EMC For High Voltage PDF

Electromagnetic compatibility in high-voltage engineering
Citation for published version (APA):

Houten, van, M. A. (1990). Electromagnetic compatibility in high-voltage engineering Eindhoven: Technische
Universiteit Eindhoven DOI: 10.6100/IR336248
DOI:
10.6100/IR336248
Document status and date:

Published: 01/01/1990
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the author's version of the article upon submission and before peer-review. There
can be important differences between the submitted version and the official published version of record. People
interested in the research are advised to contact the author for the final version of the publication, or visit the
DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page
numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
Take down policy
If you believe that this document breaches copyright please contact us:
[email protected]
providing details. We will immediately remove access to the work pending the investigation of your claim.
Download date: 04. Feb. 2019

ELECTROMAGNETIC COMPATIBILITY
.
lil
HIGH-VOLTAGE ENGINEERING
M.A. VAN HOUTEN

.
IIl
PROEFSCHRIFf
ter verkrijging van de graad van doctor

aan de Technische Universiteit Eindhoven,
op gezag van de Rector Magnificus,
prof. ir. M. Tels, voor een coi:nmissie aan-
gewezen door het College van Dekanen
in het openbaar te verdedigen op
dinsdag 23 oktober 1990 om 14.00 uur
door
Marinus Albertus van Houten
geboren te Breda
druk: wibro dissertatiedrukkerlJ. helmond.

IN
HIGH".' VOLTAGE ENGINEERING
CJP-GEGEVENS KONINKLUKE BIBLIOTHEEK, DEN HAAG
Houten, Marinus Albertus van
Electromagnetic compatibility in high-voltage engineering/

Marinus Albertus van Houten. -[S.1.:S.n.].- Fig., tab.
Proefschrift Eindhoven. Met lit. opg, reg.
ISBN 90-9002929-X
SISO 661.55 UDC 621.311.4.027.8.013.79.08(043.3)NUGI 832
Trefw.: hoogspanningsinstallaties; elektromagnetische
interferentie.
Dit proefschrift is goedgekeurd door de promotoren:
prof.dr.ir. P.C.T. van der Laan

en
prof.dr. M.P.H. Weenink
aan broer DIRK
CONTENTS
SUMM".RY 6
INTRODUCTION 9
- High-Voltage Engineering and Electromagnetic
Compatibili ty
- A basic problem
- Alm of the present work
- Outline of the thesis
ELECTROMAGNETIC COMPATIBILITY AND GROUNDING 13

2.1 Grounding 14
2.2 Criticlsm on standard deflnltions of "ground" 15
- Ground, a perfect sink or source ?
- Ground, a point of equal potentials ?
2.3 Misconceptions due to an incorrect philosophy 19
- Jumping potentials
- Single point grounding
- Multiple grounding systems
- The so-called "clean earth"
- Ground loops
2.4 Distributed inductance and grounding 24
2.5 Design rules for grounding 27
2.6 Protected regions formed by grounding structures 31
2.6.1 Protected region for leads 33
2.6.2 Protected region for instruments 36
2.7 Applications 41
- Lightning protection system
- Open substation
THE TRANSFER IMPEDANCE or GROUNDING STRUCTURES 44

3.1 The transfer impedance concept 44
3.2 The transfer lmpedance of GS's for leads 48
3.2.1 !deal GS, a tube 48
3.2.2 Practical GS's, plates or conduits 49
(model, the shape factor, Ztr-measurements,
setup, experimental results and discussion)
3.2.3 An almost ideal GS: covered conduits 59
3.3 The transfer impedance of grounding structures 65
for the protection of instruments
- An ideal GS for an instrument
- Real GS's for instruments
- Transfer lmpedance measurements
(setup, experimental results and discussion)
4
ELECTROMAGNETIC COMPATIBILIT.Y AND SIGNAL HANDLING ASPECTS 79
OF A DIFFERENTIATING/INTEGRATING MEASURING SYSTEM
4.1 High-Voltage dividers 79
4.2 D/I-system for fast rising voltage transients 82
- Principle
- High frequency problems
- Input circuit
- Sensor electrode
- Coaxial cable
- Fifty-ohm termination resistor
- Fast passive integrator
- Step r~sponse of the measuring system
4.3 EMC-aspects of a D/I-measuring system 94
ELECTROMAGNETIC COMPATIBILITY IN GIS-SUBST ATIONS 97

5.0 INTRODUCTION 97
5.1 GIS-INSTALLATION AS A HF-INTERFERENCE SOURCE 99
5.1.1 Unnecessary interruptions in a GIS-enclosure 100
- Insulated enclosures
- Continuous enclosures
- Comparison of dlssipation in enclosures
GIS-enclosures
HV-cables with a lead sheath
Insulated-phase bussystems
5.1.2 Necessary interruptions in a GIS-enclosure 108
- GIS/BV-line transition
- GIS/HV-cable transition; cathodic protection
- GIS/HV-cable transition; current-transformer
arrangement (setup, experimental results and
d1scuss1on)
CONCLUSIONS 126
REFERENCES 128
SAMENVATTING 133
DANKWOORD 135
CURRICULUM VITAE 136
5
SUMMARY
Electromagnettc --compatibility, within the field of

Electrical Engineering, is the ability of electroniè and
electrical apparatus to function correctly in each ethers
vicinity. This thesis describes new EMC-concepts for an
efficient and consistent approach to practical interference
problems. Analysis should lead to methods to eliminate and
preferably prevent interference early in the design phase.
In this thesis the concepts developed are mainly used in
high-voltage installations. The integration of modern elec-
tronics within this type of installations poses high EMC-
demands. Electromagnetic Compàtibility is obtained here by
the use of correct layout, and by the installation of leads
and additional metal at strategie places. The concepts
developed can however be used -in appropriate form- within
the whole field of Electrical Engineering.
The thesis starts with a critica! analysis of "Grounding",

a subject of fundamental importance within EMC. Due to an
incorrect understanding about what a grounding system is
supposed to do, the noun "ground" causes much confusion in
practice. This incorrect understanding is related to two
basic elements of the generally accepted definition of
"ground":
- a ground can absorb or supply current without any change
in voltage;
- a ground is an equipotential point or plane which serves
as a reference for the circuit considered.
By ignoring the concept of potential and by concentrating
ourselves on the physically meaningful currents, on the
circuits which carry these currents, and on the magnetic
fluxes, we developed a genera! protection philosophy. The
central concept within this philosophy is the "transfer
impedance" of "Grounding Structures (GS's)". By using GS's
we create protected regions in which the sensitive
electronic equipment can operate without any problems. We
distinguish between GS's for leads and GS's for apparatus.
6
In this sense, the outer conductor of a coaxial cable is
already a simple GS, which creates a protected region. An
"EMC-cabinet" is an essential part of the GS for the
protection of sensitive ~quipment.
The signif icance of the transfer impedance concept is that
it gives a meaningful criterion for the quality of GS' s
(including the "layout" of the complete "network"). The
transfer impedance of complex GS's, which can generally not
be calculated, can .always be measured with so called
"current injection test methods". Models as well as
measurements of the transfer impedance of GS's for leads as
well as GS's for equipment are discussed.
An EMC attractive solution for signal transport in high-
voltage measuring techniques is the use of differentiated
signals. The differentiated signal should be integrated
again at the input of the electronic equipment. The
components of the first part of the integrator can be of
normal size as used in electronics and passive. With a
correct design of the integrator, interference is suppresed
before it reaches the vulnerable active electronics. The
EMC-cabinet plays an important role here. It has to protect
the measuring equipment as well as the integrated signal.
The design of a "Differentiated/Integrated (D/I)"-system to
measure fast voltage transients is described. The D/I-
system has among others been used for transient voltage
measurements in the 150/10 kV "Gas Insulated Switchgear
(GIS)"-substation Eindhoven-West.
A GIS-installation is an intense source of high-frequency
interference. Measurements of steep transient voltages
across interruptions in a GIS-installation, due to
switching actions, are presented. Means to reduce the
influence of this interference source on the measuring
equipment are available, and are discussed.
7
8
CHAPTER 1
INTRODUCTION
High-voltage engineering and EHC

