presentation on momentum optimization -...

This document is owned by Agilent Technologies, but is no longer kept current and may contain obsolete or

inaccurate references. We regret any inconvenience this may cause. For the latest information on Agilent’s

line of EEsof electronic design automation (EDA) products and services, please go to:

www.agilent.com/fi nd/eesof

Agilent EEsof EDA

nstewart

Text Box

Presentation on Momentum Optimization

EM Optimization - Moving from Analysis to Design Automation 1

Momentum OptimizationEM Seminar 1999

MomentumOptimization

EM Seminar - HP Momentum Paper 1

Moving from Analysis to Design Automation

EM Optimization - Moving from Analysis to Design Automation


page 2 of 36Momentum OptimizationEM Seminar 1999

Objective

• To illustrate circuit design flowwith HP ADS

• To illustrate the use ofMomentum and MomentumOptimization as part of thedesign flow

This presentation will illustrate the design of a 3.2 GHz radial stub filter. Themicrostrip filter focuses on the use of EM simulation as a verification tool.



Overview

Microstrip Radial Stub Filter

• Analysis - Momentum

• Design Refinement - Momentum Optimization

Analysis

ModelGeneration

DesignRefinement

Momentum

Momentum Optimization

This presentation will illustrate the design of a 3.2 GHz radial stub filter. Themicrostrip filter focuses on the use of EM simulation as a verification tool.



Design Flow

Manufacturing

EM Simulation

Layout

Optimization

Synthesis

System Design

Circuit Design

EM Optimization

DESIGN

FEEDBACK

Post Processing

Sche

mat

i c-b

ased

Des

ign

Lay

out-

b ase

dD

esig

n

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

This diagram illustrates a typical design flow. The design flow is broadly separatedinto two parts: schematic-based design and layout-based design. Also represented isthe design feedback path which goes throughout the entire design process.Schematic-based DesignThe schematic-based design starts with system design, where specifications are setand system architecture is studied to understand design tradeoffs. Once the systemarchitecture is set, the specifications of each sub-module is set.

After the system design and prior to circuit design, there is an initial investigationinto the design approach. This includes literature and textbook study, as well as theuse of specialized synthesis tools. The synthesis tools provide a starting point forthe circuit design.

Circuit Design involves the details of going from the design specification to thefinal schematic-based design. This process involves the circuit simulators (linearsimulation, harmonic balance, transient, convolution, envelope,…), as well asdesign tools such as circuit optimization and post-processing of the analysis data.



Schematic-based Design

Project Manager says “… a filter isneeded to reduce the mixer LOfeed-through to the amplifiers”.

Proposed Solution:Design a MICROSTRIP radial stublow pass filterE

M S

imu

lati

on

as

an A

nal

ysis

To

ol

Some motivation starts the design process.In this case, a filter is needed to reduce the mixer LO feed-through to the amplifiersthat follow.

The proposed solution is to design a microstrip radial stub low pass filter.



• Low Pass Filter• 1 dB corner freq = 3.2GHz• > 25dB attenuation from3.9-6.0 GHz

System Design

Manufacturing

EM Simulation

Layout

Circuit Design

System Design

SynthesisLO

IF

RF 25 dB

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

From the system analysis, the low pass filter is determined to need a 1 dB cornerfrequency at 3.2 GHz, with at least 25 dB of attenuation in the 3.9-6.0 GHz.

The system design portion of this project is done on the order of hours to days.



Synthesis

• 7th order Chebychev• fc = 3.2 GHz• passband ripple 0.5 dB

Manufacturing

EM Simulation

Layout

Circuit Design

System Design

Synthesis

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

Design Synthesis for the filter is done with E-Syn to determine the starting design.It is determined that a 7th order Chebychev filter with fc = 3.2 GHz and passbandripple of 0.5 dB will work. The ideal analysis gives slightly less than 25 dB ofrejection at 3.9 GHz, but the less desirable alternative was to go to a 9th order filter.With the actual implementation in microstrip, the component values can be slightlymodified to achieve the goal.

