1. education
2. teaching and classroom resources
3. lesson plans

How to Fail EMC in 10 Easy
Lessons?
Presented By:
Elya B. Joffe
Immediate Past President, IEEE EMC Society
1
1
Alice in Wonderland…
“Would you tell me please,
which way I ought to go
from here?”
“That depends a good deal
on where you want to get
to”, said the cat.
“I don't much care where –”,
said Alice.
“Then it doesn't matter
which way you go”, said the
cat.
(Lewis Carroll, , Alice in Wonderland, Chapter 6 )
2
2
The Basic Rules in EMC
There are no rules!!!
EMI does what it wants!
where it wants!
when it wants!
3
3
3
The Basic Rules in EMC
I. You can’t
Win them All
Without a doubt, the majority
of EMI Problems are man made
and are a result of lack of
foresight, poor planning or poor
design
III. If you think you
can, go to Rule #I
4
II. You can’t even
break even
4
4
When in Distress… The 3 J’s are
at Your Service
5
5
5
!o Easy Lessons for
Failing EMC…
Let’s
10,…9…8
Go!!!
3,…2,…1
6
6
Lesson 1
Minimize Edge Rates, Particularly of Clocks
Most digital signals have extremely short
(sub-nanosecond) transition times
Spectral content mainly proportional
to edge rates
Clocks are worst:
Narrowband discrete harmonics
Steady state, continuous signals
7
7
Lesson 1
Minimize Edge Rates, Particularly of Clocks
e( f )
1
f1 =
π ⋅d
∝
1
f2 =
π ⋅ tr
1
, -20 dB/dec
f
1
∝ 2 , -40 dB/dec
f
log ( f )
Reconstruction using 1,
3, 5, 7 Harmonics
Reconstruction using 7,
15, 27 Harmonics
Effect of Wave-Shape on Spectral Content
8
8
Lesson 1
Minimize Edge Rates, Particularly of Clocks
All (almost) signals comprise of an infinitely
large spectral composition
Radiation efficiency proportional to
“electrical length” of conductors
Emax  fl   l 1 
∝ ∝ × 
I
 r  λ r
You can easily maximize emissions
(and pickup) by:
Either increasing linear dimensions, ℓ…
…or by reducing wavelength, λ
9
9
Lesson 2
Encourage Common-Mode Current Flow
Most digital signals have extremely short
(sub-nanosecond) transition times
Spectral content mainly proportional
to edge rates
Clocks are worst:
Narrowband discrete harmonics
Steady state, continuous signals
10
10
Lesson 2
Encourage Common-Mode Current Flow
“Contradictions do not exist. Whenever you think you are facing a
contradiction, check your premises. You will find that one of them is
wrong”
“Atlas Shrugged”
IC =
I1 + I2
2
IC -Common Mode
Current
ID=
I1 − I2
2
ID -Differential Mode
Current
11
11
Lesson 2
Encourage Common-Mode Current Flow
+ICM
+IDM
+ICM
Vin
I3
I1
Electric Flux
r
D
A
I2
-IDM
VG
2-ICM
How to successfully produce common-mode EMI?
