In-Situ De-embedding

(ISD)

Ching-Chao Huang
[email protected]

January 20, 2016

www.ataitec.com
Outline

 What is causality
 How to identify non-causal S parameter
 Why does S parameter violate causality
 What is In-Situ De-embedding (ISD)
 Why do other de-embedding methods give causality error
 How to use ISD to…
 De-embed crosstalk by a single trace test coupon
 De-embed long traces and extract a small DUT
 Improve results for USB Type C compliance testing
 Comparison of ISD with TRL, AFR and simulation
 Mobile-apps-like SI tools: ISD, ADK and X2D2

www.ataitec.com 2
VNA and S parameter

 Vector network analyzer (VNA) is an equipment that

launches a sinusoidal waveform into a structure,
varies the period (or frequency) of waveform, and
lets us observe the transmitted and reflected wave
as “frequency-domain response”.
 Such frequency-domain response, when normalized
to the incident wave, is called scattering parameter
(or, S parameter). S11

T (period)
f (frequency) = 1/T

S21
www.ataitec.com 3
What is S parameter

 For an n-port (or I/O) device, S parameter is an n

x n matrix:
n-1 n
1  S11 S12 S13 ... S1n 
2 S S 22 S 23 ... S 2 n 
3 [S ]
ij n × n =  21
: : : : 
 
 S n1 S n 2 S n 3 ... S nn 
 Sij is called the S parameter from Port j to Port i.
 Sij has a unique property that its magnitude is less
than or equal to 1 (or, 0 dB) for a passive device.

Sij ≤ 1
Sij (dB ) = 20 × log10 Sij ≤ 0 dB

www.ataitec.com 4
What is a Touchstone (.sNp) file

 S parameter at each frequency is expressed in

Touchstone file format.
in GHz in dB and Reference
S param phase angle impedance

! Total number of ports = 4

! Total number of frequency points = 800
# GHZ S DB R 50
0.025 -36.59296 48.77486 -41.40676 79.91354 -0.08648679 -6.544144 -49.50045 -105.618
-41.39364 79.94686 -36.35592 51.52433 -49.4886 -105.5124 -0.09038406 -6.527076
-0.08421237 -6.537903 -49.44814 -105.644 -36.0317 49.60022 -41.37105 79.91856
-49.44393 -105.8186 -0.09834136 -6.542909 -41.36758 79.9318 -36.05645 48.98348
0.05 -32.22576 48.03161 -35.59394 74.15976 -0.1277169 -12.82876 -43.90183 -112.0995
-35.58736 74.16304 -32.12694 50.92389 -43.90926 -112.0764 -0.132402 -12.7985
-0.1242117 -12.82302 -43.89 -112.0248 -32.10987 50.3115 -35.56998 74.078
-43.88424 -112.0517 -0.1381616 -12.80199 -35.56758 74.06782 -31.94136 50.49276
0.075 -29.88861 42.02766 -32.19713 68.06704 -0.1589249 -19.05252 -40.67476 -118.8188
-32.19116 68.0941 -29.7086 45.41557 -40.63857 -118.837 -0.1635606 -19.01593
-0.1603356 -19.0376 -40.63557 -118.8543 -29.89064 47.63852 -32.16917 67.94677
-40.65711 -118.8021 -0.1737256 -19.02956 -32.16865 67.93389 -29.65444 46.15548

: : :

Frequency in GHz S11, S12, …, S44 in dB and phase angle

www.ataitec.com 5
What is causality

cau·sal·i·ty
noun
1. the relationship between cause and effect.
2. the principle that everything has a cause.

In other words:

Can not get something from nothing.

www.ataitec.com 6
How to identify non-causal S parameter

 Convert S parameter into TDR/TDT.

Response before time
zero* is non-causal.

* Delay waveform by
1ns to see if tools
do not show before
time zero.

 Check phase angle. Counterclockwise

phase angle is
non-causal.

www.ataitec.com 7
Why does S parameter violate causality

 Measurement error (de-embedding), simulation

error (material property) and finite bandwidth of S
parameter all contribute to non-causality.
 Kramers-Kronig relations require that the real and
imaginary parts of an analytic function be related
to each other through Hilbert transform:
Ψ (ω ) = ΨR (ω ) + jΨI (ω )
1 ΨI (ω ')
∞
ΨR (ω ) = P ∫ dω '
π − ∞ ω '−ω
1 ∞ ΨR (ω ')
ΨI (ω ) = − P ∫ dω '
π − ∞ ω '−ω
In-Situ De-embedding (ISD) gives causal de-embedding and Advanced SI
Design Kit (ADK) extracts causal material property and/or forces S parameter
of finite bandwidth to give causal behavior of infinite bandwidth.

www.ataitec.com 8
What is de-embedding

 To remove the effect of fixture (SMA connector +

lead-in/out) and extract the S parameter of DUT
(device under test).