The synthesis part of the design process can be done on the order of hours to days,depending on the amount of research needed, and the ability of a standard design tomeet the design goals.



Circuit Design

C1 = 1.728 pFL1 = 3.129 nH

C3 = 2.624 pFL3 = 3.343 nH

C4 = 2.624 pF C2 = 1.728 pFL2 = 3.129 nH

Manufacturing

EM Simulation

Layout

Circuit Design

System Design

Synthesis

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

The ideal circuit is implemented in the schematic. This serves as a baselinecomparison for the electrical performance.

The goal for the circuit design process is to convert this ideal circuit into a physicalimplementation using microstrip components.



Convert Ideal to Physical

Microstrip ButterflyRadial StubW = 40 milRo = 218.7 milAngle = 60D = 15 mil

Microstrip ButterflyRadial StubW = 40 milRo = 276.6 milAngle = 45D = 15 mil

Radial Stub 1 - C1,C2Capacitance is 1.727pF

Radial Stub 2 - C3,C4Capacitance is 2.623pF

H = 59 mil Er = 4.3

H = open Er = 1.0

Substrate - GETEK

Method 1 - Iterative Analysis / Post Process

3.7 - 4.3 manufacturerstolerance

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

The capacitors of the filter will be implemented as microstrip butterfly radial stubs.The parameters for W and D are fixed, and through a manual process of binarysearch, the angle and radius are determined. The capacitor value is determinedthrough post processing the S-parameters (illustrated on the next slide). Thedimensions determined for the radial stub are as follows:C1= C2 = 1.727 pF { W = 40 mil, Ro = 218.7 mil, angle = 60, D = 15 mil }C3 = C4 = 2.623 pF { W = 40 mil, Ro = 276.6 mil, angle = 45, D = 15 mil }

The substrate The substrate used is 59 mil thick GETEK. Due to manufacturing tolerance, thedielectric constant is held to a range between 3.7 and 4.3.Understanding the manufacturing tolerance is important for design. A company, aspart of the design process, usually has characterized the material and the processtolerance and can provide this information to designers to improve their designaccuracy.At this point the circuit is being prototyped by an outside vendor. For this examplethe material is considered to be an unknown; so the design is based on themanufacturing specification for the material.

For the purpose of this presentation to illustrate the design process, the dielectric of4.3 is chosen.



Convert Ideal to Physical -Capacitance Calculation

Radial Stub 1 - C1,C2

Capacitance is 1.727pF

Radial Stub 2 - C3,C4

Capacitance is 2.623pF

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

Method 1 - Iterative Analysis / Post Process

The S-parameters from the analysis on the previous slide are converted intocapacitance. The S-parameter analysis is done at the 1 dB cutoff frequency 3.2 GHz.Since microstrip is dispersive, the capacitor value determined here will appear tohave different values of capacitance at other frequencies. Since it is impossiblewith microstrip to get the same capacitor values at all frequencies, the cutofffrequency is chosen as the design frequency.

Since this is a one-port S-parameter analysis, S11 can be converted into Zin with theequation:Zin = 50*[(1 + S11) / (1 - S11)]

Zin can then be converted into an equivalent capacitance with the equation:Cin = -1 / (2*pi*freq*imag(Zin))



Convert Ideal to Physical -Capacitor

The same values for the radial stub can befound through Circuit Optimization

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l Method 2 - Circuit component optimization

Circuit optimization can be used as an alternative to the binary search procedureused in the previous slides. For this circuit, the radial stub (which is set up foroptimization) is connected to S-parameter Port 1, and the ideal capacitor isconnected to S-parameter Port 2. Two goals are set up to minimize the differencebetween S11 and S22 for the real and imaginary terms.Goals:-0.001 < imag(S11) - imag(S22) < 0.001-0.001 < real(S11) - real(S22) < 0.001



