Andy Howard
Applications Engineer
RF Simulation Basics
Keysight Technologies
Agenda Page 2
–Introduction
–Applications
–S-Parameter Simulations – a closer look
–Models: The building blocks for effective
simulations
–Simulation Engines
Introduction
3
Electronic Design Automation (EDA)
IDEA
PRODUCT
CONCEPT | DESIGN
Page 4
Do you need to perform RF simulations?
If you encounter any of these phrases, you probably do.
– Maxwell's Equations – Matching Networks
– Antenna – Impedance Matching
– Distributed Circuit – Smith Chart
– Physical Layer design – S-Parameters
– S21/S11
– High Frequency – ACPR/ACLR
– Radio Frequency (RF) – Noise Figure
– Microwave (μW) – Radiation
– Power Amplifier – EMI/EMC
– Load Pull – Gain
– Source Pull – Return Loss
– X Band – TOI
– Ku Band – IP3
– Wearables – P1dBc
– EVM
– IoT
Page 5
Applications
What could you achieve if
you could do RF
simulations?
6
RF System performance
Simulating your car’s FM radio
Intermodulation & Compression
30 200
24 160
18 120
DBM[GIM3P], DB[SDR]
12 80
DBM[IIP3], PRIM3
6 40
0 0
-6 DBM[IIP3] -40
DBM[GIM3P]
-12 DB[SDR] -80
PRIM3
-18 -120
-24 -160
-30 -200
1 5 6 R
L
I 16 9 10 17 R
L
I 7 8 4
Node
Page 7
Multi Chip RF Front End Modules
Powering smart phones/tablets
LTE MMIC PA LTE MMIC PA
Switch
The Complete PA /
Switch Multi-Chip Laminate Board with
Module PA / Switch IC’s
• PA / Switch IC’s
• Bond wires
• Laminate board
• Solder bumps
• PCB test board
• Connectors
Page
Electro-Thermal Simulation
Dynamic thermal behavior
Input Power (Watts)
HBT Thermal
Power Memory
Amplifier
IC
Time scale
Ref
Amp
Dev
Dev
(Bias)
(Large)
“The Impact of Electro-Thermal Coupling on HBT Power Amplifiers”,
Matt Ozalas, Keysight, CSIC 2014 Page
Electromagnetic Analysis
Signal Propagation in arbitrary structures
Page 10
Electronic Warfare / Aerospace
3D Scenario Modeling
– Model multiple moving TX and RX platforms
including clutter and environment
Phased Array Degradations
– Model beamforming & jamming performance at the system level
Satcom Dynamic Channel Modeling
DVB-S2, QAM,
APSK, etc. AWGN
PA channel RX BER/FER/PER
TX Throughput
Atmospheric,
Terrain, Doppler,
Delay
RF/MW CIRCUITS
“software fading”
MEASURED
X,S-PARAMS
Page
DDR4 Memory
Signal Integrity and Power Integrity
Sync VRM
VCC1V2_FPGA Power Net
on Layer 9 and 10
Kintex
FPGA
GND Net
50x scaled
Page
12
S-Parameter Simulations
A closer look
13
S Parameter Simulations
You already learned about S Parameters. But how does one simulate S- Parameters
in an EDA tool?
a1 a2
|S|
b1 b2
Should we employ matrix math to
cascade S-Parameters?
Page 14
S-Parameter Simulation
– Matrix math runs out of steam if you have bigger complicated designs.
– Modern Simulators actually behave similar to Network analyzers. Excite
one “end” of the circuit (at certain frequencies) and measure incident
and reflected waves at the same “end” and also at other “ends”.
– Then compute ratios and get S-Parameters.
– For this to work, the EDA tool needs to know how each device
(transmission line, capacitor, transistor, etc..) reacts to RF excitation.
– EDA tools contain built-in models to represent the most common
devices. (More on device models later)
– And you can supplement the built-in models with measurement based
models too.
Page 15
S-Parameter Simulation
Page 16
S-Parameter Simulation
Port Count is not limited
No Calibration needed
Impedance can be any value
Page 17
Simulation Results
Page 18
Tune & Optimize– What if scenarios
Page 19
Models
The building blocks for
effective simulations
20
Models
– Previously we saw that “models” are required to represent
transmission lines, transistors, capacitors etc.
– Devices that are linear in nature (resistors, transmission lines) have
linear models.
– Devices that are non linear in nature (diodes, transistors) have both
linear and non-linear models.
