-+-+-+-+ + + + + + |

| | | | | | | | | | |6|1+-+-+-+-+-+-+-+-+ + + + + + |

| | | | | |5| + + + + + +-+-+-+ + + + + + |

| | | | | |4| + + + + + +-+-+-+ + + + + + |

| | | | | |3| + + + + + +-+-+-+ + + + + + |

| | | | | |2| + + + + + +-+-+-+ + + + + + |

| | | | | |1| + + + + + +-+-+-+ + + + + + |

| 2 2 2 2 |0+−−−−−−−−−−−−−−−−−−−−−−−−−−−−−+

1

2

=EMPOWER= Basics 29

near the end of the stub is represented by an asterisk inthe listing.

Figure 2-5 shows a closeup where you can see how metaland ports are mapped onto the borders of the cells. Thepresence of metal or conductors along the grid causes=EMPOWER= to close the connections along the grid. Thepresence of an EMPort causes the line to be opened,creating an open circuit which turns into a port in the finaldata file.

NOTE: It is possible to make a line so narrow that it mapsto one border between cells (zero cells wide). This islegal, but is not normally recommended and should beused only for very high impedance lines where accuracyis not important, such as DC power lines.

Figure 2-4 A more complex mapping example.

0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 515+−−−−−−−−−−−−−−−−−−−−−−−−−−−−−+

| |14| + + +-+-+ + + + + + + + + + |

| | | | |13| + + +-*-+ + + + + + + + + + |

| | | | |12| + + +-+-+ + + + + + + + + + |

| | | | |11| + + +-+-+ + + + + + + + + + |

| | | | |10| + + +-+-+ + + + + + + + + + |

| | | | |9|1+-+-+-+-+-+-+ + + + + + + + |

| | | | | | | | |8|1+-+-+-+-+-+-+-+ + + + + + + |

| | | | | | | | | |7|1+-+-+-+-+-+-+-+-+ + + + + + |

| | | | | | | | | | |6|1+-+-+-+-+-+-+-+-+ + + + + + |

| | | | | |5| + + + + + +-+-+-+ + + + + + |

| | | | | |4| + + + + + +-+-+-+ + + + + + |

| | | | | |3| + + + + + +-+-+-+ + + + + + |

| | | | | |2| + + + + + +-+-+-+ + + + + + |

| | | | | |1| + + + + + +-+-+-+ + + + + + |

| 2 2 2 2 |0+−−−−−−−−−−−−−−−−−−−−−−−−−−−−−+

1

2

30 =EMPOWER= Basics

The grid and the box are controlled with parameters inthe Preferences box from the =LAYOUT= File menu. Fig-ure 2-6 shows the Dimensions Tab as it was setup for themicrostrip bend in Figure 2-3. The following entries areespecially relevant to =EMPOWER=:

Show =EMPOWER= Grid - Turning on this checkboxforces =LAYOUT= to display the rectangular =EM-POWER= grid. It also allows different grid spacings in theX and Y dimensions. It is strongly recommended to turnthis checkbox on whenever you are creating a layout for=EMPOWER=.

Grid Spacing X and Grid Spacing Y - These control the cellsize for the =EMPOWER= run as well as the grid snapfeature in =LAYOUT=. When using the “=EMPOWER=Grid Style,” there will be =LAYOUT= snap points betweeneach grid line which allow lines to be centered betweentwo grid points if necessary. They are often referred to asdx and dy and should be small with respect to a wave-length at the maximum frequency to be analyzed, prefer-ably less than λ/20 and always less than λ/10. These

x

y

Ex Jx

Ex Jx

Ey

Jy

Ey

Jy

Vx Ix

Vx Ix

Vy

Iy

Vy

Iydy

dx

a) b)

z

Ez

Jz

VzIz

Figure 2-5 A closeup of two grid cells. A) Mapping of metal ontothe grid. Notice the metal is mapped onto the borders of the cells.B) Mapping of a port onto the grid. Normally for a port, terminalsin only one direction are used (Vx & Ix, Vy & Iy, or Vz & Iz).

=EMPOWER= Basics 31

parameters correspond directly to the DELTA statementin the TPL file.

Box Width and Box Height - These are the box size for=EMPOWER= simulation. In Figure 2-1, “a” is the BoxWidth and “b” is the Box Height. They correspond directlyto the SIZE statement in the TPL file. The number of cellsacross the box (equal to Width or Height divided by GridSpacing X or Y) is displayed for your convenience and canbe changed to adjust the page width. Note: Any metal putdown completely outside the box will be ignored by =EM-POWER=. This can be used to your advantage to tempo-rarily or permanently remove metal or components fromthe =EMPOWER= simulation.

Default Viahole Layers - The Start Layer and End Layercombo boxes control the default layers for the viaholes.These can be overridden individually for each viahole.

Figure 2-6 The Dimensions Tab as setup for the Microstrip Bendin Figure 2-3.

32 =EMPOWER= Basics

VIAHOLES AND Z-DIRECTED PORTS

The grid in =EMPOWER= is a truly three dimensionalgrid (rectangular lattice). Z-Directed currents and portsare mapped from the intersection points to the top orbottom cover as shown in Figure 2-5. There are twocaveats: metal and ports in the z-direction are modeled asone continuous current, so the viaholes should be small incomparison with a wavelength. Also, you cannot haveboth a port and metal along the same grid line, so youshould be extremely careful when placing a viahole directlyunderneath an internal port. You should check the listingfile (select “Show Listing File” from the =EMPOWER=simulation right-click menu) carefully to see that both theport and the viahole are represented on the grid.

The physical length of a viahole in a substrate should bekept shorter than about 1/10 to 1/20 wavelength withinthe analysis range. Longer lengths can suffer calculationinaccuracies in =EMPOWER=. For example, suppose amicrostrip circuit with a 10 mil substrate and a dielectricconstant of 2.4 is to be used. What is the highest accuratefrequency for this setup?

Note: If the subst rate layer is broken down int o twosubstrat e layers (by addin g an additiona l layer) ,each 1/2the heigh t of the original , then the viahole s wil l beaccurat e at twic e the origina l frequen cy. This procedurecan be repeate d as necessar y.

ε01288542 10= × −. F

m

µ0612566 10= × −. H

m

cr

ms= = = ×1 1

21935 10

0 0 0 0

8

ε ε µ ε µ.4.

=EMPOWER= Basics 33

EMPORTS

All circuits must contain at least one EMPort to allow datato be taken from the =EMPOWER= simulation. Thenumber of ports is equal to the number of ports in theEMPOWER network to be analyzed. They are placed inthe layout using the EMPort button and can be Normaldeembedded external ports (gray), external ports with NoDeembedding (white), or internal ports (white). Externalports are discussed in detail in Chapter 4, and LumpedElements and Internal Ports are discussed in Chapter 6.

=EMPOWER= OPTIONS DIALOG BOX

For complete details on the =EMPOWER= Options Dialogbox, see your Reference Manual or online help.

λ λ15

10 2 54 10 4 381 10 3= = × − ⇒ = × −mils m m. .

fc

max

.

.= = ×

×≈−λ

1935 10

3 81 1050

8

3 mGHz

ms

34 =EMPOWER= Basics

Chapter 3

Tips

Often, electromagnetic simulation involves tradeoffsand compromises to keep simulation times andmemory requirements as small as possible while

making accuracy as high as possible. This chapter looksat several choices and clarifies the tradeoffs. Table 3-1lists various features and gives their impact on simulationtimes, accuracy, and memory requirements. Each of thesechoices are looked at in detail below. The values areapproximate and may vary.

Choice Memory Accuracy TimeReducing Cell Size by 2 x16 x2 x64Raising Max Critical Freq x1.5 x1.5 x2Fixing Symmetry ÷4-÷16 0 ÷4-÷16Turning Off Thinning Out x16 x1.2 x64Increasing Wall & Cover Spac-ing

x1.1 x1.5 x1.1

Choosing Correct Cover 0 x4 0Including Loss x2 Lossy x4Generating Viewer Data x1.2 0 x2-x10Correcting Slot-type Structure ÷64 0 ÷256Using Preferred Box Cell Count 0 0 ÷1.1-÷10Using Thick Metal x6 x16

Table 3-1 The approximate impact of various choices on memoryrequirements, accuracy, and simulation time.

CELL SIZE

Cells should be small enough so that the result is accurate,at least 10 cells per wavelength at the maximum criticalfrequency (see below). Additionally, the cells should besmall enough that there is at least one, and preferablymore, cell across every line and gap. Decreasing the cellsize makes all stages of the solution take longer, so de-creasing cell size can be an expensive way to get moreaccuracy. Conversely, increasing cell size is a great way todo an initial run of your problem to make sure that theresult is close before you start a simulation that will takehours.

See Chapter 2 for more information on cells and theproblem geometry.

MAXIMUM CRITICAL FREQUENCY

This parameter is set in the =EMPOWER= dialog boxwhen starting a simulation. Changing this parameter hasthree (and only three) effects:

1) The maximum amount of thinning out is affected.=EMPOWER= will thin out until an area is 1/20th of awavelength at this frequency in the default thinningmode.

2) The length of line analyzed for deembedding is 1/2wavelength at this frequency in automatic mode.

3) Many parameters in the listing file are based on thisfrequency.

The most important thing to know about maximum criti-cal frequency is to keep it the same between runs of thesame problem, even if you are changing the frequencyrange which you are analyzing. If it is changed, then thethinning out is changed, and the entire problem geometryis slightly different. As an example, if you are analyzing

36 Tips

a filter with a passband from 5.1 to 5.5 GHz with areentrance mode (additional passband) around 15 GHz,you should probably set the maximum critical frequencyto 5.5 GHz. This is because the exact characteristic of thereentrance mode probably is not important (“critical”); youjust want to know approximately where the filter re-en-ters. On the other hand, you want to know precisely wherethe passband is, so you set the maximum critical frequencyabove it.

The effect of maximum critical frequency is generallysecondary. Most of the other choices in Table 3-1 have abigger effect on accuracy.

SYMMETRY

Making a problem exactly symmetrical is an easy way tomake a problem require less memory and time withoutsacrificing any accuracy. There are four types of symmetryrecognized by =EMPOWER=: YZ mirror symmetry, XZmirror symmetry, two mirror symmetry, and 180o rota-tional symmetry. These types are illustrated in Figure 3-1.

When =EMPOWER= is running, you should look at theinformation area at the top of the screen to see if symmetryis active. If it is not, recheck your problem to see if it isexactly centered on the box, and to see if it is in factsymmetrical. Two tools can help with this:

1) Using “Center Selected on Page” from the Edit menu in=LAYOUT=. This command makes it easy to make surethat your circuit is exactly centered on the page

2) Showing the listing file by selecting “Show Listing File”from the =EMPOWER= right-click menu. This file showsexactly how the problem was put on the grid, and lack ofsymmetry is often obvious.

Tips 37

Making an unsymmetrical problem symmetrical willmake it run 4 times faster in most cases, and will make it16 times faster if your problem can use two-mirror sym-metry.

See Chapter 2 for more information on cells and theproblem geometry. See Appendix C for more informationon the listing file.

THINNING OUT

For most examples, the default thinning out should beused. As a general rule, you will get better accuracy for agiven amount of time and memory when you use thinning.Thinning out helps by removing currents which have littleor no effect. This reduces the number of variables in theproblem considerably with little effect on the accuracy ofthe solution. There are a few cases where thinning outshould not be used, and they generally involve very large

YZ Mirror

Two Mirror

XZ Mirror

Rotational

Figure 3-1 Types of symmetry recognized by =EMPOWER=

38 Tips

sections of metal which are affected too much by thinningout.

WALL & COVER SPACING

Generally, the wall and cover spacing should match theproblem which you are trying to model. This will give anaccurate assessment not only of circuit performance butalso of box resonances. However, this will not be possiblein a few situations:

1) The structure will not be in a box.

2) You are analyzing part of a larger circuit, and the boxwalls would be prohibitively large to model.

3) You are designing a component (such as a spiral induc-tor) which will be reused in many different circuits, so thecover height is not known.

In these cases, you must use an approximation. Set thebox size so that the walls are separated from the circuitby at least 3 times the substrate thickness, preferably 6times. For microstrip, set the cover spacing (air above) to5 to 10 times the substrate height.

See Chapter 8 for a discussion of Box Modes.

COVER TYPE

Choosing the correct cover type is absolutely critical togetting an analysis which matches measured results. Thechoice is usually between whether to use an open cover ora closed cover. Choosing the correct cover type usually hasno effect on analysis time, so there is no reason not to setthis to the proper type. With an open cover, there will beradiation, and this can have a huge impact on circuit

Tips 39

performance. You can choose the correct cover types in theLayers Tab when starting an =EMPOWER= run.

See Chapters 2 and 8 for more information on covers. SeeExample 3 in Chapter 9 for an example of the impact thatremoving a cover has on circuit performance.

LOSSY ANALYSIS

If you do not need information about circuit loss, you cancheck the box labeled “Don’t use physical loss (Faster)”when starting an =EMPOWER= run. Turning off losseswill generally make a problem require 1/2 the memory and1/4 the time as a lossy problem.

We recommend that you define all layers with their propercharacteristics, including losses. You can then quicklychange between lossy and lossless modes as describedabove. A common technique is to analyze a circuit firstwithout losses, then turn on losses and run an analysiswith a few points in it. This allows you to determine theamount of loss and confirm that it has no other majoreffect on performance while not having to wait the addi-tional time while doing most of your analyses.

There is an additional caveat regarding loss described inthe section on Slot-type structure.

VIEWER DATA

Looking at currents in the viewer is a great way to getinsights into circuit performance. However, generatingthis viewer data requires additional time, increasing thelength of a run by a factor from two to ten, and sometimesrequiring additional memory also. Generating viewerdata has no effect whatsoever on the solution given, so youshould not have this option turned on unless you actually

40 Tips

intend to run the viewer. You can turn this option on andoff by using the checkbox labeled “Generate Viewer Data(Slower)” when starting an =EMPOWER= run.

You will not normally need viewer data, and when it isneeded, you will not normally need viewer data at everyfrequency. Our recommendation:

1) Run all problems the first time without generatingviewer data. If the answer is completely unexpected,check for errors in your description of the file. This cansave a lot of time in the experimenting stage.

2) If you decide you want viewer data, open the =EM-POWER= Options dialog box. Reduce the number of fre-quency points to be analyzed and turn on “GenerateViewer Data (Slower).” Recalculate the =EMPOWER=simulation, and you will now have viewer data at somepoints.

3) If your problem is very large, you may want to increasethe cell size or make other tradeoffs to reduce the timerequired for calculation. If you use this technique, savethe file with a new name before you generate viewer dataso that you do not corrupt your existing S-Parameter data!

See Chapter 7 for more information on running the =EM-POWER= viewer.

SLOT-TYPE STRUCTURE

In the normal mode =EMPOWER= solves for the currentsin the metal. There is an additional mode where =EM-POWER= solves for the voltages in the gaps and in lossymetals. This mode must be turned on manually by check-ing “Slot-type structure” when starting an =EMPOWER=run from GENESYS, or by using the VOLTAGE keywordwhen describing a LAYER in a TPL file.

Tips 41

In general, you should check “Slot-type structure” when-ever the metallization layer has more lossless metal thanopen space. This is often the case in a slot-type structuresuch as coplanar waveguide. The answer will always beidentical, but you will save orders of magnitude of memoryand simulation time by ensuring that this checkbox is setto the right value. Note: This setting has no effect onz-directed metal (viaholes, etc.) which is always calculatedas currents.

There is a caveat when describing lossy problems with thisoption: All non-ideal metal must be analyzed, so if themetal in your problem is lossy, turning on “Slot-type struc-ture” will result in both the air and the metal beinganalyzed, which will have a disastrous effect on memoryand time requirements. Be sure that your metal layers areset to lossless if you check the slot-type structure box.

PREFERRED BOX CELL COUNT

The first part of an =EMPOWER= run involves takingFourier Transforms of the grid. These transforms will runmuch faster if the number of cells along the each side ofthe box is of the form 2a3b5c7d11e13f where e and f areeither zero or one, and a,b,c, and d are arbitrary integers.In other words, a circuit with a box 512 cells by 512 cells(28 by 28) will analyze much faster than a circuit with abox 509 cells by 509 cells (509 is prime). Making one sidea preferred number will help, so a box 509 x 512 cells isbetter than one 509 x 509 cells. Note that only the timewhile =EMPOWER= is working on the Fourier Transformis affected, and this is normally only substantial withboxes 100x100 or larger. If you see a status with “FFT” inthe message for a long time, check to see that the box widthand height are a preferred number of cells across. Pre-ferred numbers (which fit the form given above) 10000 andbelow are:

42 Tips

1 2 3 4 5 6 7 8 9 10 11 12 13 14 1516 18 20 21 22 24 25 26 27 28 30 32 33 3536 39 40 42 44 45 48 49 50 52 54 55 56 6063 64 65 66 70 72 75 77 78 80 81 84 88 9091 96 98 99 100 104 105 108 110 112 117 120 125126 128 130 132 135 140 143 144 147 150 154 156160 162 165 168 175 176 180 182 189 192 195 196198 200 208 210 216 220 224 225 231 234 240 243245 250 252 256 260 264 270 273 275 280 286 288294 297 300 308 312 315 320 324 325 330 336 343350 351 352 360 364 375 378 384 385 390 392 396400 405 416 420 429 432 440 441 448 450 455 462468 480 486 490 495 500 504 512 520 525 528 539540 546 550 560 567 572 576 585 588 594 600 616624 625 630 637 640 648 650 660 672 675 686 693700 702 704 715 720 728 729 735 750 756 768 770780 784 792 800 810 819 825 832 840 858 864 875880 882 891 896 900 910 924 936 945 960 972 975980 990 1000 1001 1008 1024 1029 1040 1050 10531056 1078 1080 1092 1100 1120 1125 1134 1144 11521155 1170 1176 1188 1200 1215 1225 1232 1248 12501260 1274 1280 1287 1296 1300 1320 1323 1344 13501365 1372 1375 1386 1400 1404 1408 1430 1440 14561458 1470 1485 1500 1512 1536 1540 1560 1568 15751584 1600 1617 1620 1625 1638 1650 1664 1680 17011715 1716 1728 1750 1755 1760 1764 1782 1792 18001820 1848 1872 1875 1890 1911 1920 1925 1944 19501960 1980 2000 2002 2016 2025 2048 2058 2079 20802100 2106 2112 2145 2156 2160 2184 2187 2200 22052240 2250 2268 2275 2288 2304 2310 2340 2352 23762400 2401 2430 2450 2457 2464 2475 2496 2500 25202548 2560 2574 2592 2600 2625 2640 2646 2673 26882695 2700 2730 2744 2750 2772 2800 2808 2816 28352860 2880 2912 2916 2925 2940 2970 3000 3003 30243072 3080 3087 3120 3125 3136 3150 3159 3168 31853200 3234 3240 3250 3276 3300 3328 3360 3375 34023430 3432 3456 3465 3500 3510 3520 3528 3564 35753584 3600 3640 3645 3675 3696 3744 3750 3773 37803822 3840 3850 3861 3888 3900 3920 3960 3969 40004004 4032 4050 4095 4096 4116 4125 4158 4160 4200

Tips 43

4212 4224 4290 4312 4320 4368 4374 4375 4400 44104455 4459 4480 4500 4536 4550 4576 4608 4620 46804704 4725 4752 4800 4802 4851 4860 4875 4900 49144928 4950 4992 5000 5005 5040 5096 5103 5120 51455148 5184 5200 5250 5265 5280 5292 5346 5376 53905400 5460 5488 5500 5544 5600 5616 5625 5632 56705720 5733 5760 5775 5824 5832 5850 5880 5940 60006006 6048 6075 6125 6144 6160 6174 6237 6240 62506272 6300 6318 6336 6370 6400 6435 6468 6480 65006552 6561 6600 6615 6656 6720 6750 6804 6825 68606864 6875 6912 6930 7000 7007 7020 7040 7056 71287150 7168 7200 7203 7280 7290 7350 7371 7392 74257488 7500 7546 7560 7644 7680 7700 7722 7776 78007840 7875 7920 7938 8000 8008 8019 8064 8085 81008125 8190 8192 8232 8250 8316 8320 8400 8424 84488505 8575 8580 8624 8640 8736 8748 8750 8775 88008820 8910 8918 8960 9000 9009 9072 9100 9152 92169240 9261 9360 9375 9408 9450 9477 9504 9555 96009604 9625 9702 9720 9750 9800 9828 9856 9900 998410000

THICK METAL

Using “Thick Up” or “Thick Down” metal will greatlyincrease the complexity of an =EMPOWER= run, as allmetal layers must be duplicated (for the top and bottomof the thick metal) and z-directed currents must be addedalong the sides of all metal.

44 Tips

Chapter 4

External Ports

Every =EMPOWER= circuit must contain at least oneport. These ports are divided into two major catego-ries: external ports which are at a sidewall, and

internal ports which are inside the box. This chapter willcover only external ports; internal ports are discussed inChapter 6.

PLACING EXTERNAL PORTS

By now you should be familiar with the placement ofexternal ports (EMPorts). If not, you should follow thefirst example given in Chapter 1. To briefly review: Anexternal port is placed in =LAYOUT= by selecting EMPortfrom the toolbar. These ports are generally placed on theedge of the box at the end of a line.

Figure 4-1 shows a comparison between a port in circuittheory and a port in =EMPOWER=. In the circuit theoryschematic on the left, there are two ports. Each port hastwo terminals, with the bottom terminal generally beingground.

In the =EMPOWER= simulation shown on the right ofFigure 4-1, the section of line stops before the edge of thebox (generally one cell-width away) and a port begins inits place. (See Figure 2-3 in Chapter 2 to see how this ismapped onto the grid.) As in the circuit theory schematic,there are two ports, and each port has two terminals.

However, in =EMPOWER=, instead of the ground planebeing modeled as a simple short circuit, the effect ofcurrents travelling through the box is taken into account.

EMPORT OPTIONS

When you first create a port, it is automatically configuredto be an external port with the proper characteristics tobe placed on the end of a transmission line. For manyapplications, you will want to modify these characteristicswhen you place the port. These characteristics are shownin the EM Port Properties dialog box which comes upautomatically when the port is placed and which can beaccessed later by either double-clicking on the port or byselecting the port and choosing Details from the Editmenu. A typical EM Port Properties dialog box is shownin Figure 4-2. The following sections describe the entriesin this dialog box.

Draw Size - This has no effect on the simulation. Itcontrols the size that the port number appears on screenand on printouts.

Ref Plane Shift - This parameter is only available if “PortType” is set to “Normal” (see below). On most complete

1 2 1 2

Figure 4-1 A comparison between circuit theory (left) and=EMPOWER= (right).

46 External Ports

circuits, this value can be left at zero. A positive ReferencePlane shift causes the deembedding to add extra linelength to the circuit; A negative value is more common andcauses the reference planes to move inside the box. Seethe Examples manual for an example of a patch antennasimulation.

The reference plane is shown as an arrow on the layout.Additionally, when the EMPort is selected, Handles ap-pear on the reference plane, allowing it to be moved withthe mouse.

Port Number - When =EMPOWER= is run, the port num-bers specified here correspond to the port numbers in theresulting data. These port numbers must be sequential(numbers cannot be skipped), and Normal ports mustalways have lower numbers than non-deembedded andinternal ports. =LAYOUT= assigns a new port numberautomatically when an EMPort is placed, and the portnumber is displayed on the layout at the port.

Width & Length - When placing an external port on theend of a strip-type transmission line, you should normally

Figure 4-2 EM Port Properties dialog box.

External Ports 47

leave these at zero so that =LAYOUT= sizes the portautomatically. If you want to override the size, or forslot-type or internal ports, you can specify width andlength here. Note: Width and length are measured rela-tive to the line direction, so these parameters can appearto be reversed. Length is the length in the direction ofpropagation (along the line), and width is the width of thestrip.

Layer - Specifies the metal layer on which the port isplaced.

Location - specifies the edge of the port for external portsand the center of the port for internal ports.

Line Direction - Gives the direction of the line at the port.In the default mode, the nearest wall determines thedirection of the line. This value rarely needs to be over-ridden.

Current Dir - Specifies the direction of current flow withinthe port. Figure 4-3 shows the default current directionfor external ports on strip-type structures such as micros-trip and stripline. Figure 4-4 shows the default currentdirection for external ports on slot-type structures such as

0

dy

y

0 dx

x

a) b)

y

x

strip conductors

Figure 4-3 Default port current direction, used for strip-likestructures: a) Along X. b) Along Y.

48 External Ports

coplanar waveguide. For internal ports, the default cur-rent direction is “Along Z.” This value also rarely needs tooverridden.

Port Type - Specifies the basic type of port: Normal, NoDeembed, and Internal.

• Normal ports are external ports which are deembeddedand may be multi-mode. They are shown in gray on thelayout.

• No Deembed ports are external ports which are notdeembedded and cannot be multi-mode. They are shownin white on the layout.

• Internal ports are also not deembedded and cannot bemulti-mode. They are shown in white on the layout.

For more information on dembedding and multi-modelines, see below.

DEEMBEDDING

The right half of Figure 4-1 shows the structure which issimulated by =EMPOWER=. If you are actually buildingyour circuit this way, that is, if your ports consist of a line

0

dyy

0 dx

x

a) b)

y

x

slots

Figure 4-4 Current direction for ports along slot-like structures:a) Along Y. b) Along X.

External Ports 49

which stops just short of the end wall, as is often the casewith a coax-microstrip junction, then you may not need touse deembedding, because =EMPOWER= is simulatingthe circuit as you are actually going to build it.

However, you may not have this kind of construction oryou may be simulating a small segment of a larger circuit.In Figure 4-1, there is capacitance at the port due tocoupling from the open end of the line to the wall. Deem-bedding removes this extra reactance, perfectly matchingthe transmission line, modelling it as though the line andbox extend out to infinity.

Deembedding also allows you to define a reference planeshift. By default, the reference plane shift is zero, whichmeans that the resulting data is measured at exactly theside wall. If the reference plane shift is negative, then thedata is measured from inside the box, effectively subtract-ing length from the circuit. If the reference plane shift ispositive, then the data is measured from outside the box,effectively adding length to the circuit.

Figure 4-5 shows the equivalent network used whendeembedding is active. The center of the figure, labelledCIRCUIT contains the raw results from the =EM-POWER= simulation. Reactance X (shown as inductorsabove) cancels the capacitance caused by the end wall aswell as correcting other reactances. (The value of X may

CIRCUITXX

RefShift RefShift

Figure 4-5 Deembedding of the circuit simulated by=EMPOWER=.

50 External Ports

be negative, and it is frequency dependent.) The RefShiftlines at the outside move the reference planes to thecorrect location. Since the RefShift lines also help tocorrect for the discontinuity at the box wall, their lengthsare normally not zero, even if the reference shift specifiedfor the port is zero. The impedance of the RefShift lines isequal to the port line impedance, so only the phase isshifted by the addition of these lines. The magnitude ofthe reflection coefficients is not affected.

The parameters for deembedding are calculated prior tothe analysis of the circuit. =EMPOWER= does this auto-matically by analyzing two different length lines at eachfrequency for each port used, solving for the reactance andthe base RefShift value.

Note: Deembeddin g require s an additiona l lin e analysismod e at the star t of the run , so run s usin g deembeddingcan tak e substantial ly longe r.

This is especially true if the lines at the ports are wide,since a wide line is simulated across the entire length ofthe box. However, line analysis is always symmetrical,and may be symmetrical in two planes if the port lines areplaced exactly in the middle of the box. =EMPOWER=also caches the line analysis results, so if the box and portlines are not changed between runs, previous data will beused. The data for these lines are stored internally in theWorkspace (WSP) file using internal files namedEM-POWER.R1, EMPOWER.R2, etc. =EMPOWER= also hasthe intelligence to detect when two or more ports have thesame configuration (width, position, etc.), and will onlyrun the line analysis once.

External Ports 51

MULTIMODE PORTS

Until now, all ports which we have looked at have beensingle mode ports. Single mode ports act just like regularnodes in =SuperStar=, and external components can beadded directly to these ports. =EMPOWER= also supportsexternal multimode ports where two EMPorts are closeenough together that they are coupled. Figure 4-6 showsa circuit which uses multimode ports, with ports 1, 2, and3 being a 3-mode port, 4 being a normal single-mode port,and ports 5 and 6 being a 2-mode port. Multimode portshave the following features:

• They much more accurately characterize the performanceof a network with two or more lines close together on onewall.

• They cannot be used like normal =SuperStar= nodes;They can only be connected to other multimode ports,including multi-mode lines and multimode=EMPOWER= data. Further, any multi-mode elementsconnected together must have the same number ofmodes for each port.

Figure 4-6 A circuit which could use multi-mode ports.

52 External Ports

CAUTION: Do not connect standard lumped elements toa multimode port! The results will not be correct. If youwill be connecting directly to components, you shoulduse single-mode ports. Use multi-mode ports only forconnection only with other multi-mode ports and multi-mode lines.

• They can be used with decomposition to accuratelyanalyze much larger structures than would be possible in asingle =EMPOWER= circuit. See Chapter 5 for moredetails on decomposition.

