EXTRACTION OF DIELECTRIC PROPERTIES OF PCB LAMINATE DIELECTRICS ON PCB STRIPLINES TAKING INTO ACCOUNT CONDUCTOR SURFACE ROUGHNESS SPEAKER: DR. MARINA KOLEDINTSEVA, IEEE SENIOR MEMBER, ORACLE (THE WORK IS DONE IN EMC LAB OF MISSOURI S&T, SPONSORED BY CISCO AND NSF) 1
50
Embed
EXTRACTION OF DIELECTRIC PROPERTIES OF PCB · PDF file2 P I Outline I. Introduction – motivation, objectives, and state-of-the-art II. Idea of an ^effective roughness dielectric
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
EXTRACTION OF DIELECTRIC PROPERTIES OF PCB LAMINATE DIELECTRICS ON PCB STRIPLINES TAKING
INTO ACCOUNT CONDUCTOR SURFACE ROUGHNESS
SPEAKER: DR. MARINA KOLEDINTSEVA, IEEE SENIOR MEMBER,
ORACLE
(THE WORK IS DONE IN EMC LAB OF MISSOURI S&T, SPONSORED BY CISCO AND NSF)
1
2
Outline I. Introduction – motivation, objectives,
and state-of-the-art
II. Idea of an “effective roughness dielectric” (ERD)
III. PCB stripline cross-sectional analysis and roughness profile quantification
IV. Experiment-based input data for numerical electromagnetic modeling
V. Modeling results & validation
VI. Building of “design curves” regarding conductor surface roughness
VII. Conclusions
3
3 Gbps 28 Gbps
• Conductor surface roughness lumps into laminate dielectric parameters.
• Any existing analytical and numerical models of conductor surface roughness are approximations.
• Study and adequate modeling of wideband behavior of dielectrics and conductors in PCBs is important from SI point of view.
VLP
• Conductor roughness affects both phase and loss constants in PCB transmission lines and results in eye diagram closure.
STD
HVLP
Motivation
The same dielectric, the same geometry, but different copper foil profiles
Low roughness - HVLP
Medium - VLP
High - STD
VLP
STD
4
Objectives • Develop a technique to accurately measure and extract laminate
dielectric parameters (DK=' & DF=tan) removing effects of conductors.
• Develop a physics-based model, which allows for simple incorporation of conductor surface roughness in electromagnetic numerical models of transmission lines.
• Test and validate the proposed model using measurements on a multitude of various test boards with different cross-sections and roughness profiles.
• Test and validate the proposed model using electromagnetic numerical simulations with different software tools.
• Develop a database for roughness parameters, corresponding to different types of copper foils used in PCBs.
5
A. Koul, M. Koledintseva, et al, Proc.
IEEE Symp. Electromag. Compat., Aug.
17-21, Austin, TX, 2009, 191-196
'2 "24. .cos2
r rc
'2 "24. .sin2
d r rc
Measured S-parameters
Causality , Passivity &
Reciprocity check
ABCD parameters
T c d
arccos h A D
linelength
jT
Solve the system of equations
to obtain complex permittivity
d T c
Model or experimentally
retrieve conductor loss for
rough stripline conductor
OPTIONS
• Analytical Models
• Numerical Models
• Experimental
S-parameters are measured using VNA or TDR with “Through-Reflect-Line” (TRL) calibration in f-domain or t-domain, respectively
IV. Experimental separation of conductor & dielectric loss
• Koledintseva et al
Stain-proof layer
Anti-tarnish layer
Drum foilDrum foil
Dendrite plating
Protective barrier
Stain-proof layer
Oxide treatment
7
• Experiment-based Differential and Extrapolation Roughness Measurement techniques (DERM and DERM-2) have been proposed to refine wideband DK and DF from roughness.
[1] A. Koul, M.Y. Koledintseva, S. Hinaga, and J.L. Drewniak, “Differential extrapolation method for separating dielectric and rough conductor losses in printed circuit boards”, IEEE Trans. Electromag. Compat., vol. 54, no. 2, Apr. 2012, pp. 421-433. [2] M.Y. Koledintseva, A.V. Rakov, A.I. Koledintsev, J.L. Drewniak, and S. Hinaga, “Improved experiment-based technique to characterize dielectric properties of printed circuit boards”, IEEE Trans. Electromag. Compat. (to be published soon in 2014)
• An Effective Roughness Dielectric (ERD) approach has been proposed to substitute inhomogeneous roughness boundary layer by a layer with homogenized dielectric properties.
[3] M.Y. Koledintseva, A. Razmadze, A. Gafarov, S. De, S. Hinaga, and J.L. Drewniak, “PCB conductor surface roughness as a layer with effective material parameters”, IEEE Symp. Electromag. Compat., Pittsburg, PA, 2012, pp. 138- 142.