In high-voltage installations of electrical utilities, or
in HV-experiments or in testing, much capacitive energy is
present. During switching actions in HV-substations, or
during experiments, with intentional or inadvertent
breakdowns, this energy is partially converted into high
power electromagnetic waves. Oue to the resulting intense
EM-interference, modern electronics -increasingly used for
measurement and control- may not function properly in
HV-installations, or may even be destroyed. Therefore
"Electromagnetic Compatibility (EMC) ", the ability of
electronics or electrical apparatus to function correctly
in each others vicinity, is of crucial importance to high-
voltage engineering. We on purpose give a definition
different from the IEC definition: "The ability of a
device, equipment or system to function satisfactorily in
its electromagnetic environment without introducing
intolerable electromagnetic disturbances to anything in
that environment". It will become clear later that the term
EM-environment, although fashionable, can better be
avoided.
A basic problem
Electromagnetic compatibility between highly sensitive
electronic systems and HV-installations is in addition
endangered by the large geometrie dimensions of the HV-
installation. A basic EMC-problem is then that the complex
three-dimensional circuitry cannot be correctly represented
by the famil iar network theory models ( see Sec. 2 . 4) • The
discrepancy between network model and reality shows up
first of all, at fairly low frequencies already, when
significant time-varying magnetic fluxes are distributed in
space. Kirchhoff's voltage law (KVL) is then not obeyed,
which causes a considerable amount of confusion.
If electrical engineering is schematically represented by
9
the large rectangle in Fig.1 we may distinguish three
regions, from left to right, that is with increasing
frequency. At low frequency, in reqion :r, network theory
can be used, whereas at high frequency where wave length
effects are important, the full Maxwell description is
required, to describe for instance antennas and wave guides
region III). Distributed fluxes are aften already
. --------------.
I II III
!'
wavelength
~~~~~~~;~.
Network theory effects
KCL & KVL Maxwell
correct ·,_i.,,· 1
transmission linesi
eddy currents i
'
'-----------·-·--·--
grounding """." : __________ __,
~
Fig. 1: Regions within electrical engineering where
different descriptions are required; network theory at the
left, the full Maxwell equations at the right. Grounding
often falls in the difficult middle region. KVL and KCL
stands for Kirchhoff's voltage law and Kirchhoff's current
law, respectively.
important at intermediate frequencies (region II), for
instance in transmission lines, in eddy currents and
grounding. In transmission lines the problem is usually
avoided by the introduction of an equivalent network in
which only the voltage in the perpendicular cross-section
is considered, after which a 11 1umping 11 of the the
inductance is permissible. Eddy currents are clearly the
result of non-conservative E-fields, and are important
enough to make thin laminations necessary in 50-Hz
transformers. In the equivalent circuit of the transformer
the resulting losses are simulated by a resistor. In
grounding, a central problem for EMC, the leads are aften
long and have an irregular structure. Since also large
currents may flow we have to deal with appreciable
distributed fluxes. Evidently we are then not any more in
region I and cannot use the standard network theory. As a
10
consequence of the failure of the KVL extended leads no
longer "transport" potentials as they do in network
diagrams. The description should then concentrate on the
flow of currents and on the fluxes associated with the
currents for a given layout.
Aim of the present vork
The purpose of the present work is to develop concepts for
the protection of electronics or electrical apparatus which
have to function correctly in each others vicinity. To
achieve correct operation of the equipment, it is necessary
both to understand and to resolve practical electromagnetic
interference problems. A good understanding hopefully leads
to modifications early in the design phase of a setup, an
apparatus or a large installation. This is cheaper and more
efficient than modifications introduced in a later stage.
The concepts described here are mainly applied to sensitive
digital registration equipment which is used in
high-voltage research. Because of the severe interference
-often coinciding with the fast phenomena that are to be
observed- we have used much metal to obtain an adequate
EMC-protection. The concepts developed are however basic
and general; they can be easily adapted to other fields of
electrical engineering.
Outline of the thesis
The present work discusses first of all EMC-problems and
concepts in gener al and then describes practical
EMC-problems in HV-engineering in particular.
Chapter z gives an EMC-analysis for high-frequency
grounding, such as required for the protection of
electronic systems and large scale electrical
installations. It starts with a critical analysis of
"grounding", and discusses misconceptions around the noun
"ground" and related to that the validity of Kirchhoff's
laws. Design rules for the activity 11 grounding 11 are given
and explained. Grounding Structures (GS' s) for the
protection of leads and instruments are introduced. Special
attention is given to conduits made out of iron and to
11
"EMC-cabinets" for instruments.
In Chapter ~the transfer impedance concept for GS's is
introduced. This concept is used for a quantitative
comparison of GS' s, both by calculations as well a:s by
measurements. By means of current injection tests we have
measured the transfer impedance of GS's for leads as well
as the transfer impedance of simple GS's for instruments.
Chapter ,i deals with a D/I measuring system for very
fast high-voltage transients. Due attention is paid to the
design'of this measuring system and to its EMC-aspects.
Chapter .2 -EMC in a GIS substation- considers a Gas
Insulated switchgear (GIS) installation as an intense and
concentrated source of HF-interference. Proposals to
improve the EMC behavior of the GIS-installation, which
still satisfy the ether design criteria, are discussed.
Measurements on very fast high-voltage pulses are
presented.
Chapter ~ gives the main conclusions.
12
CHAPTER 2
AND
GROUNDING
For the design and production of electrical equipment much

fundamental and technica! expertise exists within the
various fields of electrical engineering. To ensure correct
operation of pieces of equipment originating from these
various fields when they are combined, requires serieus
ElectroMagnetic Compatibility design (EMC-design) efforts.
If no systematic EMC-approach is available -preferably to
be used already in the design phase- we may expect
EMC-probems to become more and more serieus with the
growing use of complex electronic systems and the
corresponding increase of different power levels.
Unfortunately no clear theory for these situations seems to
exist and EMC-problems have in fact been solved by
experiments or sometimes simply by trial and error. The
search for a clear theory which can provide practical guide
lines for EMC-correct design remains therefore very
important.
Of all the resulting EMC-problems it is recognized (Mer 76]
that the design of "correct grounding" is one of the
most important aspects of EMC-design of electrical systems.
Therefore we start in this thesis with an analysis of
"Grounding".
Positive results of this critical analysis will turn out
to be:
-we can specify more clearly what a grounding system is
supposed to do. The question whether ground wires
faithfully transport zero "potentials" becomes meaningless.
-The activity "grounding" is seen as the design and
construction of the low voltage sides of all current loops.
-Very often we have to deal with complicated existing
electrical systems. In these cases it is difficult or
13
impossible to control the entire grounding system. However,
by using metal "Grounding Structures" (GS), suitably sized
for "our" special purposes, we can locally solve all our
grounding problems.
-This analysis leads to the development of general, linear
and basic methods for protection of electronics and to a
design of grounding systems which protect (large)
interconnected electrical systems against interference: the
main aim of EMC-design.
Note that this approach does not rely on the improvement of
the overall "EM-environment" but instead, provides an
excellent local 11 climate 11 , where it is needed.
Elements of this grounding analysis have appeared in
earlier publications [Laa 78], [Ott 79], [Jon 83], [Laa 86)
[Laa 87) and [Hou 89].
2.1 GROUNDING
Grounding is interpreted in this chapter as all design and
actual construction work on the low voltage side of elec-
trical circuits. This makes grounding a very broad subject
essential for widely different fields such as lightning
protection, power engineering and micro-electronics. We may
nevertheless formulate a simple and genera! objective of
grounding: "Grounding should reduce dangerous voltage
differences between critica! points to safe values".
By grounding correctly we want to achieve the following:

a) Interference voltage differences across sensitive inputs
or across ether critica! terminals of our circuits
should r_emain low, so that correct operation of the
circuit is not affected.
b) The safety of people must be guaranteed.
c) In the case of high currents (e.g. lightning) we avoid
dissipation in a "poor conductor" by installing a metal
grounding conductor parallel to it.
The voltage differences which play a role in a) and b) are
2 . .
equal to 1
J E.dl, usually taken along the shortest
14
connection line. Note that it is this voltage difference
that is responsible for the risk to electronics, to people
or for breakdown.
Historically the objectives b) and c) were recognized first
in grounding practica. As systems became bigger and more
complex and with the growing use of electronics the typical
EMC objective a) received more and more attention [Pea 62],
[Den 73). Since often only very low interference voltages
can be tolerated in micro-electronics a) poses difficult
engineering challenges.
The available technical expertise on grounding inside an
apparatus (the internal grounding) may seem impressive, but
is more a product of art than of science, whereas the
expertise on grounding seldom includes the protection
against external interference, the EMC grounding. To
develop a more scientif ic description of grounding we have
to point out that the generally accepted definition of
"ground" is incorrect.
2. 2 CRITICISM ON STANDARD DEFINITIONS OF 11

GROUNO".
Most standard definitions of "ground" contain two
elements:
1. A ground can absorb or supply current without any
change in voltage ; in other words the ground should be
a perfect sink or source for currents.
2. A ground is an equipotential point or plane which
serves as a reference for the circuit considered~
GROUND, A PERFECT 5INK OR SOURCE ?

A ground can only act as a sink or source for current when
charges can accumulate, in other words when a capacitor
with sufficient capacity is present. This also follows from
the continuity equation for charges
div j + ap/at = o (2 .1)
When the current density j is absorbed or supplied div j
differs from zero and consequently the charge density, p,
must change.
15
In the search for the capacitor which collects this charge
we have two candidates (See Fig.2.l):
1. The Earth 1 -eomri.dered as an isolated sphere with an
average radius of 6367 km, has a capacitance of 708 µF
(Fig.2.la).
2. The capacitance between the Earth and the lower layers
of the ionosphere at for instance 50 km height, turns
out to be 91 mF (Fig. 2. lb). This large capacitor is
present, and on a world . wide scale the thunderstorms
charge -this capacitor [Vol 82] to several hundreds of
kiloVolts. This charge causes the so-called
fair-weather-field.
Bath capacitors cannot play a role in our grounding because
our local engineering activities do not influence the total
E-field around our Earth. Between our charged objects and
the Earth only a small capacitor is present (Fig.2.lc).
I
,-"o . . - . . ,\
,--x---,
1 , ,...,,..., 1 l
'T",,,
1' 1 ' J ' 1
LARGE IONOSPHERE OUR

SCALE FAIR-WEATHER-FIELD ELECTRO
LIGHTNING ENGINEERING
1a1 lbl (Cl
Fig.2.1: Electric fields on different scales. For our

electrical engineering we use only a minute part of the
Earth.