– Models can be ideal or more realistic
Page 21
Models
– Models are typically based on important works of research. For e.g.
the MLIN model:
• W. J. Getsinger, "Measurement and Modeling of the Apparent Characteristic Impedance of Microstrip," MTT-
31, August 1983.
• E. Hammerstad and O. Jensen, "Accurate Models for Microstrip Computer-aided Design," MTT Symposium
Digest , 1980.
– The research is translated into equations and then implemented in
the EDA tool.
– Models have a working “range”. For e.g. MLIN model is not
expected to work at 1 THz.
Page 22
Models
Measured or Simulated S-Parameters
Modeling Linear Behavior Using S-parameters
Page 23
Models
Nonlinear models for transistors
– Non linear models are based on complex differential equations.
Research groups (e.g. BSIM group at University of California,
Berkeley) develop these models and share with the industry.
– By feeding the right parameter values into the model “card” you can
match the software model to a real world device.
– The parameter values are typically obtained by measurements and
“fitting” algorithms.
– Typically takes several PhDs to do this right.
Page 24
https://en.wikipedia.org/wiki/BSIM
What are X-Parameters?
– X-parameters are the mathematically correct superset of S-
parameters, applicable to both large-signal and small-signal
conditions, for linear and nonlinear components. The math exists!
– We can measure, model, & simulate with X-parameters
– Each part of the puzzle has been created
– The pieces now fit together seamlessly
NVNA: Measure X-parameters PHD: X-parameter block ADS: Simulate X-parameters
Interoperable Nonlinear Measurement, Modeling & Simulation with X-parameters
“X-parameters have the potential to do for characterization, modeling, and design of nonlinear
components and systems what linear S-parameters do for linear components & systems”
Page 25
Simulation Engines
Analyze your RF design
Simulation Types
– Frequency vs Time
– Linear vs Non-Linear
– Circuit vs System
– Circuit vs EM
– Multi-Physics (Electrical Circuit + Thermal)
Page 27
Simulation Engines
• DC • EM Simulation
• AC MoM
• S-Parameter FEM
• Harmonic Balance FDTD
• Transient (High Frequency Spice) • Spectral
Channel Sim • Data Flow
• Circuit Envelope • ElectroThermal
Page 28
Harmonic Balance Setup
Page 29
Harmonic Balance Results
Page 30
Harmonic Balance Load Pull Typical Results
Page 31
Transient Analysis
Just like SPICE, only better
– Kirchoff’s current equations are derived at each node in differential
form
v(t ) dv(t )
C 0 (1)
v(t)
R dt
– The time derivatives are replaced with discrete-time approximations
(integration)
– The solution, in the case of a complex circuit, will consist of a
system of nonlinear equations which is solved using the Newton-
Raphson method
Page 32
Transient Simulation
Page 33
Circuit Envelope
Beyond CW and onto Modulated signals.
– Time samples the modulation envelope (not carrier)
– Compute the spectrum at each time sample
– Output a time-varying spectrum
– Use equations on the data
– Faster than HB or Spice in many cases
– Integrates with System Simulations & Keysight’s Ptolemy
Page 34
Circuit Envelope
Test Circuits with Realistic Signals Example CE results:
LTE, LTE-A, 802.11ac, BTLE, 5G
LTE DL
Simulations can include:
– Adjacent Channel Power Ratio
2 GHz carrier
– Noise Power Ratio
– Error Vector Magnitude
– Power Added Efficiency
– Bit Error Rate
2-tone tests and linearized models
do not predict this behavior as easily! Page 35
Electromagnetic
Method of Moments solver
• No good Circuit model exists
• Physical dimensions and properties
are known.
• The entire geometry is “meshed”
• Results are S Parameters
• Design can be optimized
Page 36
Summary
Start with your great idea, choose your EDA environment, use appropriate models to construct the
designs. Run simulations to verify performance and build great RF applications.
Page 37
Learn more about RF Simulations
– How to Design an RF Power Amplifier and other How-to videos:
https://www.youtube.com/playlist?list=PLtq84kH8xZ9HIYgBYDsP7TbqBpftidzI8
– For More Information www.keysight.com/find/eesof-ads-info
– ADS on www.keysight.com/find/eesof-ads-videos
– Genesys on
https://www.youtube.com/playlist?list=PLtq84kH8xZ9E8S_y5dmCXtJFPo14NsCtt
Page 38