Form ground loops ↔ grounding design
Allow crosstalk on PCBs and in cables ↔ circuit layout
Allow skew in differential pairs ↔ circuit layout
Create imbalance in circuits ↔ circuit design
12
etc…
12
Lesson 2
Encourage Common-Mode Current Flow
Differential-mode radiation efficiency:
2
Emax
−14  f A 
= 2.632 ×10 

I
r


V/m
Computed E[µV/m]
for r=1 m, A = 10 cm
Frequency
[MHz]
Source: Ott, H., Noise
Reduction Techniques
in Electronic Systems,
1988
2
Current [mA]
1
10
100
1000
10
1.32
13.2
132
1,320
30
11.9
119
1,190
11,900
100
13.2
132
1,320
13,200
13
13
Lesson 2
Encourage Common-Mode Current Flow
Common-mode radiation efficiency:
Emax
 fl 
= 1.26 ×10−6  
I
r 
V/m
Computed E[µV/m]
for r=1 m, ℓ = 10 cm
Frequency
[MHz]
Current [mA]
1
10
100
1000
10
630
6.3k
63k
630k
30
1.89k
18.9k
189k
1,890k
100
6.3k
63k
630k
6,300k
Source: Ott, H., Noise
Reduction Techniques
in Electronic Systems,
1988
14
14
Lesson 2
Encourage Common-Mode Current Flow
Radiation efficiency at a distance R from two
closely spaced wires carrying common-mode
Current, IC:
1
E = 2 × 6. 28 ⋅ 10−7 ⋅ ( f ⋅ L ⋅ I C ) ⋅ ( ), V/ m
r
At f=30MHz, with radiated emissions limit of
34dBµV/m (or 50µV/m) at r=1m & ℓ=2m…
Common
mmon––mode current as low as
IC=656
656nA
nA is sufficient to exceed
the above limit
15
15
Lesson 3
Always Terminate Cable Shields with “Pigtails”
(or not at all…)
"The mathematical theory of wave propagation along a conductor with an
external coaxial return is very old, going back to the work of Rayleigh,
Heaviside and J. J. Thomson”
(S. A. Schelkunoff, 1934)
Termination significantly affects total shield transfer
impedance
Vi = Z T × I S
Vi (Tot ) = ZT × I S + ( ZTS + ZTL ) × I S
Shield terminations create voltage drops over them
Increased termination impedance degrades performance
Increased emissions
Increased pickup
16
16
Lesson 3
Always Terminate Cable Shields with “Pigtails”
(or not at all…)
You could use high quality, expensive, shield terminations
EMI Backshell Termination
Adapter
Strain
Backshell
Relief
Ground
Plug
Hooks
Individual
Wire
Shields
Source: Glenair
Individual Wire Shield
Termination
D-Type Shield Termination
But then – why should you?
17
17
Lesson 3
Always Terminate Cable Shields with “Pigtails”
(or not at all…)
Do not use expensive, low impedance shield terminations
They are wasteful, and may actually help you pass EMC…!
Long “pigtail” shield terminations increase EMI emission
and coupling
It performs almost like no shield is present
50
Shielding Effectiveness [dB]
S.E. [dB] Pigtail Termination
S.E. [dB] Peripheral Termination
40
30
20
10
0
1
2
3
4
5
6
Frequency [GHz]
7
8
9
10
(*) 2” Pigtail is the absolute
minimum length recommended
to ensure EMC failures
18
18
Lesson 3
Always Terminate Cable Shields with “Pigtails”
I, dBµA
I, dBµA
(or not at all…)
You could even further reduce coupling with no
termination at all…
F, MHz
F, MHz
100Base-T Transmitter:
Pigtail Termination at
Near End
100BaseT Transmitter:
Shield Disconnected at
Near End
Source: Prof. Dr.-Ing. H. Garbe
19
19
Lesson 4
Don’t Waste your Time Balancing a Twisted Pair
Twisting of the cable pair provides:
Reduction of loop area between conductors
Effective cancellation of magnetic flux from adjacent
“mini-loops”
Source
Load
Loop j Loop j+1
I+
d
Idlj
 1  
 π  
RT = −20 ⋅ Log 
 ⋅ 1 + 2nl ⋅ sin     ; dB
 nλ   
 2nl + 1  
RT ≤ 60dB @ f ≤ 100 kHz for 30 ÷ 40 Twists/m
s
dlj+1
~ ~
rj
ur
Bj
p
rj+1
ur
B j +1
Twisting is only effective for
a differential, balanced pair or cables
Observation point,
O
20
20
Lesson 4
Don’t Waste your Time Balancing a Twisted Pair
It is very easy to degrade cable twisting…:
How about allowing part of the signal current to
return through the reference structure (IG)
Unbalanced Circuit
Source
Or - twisting together
separate single-ended cables
E
ZS
Return currents of both
(IG(1+2)) flow through the reference structure
Load
I+
I-
ZL
IG=I+-I-
21
21
Lesson 5
Use Coaxial Cables for Low-Frequency Signals
It is great to believe in effectiveness of coaxial
cables in EMI reduction
Common-Impedance Coupling in a Low Frequency
Circuit using Multi-Point Grounding
22
22
Lesson 5
Use Coaxial Cables for Low-Frequency Signals
How about “floatation” of one circuit end – that
should do (and even simplify…) it
Motor PWM Current ,
IM
PWM Motor
Driver
M
PWM
Motor
"Floatation"
Motor Current ,IM
VNG
ZG
Floatation at One End to Eliminate
Common-Impedance Coupling *
* But the shield acts like a great antenna…
23
23
Lesson 5
Use Coaxial Cables for Low-Frequency Signals
How about “capacitive grounding” of one circuit end –
that should solve it once and for all…
Capacitive Grounding Exhibits LF Single Point
Grounding and HF Multi Point Grounding *
* How about just NOT using a coaxial cable…
24
24
Lesson 6
Don’t Waste your Money on Filters for Unshielded
I/O cables
I/O cables connected to electronic equipment should not
have any costly and bulky filters on them
After all, the equipment is installed in a shielded
enclosure, even of the cables are unshielded
Noise Current
Power
Source
Load
Noise Current
Parasitic
Capacitance
CM Noise Source
in Load
Load
Power
Source
DM Noise Source
in Load
Indeed, undesired conducted energy can still couple into
the system through I/O cables, but – is that worth the
money??
