Advances in EM Modeling

for Circuit Designers

Agilent Technologies
EDA EEsof Division RF System
Architecture

Presented by: Rick Carter Time-Domain

11 Different
Circuit Synthesis
Nonlinear Modules
June 17, 2008
Agilent
Genesys

Freq-Domain Planar 3D
Nonlinear EM Simulation

Advances in EM Modeling for Circuit Designers

June 17th 2008
1
 Abstract

RF Designers today, need an EDA toolset that helps them cope with the
challenges of high frequency, highly integrated and highly outsourced
design work. Today’s large, complex RF circuits require industry-proven
electromagnetic (EM) simulators that are easy to use, have first-pass
accuracy and fast TRUE circuit co-simulation. Today’s economic
environment requires that they also be affordable. To meet these
challenges, Agilent has integrated the planar 3D EM tool from ADS -
Momentum, into the easy to use, affordable Genesys platform. Agilent
has also introduced EMDS, an affordable full 3D EM simulator.

Advances in EM Modeling for Circuit Designers

June 17th 2008
2
 RF Board Design Issues & Realities

– Passive Material parameters: uncertain baseline, poorly defined

variation
• The most popular board materials for electronics such as FR4 often
show high variation in dielectric constant (Er or Dk), Loss Tangent
(TAND or Df), and even thickness across a sample.
• Ideal rather than “RF” component models used for lumped elements
• Parts show variation in manufacturing
– Physical environment:
• Enclosure effects not considered (resonances or radiation)
• Footprints and other metalization and coupling may not be considered
• For antenna design, EM is a must Connectors and other 3D elements
hard to include in simulation
– Realities
• EM simulation is required, but including discrete parts is cumbersome,
time consuming and error prone

Advances in EM Modeling for Circuit Designers

June 17th 2008
3
 Design & Verification of Circuit Components

The effect of the layout pads on

IF filter response
– Pads add capacitance at each node
– Coupling between pads
EM Co-simulation must be used to
capture these effects
80 MHz SHIFT, REDUCED BANDWIDTH

Advances in EM Modeling for Circuit Designers

June 17th 2008
4
 Typical EM-Circuit Co-simulation – 7 steps
How to include discrete elements?
1. Generate Schematic
1 2 2. Generate Layout
3. Place internal ports on the layout (Very time
consuming and error prone).
4. Run EM, Generate EM S-parameter file
5. Create a separate circuit to combine EM File and
1
parts (again, very time consuming, error prone)
6. Run Circuit Simulation in same tool, if available
or in separate tool
7. View Combined Response
7
3 4 5
6
RUN
EM
CIRCUIT
S-Data
EM SIM
S-Data

Advances in EM Modeling for Circuit Designers

June 17th 2008
5
 RF Board Design Solutions

To simplify the design process, Agilent has introduced:

MomentumGX

– MOM based 3D planar EM simulator (2D with Z directed

currents) inside easy-to-use, affordable Agilent Genesys
– TRUE Co-Simulation with Genesys circuit simulators
– Proven Accuracy - Same engine used 10 years in ADS
– Fast & Easy to Use
• Conformal meshing
• Adaptive Frequency Sweeps
• No Box requirements
• Optimization and Monte Carlo

Advances in EM Modeling for Circuit Designers

June 17th 2008
6
 What is GENESYS?

A modern, integrated Windows environment

Easy-to-use - “Hard to forget”
Layouts Equations

Tune Vendor
Window Models

Workspace
Tree

Graphs
Schematics & Tables

Advances in EM Modeling for Circuit Designers

June 17th 2008
7
 Genesys Integrated

• Linear Simulation, Schematic, Layout

• DC Simulation
• Harmonic Balance non-linear
• Transient non-linear
• System Architecture
• 11 Synthesis modules
All tightly integrated with MomentumGX

Advances in EM Modeling for Circuit Designers

June 17th 2008
8
 MomentumGX TRUE Co-simulation
1 2
VH9
R= 0 .2 m m [Vi a _ rad]
T=0 .0 3 6 m m [Vi a _thk]

TL 41
W= 3 5 m i l [wg nd]
L = 4 0 m i l [L gnd]
VS1 {VDC}
VDC=5 V
1. Generate Schematic
TL 44

2. Generate or import Layout, ports are

W=3 5 m i l [w50]
L =1 0 0 mil

VH8
R=0 .2 m m [Vi a _ rad] VH1 0
T= 0 .0 3 6 m m [Vi a _thk] R=0 .2 m m [Vi a _ rad]
T=0 .0 3 6 m m [Vi a _thk]
1

