Post on 21-Mar-2018
11:25 a.m.Feb. 4, 2008
5th NORTHAMERICAN
USER’S FORUM
CSTCOMPUTER SIMULATION
TECHNOLOGY
The Physical Realities of Cascading
S-Parameters for Full-Path Simulations
Heidi Barnes
8 February 2008
1
Verigy / Semiconductor Test Systems
heidi.barnes@verigy.com
1. Why do we use S-Parameters (A little bit of History)
� Phase matters
� Accurate methods of measuring them
� Easily converted to the T-Matrix for cascading
2. 3D-EM Simulations for Analyzing Cascaded S-Parameters
� When 50 ohms connected to 50 ohms is not a perfect match
� The importance of transitions
� Where to put the reference planes
Outline
2
8 February 2008
� Where to put the reference planes
3. Design for Measurement Based Modeling
� Effects of reducing the size of the interconnect
� Added Capacitance vs Added Inductance
� ATE High Speed Digital Application
Telegrapher’s Equations
Voltages and Currents are changing with Time and Distance
(Magnitude and Phase)
• Create a simple model of a
transmission line.
• Utilize calculus to analyze the model
when summing a series of incremental
length sections.
3
8 February 2008
Oliver Heaviside1850-1925
length sections.
For small R and GSinusoidal Input Resulting Relationships
2-Port Network Scattering Parameters
Measuring S-Parameters
a1 a2
bbbb1 bbbb2
Scattering
Parameters
Matrix FormRelation of Incident and Reflected Power Waves
Port 1 Port 2
4
8 February 2008
Maximum Power Transfer Theorem – Termination in Zo Results in 0 Reflections
a2=0
Port 2 Terminated in Zo
a1=0
Port 1 Terminated in Zo
Cascading Two 2-Port Networks
Cascading S-Parameters
aA1 aB2
bbbbA1bbbbB2
A
NetworkPort 1 Port 2
B
Network
Multiplying S-Parameter Matrices Doesn’t Work
ST
= SAS
B
Total Network T
5
8 February 2008
Convert to Normalized Incident and Reflected Wave T-Matrix
TT
= TAT
B
From S to T From T to S
S-Parameter Data Sets – Zo=50ohms
Low Loss, Reflections <-70dB
Coaxial Cable Small 50 mil ID
S2
1 S11
Modeled S-Parameters
E-Field Plot
6
8 February 2008
Coaxial Cable Larger 140 mil ID
S2
1 S11
Low Loss Reflections <-90dB
Modeled S-Parameters
Cascading the 2 Separate Simulations
Separate Simulations Cascaded Together
S2
1
S11
Cascaded S-Parameters are Not a Physical Reality
+
7
8 February 2008
Fields do not Match
S2
1
S11
Wrong Answer
Simulating the 2 Cables Together
Re-Simulate Together
Modeled S-Parameters
S2
1
S11
8
8 February 2008
E-Field Plot
Fields Changed by the Physical Mismatch
Physical Mismatch Increased Loss Reflections >-20dB
Optimizing a Topology Transition
Optimize the Transition Modeled S-Parameters
S2
1
S11
9
8 February 2008
Impedance Matching Transition
E-Field Plot
Match is improved to <-20dB by Tapering the Transition
Placement of Reference Planes
Test FixtureReference Plane
DUT BGA
ATE Test Fixture
DUT Socket
DUT BGAReference Plane
Pogo Pin Connector
ATE Pin ElectronicsReference Plane
Typical Reference Planes
DUT DUT BGA
Reference Planes for Cascading
10
8 February 2008
ATE Pin Electronics
Test FixtureReference Plane
DUT BGA
ATE Test Fixture
DUT Socket
DUT BGAReference Plane
Pogo Pin Connector
ATE Pin Electronics
ATE Pin ElectronicsReference Plane
DUT Socket ?Reference Plane
Designing for ModelingLarge PCB Via Design – 100 mil Small PCB Via Design – 40 mil
11
8 February 2008
TOP SIDEof PCBPogo Pin Cable
Pogo Pins
Large Via Small Via
Reduce the Size of a Model by Containing the Fields
High Speed Pogo Via Ultra-High Speed Pogo Via
Ground Vias 100 mils from Signal Via Ground Vias 40 mils from Signal Via
12
8 February 2008
Ground Vias 100 mils from Signal Via Ground Vias 40 mils from Signal Via
Ming Tsai
From IMS/MTT 2007
Conclusions
� Models only work if they can connect to the real world
� 3D-EM simulations are the only way to see what is happening at the transition of one structure to another.
13
8 February 2008
another.
� Design modifications to improve simulation and measurement of sub-components in a system.
Acknowledgements
Antonio Ciccomancini Scogna, CSTJose Moreira, VerigyMing Tsai, Amalfi SemiconductorKosuke Miyao, Verigy
SPECIAL THANKS TO:
14
8 February 2008