To arrive at a correct picture we rewrite Eq. 2 .1 with

Gauss' law
div (j + 8D/8t ) = 0 (2.2)
The combined quantity, j + 8D/8t is divergence free and has
therefore no sink or source. This leads to the correct
statement that any current -where we· have to include the
capacitive current- must flow in a closed loop.
16
As a consequence of this statement we can specify more
clearly what a grounding system is supposed to do.
A grounding system never resembles a sewer system where
more and more sewage pipes converge into one main pipe with
11 unknown 11 destination (Fig. 2.2a). Instead a grounding
system is a group of interlinked current loops (Fig. 2.2b).
(8) (b)
Fig.2.2: Incorrect (a) and correct (b) picture of what a

grounding system is supposed to do. The thickness of the
lines corresponds to the magnitu~e of the current.
We make three observations. First of all, Fig. 2.2b shows

only the low-voltage part of all the circuits and is in
that sense incomplete. Secondly the connection to the Earth
in Fig. 2. 2b is not unique anymore, but is only another
part of a current loop. If a current flows into the Earth,
this current must leave the Earth somewhere else. Therefore
the connection to Earth is not essential as is demonstrated
by digi tal watches, portable radios, airplanes and
satellites.
GRDUND> A POINT OF EQUAL POTENTIAL ?

This second element of standard definitions of ground
implies that a connection to ground fixes the potential of
the connected point of the circuit, where the potential of
the ground is aften taken to be zero. A first, relatively
simple complication is caused by the resistivity of the
conductors, in particular of the soil; we may correct for
that by calculating the correct grounding resistance, which
depends on shape and size of the buried grounding rod.
17
A more basic question is whether a highly conducting, say a
metal "Earth" would form an equipotential surface. Since
the size of the sphere is not important (see Section 2.2.1)
we may -consider any metal object, suchas a ship, a screen
room or an airplane (Fig. 2.3). In electrostatics such an
object forms an equipotential surface. This seems also to
be true according to network theory, where wires are
assumed to "transport" potentials.
However, when in an airplane, struck by lightning (Fig.
2.3), we connect three voltmeters between the points A and
B, we obtain different re~dings as a result of the
distributed time-varying magnetic flux.
Fig. 2. 3: The voltage between the points A and B of an

air-plane, caused by lightning current, cannot be
described by a potential difference; each of the three
voltmeters gives a different reading, depending on the
loop enclosed by the leads.
Voltmeter 2, close to the outer surface reads plj (r2 ) ,
where p is the specific resistivity, 1 the length between
the contacts and j (r2 ) the current density at the outer
surface where r is the outer radius of the airplane.
. 2
Voltmeter 1 sees in addition to that, the voltage induced
in the outside loop. When the lightning currents are evenly
distributed around the tubular body the magnetic field
inside the airplane is zero. Voltmeter 3 then reads only
plj (r1 ) , where r 1 is the inner radius. In Fig. 2.4 the
genera! behavior of these three voltmeter readings is
shown. This picture shows first of all the streng relation
18
with the transfer impedance of coaxial structures and is
secondly reassuring for airplane electronics since the
V reading drops quickly to zero at higher frequencies.
3
For our argument here, it is important to realize that in

the loops formed by the volmeter leads, Kirchhoff's voltage
law (KVL} is not obeyed. This is also not true for the
small loops in the skin (shown enlarged in Fig. 2.3) where
j and E vary wi th depth. In these examples the simple
11 zero-dimensional" network theory cannot adequately
describe the three-dimensional reality~
V -V "-'w
lvl î 1 2
~
de ·
a
d=ó 6
().)
0
Fig. 2.4: Voltmeter readings as in Fig. 2.3 as a function

of frequency. At low frequencies the dc-resistance
determines the voltage: at frequencies where the skindepth
is smaller than the wall thickness d the readings V2 and VJ
are different. For a steel hull this takes place at quite
low frequencies because of the smaller skindepth.
2.3 MISCONCEPTIONS DUE TO AN INCORRECT PHILOSOPHY
Since the analysis of grounding is commonly based on

network theory and on potentials such as used in the
Kirchhoff Voltage Law (KVL}, many misconceptions arise, of
which we mention a few.
JUMPING POTENTIALS
A lightning discharge as in Fig. 2.5a, injects a grounding
current. The current flows to the Earth and returns to the
cloud as a displacemeht current. Of this "complete current
loop" we only control a small part: by a choice of diameter
19
and geometry and the number of lightning rods we minimize
detrimental effects. This lightning current flowing through
a lightning rod does_ not cause an unambiguous potantial
difference across a given length of the rod. The· two
voltmeters shown in Fig.2.5a illustrate that a potential in
Kirchhoff sense does not exist: they enclose with their
r--
'
1
1
3'
~1
1
1
1
1
1
1
1
1
1
(a) (b) (c) (d)
Fig. 2.5: The two voltmeters 1 and 2 (a) connected to the

same points on the lightning rod show different readings.
Therefore i t is not correct to represent a voltage tower
(b) or a communication tower (c) by a lumped network
element. In the case of an metal column tower (d) voltmeter
3 shows almost zero.
leads · different magnetic fluxes and show therefore

different voltages. Nevertheless a high-voltage tower or a
communication tower (Fig.2.5b,c) is often represented by a
lumped network element, through which the lightning current
flows [Van 80]. This is incorrect, because the voltage over
these towers is not uniquely defined. In case the rod is a
tube a voltmeter with its leads inside the tube should not
show much interference voltage because no magnetic field is
present in the tube (Fig.2.5d), just as voltmeter 3 in
Fig.2.3.
Similarly, the diverging lightning current in the soil does
not produce "step potentials", only related to the soil
resistance. The "step voltage" depends also on the flux
enclosed by the leads.
20
The same mistake is made in the modeling of lands on
printed circuit boards for the purpose of predicting
"ground shifts" between two points on a ground land
[Pau 86]. Again this is fundamentally wrong because one
cannot unambiguously localize the "lumped" inductance, or
the voltage source which represents the induced loop
voltage (See Sec 2.4.,Eq.2.3 and remark d).
SINGLE POINT GROUNDING

In the frequently recommended single point grounding one
attempts to minimize the coupling between circuits by
avoiding groundloops and by having separate returns f or the
ground currents, at least up to the "single point"
(Fig. 2.6). If all currents did flow in one direction, as
in a sewer (Fig. 2.6a) this would indeed reduce coupling,
although a possible· large common part of the circuit beyond
the single point is not considered. If the currents flow in
opposite directions, as they may do when connecting leads
are present (Fig. 2.6b), large loops are formed
unnecessarily. Actually the defendants of single point
grounding in their search for the magie zero potential,
forget that current always flows in closed loops (see
Sec.2.2.1).
(a) (b) (c)
Fig. 2. 6: Single-point grounding is based on the sewer

idea; all currents flow to the sewer (a). In fact currents
can flow in opposite directions for instance as a result of
induction, when connecting cables are present (b). Amore
compact grounding is much better; the lower horizontal line
can be a conduit (c).
21
A much better solution is to design these loops in a
compact way · \fi'±q. z ~l>c) • ~en the grouna connection is a
conduit or tube, large grounding currents can flow without
coupling in voltage over critical inputs. In such case the
connection to Mother Earth is only necessary as a safety
ground; a long thin wire is then acceptable.
MULTIPLE GROUNDING SYSTEHS

Followers of the single point grounding concept may be
concerned about the amplitude of current their grounding
system can carry, and about the resulting "potential
difference" over the grounding wire. This thought can lead
to the installation of a number of grounding systems
(Fig. 2. 7) • Of ten these systems have a long connection to
Mother Earth, to "transport" the desired potential zero.
Unfortunately, due to the large loops formed, already a
weak external magnetic field induces considerable voltages
between nearby branches of two trees. In such a case,
BONDING
INSULATION STRIPS
Fig. 2.7: Multiple grounding systems based on single point

grounding concept.
insulation is used or surge arresters or bonding strips.
Interesting is that bonding strips are supposed to
"equalize ill-behaved ground potentials". In fact these
bonding strips are shorting out induction voltages and
provide a local closing of grounding circuits. A breakdown
of insulation or a flashover in a surge arrester introduce
locally new high frequency interference voltages.
22
THE SO CALLED CLEAN EARTH
As a special case of multiple grounding systems we mention

a "clean earth" grounding system, sometimes requested for
expensive equipment.
In this concept the already available ground systems in
existing electrical installations are suspected. Therefore
a "special" ground electrode is installed to preserve a
"clean zero potential" which is kept far away from the
"dirty grounds". The long separate new connection to this
electrode, which is supposed to transport this "zero
potential" faithfully, introduces a big loop and
consequently interference voltages with respect to both the
existing "dirty" Earth and metal in the building. To
control these interference voltages one employs extra
insulation, or (rather inconsistently) surge arresters and
potential equalization strips. Clearly again nearby
conducting material as in Fig. 2. 6c would provide a much
better controlled path for the currents, and would give
lower interference voltages at a lower cost.
GROUND LOOPS
Ground loops have a bad reputation because:

- They often show up in unexpected places
- They are often large in size so that the current in the
loop can produce undesirable coupling effects over large
distances
- A part of the ground loop is often close to a signa! lead
which of course causes interference.
However a current circulating in a ground loop can be less
troublesome than a voltage of unknown value across an
interruption introduced to break a ground loop. Capacitive
currents across the interruption can give some coupling,
but much worse is a breakdown at high interference
voltages. The local breakdown generates steep transients
which may generate serious extra interference.
Ground loops are in fact rather common. Large scale
23
conductors (eg. a ground plane or a screen room) could be
considered as a collection of ground loops, which at high
frequency:-J.e.ad .:t.o-.a di-fferent internal ourrent dietribution
than at low frequency ( see Fig. 2. 8) • This redistribution
can be quite useful for EMC-purposes, since it provides a
lower inductance, a more localized flow of current, or a
lower transfer impedance of grounding structures (see
Sec.2.6).
SOURCE
Fig. 2.8: A circuit above a plane, with two connections to

the groundplane. For de and hf the current patterns in the
groundplane are different.
2.4 DISTRIBUTED INDUCTANCE AND GROUNDING

Both in electrostatics and in the network theory grounding
is simple: the potential of Mother Earth is zero and a
connection with the Ea.rth fixes the potentials everywhere
in our circuit. Because the network theory, with its simple
Kirchhoff 1 s laws is being used successfully everywhere in
electrical engineering, we are very much accustomed to this
model. In network theory all problems with E- and H-fields
are hidden in the symbols for the impedances whereas around
the terminals and near the connecting wires no fields are
anymore present. These connecting wires shown in the
diagrams are therefore zero-dimensional abstractions of the
real wires in the circuit. In the diagram each wire is
everywhere at the same potential and all induction effects
and transit times and radiation are being neglected. Within
network theory current and voltage play a dual role; we may
work with either one of the two.
24
The simple description of grounding given by the network
theory is certainly valid for electrostatics and other
dc-current circuits. For the mains frequency (50 or 60 Hz)
the network theory is in general correct; however, in the