Convert Ideal to Physical -Inductor

MicrostripTransmission LineW = 40 milL = 215 mil

Inductance is 3.129 nH

MicrostripTransmission LineW = 40 milL = 225.7 mil

Inductance is 3.342 nH

Inductor 1 - L1,L2

Inductor 2 - L3

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

In a similar way, a high impedance microstrip line is used for the inductor. Binarysearch or optimization can be used to determine the dimensions. The dimensionsdetermined for the inductors are as follows:L1= L3 = 3.129 nH { W = 40 mil, L = 215 mil }L2 = 3.342 nH { W = 40 mil, L = 225.7 mil }



Schematic with PhysicalComponents

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

Replace ideal components with radial stubs andinductive transmission lines

The major components of the circuit have been individually calculated and are nowassembled into the complete circuit. A 50 ohm microstrip transmission line andtaper have been added to each end of the filter, and the overall length is adjusted to1.5 inches. An S-parameter simulation is done of the complete circuit.



Ideal vs. Physical Comparison

Ideal

Physical

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

Difference due to lumped versus distributed model

dB(S21) and dB(S11) are shown for the comparison. Each plot shows the idealcircuit response (LC filter) and the microstrip circuit (physical) response. Theresponse at the 1 dB cutoff frequency is very close to the ideal. The response at 3.9GHz achieves the < -25 dB specification. The obvious difference is in the stop bandloss detail above 3.9 GHz.



Layout

Manufacturing

EM Simulation

Layout

Circuit Design

System Design

Synthesis

H = 59 mil Er = 4.3

H = open Er = 1.0

Substrate - GETEK

Synchronize layout from Schematic

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

The design flow now crosses the design flow boundary between schematic-baseddesign and layout-based design. A layout of the schematic is produced. At thispoint, there is design feedback to determine if the dimensions of the radial stubs andtransmission lines cause the geometry to overlap. It can also be observed if thegeometry would require EM simulation to determine the effect of parasiticcoupling.



EM Simulation

• HP Momentum - planar method of moments

• HP HFSS - 3D finite elements

HP Momentum

EM SimulationHP HFSS

Layout

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

At this point, the designer determines that EM analysis is required. Two choices areavailable: a planar EM solver or a 3D solver. HP offers Momentum as a planarmethod of moments solver, and HFSS as a 3D finite elements solver.HP Momentum is an integrated product which works directly from the Layout ofHP ADS. HP HFSS is a separate product which has translators to directly readlayout from HP ADS. If the layout is part of a solved Momentum project, then thelayout and substrate information are used to form a complete HP HFSS project. Ifthe layout is only in EGS format, then the translator provides the ability to definethe substrate definition and layer mapping.



EM Simulation - HP Momentum

• Mesh frequency 3.2 GHz• Edge mesh on• 30 cells per wavelength

Manufacturing

EM Simulation

Layout

Circuit Design

System Design

Synthesis

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

HP Momentum works directly from the layout of HP ADS. The mesh frequency isset at 3.2 GHz. Edge mesh is enabled and the mesh is set to 30 cells per wavelength.The results of the HP Momentum simulation are written out directly in the datasetformat of HP ADS, and viewed using the data display of HP ADS.



Translate to 3D EM SimulatorHP ADS

HP HFSS

Add 3D features:

•3D structures, e.g.connectors, finite dielectrics,housing features

•metal thickness

HP ADS - HP HFSSTranslator

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

Designs can be translated from HP ADS to HP HFSS. 3D features can be added tothe geometry, such as connectors, finite dielectrics, housing features, and metalthickness. HP HFSS can write out data in CITIfile or Touchstone format, whichcan be read back into HP ADS.



Circuit versus EM comparison

Momentum

Physical

Ideal

Differences due toAdded Parasitics inMomentum

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

The results of the EM simulation can be compared to the ideal results and themicrostrip schematic simulation results. Momentum shows a large deviation in thepredicted stop band performance of the filter. This is due, mainly, to the additionalparasitics present in the actual geometry and not accounted for in the schematicrepresentation.