SMA DUT SMA DUT

to VNA
lead-in trace test board lead-out trace lead-in trace test board

• The lead-ins and lead-outs don’t need to look the same.

• There may even be no lead-outs (e.g., package).

www.ataitec.com 9
Why do most de-embedding tools give
causality error
 Most tools use test coupons directly for de-
embedding, so difference between actual fixture
and test coupons gets piled up into DUT results.

DUT
- Test coupons

Phantom limbs* due to

difference in fiber weave,
etching, soldering, …
DUT

* http://www.edn.com/electronics-blogs/test-voices/4438677/Software-tool-fixes-some-causality-violations by Eric Bogatin

www.ataitec.com 10
What is In-Situ De-embedding (ISD)

 Use “2x thru” or “1x open / 1x short” as reference

and de-embed fixture’s actual impedance through
optimization.
In Situ

• Mobile-Apps-like SI
software as easy to use
2 as 1-2-3.

• Causal by construction.
3

www.ataitec.com 11
What is “2x thru”

 “2x thru” is 2x lead-ins or lead-outs.

SMA DUT

lead-in trace test board lead-out trace

2x thru for lead-ins 2x thru for lead-outs

2 sets of “2x thru” are required for asymmetric fixture.

www.ataitec.com 12
What is “1x open / 1x short”

 “1x open / 1x short” is useful when “2x thru” is

not possible (e.g., connector vias, package, …).

SMA DUT

lead-in trace test board lead-out trace

1x open 1x short 1x open 1x short

www.ataitec.com 13
Why ISD is more accurate and saves $$$

TRL calibration board ISD test coupon

 More board space - Multiple test coupons  Only one 2x thru test coupon is needed.
are required.  Test coupon is used only for reference, not
 Test coupons are used directly for de- for direct de-embedding.
embedding.  Actual DUT board impedance is de-
 All difference between calibration and embedded.
actual DUT boards gets piled up into DUT  Inexpensive SMAs, board materials (FR4)
results. and loose-etching-tolerance can be used.
 Expensive SMAs, board materials (Roger)  ECal can be used for fast SOLT calibration.
and tight-etching-tolerance are required.
 Reference plane is in front of SMA.
 Impossible to guarantee all SMAs and traces
are identical (consider weaves, etching, …)  De-embedding is made easy as 1-2-3 with
only two input files: 2x thru and DUT board
 Time-consuming manual calibration is (SMA-to-SMA) Touchstone files.
required.  More information: Both de-embedding and
 Reference plane is in front of DUT. DUT files are provided as outputs.

* TRL = Thru-Reflect-Line
www.ataitec.com 14
Example 1: Mezzanine connector
ISD vs. TRL

 In this example, we will use ISD and TRL to

extract a mezzanine connector and compare their
results.
To be de-embedded

SMA Mezzanine
connector
(DUT)

*Courtesy of Hirose Electric

www.ataitec.com 15
DUT results after ISD and TRL
Which one is more accurate?

 TRL gives too many ripples in return loss (RL) for

such a small DUT.

Non-causal
ripples

www.ataitec.com 16
Converting S parameter into TDR/TDT
shows non-causality in TRL results

Non-causal

Non-causal
?

Rise time = 40ps (20/80)

www.ataitec.com 17
Zoom-in shows non-causal TRL results
in all IL, RL, NEXT and FEXT
 TRL causes time-domain errors of 0.38% (IL),
25.81% (RL), 1.05% (NEXT) and 2.86% (FEXT) in
this case*.
Non-causal
* The percentage is
larger with single-bit
Non-causal response and/or
faster rise time.

Non-causal

Non-causal

Rise time = 40ps (20/80)

www.ataitec.com 18
How did ISD do it?

 Through optimization, ISD de-embeds fixture’s

impedance exactly, independent of 2x thru’s
impedance.

2x thru and
fixture have
different
impedance.

Rise time = 40ps (20/80)

www.ataitec.com 19
TRL can give huge error in SDD11 even with
small impedance variation*

 ISD is able to de-embed fixture’s differential

impedance with only a single-trace 2x thru.
DUT

De-embedding

TRL gives more

than 100% error
due to causality
violation.

Rise time = 40ps (20/80)

* The impedance variation between 2x thru and
fixture is less than 5%. (See previous slide.)

www.ataitec.com 20
Example 2: Mezzanine connector
Extracting DUT from a large board

 TRL is impractical for de-embedding large and

coupled lead-ins/outs.