25
25
Lesson 6
Don’t Waste your Money on Filters for Unshielded
I/O cables
If you use filters on the I/O cables, you just might
suppress the conducted EMI on the cables
But if you do use a filter, don’t bother with
its grounding; why the effort?
A filter is just a passive circuit, right?
Just for the fun of it, make sure you ground it
via long wires…
26
26
Lesson 6
Don’t Waste your Money on Filters for Unshielded
I/O cables
Keep away from well-grounded filters… they may
Ω
actually work!
π-Section Low Pass Filter
ZS=50
90
L=2µH
80
RL=1mΩ
RC=3mΩ
Insertion Loss [dB]
70
C=0.5nF
60
ES
ZL=50Ω
1V
VL
50
40
Intended Path
LB=0.5nH
1
RB=1mΩ
30
}
ZB
Unintended
Path
2
20
Poor Bond Impedance, ZB
10
0
0.01
0.1
1
10
100
1000
Frequency [MHz]
27
27
Lesson 6
Don’t Waste your Money on Filters for Unshielded
I/O cables
Of course, you could use feedthrough filters, or
ferrite beads, but,
Oh - the cost…
Protected
Line
Signal Reference Structure
Standard Plate
Capacitor
Signal Reference Structure
“Metamorphosis”
Protected
Line
Enclosure
Structure
Protected
Line
Feedthrough
Capacitor
28
28
Lesson 7
Mount Filters (Particularly Power Line Filters)
Away from Equipment Boundary
Surely it is good enough to have filters on the power and
I/O lines… Who cares where they are placed, right?
If you wish to fail EMI, take the easy way out, and in
fact mount them a long way from the entry point
This will give a great opportunity to spread the …
noise
Feed-Through
You could even do
better…
Ground filters via
long wires…
Do not isolate input
and output leads of
the filter
Mounting
Bracket
Filters
EMI Coupling
29
29
Lesson 7
Mount Filters (Particularly Power Line Filters)
Away from Equipment Boundary
Of course there is the correct way, but that will
counteract your intention to fail EMI, right?