To create a multi-mode port, click on the “Mode Setup”Button from the =EMPOWER= setup dialog box when youstart an =EMPOWER= run. You will see a box similar tothe one in Figure 4-7. To make ports multi-mode, checkthe boxes between them. In Figure 4-7, EMPorts 1,2, and3 form one multi-mode port, and EMPorts 5 and 6 formanother multi-mode port. EMPort 4 is a single mode port.To make a multi-mode port, you must follow these rules:

Figure 4-7 Multi-mode dialog box for the circuit in Figure 4-6.

External Ports 53

• All EMPorts for a multi-mode port must be on the samewall.

• All EMPorts must have the same length, line direction,current direction, and reference plane shift. The EMPortsmay (and often do) have different widths, as in Figure 4-6..

• All EMPorts must be “Normal” (not “No Deembed” or“Internal”).

• Port numbers must be sequential and in order. Forexample, in Figure 4-6, if you swapped ports 1 and 2, youcould not use a 3-mode port, because the ports would be inthe order 2-1-3 along the sidewall.

Running Figure 4-6 in EMPOWER will give 6-port data,as would be expected by glancing at the picture. However,the fourth port is the only normal, single-mode port. Inthe data file, the first three ports of data are in mode-space,and the last two ports of data are in mode-space. Forexample, in the data file:

• S41 represents the transmission of energy from mode 1 ofmulti-mode port (1,2,3) to port 4.

• S25 represents the transmission of energy from mode 1 ofmulti-mode port (5,6) to mode 2 of multi-mode port (1,2,3).

• S66 represents the reflection of energy in mode 2 ofmulti-mode port (5,6).

Multimode data should be carefully connected. Multi-mode ports should be connected only to other identicalmulti-mode port or line configuration (same box, linewidths, spacings, etc.). Otherwise, the connection is non-physical and the results are meaningless.

See the spiral inductor example in Chapter 5 for moreinformation on the use of Multimode lines.

54 External Ports

GENERALIZED S-PARAMETERS

When normal circuit-theory analysis is performed, theports are often terminated with a standard impedancesuch as 50 or 75 ohms. However, =EMPOWER= will givemuch more accurate results if you use generalized S-Pa-rameters. With generalized S-Parameters, instead of theports being terminated with 50 or 75 ohms, the ports areterminated with the characteristic impedance of the lineas calculated by =EMPOWER=. This is a more internallyconsistent representation, and the results are often farmore accurate. You should use generalized S-Parametersif the following three conditions hold:

1) You are using normal, deembedded ports. Ports marked“No Deembed” or “Internal” are not appropriate for report-ing generalized S-Parameters, so they are normalized to50 ohms if generalized parameters are requested.

2) You have calculated the impedance of the lines at theports (using =TLINE=, for instance), and they are 50 (or75) ohms.

3) You have run =EMPOWER=, but it calculated the portimpedances to be a little different (for example, 47 insteadof 50 ohms). This error is generally a result of the gridsize. (A finer grid would result in less error in the imped-ance).

In this case, you know that your port lines should be 50ohms, but =EMPOWER= reported 47 ohms. If you thenrequest Generalized S-Parameters, GENESYS will alsouse 47 ohms for the terminating impedance, and a largepart of the analysis error due to the grid will be cancelled.The results will be close to the results obtained if youmeasured the circuit in a 50 ohm network analyzer.

To get generalized S-Parameters from GENESYS, Checkthe “Generalized” box in the =EMPOWER= propertiesdialog box.

External Ports 55

When =EMPOWER= is run, it outputs a file (in the struc-tured storage when run from GENESYS) for each portwith impedance data with extensions R1, R2, R3, etc., so,for a 2-port network in file =EMPOWER= analysis EM1,using “Generalized” impedance is equivalent to using animpedance of “WSP:Simulations\EM1\EMPOWER.R1,WSP:Simulations\EM1\EMPOWER.R2".

See the examples manual f o r an example of the use of gen-eralized S-Parameters.

56 External Ports

Chapter 5

Decomposition

I n =EMPOWER=, it is possible to break down largecircuits into smaller segments which are connected bytransmission line sections. Decomposition can be tedi-

ous to implement, but its reward is that simulations canbe performed accurately in much less time and with fewerfrequency points. The principal benefits of decompositionare:

• Ability to tune single or coupled transmission line sectionsinside a circuit which was simulated by =EMPOWER=. Forexample, you can change the size of a meander line oradjust the tap point on an interdigital filter without rerunningthe =EMPOWER= simulation.

• Most circuits require far fewer frequency points for accurateanalysis. This is due to the fact that quarter-wave resonantlines are broken down into much smaller lines that do notresonate, and interpolation is possible. For example, a 7thorder interdigital filter can often be simulated with just 5frequency points in the =EMPOWER= run while 100 pointsare displayed in the output sweep!

• Ability to simulate problems too large to otherwise run.

The main disadvantages of decomposition are:

• Tedious to setup circuit. The simulation requires multiple=EMPOWER= runs combined with a schematic.

• Box modes and other phenomena related to the entireproblem are not modelled. However, since =EMPOWER=

uses mode space to model coupled line connections, this isless of a problem that it would be with other simulators.

• Losses in the connecting lines are not modelled.

BASICS

Decomposition can be applied to circuits with parts whichare connected via single or multiple transmission lines.Figure 5-1 shows some typical circuits which can be bro-ken apart. In each of the circuits, the unshaded areas aresimulated individually. The pieces are then combinedusing multi-mode transmission lines to connect the pieces,representing the lines in the shaded area.

For decomposition to be possible, you must be able to breakthe circuit down into rectangular areas which are inter-connected with transmission lines. For example, the spi-ral inductor in Figure 5-1 is broken down into fourrectangular areas, one for each corner. These sections arethen connected with multi-mode transmission lines.

Spiral Inductor

InterdigitalFilter

Edge Coupled Filter

Meander Line

Figure 5-1 Circuits which could be analyzed with decomposition.Shaded areas are not analyzed directly, but use multi-mode linesto connect the simulated areas (unshaded).

58 Decomposition

In each of the circuits in Figure 5-1, the three mainadvantages of decomposition can be seen:

• The lengths of the connecting transmission lines can bevaried. In the spiral inductor, this allows the size of thespiral (and the inductance) to be tuned or optimized inGENESYS.

• Far fewer points need to be analyzed. This is becauseeach of the pieces is simpler and interpolation works well.For example, in the edge coupled filter, each of the piecescontain only open ends and small sections of lines whichdo not resonate. As a result, this filter only needed 5frequency points for a good analysis.

• With any of these circuits, The grey areas can easily get solarge that the problem requires hundreds of megabytes toanalyze. In the meander line, if the lengths of the coupledlines (grey areas) gets very long, the =EMPOWER=simulation could take a long time. When the circuit isdecomposed, simply changing one length value inGENESYS gives a virtually instant analysis, no matter howlong the coupled sections are.

SPIRAL INDUCTOR EXAMPLE

As a first decomposition example, we will analyze thespiral inductor shown in Figure 5-1. The first step is tocome up with a plan for decomposition. Figure 5-2 showsthe plan for analysis of the spiral inductor. We stronglyrecommend that you write a similar plan on paper whenyou setup a problem for multi-mode analysis.

The first step is to create workspace with a layout for eachunique piece. In this example, there are two unique pieces:The lower left corner is the first, and each of the otherthree corners which are identical. There are two basicmethods for creating these pieces:

• Create the pieces individually, drawing only the part thatwill be simulated in each piece. In this case, eachindividual layout will look like the parts shown in Figure 5-2.

Decomposition 59

Or,

• Create a complete layout of the entire problem first. Then,make the box smaller so that only the desired piece issimulated. This is the method we will use for the spiral.

We have created a layout of the entire spiral inductor as astarting point (\EAGLE\EXAMPLES\FULL.WSP).This file was created by starting with an MRIND elementso that the layout was created mostly automatically. Theonly addition was the extra length leading to port 1 andthe EMPorts. Notice that the reference plane for port 1 isshifted to the actual start point of the spiral model. Port2 is an internal port. This circuit can be analyzed directly,but it requires minutes per frequency point and 37 mega-bytes of RAM.

This file was then saved as COMBINE.WSP. The box wasshrunk and the circuit was moved so that only the bottomleft quarter of the circuit is in the box. The number on theinternal port on the end of the spiral was changed to 10.Ports 2-5 on the right and 6-9 were added. Since these

10

1

4

4 4

Part1 Part2

M1

M2

M3

M4M

1

M2

M3

M4

M1

M2

M3

M4

M1

M2

M3

M4

Part2Part2

1

1 1

2

5

5 5

59 8

88

6

Figure 5-2 Plan for decompositional analysis of the spiral inductorfrom Figure 5-1.

60 Decomposition

ports are in the middle of a line instead of on the end, theirwidth must be set manually. Also, the reference planes onthe ports were shifted in. The resulting layout for the firstpiece is shown in Figure 5-3.

=EMPOWER= was run for Part1. The settings are asshown in Figure 5-4 Note that only 5 points are neededsince the individual parts are not resonant. The “SetupLayout Port Modes” button was clicked, and the check-boxes in the Setup Modes dialog box were set to indicatethat those inputs are modally related.

CAUTION: Do not forget to setup the modes when youare analyzing by decomposition. The Mode Setup boxturns red if any inputs are modally related. Impropermode setup is one of the most common errors in de-composition.

A similar set of steps was followed for Part2.

Figure 5-3 The layout for the analysis of the first piece of the spiralinductor.

Decomposition 61

The final step in decompositional analysis is to combinethe pieces. The Schematic (COMBINE) which does this isshown in Figure 5-5. The pieces used are NPO10 andNPO8 blocks (under “DEVICE” in =SCHEMAX=) for thedata in PART1 and PART2 plus MMTLP8 models (multi-mode physical transmission lines, found under “T-LINE”)for the interconnecting lines. The data for the NPO10 isin WSP:Simulations\EMPart1\EMPOWER.SS, and thedata for the NPO8 is in WSP:Simulations\EMPart2\EM-POWER.SS.

Note: Some users may find it easier to write a text Netlistto combine the pieces. At Eagleware, we find it easierto use =SCHEMAX= for this purpose, but you may usewhichever you feel most comfortable with.

Whenever deembedded ports are used, data files suitablefor the SMTLP and MMTLP models are automaticallycreated during the LINE portion of the =EMPOWER=

Figure 5-4 The setup for analysis of the first piece.

62 Decomposition

run. For the MMTLP8 lines, the file WSP:Simula-tions\EMPart1\EMPOWER.L2 was used. This corre-sponds to the second set of inputs for PART1. You shouldview the listing file (Right-click on EMPart1) and look atthe port numbers to determine which EMPOWER.L* filecontains the line data you need.

Note: Files with names like WSP:Simulations\EM-Part1\EMPOWER.L2 are taken from within the currentworkspace. For a complete explanation of how thesefiles are names, see the File Formats chapter in thismanual.

The substrate must also be specified, but only the UNITSparameter is used by the MMTLP8 model.

A variable was setup (LENGTH) so that the lengths of linecan be tuned in GENESYS simultaneously, changing thesize of the spiral and thus the inductance very quickly.

Figure 5-5 =SCHEMAX= file for combined analysis of the spiralinductor.

Decomposition 63

Figure 5-6 Analysis of the Spiral Inductor. Top: Results usingdecomposition with only 5 points in the =EMPOWER= runs.Middle: Decomposition results with 10 points. Bottom: Resultsanalyzing entire inductor without using Decomposition.

64 Decomposition

The results from this are shown in Figure 5-6(top). Noticethat even with only 5 analysis points across the band, theinterpolation is very good. To illustrate this, the spiralinductor was recalculated with 10 points as shown inFigure 5-6(middle). You can see quite good agreementbetween the two. To test the validity of the decomposi-tional analysis, the entire spiral was analyzed, and theresults are given in Figure 5-6(bottom). This analysis tookhours on a 266MHz Pentium II, and if the lengths of thelines in the spiral are changed, it must be rerun.

LOSSES

A current limitation of decomposition is that losses are nottaken into account in multi-mode transmission line sec-tions or in reference-plane shifts. For the spiral inductor,this means that the losses as calculated are accurate forthe nominal dimensions, but any modification to thelengths using the multimode lines will not affect thecalculated loss.

In general, if the decomposed pieces cover the circuitcompletely (as is the case in the spiral inductor), then thelosses will be accurate. If the pieces do not completelycover the circuit (if sections of line are left out of the=EMPOWER= analysis and are added with MMTLP sec-tions, then the losses will not include these sections. Thisis true regardless of the reference plane shifts used, sincethese shifts do not affect the loss.

PORT NUMBERING

You must be very careful when setting up and numberingports for decompositional analysis. The following rulesmust be followed:

• Never connect anything other than MMTLP lines or otheridentical modal inputs to inputs which are modally related.

Decomposition 65

Connecting lumped elements to modal inputs isincorrect and will give bad results.

• Ports which will be modally related must have sequentialnumbers. They must also all have the same reference shift.

• Ports for mode-space inputs must be marked type“Normal,” not “No deembed” or “Internal.”Correspondingly, their numbers must be lower than any“No-deembed” or “Internal” ports.

• The order of ports used must correspond between thepieces and the MMTLP lines used. The lowest portnumber in a modally related set of inputs should connect toMode 1 in the MMTLP line, and the highest port number inthe set should connect to Mode N on the MMTLP line.Also, port ordering should be exactly the same on bothpieces connected through the MMTLP. Figure 5-7 showsan incorrect numbering of the spiral inductor. In thisexample, PART1 and PART2 are inconsistently numbered,since on PART1 the outermost inputs (numbers 2 and 6)are the lowest number while in PART2 the innermost inputs(numbers 1 and 5) are the lowest number.

10

1

4

4 4

Part1 Part2

M1

M2

M3

M4M

1

M2

M3

M4

M1

M2

M3

M4

M1

M2

M3

M4

Part2Part2

1

1 1

5

2

5 5

56 8

88

9

Figure 5-7 ***INCORRECT***: An example of incorrect nodeordering in the Spiral Inductor example. The nodes on part 1 arereversed.

66 Decomposition

• Pieces can be connected directly together without usingMMTLP. In this case, the lowest ports in each modallyrelated set of inputs are connected to each other.

Decomposition 67

Chapter 6

Lumped Elements andInternal Ports

A s described in Chapter 4, every =EMPOWER= circuitmust contain at least one port. This chapter willcover lumped elements and internal ports (ports

inside the box). External ports (along a sidewall) werecovered in Chapter 4.

PLACING INTERNAL PORTS

The process of placing an internal port is similar to theprocess of placing an external port. To summarize: Aninternal port is placed in =LAYOUT= by selecting EMPortfrom the toolbar. Internal ports can be placed anywherein the box. When the EMPort Properties dialog box ap-pears, first select “Internal” in the Port Type combo box.Next, fill in the width and length of the pad. Press OK tocomplete the placement.

Note: The rest of the option s in the EMPort Propertiesdialo g box were covered in Chapte r 4 in the sectionentitle d “EM Port Options. ” You may want to reviewthese option s now.

Figure 6-1 shows a comparison between ports in circuittheory and internal ports in =EMPOWER=. In the circuittheory schematic on the left, there are two ports. Each

port has two terminals, with the bottom terminal gener-ally being ground.

In the =EMPOWER= simulation shown on the right ofFigure 6-1, there are two z-directed ports, one at each endof the line. These z-directed ports are mapped onto thegrid along Z, much in the same way as a viahole would bemapped. (See Chapter 2 for more information on mappingto the grid.) As in the circuit theory schematic, there aretwo ports, and each port has two terminals. The bottomterminals, which are true ground in the circuit schematic,are connected to the bottom wall (ground plane),a physicalrepresentation of ground.

Z-directed internal ports can be used in GENESYS toconnect elements, just like a node in =SCHEMAX= or ina text file. In other words, components like resistors andtransitors can be connected directly to these ports. Yousimply place a z-directed port in the center of the pad forthe component in these cases. Note: =SCHEMAX= doesthis automatically as is described later in this chapter.

1 2

1 2

Figure 6-1 A comparison between ports in circuit theory (left) andinternal ports (current direction is default, Z-directed) in =EMPOWER=(right).

70 Lumped Elements and Internal Ports

MANUALLY ADDING LUMPED ELEMENTS

Note: GENESYS will automatically add lumped ele-ments to your simulation if components are on yourlayout. This section is for background information andadvanced applications

The circuit shown in Figure 6-2 contains an =EM-POWER= circuit which was drawn completely in =LAY-OUT=. (The schematic for this network was blank.) It has4-ports; ports 1 and 2 are external, and ports 3 and 4 areinternal. =EMPOWER= will create a 4-port data file forthis circuit.

Note: Internal ports and “no-deembed” ports must al-ways have higher numbers than normal, external, deem-bedded ports. In Figure 6-2, the internal ports arenumbered 3 and 4, while the external ports are num-bered 1 and 2.

The data file created by =EMPOWER= can then be usedin GENESYS. The circuit on the right in Figure 6-2 uses

4

MYNET(1)

(2)

Figure 6-2 A simple circuit with z-ports designed for lumpedelements (left) and a network which uses the resulting data (right).

Lumped Elements and Internal Ports 71

the resulting data in a complete network. First, a FOU(four port data) device was placed on the blank schematic.The name assigned to this FOU block was the name of theinternal file from the =EMPOWER= run (“WSP:Simula-tions\EM1\EMPOWER.SS”).

An input and output were added on nodes one and two ofthe FOU block, the ground was added to the ground node,and a capacitor was connected across ports 3 and 4. Thishas the effect of putting the capacitor “into” the =EM-POWER= simulation. This capacitor can then be tunedand optimized, just like any other element in GENESYS.

When the S-Parameters of MYNET are displayed, you seethe resulting S-Parameters of the entire circuit.

AUTOMATIC PORT PLACEMENT

One advantage of =EMPOWER= is its true integration.In most electromagnetic simulators, you would have nochoice but to go through the complicated steps above.(Imagine how tedious this would be if you had 10 lumpedelements, 2 transistors, and an op-amp chip in the box!)Fortunately, when =EMPOWER= is combined with =Su-perStar=, =SCHEMAX=, and =LAYOUT=, the internalports and lumped elements can be generated and addedautomatically.

Figure 6-3 uses automatic port placement. Initially, thecircuit on the left of Figure 6-3 is drawn in =SCHEMAX=.The layout on the right of the figure was then created: Thefootprint for the chip capacitor was automatically placed.The lines and EMPorts were then manually added. When=EMPOWER= is invoked, internal ports are automat-ically added, so the circuit simulated is virtually identicalto the one on the left of Figure 6-2, and the result is a 4-portdata file.

72 Lumped Elements and Internal Ports

=EMPOWER= then automatically creates a networkwhich is identical to the network on the right of Figure6-2. This result is fundamentally the same as the resultfrom MYNET in Figure 6-2. When the capacitor in Figure6-3 is tuned or optimized, the networks MYNET andEMPOWER are both updated simultaneously.

Even if you create a file with a layout only (no schematic),you can still use automatic port placement. Simply putthe parts down onto a blank schematic, connecting theminto a dummy network. The parts will now show up in=LAYOUT= and can be moved as needed, ignoring anyrubber bands. (The rubber bands come from the meaning-less connections in the dummy network.) When you dis-play the =EMPOWER= simulation results, it will includethe components. You do not need to display the resultsfrom the schematic.

1 2MYNET(1) (2)

Figure 6-3 A circuit which uses automatic port placement.

Lumped Elements and Internal Ports 73

PLANAR (X- AND Y-DIRECTED) PORTS

Note: =EMPOWER= will create planar ports for lumpedelements if the “Use Planar Ports for one-port elements”box is checked in the =EMPOWER= options dialog. Seeyour reference manual for details.

In some situations, you may want to place internal portswith X- or Y-directed currents. These ports are muchtrickier to use manually, since they are not referenced toground. For components in your layout, =EMPOWER=will automatically place planar port and lumped elements,so this section is for background or advanced applications.

Figure 6-4 shows the configuration of these ports. Theseports can be more accurate for manually connectinglumped elements to =EMPOWER= data since the portsare a more accurately represent the physical connectionof lumped elements.

The circuit shown in Figure 6-5 contains an =EM-POWER= circuit which was drawn completely in =LAY-OUT=. (The schematic for this network was blank.) It has

1

1

Figure 6-4 A comparison between ports in circuit theory (left) andinternal ports (current direction is Along X or Y) in =EMPOWER=(right).

74 Lumped Elements and Internal Ports

3-ports; ports 1 and 2 are external, and port 3 is internalwith current direction “Along X.” =EMPOWER= will cre-ate a 3-port data file for this circuit, however, you must beaware that port 3 will be a series connected port andcannot be used in the normal manner.

The data file created by =EMPOWER= can then be usedin GENESYS as described in the previous section using“WSP:Simulations\EM1\EMPOWER.SS”. The circuit onthe right in Figure 6-5 uses the resulting data in a com-plete network. First, a THR (three-port data) device wasplaced on the blank schematic, using the EMPOWER.SSfile from the =EMPOWER= run. An input and outputwere added on nodes one and two of the THR block, theground was added to the ground node, and a capacitor wasconnected from port 3 to ground. This has the effect ofputting the capacitor across port 3 in the =EMPOWER=

1 2 3

0

0

MYNET(1) (2)

Figure 6-5 An example of the use of a port with current direction“Along X”. Note that the connection of the capacitor from the nodethree to ground on the right is the equivalent of connecting thiscapacitor across port 3 on the left, since port 3 is not referenceground.

Lumped Elements and Internal Ports 75

simulation. The rules to follow for “Along X” and “AlongY” internal ports are simple:

• Do not attempt to use them for transistors or other3-terminal (or more) devices.

• Set the Current Direction of the EMPort to “Along X” (alongthe x-axis) if the current along the component flows fromleft to right, as in Figure 6-5. Set the Current Direction ofthe EMPort to “Along Y” (along the y-axis) if the currentalong the component flows from top to bottom, as if thecapacitor were turned 90 degrees from the one on the leftin Figure 6-5.

• Connecting a lumped element in =SCHEMAX= from theport to ground when you use the resulting data isequivalent to connecting the lumped element accross thelength of the port in the =LAYOUT=. This does not meanthat the component is grounded. It simply means thatthe component is connected across the port. Thisconcept is key to understanding X- and Y- directed ports.

When the S-Parameters of MYNET are displayed in agraph, you see the resulting S-Parameters of the entirecircuit, and the results are similar to the results of thecircuits in Figures 6-2 and 6-3.

RESONANCE

Often, when a circuit contains lumped elements, you canuse very few frequency points for the =EMPOWER= runs.Since the lumped elements are not included in the =EM-POWER= data, there are generally many fewer reso-nances, and the data interpolates much more accurately.In this case, you may want to only use 2 or 3 points in theelectromagnetic analysis while showing the results of theentire network with 100 points or more (specified in theCo-Simulation Sweep in the =EMPOWER= Options Dia-log box).

76 Lumped Elements and Internal Ports

Chapter 7

Viewer

T his chapter describes how to launch the =EM-POWER= viewer program and how to use it to visu-alize and interpret currents (or voltages) generated

by =EMPOWER=. It also describes the viewer interface.

The =EMPOWER= viewer helps you visualize currentdistribution and densities in a board layout. It processescurrent density magnitude and angle and plots them astwo or three dimensional static or dynamic graphs. Theseplots provide insight into circuit behavior and often sug-gest modifications which improve the performance. Mostelectromagnetic simulators include visualization tools.The =EMPOWER= viewer has distinct advantages suchas three dimensional graphs, true animation capabilities,and precise information about current phase. The fullpotential of the =EMPOWER= viewer is realized withpractice so we encourage you to investigate your circuitswith the viewer and reflect on the results you observe.

STARTING THE VIEWER

The viewer is started by either selecting Run Viewer fromthe right-click menu of an =EMPOWER= simulation(Workspace Window in GENESYS) or by clicking on the=EMPOWER= viewer icon in the GENESYS group.

VIEWER INTERFACE

This section describes the viewer menu items and buttons.It can be used to become acquainted with the interface ingeneral as well as as a reference section.

A sample viewer screen is shown in Figure 7-1. Theobjects in this figure are described below.

A - File Menu

Open - Opens a new viewer data file.

Exit - Exits the viewer.

Toggle Background Color - Toggles the background fromblack to white, or white to black. A white background isnormally selected before a screen or window print.

Print Screen - Sends a copy of the entire screen to a bitmapfile or to a printer.

Print Window - Sends a copy of the viewer window to abitmap file or to a printer.

Figure 7-1 Sample viewer screen showing a three-resonator filter.

78 Viewer

B - View Menu

The objects in this menu affect how the current image isdisplayed.

Top (Home) - Shows a top-down view of the current image.This option can also be selected by pressing Home.

Front (Ctrl+Home) - Shows a “front” view of the currentimage. This view is from the y-axis, at z=0. This optioncan also be selected by pressing Ctrl+Home.

Side (Ctrl+End) - Shows a “side”view of the current image.This view is from the x-axis, at z=0. This option can alsobe selected by pressing Ctrl+End.

Oblique (End) - Shows an oblique view of the currentimage. This view is top-down on the x-y plane with a slightoffset. This option can also be selected by pressing End.

Rotate - The objects in this sub-menu rotate the currentimage.

Rotate - Left (Left Arrow) - Rotates the current imageclockwise in a horizontal plane perpendicular to thescreen. The center of the viewer image window is alwaysthe center of rotation. This option can also be selected bypressing ←.

Rotate - Right (Right Arrow) - Rotates the current imagecounter-clockwise in a horizontal plane perpendicular tothe screen. The center of the viewer image window isalways the center of rotation. This option can also beselected by pressing →.

Rotate - Up (Up Arrow) - Rotates the current imageforward in a vertical plane perpendicular to the screen.The center of the viewer image window is always thecenter of rotation. This option can also be selected bypressing ↑.

Viewer 79

Rotate - Down (Down Arrow) - Rotates the current imagebackward in a vertical plane perpendicular to the screen.The center of the viewer image window is always thecenter of rotation. This option can also be selected bypressing ↓.

Rotate - Clockwise (PgDn) - Rotates the current imageclockwise in the plane of the screen. The center of theviewer image window is always the center of rotation.This option can also be selected by pressing Page Down.

Rotate - Counter-Clockwise (PgUp) - Rotates the currentimage counter-clockwise in the plane of the screen. Thecenter of the viewer image window is always the center ofrotation. This option can also be selected by pressingPage Down.

Pan - The objects in this sub-menu shift the apparentlocation of the viewer window relative to the currentimage.

Pan - Left (Ctrl+Left) - Moves the viewer location to theleft (relative to the current image). This moves the imageto the right in the viewer window. This option can also beselected by pressing Ctrl+←.

Pan - Right (Ctrl+Right) - Moves the viewer location to theright (relative to the current image). This moves theimage to the left in the viewer window. This option canalso be selected by pressing Ctrl+→.

Pan - Up (Ctrl+Up) - Moves the viewer location up (rela-tive to the current image). This moves the image down inthe viewer window. This option can also be selected bypressing Ctrl+↑.

Pan - Down (Ctrl+Down) - Moves the viewer location down(relative to the current image). This moves the image upin the viewer window. This option can also be selected bypressing Ctrl+↓.

80 Viewer

Pan - Zoom In (Ctrl+PgUp) - Moves the viewer locationcloser to the current image. This increases the size of theimage in the viewer window. This option can also beselected by pressing Ctrl+Page Up.

Pan - Zoom Out (Ctrl+PgDn) - Moves the viewer locationaway from the current image. This decreases the size ofthe image in the viewer window. This option can also beselected by pressing Ctrl+Page Down.

Toggle - The objects in this sub-menu toggle the availableoptions listed below.

Toggle - Absolute Value Display - When selected, theviewer displays absolute values only. If not selected, anactual value with information about flowing direction isdisplayed. The difference is that absolute value is alwayspositive, whereas the actual current values can be positivefor forward directed currents and negative for backwarddirected currents. Negative amplitudes are drawn belowthe x-y plane. This option has a checkmark () beside itwhen selected.

Toggle - Animation - When selected, the viewer “animates”the image in real or angle mode. This is accomplished bymultiplying the individual currents by exp(jθ), where θcycles from 0 to 2π and showing a sequence of snapshotimages for increasing θ. This option has a checkmark ()beside it when selected.

Toggle - Scale - When selected, the viewer displays thescale in the lower-left of the viewer window. This optionhas a checkmark () beside it when selected.

Toggle - Value Mode (Real,Mag,Ang) - This option selectsthe current display option. The options include the Realcurrent value for current distribution snapshots and ani-mation, Magnitude for time averaged current values, andAngle for the current phase delay distribution snapshots.

Viewer 81

Toggle - Wireframe - When selected, the viewer displays awireframe version of the current plots. A wireframe iscreated by drawing the outlines of the =EMPOWER= gridcurrents without filling the resulting polygons. When thisoption is not selected, the viewer fills the polygons, result-ing in a solid surface plot of the current patterns. Thisoption has a checkmark () beside it when selected.

Load - From User View #(1-10) - Loads the previouslysaved viewer settings for the selected view #. Savedsettings can also be restored by pressing the number keycorresponding to the desired setting #.

Save - To User View #(1-10) - Saves the current viewersettings into the selected view #. The settings can berestored later by selecting the desired # from the loadsub-menu described above. The options in this menu canalso be selected by pressing Shift + the number keycorresponding to the desired save #.