Our Recently Published Works
8
Idea of “Effective Roughness Dielectric”
8
1fG
R R1C
1C
1G
1G
2C
2C
2G
2G
1fC
2fC2fG
L L
Equivalent circuit model to
calculate per-unit-length parameters
Roughness leads to the capacitance increase!
1C
2C
w
01C
02C
1C
2C
t
/ 2A
/ 2A
1fC
2fC
This effect was first noticed in:
9
Tr
Ar t r
y
x
, 1(1 ) ( )
incl matrixeff y matrix incl
matrix incl y incl matrix
vv N
.matrix m mj 0
iincl i
j
2
1ln( )yN a
a
Maxwell Garnett Mixing Rule for aligned metallic inclusions
T. Vincent, M. Koledintseva, A. Ciccimancini, and S. Hinaga,
“Effective roughness dielectric in a PCB: measurement and full-
wave simulation verification”, IEEE Symp. EMC, Aug. 2014
(accepted)
38
T. Vincent, M. Koledintseva, A. Ciccimancini, and S. Hinaga, “Effective roughness dielectric in a PCB: measurement and full-wave
simulation verification”, IEEE Symp. EMC, Aug. 2014 (accepted)
STD
Unwrapped phase of S21
S21 CST Modeled vs. Measured)
STD
39
0(1 )c c r 0
r
c
r
0r c c
Modeled & Measured Magnitude of S21 for Striplines with Different Foils
Slope of S21 as a function of frequency increases with the increase of surface roughness
0 5 10 15 20 25 30-25
-20
-15
-10
-5
0
Frequency, GHz
S2
1,
dB
Smooth conductor
STD
VLP
HVLP
21 08.686( )D csmoothS L
HVLP
STD
VLP
Smooth
40
0 0.1 0.2 0.3 0.4 0.50
0.1
0.2
0.3
0.4
QR
Ad
diti
ona
l slo
pe
R,
dB
/GH
z
Experimental points
linear
quadraticSTD
VLPHVLP
0 0.1 0.2 0.3 0.4 0.50
0.2
0.4
0.6
0.8
1
QR
Rn,
dB
/GH
z/m
Experimental points
linear
HVLP
VLP
STD
R=(S smooth- S rough)/f [dB/GHz] Rn=(S smooth- S rough)/f/L [dB/GHz/m]
Additional Slope in S21 as a Function of Roughness Factor
0 0.1 0.2 0.3 0.4 0.5-1
0
1
2
3
4
5
6
7x 10
-23
Q, m
K3~
2
SET I
SET II
QR, µm
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
3.6
3.7
3.8
3.9
4
4.1x 10
-6
Q, m
K1~
sq
rt(
)
Extrapolation to Zero Roughness
SET I
SET II
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
2
2.2
2.4
2.6
2.8
3
3.2
3.4
3.6
3.8
x 10-11
Q, m
K2~
Extrapolation to Zero Roughness
SET I
SET II
QR, µm QR, µm
41
Extrapolation to Zero Roughness in α (7-mil Lines)
SET I
RTF
STD
HVLP
RTF
STD
HVLP
RTF
STD
HVLP RTF
STDR
STDR
HVLP RTF
RTF
SET II
41
HVLP
HVLP
STDR
0 0.1 0.2 0.3 0.4
-3.6
-3.4
-3.2
-3
-2.8
-2.6
-2.4
-2.2
-2
-1.8
x 10-22
Q, m
B3~
2
SET I
SET II
QR, µm 42
0 0.1 0.2 0.3 0.4 0.56.2
6.4
6.6
6.8
7
7.2
7.4
7.6
7.8
8
8.2x 10
-6
Q, m
B1~s
qrt(
)
SET I
SET II
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.46.44
6.46
6.48
6.5
6.52
6.54
6.56
6.58
x 10-9
Q, m
B2~
SET I
SET II
SET I SET II
HVLP
HVLP
HVLP
RTF
RTF
STD
STDR
RTF
HVLP
HVLP RTF
STD
STDR
STD
RTF
HVLP
STDR
RTF
QR, µm QR, µm
Extrapolation to Zero Roughness in β
43
Refined DK and DF
0 5 10 15 20 25 303.7
3.72
3.74
3.76
3.78
3.8
Frequency, GHz
DK
DK SET I
DK SET II
0.7%
0 5 10 15 20 25 30
6.5
7
7.5
8
8.5
9x 10
-3
Frequency, GHz
DF
DF SET I
DF SET II
2.5%
Excellent agreement between the results of extraction of dielectric parameters of two independent sets of boards (3+3 boards total) with the same dielectric and the same geometry, but different types of foil roughness has been obtained! Frequency range is from 10 MHz to 30 GHz.