Put the filter at the entry point of the equipment and
isolate input from output wires
Equipment
Enclosure
Use filter connectors
Feed-Through
Filters
Equipment
Enclosure
Power Line
Input
Power Line Input to
Equipment
FILTER
Good Metal-to-Metal
Contact
Power Line Input to
Equipment
Doghouse
Entry
FILTER
Good Metal-to-Metal
Contact
Filters’ Doghouse
Mounting
Power Line
Input
30
30
Lesson 8
Allow Large Slots in Shielded Enclosures
Large slots in metallic enclosures are one of the best
ways to fail EMI
You can increase radiated coupling through such slots
and seams
Induced currents flow as long as there are no
obstacles in their path
So - any and all apertures must be arranged in such a
way as to maximize their effect on the currents
For increased effect route
wires just below the slot
31
31
Lesson 8
Allow Large Slots in Shielded Enclosures
Of course, if you’d like to pout some effort into
fixing this, there are some things you can do
But beware, they actually may work…
You can…
Break down slots into multiple smaller ones
Align cables with slot to minimize coupling
Use EMI Gaskets
And… ensure surface conductivity
32
32
Lesson 9
Use as Large as Possible Capacitors for
Decoupling
For the most ineffective high-frequency
decoupling, use the largest capacitors you have,
no smaller than 0.1µF, but preferably even
larger
Remember:
Nothing is like it seems at first…
That capacitor is bound to become inductive
soon enough…
33
33
Lesson 9
Use as Large as Possible Capacitors for
Decoupling
In general, the larger the value → the larger
the package → the larger inductance
100
LS (ESL)
C
RS (ESR)
Rp
Frequency response of
"Real World" (non-ideal)
Capacitors,
(C=10nF, LS=5nH and
RS=2mΩ)
Impedance [Ohms]
10
1
0.1
0.01
1
10
100
Frequency [MHz]
34
34
Lesson 9
Use as Large as Possible Capacitors for
Decoupling
10000
Impedance [Ohms] C=0.1uF, L=5nH, R=10mOhm
1000
Impedance [Ohms] C=10uF, L=25nH, R=100mOhm
100
Im pedance [Ohm ]
You can get even
larger capacitance, if
you use Tantalum or
Electrolytic
capacitors… but –
Do you think
performance is
improved?
Not quite!
10
1
0.1
0.01
1000
10000
100000
1000000
10000000
100000000
1000000000
Frequency [Hz]
35
35
Lesson 9
Use as Large as Possible Capacitors for
Decoupling
You can even do better, by increasing
inductance through the capacitor’s installation
Source: Wikipedia
The
Ugly
The
Bad
The
Good
Even
Better
Better
Still
Best
36
36
Lesson 9
Use as Large as Possible Capacitors for
Decoupling
But – if you change your mind, you can do
better:
Once you have chosen a package size for your
capacitor (e.g., 0603,
0402) use the largest
capacitance you can “buy”
Impedance curves for a Variety of
Values of X7R Capacitors
Impedance curves of 0.1µF Capacitors
Source: AVX Corporation
37
37
Lesson 10
Make Sure you have Large Slots and Gaps in
Return Planes on PCBs
If you REALLY want
to fail EMC, make
sure to have as many
gaps and slots in
return planes
You will not need
much more… This is
one of the “best…”
Source: Prof. T. Hubing, Clemson University
D. M. Hockanson, J. L. Drewniak, T. H. Hubing, T. P. Van Doren, F. Sha, C. W. Lam, and L. Rubin,
"Quantifying EMI resulting from finite-impedance reference planes," IEEE Transactions on
Electromagnetic Compatibility, vol. 39, no. 4, Nov. 1997, pp. 286-297.
T. Zeeff, T. Hubing and T. Van Doren, “Traces coupling across gaps in return planes,” accepted for
publication in the IEEE Transactions on Electromagnetic Compatibility.
38
38
Lesson 10
Make Sure you have Large Slots and Gaps in
Return Planes on PCBs
Antipad
Return Path in Reference
Plane
Pad
Source: Keith Armstrong, Cherry Claough
Signal Pin
Return
(GND) Pin
Signal Traces in
Signal Layer
Excessive Clearance in
Reference Plane
Reference Plane
Holes overlap each other large areas of discontinuity
39
39
Lesson 10
Make Sure you have Large Slots and Gaps in
Return Planes on PCBs
?
Top Layer
Bottom Layer
Power Plane
Return (GND)
Plane
Just make the WRONG Choice
?
40
40
Lesson 10
Make Sure you have Large Slots and Gaps in
Return Planes on PCBs
1kV ESD injected onto PCB
with and without split
Noise coupled into a test
circuit was measured
Source: “ESD and EMI Effects in Printed
Wiring Boards”, by Douglas C. Smith
41
41
Lesson 10
Make Sure you have Large Slots and Gaps in
Return Planes on PCBs
3.7V Pk
170mV Pk
Increased current
loop size increased
noise coupling
Violation of the path
of least impedance
Source: “ESD and EMI
Effects in Printed Wiring
Boards”, by Douglas C. Smith
42
42
Trace crossing
the gap
<2.5 mm from
the signal trace
Reference Plane, e.g.
