Division for Electricity and Lightning Research

Uppsala University

BLOCK II: Reasons of EMI

RT/2003

Division for Electricity and Lightning Research

Uppsala University

3. High-frequency behaviour of electrical

Components
(Non-ideal Behavior of Electrical Components)
3.1 Introduction
We use physical components such as conductors, resistors, inductors, capacitors, and switches
in electrical circuits. The symbols used in circuit diagrams to represent these components
show only their ideal properties. For example, conducting wires used between components or
in component leads have some resistance, however small it may be. While carrying high
frequency currents, the charge carriers may crowd toward the periphery of the conductor
increasing the resistance of the wire. Associated with the currents and charges in the wire
there are magnetic fields and electric fields, and hence, inductance and capacitance. We know
that at high frequencies inductance has high impedance and capacitance has low impedance.
Therefore the behavior of the conductors and circuit components used in electrical systems
may depart from the ideal behavior at high frequencies. This has several EMC consequences.
This chapter will look into the high frequency behavior of components. The following
components are examined:
Conductors
Capacitors
Inductors
Resistors
Mechanical switches
Transformers

3.2 Conductors
3.2.1 Resistance of wires
Current carrying conductors are always accompanied by electromagnetic fields and it is these
fields that determine the parasitic properties of a conductor. Consider a current carrying plate
as shown. When direct current is flowing in the conductor, the current density at same
distance from the terminals will be uniform across the plate cross section. An alternating
current will create magnetic fields inside the conductor, exerting a force in the charge carriers
and causing it to move towards the surface (the skin effect). The current density now follows
an exponential function,
x

J ( x) = J o e

(3.1)

where J o is the current density at the surface and is the skin depth, given by

RT/2003

(3.2)

Division for Electricity and Lightning Research

Uppsala University

The current I in the conductor is given by,

d

I = J ( x )dx = J o e
0

dx = J o (1 e

(3.3)

Conducting Plate
thickness

J(x)/J0
1
1/e

Fig. 3.1 Variation of current density as a function of depth in a metal plate, due to skin
effect

If the conductor is very thick compared to the skin depth, that is d >> , then I Jo . It
means that the current appears to flow uniformly in a skin of thickness . From equation
(3.3) it is evident that 99% of the current flows in a skin of thickness 5. For example, at 1
MHz 99% of the current in a copper conductor flows in a skin of 0.33 mm thick. Therefore,
from an electromagnetic point of view, it is not useful to choose a conductor much thicker
than this for carrying 1 MHz current. The d.c. resistance of a conductor is given by

Rdc =

l
A

(3.4a)

where l is the length and A is the cross sectional area of the conductor. The d.c. resistance
per unit length of a wire of circular cross section is
Rdc =

1
r 2

/m

(3.4b)

At low frequencies,

RLF Rdc

RT/2003

if

r <<

(3.5)

Division for Electricity and Lightning Research

Uppsala University

At high frequencies,

RHF =

r (r )
2

/m

if

r >>

(Almost all current confined to one skin-depth deep)

RHF

1
/m
2r

(3.6)

RHF
r
=
Rdc
2

(3.7)

Also,
RHF =

r
r 1
1
Rdc =
=
2
2
2 r
2r

f /m

(3.8)

Equation (3.8) indicates that RHF f , that is the resistance at high frequencies ( r > 2 )
increases 10 dB per decade increase in frequency.

103

RHF/RDC

Fe, d=20 mm

102

Cu, d = 20 mm

Fe, d=2 mm

101

Cu, d = 2 mm
Fe, d = 0.2 mm
Cu, d = 0.2 mm

100
101

102

103

104

105

106

107

Frequency (Hz)

Fig. 3.2 Examples of RHF/RDC for copper (thick lines) and iron (thin lines) for three
values of wire diameter. For copper r=1, r=1 and for Iron r=500, r=0.1.

RT/2003

Division for Electricity and Lightning Research

Uppsala University

3.2.2 Inductance of wires

Internal inductance of the wire is due to magnetic flux internal to the wire. For r << (at
low frequencies) the current in the wire can be assumed to be uniformly distributed. The
magnetic flux lines inside the circular conductor encloses only a portion of the total current
(x2I/r2). It can be shown that the internal inductance of the wire is,

L int =

H/m
8

(3.9)

For copper or aluminium, where r = 1, Lint = 50 nH/m.

Equation (3.9) gives the value of internal inductance for d.c. or low frequencies
where the current distribution across conductor dross section is uniform. At high frequencies
the current is concentrated in a thickness at the surface and the per-unit internal inductance
is given by,

Lint =

4r

That is, Lint =

for r >>

(3.10a)

2
Lint,dc
r

(3.10b)

dx

Lint =

1
1
H/m
4r f

From (3.10c) Lint,HF 1

(3.10c)

. That is, at high frequency the per-unit internal inductance

decreases at a rate of 10 dB/decade for r >> . High frequency internal inductance is

smaller than low frequency internal inductance.

