agilent ads deembed

8/9/2019 Agilent ADS Deembed

1/35

De-embedding Techniques in

Advanced Design System


2/35

De-embedding Techniques in ADS

Page 2

Agenda

Why de-embedding?

2-port de-embedding

Mixed mode S-parameters

Four port de-embedding

TRL based de-embedding

Summary


3/35


Page 3

Non Insertable Measurements

Die

Ball Pad

Probe SMA

Reference Plane

Reference Plane

Board Response = Total Response Probe SMA Response

Require accurate modeling of Probe SMA response


4/35


Page 4

DUT with Test Fixtures

Requires test fixture: Not a part of the back plane structure

How to remove test fixture effects from the overall measurement?


5/35


Page 5

Backplane measurements

Test Fixture

Backplane Response = Total Response Test Fixture Response

Sometimes it is not feasible to remove test fixture effects using

various calibration techniques


6/35


Page 6

De-embedding Requirements in EM Simulation

Moving Reference Plane

Sometimes it is not feasible to move the reference plane in EM tools.


7/35


Page 7

What is De-embedding?

De-embedding is a mathematical process that

removes the effects of unwanted portions of the

structure that are embedded in the measured data

by subtracting their contribution.

Real DUT SP = Measured SP Fixture Characteristic

De-embedding : Negating effects of unwanted portion

H ACTIVECHANNEL

RESPO

NSE

STIMU

LUS

ENTR

Y

INSTRUMEN

T STATE

R

CHANN

EL

THP-IB

STATUS

NETWORK

ANALYZER

50MHz-20GHz

PORT

2

PORT

1

THRUDUT

Before De-embedding

Requires accurate fixture model

Empirical models from measured data

Simulation based models

After De-embedding


8/35


Page 8

Why use S-Parameters?

relatively easy to obtain at high frequencies relate to familiar measurements (gain, loss, reflection coefficient ...)

can cascade S-parameters of multiple devices to predict system performance

can compute H, Y, or Z parameters from S-parameters if desired

can easily import and use S-parameter files in our simulation tools

Incident TransmittedS 21

S 11

ReflectedS 22

Reflected

Transmitted Incident

b 1

a 1

b 2

a 2

S 12

DUT

Port 1 Port 2

S 11 = Reflected

Incident =

b1

a 1 a2 = 0

S 21 = Transmitted

Incident=

b2

a 1 a2 = 0

S 22 = Reflected

Incident = b

2

a 2 a1 = 0

S 12 = Transmitted

Incident =

b1

a 2 a1 = 0


9/35


Page 9

De-embedding two port network

To de-embed with transfer scattering parameters, thematrix math is straightforward

11 12 12 21 11 22 11

21 22 2221

1

1

T T S S S S S

T T SS

=

Network a Network b

Ttotal = Ta Tb

Ta = Ttotal Tb-1

Tb = Ta Ttotal-1

Order is very important

11 12 12 11 22 12 21

21 22 2122

11

S S T T T T T

S S TT =

Relationship between S and T-Parameters


10/35


Page 10

Getting DUT data using De-embedding

a3

[ TA ] [ TD ] [ TB ]

2-port

DUT

a0

b0

a1

b1

b2b3

a2

MEASURED ACTUAL

[ ] [ ][ ][ ] [ ] [ ] [ ][ ]1 1;m A D B D A m BT T T T T T T T

= =

Let us build a sample network to demonstrate the de-

embedding process in ADS


11/35


Page 11

De-embedding Example - DUT

2-port

DUT

a

0

b

0

a

1

b

1

b

2

b

3

a

2

High speed

connector model


12/35


Page 12

Representing the Test Fixture 2-portDUT

a

0

b

0

a

1

b

1

b

2

b

3

a

2

Test fixture


13/35


Page 13

Test Fixture plus Connector

ConnectorTest Board In Test Board Out

b

3

a

0

b

0

2-port

DUT

a

1

b

1

b

2

a

2


14/35


Page 14

Using 2-Port ADS De-embedding Component


2-port

DUT

a

1

b

2

Comparison with original Connector Model

(Magnitude and Phase Response)

De-embedComponent

De-embedComponent


15/35


Page 15

Backplane Performance Evaluation

Backplane and

Daughter card


16/35


Page 16

The Simulation Circuit Problem

Backplane

4 TraceSMA

Connector

D/C

4Trace

SMA

4 TraceSMA

Connector

D/C

4Trac

e

SMA

Conne

ctor

Mo

del

PROBLEM: Need to remove effect of these

SMAs!

X Trace (4 or 10)

ConnectorM

odel


17/35


Page 17

Solve the Problem with De-embedding

block

Connecto

rM

odel

Con

nector

M

od

el

Effect of SMAs removed by de-embedding.