44
Dielectric Loss and Smooth & Rough Conductor Losses
0 5 10 15 20 25 300
1
2
3
4
5
6
d,
Np
/m
Frequency, GHz
SET I
SET II
0 5 10 15 20 25 300
0.5
1
1.5
c0,
Np
/m
Frequency, GHz
SET I
SET II
0 5 10 15 20 25 30-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
Frequency, GHz
ro
ug
h,
Np
/m
SET I STD
SET I RTF
SET I HVLP
SET II STDR
SET II RTF
SET II HVLP
SET II- STDR
SET I - STD
SET I - RTF
SET II - RTF
SET II - HVLP
SET I- HVLP
Total
conductor
loss
Smooth
conductor
loss
Rough conductor loss
Concentration of “roughness inclusions” decreases with distance from
zero-roughness plane
Maxwell Garnett mixing rule requires knowledge of volume concentration of
“roughness inclusions”. This volume concentration varies as a function of the
height y. Hence the dielectric properties homogenized by Maxwell-Garnett in each
incremental layer are also functions of y.
( ) 1 ( )(1 ( )) ( )
incl matrixMG matrix
matrix y incl matrix
y yy N
)(y
Effective Roughness Dielectric Approximation
Close to smooth conductor, y0=0
Deeper inside ambient dielectric, y1>y0
Deeper inside ambient dielectric, y2>y1
Closer to ambient dielectric, y3>y2
Ny is the depolarization
factor in y-direction.
and
45
d=Ar
1
2
3
Using Maxwell Garnett mixing rule the roughness could be described as gradient
layer with exponential distribution
Loss is due to non-propagating surface waves in the structure with variable permittivity.
0 12
0
01 0
( )( ) arctan ( 1)
4
tyMG eff
MG eff
MG eff
k y dy l
( ) ' ( ) " ( ).MG MG MGy y j y
The characteristic equation for wave
propagation in a gradient waveguide:
effeffeff j '''
Boundary Layer – Gradient Waveguide Model
d
0 1 z
x
)(y )(0 yk
Turn point
(y)= 0 +d f(y)
0
y
0( ) ( )MG MGy f y
0MG
Resultant effective roughness dielectric
46
47
Set 1
STD 15 6.489 0.203 12.7
6.496 0.17 12.0 13 5.05 0.124 5.02
VLP 6 2.739 0.130 8.08
2.752 0.13 9.0 5 1.932 0.100 13.58
HVLP 2 1.083 0.048 5.01
1.086 0.04 4.8 1 0.158 0.203 44.76
Extracted from Gradient Model (on Foil Sides) Extracted from Q2D
Solution #
Turning point= Ar tanrough
𝜺′𝒓𝒐𝒖𝒈𝒉 𝒆𝒇𝒇 Ar tanrough 𝜺′𝒓𝒐𝒖𝒈𝒉 𝒆𝒇𝒇
Results for the Gradient Model: Set 1, Foil Side
There is a reasonable agreement; however, the results were obtained only for a limited number of samples.
Conclusions
48
• A new improved technique DERM2 to extract dielectric properties of a
laminate dielectric for a set of five test vehicles is demonstrated.
• A semi-automatic roughness profile extraction and quantification procedure
has been applied to SEM or optical microscopy pictures of microsections of
PCB stripline.
• A metric called “roughness factor” QR to quantify roughness profiles has
been introduced.
• The correlation between the additional slope in insertion loss due to
roughness and the roughness factor QR has been established. The effective
roughness dielectric layer concept was applied to numerically model (in 2D
FEM) all the five test vehicles.
• In the numerical models, the dielectric parameters of ambient dielectric
were taken as those obtained using the DERM2 procedure; the boundary
roughness layers were substituted by Effective Roughness Dielectric.
• This model and analysis lead to the development of the “design curves”
(additional slopes of insertion loss, or additional conductor loss as a
function of roughness parameter), which could be used by SI engineers in
their designs.
Acknowledgment
49
• I am very grateful to Scott Hinaga (Cisco) for collaboration, weekly discussions at
conference calls, support of ideas, and sponsoring this research in 2008-2014.
• I would like to thank graduate and undergraduate students of Missouri S&T who
contributed to this work: Amendra Koul (Cisco), Praveen Anmula (Mentor
Graphics), Soumya De (Cisco), Fan Zhou (Semtech), Aleksandr Gafarov (Mentor
Graphics), Aleksei Rakov (Moscow Power Engineering Institute), Oleg Kashurkin
(Missouri S&T), and Alexei Koledintsev (Missouri S&T).
• I would like to thank Prof. James Drewniak for an opportunity to work with EMC
Lab of Missouri S&T in 2000-2014, motivation for this research, and useful
discussions.
• I would like to thank Missouri S&T Materials Research Center colleagues – Dr.
Clarissa Wisner , Dr. Beth Culp, Prof. Matt O’Keefe, and Dr. Signo Reis (Saint
Gobain) for their help with SEM and optical micro photographs.
• This work was also partially supported by the National Science Foundation under