RT/2003

Division for Electricity and Lightning Research

Uppsala University

r
r <<
/m
1

B
0d

H/m
L int. dc =

R dc =
=

r >>

e
cad
e
d
/

10
dB
/de
cad
e

=
8

1
r 2
2

r/

r/

Fig. 3.3 Effect of skin effect on the resistance and internal inductance of round wires.
External inductance of wires
Consider a segment of wire of length l , which is part of a closed loop where the return
conductor is at infinity (very far away). The mutual inductance of the segment of the wire
(partial mutual inductance) is given by,

Lij =

m
Ij

Bij dsi

si

(3.11)

Ij

where s i is the area bounded by conductor i and infinity and m is the flux penetrating the
surface s i due to current in conductor j , I j . The mutual inductance between the i th and j
th conductor, assuming free space as the medium between them, is given by

d
o l l
l
d
Lij =
ln + + 1 + + 1
d
l
2 d

(3.12)

where d is the distance between the two conductors. Self-partial inductance is obtained by
replacing d by the radius of the conductor r .

r
o l l
l
r
ln + + 1 + + 1
r
l
l
2 r

Lii =

RT/2003

(3.13)

Division for Electricity and Lightning Research

Uppsala University

Ij

Si

Bij

Fig. 3.4. Illustration of the concept of partial inductance

If the conductor is very much longer than the spacing d or radius r , we get the simplified
expressions,

Lij =

d
o l 2l
<< 1
ln 1 when

2 d
l

Lii =

o l 2l
ln 1
2 r

when

r
<< 1 ,
l

(3.14)

<< r

(3.15)

In the case of rectangular cross section,

Lii =

b t
o l 2l 1
b+t
, << 1
when
ln
+ + 0.22

2 b + t 2
l
l l

(3.16)

Partial inductance of wire segments can be combined to get the total loop inductance. For
example consider a rectangular loop of wire with sides p, q, r, s as shown in Fig. 3.5a. The
partial self and mutual inductance of the wires are shown in Fig. 3.5b.

Lpp
p

I
r

I
RT/2003

Lss

Lsq

Lpr

Lqq

Lrr
7

Division for Electricity and Lightning Research

Uppsala University

Fig. 3.5a. Physical dimensions of the loop

Fig. 3.5b. Equivalent circuit of the loop

The voltage across a gap in the loop is calculated as,

V = Lss

dI
dI
dI
dI
dI
dI
dI
dI
Lsq + Lpp Lpr + Lqq Lqs + Lrr Lrp
dt
dt
dt
dt
dt
dt
dt
dt

(3.17)

Note that Lpp = Lrr, Lss = Lqq, Lsq = Lqs, and Lpr = Lrp.
Therefore,
V = [ 2 ( Lpp Lpr ) + 2 ( Lqq Lqs )]
= Lloop

dI
dt

(3.18)

dI
dt

If l>>d, the loop inductance of rectangular loop is given by

L=

0l d
ln( ) H

(3.19)

which is a similar expression for the per-unit length inductance of a parallel wire transmission
line, L = (0/) ln(d/r) H/m.
The internal inductance due to magnetic flux within the wire material also contribute to the
self-inductance. However, as seen previously, internal inductance for non-magnetic materials
is about 50 nH/m, negligible compared to the external inductance. Also, if the medium
surrounding the wires and between the wires are other than free space, 0 in equations (3.12)
to (3.19) has to be replaced by 0r.

3.2.3 Capacitance of conductors

The capacitance between conductors with equal and opposite charge is defined as,

C=

[ ]

Q
F m
V

[ m]

where Q is the charge per meter C

V = V1 - V2

In conductors charges can be assumed to be confined to the surfaces. Electric field at a

distance x from a round conductor is,

Ex =

Q
2x

[ ]

V m

Gauss Law.

Potential difference between two points at D1 and D2 ,

RT/2003

Division for Electricity and Lightning Research

Uppsala University
D

V12 = Ex dx =
D1

D2

D2

2x dx = 2 ln D

D1

[V ]

Capacitance between two long conductors:

Assume r1 , r2 << D so that charge distribution on the conductor surfaces are uniform.