TE4
2

TL 40 TL 42 C1 T L 45
W=3 5 m i l [wg nd W=3 5 m i l [wg nd] c a p =2 2μF W=3 5 m i l [w50]
L =4 0 m i l [L gnd L =4 0 m i l [L gnd] L = 1 5 mil

TL 43
W=3 5 m i l [w50]
L =2 5 mil

R5
R=3 3 0Ω

associated with footprints

CP1
IDC=0 .0 7 1A

TL 24
W=3 5 m i l [w50
TL 13
L =1 0 mil W=3 5 m i l [w50]
VH7
R= 0 .2 m m [Vi a _ rad] L =2 5 mil
T=0 .0 3 6 m m [Vi a _thk]
1

TL 39 C3 TL 16
W =3 5 m i l [wg nd] c a p = 1 0 0 0 00pF W=3 5 m i l [w50] TE7
L =4 0 m i l [L gnd] L =1 5 mil
2

VH6 T L 22 TL 25
R=0 .2 m m [Vi a _ rad] W=3 5 m i l [w50] W=3 5 m i l [w50]
T=0 .0 3 6 m m [Vi a _thk] R7
L = 2 0 mil L =1 0 mil R=1 1 5 0 Ω [Rto p b a
i s]

1 2 TL 10
R2 W=3 5 m i l [w50]
TL 38 C7 TL 14 TL 23
R= 1 0 KΩ L =2 5 mil
W=3 5 m i l [wg nd] c a p =0 .1μF W =3 5 m i l [w50] W=3 5 m i l [w50]
L =4 0 m i l [L gnd] L =2 0 mil TE8 L =1 5 mil CR1
WT=3 5 m i l [w50]
WC=3 5 m i l [w50]

3. View Co-Simulation – Linear and

TL 26
W=3 5 m i l [w50
TL 27
L =1 0 mil
W=3 5 m i l [w50] R3
L =1 0 mil R=1 8 Ω [Rd ]
TL 29
W=3 5 m i l [w50]
L =1 3 6 . 2mil

R1 TP3
R=3 .3 0 KΩ VDC=3 .7 1 9V
TL 28
W=3 5 m i l [w50]
L =2 5 mil

R4
R=5 1 Ω
TL 36
W=3 5 m i l [wg nd]
L =4 0 m i l [L gnd]

non-linear circuit and EM are

1

TE6
2

VH4
R=0 .2 m m [Vi a _ rad]
T= 0 .0 3 6 m m [Vi a _thk]

TL 20 TL 15
W=3 5 m i l [w50] W=3 5 m i l [w50]
L =3 8 .1mil L =1 3 6 mil

1
VH5
R=0 .2 m m [Vi a _ rad]
VH3
T=0 .0 3 6 m m [Vi a _thk]
R=0 .2 m m [Vi a _ rad]
1

T=0 .0 3 6 m m [Vi a _thk]

TE1
1

TE2
TL 37 c d byp T L 12

combined automatically
W=3 5 m i l [wg nd] c a p =1 0 0 p F [c dbyp] W=3 5 m i l [w50]
2

TL 35 C5 TL 21
L =4 0 m i l [L gnd] L =1 5 mil
W=3 5 m i l [wg nd] c a p = 1 0 0 0 p F [c gbyp] W=3 5 m i l [w50]
L =4 0 m i l [L gnd] L =1 0 mil

TL 19 TL 17
W=3 5 m i l [w50] W=3 5 m i l [w50]
L = 4 0 mil L =4 0 mil

Lg VH2 L g1
L =1 0 0 n H [Lg] R=0 .2 m m [Vi a _ rad] L =1 0 0 n H [Ld]
Q=1 3 .2 [Ql g] T=0 .0 3 6 m m [Vi a _thk Q=1 3 .2 [Ql d]
F re q u e n c y =1 0 0 M Hz [fqg
l] Fre q u e n c y =1 0 0 M Hz [fqd
l]
M o d e=2 M o d e=2
RDC=2 .2 Ω [ rl g ] RDC=0 .2 2 Ω [rl d]
SRF=1 0 3 0 M Hz [fg l] SRF=1 0 3 0 M Hz [fd l]
TL 32
TP2 W=3 5 m i l [ w50]
VDC=0 .5 3 9V L =5 0 m i l [L ts
l]
TL4 TL7
W=3 5 m i l [w50] W=3 5 m i l [w50]
L = 4 0 mil L =4 0 mil