TIP: The save and load functions are extremely useful.If you rotate and pan to a view that you like, press Shiftplus a number (not an arrow) to save that view. Simplypress the number by itself to return to that view. Theseviews are remembered even if you exit the viewer, soyou can easily store your favorite views.

C - X / Y / Z / XY Button

Pressing this button toggles between the four possiblemodes:

X - Displays the x-directed current density distribution.

Y - Displays the y-directed currents density distribution.

Z - Displays the z-directed currents.

82 Viewer

XY - Displays additive surface current density distribu-tion function.

D - Animate Button

This button toggles viewer “animation” on the currentimage. When this option is selected, the button appearspressed. The viewer animation is accomplished by multi-plying the individual currents by exp(jθ), where θ cyclesfrom 0 to 2π and plotting snapshot graphs for sequentialtime moments. What is animated is controlled by theDisplay Option Button (see E below).

E - Display Option Button

This button selects the current display option.

Real - Displays the real portion of the current values.

Mag - Displays the magnitude or time averaged values ofthe currents.

Ang - Displays the phase delay of the current values.

F - Solid/Wire Button

This button toggles the type of surface plot to display.

Wire - Displays a wireframe version of the current pat-terns. A wireframe is created by drawing the outlines ofthe =EMPOWER= grid currents without filling the result-ing polygons.

Viewer 83

Solid - Displays a solid surface plot of the current patterns.This is created by filling the wireframe polygons.

G - Freq (GHz)

This box shows the simulation frequency (in GHz) forwhich the current image data is being displayed. This boxis restricted to frequencies that =EMPOWER= has createddata for. The value can be increased by clicking the “+”button (see I below), and decreased by clicking the “−”button (see H below).

H - Decrease Frequency Button

Decreases the current frequency (see G above). If you arealready at the lowest calculated frequency then this but-ton has no effect.

I - Increase Frequency Button

Increases the current frequency (see G above). If you arealready at the highest calculated frequency then thisbutton has no effect.

J - Clockwise Button

Rotates the current image clockwise in the plane of thescreen. The center of the viewer image window is alwaysthe center of rotation. This option can also be selected bypressing Page Down.

84 Viewer

K - Counter-Clockwise Button

Rotates the current image counter-clockwise in the planeof the screen. The center of the viewer image window isalways the center of rotation. This option can also beselected by pressing Page Down.

L - Rotate Right Button

Rotates the current image counter-clockwise in a horizon-tal plane perpendicular to the screen. The center of theviewer image window is always the center of rotation.This option can also be selected by pressing →.

M - Rotate Left Button

Rotates the current image clockwise in a horizontal planeperpendicular to the screen. The center of the viewerimage window is always the center of rotation. This optioncan also be selected by pressing ←.

N - Rotate Down Button

Rotates the current image backward in a vertical planeperpendicular to the screen. The center of the viewerimage window is always the center of rotation. This optioncan also be selected by pressing ↓.

Viewer 85

O - Rotate Up Button

Rotates the current image forward in a vertical planeperpendicular to the screen. The center of the viewerimage window is always the center of rotation. This optioncan also be selected by pressing ↑.

P - Top Button

Shows a top-down view of the current image. This optioncan also be selected by pressing the Home key.

Q - Front Button

Shows a “front” view of the current image. This view isfrom the y-axis, at z=0. This option can also be selected bypressing Ctrl+Home.

R - Side Button

Shows a “side”view of the current image. This view is fromthe x-axis, at z=0. This option can also be selected bypressing Ctrl+End.

86 Viewer

S - Oblique Button

Shows an oblique view of the current image. This view istop-down on the x-y plane with a slight offset. This optioncan also be selected by pressing End.

T - Current Plot

Shows the color-coded current patterns for the loaded=EMPOWER= generated viewer data file. The menus andtoolbar buttons control how this image is displayed.

U - Color Scale For Current Plot

This scale shows the relative current and current densitymagnitudes based on the color used to draw the plotpatterns.

VIEWER EXAMPLES

This section illustrates the use of the =EMPOWER=viewer using a number of examples. The schematic andauxiliary files for the examples are located in the subdi-rectory \EAGLE\EXAMPLES\VIEWER. You may loadthem as you follow along with the following examples.

The viewer displays current distributions as two or threedimensional graphs. The viewer has several modes thatare used to view various components of the currents fromdifferent view perspectives. The best view of most prob-lems is often found by minor adjustments of the vieworientation. The following examples include a few exam-ples of such adjustments. The examples are simple prob-lems selected because the results are predictable.Nevertheless, they are interesting and illustrate conceptswhich may be applied to more complex problems.

Viewer 87

Consider the possible graphs for a simple line segmentanalysis. The schematic file for this example isMETR16.WSP. It contains description of a segment of the50-Ohm standard stripline [Rautio, 1994] that is alsodiscussed in the Examples manual. The segment is1.4423896 mm wide by 4.996540 mm long and the box sizealong the z-axis is 1 mm. The segment length is 90 degreesat 15 GHz and 180 degrees at 30 GHz. LoadMETR16.WSP in GENESYS. Run the viewer by selecting“Run =EMPOWER= Viewer” from the right-click menu ofSimulation “EM1". The default plot seen in the mainwindow is an animated surface electric current densitydistribution function reflecting the surface currents in thestrip plane. At the initial time t=0 it will look similar tothe graph shown in Figure 7-2. To get this snapshot westopped animation by clicking the Animation camera icon,adjusted the view slightly, and toggled the backgroundcolor to white. To obtain this view, simply press the

Figure 7-2 A snapshot of the current density distribution functionfor a 90-degree segment of the 50 Ohm standard stripline.

88 Viewer

Oblique button on the toolbar after starting the viewer.All other settings are the default:

• Show XY current density distribution (XY/X/Y/Z button).

• Show Real part of the current density distribution (ViewMenu/Switches/Value Mode or Value Mode button).

• Show Absolute values of the current density quantities(View Menu/Switches/Absolute Value Display).

• Animation is off and time is set to initial (ViewMenu/Switches/Animation or Animation Camera button).

• Scale is on (View Menu/Switches/Scale)

• Solid polygons view (View Menu/Switches/Wireframe orSolid/Wire button)

Note: For printing, Toggle Background Color from theFile menu was also used to change the background towhite. To reset the time to zero, the animation wasturned off, and the Real/Mag/Angle button was clickedthree times, returning the mode to real but resetting thetime.

The resulting picture in the main viewer window is a 3Dplot of the surface current density shown with the gridgenerated to solve the problem. The axes in the metalplane (grid plane) correspond to the X and Y axes in thebox. The origin of the coordinates X and Y correspond tothe geometrical origin of the box, (0,0) in =LAYOUT=. Thez-axis perpendicular to the metal plane corresponds to theplotted current/voltage values. The red color on the axisis for high values and dark blue is for zero. The color codedscale makes it possible to evaluate actual values of currentdensity. The plotted values are an additive function ofinterpolated X- and Y-components of the current density.The current components are calculated along the cellsides, not at the corners of the cells. The X and Y current

Viewer 89

components are interpolated to the grid corners and arethen added.

The X-Y current display provides general insight intocircuit behavior. Again consider the view given in Figure7-2. The dominant eigenwave of the stripline is excited atthe left input of the structure. Observe the typical currentdistribution in the cross section X=0 (click the side viewbutton for a better look at this). At this time the currentdeclines to almost zero at the right output (click the Frontview button). This confirms a line length of 90-degrees.Next, animate the response by clicking on the Animatebutton again. Notice how the dominant stripline wavepropagates in the structure. The animation is a simple setof snapshots for the subsequent time moments. The timewill vary between zero and the period of the incident wave(1/f seconds).

The previous example illustrates the propagation of thewave. For simple evaluation of the high and low currentdensity region the time average values of the currentdensity is more practical. To obtain this plot switch toMagnitude mode by clicking on the Real button. Theviewer in this mode is shown in Figure 7-3. The resultsare as expected for a transmission line segment. Thecurrent density is highest at the edges and lowest in themiddle. Note that the absolute values of the currentdensity at the edges are greatly affected by the grid cellsize used. A smaller grid cell size increases the edgecurrent density. However, integrated values of currentdensity are nearly invariable, as they should be [Mexiner,1972]. If the exact current density values are required werecommend choosing a grid cell size equal to the metali-zation thickness.

To investigate the various current components you mayswitch from the XY mode to the X mode (XY/X/Y/Z button).You see only a small change in the graph because the

90 Viewer

current flows primarily along the line segment as ex-pected. Note however, the component visualization modes(X, Y or Z) are more accurate because the values displayedcorrespond directly to the values calculated by =EM-POWER=; no interpolation is necessary for these modes.

The absolute value of the current density is currentlydisplayed. Switch to the Real mode using the menu: ViewMenu/Switches/Absolute Value Display and selectReal mode. Animation should be turned on also (Anima-tion camera button). A snapshot of the plot is shown inFigure 7-4. The Real mode displays both current densityvalues and direction. Current flows in the positive Xdirection if the displayed values are above the metal layer(the color coded axis direction). The current flows in theopposite direction if the displayed values are below themetal plane.

To obtain even additional insight the phase of the signalalong the line may be displayed. Stay in X component

Figure 7-3 A 3D plot of the time averaged current distributionfunction for a 90° segment of the 50 Ohm stripline standard.

Viewer 91

mode, turn off animation, and switch to the Angle modeby clicking the Display Option button until it reads Ang.You may view the wireframe mode by clicking theWire/Solid button until it reads Wire. At the initial timet=0 and with a matching rotation you will a display similarto Figure 7-5. It displays delay of the current densitiesalong the structure in terms of a complex vector rotationangle. 360 degrees of phase corresponds to a one wave-length delay period. The difference of the current phasesat the input and output again confirms a 90 degrees linesegment.

The line segment example was prepared at two frequencypoints. All graphs and explanations given here used thefirst frequency point 15 GHz. The second point is 30 GHzand the corresponding segment length is a half of thewavelength. You may display results at 30 GHz by clickingthe + button and then choosing the views of your choice.

Figure 7-4 A snapshot of the X-directed component of the currentdistribution of a 90º segment of the 50 Ohm stripline standard.

92 Viewer

MULTIMODE VIEWER DATA

This example illustrates the eigenwave multi-mode exci-tation capabilities of =EMPOWER=. A three conductorcoupled microstrip line segment from [Farr, Chan, Mittra,1986] is described in the schematic file LNMIT3.WSP.Three microstrips are 1 mm wide and 0.2 mm apart. Theyare on a 1 mm substrate with relative permitivity of 10.The segment is 8 mm long. The structure has threemodally coupled inputs at opposite segment sides. Weexpect at least three propagating modes. Load the exam-ple in GENESYS. The listing file (Right-Click on the=EMPOWER= simulation in the Workspace Window andselect “Show Listing File”) gives information about thepropagating waves. The first eigenmode is an even modewith integral current distribution pattern +++, the secondeigenmode is odd (pattern +0-), and the third one is againeven (pattern +-+). To excite the odd mode as an example,select Generate viewer data and enter 2 in the “Mode

Figure 7-5 X-directed current component phase delay function fora 90-degree segment of the 50 Ohm stripline standard.

Viewer 93

Number to Excite” box of the =EMPOWER= propertiesdialog. Run the viewer. A snapshot of the calculatedcurrent density function is shown in Figure 7-6. All set-tings except two are the same as in the previous example.The initial view was set to the side view (View Menu/Sideor Side button), and the polygon view was set to wireframe(View Menu/Switches/Wireframe or Solid/Wire button).

The plot confirms that this is an odd mode and shows thetypical current density distribution. If currents on the leftstrip flows in the forward direction, the currents on theright strip flow in the backward direction and the centerstrip currents flow in opposite directions at the oppositestrip sides. For a dynamic view, turn on the animation androtate the plot for a better view of the propagating wave.

To calculate the viewer data for the other eigenwaves run=EMPOWER= and the viewer twice more with “ModeNumber to excite set to 1 and 3. Note that newly calcu-lated data will overwrite the previous ones. To avoid this

Figure 7-6 A snapshot of the current density distribution functionfor the second mode (the first odd mode) of the three conductormicrostrip line.

94 Viewer

and to keep viewer data for all excitation experiments youneed to save a copy of the existing workspace(LNMIT3.WSP in this case) before the next run.

Via Hole Viewer Example

The last visualization example shows a structure withnon-zero X, Y, and Z current components. A segment ofmicrostrip line terminated by a via hole from [Swanson,1992] is described in the file VIA.WSP. The line is 12 milwide and is terminated by a metal square 24 by 24 milwith a 13 mil diameter circular via hole in the center. Thesubstrate height is 15 mil and the relative permitivity is9.8. The box size is 120 by 120 mil. Load this example inGENESYS and run the =EMPOWER= viewer. Figure 7-7shows the time averaged plot (View Menu/Switches/ValueMode or Value Mode button) for additive (XY) currentdensity distribution. The view point is the oblique viewwith a few minor adjustments. The plot shows how thedominant microstrip line mode currents spread across thesquare metal pad. You can see the typical peaks in the

Figure 7-7 Time averaged values of the additive current densitydistribution for a microstrip line terminated in a via hole.

Viewer 95

current density function in the vicinity of the metal inter-nal corners where the surface current changes flowingdirections. Toggling to the X and Y components of thecurrent (XY/X/Y/Z button), you can investigate how thesurface currents change direction in different parts of thestructure. Switching to the Z-current visualization modewill show a plot like Figure 7-8. Note that the scale forthe Z-directed currents is in Amperes and not currentdensity. Each current represents a volume current den-sity integrated across the grid cell. They are shown aslines connecting the corresponding geometrical point inthe grid plane and the point corresponding to the actualcurrent values. If a via hole surface shape is known, usingthe current in Amperes it is possible to estimate a currentdensity on the via hole surface. It is obvious from thepicture that the current density is higher on the via holeside that is closer to the microstrip line segment.

We intentionally did not use the thinning out proceduresin the visualization examples. The thinning out algo-rithms add numerical artifacts that do not substantially

Figure 7-8 Time averaged values of the integral via hole surfacecurrents for a microstrip line terminated in a via hole.

96 Viewer

affect the integral characteristic of a problem (S-,Y-, orZ-matrices). At the same time they can distort the currentdensity distribution function. With these distortions it isoften hard to distinguish between the artifact effects andreal processes. We recommend only general analysis forthinned out problems using the magnitude mode and thetop view.

VIEWER THEORY

The =EMPOWER= viewer is a program designed to read,to process and to visualize the current distribution datacreated by =EMPOWER=. To obtain a current distribu-tion inside a structure the excitation condition must bedefined. This mirrors a real measurement where there areincident and reflected waves. The viewer depicts the casewith one incident wave at one input. The excitation con-ditions are passed to =EMPOWER= in the command linewhen running =EMPOWER= text files. When =EM-POWER= is launched from GENESYS the excitation con-ditions are automatically defined from the =EMPOWER=Setup dialog box when the Generate Viewer Data checkbox is active. If Generate Viewer Data is selected, thedefault incident wave is the first eigenwave of the firstinput. The input number can be changed in the Portnumber to excite box of the =EMPOWER= Setup dialog,and the input mode number can be changed in the Portmode to excite box. The control information about whatinput and what mode are actually excited in the structureis printed out in the listing file (see “PPLT: Input __ mode___ will be incident.” in the listing file).

An output binary file with the extension .EMV is createdby =EMPOWER= to pass data to the viewer program. (Ina GENESYS Workspace, the internal name of this file isEMPOWER.EMV) An optional self-documented ASCIIdata file with extension .PLX can also be written for

Viewer 97

import into other programs. The options -IT and -IBcontrol the format; see Chapter 10.

To understand the viewer, a review of =EMPOWER= inputand mode representations is helpful. A circuit can haveexternal and/or internal inputs. External inputs aretransformed to eigenmode space, de-embedded, and nor-malized to characteristic impedances of eigenmodes. Theycould be one-mode or multimode (modally coupled) andthe incident wave for these inputs can be one of the inputeigenmodes. The incident wave is a harmonic function oftime. Its magnitude is unity, and it corresponds both toone Watt instantaneous power and 1/2 Watt time averagedpower. The initial phase of the incident wave is zero.Other eigenmodes of the structure are terminated by theircharacteristic impedances and are perfectly matched. Itnumerically represents a row of the generalized scatteringmatrix.

The internal ports are often locations where lumped ele-ments will be included by GENESYS. Parameters of thelumped elements are not required for the =EMPOWER=simulation. Thus internal ports default to 1 ohm normali-zation. In this case, the viewer data may not be as useful,since the lumped elements are not taken into account bythe viewer.

It is also possible to use an internal port as a source ofenergy to excite a structure. The termination impedancecan be specified using the option -NI<n>. In this case, theinternal inputs are terminated by virtual transmissionlines with the specified characteristic impedance. Theunit incident wave is excited at the specified input. Notethat if option -NI<n> is used then the external inputs arealso terminated by transmission lines (or loads) with thisimpedance after de-embedding and transformation intothe mode space if necessary.

98 Viewer

If the excitation conditions are defined, =EMPOWER=calculates the scattering matrix S with default or definednormalization first. Then it creates an excitation vectorA=[0...,1,....,0] that contains only one unit element corre-sponding to the specified input. The other elements of thevector are zeros. Reflected waves vector B are calculatedfrom the equation:

Then the simulator defines normalized voltages and cur-rents in mode space, denormalizes them, and restores thegrid currents and voltages inside regions corresponding toall input surface current regions. Finally, using the inputregion variables, the program calculates non-zero gridcurrents Ig for strip-like structures or voltages Vg forslot-like structures. The grid currents and voltages arelocally defined model currents and voltages (see the The-ory Chapter), and their units are Amperes and Voltsaccordingly. The grid currents and voltages together withtheir coordinates on the grid are stored in the .EMV file.(The same data can be written in the self-documented textfile with the extension .PLX) The viewer reads the .EMVfile and to displays data.

Note that the initial current (voltage) distribution is amodel representation and is treated using complexnumber conventions. The currents (voltages) are complexquantities and harmonic functions of time. So, their mag-nitudes are maximal values for the excited wave period.The real component corresponds to instantaneous valuesof currents, and their phases reflect the phase delays ofcurrents at the initial time t=0. Using these initial datathe current distribution is calculated versus time. If f isthe incident wave frequency, the current distribution Ig(t)at time t is given by expression:

B S A= *

I t I j ftg ga f a f= ⋅0 2exp( )π

Viewer 99

The same formula is valid for the voltage distributions.Advancing time displays snapshots of the current or volt-age distribution thus animating the display.

As we mentioned above, the viewer reads the grid currents(or voltages) with their coordinates and prepares them forplotting. The preparation stage includes a transformationof the grid variables to more general current densityfunctions (surface electric current density function forstrip-like problems or surface magnetic current densityfunction for slot-like problems). The units for the electriccurrent density magnitudes are Amperes per millimeter(A/mm). The units for the magnetic current densities areVolts per millimeters (V/mm). We choose millimeters toscale graphs to more readable values. The current densityfunctions are created only for the currents in the signal ormetal layer. Viaholes and z-directed ports are alwaysrepresented as z-directed currents in Amperes.

To summarize viewer behavior:

• If Generate Viewer Data is selected, the default incidentwave is the first eigenwave of the first input.

• Define the input number and mode number in the=EMPOWER= properties dialog

• An incident wave is a time harmonic function with unitmagnitude and zero initial phase.

• The external ports are terminated by corresponding modecharacteristic impedances while the internal ports areterminated by 1 Ohm if another termination is not definedby the option -NI<n>.

• The instantaneous power of the incident wave is 1 Wattand time average power is 1/2 Watt.

• Surface current density functions are used for the signal ormetal layer and integral currents are used for viaholes andz-directed inputs.

100 Viewer

Chapter 8

Box Modes

A fully enclosed rectangular box acts as a cavity reso-nator. At frequencies near each resonance modesignificant coupling exists between the desired sig-

nal metalization and the cavity. Because this coupling isreciprocal coupling occurs between segments of the signalmetalization. This is nearly certain to perturb the circuitresponses as the operating frequency approaches or ex-ceeds the first resonant frequency of the cavity. While=EMPOWER= inherently predicts these effects, they mayhave a significant destructive effect on the performance ofyour designs. Box modes are clearly illustrated in an ex-ample in the Examples manual.

HOMOGENEOUS RECTANGULAR CAVITY

In the formulation which follows we use definitions fromChapter 2, Geometry. The height of the box in the zdirection is h, the length of the box in the x direction is aand the width of the box in the y direction is b.

The resonant wave number for a rectangular cavity is

(MKS units) and the resonant frequency when homogene-ously filled with material with a relative dielectric con-stant of εr is

km

a

n

h

p

bmnp = FH

IK + F

HIK + F

HIK

L

NM

O

QP

π π π2 2 212

where c is the velocity of light in a vacuum,2.997925×108m/sec. The frequency of the dominant modeis f101 (lowest resonant frequency) and in a vacuum wehave

In air, with linear dimensions in inches and the frequencyin megahertz

With linear dimensions in millimeters and the frequencyin gigahertz

For example, in air (εr=1.0006) with a 2×4 inch, 0.5 inchhigh box, b=101.6mm, a=50.8mm and h=12.7mm. Thenk101=69.14 and f101=3297 MHz.

HIGHER ORDER BOX MODES

It is interesting to note that if h<a and h<b then thefrequency of the dominant mode is not a function of thecavity height. This is not the case for certain higher ordermodes. The mode which is next higher in frequency thanthe dominant mode is a function of the relative values ofh, a and b.

Consider for example the previous 2×4×0.5 inch box (or,any size box with the size ratios b=2a and h=a/4). There-fore the wave numbers are

fk c

mnpmnp

r

=2π ε

fc

a b101 2 22

1 1= +

f MHz inchesa b101 2 2

59001 1= • +

f GHz mma b101 2 2

149 81 1= • +.

102 Box Modes

The wave numbers for the lowest frequency modes for thisshape box and the resonant frequencies with a=2 inchesare listed here:

MODE WAVE # FREQ(MHz)@a=2inches101 1.118π/a 3299102 1.414π/a 4173103 1.803π/a 5319201 2.062π/a 6083104 2.236π/a 6598105 2.693π/a 7945301 3.041π/a 8974106 3.162π/a 9331

Notice that higher order modes occur frequently afterdominant mode resonance. It is possible to minimizeperturbations in narrowband applications by operatingbetween resonant frequencies. However, the above analy-sis assumes a pure homogeneous rectangular cavity anddielectric. Partial dielectric loading and signal metalwithin the cavity will influence the frequency. A moreconservative and safer approach is to enclose the circuitin a box with the dominant resonant mode higher than thehighest frequency of interest.

PARTIAL DIELECTRIC LOADING

If the cavity is not homogeneous, but instead is partiallyfilled with a dielectric and the remainder of the cavity isfilled with air then the dominant mode resonant frequencyis reduced and may be approximated using a filling factor[Johnson, 1987]. Assuming the substrate is mounted onthe floor of the cavity, the resonant frequency of a partiallyfilled rectangular cavity, fpartial, is

ka

m np

mnp = + +FHG

IKJ

π 2 22

164

Box Modes 103

where t is the thickness of the substrate and h is the heightof the cavity without a substrate. For example, f101 forthe 2×4 inch box is reduced from 3299MHz to 3133MHzwith t=62mils and εr=4.8.

This expression is approximate because the electric fieldlines are not parallel to the z axis and a component of theselines terminate on the side walls. This mode is referred toas a quasi-TEM101 mode.

SIGNAL METAL EFFECTS

Relatively sparse signal metal has little effect on theresonant frequency. Larger metal segments, particularlywhen grounded, significantly reduce the resonant fre-quency. To obtain a feel for the significance of signal metalyou may add extraneous metal to the substrate in Exam-ple 10, Box Modes, and observe the shift in the transmis-sion peaks.

TOP COVER

Transmission line discontinuities disturb current flowand energy is lost from the transmission structure. Whilethis lost energy is typically small, the Q of the resonantcavity is high and coupling at these frequencies is signifi-cant. Removing the cover of the enclosure causes energyto be lost to free space and resonance effects are reduced.This greatly reduces coupling between metal segments ofthe circuit and it is evident in the responses given in theBox Mode example cited earlier with the cover removed.Effects of removing a top cover are illustrated in theExamples “EdgeCoupledOpen.WSP” and “BoxModes.WSP”. See your Examples manual for details.

f ft

hpartialr

r

= − − FH

IK101 1

1εε

104 Box Modes

CAVITY ABSORBER

A similar benefit may be derived by placing absorbermaterial on the cover or in the cavity. While the poorultimate rejection in the stopbands of filters is not recov-ered, heavy coupling between segments is avoided. Thisis sometimes necessary to eliminate oscillations of highgain amplifiers in oversize enclosures.

By far the most elegant and safest approach to minimizingbox mode problems is placing circuits in small enclosures.

Box Modes 105

Chapter 9

Examples and Benchmarks

T he example are now completely contained in the=EXAMPLES manual. Examples which illustrate=EMPOWER= include:

• Microstrip Line.WSP

• Stripline Standard.WSP

• Spiral Inductor 2.WSP

• Box Modes.WSP

• Film Atten.WSP

• Edge Coupler.WSP

• Dual Mode.WSP

• 8 Way.WSP

• Edge Coupled.WSP

• Coupled Stepped Z.WSP

• Tuned Bandpass.WSP

• Patch Antenna Impedance.WSP

The required RAM specified in the Examples manual isthe value estimated by =EMPOWER=. They are approxi-mate and are determined by algorithm rather than a testof memory used. The execution times are for a 266 MHzPentium II with 256Mbytes of RAM operating under Win-dows 98. In most cases execution time is for the disconti-nuity mode.

Chapter 10

Batch Runs/Console Mode

T he =EMPOWER= program operates in a 32 bit con-sole mode. It is run from a command prompt and itdoes not display graphics. It is launched automat-

ically by GENESYS when an =EMPOWER= run is re-quested (with its results redirected to a window). It mayalso be run by batch files which is convenient for sequen-tial overnight runs.

This chapter describes the use of =EMPOWER= from thecommand line, and how batch files are written to performsequential runs. Also given is a console window descrip-tion and detailed command line option information.

BATCH RUNS WITHIN GENESYS

Starting with GENESYS Version 7.0,multiple workspacescan be loaded, and all =EMPOWER= simulations can beupdated sequentially. This new capability makes the tech-niques given in this chapter much less important for mostusers. Simply open as many Workspace files as you need(Select Options from the Tools menu and check “AllowMultiple Open Workspaces”). Right-click on any of the=EMPOWER= simulations and press Recalculate Now.You will then be asked if you want to recalculate allsimulations; select “Yes”.

Note: You should probably check “Automatically saveworkspace after calc” if you are running long or over-night batches so that if there is a power outage you willnot lose your results.

THE CONSOLE WINDOW

Figure 10-1 shows a sample console window.

The objects on the second line are:

FREQ - The current calculation frequency.

Mode - DISC (discontinuity), LINE (line analysis), orLN+D (both).

View - Checked if viewer data is to be generated.

Loss - Checked if physical loss is being modeled.

Thin - Checked if thinning is enabled.

Symm - Displays the type of symmetry possessed bythe circuit being analyzed. This option can be XZ,YZ, Mirror, 2-way mirror, or 180o rotational.

The objects on the third line are:

Figure 10-1 Example of an =EMPOWER= console window.

110 Batch Runs/Console Mode

Estim Time - The estimated total time to completethe current calculation mode.

Each frq - The estimated calculation time per fre-quency in the current mode.

Estim RAM - The estimated total memory requiredfor the current simulation.

The fourth line displays the simulation time of the currentfrequency and symmetry, plus the symmetry stage.

The fifth line displays the calculation stage.

The lines below the fifth line describe the calculated datafor each frequency. During line analysis, the impedance(Z) and propagation constant (G) are displayed for eachfrequency. In the discontinuity calculation mode, the firstrow of the s-matrix is displayed at each frequency.

THE COMMAND LINE

In Windows 95 and Windows NT, the MS DOS prompt canbe launched from the Programs option on the Start menu.

GENESYS automatically sets the PATH variable to in-clude EAGLE\BIN, where =EMPOWER= resides. Tolaunch =EMPOWER=, use the following command:

EMPOWER [options] FILENAME.TPL

where the [options] parameter consists of any of the op-tions described in this chapter, and FILENAME[.TPL] isthe name of the TPL file to be simulated. If this file is notin the current directory, the entire path (e.g.C:\EAGLE\MYFILES\) must precede the TPL file name.

ASCII TEXT FILES

An ASCII text file is simply a file with no binary datainside. This type of file is created by the DOS EDITprogram, or the NOTEPAD program included with Win-

Batch Runs/Console Mode 111

dows. Word processors typically create binary data filesby default, but can create an ASCII text file if instructedto do so. For example, to use MS Word to create a text file:

1. Create a new document (or open an existing one)

2. Type the desired text

3. Select Save As from the File menu

4. Under Save As Type, select Text Only

5. Assign a name to the file, and click OK.

BATCH FILES

Note: See the earlier section on Batch Runs for a mucheasier method of running multiple analyses.

A batch file is simply a text file composed of a series ofDOS commands. The extension of a batch file is always.BAT. Any valid DOS command can be used in a batch file,and the whole list of commands can be executed by typingthe name of the batch file at the command prompt, andpressing ENTER. For example, suppose a file namedEMRUN.BAT has been created. The following commandloads the batch file and executes each command:

EMRUN

For your convenience, GENESYS writes a batch file withyour selected options each time your “Write Internal DataFiles” for an =EMPOWER= simulation.