Digital Return, DGND
Reference Plane, e.g.
Analog Return, AGND
Single Point (Star) Connection between
AGND and DGND Reference Planelets
If you want to fix this …
Rearrange layers
Change routing, or –
bypass
(Source: Keith Armstrong,
Cherry Clough)
Plane Stitching
Capacitors
Trace crossing
the gap
Split in Plane
<λ/10
<λ/10
<λ/10
λ/10
Pads for fitting
small capacitors
Slot
Lesson 10
Make Sure you have Large Slots and Gaps in
Return Planes on PCBs
Galvanically
<2.5
mm from
Isolated
the
signal
Referencetrace
Planelet
Main Reference Plane
Stitching Capacitors across a
Slot in a Reference Plane
43
43
Lesson 10
Make Sure you have Large Slots and Gaps in
Return Planes on PCBs
C1
C3
C2
Signal Trace
L1
Signals
Gap
L2
L3
Increased Loop
Power
Gapped Power Plane
Return
Solid Return Plane
L4
Signals
Preferred Return Current Path
(Through non-existing C2)
Actual Return Current
Path (Through C3)
Although a solid ground plane is present, the RETURN
current must seek its way through when the adjacent
power planes are slotted/split
44
44
Lesson 10
Make Sure you have Large Slots and Gaps in
Return Planes on PCBs
Comparison of Maximum Radiated E-Field for Microstrip
With and without Split Ground Reference Plane
120
But…
even
bypassing has
its limits
110
Maximum Radiated E-Field (dBuv/m)
100
90
80
Comparison of Maximum Radiated E-Field for Microstrip
With and without Split Ground Reference Plane and Stiching Capacitors
70
120
No-Split
60
Split
110
50
100
30
20
10
100
Frequency (MHz)
Source: Dr. Bruce Archambeault
1000
Maximum Radiated E-Field (dBuv/m)
40
90
80
70
60
No-Split
Split
Split w/ one Cap
Split w/ Two Caps
Split w/One Real Cap
Split w/Two Real Caps
50
40
30
20
10
100
1000
Frequency (MHz)
45
45
“Bonus” Lesson
Count on “Rules of Thumb”
Just tell me what rules I need to
follow to ensure that I don’t have
EMC-related problems with my
system design.
Just tell me what rules I need to
follow to ensure that I don’t have
health-related problems with my
brain surgery.
Source: Prof. T. Hubing
Clemson University
46
46
Summary of Lessons
1)
2)
3)
4)
5)
6)
Minimize Edge Rates, Particularly of Clocks
Encourage Common-Mode currents Flow
Always Terminate Cable Shields with “Pigtails”
Don’t Waste your Time Balancing a Twisted Pair
Use coaxial cables for low-frequency signals
Don’t waste your money on filters
for unshielded I/O cables
7) Mount filters (particularly power
line filters) away from equipment
boundary
8) Allow large slots in shielded enclosures
9) Use as large as possible capacitors for decoupling
10) Make sure you have large slots and gaps in return planes on
PCBs
47
47
“Bonus” Lesson
Count on “Rules of Thumb” and
forget Maxwell…
Just tell me what rules I need to
follow to ensure that I don’t have
EMC-related problems with my
system design.
Just tell me what rules I need to
follow to ensure that I don’t have
health-related problems with my
brain surgery.
Source: Prof. T. Hubing
Clemson University
48
48
“Bonus” Lesson
Count on “Rules of Thumb” and
forget Maxwell…
49
49
“Bonus” Lesson
Count on “Rules of Thumb” and
forget Maxwell…
In the (very) beginning, God created the Heaven
uur
and the Earth …
∇⋅ D = ρ
uur
… and God Said, Let…:
∇⋅ B = 0
uur
∂B
∂t
ur
uur ur ∂ D
∇× H = J +
∂t
is based on
And there EMC
Maxwell’s Equations
Forget these and
NOTHING ELSE
was light!
WILL HELP YOU!
uur
∇× E = −
50
50
QUESTIONS
51
51
for your Attention!!!
I am glad to have Participated
in this meeting
52
52