V12 =

r
Q D
ln ln 2 [V ]
D
2 r1

V12 =

Q
D2
ln
2 r1 r2

C 12 =

Q
=
V12

[ ]

2
F m
2
D
ln

r1 r2

If r1 = r2 ,

C 12 =

D
ln
r

[ ]

F m

when D >> r

Note that the capacitance of a conductor depends on:

1) The shape of the conductor cross section. Larger surface area can give greater capacitance.
2) Configuration (topology) of the circuit. Two-wire circuit, one wire with ground return, etc.
3) Surroundings of the circuit. Influence of . The presence of a conductor plane near a
circuit, etc.
Also note that unlike an internal inductance, we do not have an internal capacitance,
because the charges are on the surface and electric fields are assumed not to penetrate the
conductor. Each conductor cross-section is assigned a unique potential.
Expressions for capacitance for various configurations

RT/2003

Division for Electricity and Lightning Research

Uppsala University

d
C12
D
h
Cr

Cr

C 12 =

2h
2 ln + 1
D
4 h 2h 2

ln + 1
d D

[ ]

F m

The total capacitance between the two conductors is,

C t = C 12 +

Cr
=
2

[ ]

4h
ln

F m

d
2

2 h
+1
D

Examples:
a) Two long round conductors of length l in free space (no surrounding circuits or objects)

C 12 =

l
2D
ln

[ ]

F m

when D >> d

Let us consider two long conductors in free space with a diameter d = 0.001 m each
conductor. The capacitance C 12 in between them varies according to the separation D .
Fringing effects at the ends are neglected.

D = 0.005 m, then C 12 = 12 pF/m

D = 0.01 m,

then C 12 = 9.3 pF/m

D = 0.02 m,

then C 12 = 7.5 pF/m

C12

b) Introducing a ground plane close by,

RT/2003

10

Division for Electricity and Lightning Research

Uppsala University

C t = C 12 +

[ ]

Cr
F m
2

Let us consider two long conductors in free space with a diameter d = 0.001 m each
conductor and h = 0.002 m over the ground. Then for D = 0.005 m, the capacitance
involved in the circuit are,

C 12 5 pF/m

C r 22 pF/m

C t 16 pF/m
For D = 0.01 m, the capacitance involved in the circuit are,

C 12 2 pF/m
C r 26 pF/m

Cr

C12
Cr

C t 16 pF/m
For D = 0.02 m, the capacitance involved in the circuit are,

C 12 5 pF/m

C r 22 pF/m
C t 16 pF/m
The proximity of ground increases the total capacitance Ct of the system, which is mainly
determined by capacitance Cr to ground for large values of D compared to h . However,
capacitance C12 between the conductors is reduced due to the proximity of the ground.

Rule of thumb:
Capacitance values of conducting wires are usually 5-15 pF/m. The capacitance depends
strongly on distance D up to

D
= 10 . Beyond that capacitance values fall off slowly.
d

Characteristic impedance, Z0
Configuration:

RT/2003

D/d or h/d

Z0 ( )

166

276

10

359

11

Division for Electricity and Lightning Research

Uppsala University

r 2D
L
= 120
ln

d
C
r

Zo =

One conductor above ground,

Z o = 60

125

180

10

221

r 4 h
ln
d
r

where r and r are the properties of the medium between the conductors and ground plane.
The external inductance and capacitance of component leads or connecting wires can form
transmission lines with characteristic impedance Z0.

RT/2003

12

Division for Electricity and Lightning Research

Uppsala University

Example: Comparison of inductance of circular and rectangular wires

<<d
Long circular conductor in
free space (return far away

L=

l
2

4l
ln( ) 1

Long rectangular conductor in free

space (return far away)

L=

l
2l
1
b + t
) + + 0.22
ln(

2 b + t 2
l

Long circular conductor (return

close by)

L=

l 2 D D
)
ln(

d
l

D=1 mm, L=1.5 H/m

b= 10 mm, t=2 mm, L=1.12 H/m

(S=0.024 m)

d=1 mm, D=4 mm, L=0.83

H/m

D=10 mm, L=1.0 H/m

b=40 mm, t=2 mm, L=0.87 H/m

(S=0.084 m)

d=10 mm, D=40

mm,L=0.82H/m

D=7.65 mm, L=1.05

H/m (S=0.024 m)

b=80 mm, t=1 mm, L=0.74 H/m

(S=0.162 m)

Note: = 0r - property of surrounding medium, S = perimeter

Observations:
Large surface area tend to reduce external inductance
(Easier to achieve large surface area by strip/tape)
Ground wire inductance can be reduced if run close to an earth plane or metal cabinet or
return conductor
{Reduce loop area to reduce induced voltages
Remember:
v = loop areadB/dt (due to external field)
v = Ldi/dt (due to current in the wire)}
Rule of thumb:
Inductance of a conductor 1 H/m (1 nH/mm)
Reducing the length is most effective in reducing the inductance (impedance) of earth wires

1 m earth wire has about 126 impedance at 20 MHz compared to < 1 at power
frequencies
Safety earthing is not enough for EMC

RT/2003

13

Division for Electricity and Lightning Research

Uppsala University

1.6

400

1.4

Wire diameter

300

1.0
0.8

200
L strip

0.6

L wire

0.4

Wire diameter (mm)

Inductance (H)

1.2

100

0.2
0.0

100

200

300

400

500

Strip width (mm)

Fig. 3.6 The inductance of a strip (foil) compared with that of a wire of circular crosssection, when strip and wire have same circumferences and length 1 m. Strip thickness = 0.2
mm.

