EMC modeling

EMC
modeling
04.12.2006, KOTEL TR7 meeting
© Esju Oy 2006 page: 1
www.esju.fi
Contact information
Esju Oy
Konekuja 8
90630 Oulu
tel: 0424 54321
Timo Tarvainen
email: timo.tarvainen@esju.fi
tel: 0424 5432 216
© Esju Oy 2006 page: 2
Contents
• Very brief overview of Esju Oy (from where I come)
• The tools with examples
• 3-D models
• 2&2.5-D models
• Circuit simulations
• Analytic models
• Combined models
• Thoughts on future
• Summary
© Esju Oy 2006 page: 3
Esju Oy
•
Esju Oy is a professional provider of engineering design and
research services
•
Services mainly for
• Telecommunications
• Defence industry
• Health care technology
•
•
•
Founded 1991
About 40 employees
Turnover 3.1 million euros
(year 2005)
© Esju Oy 2006 page: 4
www.esju.fi
Esju Oy
EMC laboratory
• EMC testing
services
• EMC specialist
services
Engineering
• RF technology
• Analog and digital tech.
• EMC and isolation design
• Electromagnetic design
• Measuring and testing
• Embedded systems
© Esju Oy 2006 page: 5
Research and training
• EMC, EMI and isolation studies
• RF studies
• EMC, EMI and isolation training
• Customer specific cources
3-D tools
•
•
Volume vs. surface
TD vs. FD
TL
M
1a
2
1b
4
5
© Esju Oy 2006 page: 6
3
1b
FI
T
? 4
PE
EC
???????
?
5
FE
M
M
oM
3
???
????????
??????
???
?????
?? ?
2
?
FD
TD
1a
110
0
101
0011
0
0
1 100
1
1110101001 1
10
01
1
Examples of 3-D EMC problems
• Radiated emissions calculations at desired distance
• Radiated immunity calculations with plane waves or
antennas
• Isolation/ crosstalk/ shielding effectiveness
evaluations
• Filtering evaluations with PWBs
• Mechanics and PWBs are often well suited to 3-D
simulations
• Parameter extractions from the measured data
• Troubleshooting – finding the root cause of an error
© Esju Oy 2006 page: 7
Some difficult areas for 3-D
• Non-linearity is not easy today, but some tricks can be
used (e.g. local parameter variations based on
analytic models and non-linear circuit models)
• Models that can adequately treat non-linearity, thermal
behaviour and 3-D electromagnetics are under
development
• Poor or random contacts between parts
• Structures that leak and are very thin vs. wavelength
• Very big size vs. wavelength
© Esju Oy 2006 page: 8
Problem meshing
Surface (mainly)
Volume
(CST MWS)
Further information at:
www.csee.ltu.se/peec
© Esju Oy 2006 page: 9
PEEC example on visualization
•
•
•
© Esju Oy 2006 page: 10
Screenshot from 50 ns,
Gaussian, current source
excitation at the front.
Arrows indicate the
current distribution.
Further information at:
www.csee.ltu.se/peec
FIT large device EMP example 8m
Outside
E [kV/m]
Desire: big problems
from DC up to 40 GHz
•
•
•
50kV/m radiated EMP pulse
Typically Ein << Eout,
but not everywhere
How to define SE?
© Esju Oy 2006 page: 11
Inside
Power divider radiation (Pin = 1W) 1/2
© Esju Oy 2006 page: 12
Power divider radiation (Pin = 1W) 2/2
@3m, front
[dBV/m]
@3m, max,
[dBV/m]
© Esju Oy 2006 page: 13
Detonator immunity
Transfer function S = I/E [A*m/V]
1.00E+00
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
1.00E-01
S = 0.00245
1.00E-02
20mm
1.00E-03
1.00E-04
S, Straight
S
Ignition wire, R = 1Ω
S, Dipole
1.00E-05
1.00E-06
S = 7.87E-6
1.00E-07
30mm
1.00E-08
1.00E-09
f [GHz]