case of long lines, large currents and grounding,
distributed fluxes cause problems. At high frequencies this
is even more the case and the simple network model fails
completely.
Although these problems arise because network theory is no
langer adequate, we hesitate to return to the full Maxwell
laws and to the resulting complicated boundary value
problems. Fortunately this is seldom necessary, we already
gain much insight by a study of the parasitic impedances of
the relatively long ground, supply and signa! wires. The
parasitic resistances and capacitances can, if necessary,
be represented by network elements which can be added to
our circuit, without essential problems. As will be
explained and was already illustrated in the previous
sections, the parasitic inductance is basically much more
complicated, because its effect cannot always be
represented by a lumped circuit element.
Whereas in network theory the magnetic flux of an
inductance is always supposed to be confined within the
impedance symbol, the fluxes associated wi th the ground
leads, the supply leads or the signa! leads are essentially
distributed in space. We then have to think in terms of
current loops and we apply Maxwell's laws for instance to
these closed contours. The induction law in its integral
form is useful; we employ the customary notation.
! E.dl = -;J Bt
BB
.dS =- d</I
~ (2.3)
Consequences of this equation are:
a. Kirchhoff's voltage law (KVL) is based on this equation
without enclosed flux. With enclosed time dependent flux
the contour integral of E is no langer zero and the KVL
fails: the sum of the voltages across circuit elements
in a circuit mesh is not zero.
25
b. Potentials as used in the I<VL, do not exist when
chan<Ii°-g magneti~fluxes are present (see also [Laa 78]).
c. In an actual circuit there are still voltage differences,
which may lead to breakdown, or to interference, or to a
voltmeter reading. The voltage between points 2 and 1
(2.4)
depends on the actual integration path (compare Eq.2.3).

The lay-out of the wiring and the lay-out of the
voltmeter leads are now important; remember that this is
not the case in network theory.
d. Equation 2.3 also shows that an inductance can only be
defined for an (almost) closed current loop. Since this
inductance is a property of the loop as a. whole, one
cannot unambiguously localize the "lumped 11 impedance, or
the voltage source which represents the induced loop
voltage. A lot of cónfusion is caused by authors who
(incorrectly) define inductances of single straight
wires [Van 80, Pau 87, Ott 88]
e. The duality of current and voltage is lost, because a
potential cannot be defined (b) and voltages depend on
the integration path (c). The current remains a much
clearer physical quantity; its value is found from
JJ(j+:~).dS (2.5)
5
where we have to integrate over the cross section of the
wire or in case 8D/8t is important also over the area
where the displacement current flows. We may describe
this by saying that Kirchhoff's current law (KCL)
remains valid.
f. As a result of e) and in agreement with ott's statement
(Ott 79) we should concentrate on the currents in our
analysis of grounding problems. The grounding currents
circulate in a loop. Our task is to design, or to
identify and if necessary modify these loops in such a
way that interfering voltage differences, calculated
with Eq.2.4 remain low enough at critical terminals.
26
2.5 DESIGN RULES FOR GROUNDING
The wish to keep interfering voltage differences across
sensitive inputs or across ether critica! terminals low is
best served when we (re)arrange the current loops and close
them as compactly and locally as the circumstances allow.
To design a grounding system according to the correct
picture of what a grounding system is supposed to do (see
Fig.2.2b), we follow a number of steps.
- We must ignore potentials, particularly when they seem to
behave wildly, according to the naive picture of network
theory.
- We concentrate on the currents in our various circuits.
- we design new, or modify existing current loops such that
impedances and coupling to neighboring circuits are
minimized. We do this by closing the circuits as
compactly and locally as the circumstances al low; this
also results in a clearer design.
- We start closing the circuits for the grounding currents
in the smallest subsystem. Only after we have solved the
local grounding problems we move outward to the next
larger system.
- The largest and final ground system ( see Fig. 2. 2b) is
aften partly formed by Mother Earth. We limit the
currents to and from Mother Earth as much as possible and
let her only play a rele when it is absolutely necessary.
In fact the connection with Mother Earth only carries
current in case of lightning or -depending on the
regulations- for safety grounding.
- Finally we check by means of Eq. 2. 4 and Eq. 2. 3 whether
the voltages at critica! inputs are indeed low enough.
This design method keeps magnetic fluxes (self and mutual)

small so that we will have f ewer deviations from the KVL,
than a less compact design would give. Moreover the compact
and local approach reduces capacitive and resistive
coupling. Generally speaking we may expect the interfering
voltages to be small; if not we retrace the steps outlined
above.
27
Making grounding circuits "as compact as possible" is
rather straightforward in an Integrated Circuit (IC), on a
Printed Circuit Board (PCB), or inside an apparatus
[Laa 86}.
Goed grounding connections are formed by copper ground
planes, or "gridded grounds" as being used on PCB's [Ott
81, Ger 85], to offer ground currents a path with a low
impedance [Ott 79]. This is especially important at the
high frequencies occurring in digital circuits. Also
decoupling capacitors, e.g. for digital circuits, must be
mounted in close proximity to the IC in order to provide a
compact path for switching currents [Dan 87]. Inside an
apparatus, wherever possible, we close the current loops
already within the apparatus itself, but where this is not
possible (as for input, output or supply lines) short
grounding connections formed by a wide metal strip (chassis
or a front panel) are often excellent. Such a metal strip
introduces very little extra resistance or extra flux in
the current loop.
In these cases the completion of the low voltage sides can
be solved locally. This is one of the reasons that so many
correctly operating apparatus have been built.
More grounding problems show up in the following cases;

- large distances in a network between sub-systems,
- when a new installation, e.g. a computer network, has to
be installed in an existing building with an older
electrical installation,
- electrical installations, with sensitive electronics for
measurements or control, which have to operate properly
in the presence of intense interference, e.g. lightning
events or interference due to switching events in
HV-installations.
These serieus problems can be sol ved by creating
"protected regions" formed by "grounding structures" with a
low "transfer impedance". The protection philosophy bebind
this statement,- still based on one point of view; the
control of current flow to avoid interference voltages
28
across critica! terminals-, is discussed in the next
sections.
Before we start with this discussion we analyze first the
problem of large distànces in a network between sub-
systems. In this analysis we introduce at the same time
terms which will be important in our protection philosophy,
namely "Differential-Mode (DM)" circuit, internal and
external "Common-Mode (CM)" circuit, protected region,
grounding structure, and the transfer impedance.
For simplicity we consider two subsystems which have metal
housings and which are connected to each ether by only one
cable (See Fig.2.9a).
The differential-mode circuits for DM-signal transport are
usually coaxial cables or several wires in a common braid.
These cables provide a correct DM-signal transport over a
wide frequency band. The regular geometry and the
compactness of such DM-circuits reduce not only the
1 DM-CIRCUIT Il
\
\J
cD 11
(a)
GROUNDING
lli=;a:zz=:z~ STRUCTURE·~:x=z=:z::czdl
?
(b) (c)
Fig. 2.9: a) A network consisting of two subsystems

interlinked by a coaxial cable for DM-signal transport.
b) Ground loops wil be formed by external ground
connections or by supply lines to the mains and c) an extra
nearby ground connection creates an additional but useful
ground loop.
29
impedance of these circuits but also the coupling to
neighboring circuits.
However, Fig.2~9a-·is incomplete. Oue to external ground
connections or by supply lines, we create at the same time
an irregular common-mode circuit with often unknown but
usually a large loop area (see Fig.2.9b). Magnetic fluxes
enclosed by this loop induce large interference currents,
which because of the transfer impedance of the signal cable
may introduce undesired DM-voltages.
A qrounding structure connected between the subsystems,
close to the cable creates a new well defined, regular and
compact internal CM-circuit (see Fig.2.9c). When we give
this grounding structure the form of a metal strip, a metal
conduit or tube, we create thereby a protected region which
can hold all leads and the fields of internal DM- and
internal CM-currents (see Fig.2.10).
sub
systeml
Fig. 2.10: A conduit-shaped grounding structure which

connects two subsystems
On the outside is formed a new external CM-circuit, now

even better closed by the grounding connection. Depending
on the transfer impedance of the grounding structure the
interference voltages due to the external CM-current can be
calculated. With a proper designed grounding structure
arbitrarily low values can be reached. In the case of a
tube, especially at high frequencies, these voltages remain
very small (see Fig. 2.11).
30
/
BRAIDED COAXIAL CABLE
Fig.2.11:
Typical behavior of the
transfer impedance for a
good coaxial cable and
for a tube.
logw--
2.6 PROTECTED REGIONS FORMED BY GROUNDING STRUCTURES

Grounding structures (GS) can be used, -and for many years
have been used-, in electrical engineering at several
locations and in many farms. Depending on our position in
electrical engineering we apply these grounding structures
in complicated networks to create a local protected region,
suitably sized for our "special" purpose (Fig.2.12).
For example, an experimentalist who has to use sensitive
measuring equipment near intensive interference sources
should use an extra suitable grounding stucture (e.g. an
(b)
(c)
Fig 2.12: With grounding structures we decouple regions in

the "grounding tree". As in Fig. 2. 2, this figure shows
only the low voltage part of all the circuits.
31
EMC-cabinet as shown in Fig.2.19). Attempts to improve the
EMC-properties of the measuring instrument or of the
elaçt.riGal . instal±atie!T iR t:he buildin<J are usually more
difficult and also less certain to lead to
EM-compatibility.
Examples of grounding structures at different locations in
a complicated electrical system are (Fig.2.12):
- ground planes in an integrated circuit (IC) or on a
printed circuit board (PCB) (a)
- a metal plate or chassis for the assembly of components,
plugs and ether connections (b)
- an EMC-cabinet for instruments (c)
- a conduit, for instance between EMC-cabinets (d)
- a metal floor or a completely shielded room (e).
We can also combine the protected regions, for instance
those created by cabinets and a conduit (see also
Fig. 2.10).
Metal, originally used for mechanica! rigidity of an
apparatus can often also be utilized for EMC-purposes.
Already existing metal constructions, e.g. of a
telecommunication tower or of a transformer housing (see
Fig. 2.22b and Fig. 2.22c), can play this double role.
In this section we discuss how we, by means of GS's, create
protected regions in which both sensitive signal leads and
electronic equipment can operate without interference (see
Fig.2.13). To utilize all benefits of these protected
SIGNAL CABLE
--<re:========:::=>-- PROTECTED
REGION
MAINS
EXTERNAL INTERNAL
GROUNDING GROUNDING
(b)
Fig. 2 .13: Protected regions formed by conducting metal
provide a local 11 climate" both for leads (a) and electronic
equipment (b).
32
reqions they should be given a correct shape and are to be
properly interconnected. Also all points where leads enter
the protected region should be correctly treated (see
Sec. 2.6.2).
The leads that reach the protected region, such as signal
cables may already carry differential mode interference
superimposed on the legitimate signals. Several options are
open to improve the signal to interference ratio: cables