Manufacturing

Manufacturing

EM Simulation

Layout

Circuit Design

System Design

Synthesis

Layout translator(Gerber) used to linkto Manufacturing.

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

Additional geometries needed for manufacturing are added to the layout. Layouttranslators, such as Gerber, GDS-II, IGES, or DXF are available to link tomanufacturing. The completed prototype circuit is shown.The prototype process time for this design was on the order of 3-4 weeks.



Design Flow

Manufacturing

EM Simulation

Layout

Optimization

Synthesis

System Design

Circuit Design

EM Optimization

DESIGN

FEEDBACK

Post Processing

Sche

mat

i c-b

ased

Des

ign

Lay

out-

b ase

dD

esig

n

Learning

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

An important aspect of the design flow is the design feedback process. As tests andprototypes are completed, it is important to incorporate the new learning into thedesign. Feedback can affect all stages of the design flow, from the em simulationup through system design.



Measured vs. Modeled Results

MeasuredHP Momentum

Ideal

•Does not meetspecification

•Need to account for thedifferences

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

The prototype is measured and compared to the Momentum simulation. In thisexample, there is a significant difference in the filter response from the predicted.



Measured vs. ModeledResolution

Check manufacturing

• Actual Geometry vs. Ideal Geometry

•Actual Substrate vs. Ideal Substrate

• EM Optimization

Modify Design

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

Search for causes of discrepancy

Since this is an unknown substrate and manufacturing process, there are severalpossible causes for the differences in the simulation and measured results. It isimportant to check the manufacturing process for the dimensions of the actualgeometry produced as well as the actual substrate properties. If both of these are inagreement with the simulation assumptions, then the designer can look intomodifying the design.



Design Feedback - Geometry

Geometry is from 0.5 milto 1 mil over-etched in

different areas

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

The first design feedback is to check the geometry. Using a machinist microscopeto measure the circuit, it is observed that the etching process has over-etched thecircuit by varying amounts. The over-etched circuit is from 0.5 mil to 1 mildifferent than the intended design. This can contribute to the difference in thefrequency response.



Design Feedback - Dielectric

Dielectric Measurement - HP 85070B Dielectric Probe

The measured dielectricconstant:

2.7 GHz - 3.75

3.2 GHz - 3.8

3.4 GHz - 3.85

Different fromer= 4.3 used insimulation

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

Using the HP 85070B Dielectric Probe, a sample of the board was tested fordielectric constant. The measured dielectric value was approximately 3.8, and not4.3 that was used in the design assumption. The dielectric constant is also seen tovary with frequency. In the range from 2.7 GHz to 3.4 GHz, the dielectric changesby 0.1 .



HP Momentum HP Momentum

Measured

HP HFSS

Measured vs. New Modeled Results

•Dielectric constant= 3.8

•Modified Geometry

EM

Sim

ula

tio

n a

s an

An

alys

is T

oo

l

(thick metal)

•Have determined the processparameters

•Have verified Momentum’sability to analyze performance

•Need to change design tomeet specification

goal

The layout geometry is modified based on the microscope measurements, and thenew dielectric constant is used from the dielectric probe measurement. Theresulting Momentum analysis now compares favorably with the measured results.

This filter design does not meet the original design specification, but the learningthat resulted from the design feedback will help with the further design iterations.The design process is a series of design and re-design steps that are used to gain anunderstanding of the circuit and of the manufacturing considerations.

Simulation time:Momentum - 4.5 min/freq 10 frequencies, total time 45 minutes (23 MB)HP HFSS - 11 min/freq 6 frequencies, total time 70 minutes, thick metal (560 MB)



Overview

Microstrip Radial Stub Filter

• Analysis - Momentum

• Design Refinement - Momentum Optimization

Analysis

ModelGeneration

DesignRefinement

Momentum


EM

Op

tim

izat

ion

as

a D

esig

n T

oo

l

The existing design can now be used as a starting point for design refinement.Momentum Optimization automates EM simulation and controls the geometricparameters to improve the circuit performance toward the design goals. This filterwill be used to illustrate the Momentum Optimization process.