BATCH FILES FOR MULTIPLE RUNS

If several =EMPOWER= runs are needed, a batch file canbe created to call a sequence of files. Suppose that thefollowing TPL files are to be run:

112 Batch Runs/Console Mode

1.C:\FILES\HAIRPIN.TPL2.C:\FILES\AMP.TPL3. F:\WORK\DATA\SPLITTER.TPL4.T:\USER\PUB\COMBO.TPL

To run these files sequentially, a batch file could be createdwith the following commands:

EMPOWER [options] C:\FILES\HAIRPIN.TPLEMPOWER [options] C:\FILES\AMP.TPLEMPOWER [options] F:\WORK\DATA\SPLITTER.TPLEMPOWER [options] T:\USER\PUB\COMBO.TPL

Where [options] describes the desired =EMPOWER= com-mand line options for each file.

All files generated by =EMPOWER= will be in the direc-tory where you run the batch file except one case. If aPORT descriptor contains a file name for the line parame-ters and de-embedding data, then corresponding files(*.RGF or *.Ln) will be created either in the same directorywith the TPL file (file name without a path) or in someother specified directory (file name with a path).

Calling Other Batch Files

By default, whenever another batch file is loaded, thecurrent batch file is terminated. The call command forcesDOS to return to the current position after executing thecalled batch file. For example, suppose the list abovecontains batch files instead of TPL files. It seems naturalto create a batch file containing the following commands:

C:\FILES\HAIRPIN.BAT ***INCORRECT!***C:\FILES\AMP.BATF:\WORK\DA TA\SPLITTER.BATT:\USER\PUB\COMBO.BAT[...more commands]

However, HAIRPIN.BAT is the only file that will be loadedbecause the current batch file is terminated when HAIR-

Batch Runs/Console Mode 113

PIN.BAT is loaded. To force DOS to return to the currentfile after executing HAIRPIN.BAT, use:

call C:\FILES\HAIRPIN.BATcall C:\FILES\AMP.BATcall F:\WORK\DATA\SPLITTER.BATcall T:\USER\PUB\COMBO.BAT[...more commands]

This will make the batch file work properly.

Again, as in the case of the simple batch file, all result andauxiliary files will be created in the directory where yourun the batch file calling other batch files. If a file nameis specified for the line parameters and de-embedding datain a PORT descriptor, corresponding file (*.RGF or *.Ln)will be either in the directory with the TPL file or in someother directory if a path is specified.

=EMPOWER= COMMAND LINE OPTIONS

Options may be passed to =EMPOWER= via the Cmd Lineentry cell in the =EMPOWER= Setup dialog box. Com-mand line options provide a means to request auxillaryinformation about the problem, to eliminate undesiredinformation, specify custom parameters for processingalgorithms and to enter additional data.

There are three ways to specify the command line options:

• Entering them in the MS DOS prompt before the TPL filename.

• Preparing a batch file with desirable options or editing anexisting batch file created by GENESYS.

• Specifying options in the GENESYS / =EMPOWER=Setup/Options menu either using check boxes or typingthem directly in the “Cmd Line” window.

Either - (minus) or / (slash) must precede all options.Options can be typed in uppercase, lowercase or mixtureand must be separated by at least one space. You can

114 Batch Runs/Console Mode

obtain a list of possible options using the following com-mand in the MS DOS prompt:

EMPOWER -HE

=EMPOWER= command line options are broken into thefollowing sections:

• General options (HE,CH,NT, NW, NL, NIr, VM, ME, FI, PT).

• Processing algorithm options (St, Ta, Tn, T, TM, TS,TX,TY,RT).

• De-embedding options (O, Oa, On).

• Current/Voltage data file options (Ini, Imj, IT, IB).

• Listing file options (Lpn, Ln, Lw, La, Ly)

• Debug options (Dceriu, Mi, Rl, Sp, Sf, Zs, Nd).

GENERAL OPTIONS

The general options are shown in Table 10-1.These optionsdo not fit any particular category.

-HE - HELP, described above.

-CH - Check memory requirements. This option providesa very important means both for checking the grid map-ping and required memory. We recommend using it everytime for a new or substantially modified problem. =EM-POWER= just maps the problem onto the grid and calcu-lates the required number of the grid variables for eachfrequency. Check the “map of terminals” in the listing fileto see the grid model of the problem and check the MEM-ORY lines in the listing file to get some idea about problemcomplexity and probable simulation time.

-NC - If this option is checked, =EMPOWER= will allowde-embedded ports to be away from the wall. This optionis especially useful for finline and slotline configurations.

Batch Runs/Console Mode 115

-NT - the program does not print information about thestage of solution (STAGE:) in the console window statusline. Since time is spent showing status information,small problems can be sped up somewhat with this option.

-NW - No warnings. If you run =EMPOWER= from a batchfile (especially overnight) you should use the option NW.The program prints out all messages in the listing file andin the console window. The console warning messagesgenerally wait for a keystroke to continue. With the optionNW the informative messages and warnings are justprinted out to the listing file. If there are choices in awarning, a default value will be picked. Be sure to checkthe Errors/Warnings/Info lines in the listing if you usedNW.

Option Default Purpose

HE Output help options list.

CH solve Just check problem and estimate memory required.

NC off Less checking of port positions

NT trace on Trace off (accelerate calculations).

NW warn No warnings (output warnings only in listing).

NL Loss(X) Do not use physical losses (for preliminary analysis).

NIr GEN&1Normalize all ports to “r” ohms before calculatingS-matrix to be printed in the log and listing file.

VM RAMPermit use of virtual memory to solve a largeproblem.

ME rewrite Merge new data with existing descriptor file.

FI no filesOutput characteristic impedances of modes to files*.Rn (for GENESYS).

Table 10-1 General =EMPOWER= options

116 Batch Runs/Console Mode

-NL - No Losses. This option is especially useful forpreliminary solutions. If you entered a problem that con-tains some media layers with physical losses or largemetal regions with losses described as physical and youdecide to ignore them for a preliminary analysis, you canjust use the NL option. This is much easier and safer thantrying to remove losses manually from all substrates andmetal layers. If you do not remember whether or not youused this option, look at the Loss( ) entry in the consolewindow status line. If checked, losses are being used.

-NIr - Normalizes all inputs to r Ohms. After de-embed-ding the external inputs, all external and internal inputsare normalized to the same specified impedance. The re-sult scattering matrix is used to generate data for thevisualization. The value is also printed out to the listingfile and to the console window. Without this option, theexternal inputs are normalized to their eigenmode imped-ances and the internal inputs are normalized to 1 Ohm.Note: The data files with extensions .SS always contain50-Ohm normalized S-matrices.

-VM - Allow virtual memory usage. To solve a complexproblem, =EMPOWER= always limits usage of computervirtual memory (hard disk space) in a rational way. It willnot use it for some numerically intensive parts of thesimulation. The option VM tells =EMPOWER= to usevirtual memory more freely. But even with this option, theprogram stops calculations if substantial hard disk spaceis involved in some parts of the simulation. Check theMEMORY lines in the listing file to have an idea how muchmemory your computer lacks or how to reduce the prob-lem.

-ME - Merge Data. If you need to add some frequencypoints to existing descriptor files, use this option. Theprogram checks the corresponding binary descriptor file(.Y), decides what frequencies need to be added, and solves

Batch Runs/Console Mode 117

the problem only at these frequencies. The resulting .SSfile contains both old and new data. It works only if theproblem geometry is not changed, otherwise all outputfiles will be rewritten. Use this option with care: theprogram does not track all geometry parameters so if youchange anything except the frequency sweep you shouldnot use it.

-FI - Output characteristic impedance files with exten-sions .Rn. Basically, these files are needed by =Super-Star= simulator if generalized S-Parameters will be

Option Default Purpose

St wireSolid model of thinning out (linear supergrid),default is wire model (keeps capacitance).

Ta -TaDetermine thinning out threshold automatically asN/20. N is the minimum wavelength in themedium at the critical frequency to cell size ratio.

Tn -Ta

Set thinning out or super-grid parameter, acceleratecalculations and decrease accuracy.If n=0, no thinning is performed.If n=10 (max. value) - corresponds to N/10.(N is defined as in the Ta description)

T -TaDo not use thinning out procedure (most accurateand slow).

TM -TM“Maximal” steps in thinning out algorithm (2^n),accelerate dramatically.

TS -TM“Smooth” steps in thinning out algorithm (n), use tocheck trustworthiness.

TX -TX -TY Thinning out along X axis only.

TY -TX -TY Thinning out along Y axis only.

Sg -Mesh generation based on complete geometry ratherthan on elements of a circuit (for consistency withversions 7.03 and earlier).

Table 10-2 Processing options

118 Batch Runs/Console Mode

displayed. Files with extension .Rn contain a table ofcharacteristic impedances of an input mode versus fre-quency. The number n corresponds to the input position inthe .SS file.

PROCESSING ALGORITHM OPTIONS

The processing algorithm options make it possible tochange and to manually optimize some parts of the =EM-POWER= algorithms. Some of these techniques are illus-trated in Chapter 9, Examples. A list of the options withshort descriptions is given in Table 10-2. All these optionsare related to the thinning out algorithm which basicallyreduces a number of variables in the problem solution andaccelerates calculations. However, this problem reductionincreases computational errors because currents forthinned out regions are only approximated. Thus, thethinning out process can be customized by the userthrough various options. Accuracy and simulation speedare tradeoffs. Intellegent use of these options requiresexperience. If accuracy is paramount and sufficient mem-ory is available, run a few frequencies without thinningout.

If the thinning out algorithm is turned on, the correspond-ing position of the console window status line is checked[Thin(X)].

-St - Two models are available for a secondary grid gen-eration: wire thinning out and linear super-grid or solidthinning out. The default is wire model. It is faster butmay not be as accuracte for structures with large metal-lized regions. For these problems the solid model or linearsuper-grid should be used. It is not as fast but is moreaccurate with large metallized regions. It is a consolida-tion algorithm because of it unites grid cell groups. Theoptions St allows you to switch to this mode, and all otheroptions work with the solid model.

Batch Runs/Console Mode 119

-Ta, -T, -Tn - Specify thinning out amount. The thinningout threshold controls the largest number of grid cells thatare allowed to be processed together. This value is relatedto the wavelength in the medium at the critical frequency.The default value of -T5 corresponds to a maximum lengthof λ/20. Therefore, with -T5 the maximum number of cellsthat are thinned together is equal to λ/(20d). You canchange this value using the option Tn. If n=0 (-T0, orsimply -T), the problem will not be thinned out at all.Values of n from 1 to 10 correspond to the following valuesof the threshold: λ/80, λ/60, λ/40, λ/30, λ/20, λ/18, λ/16, λ/14,λ/12, λ/10. Use more aggressive thinning out (such as-T10) with the solid model, and less aggressive with thewire model (such as the default -T5).

-TM, -TS - Control advanced thinning parameters. (Nor-mally, users do not need to use these options.) The “maxi-mal” and “smooth” thinning out algorithms define thesequence used in choosing the number of cells to thin outtogether. The “maximal” algorithm (default) correspondsto the increment 2^n, where n=1,2,3,...until 2^n < thethreshold. The “smooth” algorithm corresponds to the sim-ple increment n=1,2,3,...until n < the threshold.

-TX, -TY - Allow thinning out only along one coordinateaxis. (Normally, users do not need to use these options.)Use it to suppress the thinning out along the other axis ifa problem is sensitive to grid changes in this direction.

Option Default Purpose

Oa -OaDetermine automatically line segment lengths fordeembedding (default)

On -OaSet line segments for deembedding at n timesshorter than box sizes.

Table 10-3 De-embedding options

120 Batch Runs/Console Mode

-Sg - generate secondary mesh for the thinning out andsuper-grid prcedures on the base of a complete geometryof a circuit. Starting from the version 7.5 the mesh gen-eration procedure is based on meshing circuit elementsseparately taking into account their connections. It pro-duces more efficient meshes that leads to faster solutions.The option is added for consistency with versions 7.03 andearlier.

DEEMBEDDING OPTIONS

The deembedding options (see Table 10-3) define thelength of line that is created to calculate the line parame-ters and de-embedding data for an external port. By de-fault (or with the option -Oa), =EMPOWER= sets thesegment length equal either to the box size along the linesegment or to one quarter of the minimum wavelength inthe medium at the maximal critical frequency, whicheveris smaller. To rewrite this length value, use the option -On,where n=0,1,2, .... . Here, -O0 and -O1 as well as the simpleoption -O correspond to the line segment length exactlyequal to the box size along the line. Values of n > 1 will setthe lengths n times shorter then the corresponding boxsize and are useful if a problem contains a relatively widestripline input and the corresponding line analysis withthe default line segment length would take a long time.You can check the default length and complexity of the lineanalysis problem using the option -CH. Usually the linesegment can be two or even four times shorter then thedefault value without considerable effect on accuracy. Itis difficult to give a recommendation about the optimallength. It basically depends the high order modes thatcould result in parasitic coupling between the line seg-ments and adjacent ports. If you are sure that the attenu-ation of the first high order mode is large enough to haveno effect on a segment length, override the default length.It could save significant computational time. On the other

Batch Runs/Console Mode 121

hand, if the first high order mode is close to the cutoff or,even worse, it is propagating, the line analysis results andde-embedding data could be absolutely wrong even forsegments longer than the default.

The line analysis and de-embedding data are stored inbinary files with .RGF or .Ln extensions. These files alsocontain information about the line segment geometry. Theprogram tracks them and if the line geometry is changed,the data will be recalculated and the file will be rewritten.The program does not control the line segment lengthwhen options -Oa or -On specified. So, if you changed themaximal critical frequency and want to see that affect onthe de-embedding line segment length, delete correspond-ing .RGF files.

VIEWER DATA FILE OPTIONS

Options related to viewer data are listed in Table 10-4. Togenerate viewer data, you need to specify an input and anincident eigenwave using the options -Ini and –Imj. Theinput number i corresponds to an input using the orderfrom in the DEFnP line. Modally coupled inputs arecounted as one. If the input is multimode (modally cou-pled), there is a choice of modes to be excited. The defaultmode is the first one (with the largest propagation con-stant). The modes are ordered in accordance with their

Option Default Purpose

Ini i=1 input number i.

Imj j=1 eigenwave number j.

IT ASCII Output plot data file in text format (.PLX).

IB binaryOutput plot data file in both binary (.EMV) and textfile (.PLX) formats.

Table 10-4 Viewer data file options.

122 Batch Runs/Console Mode

propagation constants. See the line analysis mode resultsin the listing file for more details (characteristic imped-ances, integral current/voltage distributions across theline and so on). The maximal mode number is limited bythe number of PORTs or excitation regions in the input.Only one mode can be exited at a time.

If the option -Ini is used, the program creates a binary filewith the extension .EMV to store and to pass the visuali-zation data to the VIEWER program. If the option -ITspecified the current/voltage distribution data are printedout in an ASCII text file with the extension .PLX. Use thisfile if you need to process the data using another program.With the option -TB, the program creates both the binaryand the text files.

LISTING FILE OPTIONS

The listing file options are shown in Table 10-5. Theyconfigure the data printed into the listing file.

-Lpn - sets the length of a listing page. The pages aredivided by a page break symbol and contain n text lines.By default the listing is not broken into the pages. This

Option Default Purpose

Lpn no pagesSet listing file page size. n is a number of lines in apage (n>10).

Ln num. off Numerate lines in the listing file.

Lm map Do not print out maps of terminals.

Lw no wrap Wrap maps of terminals.

La no info. Output additional checking information.

Ly S-matr. Print out generalized unnormalized Y-matrices.

Table 10-5 Listing file options

Batch Runs/Console Mode 123

option is only necessary if you will be dumping the listingfile directly to a printer and page breaks are required.

-Ln - Number all lines in a listing file. It is useful forreference.

-Lm - Do not print terminal mapping. The “map of termi-nals” that reflects the grid representation of a problem issometimes the largest portion of a listing file. Thus, if youchecked the map already and do not need it in the repeti-tive runs, use this option. The maps are not printed andlisting files are much smaller.

-Lw - Wrap terminal maps at 80 columns. A possibleproblem with the maps of terminals is that they are

Option Default Purpose

Dc no printPrint terminal parameters and renumerationtables.

De no printPrint eigenvalues of the Grid Green’s Function(GGF) matrix.

Ds no print Print general sums of the GGF matrix eigenvalues.

Dr no print Print reduced GGF and inverted GGF matrices.

Du no print Print unfolded multiport immitance matrix.

Mi skipping Use complete Gauss’ matrix inversion algorithm.

Sp accountSolve problem without accounting geometricalsymmetry.

Sf impose Do not superimpose fictive symmetry.

Zs hole outDo not hole out metal regions with z-directedcurrents.

Nd re-expand Do not re-expand non-divergent currents.

Table 10-6 Debug options

124 Batch Runs/Console Mode

usually wide. It means that listing lines corresponding tothe map are long. You can print the maps as 80-columnsfragments using this option.

-La - Add additional information to the listing file. Themain parts affected are:

• Information about the shielding box and media layers(Package Structure).

• Tables of surface impedances in Ohms per square versusfrequency (Impedances of Surface ...).

• The general grid parameters of inputs (Input number ...general data) and line and de-embedding parameterstaken from the corresponding RGF files (Input ...parameters from the file ...).

-Ly - Print out generalized unnormalized Y-matrices. Thisis an intermediate descriptor matrix after transformationto a mode space and de-embedding.

DEBUG OPTIONS

The debug options (see Table 10-6) provide numericalalgorithm insight and select to more reliable but lessnumerically efficient forms of the algorithms.

-Dc, -De, -Ds, -Dr, -Du - Options starting with D are solelyfor debugging purposes and are listed here for complete-ness. Users will seldom use them.

-Mi - Use standard Gauss inversion algorithm. To invertthe reduced Grid Green’s Function (GGF) matrix we nor-mally use a modified Gauss’ algorithm which is much morenumerically efficient.

-Sp, -Sf - Disable the geometrical symmetry processingalgorithm.

-Zs - Do not “hole out” metal regions with z-directedcurrents. To decrease the number of variables related witha viahole, a hole out algorithm was introduced. Basically,

Batch Runs/Console Mode 125

it leaves only currents corresponding to the viahole sur-face. It works only if a viahole is coated by metal in the XYplane. You can see the results of the holing out on the “mapof terminals” in the listing file. If this algorithm is notsuitable, turn it off with this option.

-Nd - The base element of the wire and solid thinning outalgorithms is a linear re-expansion procedure. It reducesthe number of variables (currents and voltages) alongsome lines to two by assuming linear dependencies ofcurrents along the line. It suppresses divergence of cur-rents along the lines and gives staircase plots for currentdensities. This option provides smoother current plots.

126 Batch Runs/Console Mode

Chapter 11

TPL File Format

T his chapter contains information on writing text in-put files for =EMPOWER=. It is for users want tocreate circuits with a text file from internal company

programs, but it may also be useful for others who wantto know the detailed structure of the input data.

For each electromagnetic problem to be solved, a text fileis created using =EMPOWER=’s proprietary input lan-guge. The input file a file is a text file and in general issimilar to a text circuit or netlist file. The suggestedextension for these files is .TPL, thus they will also bereferred to as TPL (Topology) files. If you are usingGENESYS and =LAYOUT= to create =EMPOWER= cir-cuits then a TPL file will always be automatically gener-ated in the structured storage. If you write internal=EMPOWER= files, EMPOWER.TPL will be writtenalong with EMPOWER. BAT with options chosen to run=EMPOWER=. The batch file contains the command ac-tually run by GENESYS, allowing you to easily rerunEMPOWER.EXE later from the command prompt withthe same options.

Note: To see a TPL File automatically created inGENESYS, check the “Additional Information in ListingFile” box in the =EMPOWER= dialog and choose “Viewlisting” from the =EMPOWER= right-click menu.

TPL files can also be entered manually using a text editor(such as EDIT or NOTEPAD).

TPL FILE FORMAT OVERVIEW

The general structure of a TPL file is shown in Table 11-1.Each line of the file begins with a keyword identifying thefollowing parameters. Optional lines and parameters areshown in square brackets, […]. Possible choices of pa-rameters are separated by slash signs and are listed incurly braces …/…/…. Default parameter values are un-derlined. A long line can be continued on the next line,marking the start of the next line with an ampersand (&).Parameter names are optional if the parameters are en-tered in the order shown in Table 11-1, however, parame-ters with names can be entered in any order. ! and ‘ arecomment markers, so anything on a line following them isignored. Blank lines are legal. You can type TPL files inuppercase, lowercase or a mixture.

Each TPL file can contain three main blocks: dimension(DIM), equation (EQUATE), and circuit (CIRCUIT). TheDIM and EQUATE blocks are optional, while the CIR-CUIT block is required.

The dimension block begins with DIM and should precedethe CIRCUIT block. The frequency and length units canbe defined in the DIM block.

128 TPL File Format

! Comments.

[DIMFREQ HZ/KHZ/MHZ/GHZ/THZLNG UM/MM/CM/M/MIL/IN]

[EQUATEequations]

CIRCUIT[EMLINE] ‘More comments.PACKAGEEMFRQS F0=f0 F1=f1 COUNT=nf[MAXFRQ fmax][TOLERANCE T=t]DELTA X=dx Y=dySIZE X=a Y=b[TOP_W TYPE=METAL/MAGNETIC/OPEN RHO=rho TMET=tmet

& ROUGH=rough / Z=file/impedance]MEDIA H=hP [E=eP] [U=mP] [TAND=tandP/SIGMA=Sp]…LAYER N=1 [TYPE=CURRENT/VOLTAGE]…[MEDIA H=h2 [E=e2] [U=m2] [TAND=tand2/SIGMA=S2]][LAYER N=2 [TYPE=CURRENT/VOLTAGE] ]MEDIA H=h1 [E=e1] [U=m1] [TAND=tand1/SIGMA=S1][BOTTOM_W TYPE=METAL/MAGNETIC/OPEN RHO=rho TMET=tmet

& ROUGH=rough / Z=file/impedance]END_PACKAGE[LAYER N=1]PORT k1 [F=file] [CD=X/Y] [LD=X/Y] [RPL=shift] X1=x1 Y1=y1 X2=x2 Y2=y2…PAD kn X1=x1 Y1=y1 X2=x2 Y2=y2 CD=X/Y/Z [TOL=BOTTOM/TOP]…[LOSSLESS][LOSSLESSX][LOSSLESSY][LOSSLESSZ][SURFACE RHO=rho TMET=tmet ROUGH=rough / Z=file/impedance][SURFACEX RHO=rho TMET=tmet ROUGH=rough / Z=file/impedance][SURFACEY RHO=rho TMET=tmet ROUGH=rough / Z=file/impedance][SURFACEZ RHO=rho TMET=tmet ROUGH=rough / Z=file/impedance]…[RECT X1=x1 Y1=y1 X2=x2 Y2=y2 [CD=X/Y/XY/Z/XZ/YZ/XYZ]

&[TOL=BOTTOM/TOP]][POLYGON X1=x1 Y1=y1 X2=x2 Y2=y2… X99=xn Y99=yn

&[CD=X/Y/XY/Z/XZ/YZ/XYZ ] [TOL=BOTTOM/TOP]]…

DEFnP k1... (ki,...,kj) … kn EMPOWER

Table 11-1 General TPL-file structure.

TPL File Format 129

DIMStarts the dimension block. Sets dimensions for all valuesentered into TPL file.

Format:DIM[FREQ HZ/KHZ/MHZ/GHZ/THZ][LNG UM/MM/CM/M/MIL/IN]

The optional dimension block begins with DIM. The de-fault frequency and length units can be overridden in thisblock. Possible choices for them are shown in Tables 11-2and 11-3.

The default units are MegaHertz (MHZ) and millimeters(MM). In =EMPOWER=, all units will be converted inter-nally to Hertz and meters. These internal units will alsobe used to print out parameters in the listing file. Fre-

UnitConversion toHz

Keyword

Hertz 1 HZ

Kilohertz 1.0e3 KHZ

Megahertz 1.0e6 MHZ

Gigahertz 1.0e9 GHZ

Table 11-2 Available frequency units.

Unit Conversion to M Keyword

Micrometer 1.0e-6 UM

Millimeter 1.0e-3 MM

Centimeter 1.0e-2 CM

Meter 1 M

Mil 2.54e-5 MIL

Table 11-3 Available length units.

130 TPL File Format

quencies in *.SS files and printed on the console screenwill be in Megahertz (the GENESYS default).

Example 1:

The following dimension block sets the frequency units tohertz and length units to meters:

DIMFREQ HZLNG M

Example 2:

The following dimension block sets the frequency units toGigahertz and length units to millimeters:

DIMFREQ GHZLNG MM

TPL File Format 131

EQUATEStarts the optional equation block.

Format:EQUATEequations

Example:EQUATEw=1dx=w/8dy=w/8a=5*wb=5*w

=EMPOWER=’s equation block is virtually identical toGENESYS’s global equation equations. There are threefeatures from GENESYS which are not available: tunablevariables, function definitions, and post-processing/fre-quency dependent variables (FREQ keyword).

The equation block example above shows how some pa-rameters can be conveniently interrelated. This examplecontains the definition of one independent variable w andfour variables dx, dy, a, and b defined from w.

132 TPL File Format

CIRCUITStarts the circuit block. Must be present in all =EM-POWER= files.

Format:CIRCUIT[EMLINE]PACKAGE…

END_PACKAGE…DEFnP k1... (ki,...,kj) … kn EMPOWER

Example:CIRCUITPACKAGE.EMFRQS ...DELTA ...SIZE ...MEDIA ...LAYER N=1MEDIA ...

END_PACKAGE...PORT 1 X1=...PORT 2 X1=...PAD 3 X1=......DEF3P 1 2 3 EMPOWER

The circuit block begins with CIRCUIT and ends with theDEFnP line. It contains a complete description of theproblem geometry. It can be divided into two main sec-tions: package sub-block and geometry section.

The package sub-block begins with PACKAGE and endswith END_PACKAGE lines. Frequencies for analysis,

TPL File Format 133

grid parameters (cell sizes), and specifications of the boxand its media layers must be specified in this sub-block.

The Geometry section is the rest of CIRCUIT block afterthe package sub-block and before the DEFnP line. Itcontains a description of the planar regions in the signallayer (metallization and resistive films), via-holes to thetop or bottom cover of the box, and inputs. Inputs to bede-embedded are specified as PORTs, while inputs with-out de-embedding and places to connect lumped elementsare described as PADs. The only required entry in thegeometry section is at least one PORT or PAD.

The DEFnP line (n-port network definition line) ends thecircuit block. It contains a list of all numbers specified inthe PORTs and PADs in the geometry section as well asthe fixed network name EMPOWER. PORT numbersconfined in parenthesis are treated as coupled (modallyrelated). In the resulting listing and data files all inputswill be renumbered according to their order in the DEFnPline.

EMLINE/LINE ANALYSIS MODE

Structures are classified in two categories: discontinuitiesand line segments. Line segments are defined as struc-tures with a set of identical PORTs on opposite sides of theshielding box interconnected by an arbitrary number ofconductors. All ports at the side of the box should bedescribed as coupled in the DEFnP line. The number ofports at one side of the box is the same as the number ofeigenwaves (modes) to be evaluated.

By default, =EMPOWER= performs both line and discon-tinuity analysis. If the keyword EMLINE is found in thecircuit block, then only the line analysis will be performed.

134 TPL File Format

Format:CIRCUITEMLINE...

In this case the propagation constants and impedances ofthe eigenwaves are evaluated as well as the de-embeddingparameters. They will be printed out in the listing file andrecorded in a binary file with the extension .RGF. Themain part of the file name may be specified with the firstPORT descriptor, otherwise the name of the current TPLfile is used.

Without the EMLINE operator, de-embedded descriptor(Y and S) matrices of the line segment will be calculatedin addition to the line and de-embedding parameters.

DISCONTINUITY ANALYSIS

In discontinuity analysis, a text file with extension .SSadded to the name of the TPL file is also written. Itcontains 50 Ω normalized S-parameters in an industrystandard format which can be used directly by GENESYS.Together with input impedance files with extensions .Rn(where n is the input number) they are interface files forGENESYS and can also be used to pass =EMPOWER=results to other circuit simulators.

Structures that do not meet criteria of the line segmentare treated as discontinuities. In this mode the de-embed-ding parameters of all PORTs will be automatically pre-calculated and written in binary files with names specifiedin PORT lines (with default extension .RGF). If the namesare omitted, they will be generated from the main nameof the TPL file by changing the extension to .Ln, where nidentifies the set of PORT numbers.

After the line analysis is finished, the discontinuity (com-plete circuit) itself is simulated. The resulting de-embed-

TPL File Format 135

ded un-normalized admittance matrix is stored in a binaryfile with extension .Y added to the main name of the TPLfile. The generalized S-matrix (with PAD type inputsnormalized to 1Ω) is also calculated and printed out in thelisting file.

For a more detailed description of these files see AppendixB.