with a lower transfer impedance (see Ch.3), larger
amplitude sigals or signals transported with H.F.-emphasis
(see Sec. 2.6.2).
Note:
-that the creation of protected regions does not rely on
the improvement of the overall "EM-environment" but
instead, only provides an excellent local "climate" where
it is needed.
-that the GS's not only provide protection but also give an
opportunity for internal and external grounding!
-that there do not have to be conflicts between connections
for safety reasons and the internal grounding facilities.
2.6.1 PROTECTED REGION FOR LEADS

Long GS's between electronic systems, in the form of metal
strips, metal conduits or metal tubes, create protected
regions which can hold all the leads and the fields of
DM- and internal CM- currents (See Fig. 2.14). In addition
the GS provides a preferred current path for external
(a) (b) (c) (d)
Fig. 2.14: Grounding structures (GS) for leads with their

current density distribution and corresponding H-field for
a CM-current carried by the GS.
33
CM-currents and acts as a shield against external E-fields
and B-fields.
The üsê öf metal f or a GS guarantees a conducting path f or
external CM-currents. The E-field in the metal is very low.
For de- and low frequency currents the current density is
homogeneously distributed over the cross section. At
suff iciently high frequencies the CM-current is confined to
the outer skin of the GS. The magnetic field lines due to
this current have to go around the structure. The GS in the
shape of a wide strip reduces the current density and the
flux between leads and strip in the middle of the strip.
Moreover the shape of a conduit, or better still that of a
tube, forces the field lines away from the leads. The
weakened magnetic fields then produces not much flux and
reduce the induced interference voltage between DM-leads
and GS.
A different way to describe this is to say that the
external CM-currents in the GS couple in interference
voltages via the transfer impedance (see Fig. 2.15). At
high frequencies the transfer impedance of an open GS
increases linearly with frequency because the magnetic
field lines can penetrate between the leads and the GS. For
an u-shaped GS, a larger height to width ratio helps; for a
tube the transfer impedance becomes negigibly small at high
frequencies. Note that even a small transfer impedance can
be important because the coupling takes place over the
entire length of the GS.
·l
->;------
[!l/m J
_, Fig.2.15:
The transfer impedance of a
-· few grounding structures. The
curves for the conduits (for
-ll
~<<d) are calculated with the
• model given in Chapter 3.
lO'J <.> ~
34
In most cases the transfer impedance is the most important
parameter of a GS; here we also discuss the role of a GS as
a shield against external E- and B-fields.
The role of the GS as a shield
a. The GS intercepts external E-lines very effectively
(Fig. 2.16a,b). The current coupled in by the E-field
can be carried away by a connection to ground
(Fig.2.16c), and then may induce an interference voltage
due to the transfer impedance, if the current has to
flow along the GS over an appreciable length.
The signal leads inside the GS intercept only a few
E-lines from the outside, which are responsible for the
so called "transfer admittance". If the internal leads
are coaxial cables with the braid connected to the GS or
tubes the transfer admittance is negligibly small.
Note that the region of interception of E-f ield lines is
often limited in size in contrast to the CM-currents
which couple in over the whole length of the GS.
----!---- E
3 · E
--------(a) (b)
®L;J®
®
(d) (e) ( f)
Fig. 2.16: The role of a GS as a shield. The GS intercepts
E-lines (a,b,c) very effectively. The induced voltage due
to a B-field perpendicula+ to the side walls (d) of the
conduit could be of the same order as that due to the
transfer impedance. B~fields perpendicular to the bottom of
the GS or parallel to the GS (e,f) are not harmful.
35
b. The GS has a shielding effect on external magnetic
-1.
fields. A B-field perpendicular to the side walls of a
conduit (Fig. 2.16c) couples in some flux depending on
the height to widtb ratio and and how close the leads
are to the bottom of the conduit. This coupling
process could be of the same order as the transfer
impedance effect (compare the fields in Fig. 2.16d and
Fig. 2.14). However fairly streng sources are required
to generate a perpendicular field comparable to the self
field. If coils generate the perpendicular field, the
effect still does act over a limited length, rather than
along the entire length of the conduit, as in case of
the transfer impedance.
Magnetic-fields perpendicular to the bottom of the GS or
parallel to the GS (Fig. 16e,f) are not harmful; in
Fig.16e the signal lead is assumed to be in the middle
of the grounding structure.
2.6.2 PROTECTED REGION FOR INSTRUMENTS

Most measuring instruments have a metal housing or are
installed in a metal cabinet on location and the question
shoµld be asked whether this metal housing creates a
"protected region" for the electronics. Although this is
true to some extent, the protection is not sufficient in
situations with high interference, e.g Pulsed Power or
HV-switching events. High intensity EM-waves cause
undesirable interference signals in these instruments. The
shielding provided by the metal housing should protect the
interior, but unfortunately due to slits, paint, plug-in
units and panels fastened with a few screws the shielding
is poor and cannot easily be improved. A limited amount of
shielding is however already adequate in case the distance
from source to victim is sufficiently large. Moreover the
apparatus proper is a relatively small size antenna.
More problems are caused by the leads, which carry signals
or power into an electronic instrument from far away. These
long leads can be efficient antennas to bring in large
common mode currents (Fig. 2.17).
36
signal cable
Electronic
Instrument
grounding mains power

connection
Fig. 2.17: Leads acting as large antennas. All leads carry

a CM-current to or from the system. The sum of the currents
is zero.
Most instrument housings cannot safely carry these
currents. Interference couples in because of the poor
"transfer impedance" of the instrument housing. With extra
metal ( a GS) we can create a protected reg ion for the
instrument, by diverting the external CM-currents away from
the instrument housing. The GS works like a "current
splitter"; DM-currents enter the protected region without
problem while the CM-currents stay in the outside circuits
formed by cable braids, the ground connections and the GS
(see Fig. 2.18). The EMC cabinet, depicted in Fig.2.19 can
serve this purpose. Since the currents flow mainly on the
raar panel the front can often be left open.
_J
Fig. 2.18: Grounding structure werking as a 11 current-

splitter11. DM-currents can flow into the protected region
while the external CM-currents are diverted by the GS.
37
Fig. 2. 19: An EMC cabinet di verts the CM-currents, and
strongly reduces the combined transfer impedance of cabinet
and electronic instrument (eg. an A/D converter
or computer). Note that the ground does not provide a
potential reference but is only a current path.
To keep the transfer impedance of the EMC-cabinet low, all

cable braids are connected all around to the rear panel. A
mains power filter, as well as attenuators, integrators or
buffers are well bonded to the same continuous panel.
Since the cables for signal transport have sizeable
transfer impedances, DM-interference signals may already
leak into these cables before they enter the EMC cabinet.
To improve the signal to interference ratio several options
are open. We can use cables with a better braid, or
transport signals with a large amplitude.
If signals of high amplitude are transported an input
attenuator is required at the point of entry. This
attenuator should have a flat response in the frequency
band of the signal and at least a non-increasing response
at the frequencies of the interference.
An alternative and attractive solution for wide band signal
transport is the use of differentiated signals that are
integrated again at the input of the electronic equipment.
The advantage is that interference, superimposed on the
differentiated signal is reduced by a factor proportional
to the frequency. The integrator should have a 1/w-response
38
in the frequency band of the signal and at the interference
frequencies a 1/w or lower response.
Note that many simple sensors, such as Rogowski coils,
magnetic probes and some voltage sensors [Wol 83] give a
differentiated signal.
The use of differentiated signals is comparable to the pre-
and de-emphasis noise reduction system for FM-broadcast or
to the RIAA correction system in record players.
Important for both options of signal transport is that
passive components are used in the first section of the
"receiver" (attenuator or integrator), i.e. where the
interference is to be suppressed. Thus interference signals
cannot reach the essentially non-linear electronics. The
passive first section should be mounted in a shielded
compartment next to the rear panel of the EMC-cabinet.
In cases where the integrator (or attenuator) must be
connected to the front panel of the instrument directly, to
maintain a correct termination, the lead can be fed in
through a.n "inside extension" as shown in Fig.2.2oa.
If the signal lead runs to the rear of the EMC cabinet
close to the side panel of the EMC cabinet the CM-current
in the side wall is a localized image current and
interference due to external CM-currents is reduced as much
as possible.
FRONT OF THE FRONT OF THE

INSTRUMENT INSTRUMENT
"
Fig. 2.20: Permitted side panel extensions:
(a) to the inside:
(b) the outside for instance for an attenuating
probe.
A horizontal cross-section is shown.
39
An other input problem shows up with an attenuating probe
such as t._1151 ~J,: lOOO~Tg}ctronix. It is difficult, to connect
the cable braid all around to the GS without damage to
these special probe cables. A solution is a 11 flexible
extension 11 also at the side panel as depicted in Fig.20b.
An different protection problem shows up in the case of

consumer electronics, where usually only a plastic housing
is present. In this case, where external CM-currents pass
freely through the instrument, interference couples in via
the various loops on the printed circuit boards (see
Fig. 2.21a) •
(a) (b)
Fig. 2.21: External CM-currents have free passage through

the instrument with a plastic housing (a). A simple
grounding structure reroutes the CM-current out of the
instrument (b).
A cheap solution to this interference is to apply the

concepts of compactness during the design phase (as
described in Sec.2.5). This could result in a simple
grounding structure and "connector panel" where the
external CM-currents are kept out of the instrument. An
extra rim all around the grounding structure reduces the
magnetic flux entering the protected region even more. A
further reduction of the flux is obtained with a ground
plane parallel to the printed circuit board.
40
2.7 APPLICATIONS
To illustrate the use ,of grounding structures to achieve
correct operation of sensitive electronic equipment in
situations with high interference we briefly discuss two
examples:
1. A lightning protection system for a telecommunication
installation [Deu 89].
2. The protection of control equipment in an open
substation [Hou 89].
Both are examples of the same basic solution; namely the
transport of a signa! over a long cable which is locally
protected against external CM-currents by a sui table GS
(see Fig.22a). In Ex.l the external CM-current is caused by
a lightning stroke, and in Ex.2 due to switching events in
a primary circuit of an HV-installation.
LIGHTNING PROTECTION SYSTEH

The considered telecommunication installation consists out
of a tower and transmitters located in a building at some
distances from the tower. The HF-signals are carried to the
smaller dipole antennas by coaxial cables. Figure 2.22b
shows the situation.
Normal lightning protection systems reduce Ohmic heating
( fire-protection) and prevent 1.ocal spark-over. For these
purposes sufficient contra! of lightning current flow is
provided by the metàl towèr and by grounding pins connected
to the footing of the tower. For the more critica!
protection of leads ana electronic equipment we'must reduce
the . fraction of the lightning current going to the
building. For this we need a grounding structure to create