RT/2003

14

Division for Electricity and Lightning Research

Uppsala University

3.3 Capacitor
Capacitor is a discrete component ideally providing an impedance inversely proportional to
the frequency

Z ( jw ) =

Z =

1
1
=
90
jwC wC

1 1

2C f

1
20 log Z = 20 log
20 log( f )
2C

|Z| dB
0
-20 dB/decade

-90o

f
Fig. 3.7 Impedance magnitude and
phase of an ideal capacitor

Magnitude of the impedance of an ideal capacitor decreases by 20 dB per decade increase in

frequency. Capacitors are often used to provide a short-circuit path (very low impedance) at
frequencies beyond certain value. A capacitor uses conducting wires and dielectric in its
construction. Therefore high frequency behaviour of capacitors are far from ideal. The
general high frequency equivalent circuit of a capacitor is as shown,

L le

L li

L le

Cl
where

RT/2003

15

Division for Electricity and Lightning Research

Uppsala University

Lle is the inductance of the external connecting lead.

Lli is the inductance of the internal connecting lead and other parasitic inductance due to
method of construction.

R is the resistance of the capacitor plates and connections.

C is the nominal value of the capacitance
Rd Bulk resistance of the dielectric

C l is the capacitance of the external lead.

Usually C l is very small and Rd is very large. Therefor they can be neglected. In short, the
equivalent circuit for a capacitor is an inductance, capacitance and resistance connected in
series

-20 dB/decade

||Z| dB

+20 dB/decade

Capacitive

Inductive

RS
fr

fr

Z
+90o
0
-90o

Fig. 3.8 Impedance magnitude and phase of a real capacitor

Z ( jw ) = R + jwL +

1
jwC

At series resonance, wL =

RT/2003

1
, thus
wC

16

Division for Electricity and Lightning Research

Uppsala University

resonance frequency, f r =

1
2 LC

The impedance at resonance is equal to R . Below resonance frequency,

1
> wL
wC

and capacitance effect is dominant. Above resonance frequency, wL >

1
wC

and inductance effect is dominant.

The parasitic inductance of a capacitor is mainly determined by the length of the leads and
hence do not change. Therefore, for a given type of capacitor, the resonant frequency tend to
decrease with increase in capacitance value. This effect has important practical consequences
which is shown in the following example. Suppose a 470 pF ceramic capacitor is used as a
shunt element to divert high frequency noise currents. The loop formed by the component
leads (12 mm length and 6 mm separation) has a parasitic inductance of about 14 nH. The
capacitor will resonate at a frequency of 62 MHz and at this frequency it has the minimum
impedance. As the frequency is increased beyond the resonance frequency, the impedance of
the capacitor start increasing and it behave more like an inductor. The above capacitor may
be suitable for providing a low impedance path to noise currents at, say 60 MHz, but may not
be suitable for noise currents at, say 200 MHz.
There are many practical considerations in the use of capacitors to divert noise currents. They
are,
1) Self-resonant frequency of the capacitor
2) In low-amplitude signal applications, the capacitor should not be placed in such a way so
that it forms a loop with other components serving as a receiving antenna for radiated
electromagnetic Interference (EMI).
3) Capacitor in parallel with the inductance of the cable it is protecting can form a parallel LC circuit and can produce ringing due to resonance.
4) Shunting capacitors work best in high impedance circuits (noise current division between
the circuit impedance and the capacitor impedance).

RT/2003

17

Division for Electricity and Lightning Research

Uppsala University

3.4 Inductors
The impedance of an ideal inductor is given by,

Z L = jwL = wL + 90
20 log Z L = 20 log( 2L ) + 20 log( f )

Z
|Z| dB

+90o

+20 dB/decade
f

Fig. 3.9 Impedance magnitude and phase of an ideal inductor

The magnitude increases at a rate of 20 dB per decade increase in frequency. The general
circuit for the inductance is as shown,

Rs

Cp

|Z| dB

Inductive

Capacitive

Resistive
RS
Z

RS
2L

fr

+90o
0
-90

RT/2003

fr

18

Division for Electricity and Lightning Research

Uppsala University

Fig. 3.10 Impedance magnitude and phase of a real inductor

where,

L is the nominal value of the inductance (much higher than lead inductance).
Rs is the loss resistance (eddy current losses resistance of the wires).
C p is the effective parasitic capacitance (capacitance between turns, between turns and the
core). This also known as lead capacitance.
Inductors behave as a parallel-resonant circuit,

Z L ( jw ) = Rs + jwL / /

1
jwC

1
jwC
Z L ( jw ) =
1

Rs + j wL

wC

( Rs + jwL)

At resonance, wL =
given by

fr =

1
and the impedance Z L is a maximum. The resonant frequency is
wC

1
2 LC

1
and the inductor predominantly has an
wC
1
and the inductor behaves
inductive character. Above the self-resonant frequency, wL >
wC
Below the self-resonant frequency, wL <

as a capacitance. At very low frequencies the inductive impedance may be even lower than
the series loss resistor, and the inductor behaves more like a resistor.