Backplane

4 Trace

Connector

4Tra

ce

SMA

4 Trace

Connector

4

Trace

SMA

X Trace (4 or 10)

D/C D/C


18/35


Page 18

Simulation vs. Measured Data( Only two port de-embedding element was used)

ScopeEye

V=0.415 V

Teradyne + 18 at 2.5Gb/s

D(sim vs. scope) < 1%

Teradyne + 18 at 3.125Gb/s

(sim vs. scope) = 6%

V=0.251 V

ADS Eye

Simulation

V=0.236 V

ADS Eye

Simulation

V=0.417 V


19/35


Page 19

DUT

Test Board To Be De-embed

Differential Pair

Measurement

Reference Plane

Four Port Measurements

Four Port Fixture: Four Port De-embedding

Need to account for

coupling

Differential-mode signal Common-modesignal

Differential to

common-mode

conversion

DUT Reference Plane


20/35


Page 20

Single Ended Simulation of Differential Pins


21/35


Page 21

44434241

34333231

24232221

14131211

SSSS

SSSS

SSSS

SSSS

Nodal S-parameter

SDD11 SDD12

SDD21 SDD22

SDC11 SDC12

SDC21 SDC22

SCD11 SCD12

SCD21 SCD22

SCC11 SCC12

SCC21 SCC22

Diff Mode

Common Mode

Mixed Mode S-parameter

What is Mixed Mode S-parameters?

reciprocal


22/35


23/35


Page 23

Direct Mixed Mode S-Parameter Simulation

Differential Mode to

Differential Mode S-

parameter

Common Mode to

Common Mode S-parameter

Balun Balun

Comparing two techniques

Advantage Mixed Mode S-Parameter Optimization

C Wi h Fi M d l


24/35


Page 24

Connector With Fixture ModelConnectorTest Board In Test Board Out

b

3

a

0

b

0

2-port

DUT

a

1

b

1

b

2

a

2

Using ADS Four Port De embedding Component


25/35


Page 25

Using ADS Four Port De-embedding Component

Comparison with

Connector Model

Only Response


De-embed

Component

De-embed

Component

Accounts for coupling between feed transmission lines


26/35


Page 26

De-embedding Components

Standard de-embedding components in ADSOne Port de-embedding

Two Port de-embedding

New standard de-embedding components in ADS2006AAggressor and Victim differential pair characterization requires

multi-port De-embedding

Four port de-embedding

Six port de-embedding

Eight port de-embedding

Twelve port de-embedding

Requirements:

Measured/Simulated S-parameters to be available


27/35


Page 27

Test Cases?

Fixture S-parameter cannot be obtained directly?

Can one use TRL calibration standards?


28/35


Page 28

TRL measurement based De-embedded

Component Used when the test fixture S-parameter response is not available

Ability to fabricate test fixtures and calibration standard

Test fixtures are created using the same feed line structure as in the TRL

standard.

DUT & Fixture/Calibration Kit


29/35


Page 29

DUT & Fixture/Calibration Kit

TRLCalibration Kit DUT Board

DUT

Material - FR4

Er=4.2

Thickness=0.6mm

Tand=0.025

T=18[mm] (Metal Thickness)

S=5.8*107[S/m](Conductivity)

TRL calibration standard.


30/35


31/35


Page 31

Why ADS based TRL De-embedding?

Feed line

coupling notaccounted for

Some test fixture might have significant coupling.

How one can account for the coupling effects?

C t D b ddi i TRL C lib ti


32/35


Page 32

Custom De-embedding using TRL Calibration

Standards

Coupled Feed line1

DUT with Coupled feed lines 4-port De-embedding

Our New 4-port De-embedding model

created using C code

Custom solution-Not available as standard

feature of ADSContact EEsof for details

Coupled Feed line2

DUT

Feed lines can be

Asymmetric Can not use TRL on Si Substrate

N 4 t D b ddi ifi ti lt


33/35


Page 33

New 4-port De-embedding verification results

Comparison between De-embedded S-parameters & DUT withoutfeed lines.

2 types of S-parameters are completely identical!!

Compare to ADS model


34/35


Page 34

Compare to ADS model

DUT

DUT

Simulated (ADS Coupled line model)

Measured

Conclusion


35/35


Page 35

Conclusion

Demonstrates the use of ADS for a typical SI problem

Provide powerful measurement based modeling

ADS De-Embedding capabilities

Two port de-embedding

Powerful Multi-port de-embedding essential for bus/ differential bus

( one pair, two pair, & three pair )

TRL calibration based de-embedding

ADS has been used for SI design for over 20 years

ADS has a multitude of accurate built in models

ADS allows you to build accurate physical models

ADS brings IP, simulation and measurement together

Questions?

agilent ads deembed

Documents