Po rt_1
TE5
PH=0 d eg
F= 4 0 0 ;4 0 1 M Hz [[F1 ,F2] TP1
PAC= -1 4 .2 ;-1 4 .2 d Bm [[Pi n _ e a c h to n e ,Pi n _ eachtone] VDC=2 .6 7 6V
2 1
1 2
T L1 Ci n TL2 TL3 TL 34 TL9 Rs TL5 TL6 c o ut TL8 Po rt_2
W=3 5 m i l [w50] c a p = 1 0 0 p F [Cn
i] W=3 5 m i l [w50] W=2 5 mil W= 3 5 m i l [w50] W=3 5 m i l [w50] R=1 5 Ω [Rs o u t] W= 3 5 m i l [w50] W=3 5 m i l [w50] c a p =1 0 0 p F [Cout] W=3 5 m i l [w50] ZO=5 0 Ω
L = 2 0 0 mil L =2 0 mil CR2 L =7 5 mil L =4 0 mil TE3 L =2 0 mil L =2 0 mil L =2 0 mil L = 2 0 0 mil
SUBST=L L _ s u b s [Su b s trate] WT=3 5 m i l [w50]
WC=3 5 m i l [w50]
Q {ATF5 4 1 4 3 _ L L m o d e l @L a n c eModes l}
Su b n e t : 5 4 1 4 3 _ s 2 p _ m o d e_
l 4e
l ads
TL 18
W =3 5 m i l [w50]
L =5 0 m i l [ L ts
l]

TL 30
W=3 5 m i l [w50]

3
L = 4 1 mil
VH1
R=0 .2 m m [Vi a _ rad]
T=0 .0 3 6 m m [Vi a _thk

TL 11
W=3 5 m i l [w50]
L =4 0 mil
rfb1
R=3 3 0 Ω [rf b _ h a fl]

rfb2
R=3 3 0 Ω [rfb _ h a
TL 31
W =2 0 m i l [wfb trace]
TL 33
L = 5 0 mil
W=2 0 m i l [wfb t race]
L =1 7 5 .9mil

c fb
c a p =2 .7 p F [cfb]

Advances in EM Modeling for Circuit Designers

June 17th 2008
9
 EM Speed and Accuracy

Typical Grid-based EM solvers Momentum GX:

Suffer from “big/small” problem
Large distances and small
dimensions
Grid for smallest = large memory
Grid for largest = inaccuracy
Automatic Conformal Mesh
Accuracy and Speed for optimizations and Monte
Carlo analyses

Advances in EM Modeling for Circuit Designers

June 17th 2008
10
 RF Board Design Issues & Realities

– Realities
• EM simulation is required, but including discrete parts is
cumbersome, time consuming and error prone
• Accurate discrete models must be included with the EM
simulation, but can not be used with many EM tools

Advances in EM Modeling for Circuit Designers

June 17th 2008
11
 Design & Verification of Circuit Components Bandpass [Butterworth], Order 3
0 0

-3 -7

Genesys IF filter design:

-6 -14

-9 -21

Return Loss, Original:Return Loss (dB)

-12 -28

Gain, Original:Gain (dB)

• Substitute advanced models -15 -35

for ideal parts -18 -42

-21 -49

• Effective inductance and capacitance -24 -56

values are shifted XL-XC -27 -63

-30 -70

• Parasitic resistance affects insertion loss 437.5

Gain
450 462.5 475 487.5

Return Loss
500
Frequency (MHz)
512.5

Original:Gain
525 537.5 550 562.5

Original:Return Loss

Advances in EM Modeling for Circuit Designers

June 17th 2008
12
 Design & Verification of Circuit Components

Genesys IF filter design:

• Substituting measured s-data
components into the filter
• Vendor measured data
• Includes parasitic effects

Advances in EM Modeling for Circuit Designers

June 17th 2008
13
 Design & Verification of Circuit Components

Genesys Microwave filter design:

• Note shift between synthesized filter in
MomentumGX response
• Both center frequency and bandwidth differ
• Correction is made to filter dimensions
to compensate for offsets

Non-adjacent

Advances in EM Modeling for Circuit Designers

June 17th 2008
14
 RF Board Design Issues & Realities

Realities
• EM simulation is required, but including discrete parts is
cumbersome, time consuming and error prone
• Accurate discrete models must be included with the EM
simulation, but can not be used with many EM tools
• Optimization and Monte Carlo analysis of both discrete and
distributed parts and substrate parameters are necessary to
verify manufacturability

Advances in EM Modeling for Circuit Designers

June 17th 2008
15
 Design & Verification of Circuit Components

What else is needed to ensure manufacturability?