136 TPL File Format

PACKAGEFormat:

PACKAGEEMFRQS F0=f0 F1=f1 COUNT=nf[MAXFRQ fmax][TOLERANCE T=t]DELTA X=dx Y=dySIZE X=a Y=b[TOP_W TYPE=METAL/MAGNETIC/OPEN RHO=rho

& TMET=tmet ROUGH=rough / Z=file/impedance]MEDIA H=hP [E=eP] [U=mP] [TAND=tandP/SIGMA=Sp]…LAYER N=1 [TYPE=CURRENT/VOLTAGE][MEDIA H=h2 [E=e2] [U=m2] [TAND=tand2/SIGMA=S2]][LAYER N=1 [TYPE=CURRENT/VOLTAGE]]MEDIA H=h1 [E=e1] [U=m1] [TAND=tand1/SIGMA=S1][BOTTOM_W TYPE=METAL/MAGNETIC/OPEN

& RHO=rho TMET=tmet ROUGH=rough/Z=file/impedance]END_PACKAGE

z

x

y

h

h...

h

2

1

P

εε

ε

1

2

P

TOP_W

BOTTOM_W

SIDEWALLS

µ

µ

µ

1

2

P

a

b

(MEDIA) LAYER

Figure 11-1 General structure of a 3-dimensional rectangularregion and its correlation with the package sub-block.

TPL File Format 137

The Package sub-block is everything between the PACK-AGE and END_PACKAGE lines as shown above. Squarebrackets […] indicate optional information, curly bracketswith slashes …/…/… means that you are to select one ofthe choices. Fig. 11-1 gives a general idea of the corre-spondence between a real circuit and the package sub-block descriptors. In this sub-block the followingparameters of the problem are defined:

• Frequency range and number of frequency points (EMFRQ)

• Maximal critical frequency (MAXFRQ)

• Input position control parameter (TOLERANCE)

• Grid cell sizes (DELTA)

• Shielding box size in signal layer plane (SIZE)

• Parameters of the top box cover (TOP_W)

• Parameters of homogeneous media layers (MEDIA)

• Type and position of signal layer (LAYER)

• Parameters of the bottom box cover (BOTTOM_W)

See Chapter 2, Geometry, for more information on packageparameters.

138 TPL File Format

EMFRQSpecifies the frequencies for =EMPOWER=analysis.

Format:EMFRQS [F0=]f0 [F1=]f1 [COUNT=]nf

This line specifies a set of frequencies at which the circuitis to be simulated. Frequency units are defined in the DIMblock. (Default units are MHz.) The frequency range isdefined by start (f0) and stop (f1) frequencies and must bepositive (non-zero) numbers. nf is the number of frequen-cies to be analyzed (a positive integer). The problem willbe analyzed starting from the frequency f0 up to thefrequency f1 with equidistant steps of size (f1-f0)/nf.

Example:

The following sample defines a frequency set of 6 pointsfrom 5 to 20 GHz. It is assumed that the units are definedas GHz in the DIM block.

…EMFRQ F0=5 F1=20 COUNT=6…

The same result can be achieved with the next line:

…EMFRQ 5 20 6…

TPL File Format 139

MAXFRQSpecifies the maximum critical frequency.

Format:[MAXFRQ fmax]

MAXFRQ is specified in the units defined in the DIMblock. (The default units are MHz.) Parameters of thesolution quality, thinning out thresholds, and lengths oflines for de-embedding are based on the maximum criticalfrequency value. In other words, this value influencesboth accuracy of simulation and calculation time. De-creasing the value accelerates simulation but may in-crease model error, especially at frequencies above thevalue. On the other hand, an unnecessarily high valuemay slow down the solution without visible improvementsin accuracy.

An important reason to specify MAXFRQ:By default thisvalue is set equal to the highest sweep frequency speci-fied in EMFRQ. Even a small change of its value maycause the grid to change, forcing recalculation of de-embedding parameters and unnecessarily increasingsimulation time as a consequence. This change willalso change the answer slightly, with disastrous resultsif you are merging data. This will not happen if you useMAXFRQ. It is also important to remember to update itif you change the frequency range substantially.

Example:

The following sample sets maximum critical frequencyequal to 20 GHz. It is assumed that the frequency dimen-sions are set to GHz in the DIM block.

…MAXFRQ 20…

140 TPL File Format

TOLERANCE

Specifies allowance for shift of PORTs from the box side-walls.

Syntax:[TOLERANCE [T=]t]

Units for t may be defined in DIM block. (The defaultunits are millimeters.) PORT inputs should almost al-ways touch a sidewall, as any other situation is normallyan erroneous definition of a PORT. But if you are sure thatyou need to shift PORT from the sidewall by value t, youmust to specify t in TOLERANCE line. This value may bereferred to as the input position control parameter.

Example:

The following sample line allows PORTs to be specifiedwhich are up to 2 mils from the sidewall. It is assumedthat MIL is specified for the length unit in the DIM block.

…TOLERANCE T=2…

TPL File Format 141

DELTASpecifies the grid cell size.

Syntax:DELTA [X=]dx [Y=]dy

This line specifies grid cell sizes (discretization distances)in the plane of the signal layer (xy-plane). (See Figure11-1.) Parameter X specifies the cell size along the x-axis(dx), while parameter Y specifies the cell size along they-axis (dy). Units for dx and dy are defined in DIM block.(Without a DIM block they are in millimeters.)

The specified DELTA has the following effects on problemgeometry:

• The entered electromagnetic problem will be mapped ontothe grid with cell size dx by dy. This imposes somerestrictions on the problem geometry.

• All dimensions of the box must be defined in accordancewith the cell sizes and must fit to the grid in the same way(or vice versa). In other words, the x and y dimensions ofthe box must be proportional to the grid cell sizes.

• Metallization edges will preferably be positioned on the gridin the same way. As a consequence, changes indimensions will have influence on the problem solution onlyif they are larger than the cell sizes.

The cell size values also directly affect accuracy: Theeffective simulation accuracy depends mainly on the ratioof the wavelength to the larger of the discretization steps.You can find this value at at the maximum critical fre-quency in the section of listing file named “QCHK: Pa-rameters of the solution quality”. Respective convergencestudies have established that a ratio exceeding 40 isusually sufficient for reliable results. If the ratio is lessthen 20 =EMPOWER= issues a warning. Each particulargeometry type, however, requires its own tentative conver-gence analysis with various cell sizes. It must be men-

142 TPL File Format

tioned that enhanced accuracy (reduced cell size) alwayscoincides with increasing calculation time in approxi-mately cubic proportion. Therefore it is recommended toperform the first stages of a device design at a low accuracylevel (with grid cells as big as possible) and to increaseaccuracy (decrease cell size) at the final stage while refin-ing the analysis results.

Example.

The following line in the package block sets both cell sizesto 0.2 mm. It assumes that the length units have been setto mm:

…DELTA X=0.2 Y=0.2…

The next line will have the same effect:

…DELTA 0.2 0.2…

TPL File Format 143

SIZESpecifies the dimensions of the shielding box along the xand y axes. (See Figure 11-1.)

Format:SIZE [X=]a [Y=]b

The dimensions of the shielding box along x and y axes inthe plane of partial metallization are defined by the X andY parameters in the SIZE line. Length units are millime-ters unless they are overridden in DIM block. a and bmust be multiples of the respective grid cell sizes definedin the DELTA line.

When analyzing a device as a whole, the it is best to set aand b equal to the sizes of the physical shielding box.=EMPOWER= accounts for the shielding box by simulat-ing the interaction between the device elements and thebox in a modal sense.

When analyzing a piece of a complex device or a simplediscontinuity, the choice of a and b is not so well defined.The sidewalls must be far enough away to minimize theeffect of the shielding walls on the element or discontinu-ity characteristics. It must be stated that extension of theregion size in the metallization plane increases simulationtime and can cause some undesirable effects like shieldingbox resonances. Hence the region size should be kept closeto adequate minimum. On the other hand, placing theshield walls too near to the discontinuity may substan-tially distort its characteristics. Consequently, for criticalapplications, a small-scale case study of such influencesmight be expedient.

Example:

To describe a shielding box 5x5 mm in the signal plane,the following line has to be entered in the package sub-block:

144 TPL File Format

…SIZE X=5 Y=5…

The next line will have the same effect:…SIZE 5 5…

TPL File Format 145

TOP_W / BOTTOM_WOptional lines describing the top and bottom box walls (topand bottom covers). The default is lossless metal.

Format:TOP_W TYPE=METAL/MAGNETIC/OPEN [RHO=rho& [TMET=tmet] [ROUGH=rough] / Z=file/impedance]

BOTTOM_W TYPE=METAL/MAGNETIC/OPEN [RHO=rho& [TMET=tmet] [ROUGH=rough] / Z=file/impedance]

The characteristics of the top and bottom walls of theshielding box are given in the lines TOP_W and BOT-TOM_W (see Figure 11-1). The type of the walls can bedescribed as METAL, MAGNETIC or OPEN.

The METAL type is the default and defines either a purelossless electrical cover, or physical metal if parametersRHO/TMET/ROUGH or Z are entered. RHO is metalliza-tion resistivity relative to copper, ROUGH is surfaceroughness, and TMET is just for consistency with theSURFACE keyword and will be ignored. Default lengthunits are millimeters, unless they redefined in the DIMstructure. If RHO=0 or is not entered at all, the corre-sponding cover will be treated as an ideal electric wall(zero of tangential electric field). For a more detaileddescription of these parameters see the SURFACE entryin the geometry section.

Surface impedance of the box covers can be defined as avalue in Ohms per square using Z parameter. You canspecify it directly as a number or enter a file name, wherefrequency-dependent values of the surface impedance arestored in a table in a file. The file format corresponds tothe format of the standard impedance file and is describedin the Device Data section of the Reference manual.

On a technical note, if loss is used, the box covers aremodeled as a boundary with Schukin-Leontovich condi-

146 TPL File Format

tion. They could also be modeled as semi-infinite lossylayer with option -RL defined in the command stringbefore the TPL file name. Both conditions are almostexactly identical for typical metal covers.

If the type of the cover is defined as MAGNETIC, it willbe simulated as ideal magnetic wall (zero of tangentialmagnetic field). Use this type to model symmetrical prob-lems with reflection symmetry over the plane parallel tothe signal layers (broadside coupled stripline, bilateralfinline and so on.). No parameters are allowed with thistype.

Open boundary conditions can be simulated using OPENfor the top or bottom wall. It simply terminates the box inthe top or bottom plane with a semi-infinite rectangularwaveguide with the same cross-section as the box. Inother words, it matches all waveguide modes (propagatingand evanescent) on the grid. There will be loss of energyin the structure due to radiation if some modes are propa-gating, otherwise it just models a structure with the coverfar from the signal plane. To model a completely open orradiating structure on purpose, you need to put sidewallsof the box as far away as possible to provide a sufficientnumber of waveguide modes propagating along z axis torepresent a radiation pattern. There are no other parame-ters for covers of OPEN type. A structure with openedboundary conditions could also be simulated using char-acteristic impedance of TEM wave in the free space 377Ohms per square as the boundary condition for the coverof METAL type. It does not match all rectangularwaveguide eigenmodes and works more correctly if youneed to simulate a boxed structure with an open top cover(as opposed to having no box at all).

Example:

To describe a shielding box with an ideal metal top cover,the following line can be entered in the package sub-block:

TPL File Format 147

…TOP_W TYPE=METAL…

Absence of this line will have the same effect.

Example:

The following lines in the package sub-block describe topand bottom walls of the box as copper surfaces withroughness 0.002 mm.

…TOP_W TYPE=METAL RHO=1 ROUGH=0.002…BOTTOM_W TYPE=METAL RHO=1 ROUGH=0.002…

Example:

Open boundary conditions in the plane of the top wall (“nocover”) are defined by this line:

…TOP_W TYPE=OPEN…

A box without cover could be also modeled as a structurewith 377 Ohms per square resistive surface in the planeof top wall:

…TOP_W TYPE=METAL Z=377…

148 TPL File Format

MEDIA /LAYER

Defines the layers in the box

Format:MEDIA [H=]hP [[Er=]eP] [[Ur=]mP] [TAND=tandP/SIGMA=S]…LAYER [N=]1 [[TYPE=]CURRENT/VOLTAGE][MEDIA [H=]h2 [[Er=]e2] [[Ur=]m2] [TAND=tand2/SIGMA=S]]MEDIA [H=]h1 [[Er=]e1] [[Ur=]m1] [TAND=tand1/SIGMA=S]

A line beginning with MEDIA describes a homogeneousmedia layer (substrate, air, or other dielectric) inside ashielding box as shown in Figure 11-1. The LAYER lineindicates the position and type of the signal layer. Thissignal layer is infinitely thin. There must be at least twolayers of MEDIA type in a structure and can only be onesignal LAYER. The signal LAYER must be between ME-DIA layers. The total number of media layers is unre-stricted. The sequence of MEDIA and LAYER statementsdefines the structure of the box along z-axis. Thus, theorder of the MEDIA and LAYER descriptors is importantand must coincide with the order of the layers in the actualproblem as shown in Figure 11-1.

Note: The metal LAYER is always modeled as infinitelythin, even if TMET is specified in the SURFACE parame-ters. The SURFACE TMET parameter only affects thecalculation of loss.

The MEDIA line has one mandatory parameter H thatspecifies the thickness (height) of a media layer. Units forthickness are millimeters if they are not overridden in theDIM block. Relative dielectric permittivity and relativepermeability of a media layer are specified by parametersEr and Ur respectively. Default values of E and U are 1.TAND specifies the loss tangent of a media layer. The

TPL File Format 149

default value for TAND is zero (lossless media). An alter-native loss descriptor is SIGMA, which specifies the bulkconductivity of a media (in units Siemens/meter). Allvalues in MEDIA must be real (not complex). Strictlyspeaking, all these parameters are frequency dependent,but in practical problems they usually do not change muchin wide frequency ranges, so you can just pick the valuesfor your frequency range and use them as constants.Typical parameter values are given under the Tablessection of the Reference manual.

The LAYER line includes a mandatory layer number N(always set to one in this version) plus an optional TYPEparameter. The TYPE can be specified as CURRENT orVOLTAGE and defines initial type of variables on theright hand side of the system of linear equations. The caseCURRENT (also assumed by default) is for structureswith a “strip-like” metallization in the layer or for anystructures with overall area of metallization and otherconductive regions in the signal layer less than half of thetotal box cross-section in the plane xy. From mathematicalpoint of view, the problem with TYPE=CURRENT will beformulated in respect to currents inside regions wherethey may differ from zero. To solve dual problems inrespect to grid voltages or magnetic currents, the parame-ter TYPE has to be defined as VOLTAGE. Use this valueto solve “slot-like”problems or problems with large regionsof ideal metallization (slot lines, coplanar waveguides,fin-lines etc.). Select this type of layer if the overall areaof ideal (lossless) metallization is bigger then half of thebox cross-section in the plane xy. Properly chosen, theTYPE parameter reduces simulation time substantially.

Example:

A box with an Alumina substrate 1 mm thick, with strip-like signal layer on it and 5 mm of air between signal layer

150 TPL File Format

and top cover is defined by three lines in the packagesub-block:

…MEDIA H=5LAYER N=1 TYPE=CURRENTMEDIA H=1 E=10 TAND=1.0e-4…

The next fragment defines the same structure:

…MEDIA 5LAYER 1MEDIA 1 10 1 1.0e-4…

Example:

To define a box for coplanar waveguide problem withsuspended Alumina substrate, use the next four lines:

…MEDIA H=5LAYER N=1 TYPE=VOLTAGEMEDIA H=1 E=10 TAND=1.0e-4MEDIA H=3…

Here the signal layer of slot-like type is deposited on 1 mmAlumina layer and both are between 3 and 5 mm thicklayers of air.

TPL File Format 151

Geometry sectionDefines the pattern of traces on the conducting (signal)layer.

Format:CIRCUITPACKAGE…END_PACKAGE[LAYER [N=]1]PORT k1 [[F=]file] [[CD=]X/Y] [[LD=]X/Y] [[RPL]=shift]

& [X1=]x1 [Y1=]y1 [X2=]x2 [Y2=]y2…PAD kn [X1=]x1 [Y1=]y1 [X2=]x2 [Y2=]y2 [CD=]X/Y/Z

& [[TOL=]BOTTOM/TOP]…[LOSSLESS][LOSSLESSX][LOSSLESSY][LOSSLESSZ][SURFACEX [RHO=]rho [TMET=]tmet [ROUGH=]rough /

& Z=file/impedance][SURFACEY [RHO=]rho [TMET=]tmet [ROUGH=]rough /

& Z=file/impedance][SURFACEZ] [RHO=]rho [TMET=]tmet [ROUGH=]rough /

& Z=file/impedance]…[RECT [X1=]x1 [Y1=]y1 [X2=]x2 [Y2=]y2

& [CD=X/Y/XY/Z/XZ/YZ/XYZ] [[N=]n] [[M=]m]& [[TOL=]BOTTOM/TOP]][POLYGON X1=x1 Y1=y1 X2=x2 Y2=y2… X999=xn Y999=yn

& [CD=X/Y/XY/Z/XZ/YZ/XYZ ] [TOL=BOTTOM/TOP]]…

DEFnP k1... (ki,...,kj) … kn EMPOWER

Positions of internal and external inputs, shape, positionand properties of planar regions in the signal layer andposition and shape of via-holes are entered in the geome-

152 TPL File Format

try definition section. The following keywords can be usedto describe a problem:

PORT - An external input to be de-embedded.

PAD - internal port or place to connect a lumped element.

RECT - A rectangular region of metallization, resistivefilm or rectangular via.

POLYGON - A polygonal region of metallization, resistivefilm or polygonal via.

LOSSLESS - Descriptors RECT and POLYGON enteredafter this line are specifying a lossless metallization.

SURFACE - Physical properties of regions described byRECT and POLYGON entered after it.

All descriptors of the geometry section must be enteredinside circuit block after the package section. The sectionends with a DEFnP line that contains a list of all inputs(PORTs and PADs) and the name of the network (EM-POWER). Ports in the final calculated admittance orscattering matrix in the same order as inputs are listed inthe DEFnP line. The number n in DEFnP must be equalto the total number of inputs in the structure. PORTs thatcorrespond to coupled or multimode inputs have to benumbered sequentially and put in parentheses in theDEFnP line for proper de-embedding.

All entities in the geometry section will be mapped ontothe grid (see grid definition in the package section). Hav-ing a clear perception of how the original problem ismapped onto the grid is of great importance for competentproblem solution. The simplest way to fit a problem to thegrid is by defining all dimensions as multiple of grid cellsizes and positioning all edges exactly on the grid as shownin Figure 11-2(a). From theoretical point of view thispositioning is not exactly the best one (see Example “Mi-crostrip Line.WSP”), but following it gives you monotonic

TPL File Format 153

convergence of simulated data and in some sense predict-able error of modeling. Another one way to position edgesis to put them exactly between the grid nodes as shown inFigure 11-2 (b). This gives the same rate of convergencebut opposite sign of computational error.

Figure 11-2 (c) shows an intermediate positioning andgives some kind of combination of the two previous cases.The value of d has to be kept the same for whole problemand the best choice for it according to U. Schulz is 1/4 ofthe grid cell size. (See the Standard Stripline example inthe examples manual.) Any coherent way of positioningthe problem on the grid eventually gives systematic andthus controllable simulation errors. It must be mentionedthat all problems with d up to 1/2 cell size correspondexactly to the same grid problem. This also means thatthere is no exact one-one equivalence between an actualproblem and the grid one. It is easy to follow one patternof edge positioning if you start designing a device with theelectromagnetic simulator from scratch. Problems mayarise in analyzing already completed structures. Usuallyit is impossible to fit all regions to a grid following onepattern. In this case try to determine most significantparts of the structure and fit them. Other less significantand smaller parts must be at least commensurate with thegrid cell size and must be represented by at least one grid

region

grid

region

grid

region

grid

a) b) c)node d

Figure 11-2 Possible ways to position a region on the grid

154 TPL File Format

variable, otherwise they are ignored in the grid repre-sentation.

There is a map of terminals or grid variables in the listingfile to show the correlation between initial problem andgrid one (see description of the listing file). Figure 11-3 isalso provided for better understanding of how grid vari-ables are distributed on a grid cell. Two ways exist ofrepresenting grid variables. The first way is shown inFigure 11-3(a) illustrating the reference between the gridvariables and components of the electric field and surfaceconductivity currents and z-directed currents. This iscalled a “strict” representation of the grid problem. Thesecond way is shown in Figure 11-3(b) and represents thelayout of the terminals of an informational multiportcircuit (probe) on the grid. This is called the impedanceinterpretation of the grid problem. Figure 11-3 also pro-vides for matching the variables of these two repre-sentations. Values dx and dy denote cell sizes along the xand y axes.

x

y

Ex Jx

Ex Jx

Ey

Jy

Ey

Jy

Vx Ix

Vx Ix

Vy

Iy

Vy

Iydy

dx

a) b)

z

Ez

Jz

VzIz

Figure 11-3 Two methods of grid cell representation. a) Placingthe grid functions of electric field strength (E), surface current (J)and z-directed current (I) on the grid cell. b) Correlation of theterminals of a port to the grid cell. V and I designate respectivelythe terminal’s voltage and current.

TPL File Format 155

If a structure to be analyzed has edges that are not parallelto either the x- or y-axis, a stairstep approximation isimplemented to map the problem on the grid. Figure11-4(a) shows the intial position of a slanted edge on thegrid, and Figure 11-4(b) shows how it is approximated bya stairstep boundary in the =EMPOWER= simulation.Again, a small movement of the slant edge will not affectthe solution until it has moved enough that the gridrepresentation changes.

Summarizing the above, a few tips can be given:

• Chose grid cell sizes taking problem dimensions intoaccount. All dimensions of the problem and their changesmust be commensurate with the grid cell sizes.

• Try to position edges of all regions in the same way forentire problem.

• To see the problem mapping, look at the terminal map inthe listing.

See the Geometry and Theory Chapters for more detailsabout mapping on the grid.

a) b)

Figure 11-4 Stairstep approximation of a region with an edge thatdoes not lie along a grid coordinate axis.

156 TPL File Format

PORTPlaces an external, deembedded port in the circuit.

Format:PORT k [[F=]file] [[CD=]X/Y] [[LD=]X/Y] [[RPL]=shift]& [X1=]x1 [Y1=]y1 [X2=]x2 [Y2=]y2

All inputs of a structure to be de-embedded are describedin lines beginning with the word PORT. Inputs of this typeare simulated as surface current sources in one or morerectangular regions in the plane of the signal layer andare usually positioned in places where line approaches thebox sidewalls. They are always subjected to de-embed-ding. De-embedding is a numerical procedure thatmatches perfectly chosen propagating modes of the lineand finally extracts generalized admittance and scatter-ing matrices from the problem. This simply means that,after de-embedding, a problem in a closed box turns intoa problem with semi-infinite lines approaching it. Exter-nal inputs of an MIC or inputs to be de-embedded fordecomposition could be described as PORTs.

An input of a line with several propagating modes has tobe described as a set of PORTs grouped in parentheses inthe DEFnP line. The number of PORTs should be equalto the number of propagating eigenmodes to be extracted.The =EMPOWER= simulator contains a special procedurebased on the method of simultaneous diagonalization thattheoretically makes it possible to extract multimode pa-rameters for any line of arbitrary cross-section. You justneed to choose and properly place a number of surfacecurrent regions and describe them as PORTs. The simu-lator then extracts a transmission line with correspondingPORTs, calculates parameters of the line eigenmodes andde-embedding parameters and finally analyzes and ex-tracts a generalized descriptor matrix of the problem.Each input region described as a PORT will first be

TPL File Format 157

reduced to a pair of terminals as shown in Figure 11-5 byintegrating currents across the region and electric fieldalong the region. It will be represented by one row andcolumn in the admittance matrix. After this, if the inputconsists of two or more regions or PORTs, the simulatortransforms their currents and voltages (Vi and Ii in Figure11-5) into a mode space, eliminates reactance of excitationregions for each mode and finally normalizes and trans-forms the Y-matrix to scattering matrix.

The input number k and coordinates of the rectangularregion with surface current sources are mandatory pa-rameters in the line PORT. (X1, Y1) are x and y coordi-nates of the bottom left corner of the region, and (X2, Y2)are x and y coordinates of the top right corner of the regionas shown in Figure 11-5. As is the case with other coordi-nates, they should be aligned with the grid, and each inputregion must be represented by at least one pair of gridterminals after mapping (see the Theory and Geometrychapters). The following rules apply to ports:

• The length of the regions along the line should be uniformin the problem, preferably one grid cell.

a) b)

X1

Y1

X2

Y2

X1

Y1

X2

Y2

Vi,Ii

Vi,Ii

Figure 11-5 a) An input region oriented along the x-axis. b) Aninput region oriented along the y-axis.

158 TPL File Format

• The regions must touch one of the sidewalls. (Position ofthe input regions with respect to the sidewall is controlledby the TOLERANCE parameter defined in the packagesection.)

• The widths of the regions should be set equal to thecorresponding width of the line conductors in the strip-typelines or to dimensions between line conductors in slot-likelines.

• Overlaying ports is not permitted.

Optional parameters of the PORT line are:

• Name of the file to store line parameters and data forde-embedding (F).

• Direction of the surface currents in the region (CD).

• Direction of the line to be excited or de-embedded (LD).

• Shift of the input reference plane (RPL).

The filename parameter (F) specifies the file name forstoring calculated characteristics of the line and de-em-bedded data. The simulator creates these files in the lineanalysis mode and uses de-embedding data from them inthe discontinuity analysis mode. =EMPOWER= runs theline analysis mode if you described only a segment of a lineor before a discontinuity analysis in the automatic processof de-embedding data estimation.

In line analysis mode (EMLINE), the file name may beentered only in the first PORT descriptor. The namesentered for other PORTs will be ignored. In its absence,the default name is generated from the current input filename and the extension .RGF. The same extension will beadded to the file names specified without extensions.

When an entered structure is analyzed as a discontinuity,the de-embedding process either takes specified file nameswith default .RGF extensions or, if no name is specified,creates file names from the base name of the TPL file and

TPL File Format 159

extensions .Ln. Here n is input number (coupled inputscount together as one number and its value can be figuredout from the order of PORTs enumerated in the DEFnPline.

The parameters CD and LD normally should not be nec-essary, but, if needed, may be set to either X (directionalong x-axis) or Y (direction along y-axis). In excitingstrip-type lines, these parameters should correspond tothe actual direction of the line. To excite a slot likestructure the currents in the source regions are usuallydirected across the line. Default values for these parame-ters are assumed only if regions are positioned accordingto recommendations for the multiconductor microstripand slot-like lines. Positions of the current regions andpolarization of the surface currents should be chosen withrespect to facilitate optimally the excitation of line eigen-waves.

The RPL parameter is used to fix the shift of the input’sreference plane. In its absence the origin plane of phaseis assumed to coincide with the plane of the nearestsidewall. Positive values of the RPL parameter refer tooutward shift of the reference plane (adding length to theproblem), while negative values refer to inward shift withrespect to the box (subtracting length from the problem).All values are in millimeters if the units are not overrid-den in the DIM section.

Example:

Let us consider, for instance, exciting a coupled microstripline oriented along the x-axis. The number of conductorsin the line determines the respective number of dominanteigenwaves to be launched and, consequently, the mini-mum number of input regions. Hence, in the present casetwo PORTs are needed to excite one coupled (multimode)input. Following the above guidance, the regions aredisposed as shown in Figure 11-6(a). The arrows indicate

160 TPL File Format

directions of the surface currents (parameter CD) at inputregions. Assuming that there is one other coupled port(numbers 3 & 4), for a total 4 ports, the reference plane ascoincides with the box wall and the file with line charac-teristics is named LNCPL.RGF, we may describe thePORTS as follows:

…PORT 1 F=LNCPL CD=X LD=X RPL=0 X1=0 Y1=by1 X2=dx& Y2=ty1PORT 2 F=LNCPL CD=X LD=X RPL=0 X1=0 Y1=by2 X2=dx& Y2=ty2…DEF4P (1,2) (3,4) EMPOWER

Since CD and LD automatically sense the correct settingsfor most cases and RPL defaults to zero, this examplecould also be entered as:

…PORT 1 F=LNCPL X1=0 Y1=by1 X2=dx Y2=ty1PORT 2 X1=0 Y1=by2 X2=dx Y2=ty2…DEF4P (1,2) (3,4) EMPOWER

lx1 rx1 lx2 rx20

dy

yby1

ty1by2

ty2

0 dx

x

a) b)

y

x

strip conductors

Figure 11-6 Sample positions of input regions to launch a coupledline oriented along the axes x (a) and y (b). Symbols dx and dydenote respective grid cell sizes.

TPL File Format 161

To excite a coupled microstrip line oriented along they-axis, the regions are setup as shown in Figure 11-6(b).Using assignments of the previous example, the input maybe described as follows:

…PORT 1 F=LNCPL CD=Y LD=Y X1=lx1 Y1=0 X2=rx1 Y2=dyPORT 2 X1=lx2 Y1=0 X2=rx2 Y2=dy…DEF4P (1,2) (3,4) EMPOWER

Or, even simpler:…PORT 1 F=LNCPL X1=lx1 Y1=0 X2=rx1 Y2=dyPORT 2 X1=lx1 Y1=0 X2=rx1 Y2=dy…DEF4P (1,2) (3,4) EMPOWER

If the name of the file dropped, it would be either TPL-file-name.RGF if a only segment of coupled microstrip linesare described or TPL-file-name.L1 if the structure is somediscontinuity containing coupled line inputs.