a protected region to keep interference low at sensitive
inputs.
This protected region is formed by:
- a locally grounded EMC-cabinet for the transmitters,
- a conduit firmly connected between EMC-cabinet and the
tower.
- a conduit or a special tube at the inside of tower. Also
41
•
0
..... __ - - __ ,,,,,,,
(a)
...-TELECOMMUNICATION
TOWER
(b)
TRANSFORMER INSTRUMENT
-RACK
Fig. 2.22: (a) The general principle of protection by means

of a GS. (b) Utilized in a lightning protection system for
a teleco:mmunication installation. (c) Utilized for the
protection of control equipment in an open substation.
42
the inside of the ang,le iron leg of the tower can serve
as a part of the GS.
A further control of the lightning current flow is provided

by the placement of additional grounding pins. Extra pins
can be connected to the footing, but also to the outside of
the local grounding structure. The current is peeled off,
the higher frequency part already near the tower due to the
localized flux around the grounding structure.
Note that there is no need for overvol tage protectors or
spark gaps. A good grounding system keeps the interference
voltages low by itself.
OPEN SUBSTATION
Existing signal and control leads from a 150 kV / 10 kV
transformer in an open substation were reported to show
intense interference signals (a few kV) near electric
circuitry in the control room of the station. Extra
grounding pins installed near the transformer could not
solve the problem.
Figure 2.22c explains the situation. At energization of the
transformer from the 150 kV side, hf-currents in the
primary circuit (1) cause a magnetic flux that couples into
the secondary (signal/control) circuit (2).
To obtain substantial improvements the signal and control
leads are installed in a protected region. For this
situation the protected region was partly formed by the
transformer housing, a conduit firmly connected between
transformer and the instrument rack. In addition the
instrument rack was grounded locally.
Results of many experiments show that interference can
be reduced from kilovolts to less than 1 volt [Hou 89].
43
CHAPTER 3
THE TRANSFER IMPEDANCE OF GROUNDIRG STRUCTUR.ES
An EMC-analysis for the protection of electronic systems

and large scale electrical installations was given in
Chapter 2 of this thesis. The present chapter considers a
specific aspect of this analysis namely, the transfer
impedance of grounding structures (GS).
3.1 THE TRANSFER IMPEDANCE CONCEPT

The transfer impedance is a much used concept to specify
the interference properties of for instance coaxial cables
and cable connectors (see Fig.3.1).
RG 58/u
10-i' RG213/u i'trl 1
l2t-rl In1
RG214/u
['X.1
10- 3 KXSO 10 -•
EPD2205.9A
- - EPD214B3/A
Fl209/8
CELLFLEXiu
10~
7 L-_L~~...L..~__l'::--~....1...~~.L.-~~~~'-' 10· 1
10 3 107 f(Hz)- 10 1 10 8 1~ 9 f(:o~t

Fig.3.1: The transfer impedance of (a) coaxial cables
[Sev 86] and (b) cable connectors [Eic 85].
The transfer impedance of a coaxial cable is generally

defined as fellows (see Fig.3.2):
if an interference current IcK flows through the outer
conductor of a coaxial cable and it causes a voltage
difference V0 M at the open end of the cable which is short
circuited at the other end, then the transfer impedance Ztr
is given by the relation
VDM(W)
[O/m], l«A/4 ( 3. l)
ICK(W) l
where l is the length of the cable.
44
very large plate
short !CM ____) ~eet bonds
e_:r:_u~t--------- - - - - - - - - - - - ___ _
)
l
Fig.3.2: Sketch illustrating the definition of the transfer

impedance of a coaxial cable.
For cables longer than a quarter wave-length the DM-voltage

and DM-current due to the interference current can be
calculated with a basic transmission-line model which
contains per section dz an extra distributed internal
series-voltage source E(z)dz where E(z) equals the product
of the transfer impedance and the common-mode current (See
Fig.3.3).
I+~~z1 d z
I (z)
+ + dV + ZI
E(z)dz a:z E(z)=ZtrICH
V(Z) Ydz v+:~dz dI + YV
a:z = 0
dz
Fig.3.3: Equivalent circuit including the transfer

impedance Ztr, where Z is the series impedance and Y is the
shunt admittance (Van 78).
Schelkunoff was the first to calculate the transfer

impedance of a coaxial cable with solid outer conductor.
Instead of "transfer impedance" he used the term "mutual
impedance 11 [Sch 37).
Kaden calculated the transfer impedance of coaxial
structures for various simple outer conductors. He used the
term "Kopplungswiderstand" [Kad 59].
45
The value of Ztr' both for braided coaxial cables and
connectors is usually not calculable, but can be determined
experimentally from a measurement of the DM-voltage caused
by an in]eétea--ëj.f":current.
The concept of transfer impedance, as def ined for coaxial
cables, can be used for a quantitative comparison of
grounding structures, both by calculations as well as by
measurements.
Clearly a properly designed grounding structure should be a
structure with a low transfer impedance. Through this
stucture large external CM-currents may flow, and still no
dangerous voltage differences are induced: the main aim of
"grounding", or more general, of EMC-practice.
All grounding structures described in this thesis, fall
into two categories: "Grounding structures to protect
leads" and 11 Grounding Structures to protect instruments" as
introduced in Section 2.6. The concept of transfer
impedance will be applied to both structures.
First we def ine the transfer impedance of a GS to protect
leads:
When an interference current IcH flows through an GS (see
Fig.3.4), it causes a voltage .difference
·
VDM between the
points 1 and 2. The transfer impedance ztr of the GS is:
VDM(W}
ztr (w) ICM (w} 1 [O/m] I l«i\/ 4 (3. 2)
where 1 is the length of the measuring loop.
GROUNDING STRUCTURE

impedance z tr of a GS to protect leads.
46
Secondly we discuss the transfer impedance of a GS for
instruments.
Due to for instance the finite size of a GS and an
imperfect cable bond, CM-currents carried by leads coming
from the "outside world" cause a CM-voltage in the
protected region (see Fig. 3.5). The transfer impedance
will be given by:
VCH
z • [O] (3.3)
tr ICK
Note that the voltage vcK in Fig. 3.5 can drive a common
mode current in the right hand part of the cable if its
bra id is grounded at the far right. In that case the
behavior of the GS can also be described by a ratio of
currents.
FINITE PLATE
____) CABLE BOND

------------- -------------

impedance of a GS to protect an instrument.
These definitions are a starting point for an evaluation of

the transfer impedance of GS' s. Measurements as well as
calculations will be presented for the GS for leads and GS
for instruments in sections 3.2 and 3:3, respectively.
The signif icance of the transfer impedance concept for

grounding structures is that simple standard tests can
provide its value and thereby a meaningful criterion for
the quality of the grounding structure (including the whole
"layout" of the "network"). The non-calculable overall
transfer impedance of a large extended grounding structure
can be determined by (CM-) current-injection test methods
in which the voltage difference over critica! terminals is
47
measured. Examples of such tests are presented in:
"Measurements of Currents Around and in Large Grounded
Str~'9& 11 {l>eu -33}-, and in; "I.oGal Proteotion of
Equipment in HV-Substations" [Deu 89]. Also Bersier's
interference e~eriments with TV-sets presented in
"Measurement of the. immunity of TV-Receivers to AM,
RF-fields in the 3 to 30 MHz Range, including the influence
of connected cables" [Ber 81] is an example of a
common-mode current-injection test.
3.2 THE TRANSFER IMPEDANCE OF GS's for leads

A GS for leads provides a controllable current path for
external CM-currents and creates a protected · region for
leads by means of its shape.
3.2.1 AN IDEAL GS, A TUBE

The best, but in real situations not always a practical GS,
is a completely closed tube (see Fig.3.6). The transfer
impedance for a tube with a wall thickness d small compared
to its radius r i s given by [Kad 59, pg.294]:
k p
(3.4)
2 rr r sinh kd sinh kd
1/2
where k = 4L· a = (
2P )
W µOµr
is the skindepth, p the
resistivity, µ 0 µr the permeability and R0 the
DC-resistance.
The magnitude of the transfer impedance IZtrl is:
2 ad·Ro
1ztr 1 = ....(_c_o_s_h_2.... )1-/-
~----c-o-s--2i-=--..... 2 "'
(3. 5)
d<<S
èl<d
48
A B
i------------------
Cu
Fig. 3.6: The transfer impedance of a thin-walled tube made

out of copper or iron.
We may characterize two frequency regions A and B.

-In region A, O<f<f 1= p/(rrµ0 µ r d 2 ) , in ether words where
d<3, the transfer impedance remains close to R0 .
-In region B , f>f 1 , the transfer impedance relates the
electric field which has "diffused" to the inner tube
surface to the total current. For f>>f 1 the ztr is almost
zero in this nice, closed geometry.
The right hand part of Fig.3.6 depicts a comparison between

iron and copper tubes. Although the R0 of an iron tube is
higher than that of a copper tube, the permeability of iron
gives already at low frequencies a small skin depth, so
that the ztr of an iron tube falls below that of a copper
tube.
3.2.2 PRACTICAL GS's, PLATES OR CONDUITS

More practical GS's have the shape of a plate or a conduit.
The transfer impedance of these GS's can be determined from
Eq.3.2: see Fig.3.4 and Fig 3.7:
zt"= VDH/(ICH 1) [Q/m]
where V0 H is the interference voltage between the points 1
and 2, IcH the interference current flowing through the
grounding structure, and 1 is the length of the measuring
loop, L ( see Fig.3.7).
49
Fig. 3. 7: The condui t shape of the grounding structure
leads to a rearrangement of current and field and
consequently to a low transfer impedance.
The voltage is given by:

2 1
V =-JE. dl =jwµ JJ H dS + JE dl (3. 6)
DM l OS x s
4
2
where S 2 is the area enclosed by loop L. Assumed is an
exp(jwt) time-dependance; the E-field along the integration
path 2-3-4, in the measuring wire, is zero. In this
expression, the electric field E 8 on the inner surface
(y=O) of the conduit as wel! as the magnetic field H in
' x
loop L, are unknown. When the skindepth a in the conduit is
less than the thickness d of the structure the electric
field in the bottom of the conduit can be approximated by
the expression for the electric field in a conducting half
space:
E:z ( Y ) = k PHseky ' where k- 1 + J
- T ' <15
=
Wil Il
1/2 (2..2...)
o r
so that Es= E:z(O)= kpH8 (3.7).
H
s
is the magnetic field strength at point P on the
inner surface of the structure. The magnetic field through
surface S 2 is assumed to be constant and equal to Hs . With
these simplifying assumptions the transfer impedance
follows from eq.(3.6):
H
Z = -I• (kp +jw1.1. Ay) , (3. 8)
tr CM 0
which is valid for a << d. The expression between brackets

depends on the resistivity p and the permeability µr of the
50
grounding structure, the frequency and the height 6.y. The
ratio Hs /I CM depends on the shape of the conduit.
Figure 3.8 shows the magnitude of the transfer impedance of
an open GS as a function of frequency for typical conduits.
lztrl lzt
A B
l~J 1
1 I;J ,"
1
1 Fe 1
-i, '1
'
1 \
1\ c"
1
1
'" \ ... _
1
1
fl f2
( a)
-
f3 log f CHiJ
(b)
--
log f [1-1.a]
Fig.3.8: Magnitude of the transfer impedance of (a) an open

GS. (b) A comparison of an iron and a copper conduit with
the same dimensions.
We characterize four regions A, B, C and D (Equation 3.5 is

valid for region A and and gives an estimate for region B,
Eq. 3.8 is valid for the regions C and D).
p
-Region A, O<f<f 1= 2'
in ether words d<a ,the
rr µo IJ." d
transfer impedance is close to the DC-resistance R0 •
-Region B, fff<f 2 , where f 2 "' 25f 1 , which means ll (f2 )"' o. 2d.
In this region two phenomena characterize the behavior of
the transfer impedance. The skin effect leads to a
decrease of ztr , whereas Eq. 3. 8 predicts an increase of
Ztr' Depending on the height to width ratio of the
conduit, and the value of p and µr the IZtrl shows a
minimum or a monotonie increase.
pµ
-Region c, f 2 < f < f 3 , f = r The transfer
3 2rrµ (6.y)2
H o H p
impedance is given by Z "' Is kp,
tr CM
andlZt l"'
r CM
v'2' ;s- • i-
-Region D, f > f 3 . The transfer impedance is given by
H
Z "' - 1 8 jwµ 6.y and is independent of the properties p
tr CM 0
and µr of the metal of the conduit.
51
Figure 3.Sb also compares an iron and a copper conduit with
the same dimensions.
-For low nequenc1es th1s c0111par1son is easy: the
dc-resistance of the copper conduit is smaller than, that
of the iron conduit.
-For the intermediate frequency region the comparison is