Inductors are used to block noise current and are used in series, where as
capacitors are used to divert (shunt) noise currents and are used in parallel. Inductors behave
as a capacitance above the self-resonant frequency whereas capacitor behaves as an
inductance above the self-resonant frequency. Increasing the value of an inductor do not
necessarily give a higher impedance at a given frequency, since the large value of the
inductance lower the self-resonant frequency of the inductor. Practical considerations in the
use of inductor to block noise currents are:
1) Self-resonant frequency of the inductor

RT/2003

19

Division for Electricity and Lightning Research

Uppsala University

2) The impedance of the inductor at noise frequencies should be larger than the impedance of
the circuit it is protecting. That is, blocking series inductors are most effective in the lowimpedance circuits.
3) The possibility of ringing due to resonance formed by parallel or series (parasitic)
capacitance of the circuit.
I

I
Zhigh
V

Zhigh

Zlow

Zlow

In a series circuit I is mostly

In a parallel circuit I is mostly

determined by Zhigh

determined by Zlow

4) In inductors with ferromagnetic cores, the non linear effect of saturation must be
considered.

B
I
N

H or NI

B
H

where is the slope of the B H curve and is also a function of B ( B ) . At low values

of magnetic flux density, slope of the B H curve is large, that is r is large. As the flux
density is increased the core saturates and r is small. Therefore in inductors with Ferromagnetic core the inductance value decreases with increasing current.

RT/2003

20

Division for Electricity and Lightning Research

Uppsala University

3.4.1 Chokes and ferrites:

Magnetic fields tend to concentrate on high permeability materials. Ferro-magnetic rings are
very useful in suppressing the unwanted common-mode (CM) noise in cables. The wires are
wound through the core in such a way that the fluxes due to the CM currents add in the core
whereas the fluxes due to differential mode (DM) currents or signal currents subtract in the
core. Since almost all the flux is confined to the core, the self and mutual inductance of the
winding are the same.

LM

IIDD
ID

IC
ID

ID
IC

ID

ID

L
M
L

Z DM = j ( L M )

per winding

Z CM = j ( L + M )

per winding

IC
IC

IC
IC

RT/2003

21

Division for Electricity and Lightning Research

Uppsala University

The differential mode currents produce fluxes in opposite direction, and in the ideal case they
exactly cancel each other and therefor the mutual inductance is negative and cancels the self
inductance part. Therefore the choke do not present any impedance to differential-mode
currents. In the case of common-mode currents, the fluxes set up by the individual currents
add up and therefor the mutual inductance is positive. The choke presents a series impedance
of Z CM = jw( L + M ) = jw 2 L per winding to common-mode currents. This type of choke
is called common-mode choke.
Generally the differential-mode currents are much higher than common-mode currents. The
fluxes due to large DM currents cancel in the core. Therefore the core is usually not driven
into saturation. The core of the choke is made of high permeability materials. Permeability is
a function of frequency. For example, r of two materials are shown in the figure.
Depending on the frequency of application, proper materials for the ferrite core has to be
selected

RT/2003

22

Division for Electricity and Lightning Research

Uppsala University

3.5 Resistors
There are basically three types of resistors, They are a) carbon composition, b) wire wound
and c) thin metal film. Ideally a resistor has an impedance equal to the resistance value and a
phase angle zero at all to the frequencies. That is, Z = R0 . Therefore the voltage across
the resistor is proportional to the current through it. This property is used in the measurement
of current in a circuit using shunt, and for deriving control signals proportional to the current.
However, the behaviour of a practical resistor is different from ideal and very much depends
on the construction method. The equivalent circuit of a real resistor is given below, where

L we

R
Cp
LC

R is the nominal value of resistor

Lwe is the inductance of connecting leads. Its typical value is about 15 nH.
LC is the inductance of the resistance element itself. For example, inductance of the wire in a
wire-wound resistor. Usually negligible except for wire-wound resistors.