Example:

Let us consider exciting a slot- or fin-line oriented alongthe x-axis. To de-embed a dominant eigenwave of the slottype in these lines it might be recommended to use oneinput region placed in the line gap as shown in Figure11-7(a). Surface currents in the input region should bedirected across the slot. Assuming the input number is 1,the reference plane coinciding with the nearest sidewall,the file with input characteristics is named LNSLT.RGF,and one other port is present in the circuit, we maytranscribe the respective description as follows:

…PORT 1 F=LNSLT CD=Y LD=X X1=x1 Y1=by1 X2=x2 Y2=ty1…DEF2P 1 2 EMPOWER

162 TPL File Format

The result will be the same with default CD and LD:…PORT 1 F=LNSLT X1=x1 Y1=by1 X2=x2 Y2=ty1…DEF2P 1 2 EMPOWER

To excite a slot- or fin-line oriented along the y-axis, theports are oriented as shown in Figure 11-7(b). Usingassignments of the previous example, the input may bedescribed as follows:

…PORT 1 F=LNSLT CD=X LD=Y X1=lx1 Y1=y1 X2=rx1 Y2=y2…DEF2P 1 2 EMPOWER

In contrast to multiconductor microstrip lines, relativeposition of the slot-like type input regions and line conduc-tors and sidewalls could not be uniquely defined for allproblems of this kind. It has to be mentioned, that theaccuracy of calculated line parameters and de-embeddingdata depends a lot on it. To avoid the short circuiting effectof the sidewall that could spoil all data, either the inputregions should be shifted from the sidewall or a gap shouldbe left between line conductors and the sidewall.

lx1 rx1y1

y2

y

by1

ty1

x1 x2

x

a) b)

y

x

slots

Figure 11-7 Sample orientation of inputs to excite a dominantmode of a slot- or fin-line oriented along the x-axis (a) and y-axis(b). Arrows depict current polarization.

TPL File Format 163

PADDescribes a non-deembedded port or internal port.

Format:PAD k [X1=]x1 [Y1=]y1 [X2=]x2 [Y2=]y2 [CD=]X/Y/Z[[TOL=]BOTTOM/TOP]

Parameters:k = Port NumberX1,Y1 = coordinates of the bottom left corner of the PADX2,Y2 = coordinates of the upper right corner of the PADCD = Current directionTOL = Destination layer for Z-directed port (optional)

Places to connect lumped elements and inputs that do notneed automatic de-embedding are described as internalports in a line beginning with PAD. An Internal input isa rectangular region with currents flowing along one ofthe coordinate axes. See Figures 11-5 and 11-8. =EM-POWER= integrates current across the region and electricfield along the region, creating a pair of terminals thatcould be used in GENESYS to connect a lumped elementor a source of energy. The transverse components of thecurrent and the electric field are assumed to be zero insidethe region. The internal port makes it possible to estimatethe influence of the lumped element connection region oncharacteristics of a device at electromagnetic level.

The size and position of the region have to be aligned withthe grid. If the region of the internal port is smaller thanthe grid cell or positioned improperly, it will be displacedto the nearest grid terminals. Overlapping of PADSand/or PORTS is not permitted.

The current direction CD in the PAD region is a manda-tory parameter. In essence, this parameter determinesalong which axis a lumped element or input is connectedand may assume the values X, Y or Z for directions alongcorresponding coordinate axes.

164 TPL File Format

The last (optional) parameter, TOL, is the definition of theplane to connect z-directed inputs to. It may assumevalues BOTTOM or TOP, meaning connection to the layerbelow or the layer above, respectively (or to the bottom orto the top covers of the box, respectively, if there is no othermetal layer below or above). By default, a z-directedinternal input will be connected to BOTTOM. Note thatz-directed ports can be connected across multiple medialayers, but not across multiple metal layers.

Example:

A one-port pad number 3 connected along the x-axis asshown in Figure 11-9(a) may be described as follows:

…PAD 3 X1=lx Y1=by X2=rx Y2=ty CD=X…DEFnP … 3 … EMPOWER

X1 X2

Y2

Y1

y

x

Vl,Il

Figure 11-8 An internal port connected along the z-axis. Note thatthe port may go to another level instead of ground.

TPL File Format 165

CAUTION: If X- or Y- directed ports are created, then theyrepresent series interconnected ports and should beused with great caution in GENESYS. See Chapter 6,Internal Ports, for details.

Example:

An one-port pad number 4 connected along the y-axis asshown in Figure 11-9(b) may be described as follows:

…PAD 4 X1=lx Y1=by X2=rx Y2=ty CD=Y…DEFnP … 4 … EMPOWER

Example:

To connect a two-port lumped element it is possible to usetwo separate internal ports directed along z-axis andconnect the two-port between them as shown in Figure11-10. Numbering the pads as 5 and 6, we can enter theproblem as follows:

…PAD 5 X1=xl1 Y1=yb1 X2=xr1 Y2=yt1 CD=Z TOL=BOTTOM

lx rx

ty

by

y

x

Lumpedelement

lx rx

y

x

ty

by

Lumpedelement

a) b)

Figure 11-9 a) A region with a lumped element connected alongthe x-axis. b) A region with an lumped element connected alongthe y-axis.

166 TPL File Format

PAD 6 X1=xl2 Y1=yb2 X2=xr2 Y2=yt2 CD=Z TOL=BOTTOM…DEFnP … 5 6 … EMPOWER

After the electromagnetic analysis of a problem containingthese two internal inputs, you will get a descriptor matrixwith two additional ports corresponding to the internalinputs. Their position in the matrix depends on the posi-tions of 5 and 6 in the DEFnP line. If their numberscorrespond to their positions in the DEFnP line, they willbe port number 5 and 6 in the final matrix. Actual pa-rameters of the lumped element could be placed in theseports in GENESYS to obtain the results of the entire=EMPOWER= circuit, including the lumped elements.

When solving an electromagnetic problem with internalports, we obtain a descriptor matrix that accounts forcurrent redistribution caused by connection of arbitraryelements, allowing us to connect and tune values oflumped elements in a circuit simulator without any addi-tional electromagnetic analysis.

xl1 xr1

yt1

yb1

y

xl2 xr2

yt2

yb2x

Lumpedelement

Region 1

Region 2

Figure 11-10 A two-port lumped element connected between twoz-directed internal inputs.

TPL File Format 167

LOSSLESS / SURFACEDescribe loss parameters of the signal layer

Format:LOSSLESSLOSSLESSXLOSSLESSYLOSSLESSZSURFACE [RHO=]rho [TMET=]tmet& [ROUGH=]rough / Z=file/impedanceSURFACEX [RHO=]rho [TMET=]tmet& [ROUGH=]rough / Z=file/impedanceSURFACEY [RHO=]rho [TMET=]tmet& [ROUGH=]rough / Z=file/impedanceSURFACEZ [RHO=]rho [TMET=]tmet& [ROUGH=]rough / Z=file/impedance

Physical parameters of planar regions in the signal layerand parameters of viaholes are defined using descriptorsLOSSLESS and SURFACE. The main concept is thateach keyword just modifies values specified previouslyand affects all conductors described until the nextLOSSLESS/SURFACE keyword is used. Any metal en-tered before LOSSLESS/SURFACE uses LOSSLESS bydefault.

LOSSLESS corresponds to ideal metallization. To de-scribe a surface with different properties, the SURFACEdescriptor has to be used. The first “optional parameter”(actually part of the keyword) in the SURFACE line is thedirection of current for which the properties are specified.X, Y and Z values are for x-, y- and z-directed currentsrespectively. If this parameter is omitted, the parametersdefined in the descriptor are applied to the currents in alldirections. Using these directions, anisotropic conductorscan be created

There are two ways to define properties of a region. Thefirst one is suitable only for metal regions with known

168 TPL File Format

physical parameters. They can be described using metal-lization resistivity relative to copper (RHO), thickness ofmetallization layer (TMET), and surface roughness(ROUGH) parameters. Length units for TMET andROUGH are millimeters, unless they were redefined inthe DIM block. If RHO=0, the statement is equivalent toLOSSLESS.

CAUTION:The thickness, specified in the SURFACE lineis only used to calculate the surface impedance. Allsimulations always use infinitely thin metal.

An alternative way to describe the properties of a regionof any kind is to make a file with a table of surfaceimpedances (ohms per square) and enter the filename inparameter Z in the SURFACE line. The format of the fileis the same as is used to describe a Z-matrix of a two-portin GENESYS. The first line in the file must is usually inthe form:

# units Z RI R 1

where units is frequency units (default is MHZ). The restof the file contains three columns - frequency in specifiedunits and real and imaginary parts of the surface imped-ance in Ohms per square. Here is an example of a file withsurface impedances of some surface specified for n fre-quencies:

# GHz Z RI R 1frq1 Zr1 Zi1frq2 Zr2 Zi2…frqn Zrn Zin

Where frq is the frequency value and Zr and Zi are thereal and imaginary parts of the impedance (ohms per

TPL File Format 169

square = Zr + jZk). Frequencies must be specified inascending order. If some frequencies are omitted, theimpedance will be linearly interpolated or extrapolated.

If a surface impedance is real and it is not frequencydependent or does not change a lot in some frequency bandthen the parameter Z can be described as a real numberspecifying value in Ohms per square. As an example, thistype of description is suitable for resistive films.

Examples:See the RECT and POLYGON sections following.

170 TPL File Format

RECTDefines a rectangular conductor on the signal layer.

Format:RECT [X1=]x1 [Y1=]y1 [X2=]x2 [Y2=]y2 [[N=]n] [[M=]m]& [CD=X/Y/XY/Z/XZ/YZ/XYZ] [[TOL=]BOTTOM/TOP]

The RECT descriptor is used to specify rectangular met-allized regions and regions with surface impedance as wellas via-holes. The description must begin with keywordRECT and contains at least coordinates of the rectangle.(X1,Y1) are the coordinates of the bottom left corner of therectangle, and (X2,Y2) are the coordinates of the top rightcorner as shown in Figure 11-11.

The coordinates should be consistent with the grid (see theintroduction to the circuit block section). Rectangularregions with a dimension smaller than the grid cell sizemight not be reflected by the grid, causing a differencebetween the actual circuit and the one modeled by =EM-POWER=. You should view the listing file to see if this isa problem.

The RECT line can optionally include the current direc-tion parameter CD and the destination parameter TOLfor currents flowing along the z-axis.The N and M parame-ters are reserved for future use and should be set to one ifused.

The current direction CD takes on values X, Y and Z forcurrents flowing along the coordinate axes. The CD pa-rameter can be also defined as the mixture of the maindirections XY, XZ, YZ and XYZ, that corresponds to asuperposition of currents flowing along correspondingaxes. This parameter may be used to reduce model com-plexity and computation time for problems with priorknowledge on zero components of the currents.The defaultvalue of the CD is XY (x- and y-directed currents in a

TPL File Format 171

planar region). Regions with z-directed currents are basi-cally for the description of via-holes. There is a smalldifference in mapping the z-directed currents in regionswith currents directed only along the z-axis, and regionswith superposition of other currents; see the detaileddescription in the POLYGON section.

The last parameter in the RECT line is the destinationparameter TOL. It works only if the current directionparameter contains Z, otherwise it will be ignored. TheTOL parameter can be defined either as BOTTOM for thez-directed currents flowing down one layer toward thebottom cover of the box (default value) or as TOP forz-directed currents flowing up one layer toward the top.

To specify the physical parameters of a rectangular region,the SURFACE and LOSSLESS descriptors are used (seethe previous section). If there are no SURFACE linespreceding the RECT descriptor, the region is considered tobe lossless (perfect metal). Otherwise most recent occur-rences of the SURFACE and LOSSLESS lines are used todetermine properties of a surface.

CAUTION: For anisotropic materials, losses can be eas-ily confused by the user unless surface properties of alldirections are set each time they are redefined.

X1 X2

Y2

Y1

y

x

Figure 11-11 A rectangular region.

172 TPL File Format

Example:

A structure contains two segments of a strip line made ofcopper with resistive film connected between them asshown in Figure 11-12. The strip line is w mm wide and0.01 mm thick and roughness of the copper metallizationis 0.001 mm. The sizes of the box along x- and y-axis area and b respectively. The surface resistance of the resistivefilm is 50 Ohms per square and its dimensions are w byw. The film is centered in the box. The next fragment maybe used to enter the problem (the description of inputs isomitted):

…SURFACE RHO=1 TMET=0.01 ROUGH=0.001! The first segment of the strip:RECT X1=0 Y1=(b-w)/2 X2=(a-w)/2 Y2=(b+w)/2! The second segment of the strip:RECT X1=(a+w)/2 Y1=(b-w)/2 X2=a Y2=(b+w)/2…SURFACE Z=50! The resistive film region:RECT X1=(a-w)/2 Y1=(b-w)/2 X2=(a+w)/2 Y2=(b+w)/2…

a

w

w

b

Resistivefilm

Metal regions

Figure 11-12 The geometry of a signal layer containing two stripline segments and a resistive film region.

TPL File Format 173

Example:

A rectangular copper viahole with size D by D and withcoordinates of the center in xy plane (Xv,Yv) and connectedbetween the signal layer and the bottom cover of the boxcan be described as follows:

…SURFACE RHO=1 TMET=0.1RECT X1=Xv-D/2 Y1=Yv-D/2 X2=Xv+D/2 Y2=Yv+D/2 CD=Z…

It is assumed that there will be a planar conductive regionin the signal layer over the via-hole to connect it with otherparts of the circuit.

174 TPL File Format

POLYGONDefines an arbitrarily shaped conductor on the signallayer.

Syntax:POLYGON X1=x1 Y1=y1 X2=x2 Y2=y2… [X999=xn Y999=yn][CD=X/Y/XY/Z/XZ/YZ/XYZ ] [TOL=BOTTOM/TOP]

To describe arbitrarily shaped metallized regions, regionswith surface impedance, and via-holes, a polygon can beused. It begins with the POLYGON keyword and containsa list of coordinates of the polygon vertexes, polarizationof the conductive currents inside region and the destina-tion layer for currents flowing along the z-axis. The verti-ces must be numbered and entered sequentially as shownin Figure 11-13. The actual number of vertexes on thefigure is N. (Xk,Yk) corresponds to the coordinates of thek-th vertex in the xy plane. The minimum number ofvertices in a polygon is 3; the maximum is 999.

The direction or polarization of currents inside the region(CD) can be X, Y, Z, XY, XZ, YZ, or XYZ, with the samemeanings as for the RECT descriptor. The default valueis XY. The parameters X, Y and Z correspond to thedirections along the respective coordinate axes.

The last parameter in the POLYGON line is the TOLparameter that defines the destination plane for z-di-rected currents. It can assume the values TOP and BOT-TOM to connect the polygon to the next layer above orbelow (or to the top/bottom cover if no metal layer ispresent). If a region does not have currents in z-direction,the destination parameter is ignored.

A few words about positioning and mapping a polygonalregion on the grid. The vertexes must be positioned prop-erly in accordance with rules described in the introductionto this chapter. Sides of the polygon that are not parallel

TPL File Format 175

to a coordinate axis are approximated by stairsteps. Thereis a simple rule to predict the result of the polygonal regionmapping: If the center of a grid cell is inside the regionthen all four currents surrounding the cell will be includedin the grid model to represent the actual region (seeFigures 11-2 to 11-4). This rule applies for all regionscontaining x and y directed currents. If a region containsonly z-directed currents, the rule is slightly different. Az-directed current is included if the corresponding gridnode (as opposed to the center of a grid cell) is inside theregion.

A curvilinear region must be approximated first by apolygonal region with sufficient numbers of the vertexesfor the desired accuracy. A simple criteria to determinethe number of vertexes can be given. If the distancebetween the actual region border and the nearest polygonside approximating it is less then 1/4 cell size, the numberof sides is usually sufficient. For less critical applications,fewer sides can be used.

y

x

x1

y1

x2

y2

1

2

xN

yN N

Figure 11-13 A polygonal region.

176 TPL File Format

Example:

To describe a mitered bend in a strip line we can introducethe pentangular region as shown in Figure 11-14. Thewidth of the stripline is w, and the chamfer is positionedaccording to the figure. Numbering vertexes counter-clockwise starting from the internal corner we can enterit as the next polygon with both x- and y-directed currents:

POLYGON X1=(a-w)/2 Y1=(b-w)/2 X2=(a-w)/2& Y2=(b-w)/2-(d-w) X3=(a+w)/2 Y3=(b-w)/2-(d-w)& X4=(a-w)/2-(d-w) Y4=(b+w)/2 X5=(a-w)/2-(d-w)& Y5=(b-w)/2

a

w

w

b

Polygonalregion

d

1

2 3

4

5

Figure 11-14 A pentangular region used in the description of achamfered bend in stripline.

TPL File Format 177

SIMPLE LINE ANALYSIS EXAMPLE

TPL files: EAGLE\EXAMPLES\TPLFILE\LINE1.TPL,EAGLE\EXAMPLES\TPLFILE\LINE2.TPL.

This example demonstrates how to compose a TPL file toanalyze a line segment. It also shows two ways to enterparameters of the problem: 1) direct description of alldimensions, and 2) indirect definition of key parametersin the EQUATE block. All files prepared for this exampleare in the sub-directory EAGLE\EXAMPLES\TPLFILE.Files created during the =EMPOWER= run will be placedin this directory.

Let us consider the electromagnetic analysis of the micros-trip line shown in Figure 11-15 in the frequency rangefrom 5 to 20 GHz. To analyze it we need to describe asegment of the line enclosed in a rectangular shieldingbox. It is also necessary to setup the external ports (exci-tation regions) and to set the appropriate operating mode.Additionally, appropriate grid cell sizes in the plane of themetallization layer must be chosen.

5

5

1

1

ε = 9.6

ε = 1.0

Figure 11-15 The cross-section and dimensions of the microstripline.

178 TPL File Format

Let us assume the line length to be equal to the packagewidth. Figure 11-16 depicts a segment of the line. Alldimensions in the figures are given in millimeters. We willchoose a grid cell size 0.2 mm×0.2 mm. It is recommendedto set the width of input regions equal to the line widthand their length equal to one grid cell along the line. Themicrostrip conductor is rectangular region of perfect met-allization. The substrate is Alumina with Er=9.6 andH=1mm. There is 5 mm of are between the line and thetop cover. Now that the problem has been completelyconceptualized, it can be transcribed into a correspondingTPL file for =EMPOWER= as given in Table 11-4. Thisfile is LINE1.TPL. You can run it using the batch fileLINE1.BAT.

Now we will consider a way to allow the input data to bemore easily modified using the =EMPOWER= EQUATEblock. To facilitate resetting the parameters, they aredeclared as variables at the beginning of the input file as

50.2 4.800

2

3

5

11 2

x

y

Figure 11-16 A segment of microstrip line in the signal layer plane.Shaded regions represent the external input regions.

TPL File Format 179

it is shown in Table 11-5. In the PORT and RECT lines,instead of numbers being given directly, the parametersare given using equations based on these variables.

Running =EMPOWER= with this input data file (batchfile LINE2.BAT can be used) results in recording the lineand de-embedding data in the binary file LINE2.RGF aswell as printing them to the listing file LINE2.LST. Table11-6. presents the most important fragments of the listingfile. A detailed description of a listing file contents is givenin Appendix B, so here we will just point out a few sectionsof the listing.

! Microstrip line analysis example.! Direct description of the input data.

! Eagleware Corporation (C) 1998.DIMFREQ GHZLNG MM

CIRCUITEMLINE !Line analysis mode.

PACKAGEEMFRQS 5 20 6MAXFRQ 20DELTA X=0.2 Y=0.2 !Grid cell sizes.SIZE X=5 Y=5 !Package dimensions in X-Y plane.MEDIA H=5 !Free space layer.LAYER N=1 !Metallization layer.MEDIA H=1 E=9.6 !Alumina layer.

END_PACKAGE

! The first input description.PORT 1 X1=0 Y1=2 X2=0.2 Y2=3

! The second input description.PORT 2 X1=4.8 Y1=2 X2=5 Y2=3

! Line segment is described as the rectangle.RECT X1=0.2 Y1=2 X2=4.8 Y2=3

DEF2P 1 2 EMPOWER

Table 11-4 The input for analyzing the line shown in Figures11-15 and 11-16 (EAGLE\EXAMPLES\TPLFILE\LINE1.TPL).

180 TPL File Format

The section that begins with the keyword QCHK containsproperties of the problem at the specified maximum criti-cal frequency (MAXFRQ). It shows that the grid cell sizeis small enough for accurate analysis of the structure atthis frequency (minimal media wavelength to mesh sizeratio is 24). At the same time the problem will not bethinned out because the cell size is relatively large (thin-

! Microstrip line analysis example.! Parametric description of the input data.

! Eagleware Corporation (C) 1998.

DIMFREQ GHZLNG MM

EQUATEw=1 !Strip width.a=5*w !Box size along X.b=5*w !Box size along X.dx=w/5 !Cell size along X.dy=w/5 !Cell size along Y.

CIRCUITEMLINE !Line analysis mode.PACKAGEEMFRQS 5 20 6MAXFRQ 20DELTA X=dx Y=dySIZE X=a Y=bMEDIA H=5LAYER N=1MEDIA H=1 E=9.6

END_PACKAGE

! The first input description.PORT 1 X1=0 Y1=(b-w)/2 X2=dx Y2=(b+w)/2

! The second input description.PORT 2 X1=(a-dx) Y1=(b-w)/2 X2=a Y2=(b+w)/2

! Line segment is described as the rectangle.RECT X1=dx Y1=(b-w)/2 X2=(a-dx) Y2=(b+w)/2

DEF2P 1 2 EMPOWER

Table 11-5 The input data file for microstrip line analysis usingthe EQUATE block. The data are stored in the fileEAGLE\EXAMPLES\TPLFILE\LINE2.TPL.

TPL File Format 181

C:\EAGLE\EXAMPLES\TPLFILE\LINE1.TPL Page 1Time 12:57:25

Wed 29 Jul 98=EMPOWER= EM MIC Simulator. Copyright (c) 1998 Eagleware Corporation.

QCHK: Parameters of the solution quality.Min media wavelength to mesh size ratios: 24(X), 24(Y).Thinning out thresholds: 1(X), 1(Y).Max box size to media wavelength ratios: 1.03351(X), 1.03351(Y)!Lines difference: 1(X), 1(Y).

!!!!! Line analysis mode results !!!!!SDTC: Two plane symmetry (YZ and XZ) ***

*** MAP of the terminals used in the LAYER 1 ***

...=== THE TERMINAL MAP HAS BEEN OMITTED ====== SEE FIGURE 2-1 FOR AN EXAMPLE ===...MEMORY: GGF Matrix: 22K (270 variables).MEMORY: Estimated Total: 44K (EGV+GS=20K, GGF=22K, CDM=1K, CPM=0K)

**** Line parameters at frequency 5.000e+009 Hz. ****+—-+——+———————-+————————+————————+——————+|Nm |Type | Zo( Ohm ) | Gw(rad/m) | Gw/Go | Comp.phase |+—-+——+———————-+————————+————————+——————+| 1 | re. | 45.275677 | 263.477971 | 2.51428868 | 0.00013372 |+—-+——+———————-+————————+————————+——————+Compensation admittance (imaginary) 0.0029915642 1/Ohm

**** Line parameters at frequency 8.000e+009 Hz. ****...

**** Line parameters at frequency 1.100e+010 Hz. ****...

**** Line parameters at frequency 1.400e+010 Hz. ****...

**** Line parameters at frequency 1.700e+010 Hz. ****...

**** Line parameters at frequency 2.000e+010 Hz. ****...

!!!!! End of line analysis data !!!!!

Table 11-6 Fragments of the listing file generated from an=EMPOWER= run with the input data from the Tables 11-2 or 11-3.

182 TPL File Format

ning out threshold is 1). The box size looks too large foranalysis at 20 GHz (maximal box size to media wavelengthratio is about 1). This happened because we specified toowide of a frequency range and thus can not satisfy alldemands.

The next large section in the listing file is the line analysismode results. The program determined that the structureis a line segment and that it has two-mirror symmetry (seeSTDC in the listing). The map of the terminals used inlayer 1 gives a representation of the grid problem that issolved to model the actual one. The signs + correspond tothe grid nodes, the signs - and | correspond to the boxsidewalls and to conductivity currents on the microstripsurface, and numbers 1 and 2 correspond to the gridterminals in the input regions. The map is a very impor-tant control tool to check for user error.

Note: Whenever you enter a TPL file manually, youshould check the terminal map very carefully for errors.

The lines beginning with the keyword MEMORY informabout memory usage by particular parts of the algorithmand total estimated memory to solve the problem. Thisproblem was solved with 270 grid variables (the order ofthe matrix of the system of linear equations) and occupied44 Kbytes of RAM for data in the process. The rest of thelisting file gives the results of the analysis. Since we haveanalyzed a line, we have tables with the propagationcoefficients,wave impedances, and de-embedding parame-ters. A reference on the effective accuracy and the waveimpedance evaluating method may be found in the Appen-dix A (Theory). The columns Zo gives the wave imped-ances multiplied by the grid transformation coefficients,the Gw columns contain the absolute propagation con-stants in radian per meter units while the normalized(with respect to the free space) propagation constants are

TPL File Format 183

given in columns Gw/Go. The compensation phase and thecompensation admittance are auxiliary parameters forde-embedding purposes.

184 TPL File Format

SIMPLE DISCONTINUITY ANALYSIS EXAMPLE

TPL files: EAGLE\EXAMPLES\TPLFILE\BEND1.TPLEAGLE\EXAMPLES\TPLFILE\BEND2.TPLEAGLE\EXAMPLES\TPLFILE\BNDCHM.T PL

This example illustrates simulation of a discontinuity ina planar line starting from a TPL file. Basically, almostall problems can be classified as the discontinuities be-tween input and output lines, so this example can serveas the basis for any other file you want to create. All filesprepared for this example are in the sub-directoryEAGLE\EXAMPLES\TPLFILE.

Let us consider a bend in the microstrip line from the lineanalysis example shown in Figure 11-15 (see the previousexample). We assume the same frequency range as for theline analysis, from 5 to 20 GHz in 2 GHz steps. We willenclose the bend in a rectangular shielding box with thesame dimensions as in the line analysis. The grid cell sizes

50.200

2

3

5

11

2 x

y

2 3

0.2

1

RP

RP

1

2

Figure 11-17 A bend in the microstrip line (other dimensions areshown on Figure 11-15). Shaded areas mark the external ports.

TPL File Format 185

are set 0.2 mm along both axes. The size and locations ofthe input regions are also replicate the case of the lineanalysis, with the output on a different wall. The formu-lated problem is depicted in Figure 11-17. All dimensionsare given in millimeters.

The RP1 and RP2 lines shown in the figure indicate thereference planes for transforming the phases of the scat-tering matrix elements. Having specified all parameters,the corresponding input data can be encoded into a TPLfile. Again, as in the line analysis example we can specifyall data directly. The file with this kind of setup is inBEND1.TPL (not shown here). Table 11-7 shows amoreconvenient way to code the problem using the EQUATEblock. You can find this data in the file BEND2.TPL andcan run =EMPOWER= to solve this problem using thebatch file BEND2.BAT. =EMPOWER= will generate thelisting file BEND2.LST (some fragments of it are shownin Table 11-8), write data files BEND2.SS (50-Ohms nor-malized scattering matrix) and BEND2.R1 andBEND2.R2 (normalization impedances of the ports). Inaddition it will write a binary file BEND2.L1 with pa-rameters of the approaching to the discontinuity line andthe de-embedding data and a binary file BEND2.Y withunnormalized admittance matrices of the problem.

Let’s look at the generated listing file closely. As in thecase of the line analysis it shows some general charac-teristics of the problem at the maximum critical frequencyin the section QCHK. The symmetry detector classifiedthe problem as a nonsymmetrical. (Mirror symmetryacross a diagonal plane does not count.) The map of theterminals in the signal layer shows that problem ismapped on the grid properly, and the estimated memoryto solve the problem is 297 Kbytes (270 grid variables).Before analyzing the discontinuity itself, =EMPOWER=extracts the line approaching the structure and analyzesit. The results are skipped in the Table 11-8, but you

186 TPL File Format

should notice when you look at the actual listing file thatthey are a little bit different from the results obtained inthe line analysis mode (Table 11-6). This is because of thelength of the line segment to be analyzed before disconti-nuity was detected automatically and is not equal to thebox size (as was the case in the first example). It couldcause some differences especially at the highest or lowestfrequencies. You can prohibit this automatic length detec-

! Microstrip bend analysis example.! Parametric description of the input data.

! Eagleware Corporation (C) 1998.

DIMFREQ GHZLNG MM

EQUATEw=1 !Strip width.a=5*w !Box size along X.b=5*w !Box size along Y.dx=w/5 !Cell size along X.dy=w/5 !Cell size along Y.