• Transfer function (S=I/E) can be multiplied with different radiated threat
•
levels to find the ignition wire RF current
For example scaling to 5670V/m threat level @1.8GHz yields 0.044A and
13.9A ignition wire RF currents for straight and dipole wire arrangements for
the same detonator, respectively
© Esju Oy 2006 page: 14
Coupling between antennas
•
Two 3.1m long antennas on the opposite corners of the bus
© Esju Oy 2006 page: 15
What happens on the coupling?
1. The bus resonance of λ/2 type.
3. Slight downward
shifting of λ/4
monopole response
(nom. ≈ 24.2MHz)
1
3
2
2. Strongly attenuated
wave propagation.
© Esju Oy 2006 page: 16
Internal shielding effectiveness and gaskets
0
-20
S21 [dB]
-40
-60
-80
-100
-120
0
1
2
f [GHz]
measured
cond 450
Microwave Testing of Electromagnetic Shielding On a Printed Wiring Board Level
Computationally Efficient FIT Characterization and Measurement of EMC Gaskets
- ICEAA 01 & ICEAA05, Turin, Italy
© Esju Oy 2006 page: 17
3
2 and 2.5D approximations
Hnorm
Enorm
Scaled resonant frequency
4,5
4,0
3,5
3,0
2,5
2,0
1,5
1,0
0
50
Sample number
20 vias
50 vias
100 vias
100
200 vias
Statistical Analysis of
the PCB Resonance Suppression
with Randomly Positioned Ground Vias
- EMC Europe 2006, Barcelona, Spain
• Good for printed wiring boards and insight
© Esju Oy 2006 page: 18
Circuit simulations
• Good when the design is not too complex and
component models are available
• Handles non-linear components (e.g. transient
protectors)
• Typical applications:
• Signal integrity
• Filtering
• ESD, EFT and other transients
© Esju Oy 2006 page: 19
Signal integrity example
APLAC 8.10 User: ESJU Oy
Nov 22 2006
0.30
V
0.12
-0.05
-0.23
-0.40
0.000
10.000n
20.000n
t/s
30.000n
V_Rx_diff_1
V_Rx_diff_2
VRx_1
SSub FR4Buried
ER = 4.4
B = 1mm
T = 0.017mm
TAND = 0.015
RHO = 0.707
LEVEL = 2
FILE="90c031tm.ibs"
COMPONENT = "DS90C031TM"
PIN = "2"
EN = "V_En"
W = Width
S = Separation
L = Length
Zterm
© Esju Oy 2006 page: 20
FILE="90c032tm.ibs"
COMPONENT = "DS90C032TM"
PIN = "2"
IBIS (Diff Input)
V_In
IBIS (Diff Output)
VRx_2
Sclin
VRx_3
Var V0 = 3 opt
Var Trise = 200p opt
Var Tfall = 200p opt
Var Tbit = 2ns opt
Var ZTerm = 100 opt
Var Width = 360u opt
Var Separation = 600u opt
Var Length = 50mm opt
optimmethod tuning
40.000n
V_In1
Zterm
IBIS (Diff Output)
FILE="90c031tm.ibs"
COMPONENT = "DS90C031TM"
PIN = "2"
EN = "V_En"
VRx_4
FILE="90c032tm.ibs"
COMPONENT = "DS90C032TM"
PIN = "2"
IBIS (Diff Input)
Filtering example
SAMPLE CASE: The same
filter assembled to 50 ohm
and 1 ohm systems
0
• Transients via wires
(power, control,
signal)
• How much filtering?
• Relying on 50 ohms
specifications may
yield surprises
S
-30
[dB]
-60
-90
-120
0
250M
S21@50R
© Esju Oy 2006 page: 21
500M
f [Hz]
750M
S21@1R
1G
Analytic models
• Good for yielding insight and fast
• Multiplex calculation methods
• Typical applications:
• First approximations for coupling problems (e.g.
EMP)
• Radiated emissions approximations
• Shielding effectiveness calculations (but typically
inaccurate, because of closed volumes, although
approximations for these exist also)
© Esju Oy 2006 page: 22
EMC laboratory absorbers
Already layered = “easy”
Not layered = “bit tricky”
ε z = (1 − g) ε 0 + gε a
µ z = (1 − g) µ 0 + gµ a