more complicated than fora tube (Fig.3.6). The magnitude
of the transfer impedance depends on material constants and
geometry as explained before for the Regions B and c in
Fig.3.Sa.
- Finally for high frequencies, 1 Ztr 1 Cu 1Z 1 ~ and is
tr Fe
independent of material constants
Note:
(1) Due to the higher surface impedance of iron, in
comparison with copper, the external CM-current might be
reduced more by an iron conduit than by a copper one.
(2) For large currents iron will be saturated. For an
U-shaped conduit saturation occurs where the magnetic field
is high. This phenomenon resul ts in a higher leakage of
H-f ield into the protected region and thereby in a higher
transfer impedance. Note that for a tube the fields near
the inside are small, so that in the case of an iron tube
saturation will not reàdily occur there. In the symmetrie
iron tube the H-field remains always zero at the inner
surface.
THE SHAPE-FACTOR g.
The ratio Hs /I CM in Eq.(3.8) depends only on the shape of
the GS. When we bend a plate in a u-shape the field HB (c)
at point P of a conduit will be considerably less than
H (p) at point P on the surface of the original plate (see
s
Fig.3.9). This reduction can be expressed by a
dimensionless shape-factor g, defined as:
H
.s
(c)
g = (3.9)
where IcM' d and 2h+2w=2b are kept constant.
52
2b
p H e 'A.
H
ellipse
p Hs (p)
plate
p Hs(c)B'
B
p
c c•
Fig.3.9: The field strength H along the horizontal surface

of several GS's. Note that H is zero in an inside corner
and very large at the outside corner of the plate and the
ellipse.
The transfer impedance at higher frequencies becomes:
H (p)
Z
tr
=g _ I s (kp+jwµ Ay) ,
O
8<<d. ( 3. 10)
cm
The field H (p) at point P on the surface of a thin plate
"
is well approximated by the field H s (e) in point P, for a
structure in the shape of an ellipse [Kad 59]:
IC)(
H5 (e)= rrb , 8<<d. (3.11)
2
The Z tr of a conduit formed out of a thin plate with
original width 2b is therefore given by the simple
expression
(3.12)
1 --·o-.
... " ........ kept constant:
0.92 'o....
O.B4'-..._ 0.71 I
cm
'·et.. 2h + 2w
' ' , 0.49
0. 5· d
'o.
' , 0.38
',o.......
...... --o..
0.26
0.19
... __ -<>---.. . .
0.135
0
0 1 h/w- 2
Fig.3.10: The shape factor g as a function of h/w. The
o-dots are calculated. The return conductor for Icx is
supposed to be far away.
53
Figure 3.10 gives the calculated shape factor g, for some
structures. The field strengths in points P on the surface
of the .U-shaped H-~S. .. lrlere determined by means of a
Boundary Element Method (BEM)-computer program [Oef 90].
TRANSFER IMPEDANCE MEASUREMENTS

By means of a current-injection test we also measured the
transfer impedance of a plate and two conduits, shown in
Figure 3. 11. The condui ts were made out of a 1 llll!l thick
aluminum plate of 270 mm width. The 4 m long conduits were
placed along the axis of a metal tube with a inner diameter
of 310 mm which served as the return conductor for IcK·
90
Fig. 3.11: GS's placed in a metal return cylinder for the

measurement of Ztr .
SETUP
Figure 3.12 shows a schematic diagram of the setup.
' ~ D~TA'I=L==B======~
~ r. EMC-
CABINET
-CONDUIT
'·tl
.
L:JB /
100 DETAIL A
Fig.3.12: setup for transfer impedance measurements of GS's.
54
To obtain a reproducible injection current a 0.5 µF, 25 kV
capacitor was discharge,d through a spark-gap into a 3: 1
pulse transformer. The,. secondary of the transformer was
connected to the te~n conduit and to the outer tube, which
were short circuitéd at the right hand end by a metal disk.
The injected current I(t) through the conduit, measured
with a Pearson probe, is shown in Fig.3.13a. The magnetic
field Hs (c) was determined by measuring the change of the
magnetic flux within a single turn loop (See Fig.3.12,
detail A). The voltage signal VA(t) induced in this loop is
shown in Fig.3.13b • The output signals were transported by
RG-214 cables surrounded by copper tubes, "grounded" on
both ends, and terminated with 50 n. The signal data were
registered with a Nicolet 4094C digital oscilloscope (200
MHz max. sampling rate, 8 bit resolution), which was placed
in an EMC-cabinet.
4 I
0
e.5 VAO
2
-e e
-2
-e.5
-4
-:6 -1
-10 e 10 20 ao 40 -10 e 10 20 ao 40
t [µs] t [µsJ
~
Fig. 3.13: Waveforms of the injected current I(t) and the

voltage VA(t) induced in a single loop.
EXPERIHENTAL RESULTS AND DISCUSSION

Table 3A gives the results of the measurements. The values
of VAo' I 0 , the amplitudes extrapolated to t=O, and w are
obtained from waveforms as in Fig.3.13a by curve fitting
[Sta 87).
The magnetic field measurements are in good agreement with
the calculations, under the same geometrical condi tions
(Table 3A,. column 7) •
55
H-FIELDS measured calculated
Return cyl1nder R-0.155m R-lOOm
h VAD HBEH HBEH
w I0 v AD w H =Cl.
.I 0 .w
• 0
A
. jJ
o
0. 155 1DD
"v rad/s A/ m for 1 Amp A /m A /m
····ex
r r f . case
0 915 825
9. 22
x1D
5 0.95 0.95 1.17
8
9. 17
1
1012 856 5 0.81 0.79 0.98
2 "1 D
···········--···- ·-··-···-····-- ·························-··-···"··· . ··········-····-·-·-····-·····- ····-·····--········-···- ·-·-·"'
0 1
2
2
824
3
121. 7
4
9, 21
"1 D
5
5 0.156
6
0.153
7
0.158
TABLE 3A: Measured H-field values with loop A at point P

(column 6, radius return cylinder R = 0.155 m) for a plate
and two conduits (column 1). Column 7 gives H-field values
obtained from BEM-calcuations under the same geometrical
conditions. Column 8 gives calculated H-field values but
now for the situation with R = 100 m. VAo and Io, the
values of the amplitudes of V(t) and I(t) extrapolated to
t=O, and w are obtained from measurements and by curve
fitting. The H-fields are given for a total current of 1
Amp.
return cyl1nder R=0.155 R= lOOrn
SH AP E
M BEM BEM
FACTOR g g0.155 glOO
re f c a se
0 1 1 1
8 0.85 0.83 0.84
0 1
0.164
2
0.166
3
0.135
4
TABLE 3B: Shape factors (column 2) for a plate and two
conduits (column 1) calculated from measured H-field values
with loop A. Columns 3 and 4 give shape factors, calculated
from H-field values obtained by BEM-calcuations, in the
cases of return cylinder radii R = 0.155 m and R = 100 m,
respectively. Note that the shape factor for the plate is
kept equal to unity, both for small and large outer cylinder.
56
Experimental parameters such as the effective area , o A, of
the pick up loop were not very accurately known. Therefore
we introduced a correction factor, Cl, to adjust all
measured H-fields; this factor (which turned out to be
0.92) is obtained from the case of the flat plate, where
the calculated field should be very reliable.
Because the return cylinder of radius 0.155 m is fairly
close to the centra! conductor, a "proximity effect"
modifies the current flow in the conduit, and also the
field around it. In practice conduits may be at a large
distance from the return conductor. Therefore a second
series of BEM-calculations (also used in Fig.3.10) was
carried out for conduite with the return conductor at a
larger distance (100 min the BEM calculations). The column
at the far right of Table 3A shows the resulting H-fields.
As expected, we see an appreciable difference for the plate
and a much smaller one for the more compact conduit with
h/w=2.
Similar differences show up when the results are expressed
as shape factors (Table 3B), where the H-fields are
normalized to make them unity for the plate. ~ote that the
shape factor for the plate is kept equal to tinity, both for
the smaller and the larger cylinder.
The above mentioned 11 proximity effect" also influences the
current density distribution at the insi._de of the return
cylinder. Calculations show the influence of~ the proximity
effect on the current density distribution at the inner
surface of the return cylinder (See Fig.3.14).
To illustrate this effect we determined the current density
j 8 (t) by measuring the Change Of the magnetic flUX Within a
single turn loop close to the cylinder surface (see
Fig. 3 .12, detail B). The voltage signal V8 (t), induced in
this loop has the same damped eosine behavior as VA(t) in
Fig.3.llb.
Table 3C gives the results. of the measurements. The values
of VBo , I o and w are again obtained from wave forms as in
Fig. 3.13 by curve fitting [STA 87].
57
SURF ACE C U RRE NT DEN SITY measured
return cy11nder R==O.l55m
~
h Veo jBEM
w I
0
v BO w j
8
=C2. 0 .µ
.I 0 .w B
B o
A n;V t ad Is A /8' for l A lll p A /M
r e{ . ~ ~ ..~" ..~ '!l. ll!I!
1
~
8 55 48.8 XIO
s l. 026 1.026
8
9.22
0 9 18 56.3 x 1 ()
s 0.52 0.47
g
........ "" ·······-····· .............. f··-"····-·'"-· .. ..." . ".."·--·-······---··
"
9. 17
1
l 0 12 26.9 5 0.434 0.429
2 ><10
...... .. ····- ..................... . .... "".
,..,".,_""." .., . --·····-··-····-•·<>·--·--·····"--·-~-~
@ 1
2
2
8 24
3
4 3.4
4
9' 21
)< 10
5
5 0.94
6
0.88
7
TABLE 3C: Surface current density jB, measured (column 6)
for a tube, a plate and two conduits (column 1) with a
return cylinder with radius R = 0.155 m. Column 7 gives
jB-values obtained from BEM-calculations under the same
geometrical conditions. VAo and·· Io, the values of the
amplitudes extrapolated to t=O of the induced voltage VA(t)
and the injected current I (t), and w are obtained from
measurements by curve fitting. Thè surface current is given
for a total current of 1 Amp. ~
/ /;:~=::~<,''~
((,~)\)
\ ~/
-~ /
(a) (b)
a- 2n
IHIÎ ljdl
__ .cJty_____
l~ !~
A (c) A' (d)
Fig.3.14: FEM plots show the H-field lines for a strip in
an return cylinder with R=0.155m (a) and with R=.lOOm (b).
BEM-plots (c) show the H-field values at the surface of the
strip for both situations. The surface current density at
the inner surface of the return cylinder is shown in (d) as
a function of the angle e.
58
We introduced also here a correction factor to adjust all
measured current densities: thïs factor (which turned out
to be 0.94) is obtained from the case of a tube as central
conductor, where the current density is homogeneous.
Excentricity of the inner conductor may be the cause of the
slight discrepancies found for the plate and the conduits .•
3.2.3. ALMOST IDEAL GROUNDING STRUCTURES: COVERED CONDUITS

Metal plates and conduits provide an excellent conducting
path for external CM- currents and create well protected
regions. In situations with intense interference, the
transfer impedance of these grounding structures can be
further reduced by the use of non-touching metal covers of
finite length in addition to the conduits, as depicted in
Fig.3.15. o.1sa
... A lp A'
·=
A
Fig. 3.15 : A conduit with h/w=2 with and without a cover.
The graph shows the reduction of the H-f ield at the bottom
of the conduit c.aused by the non-touching metal cover.
With this configuration an almost ideal GS, can be

constructed, which is much more practical than a tube. At
high frequencies the metal cover changes the distribution
of currents and forces the already weakened magnetic field
further away from the leads. Induced currents flowing at
the outside surface of the cover are concentrated at the
corners. These currents are parts of loops which will be
closed by currents which return as current density
concentrations at the overlap of cover and conduit (See
Fig. 3.16).
59
9 10
ît .f 30 mm
Ijl! 6,.!55
I• 1
1 1
~-"---- -,
"1
[~ conduit 1 '
surf ace I1 '\
7 4 r.
cover
surf ace
Fig. 3.16: The surface current density in the side wall of

the conduit (points 1 to 5) and in the cover (points 6 to
10), normalized to a total current of 1 Amp. The
concentration of the current density at the overlap of
cover and conduit causes a concentration of the H-field in
the overlap gap.
In the overlap-gap we get a concentration of the H-field.

since §H.dl around the cover is zero, the magnitude of the
H-field in each of the two gaps, Hgap' is approximately:
Hgap I'> o. 5
-1- I (3 .13)
lnd
0V
where I lnd is the current induced in the outer surface of

the cover and l ov is the length of each overlap. The
induced current is roughly equal to o. 4xicK' since the
cover represents about 40% of the total circumference,
through which the net interference current IcM is flowing.
We can rewrite Eq. 3.13:
H
gap
~ 0.2
l
ICK (3.14)
ov
The total interference flux per meter, •int, passing the
gap can be estimated by:
d
41 Int = Hgap . d i;, 0. 2 - ICM (3.15)
1
ov
where d is the width of the gap.
According to this formula we obtain a reduction of •int
60
when we use a large overlap length and a small gap width.
We get also a reduction of ~ lnt when we can reduce 1 1nd in
the cover. This in fact happens when we place the cover in
the conduit instead of over it. This statement can be
confirmed by BEM-calculations (See Fig.3.17).
B
3
c
4
• "--- -- D
-0
j9!!P,,! - 2_ _ _ ___