CIRCUITPACKAGEEMFRQS 5 20 6MAXFRQ 20DELTA X=dx Y=dySIZE X=a Y=bMEDIA H=5LAYER N=1MEDIA H=1 Er=9.6

END_PACKAGE

! The first input description.PORT 1 RPL=-(a-w)/2 X1=0 Y1=(b-w)/2 X2=dx Y2=(b+w)/2

! The second input description.PORT 2 RPL=-(b-w)/2 X1=(a-w)/2 Y1=0 X2=(a+w)/2 Y2=dy

! Line segments are described as the rectangular regions.RECT X1=dx Y1=(b-w)/2 X2=(a+w)/2 Y2=(b+w)/2RECT X1=(a-w)/2 Y1=dy X2=(a+w)/2 Y2=(b+w)/2

DEF2P 1 2 EMPOWER

Table 11-7 The file to analyze the bend shown in Figures 11-15and 11-17 (EAGLE\EXAMPLES\TPLFILE\BEND2.TPL).

TPL File Format 187

C:\EAGLE\EXAMPLES\TPLFILE\BEND2.TPL Page 1Time 13:27:00

Wed 29 Jul 98=EMPOWER= EM MIC Simulator. Copyright (c) 1998 Eagleware Corporation.

QCHK: Parameters of the solution quality.Min media wavelength to mesh size ratios: 24(X), 24(Y).Thinning out thresholds: 1(X), 1(Y).Max box size to media wavelength ratios: 1.03351(X), 1.03351(Y)!

SDTC: Unsymmetrical structure ****** MAP of the terminals used in the LAYER 1 ***

...!!!!! Line analysis mode results !!!!!…

!!!!! End of line analysis data !!!!!

MEMORY: GGF Matrix: 285K (270 variables).MEMORY: Estimated Total: 297K (EGV+GS=10K, GGF=285K, CDM=1K, CPM=0K)

S-matrix table in form “magnitude-angle” at frequency: 5.000e+009 Hz.+——————-———+———————————————+—————————+| | | S-matrix elements | L or VSWR |+ input | mode +———————+———————-+—————————+| number | number | ‘S’-magnitude | ’S’-phase(rad.) | L in dB |+——+——+——+——+———————+———————-+—————————+| 1 | 1 | 1 | 1 | 0.108308 | -1.71733 | VSWR=1.243 |+——+——+——+——+———————+———————-+—————————+| 1 | 2 | 1 | 1 | 0.994117 | -0.146531 | L = 0.05125 |+——+——+——+——+———————+———————-+—————————+| 2 | 1 | 1 | 1 | 0.994117 | -0.146531 | L = 0.05125 |+——+——+——+——+———————+———————-+—————————+| 2 | 2 | 1 | 1 | 0.108308 | -1.71733 | VSWR=1.243 |+——————————————————————————————————-+

S-matrix table in form “magnitude-angle” at frequency: 8.000e+009 Hz....

S-matrix table in form “magnitude-angle” at frequency: 1.100e+010 Hz....

S-matrix table in form “magnitude-angle” at frequency: 1.400e+010 Hz....

S-matrix table in form “magnitude-angle” at frequency: 1.700e+010 Hz....

S-matrix table in form “magnitude-angle” at frequency: 2.000e+010 Hz....

Table 11-8 Fragments of the listing file generated by an=EMPOWER= run with the input data from the Table 11-7.

188 TPL File Format

tion by using option -O in the command string. In this casethe lengths of all lines to be de-embedded will be set to theactual lengths of the shielding box. The results of de-em-bedding will be slightly more accurate, but it may takemuch more time to prepare the de-embedding data.

Finally in Table 11-8, we have simulation results: thegeneralized scattering matrices of the bend with referenceplanes shifted toward the corner are printed out in formof tables. The first four columns of the tables show thenumber of the mode and input for a particular scatteringmatrix element. The next two columns contain the ele-ment value magnitude and angle in radians. And the lastcolumn gives the element value as the attenuation fortransmission coefficients, or as the wave standing ratio forreflection coefficients.

To show how some minor modification of the problemaffects on the input data, we prepared file BNDCHM.TPLwith the description of the same bend but with a chamferto decrease reflection from the bend (see Table 12.7). Inaddition to the two line segments entered as rectangularmetallized regions, the chamfer region is described aspentagonal polygon as shown in the illustrated exampleto the POLYGON section earlier in this chapter. By chang-ing the chamfer position and recalculating you can evalu-ate the effect of the chamfering.

TPL File Format 189

! Microstrip chamfered bend analysis example.! Parametric description of the input data.

! Eagleware Corporation (C) 1998.

DIMFREQ GHZLNG MM

EQUATE

w=1 !Strip width.a=5*w !Box size along X.b=5*w !Box size along Y.dx=w/8 !Cell size along X.dy=w/8 !Cell size along Y.d=w !Chamfer position parameter.

CIRCUIT

PACKAGEEMFRQS 5 20 6MAXFRQ 20DELTA X=dx Y=dySIZE X=a Y=bMEDIA H=5LAYER N=1MEDIA H=1 Er=9.6

END_PACKAGE

! The first input description.PORT 1 RPL=-((a-w)/2-(d-w)) X1=0 Y1=(b-w)/2 X2=dx Y2=(b+w)/2

! The second input description.PORT 2 RPL=-((b-w)/2-(d-w)) X1=(a-w)/2 Y1=0 X2=(a+w)/2 Y2=dy

! Line segments are described as rectangular regions.RECT X1=dx Y1=(b-w)/2 X2=(a-w)/2-(d-w) Y2=(b+w)/2RECT X1=(a-w)/2 Y1=dy X2=(a+w)/2 Y2=(b-w)/2.0-(d-w)

! Chamfered region is described as pentagonal piece of metal.POLYGON X1=(a-w)/2 Y1=(b-w)/2 X2=(a-w)/2 Y2=(b-w)/2-(d-w)& X3=(a+w)/2 Y3=(b-w)/2-(d-w) X4=(a-w)/2-(d-w) Y4=(b+w)/2& X5=(a-w)/2-(d-w) Y5=(b-w)/2

DEF2P 1 2 EMPOWER

Table 11-9 The input data to analyze the chamfered microstrip bend(EAGLE\EXAMPLES\TPLFILE\BNDCHM.TPL).

190 TPL File Format

Appendix A

Theory

T his chapter gives a technical description of the basic=EMPOWER= algorithms. Unlike most similar toolson the market, =EMPOWER= is based on the method

of lines (MoL) and comprises a set of numerical techniquesdesigned to speed up calculations while increasing accu-racy of computations. Incorporation of geometrical sym-metries (including rotational), reduction of problemcomplexity using thinning out and linear re-expansionprocedures, and multimode deembedding by the simulta-neous diagonalization method are outlined here. Thistheory chapter is for =EMPOWER= users familiar withnumerical electromagnetics foundations. We have addedthis chapter because MoL is less well known than themethod of moments or the finite difference method.

MoL can be represented as a simple combination of bothmethod of moments and finite difference method. Thuswe have skipped common parts and given our attention tothe original parts of the algorithm. More details on par-ticular algorithm parts, accuracy and convergence inves-tigation results can be found in publications listed in theReferences chapter in the =EMPOWER= Engine Theoryand Algorithms section.

Basically, the theory behind the simulator can be reducedto the following: An initial 3D problem in a layered me-dium is reduced to a 2D problem through a partial dis-cretisation of the Maxwell’s equations and its solution for

a homogeneous layer in a grid spectral domain. Theresultant matrix relating local grid currents and voltagesis reduced to an immitance matrix relating integral cur-rents and voltages in ports. To extract a generalizedscattering matrix of the problem from the immitancematrix, the method of simultaneous diagonalizations isused.

After this introduction we are ready to formulate thereasons for using MoL as a basis for an electromagneticsimulator. The 3D problem is discretized only in twodirections and reduced to a 2D one that correspondsnaturally to the planar MIC structures. In contrast withthe method of moments, the MoL gives a self-regularizedsolution with only one variable (grid cell size) defining allparameters of the numerical model. That eventually leadsto monotonic convergence of calculated data and predict-able errors of calculations. The high grade of internalsymmetries of the MoL based algorithms makes it possibleto substantially reduce the numerical complexity of themain matrix computation stage. The main restriction ofusing a regular grid related with its potentially excessivenumber of variables has been overcome by introducingthinning out and re-expansion procedures. Basically, thediscrete analogue of a problem is processed in a waysimilar to the method of moments but in discrete spacelike the finite difference approach which facilitates differ-ent aspects of the solution and programming.

Thus, the main advantages of the MoL are reliable solu-tion with the predictable calculation error, relativelystraightforward algorithms that facilitate development ofgeneral purpose programs, and a lot of possibilities tospeed up calculations and to increase accuracy of solu-tions. For these reasons and others, we decided to use itfor the electromagnetic simulator. This chapter summa-rizes the theoretical backgrounds with emphasis on theproblem formulation and acceleration techniques.

192 Theory

ELECTROMAGNETIC PROBLEM FORMULATION

This section describes a general mathematical formula-tion of the boundary value problem to be solved. It definesall restrictions in the problem domain. You can use thissection to decide whether your particular problem fits theformulation or not.

For analysis, a passive MIC structure is confined inside athree dimensional rectangular volume bounded by electricor magnetic walls. The volume is filled by a layeredmedium that may consist of an arbitrary number of iso-tropic homogeneous dielectric or magnetic layers as shownin Figure A-1.

The electric (E) and magnetic (H) field vectors are relatedby Maxwell’s system of equations:

z

xy

d

d

d

0

1

P

Ω

Ω

Ω

1

2

P

Figure A-1 A layered medium in a three-dimensional rectangularvolume.

Theory 193

(A-1)

Here Jz is the volume density vector of z-directed currentsinside a media layer. εp and µp are permittivity and per-meability of the media layer. εp is a complex value for alossy media. The z-directed currents are constant valuesinside a layer, but they can change from layer to layer,which gives a possibility to discretize the problem alongthe z-axis. Thus we have all six components of the electricand magnetic fields inside a layer with constant currentacross it. X and y current components can exist only in asignal layer z=dj, parallel to medium layer interfaces.Generalized boundary conditions for the signal layer are:

(A-2)

The signal layer plane can contain arbitrarily shapedregions of perfect metallization, regions with complexsurface impedances (lossy metal), resistive films, and re-gions modeling lumped element connections. All regionshave zero thickness. The top and bottom walls of the boxcan be ideal electric & magnetic walls or walls with surfaceimpedance. The structure can also be terminated by semi-infinite rectangular waveguides in the planes of the boxtop and bottom walls. A clarification of the boundaryconditions for the media layer interfaces (A-2) are givenin the following Table A-1.

rot i

rot i

div

div

x y z

p

pp

H E Jz

E H

E

H

, ,

= += −

==

U

V||

W||

∈

ωεωµ

0

0

a f Ω

1 * H( ) H( )

1 * E( ) E( )

z

z

dj dj

dj dj

+ − − =

+ − − =

k p

k p

η

0

194 Theory

1. Region without metallization η = 0

2. Lossless metallization

3. Surface Impedance

4.

Port Region along X-Axis or InternalPort (Lumped Element Region) alongX-Axis (the same for y-axis)C is region cross-section, l is regionlength.

5. Internal Port along Z-axis

Table A-1

Input ports in the structure are modeled by line segmentsapproaching the outer boundaries (line conductors) andsurface current sources in the regions where line conduc-tors approach the walls of the volume. It is assumed thatthe currents inside the input and the lumped elementregions are constant in the direction of current flow andthe corresponding electric field component along the re-gion is constant across it. Thus, the integral of currentacross the region gives an integral current, and integralof the electric field along the region gives an integralvoltage for the region.

The desired solution of the electromagnetic problem is animmitance matrix relating the integral voltages and cur-rents in the port and lumped element regions. This isactually a kind of Green’s function contraction on the portand lumped element regions. After connection of thelumped elements the immitance matrix can be trans-formed into a generalized Y- or S-matrix using the simul-taneous diagonalization method (see the de-embeddingsection).

1 0z jd* E( ) =

η σ= E

ηdy Y dxi x

lC

= zz E

j Ez

C

l z

l

dxdy Y dzzz z=

Theory 195

Thus, we have a problem formulation that is appropriatefor a wide range of microwave and mm-wave devices suchas planar filters, dividers/combiners, matching circuits,phase-shifters, attenuators, diplexers, amplifiers as wellas their components.

METHOD OF LINES

The method of partial discretization (later called themethod of lines [MoL]) is as old as partial differentialequations and the finite difference approach to their solu-tion. Traces of it can be found in the 18th century worksof J.-L. Lagrange. Its first conscious usage for the numeri-cal solution of elliptical problems could be attributed toM.G. Slobodianskii [1939]. An almost complete referenceon the MoL development and applications in the periodfrom the beginning up to sixties are given in Liskovets’paper [1965].

The network analogue method of B.L. Lennartson [1972]is probably the first technical application of the MoL tothe static numerical analysis of planar multiconductorlines. It was not quite straightforward when it was pub-lished, and the actual exploration of the method for micro-wave integrated circuit structures began in the earlyeighties in works of German scientists H. Diestel, R.Pregla, U. Schulz, S.B. Worm and others [Pregla, Pascher,1989].

The =EMPOWER= algorithms can be also classified asMoL because of its semi discrete nature. Originally thenetwork impedance analogue method [Kron, 1944/ Ses-troretzkiy, 1977] and a grid spectral representation insidehomogeneous layers were used to analyze the layeredthree dimensional structures [Sestrorezkiy,Kustov, Shlep-nev,1988] that correspond to a combination of the 3D finitedifference approach and the spectral domain technique.Later, only the discretisation in the metal plane was left,

196 Theory

but the method still retains some advantages of the net-work impedance analogue method. That is why we some-times refer to the =EMPOWER= numerical techniques asthe impedance interpreted method of lines.

Here are the main solution stages of the impedance inter-preted MoL:

• Partial discretisation of the Maxwell’s equations, only in theplane of metallization (x-y plane).

• Grid spectral representation of the EM fields in thehomogeneous layers.

• Building Grid Green’s Function (GGF) matrix in spectraldomain using impedance form of the solution in a layer.

• Representation of each GGF matrix element as a sum offour elements of an auxiliary array obtained using DFFTtechnique.

• Equidistant grid transformation to a non-equidistant gridusing thinning out and linear re-expansion procedures.

• Automatic detection of symmetry for symmetrical andnearly symmetrical problems (reflection and 180°rotational).

• Solution of the main system of linear algebraic equationsusing partial inversion.

• Resolution to Y- or Z-matrix relating integral grid currentsand voltages in the input and lumped element regions.

MAPPING ON THE GRID

To map a boundary value problem for a partial differentialequation on the grid basically means to substitute theproblem with solution defined in a space of continuousfunctions by a problem with a solution defined in a discretespace. The model solution must be as close to the continu-ous one as possible. To solve the problem we approximatedthe partial derivatives in the signal plane by finite differ-

Theory 197

ences applied to grid ana-logues of the field compo-nents. The correspondinggrid is shown in FigureA-2.

There are L+1 equidis-tant cells along the x-axisand M+1 cells along they-axis. The grid equiva-lents of the electric (e) andmagnetic (h) fields are de-fined as correspondingcontinuous function val-ues in offset grid points asis shown for a grid cell inFigure A-2. The grid func-tions are continuousalong the z-axis. Grid x-and y-directed currentvariables (Jx,Jy) are de-fined as integrals of the

surface current in the metal plane across the grid cell.Grid z-directed currents (Jz) are defined as surface inte-grals of the volume current density jz across the grid cell.

The first offset model of Maxwell’s equations was appar-ently proposed by G. Kron [1944]. Figure A-3 shows asummary of the similar models implemented by differentauthors. The resultant system of differential-differenceequations approximates the initial system with the sec-ond order locally inside a layer. The initial boundary valueproblem can contain infinitesimally thin metal regionswith consequent singularities of the field and conductivitycurrents at the metal edges [Meixner, 1972]. That is whya global approximation order of the problem is usuallylower and the largest calculation error part for integralparameters of a structure (Y, S-matrix elements, charac-

Figure A-2 The grid and itsvariables.

198 Theory

G. Kron, 1944

B. Sestroretzkiy, 1977

K. S. Yee, 1966Also T. Weiland, 1977

S. B. Worm,R. Pzegla, 1984

=empower=

Figure A-3 The offset finite difference scheme analogues.

Theory 199

teristic impedance) decreases usually proportionally tothe grid cell size. That is, the monotonic convergence wasobserved for almost all problems solved on the initialequidistant grid. This makes it possible to use such pow-erful convergence acceleration techniques as Richardson’sextrapolation [Richardson, 1927; Marchuk, Shaidurov,1979]. Note that this is an observation and it cannot beproven to work for all problems. The technique used herefor the descriptor matrix evaluation using current sourcesin the metal plane is empirical. The evaluation accuracydepends on parasitic high order modes that could beexcited by current sources and if they are close to theircutoffs or even are propagating, the estimated descriptormatrix could be far away from the correct one. This canbe expected, however, since real circuits which have unex-pected high order modes near the cutoff usually do notwork properly either.

GRID GREEN’S FUNCTION

The Grid Green’s Function (GGF) has been mentionedquite a few times. The GGF is a solution of the differen-tial-difference analogue of Maxwell’s equations (A-1) ex-cited by a unit grid current (Jx,Jy or Jz). The solution orresponse function is a discrete function in the xy plane andcontinuous inside layer along the z-axis. Actually to solvethe formulated problem we need just a contraction of theGGF to the signal plane and to the regions with non-zeroz-directed currents. This contraction is a matrix due tothe discretization.

To find the GGF matrix we used a spectral approachsimilar to one used in the spectral domain technique or inthe method of moments [Nikol’skii, 1982; Vesnin, 1985;Jansen,1985; Rautio, Harrington,1987; Dunleavy,Katehi,1988]. Instead of continuous TE and TM rectan-gular waveguide eigenwaves [Samarskii, Tikhonov, 1948],their grid analogues are used as a basis to expand the

200 Theory

electromagnetic field inside a layer. The number of thegrid TE and TM waves is finite and their system is com-plete. This means that instead of a summation of seriesas in the spectral domain approach we have finite sums.Moreover each basis grid eigenwave has a grid correctionthat provides convergence of sums to the series obtainedby the continuous spectral domain approach. Note that abackward process is impossible and a simple truncation ofthe series does not give the same answer as the gridtechnique. The finite sums and the grid corrections arethe most important things for monotonic convergence ofthe algorithm.

To construct the GGF matrix in the grid spectral domain,the impedance form of the solution for a layer was used.The base of the solution is a layer admittance matrix inthe grid spectral domain. This matrix relates the gridanalogues of the tangential electric and magnetic fieldcomponents at opposite surfaces of the layer, z-directedcurrents and integrals of z-directed grid electric field alongthe z-directed current inside the layer. All of these are inthe basis of the grid eigenwaves, thus we have a set ofindependent matrices for each pair of grid eigenwaves.Uniting those matrices for all layers in a structure givesa grid spectral GGF representation. The construction pro-cedure is completely automated for arbitrarily layeredconfigurations. This technique is similar to the impedanceapproach in the spectral domain [Uwano, Itoh, 1989]. Thegrid spectral GGF representation was also called a GGFeigenvalue vector, but that term is not quite correct. Thedimension of the vector is about 3*L*M if there is only onesignal layer.

All we need now to get the GGF matrix in the initial spaceis to perform a backward transformation of the GGFeigenvalue vector from the grid spectral domain to thespatial domain. To do it an auxiliary array called generalsums array is introduced. The dimension of the general

Theory 201

sums array is also about 3*L*M. Each element of the GGFmatrix can be obtained as a sum of four elements of thegeneral sums array. The general sums array depends onlyon the box and media structure and the grid cell size. Itselements are calculated via the discrete Fourier trans-forms of the GGF eigenvalue vector using the Prime Factoralgorithm. This stage is based on the maximal utilizationof internal symmetries of the bounded equidistant gridand usually takes negligibly small CPU time. Moreoverit can be done only once for all structures with the samebox, media and grid. The described technique is quitesimilar to the main matrix filling procedure designed forthe spectral domain technique [Hill, Tripathi, 1991], ex-cept that it has been done here in finite space and wecalculate the GGF matrix elements without additionaltruncation (or series summation) errors. It can also bereformulated in matrix form in accordance with [Pregla,Pascher, 1989]. The GGF matrix can be represented by asum of Toeplitz and Hankel matrices and their rows canbe obtained directly from the general sums arrays.

INFORMATIONAL MULTIPORT

The informational multiport term was introduced by B.V.Sestroretzkiy [1987] and in a nutshell means a modelmultiport that reflects electromagnetic properties of anobject before superimposing an additional boundary con-dition. It comprises information about all possible struc-tures that could be formed by different combinations of theadditional conditions. The boundary condition superim-posing can be represented as a set of simple manipulationswith the informational multiport terminals. We haveadded this section to clarify connections of the numericalelectromagnetic solution with the circuit theory. Thistechnique is also known as the impedance interpretationof boundary condition superimposition.

202 Theory

The GGF matrix obtained in the previous section can berepresented as an impedance matrix Z of a multiportshown on the left in Figure A-4 (compare with Figure A-2).

The multiport terminals are conceptual and their posi-tions are just a schematic representation. Four concep-tual ports or pairs of terminals correspond to a grid cell asshown in the Figure A-4. The total number of portsoriented along the x-axis is M*(L+1). The total number ofports oriented along the y-axis is L*(M+1). The multiportcan also have a set of z-directed ports corresponding tovia-holes or z-directed internal inputs. Note that we donot need to calculate all elements of the multiport imped-ance matrix and its order can be reduced taking intoaccount that some ports are no-loaded or short circuited.The no-loaded terminals correspond to regions of thesignal layer without any conductivity currents. The right

Figure A-4 Left: The informational multiport for a signal layer andcorrespondence of the grid variables to voltages and currents atthe multiport terminals. Right: The boundary conditions andcorresponding operations with the informational multiportterminals.

Theory 203

half of Figure A-4 illustrates the correlation of other typesof the boundary conditions to operations with the informa-tional multiport terminals. Operations with the z-di-rected terminals are similar.

The operations in a discrete space of the informationalmultiport terminals are completely in accordance with theusual electromagnetic theory. To connect a lumped ele-ment for example (see Figure A-4), we performed bothserial connections of terminals along the element thatcorresponds to the electric field integration along theelement and parallel connections that corresponds to thesurface current integration across the element (see TableA-1). The analogies described are meant to facilitateunderstanding of numerical electromagnetics. Note thatthe examples given are not the only possible manipula-tions with the terminals with physical electromagneticequivalents.

NUMERICAL ACCELERATION PROCEDURES

Before filling the reduced GGF matrix we can additionallydecrease the GGF matrix order and required storage spaceby means of thinning out with linear re-expansion proce-dures and by incorporating a geometrical symmetry intothe problem.

Thinning out is a simple elimination of the grid currentsin metallized regions that can be represented by a smallernumber of currents without loosing accuracy. As an illus-trative example, the left half of Figure A-5 shows a threeresonator filter mapped on the grid. The grid cells withpossible non-zero conductivity currents (metallization re-gions) are depicted by the thick lines. The thinning outprocedure decreased the number of the currents in theproblem and leaves the currents that are shown by thethick lines in the right half of Figure A-5. This looks likea pseudo-non-equidistant grid over the regular grid that

204 Theory

is finer near edges, corners and via-holes and coarserinside the solid metal regions. The enlarged secondarygrid cells after the thinning out consist of non-divergentcurrent borders along each side that can be substituted bytwo variables on the grid using linear re-expansion. Com-bination of these two procedures makes it possible toovercome restrictions of the MoL with a regular grid whilekeeping the main advantages of the equidistant grid.

The described procedure with total elimination of somecurrents inside the solid metal regions is called the wiremodel. It basically substitutes a problem with another onewith removed small metallization pieces. It certainly givesan additional error, but fortunately this error is oppositeto the regular grid model error. In other words, the wirethinning out model actually increases the solution accu-racy if the structure is thinned out properly. However, if

Figure A-5 The grid mapping of an interdigital bandpass filterbefore (left) and after (right) thinning out.

Theory 205

too much metal is removed, the thinning out error domi-nates. Thus, a solid thinning out model procedure wasintroduced to avoid it. The solid model can be representedas a simple modification of the wire model. To explain itwe start from the pseudo-non-equidistant grid of currentsformed for the filter and shown in the Figure 7. Insteadof complete elimination of the currents inside the enlargedgrid cells we leave some of them to keep metal surfacesolid. Those currents left are also replaced with just twovariables by means of linear re-expansion. The solidmodel is more correct but gives a larger number of vari-ables for similarly thinned out problems in comparisonwith the wire model. (The solid model is actually a wayto form a non-equidistant grid with the grid functionre-expansion in a discrete space.)

The GGF matrix of a symmetrical problem could be re-duced to a centrosymmetrical matrix (with centrosymmet-rical blocks in the case of two-plane symmetry) and it istreated in the way similar to described in [Weeks, 1979].This reduces required CPU memory from 4 to 16 times(serial allocation of partial matrices) and speeds up calcu-lations from 4 to 16 times. One plane, two plane, and 180°rotational symmetries are included in the program.

Thereafter, the classic Gauss’ inversion algorithm is usedwith a few changes. The result of this stage of solution isa matrix (Y- or Z-matrix) relating the grid currents andvoltages in the input source regions, and thus we need toget only a small part of the inverted matrix correspondingto these variables. A partial inversion procedure performsit and gives an additional acceleration.

DE-EMBEDDING ALGORITHM

The method of simultaneous diagonalization (MoSD)[Shlepnev, 1990, 1998] is used to extract a multimode orgeneralized S-matrix. The MoSD is based on the electro-

206 Theory

magnetic analysis of two line segments corresponding toan MIC structure port to be de-embedded. The segmentshave different lengths and the same surface currentsource regions as in the initial structure. The result of theEM analysis is two Y-matrices relating integral grid cur-rents and voltages in the source regions. These matrices,transformed from the space of the grid functions to a spaceof the line eigenmodes, are set equal to Y-matrices describ-ing independent modes propagated in continuous part ofthe line segments. It gives the basic non-linear system ofequations relating eigenwave propagation constants andcharacteristic impedances, a matrix of transformationfrom the grid functions space to the mode’s space (trans-formation matrix) and an auxiliary matrix that helps tomatch propagated modes perfectly (compensation matrix).Solution of the system is based on simultaneous diagonali-zation of Y-matrix blocks. Each port of the MIC structureor discontinuity can be de-embedded using the pre-calcu-lated line parameters and the transformation and com-pensation matrices. The main advantages of thisapproach are the possibility of multimode deembeddingwithout direct spectral analysis of the line cross-sectionand ideal matching of line eigenmodes in the analysis ofthe line segment that increases the accuracy of disconti-nuity analysis.

Note that despite the theoretical ability to excite and tomatch any propagating line eigenwave using the surfacecurrent sources in the metal plane, it does not always workin the discrete models. Using a limited number of vari-ables in the source regions it is sometimes impossible toseparate different modes completely. Moreover, the suc-cess of the MoSD application depends on the high ordermodes that could substantially influence the result. Thisis the main drawback of the described MoSD applicationto planar structures.

Theory 207

Appendix B

File Descriptions

I n performing its tasks, =EMPOWER= creates manydifferent types of files. An understanding these differ-ent files is very helpful in understanding the operation

of =EMPOWER=. These files contain the topology of thecircuit, external port line data, generalized S-Parameternormalizing impedances, output information, S-Parame-ter data, batch commands, Y-Parameter data, viewerdata, and backup data.

WHERE ARE THESE FILES?

Starting with Version 7.0, GENESYS uses OLE Struc-tured Storage for its workspace files. These files aresometimes called “file systems in a file.” Structured Stor-age files contain internal directories and files, and copyingone workspace file copies all internal files contained in it.Figure B-1 shows the structure of a typical workspace file.Notice that within each simulation, all filenames beginwith “EMPOWER”.

Note: Previou s version s of GENESYS used actua l diskfile s for all interna l =EMPOWER= files an d separatesubdirectories were recommended for each circu it.This is n o longe r necessary for typical us age.

If you need to access these internal files in a workspace,you have two options:

• Right-click on the =EMPOWER= simulation on the tree andselect “Write Internal Data Files”. This automaticallycreates a directory with the same name as the simulationand places copies of the files there.

• In the same workspace, you can access internal files usinga special file prefix, “WSP:”, followed by folder names andthe filename. For example, to access the EMPOWER.SS,use the name “WSP:Simulations\EM1\EMPOWER.SS”.

The second method has the advantage of automaticallyupdating whenever the =EMPOWER= simulation is re-run. The first method requires you to re-write the datafiles whenever you need an updated version.

Figure B-1 The internal contents of a typical workspace file.

210 File Descriptions

TEXT FILES VS. BINARY FILES

There are two basic types of data files: text (sometimescalled ASCII) and binary. Text files are human readablefiles. They are universal and can be edited with manydifferent programs such as NOTEPAD or DOS EDIT.Among the text files used by =EMPOWER= are batch,topology, listing, and S-Parameter files.

Note: Word processors can also edit text files, however,they will store binary formatting information in the fileunless explicitly told not to (“Save as...Text”), so we donot recommend their use for editing text files.

Extension Type Purpose

BAT Text Batch file used to launch =EMPOWER=

EMV Binary Viewer data

L1, L2, etc. BinaryPort deembedding and line data for port 1, 2,etc.

LST TextListing file summarizing all =EMPOWER=data

PLX Text Text listing of viewer currents.