Define Parameters

Optimize L1 (L3 not shown) Optimize L2

Optimize C3 (C4 not shown)Optimize C1 (C2 not shown)

L2L1

C3C1

ind1 ind2

rad2 rad1

EM

Op

tim

izat

ion

as

a D

esig

n T

oo

l

The parameters for this optimization are illustrated. Each graphic shows andoverlay of the nominal geometry with the perturbed geometry. Each of theseperturbed geometries exists in their own separate layout file. The four parametersfor optimization are:•The length of inductor L1 (L3 is changed at the same time as L1)•The length of inductor L2•The radius of the radial stub C1 (C2 is changed at the same time at C1)•The radius of the radial stub C3 (C4 is changed at the same time at C3)

The two rules to remember for defining an optimization parameter are:•There must be no change in the number of vertices.•The defined parameter represents a linear translation of the vertices



The Solution Process - Parameters

Parameter for Radius is defined:

•Allow to vary during optimization

•Lower and Upper Bound for parameter

•‘Add’ makes a copy of the nominal design

EM

Op

tim

izat

ion

as

a D

esig

n T

oo

l

The Momentum Optimization process is started from the Momentum menu. Thegeneral process to start an optimization is the following:•Define candidate parameters for optimization•Specify the design goals•Setup and run the optimization

The parameters for this example are the radius of the radial stubs, and the length ofthe inductive transmission lines. From the parameters dialog, define the following:

•variable name•nominal value•perturbed value•starting value.

With the advanced options, the lower and upper limit can be set for the variable.When ‘Add’ is selected, a copy of the nominal design is automatically made todefine the perturbed design.



Momentum Optimization -Geometry Capture

• Define Nominal Design

• Define a Perturbed Design

• Move the affected verticeson the Perturbed Design todefine the parameter foroptimization

EM

Op

tim

izat

ion

as

a D

esig

n T

oo

l

Momentum Optimization uses a process called geometry capture to specify acandidate parameter for optimization.

Geometry capture can be defined by the following process:• Define the nominal design - the geometry of interest will serve as a start• Define the perturbed design - simply define the parameter name to add, andMomentum makes a copy of the nominal design in a new layout window.• Move the affected vertices on the perturbed design to define the parameter foroptimization. The difference between the nominal design and the perturbed designwill define the parameter.



The Solution Process - Specification

Specification goals for theoptimization:

•frequency point or sweep

•Goal and Weight

EM

Op

tim

izat

ion

as

a D

esig

n T

oo

l

The next step in the Momentum Optimization process is to define the specificationgoals. For this radial stub filter example, we need to express the two design goalsof interest:dB(S21) at 3.2 GHz = -1dBdB(S21) at 3.9 GHz = -25 dB

Other frequencies (or ranges of frequencies) as well as other inequalities could bespecified in the above goals were not sufficient to achieve the needed performance.



The Solution Process - Run

Optimization setup:

•Optimization type

•Interpolation type

•Stopping criteria

EM

Op

tim

izat

ion

as

a D

esig

n T

oo

l

The last step for Momentum Optimization is to define the type of optimization andstopping criteria. Then select the ‘start’ button.

Momentum Optimization starts the optimization process, and displays a windowthat shows the error function result of each iteration.



The Solution Process - Results

optimal values displayed•increase rad1 by 17.51•increase rad2 by 4.26•increase ind1 by 5•increase ind2 by 5

EM

Op

tim

izat

ion

as

a D

esig

n T

oo

l

Before and afteroptimization

After the stopping criteria is met, the optimization parameter dialog is updated toshow the optimal value.