C p is the parasitic capacitance which include the lead capacitance and leakage capacitance of
the resistor body. Its typical value is in the range of 1-2 pF
Behaviour of resistor at various frequencies

RT/2003

23

Division for Electricity and Lightning Research

Uppsala University

L we

Cp

At d.c., inductance acts as a short circuit and capacitance as an open circuit. As the frequency
is increased, the impedance of the capacitor decreases and at a frequency

f =

1
2RC par

and the impedance of the capacitance and resistance became equal. Above this value of
frequency the capacitor impedance become dominant (i.e. less than R ) and the net impedance
decreases by approximately -20 dB/decade and the phase angle approaches 90 . The
model inductance and capacitance resonate at

fo =

1
2 Lwe C p

and the impedance is a minimum. Above the resonant frequency, the inductive impedance
become dominant and the magnitude of the impedance increases by +20dB/decade and the
phase angle approaches + 90 .

|Z|
+90
R

0
Resistive

Capacitive

1
2RCp

RT/2003

-90

Inductive

1
2 L we C p

Resistive Capacitive Inductive

f
1
2RCp

1
2 L we C p

24

Division for Electricity and Lightning Research

Uppsala University

Starting from d.c., as the frequency is increased the inductive impedance in the resistor model
is also increasing, while the capacitive impedance is decreasing. For a given value of the lead
inductance and parasitic capacitance, the behavior of the resistor depends on the nominal
value of the resistance. For low value resistors, the capacitance become dominant in the
parallel RC circuit only at very high frequency. In this case, the resistor goes from the
resistive phase to the inductive phase directly. A numerical example is given below.

C p = 2 pF
Lwe = 20 nH

fo =

1
= 796 MHz
2 Lwe C p

1
2RC

79.6 GHz

> f o Resistive Inductive

100

796 MHz

= f o Resistive Inductive

1000

79.6 MHz

< f o Resistive Capacitive Inductive

For low value resistors the inductive reactance became dominant even before the resonant

1
is reached. For example, at 1 MHz, a 20 nH inductance has an
2 LC
impedance of 0.125 . If the nominal value of the resistor is 0.1 , the resistor is behaving
frequency

mostly as an inductance.

At high frequencies and below the resonant frequency:

1) Capacitance is the main problem for high value resistors (inductive impedance is only a
small fraction of the nominal resistance value.
2) Inductance is the main problem for low value resistors (capacitive impedance is many
times the nominal value of the resistor).
To avoid the above two problems, the leads can be shorted as much as possible and connect
many smaller value resistors in series to get high resistance value (parasitic capacitance in
series). Also, connect many higher resistors in parallel to get low resistance value (lead
inductances in parallel).

RT/2003

25

Division for Electricity and Lightning Research

Uppsala University

3.6 Mechanical switches

The most common type of mechanical switches employs solid metallic contacts operating in
air at normal atmospheric conditions. During switch closure and opening a spark gap with
variable gap length is created between the contacts. Arcing at the contacts is the primary
interference problem in switches. Arc current/voltage waveforms contain high frequency
spectral components of large magnitude and produce conducted and radiated emission
problems. In a real switch, a single closing or opening operation is actually a rapid series of
sliding and rocking. Moreover, the electrical characteristics of the load and switch
connections also play an important role in influencing the noise characteristics.

During the operation of a switch, two types of electrical discharges can happen between the
switch contacts: the glow discharge and the arc discharge. In general glow discharge is a
function of gas in between the contacts and is characterized by relatively high voltage and
low current. In contrast, arc discharge is a function of contact material (can happen even in
vacuum) and is characterized by relatively low voltage and high current.

Glow phase
The voltage necessary to initiate a glow discharge is a function of the nature of gas, gas
pressure, and distance between contacts. The minimum voltage VB, required to start a glow
discharge is about 320 V, in air at normal temperature and atmospheric pressure and for a
spacing of about 8 m. VB is higher for contact spacing less than or more than this value.
After the initiation of glow discharge, a slightly lower value of voltage VG is required to
sustain the glow discharge. In air and at normal atmospheric pressure VG is approximately
300 V and is nearly constant with varying contact spacing. Therefore to avoid a glow
discharge, the voltage across the contacts should be kept below 300 V.

Arc phase
An arc discharge or metal vapor discharge is initiated by field-induced electron emission and
requires a voltage gradient of approximately 0.5 MV/cm. This translates into about 400 V
for a gap distance of 8 m. The electric field is enhanced at microscopic projections on the
cathode surface and act as efficient sources of electrons. The current density at small regions
can be very high, raising the local temperature to a few thousand degrees Kelvin. The metal
melts and evaporates forming a bridge between the contacts. Regions on the anode may also
melt due to electron bombardment. In general, either the anode or cathode may vaporize first
depending on the heat generation and heat removal rates at the contacts. The transition from
field emission to a metal vapor arc can happen within a nanosecond. The minimum voltage
VA and current IA to sustain an arc for different contact materials are given below. Note that
the minimum arcing currents are applicable to clean undamaged contacts and for
contaminated damaged contacts this value may be as low as one-tenth of the value given in
the table.
RT/2003