Fixture

Mezzanine
connector
2x thru (DUT)
(12” total)

www.ataitec.com 21
ISD can use a .s4p file of 2x thru for
de-embedding
 TRL would have required many long and coupled
traces.

www.ataitec.com 22
ISD can even use a .s2p file of 2x thru to de-
embed crosstalk…

 And the results are similar!

www.ataitec.com 23
ISD allows a large demo board to double
as a characterization board
 ISD de-embeds fixture’s impedance regardless of
2x thru’s impedance.

Rise time = 40ps (20/80)

www.ataitec.com 24
Example 3: USB type C mated connector
ISD vs. AFR

 Good de-embedding is crucial for meeting compliance

spec.

3
7
5 2x thru

3
1
1

DUT 2
4 6 Fixture

8
2 4

www.ataitec.com 25
DUT results after ISD and AFR
Which one is more accurate?

 AFR gives too many ripples in return loss (RL) for

such a small DUT. Non-causal ripples

www.ataitec.com 26
Converting S parameter into TDR/TDT
shows non-causality in AFR results
 Counter-clockwise phase angle is another indication
of non-causality.
Ripples

Non-causal

Counter-
clockwise

www.ataitec.com 27
De-embedding affects pass or fail of
compliance spec.
 ISD improves IMR and IRL (from compliance tool).

ISD AFR
Spec
-0.6
-0.8
-1.0
-40
-18
-44
-44

PASS FAIL

IL RL IL RL

www.ataitec.com 28
Example 4: Resonator
ISD vs. AFR vs. simulation

 Good de-embedding is crucial for design verification

(i.e., correlation) and improvement.

2x thru

feed line feed line

DUT

www.ataitec.com 29
SDD21
ISD correlates with simulation better

Ripples

Red – Simulation
Yellow – After AFR

Red – Simulation
Yellow – After ISD

www.ataitec.com 30
SDD11
ISD correlates with simulation much better

Red – Simulation
Yellow – After AFR

Red – Simulation
Yellow – After ISD

www.ataitec.com 31
SCC11
ISD correlates with simulation much better

Red – Simulation
Yellow – After AFR

Red – Simulation
Yellow – After ISD

www.ataitec.com 32
ADK and X2D2
Mobile-apps-like SI tools

 ADK is a collection of many SI programs to convert

S parameter into TDR/TDT, SPICE model or eye
diagrams, compare with industry spec. (ICR, ICN,
…) and extract DK, DF and roughness, …
 X2D2 models the effect of surface roughness on all
IL, RL, NEXT and FEXT by effective conductivity.

VNA 3D Solver
Material property
S S S

ISD ADK X2D2

S S
S S S
Channel
Simulation S = S parameters

AtaiTec enhanced flow

www.ataitec.com 33
Automated extraction of DK, DF,
roughness and 2D cross section
 Built-in templates for microstrips and striplines.
 Other templates (such as cable) can be easily added.
 Easily create trace S param for any length and to
any frequency.

Note: We can fit RL from ISD but not from TRL because TRL’s RL is often non-physical.
www.ataitec.com 34
RL is crucial for DK extraction
Use ISD instead of TRL results for extraction

 TRL gives non-physical RL and will be impossible to fit.

Matching RL is crucial because it affects DK and cross section
(and therefore length, DF and roughness).

www.ataitec.com 35
Summary

 In-Situ De-embedding (ISD) solves the causality

problem commonly encountered in de-embedding.
 ISD is causal by construction and it can…
 give more accurate DUT results,
 de-embed crosstalk using a single-trace coupon,
 de-embed long traces and extract a small DUT,
 give better DUT results for compliance testing,
 correlate with simulation better and make design
refinement possible.
 Mobile-apps-like ISD is easy to use and it helps
save $$$ from board material, correlation and
design cycle time, …

www.ataitec.com 36
 Appendix

1x Open De-embedding

www.ataitec.com
ISD’s new 1x open de-embedding needs
only one 1x open test coupon

Fixture + DUT 1x open DUT

-
0.5” 4.5” 0.5” 0.5” 1.5” 1.5”
microstrip stripline microstrip microstrip stripline stripline

DUT

www.ataitec.com 39
ISD can reconstruct 1x thru’s IL and RL
from 1x open’s RL

SDD12

SDD11
SDD11 SDD22

open

www.ataitec.com 40
ISD’s “in-situ” technology matches the
fixture’s impedance for de-embedding

www.ataitec.com 41
IL and RL extracted by ISD match the
actual values very well

1.5”
stripline

DUT

www.ataitec.com 42