Select ‘back annotate optimal values’ to make the starting values the same as theoptimal values. Then select ‘view start design’ to open a layout window with thegeometry that represents the new starting values.

This process can be done for each geometry and each parameter of interest. EMoptimization could also be performed for the width of the high impedancetransmission lines (the inductors), or for the angle parameter of the radial stubs.



EM Analysis of entire structure

Goals:1dB loss at 3.2 GHz> 25 dB loss 3.9-6.0 GHz

IdealMomentumafter opt

EM

Op

tim

izat

ion

as

a D

esig

n T

oo

l

Momentumbefore opt

goal

The simulation of the optimal filter shows good agreement with the design goal.



Design Flow - Time Optimization

Manufacturing

EM Simulation

Layout

Optimization

Synthesis

System Design

Circuit Design

EM Optimization

DESIGN

FEEDBACK

Post ProcessingSc

hem

ati c

-bas

edD

esig

nL

ayou

t-b a

sed

Des

ign

~ 2-4 weeks

~ 1-2 days

~ 1-2 hrs

~ 1 day

~ 1-2 days

EM optimization has been shown to be a vital part of the complete design flow.



Summary

• EM Analysis is a vital part of the design flow

• Improve Time To Market

• Integrated with HP Advanced Design System

•With the addition of Momentum Optimization,Momentum moves from being an analysis tool to adesign tool.


Momentum

SummaryMomentum- EM Analysis is a vital part of the design flow - Momentum simulation was usedfor design verification, and served as the basis for uncovering the incorrect designassumptions. Once the correct assumptions were used for the Momentumsimulation, there was good agreement between measured and modeledperformance.- Improve Time to Market - Momentum, when used as part of the complete designflow, can help reduce the number of design iterations and thus improve time tomarket- Integrated with HP Advanced Design System - Momentum was shown to beintegrated with the complete design environment, and so saves time and effort inthe design flow.Momentum Optimization- moves Momentum from Analysis to Design Refinement - With the addition ofMomentum Optimization, Momentum is no longer simply an analysis tool: it is adesign tool.- Automation improves time to market- Integrated with Momentum

www.agilent.com/fi nd/emailupdatesGet the latest information on the products and applications you select.

www.agilent.com/fi nd/agilentdirectQuickly choose and use your test equipment solutions with confi dence.

Agilent Email Updates

Agilent Direct

www.agilent.comFor more information on Agilent Technologies’ products, applications or services, please contact your local Agilent office. The complete list is available at:www.agilent.com/fi nd/contactus

AmericasCanada (877) 894-4414 Latin America 305 269 7500United States (800) 829-4444

Asia Pacifi cAustralia 1 800 629 485China 800 810 0189Hong Kong 800 938 693India 1 800 112 929Japan 0120 (421) 345Korea 080 769 0800Malaysia 1 800 888 848Singapore 1 800 375 8100Taiwan 0800 047 866Thailand 1 800 226 008

Europe & Middle EastAustria 0820 87 44 11Belgium 32 (0) 2 404 93 40 Denmark 45 70 13 15 15Finland 358 (0) 10 855 2100France 0825 010 700* *0.125 €/minuteGermany 01805 24 6333** **0.14 €/minuteIreland 1890 924 204Israel 972-3-9288-504/544Italy 39 02 92 60 8484Netherlands 31 (0) 20 547 2111Spain 34 (91) 631 3300Sweden 0200-88 22 55Switzerland 0800 80 53 53United Kingdom 44 (0) 118 9276201Other European Countries: www.agilent.com/fi nd/contactusRevised: March 27, 2008

Product specifi cations and descriptions in this document subject to change without notice.

© Agilent Technologies, Inc. 2008

For more information about Agilent EEsof EDA, visit:

www.agilent.com/fi nd/eesof

nstewart

Text Box

Printed in USA, November 01, 2000 5989-9093EN

presentation on momentum optimization -...

Documents