26

Division for Electricity and Lightning Research

Uppsala University

Table 3.1: Contact arcing voltage and current values

Material

Minimum arcing voltage, VA

(Volts)

Minimum arcing current, IA

Silver

12

400

Gold

15

400

Palladium

16

800

Platinum

17.5

700

(mA)

To extinguish the arc the voltage or current has to be brought below the arcing value. This is
easier in ac operation because of the natural zero crossing. The damage due to the arc on the
contacts is proportional to the energy, that is, proportional to the product of voltage, current,
and time. A given set of contacts can handle higher ac voltages than dc voltages. For
example, a contact rated for 30 V dc can typically handle 115 V ac.
Voltage-distance relationship for contact breakdown
The relationship between the voltage across the contacts and the distance between the
contacts governing the contact breakdown is shown in the figure 12. For transition into arc
phase, a minimum voltage gradient of 0.5 MV/cm is assumed. The heavy line represents the
composite requirements for producing contact breakdowns. Note that the separation distance
scale (x-axis) can be converted into time if we assume a speed for the moving part of the
contact.

Arc Discharge
Breakdown
LOG of
Contact
Voltage,
Vc
300 V

Glow Discharge
Breakdown
Glow Discharge
Sustaining Voltage
Composite Breakdown Characteristic

10 V

Arc Discharge Sustaining Voltage, VA

LOG of Time or LOG of Separation Distance

Fig 3.12. Voltage-time (or distance) relationship for contact breakdown

The contacts breakdown if the available voltage across the contacts is greater than the
breakdown characteristics (heavy line). Breakdown forces the voltage across the contacts to a
lower value given by breakdown characteristics. If the available voltage across the contacts is
RT/2003

27

Division for Electricity and Lightning Research

Uppsala University

always below the breakdown characteristics, the contacts open or close without any
breakdown. The actual operation of the switch is more complicated due to load
characteristics and parasitic inductance and capacitance of the wiring.
Load characteristics
Some loads draw high inrush currents when switched on, for example, cold filament lamp,
motors, and capacitors. The inrush current in a cold filament lamp can be 10-15 times higher
than the operating current when the filament is hot. Motor starting currents may be 5-10
times higher than normal rated currents. Capacitor inrush current is only limited by series
resistance.
If the load is inductive, opening the switch create large di/dt and associated high voltage
inductive transient. For example, a typical inductive load operating from a 26-V dc supply
can generate maximum voltages in the range of 500-5000 V when the current is interrupted
suddenly.
Real mechanical switch
A complete switching circuit is shown below.

L2

R2

C2

VS

S1

V2

V1

C1

R1
L1

S1 is the moving part of the switch. L 2 , R 2 and C 2 represent the parameters of the line
leading up to the switch contacts and L1 , R 1 and C1 represent the parameters of the line after
the switch and the load. Typically, L 2 , R 2 and C 2 are much less than L1 , R 1 and C1 . The
resonant frequencies associated with the source side and the load side of the switch contacts
are:
f2 =

1
,
2 L 2 C 2

f1 =

1
2 L1C1

f 2 > f1

The available voltage across the switch is the vector sum of voltages V1 and V2. When the
switch is first closed, the initial current may rise rapidly to high peaks due to C1 and C 2 .
After that it decays to normal load current as damped oscillatory current with an approximate
1
frequency of f 3 =
. Since the first area of contact may be very small, the rapidly
2 L 2 C1
rising current may melt and vaporize tiny amounts of metal from the contacts producing an
arc and the voltage across the switch drop to typical arc voltages (e.g. 12 V). The bouncing
RT/2003

28

Division for Electricity and Lightning Research

Uppsala University

(jumping) of the switch contacts (due to vapor pressure, elasticity, etc.) make and break the
contact many times before finally the switch is closed. Damped oscillations with frequency f 3
may be repeated with each bounce, but progressively reducing in maximum amplitude in
succeeding bounce.