2( ε a − ε 0 )
ε t = ε 0 1 + g
(1 + g)ε 0 + (1 − g)ε a 

 z
g= 
 L
2


2( µ a − µ 0 )
µ t = µ 0 1 + g
(1 + g)µ 0 + (1 − g)µ a 

Zi , in =
Tarvainen Timo (1997) Häiriösuojatun radiokaiuttoman
huoneen suunnittelu. Oulun yliopisto, sähkötekniikan
osasto. Lisensiaatintyö, 86 s.
© Esju Oy 2006 page: 23
Zi +1, in + jZi , ⊥ tan(φ i )
Zi ,⊥ , i = 2, 3, ... , n-1,
Zi , ⊥ + jZi +1, in tan(φ i )
Γ=
Z2 , in − Z1,⊥
,
Z2 ,in + Z1, ⊥
Some absorber reflection examples
o = 0°, + = 30° TE,
x = 45° TE ja * = 60° TE
+ = kohtisuora tulokulma,
* = 45° TE, o = 45° TM
+ = kohtisuora tulokulma,
* = 45° TE, o = 45° TM
© Esju Oy 2006 page: 24
Tarvainen Timo (1997) Häiriösuojatun radiokaiuttoman
huoneen suunnittelu. Oulun yliopisto, sähkötekniikan
osasto. Lisensiaatintyö, 86 s.
Combined tools
•
•
3-D electromagnetics + circuit simulations + measurements
Example: a concept high-speed connector to antenna coupling
APLAC 8.10 User: ESJU Oy May 18 2006
-70.00
P
-80.00
[dBm
-90.00
-70
0
500
1000
1500
2000
-80
measured
simulated
-90
-100.00
-100
-110.00
0.000
-110
750.000M
1.500G
f/Hz
2.250G
3.000G
P coupled ant[
ON THE EFFECT OF HIGH SPEED CONNECTOR SHIELDS
DISTRIBUTED CONTACT RESISTANCE TO RADIATED
EMISSIONS, EMC Europe 2006, Barcelona, Spain
On the Effect of Mobile Device Shape Characteristics to
Interconnection Noise Coupling to an RF Chip Antenna,
EPEP 2006, Scottsdale, Arizona, USA
© Esju Oy 2006 page: 25
frequency [MHz]
2500
3000
PWB filter assembly example
•
•
Circuit models are often computationally quite efficient
•
Combine with electromagnetic models for accuracy + efficiency
BUT, it can be impossible to get accurate results when 3-D
geometry counts
Parasitic examples (Pi filter: C = 5.6p, L = 56n, 5o ohm ports)
Aplac 7.61 User: ESJU Oy
Feb 20 2002
0.00
S21
-30.00
Cpar
dB
IN
-60.00
Cfilt
Rfilt
Lpar
-90.00
-120.00
1
0.000
OUT
Cfilt
Lpar
Lpar
Rpar
Rpar
750.000M
Aplac
CST coplnar
© Esju Oy 2006 page: 26
1.500G
f/Hz
2.250G
CST Vias
3.000G
Computational electromagnetics future
• Parallel processing
• Integration of tools
• Multiple scales
• EMC fingerprints, MOR (Model Order Reduction)
• Handling non-linearity
• Total CEM solutions
• Multiphysics
Further information at:
http://www.ewh.ieee.org/soc/cpmt/tc12/
© Esju Oy 2006 page: 27
Look for presentations from FDIP'06 workshop
Summary
• Huge amount of EMC type problems can be solved
today
• Complete device models are coming, but not yet quite
there
• Computational capacity increases very quickly with
respect to time
© Esju Oy 2006 page: 28