R1, R2, etc. Text Port normalizing impedances

RGF BinaryPort deembedding and line data for a port witha user-specified deembedding file name.

RX Text Frequency vs. impedance data

WSP Binary GENESYS Workspace File

SS Text S-Parameter results

TPL Text “Netlist” for =EMPOWER=

Y Binary Y-Parameter results

Table B-1 File types used by =EMPOWER=.

File Descriptions 211

In contrast, binary files are not human readable. Theycontain information encoded into the numbers whichmake up the file which are ultimately turned into ones andzeros, thus the name “binary.” Unlike text files, binaryfiles are not universal and should only be edited by aprogram designed for the particular type of binary file youare using. Editing a binary file in a regular wordprocessor or text editor will undoubtedly destroy it!Some binary files used by =EMPOWER= and GENESYSare workspace, line, and Y-Parameter files.

FILE EXTENSIONS

You can normally tell the kind of file you have by lookingat its extension (the part of the name after the last period).Some commonly used extensions include EXE (ex-ecutable), TXT (text), and HLP (help). Each kind of fileused by =EMPOWER= has its own unique extension.These extensions are shown in Table B-1. Each of thesetypes will be discussed individually below.

Note: Unfortunately, Windows can be setup to hide filesextensions, as well as actual files, from the user. Wewould recommend that you turn off this “feature”: Dou-ble click on “My Computer,” Select Options from theView menu, Click the Viewer tab, Click “Show all files,”Deselect “Hide MS-DOS file extensions for file typesthat are registered”, and click OK. Different versions ofWindows may have slightly different procedures.

.BAT (BATCH) FILES

Written by: GENESYSType: TextCan be safely edited: Yes*Average size: 50 bytesUse: Running =EMPOWER= from the command line

212 File Descriptions

When GENESYS runs =EMPOWER=, it creates a batchfile with that can be used to run =EMPOWER= from thecommand line with the same options. You can edit this fileto customize the options, but if you rerun =EMPOWER=from GENESYS, your changes will be overwritten, so it isbetter to update options from the “Cmd. Line:” prompt inGENESYS. For more information on batch files, see Chap-ter 10, Command Line Options.

.EMV (VIEWER) FILES

Written by: =EMPOWER=Type: BinaryCan be safely edited: NoAverage size: 10 to 100Kbytes, but may be largerUse: Data for viewing currents or voltages

EMV files (=EMPOWER= Viewer) files are completelyself-contained files containing all information needed bythe viewer to display currents and voltages for a circuit.These files contain information about the box and the gridmapping of the circuit as well as actual complex currentor voltage values at each frequency. =EMPOWER= cre-ates an .EMV file whenever “Generate Viewer Data” ischecked or the -In option is specified.

.EMV files can only be read by the =EMPOWER= viewer.If you want to generate viewer data for import into otherprograms, you should generate a .PLX text file. For moreinformation on viewer files, see Chapter 7, Viewer.

.L1, .L2, ... .Ln (LINE DATA) FILES

Written by: =EMPOWER=Type: BinaryCan be safely edited: NoAverage size: 1 to 5Kbytes, but may be largerUse: Internal file for =EMPOWER= but can also be usedin the SMTLP and MMTLP models in GENESYS

File Descriptions 213

=EMPOWER= must perform a separate line analysis forall external ports. If no filename is specified by the user,then the results from the line analysis are stored in .Lnfiles. These files also store all information about the boxand port and are intelligent: They are only recalculated ifnecessary, and even then only at frequencies necessary.Even if the circuit changes they are only recalculated ifthe change affects the line analysis. Notes: When thesefiles are numbered, modally related groups of ports arecounted as one. Also, if two ports are identical, then onlythe first one will create a .Ln file.

Tip: If you rename or copy a TPL file or SCHEMAX file,you may want to also copy or rename all other files,especially line data files. Since these files are intelli-gent, you can save a lot of line analysis recalculationtime if you use this technique.

.LST (LISTING) FILES

Written by: =EMPOWER=Type: TextCan be safely edited: Yes*Average size: 50K to 200K, but may be largerUse: Gives all calculated data and grid mapping from=EMPOWER= in human readable form

This file is overwritten whenever =EMPOWER= is run. Itshould be carefully checked whenever a new circuit isanalyzed, especially if that circuit was described manuallyfrom a text TPL file. The following sections describe thecontents of a listing file. Note: Some of the informationdescribed below is only output if “Output additional infoin listing file” is checked or -La is specified.

214 File Descriptions

QCHK SECTION

This section allows you check the quality of the solution.Entries include:

Min. media wavelength to mesh size ratios - should be atleast 20.

Thinning out thresholds - Specifies the maximum numberof lines in a row which can be thinned out.

Max box size to media wavelength ratios - If the box is toolarge, you will have box resonances. If this line ends withan exclamation mark (!), it may be too large. See Chapter8, Box Modes, for more details.

PACKAGE STRUCTURE

This section is only present when the -La option is used.It gives a summary of the substrate and metal layers usedas well as cell sizes.

MEMORY SECTIONS

Several memory sections throughout the listing file givememory requirements for different parts of the simula-tion.

MAP OF TERMINALS

This section shows the grid representation of the problem.The following symbols are used:

File Descriptions 215

Symbol Meaning

+ Grid line intersection.

- Lossless metal along x direction.

| Lossless metal along y direction

#Electrically lossy metal. Also, when in place of a plus (+)sign, it is Z directed metal going to the box top

^ Metal along x direction described with physical loss

> Metal along y direction described with physical loss

1-9 External deembedded port.

A-Z Internal or non-deembedded external port.

* Z directed metal (viahole) going to box bottom.

0Z directed metal (viahole) going to both the box top andthe box bottom.

When the -Dc option is used, there are additional tablesbelow the map of terminals. The tables “Terminals alongX/Y-axis” and “Symmetry processing variables” are fortechnical support use only. The input thinning out pa-rameters show which lines along external ports have beenthinned out.

SDTC SECTION

Symmetry detection sections specify whether the struc-ture is symmetrical. The symmetry processing addition-ally shows where any differences occurred and can be veryuseful in finding out where the structure is not symmet-rical. The coordinates specified refer to the terminal mapshown above.

LINE ANALYSIS MODE RESULTS

This area of the listing contains sections identical to thosedescribed above which pertain to the line analysis. Below

216 File Descriptions

these sections you will find a table of line parameters foreach frequency. The entries are:

Nm - port number

Type - impedance type, real (re) or imaginary (im). Nor-mal lines should have a real impedance.

Zo (ohm) - Line impedance

Gw(rad/m) - propagation constant

Gw/Go - propagation constant relative to free space

Comp Phase, Compensation Admittance - value of phaseand impedance compensation for deembedding.

S-MATRIX TABLES

Each table gives the circuit’s s-parameters at one fre-quency. For normal, non-multimode inputs, as an exam-ple, S21 is found in the row with input numbers 2 and 1(in that order).

.PLX (CURRENT/VIEWER DATA) FILES

Written by: =EMPOWER=Type: TextCan be safely edited: Yes*Average size: 200 Kbytes to 2Mbytes, but may varyUse: Importing current data from =EMPOWER= intoanother application, such as Matlab or Excel.

This file contains two tables per frequency, one each for x-and y-directed currents. Each table contains 4 columnscontaining the x and y coordinates followed by the real andimaginary part for each current. These tables could beedited, but it would be best to leave them alone since theywould be very tedious and error-prone to edit them byhand. These files should be very useful in other applica-tions, as the engineers at Eagleware used third party

File Descriptions 217

applications to graph currents before our =EMPOWER=viewer was completed.

.R1, .R2, ... Rn (PORT IMPEDANCE) FILES

Written by: =EMPOWER=Type: TextCan be safely edited: Yes*Average size: 1KbyteUse: Read by GENESYS when Generalized S-Parametersare requested

These files contain each port’s impedance versus fre-quency. These ports are read by GENESYS if the keywordGEN is used in place of a termination impedance. The filesare formatted just like RX files in GENESYS; see the Dev-ice Data section of the Reference manual for a descrip-tion of these files. GENESYS always requests these fileswhen =EMPOWER= is run from GENESYS. From thecommand prompt, use option -FI to get these files. Eventhough these files can be edited, they will be overwrittenwhenever =EMPOWER= is rerun. Notes: These files arenumbered differently than Ln files. When these files arenumbered, each port in a related group of ports is countedindividually. All ports, even if some are identical, willoutput Rn files if the -FI option is used.

.RGF (LINE DATA) FILES

Written by: =EMPOWER=Type: BinaryCan be safely edited: NoAverage size: 1 to 5Kbytes, but may be largerUse: Internal file for =EMPOWER= but can also be usedin the SMTLP and MMTLP models in GENESYS

These files are used in place of .Ln files if a filename wasgiven on the PORT line in the TPL file. When run fromGENESYS, this file type is not available; use the .Ln files

218 File Descriptions

instead. Otherwise, they are completely identical to the.Ln files described earlier.

.RX (FREQUENCY VS. IMPEDANCE) FILES

Written by: UserType: TextCan be safely edited: YesAverage size: 1KbyteUse: Specifying electrical losses

These files are used to specify the impedance of conductorsin ohms per square. These files are used in the =EM-POWER= layers setup dialog box or in the TPL file. Thefiles are formatted just like RX files in GENESYS; seethe Device Data section in the Reference manual for adescription of these files.

.WSP (WORKSPACE) FILES

Written by: GENESYSType: BinaryCan be safely edited: Yes, but only using GENESYSAverage size: 10 to 2,000 KbytesUse: Contains complete simulation, graph, schematic, andlayout information from GENESYS

See the User's Guide for more details on .WSP files.

.SS (S PARAMETER) FILES

Written by: =EMPOWER=Type: TextCan be safely edited: Yes*Average size: 5 to 50 Kbytes, but may be largerUse: Contains S-Parameter data calculated by =EM-POWER=

This file contains the S-Parameter data written by =EM-POWER=. It is in the industry standard format and can

File Descriptions 219

be loaded into most RF and Microwave simulators. Theformat is described in the Device Data section of theReference manual. Even though these files can be ed-ited, they will be overwritten whenever =EMPOWER= isrerun.

.TPL (TOPOLOGY) FILES

Written by: User or GENESYSType: TextCan be safely edited: Yes*Average size: 1 to 5KbytesUse: Describing circuit to =EMPOWER=

This file contains a complete description of the circuit tobe analyzed by =EMPOWER=. GENESYS will create thisfile automatically whenever =EMPOWER= is run fromthe =EMPOWER= menu in GENESYS. Even though thisfiles can be edited, it will be overwritten if =EMPOWER=is rerun from within GENESYS. For complete details onthis file, see Chapter 11, TPL File Format.

.Y (Y-PARAMETER) FILES

Written by: User or GENESYSType: BinaryCan be safely edited: NoAverage size: 2 to 25Kbytes, but may be largerUse: Internal data file for =EMPOWER=

This file contains the calculated Y-parameters beforedeembedding. If merge (-ME) is specified, the previousdata stored in this file is combined with the newly calcu-lated data, and the .SS (S-Parameter) file is rewritten.

~SS, ~RG, ETC. (BACKUP) FILES

All files with a name or an extension starting with tilde(~) are backup files and can be safely deleted. Examplesof these files are ~OMBINE.TPL and COMBINE.~RG.

220 File Descriptions

Appendix C

References

GENERAL BACKGROUND

J.A. Stratton, Electromagnetic theory, McGraw-Hill Co.,New-York, 1941

G. Kron, “Equivalent circuit of the field equations of Max-well.-Part I.,” Proc. of IRE, 1944, May, p. 289-299.

C.G. Montgomery, R.H. Dick, E.M. Purcell, Principles ofmicrowave Circuits, McGraw-Hill Co., New-York,1948.

O. Heaviside, Electromagnetic theory, AMS Chelsea Pub-lishing Co., New-York, 1950.

A.A. Samarskii, A.N. Tikhonov, “About representation ofwaveguide electromagnetic fields by series of TE and TMeigenwaves” (in Russian), GTF (Journal of TheoreticalPhysics), 1948, v. 18, p. 959-970.

P.I. Kuznetsov, R.L. Stratonovich, The propagation of elec-tromagnetic waves in multiconductor transmission lines,Pergamon Press, Oxford, 1964 (originally published inRussian, 1958).

K.S. Yee, “Numerical solution of initial boundary valueproblems involving Maxwell’s equations in isotropic me-dia,” IEEE Trans. v. AP-14, 1966, p. 302-307.

V.V. Nikol’skii, Variational approach to internal problemsof electromagnetics (in Russian) , Moscow, Nauka, 1967.

J. Meixner, “The behavior of electromagnetic fields atedges”, IEEE Trans, v. AP-20, 1972, N 7, p.442-446.

B.V. Sestroretzkiy, “RLC and Rt analogies of electromag-netic space” (in Russian), in Computer aided design ofmicrowave devices and systems, Edited by V.V. Nikol’skii,Moscow, MIREA, 1977, p. 127-128.

T. Weiland, “Eine Methode zur Losung der MaxwellschenGleichngen for Sechskomponentige Feleder auf DikreterBasis”, Arch. Electron. Uebertragungstech., v. 31, N 3,1977, p.116-120.

Computer-aided design of microwave devices (in Russian),Edited by V.V. Nikol’skii, Moscow, Radio i Sviaz’, 1982.

R.H. Jansen, “The spectral-domain approach for micro-wave integrated circuits,” IEEE Trans., v. MTT-33, 1985,N 10, p. 1043-1056.

S.G. Vesnin, Electromagnetic models for design of micros-trip microwave structures (in Russian), Ph.D. Thesis,MPEI, Moscow, 1985.

E.F. Johnsom, “Technique Engineers the Cavity Reso-nance in Microstrip Housing Design,” MSN & CT, 1987,Feb., p. 100-102, 107-109.

J.C. Rautio, R.F. Harrington, “An electromagnetic time-harmonic analysis of shielded microstrip circuits,” IEEETrans., v. MTT-35, 1987, N 8, p. 726-730.

B.V.Sestroretzkiy,V.Yu.Kustov, “Electromagnetic analysisof multilevel integrated circuits on the base of RLC-net-works and informational multiport approach” (in Rus-sian), Voprosi Radioelektroniki, ser. OVR, 1987, N 1, p.3-23.

L.P. Dunleavy, P.B. Katehi, “A generalized method foranalyzing shielded thin microstrip discontinuities”, IEEETrans., v. MTT-36, 1988, N 12, p.1758-1766.

222 References

T. Uwaro, T. Itoh, “Spectral domain approach,” in Numeri-cal techniques for microwave and millimeter-wave passivestructures, Edited by T. Itoh, John Willey & Sons, 1989.

R.H. Jansen, “Full-wave analysis and modeling for CADof mm-wave MMICs,” Alta Frequenza, v. LVIII, 1989, N5-6, p. 115-122.

A. Hill, V.K. Tripathi, “An efficient algorithm for the three-dimensional analysis of passive microstrip componentsand discontinuities for microwave and millimiter-waveintegrated circiuts”, IEEE Trans., v. MTT-39, 1991, N 1, p.83-91.

THE METHOD OF LINES

M.G. Slobodianskii, “A new method of approximate solu-tion of partial differential equations and its application tothe theory of elasticity” (in Russian), PrikladnaiaMatematika i Mekhanika (Applied Mathematics and Me-chanics), v. 3, 1939, N 1, p. 75-82.

O.A. Liskovets, “The method of lines, Review” (in Russian),Differenzial’nie Uravneniya, v. 1, 1965, N 12, p. 1662-1668.

B.L. Lennartson, “A network analogue method for comput-ing the TEM characteristics o planar transmission lines,”IEEE Trans., v. MTT-20, 1972, N 9, p. 586-590.

U. Schulz, “On the edge condition with the method of linesin planar waveguides,” Arch. Electron. Uebertragung-stech., v. 34, 1980, p.176-178.

U Schulz, R. Pregla, “A new technique for the analysis ofthe dispersion characteristics of planar waveguides andits application to microstrips with tuning septums,” RadioScience, v. 16, 1981, Nov.-Dec., p. 1173-1178.

S.B. Worm, R. Pregla, “Hybrid-mode analysis of arbitrarilyshaped planar microwave structures by the method oflines,” IEEE Trans., v. MTT-32, 1984, N 2, p. 191-196.

References 223

R. Pregla, W. Pascher, “The method of lines,” in Numericaltechniques for microwave and millimeter-wave passivestructures, Edited by T. Itoh, John Willey & Sons, 1989.

S.B. Worm, “Full-wave analysis of discontinuities in pla-nar waveguides by the method of lines using a sourceapproach,” IEEE Trans., v. MTT-38, 1990, N 10, p.1510-1514.

RICHARDSON’S EXTRAPOLATION

L.F. Richardson,"The differed approach to the limit. 1:Sin-gle lattice," Philos. Trans. of Royal Society, London, ser. A,226, 1927, p.299-349.

A. Premoli, “A new fast and accurate algorithm for thecomputation of microstrip capacitances,” IEEE Trans. v.MTT-23, 1975, N 8, p. 642-647.

G.I.Marchuk,V.V.Shaidurov,Difference methods and theirextrapolations, Spr.-Verlag, 1983 (originally published inRussian, 1979).

A.G. Vikhorev, Yu.O. Shlepnev, “Analysis of multiple-con-ductor microstrip lines by the method of straight lines,”Journal of Communications Technology and Electronics,1991, N 12, p. 127-129, originally published in Ra-diotekhnika i Elektronika, v. 36, 1991, N 4, p. 820-823.

SYMMETRY PROCESSING

M. Hammermesh, Group theory and its application tophysical problems, Pergamon Press, Oxford, 1962.

I.J. Good, “The inverse of a centrosymmetric matrix,”Technometrics, Journal of Statictics for Physical Chemicaland Engineering Science, v. 12, 1970, p. 925-928.

P.R. McIsaac, “Symmetry-induced modal characteristics ofuniform waveguides,Part I:Summary of results,PartII:Theory,” IEEE Trans., v. MTT-23, 1975, N 5, p.421-433.

224 References

W.T. Weeks, “Exploiting symmetry in electrical packaginganalysis,” IBM Journal of Research and Development, v.23, 1979, N 6, p.669-674.

A.B. Mironov, N.I. Platonov, Yu.O. Shlepnev, “Electrody-namics of waveguiding structures of axisymmetrical mi-crowave integrated circuits,” Journal of CommunicationsTechnology and Electronics, 1990, N 7 p. 71-76, originallypublished in Radiotekhnika i Elektronika, v. 35, 1990, N 2,p. 281-286.

E.V. Zakharov, S.I. Safronov, D.P. Tarasov, “Abelian Groupsof finite order in numerical solution of potential theoryboundary value problems” (in Russian), GVM & MF (Jour-nal of Computational Mathematics and Mathematical-Physics), v. 32, 1992, N 1, p. 40-58.

=EMPOWER= ENGINE THEORY ANDALGORITHMS

B.V. Sestroretzkiy, V.Yu. Kustov, Yu. O. Shlepnev, “Analysisof microwave hybrid integrated circuits by informationalmultiport network method” (in Russian), Voprosi Ra-dioelektroniki, ser. OVR, 1988, N 12, p. 26-42.

B.V. Sestroretzkiy, V.Yu. Kustov , Yu.O. Shlepnev, “Tech-nique of electromagnetic analysis of microstrip devicesusing general purpose programs” (in Russian), VoprosiRadioelektroniki, ser. OVR, 1990, N 1, p. 3-12.

Yu.O. Shlepnev, Method of lines in mathematical modelingof microwave integrated circuit planar elements (in Rus-sian), Ph.D. Thesis, NEIS, Novosibirsk, 1990.

V.Yu. Kustov, B.V. Sestroretzkiy, Yu.O. Shlepnev, “Electro-magnetic analysis of planar devices with resistive filmsand lumped elements,” Proc. of Europ. Symp. on Numeri-cal Methods in Electromagnetics (JEE’93), Toulouse,France, 17-19 November, 1993, p. 227-234.

References 225

V.Yu. Kustov, B.V. Sestroretzkiy, Yu.O. Shlepnev, “Three-di-mensional electromagnetic analysis of planar devices withresistive films and lumped elements,” Proc. of 27th Con-ference on Antenna Theory and Technology (ATT’94),Mos-cow, Russia, 23-25 August, 1994, p. 352-356.

K.N. Klimov, V.Yu. Kustov, B.V. Sestroretzkiy, Yu.O. Shlep-nev, “Efficiency of the impedance-network algorithms inanalysis and synthesis of sophisticated microwave de-vices,” Proc. of the 27th Conference on Antenna Theoryand Technology (ATT’94), Moscow, Russia, 23-25 August,1994, p. 26-30.

V.Yu. Kustov, B.V. Sestroretzkiy, Yu.O. Shlepnev, “TAMICpackage for 3D electromagnetic analysis & design ofMICs,” Proc. of the 5th Intern. Symp. on Recent Advancesin Microwave Technology (ISRAMT’95), Kiev, Ukraine,September 11-16, 1995, p. 228-233.

Yu. O. Shlepnev, B.V. Sestroretzkiy, V.Yu. Kustov, “A newmethod of electromagnetic modeling of arbitrary trans-mission lines,” Proc. of the 3rd Int. Conference Antennas,Radiocommunication Systems and Means (ICARSM’97),Voronezh, 1997, p.178-186.

Yu.O. Shlepnev, B.V. Sestroretzkiy, V.Yu. Kustov, “A newapproach to modeling arbitrary transmission lines,” Jour-nal of Communications Technology and Electronics, v. 42,1997, N 1, p. 13-16, originally published in Radiotekhnikai Elektronika, v. 42, 1997, N 1, p. 13-16.

Yu.O. Shlepnev, “A new generalized de-embedding methodfor numerical electromagnetic analysis,” Proceedings ofthe 14th Annual Review of Progress in Applied Computa-tional Electromagnetics, Monterey, CA, March 16-20,1998, v.II, p. 664-671.

Yu.O. Shlepnev, “Extension of the method of lines forplanar 3D structures,” Proceedings of the 15th Annual

226 References

Review of Progress in Applied Computational Electromag-netics, Monterey, CA, 1999, p. 116-121.

TEST EXAMPLES AND COMPARISONS

E.G. Farr, C.H. Chan, R. Mittra, IEEE Trans., v. MTT-34,1986, N 2, p. 307.

G. Gronau, I. Wolff “A simple broad-band device de-embed-ding method using an automatic network analyzer withtime-domain option“, IEEE Trans., v. MTT-37, 1989, N 3,pp. 479-483.

D.J. Swanson, “Grounding microstrip lines with via holes”,IEEE Trans., v. MTT-40, 1992, p. 1719-1721.

J.C. Rautio, “An ultra-high precision benchmark for vali-dation of planar electromagnetic analysis”, IEEE Trans.,v. MTT-42, 1994, N 11, p. 2046-2050.

T. Kawai, I. Ohta, “Planar-circuit-type 3-dB quadraturehybrids“, IEEE Trans., v. MTT-42, 1994, N 12, p. 2462-2467.

Y. Gao, I. Wolff, “Miniature electric near-field probes formeasuring 3-D fields in planar microwave circuits“, IEEETrans., v. MTT-46, 1998, N 7, p. 907-913.

References 227

Appendix D

=EMPOWER= Messages

T his chapter contains a description of all error mes-sages which are either displayed by =EMPOWER= orare displayed by GENESYS but are relevant to =EM-

POWER=. These error messages are given in alphabeticalorder. For all other GENESYS, =SCHEMAX=, and =LAY-OUT= error messages, see Appendix G, Error Messages,in the GENESYS Simulation Manual.

All error messages which refer to line numbers are indi-cating the line number in the TPL file. If you are creatinga file from =LAYOUT=, you should first try to fix the errorwithout regard to the line number. If you still cannot findthe error, then look at the specified line in the TPL file.

An attempt of a problem solution with some physicalparameters that are far outside their usual values couldcause a lot of error messages. =EMPOWER= can handlea very large range of physical parameters, thus it usuallyhappens only when the problem is not properly described.First, check all specified data thoroughly and the modelrepresentation of the problem (mapping on the grid).Print out all available information about the problem inthe listing file using options -CH and -LA and proof it. Ifproblem is found, delete all auxiliary files and rerun thesimulation. Otherwise, try to figure it out what parametercaused the failure.

Another common source of induced errors is the de-embed-ding parameters. Keep track of them, especially the char-acteristic impedances of the eigenwaves (normalizationcoefficients). They must be real values for propagatingwaves and do not usually change much in a wide frequencyrange. If they are not real or change drastically across thefrequency range, then something is wrong. Check the boxgeometry, the extracted line segments’ geometry and theposition of the line inputs on the grid. If everything is OK,turn off de-embedding or use the -O option to change thebox size for line analysis.

Note: The =EMPOWER= Error Messages have beenmoved to the reference manual.

230 =EMPOWER= Messages

INDEX

!2 1/2-D simulators 232-D simulators 233-D simulators 23

AAdd More Frequencies 15Air, Above and Below 26ASCII text files 111

BBatch files 112

Calling other batch files 113Bibliographt 221Binary files 211Box Height 32Box modes

Absorber material 105Dielectric loading 103Higher orders 102Introduction 101 - 105Metal loading 104Top cover 104Wave number 101

Box settings 4Box Width 32

CCavity modes 101Center selected on page 10Command line options

Debug 125Deembedding 121General 111, 115Listing file 123Processing algorithms 119TPL file 111Viewer Data File 122

Console mode 109 - 126ASCII text files 111Batch files 112

Multiple sequential runs 112Contacting Eagleware viiCover, Top and Bottom 26

Electrical Description 26From =SCHEMAX= metal 26Lossless 26Magnetic Wall 26Physical Description 26Semi-Infinite Waveguide 26

DDecomposition

Basics 58First step 59Losses 65Lumped elements 66Mode Setup 61Order of ports 66Port Numbering 65Spiral 58Typical circuits 58

Default Viahole Layers 32Discontinuity analysis mode 135Dominant box modes 102

EEagleware viiEM Port

Current Direction 48Deembedding 49Draw Size 46General info 45 - 56Layer 48Line Direction 48Location 48Mode Setup dialog 53Multimode 52Port Number 47Properties dialog box 46Reference Plane Shift 46Type of Port 49

Width & Length 47EMports 34Error messages 229Example files

BEND1.TPL 185BEND2.TPL 185BNDCHM.TPL 185COMBINE.WSP 62FULL.WSP 60LayoutOnly.WSP 2LINE1.TPL 178LINE2.TPL 178LNMIT3.WSP 93METR16.WSP 88PART1.WSP 60PART2.WSP 61

External portsIntroduction 45Placing 45

FFeature Overview xFile extentions 212File formats 209 - 220

.BAT 212

.EMV 213

.L1, .L2,.. 213

.LST 214

.PLX 217

.R1, .R2... 218

.RGF 218

.RX 219

.SS 219

.TPL 220

.WSP 219

.Y 220~xx (backup files) 220

File types table 211Finline 115

GGeneral Information viiGeneralized S-parameters 55Geometry, basic 24Grid 28Grid settings 4Grid Spacing 31Grid, =empower= Style 31

HHigher order modes 102Historical background ix

IInternal Ports

Adding lumped elements 71Automatic placement 72Basics 69Directed port rules 76Numbering ports 71Placement and setup 69Series port 75X- and Y-directed 74z-directed ports 70

LLayers, =empower= 7Layers, air 7Layers, general 5Line analysis mode 134Lumped Elements 20

MMap of terminals 215Mapping on grid 30Max box size 215Messages 229 - 230Metal Layers 27Min. media wavelength 215Mode Setup 53Multimode Ports 52Multiple sequential runs 112

NNew =empower= run 11

OOpen Schematic File 17

PPorts, placing 10Program

Feature Overview x

QQCHK 215

232 Index

RReference Plane 46References 221 - 227Resonance, advantage of avoiding 76Run Viewer 15

SSDTC 215Sequential runs 112Setup Dialog Box 34Slotline 115Stairstep 156Starting

With a schematic 1Without a schematic 1

Substrate/Media Layers 27

TTechnical Support viiTerminal map 215Thinning thresholds 215Tips

Cell size 36Cover type 39Lossy analysis 40Maximum critical frequency 36Preferred box cell count 42Slot-type structures 41Spacing, wall and cover 39Symmetry 37Thinning 38Viewer data 40

TPL file format 127 - 190CIRCUIT block 133Delta 142DIM block 130EMFRQ 139EQUATE block 132Example file 128Example, discontinuity analysis 185Geometry 152Grid 153Lossless 153, 168MAXFRQ 140Media/Layer 149Memory 183Overview 128Package 137

Pad 153, 164Polygon 153, 175Port 153, 157Rect 153, 171Simple example 178Size 144Surface 153, 168Terminal map 183TMET 169Tolerance 141Top_W/Bottom_W 146

Tuning 20

VViahole Layers, Default 32Viaholes 33Viewer

Animation 91Buttons listed 82Display controls listed 79EMport to excite 97Generate Viewer Data 97Introduction 77 - 100Magnitude/Real/Img 90Multimode 93Normalizing impedance 98Pan controls listed 80Phase 91Plots 87Print screen or window 78Starting 77Thinning distortion 96Time averaged 95Toggle Background Color 78, 89Toggle Colors 89Toggle controls listed 81Toggle Origin coordinatesBackground

Color 89

WWave number 101

ZZ-directed ports 33

Index 233