When the switch is opened the current density in the last area of contact may be high enough
to cause metal vaporization. When this happen an arc-impedance is introduced in the circuit.
The sudden voltage drop across the arc excites the resonant circuits, primarily f3. Subsequent
reclosures establish load currents again. When the current is being interrupted, the voltage
across the switch is changing and it can be greater than either V1 or V2. The voltage across
the switch may alternatively be below and above the breakdown characteristics breaking and
making the load currents. This may have a frequency f4 (load relaxation oscillation) decided
by V1 at f1 and V2 at f2.
If the load current is below vaporization level, no arc is established, but opening of contacts
initiates resonances with frequencies f1 and f 2 . The resonant voltages on both sides of the
switch contacts can produce voltages in excess of the breakdown voltage of the switch
contacts in open position, which results in sparking.
Noise suppression:
Switch operation produce conducted and radiated emission. The fundamental requirements
for avoiding contact breakdown are 1) keeping the contact voltage below 300 V to prevent
glow discharge, and 2) keeping the initial rate of rise of voltage below the value necessary to
produce an arc discharge (A value of 1V/s is usually sufficient for most contacts). If the
breakdown happens the next effective strategy to reduce noise is to make the current available
always below that necessary to sustain the breakdown.
Usually the action taken to protect the contacts also reduces conducted and radiated emission.
To protect the contacts that control inductive loads, shunt elements, such as diodes, avalanche
diodes, varistors and capacitors, are connected across the load, which divert the transient
currents generated in the load away from the contacts.

RT/2003

29

Division for Electricity and Lightning Research

Uppsala University

Switch contacts can be directly protected by connecting protecting networks across them.
Some examples of protecting networks used across switch contacts are given below.

C
RL
Vdc

C
RL

Vdc

L
C
R
RL

Vdc

RT/2003

30

Division for Electricity and Lightning Research

Uppsala University

3.7 Transformers
A transformer is ideally a four-terminal device. However, if the influence of the surroundings
is included, it becomes a six-terminal device. Transformers are meant to transfer differentialmode (DM) signals from one side to other. In practice, common-mode (CM) signals are also
transferred due to parasitic capacitance. Transfer of CM signals are considered next.
1

6
Reference or Ground

Parasitic capacitance in a transformer

Cps

3
Cp

Lp

Csc

Cpc

Cs

Ls

2
Cpr

Ccr

Csr

6
Reference or Ground

Equivalent circuit diagram of a mains transformer.

p - primary
s - secondary
c - core
r reference

CM-CM Transfer
In an ideal CM-CM transfer situation, same current flows in both terminals of the winding
and in the same direction. There is no current in the windings, no voltage difference between
terminals, and no magnetic flux in the core. There is a voltage between the reference and the
windings.

Ip
RT/2003

Is
4

31

Division for Electricity and Lightning Research

Uppsala University

If the transformer core is not connected to ground, the equivalent circuit is as follows,
Cps

3, 4

1, 2
Csc

Cpc

Cpr

Csr

Ccr

5
Reference or Ground

If the core is connected to ground, C cr is shorted out. Then, the equivalent circuit is as
follows,
Cps

3, 4

1, 2

Csc+Csr

Cpc+Cpr

5
Reference or Ground

It is obvious that the most important capacitance in the CM-CM transfer is the coupling
capacitance between the primary and secondary windings. If the core is not connected to
ground, the coupling capacitance between the core and the windings also contribute to the
transfer ( C pc and C sc ). Therefore in most cases, it is admissible to connect the core to the
reference, via a low impedance.
DM-DM Transfer:

3
Ip

2
RT/2003

Is
4
32

Division for Electricity and Lightning Research

Uppsala University

Core directly connected to reference. Equivalent circuit is as shown.

Cps
1

Cp

Lp

Ls

Cs
4

Cpc+Cpr

Csc+Csr

Reduction of unwanted transfer

We have seen that the coupling capacitance ( C ps ) between the primary and secondary is
important in the transfer of unwanted high frequency signals between the primary and
secondary. Any measures taken to reduce C ps will reduce the unwanted transfer. If a metallic
screen (made of foil) is introduced between the primary and secondary windings, with the foil
connected to reference, the capacitance C ps is effectively reduced. Care should be taken to
keep the impedance (mainly inductance) of the connection between the foil and reference to a
minimum and also not to form conductively closed loops with the foil. A carefully installed
screen can reduce C ps to as low as 1.0 pF. Without the screen C ps is usually many tens of
pF.

Cpf

Csf
3, 4

1, 2

Foil (screen)
Csc+Csr

Cpc+Cpr

5
Reference or Ground

Reduction of CM-CM transfer with a CM screen

CM screen is formed by surrounding one of the windings with foil and connecting the foil to
reference. Screening can also reduce high frequency DM-DM transfer. In this case the foil
surrounding the winding is connected to either the primary side or secondary side (not to
reference) depending on from where the noise is originating. Transformers are commercially
available with 3 screens: one to control CM-CM transfer and two (one on each side) to
control DM-DM transfer at high frequencies.

RT/2003

33

Division for Electricity and Lightning Research

Uppsala University

The DM screen increases the capacitance between terminals 1 and 2 ( C p + C pf ). In the

absence of DM screen external capacitance can also be used to perform the same task.

Cpf

Csf

Cp
2

Lp

Ls

Screen

Cs
4

Reduction of DM-DM transfer with a DM screen for disturbance source on the

primary side.

RT/2003

34