Accounting for device and board variations
– Monte Carlo analysis of component tolerances
– Sweeping the expectedεTanD and Thickness variation for
r
our substrate
– Specify the required tolerances for substrate material and
components

Advances in EM Modeling for Circuit Designers

June 17th 2008
16
 Design & Verification of Circuit Components

Ensuring manufacturability
9 Automated Multiple EM sweeps required

Momentum Sweep of Er Linear Monte Carlo 5% Std Dev Spacing dimensions

Er sweep 4.1- 4.9

Fo 3.3 – 3.6

Advances in EM Modeling for Circuit Designers

June 17th 2008
17
 Design & Verification of Circuit Components

Ensuring manufacturability
• Statistical Distributions
• Histograms
• Yield Analysis

Advances in EM Modeling for Circuit Designers

June 17th 2008
18
 Patch Antenna
Genesys - planar antennas can be tuned and optimized

Advances in EM Modeling for Circuit Designers

June 17th 2008
19
 Antenna Design with Momentum GX

Momentum GX now
supports producing Far
Field patterns and
Antenna characteristics
calculation

Advances in EM Modeling for Circuit Designers

June 17th 2008
20
 RF Board Design Issues & Realities

Realities
• EM simulation is required, but including discrete parts is
cumbersome, time consuming and error prone
• Accurate discrete models must be included with the EM
simulation, but can not be used with many EM tools
• Optimization and Monte Carlo analysis of both discrete and
distributed parts and substrate parameters are necessary to
verify manufacturability
• FULL 3D EM simulation is required for connectors, packages,
external environment, etc.

Advances in EM Modeling for Circuit Designers

June 17th 2008
21
 EMDS Solves Complex 3D Structures

• Standalone 3D finite-element EM
simulator
• Applications: modeling, transitions,
mechanical component design, antennas,
and more
• Used within Agilent for over 10 years;
proven accuracy
• Affordable, 1/3 the cost of other tools

Advances in EM Modeling for Circuit Designers

June 17th 2008
22
 Design & Verification of Circuit Components

Radiation pattern for single patch antenna and verification

• Gain 3.6 db
• Efficiency 81%

Advances in EM Modeling for Circuit Designers

June 17th 2008
23
 Lessons Learned

• Fast, Easy EM Co-simulation is required for accurate

simulations and manufacturability analysis
• Using ideal parts and circuit simulation alone is not
practical in many cases
• Closed-form distributed models are limited
• Statistical and swept analyses are required to ensure
manufacturability
• An accurate Full 3D EM simulation tool is indispensable
and an affordable tool is now available

Advances in EM Modeling for Circuit Designers

June 17th 2008
24

Contact Agilent at:
For more information about
Agilent EEsof EDA, visit:
Americas
www.agilent.com/find/eesof
Canada (877) 894-4414
www.agilent.com
Latin America 305 269 7500
United States (800) 829-4444
For more information on Agilent Technologies’
products, applications or services, please
Asia Pacific
contact your local Agilent office. The
Australia 1 800 629 485
complete list is available at:
China 800 810 0189
www.agilent.com/find/contactus
Hong Kong 800 938 693
India 1 800 112 929
Thank you to Lance Lascari at www.rfdude.com for
Japan 0120 (421) 345
supplying content for this paper.
Korea 080 769 0800
Malaysia 1 800 888 848
Singapore 1 800 375 8100
Taiwan 0800 047 866
Thailand 1 800 226 008
Advances in EM Modeling for Circuit Designers
25
www.agilent.com
Europe & Middle East
Austria 01 36027 71571
Belgium 32 (0) 2 404 93 40
Denmark 45 70 13 1515
Finland 358 (0) 10 855 2100
France 0825 010 700*
*0.125 €/minute
Germany 07031 464 6333
Ireland 1890 924 204
Israel 972-3-9288-504/544
Italy 39 02 92 60 8484
Netherlands 31 (0) 20 547 2111
Spain 34 (91) 631 3300
Sweden 0200-88 22 55
Switzerland 0800 80 53 53
United Kingdom 44 (0) 118 9276201
Other European Countries:
www.agilent.com/find/contactus
Product specifications and descriptions
in this document subject to change
without notice.
© Agilent Technologies, Inc. 2007
Printed in USA, January 7, 2009
5990-3478EN
June 17th 2008