2.9
5
cover.in cover over

condu1t conduit
Fig.3.17: The current surface density j in the side wall of

the conduit for (a) from points A to E, and for (b) from
points 1 to 5, normalized to a total current of 1 Amp. The
current density in the gap is lower for the "inside" cover.
Although the interference flux t 1nt in Fig.3.17b streams in

closer to the bottom surface of the conduit than in
Fig.3.l7a, the magnitude of t lnt is much smaller in case b.
The H-field at the bottom of the conduit is therefore
smaller for the inside cover. Figure 3.18 gives the
fieldstrength at point P as a function of the overlap-
length for a few conf igurations of covers placed over and
in conduits.
2
-~
0.158
p °'
1 90
H-values for 1 Amp
0.045
~1 ~0.031
o.~
0.024
~D
"'--
o.oit I' 'I
0 10 20 30
---overlap [mm]
Fig. 3 .18: The H-field at point P as a function of the
total overlap length for both outside and inside covers.
61
Because of the time variation of the flux in the gap a
voltage difference is induced between the non-touching
oeve:r- and the conduit (See Pi9.3.l~). streng time varyinq
0
"" •" ."

+L
'
-L +L
+V
-v
-L
Fig. 3.19: The magnetic flux in the gap induces a voltage

between cover and conduit. (b) The current through the bolt
in the middle of the cover is zero for synunetry reasons.
fields may produce sparking between cover and conduit. To
reduce this voltage difference and to avoid sparking and
its subsequent interference two solutions can be
considered.
1. We can use short segments of non-touching covers to
limit the flux in the gap. To fasten the cover, one
bolt in the middle at both sides of the cover is
permitted. No current will flow from cover to conduit
because the bolt does not change the voltage
distribution between cover and conduit (See Fig.3.19b).
2. A more expensive solution is to make full metallic
contact between conduit and cover over the entire length
of the cover. When we use bolts we need many to reduce
the resulting current density concentrations.
As a result of the above analysis we show in Fig.3.20a an

example of an almost ideal grounding structure which could
carry leads between 11 subsystems 11 • A cover is fitted in the
inside of a conduit, a bit below the top of the vertical
sides of the conduit. Contact-springs allow currents to
cross over the entire length of the conduit. We use round
edges and more metal at places where we expect current
density concentrations. The realisation of such a grounding
62
structure for protection of electric circuits against
unwanted electric signals may be costly, but is certainly
feasible. Many other configurations can be designed (See
Fig. 3.20b).
m
(a) (b)
ca. structure
Fig. 3.20: (a) An almost ideal GS, and (b) some other
possible GS' s.
The combined use of a coaxial cable and a grounding

structure resul ts in a very low total transfer impedance
Z
tr,tot
• This ztr,tot can be easily calculated (see
Fig.3.21a).
The external CM-current, I CH,ext , induces a current I CH,int
in the internal CM-circuit equal to:
I CM, int= (I C H , ex t
Zt r , GS
)/Z C H , 1nt
(3.16)
where Ztr,GS is the transfer impedance of the GS and

zCM, int is the impedance of the internal CM-circuit.
The induced interference voltage V0 H in the cable becomes:
I CH,int Ztr,cable (3.17)
where
·
ztr,cable is the transfer impedance of the cable.
So the total transfer impedance becomes the expression:
Z - (Z /Z ) Z ( 3. 18)
tr,lot tr,GS CH,lnt tr,cable
For high frequencies this expression reduces to:

Ztr,tot = (M/L) Ztr,cable (3. 19)
where M is the mutual inductance between the external and

internal CM-circuit, and L is the self inductance of the
internal CM-circuit. When we use a proper grounding
structure M is much smaller than L.
63
Further reduction of Ztr,tot is achieved when we increase
zCM, int by e.g. a ferromagnetic layer around the coaxial
cable (see Fig. 3 • 2 lb) . This layer could consist of iron
tape or of ferrite rings which increase either the
inductance or the resistance in the internal CM-circuit.
coaxial cable, Ztr cable
Internal CM-circuit
I
CM,Ext
External CM-circuit
Ferromagnetic tape
Fig. 3.21: (a) A coaxial cable in the protected region of a

GS. The voltage VnM due to the current IcM, tnt is very low.
(b) A ferromagetic tape wound around the cable increases
the internal CM-circuit impedance.
Note:
(1) The GS's described above offer protection over a wide
frequency range: in the literature protection is only
judged possible over a small frequency range. A quote from
(Den 74) is "neither of the four fundamental methods
(isolation, single-point grounding, multiple-point
grounding, or noise reduction) of minimizing the noise in
the signal transfer loop will by itself provide
interference-free operation over the entire range of
amplitudes and frequencies encountered in many
installations".
( 2) For the low frequencies (e.g. 50 or 60Hz) the
transfer impedance of the GS is about equal to the
DC-resistance. This DC-resistance .is generally low because
of the bulky metal GS.
64
(3) An iron conduit has a number of advantages over an
aluminum or copper conduit:
- The small skindepth leads to a rearrangement of the
current, already at low frequencies, as shown by the drop
in Fig.3.8.
- The external IcK can be reduced by the extra impedance (R
and L) of the thin outside skin.
- The inside V0 K sees a higher ZCM, int as a resul t of the
extra impedance.
3 •3 THE TRANSFER IMPEDANCE OF GROUNDING STRUCTURES FOR

PROTECTION OF INSTRUMENTS
In Section 2. 6 we mentioned that long leads, which carry

signals or power towards an electronic instrument can be
eff icient antennas which bring in large common mode
currents. Interference may couple in when these currents
are not correctly treated. For this protection of
instruments we need a GS through which the useful
DM-currents can flow into the protected region, whereas the
external CM-currents are diverted (See Sec.2.6, Fig.2.18).
In situations with a high interference level where much
protection is necessary an EMC-cabinet can be an eff icient
solution. For the protection of instruments in situations
with a moderately high interference level a search for a
solution with less metal (a simple GS) is meaningful. Such
simple GS's are of course conceptually related to the
EMC-cabinet. The transfer impedance of these simple GS' s
will be discussed in this section.
Note that the transfer impedance is only meaningful when
primary and secundary circuit are defined. In this section
we keep these circuits the same and study the influence of
the size and geometry of the GS on the transfer impedance.
As a step in our analysis of the transfer impedance of
simple GS's for the protection of instruments we consider a
single coaxial cable with its cable braid connected all
65
around to the GS (A) and one additional grounding lead (B)
connected to the same GS (see Fig.3.22). These leads are
mounted perpendicularly to the GS, at a fixed distance of
each ether. . ]<'()~ this type of GS we neg leet the small
"resistive" part of the transfer impedance as is reasonable
at higher frequencies. We consider only the mutual
inductance M between the external CM-circuit and a
"testloop" . placed at a fixed position in the protected
reg ion. Thus for the transfer impedance of the grounding
structure we may use the expression Z= jwM.
GROUNDING STRUCTURE ~
PROTECTED
REGION
---------
coaxial cable
---- --------- _, ___ _
grounding lead
EXTERNAL CM-CIRCUIT 1f' INTERNAL CM-CIRCUIT
Fig. 3.22: A coaxial cable and one grounding lead mounted

perpendicularly to the GS. The 11 testloop 11 is located in the
protected region, in the same plane as the leads.
AN IDEAL GS FOR AN INSTRUMENT

An ideal grounding structure in this situation would be an
infinitely large metal plate or a completely closed metal
box. In the case of a infinitely large metal plate the
external CM-current spreads over the left hand side of the
plate in a "line dipole"-pattern as shown in Fig. 3.23.
The components of the current density are given by [Ram 84]:
2 2
I a(r -a )sine
J= (3.20a),
TC
r
r +2 .r2
2
-I a (r2 +a ) cose (3. 20b)
TC 2 2
r.r
+
r~ = r +a +2arsine: the
2 2 2 2 2
in which r r +a -2arsine,
+
coordinate system is given in Fig. 3.23. These results are
used later as an approximation for a finite size plate. For
66
the infinitely large plate or the closed box the mutual
inductance, between the external CM-circuit and the
internal circuit, is zero just as the transfer impedance of
the GS.
Fig. 3.23: The current density distribution for an

infinitely large plate has the shape of the field lines of
a line dipole.
REAL GS 1 s FOR INSTRUMENTS

In reality only plates of finite size are used. The
transfer impedance of such a GS with finite size is
analyzed for the example of a circular disk as grounding
structure (see Fig. 3.24). The analysis is based on
measurements with the setup shown in Fig. 3.27. In addition
to the measurements we derive an approximate relation
between the mutual inductance, -the M between the external
CM-circuit and the testloop-, and the diameter of the GS.
2a
Fig. 3. 24: A GS shaped as a disk. The disk serves as a

simple geometry for which approximate calculations are
possible. The transfer impedance depends on the radius R,
the distance 2a between the injection wires and the
location and size of the test loop at the right.
67
To obtain an approximate expression for the mutual
inductance M, we consider a disk with radius R, where R is
assumed to be large with respect to the distance 2a between
the leads.
For the surface current density distribution, at the left
side of the disk, we use -as an approximation- the
solutions J r and J e for the infinitely large plate in the
region r<R (Eqs. 3.20a,b). The values of J (R) and J (R) at
" e
the boundary of the left side are:
J (R)=
r
!n ~2
R
sine and {3.2la,b)
obtained from Eqs.3.20a,b with R>>a.
Continuity of J{R) between left and right hand side
requires:
J~(R)= -J~{R) and J~{R)= J:(R) {3.22a,b)
A approximate solution is that the current density J at the
right hand side is constant over the entire right hand
I a
surface with the value
R
Jy(x,y)= n R2 (See Fig. 3.25).
x X' X' x
Left hand side Right hand side
Fig. 3.25: Approximate surface current density distribution

patterns for both sides of a disk.
To calculate the mutual inductance between the external

CM-circuit and the testloop we assume that the field
strength at the surface, Hx which is equal to J y , is
constant over the small testloop area STL.
This derivation for M leads to:

µoa
M= 1l R2 STL (3.23)
This approximate expression shows that the mutual
68
inductance, -and thereby the transfer impedance-, will be
reduced when we reduce the distance between the leads or
enlarge the area of the disk.
We expect a further reduction of the transfer impedance
when a cylinder is attached to the disk (Fig.3.26b) or when
the disk is enlarged toa strip (Fig.3.26c). For the strip
the question can be asked in which direction (x- or
y-direction) the disk should be expanded to achieve the
lowest transfer impedance. No calculations were possible
for the disk with cylinder and for the strip. To answer
these questions we have measured the transfer impedance of
a disk, a disk with cylinder and the strip as shown in
Fig.3.26; all by means of a current injection test •
• y • y
x x
Fig. 3.26: A disk, a disk with an attached cylinder, and a

strip, with injection wires in the y-direction.
TRANSFER IHPEDANCE HEASUREHENTS.

Three series of measurements have been done:
-we measured the Zlr of a brass disk as a function of the
disk radius (R= 25, 50, and 75mm.),
-we measured the Zt.r of the same disks now with cylinders
attached; the dimensions of the cylinder were: r= 18 mm
and the length was taken to be 10, 20, 30, 50 and 70 mm,
-we measured the zt.r of two brass strips with dimensions 50
by 400 mm and 100 by 400 mm, both with the connections
perpendicular or parallel to the long direction of the
strip.
The distance between current injection wires is 16.9 mm for
all measurements.
69
By measuring the injected current I inj (See the setup shown
in Fig.3.27) and the induced voltage Vind in a test loop we
determined the mutual inductance M between injection
circuit and testloop. The mutual inductance is obtained
from the relation M = JV lnddt/I lnJ . The integration of V
and calculation of M is carried out by means of the
software package vu-point [VU- 87].
SETUP
Fiqure 3.27 shows a diagram of the setup.
Pearson test loop

probe EMC-
v
I~jection
~~- CABINET
RG 214/
Pul se
transformer
Fiq. 3. 27: current inj eet ion test for transfer impedance
measurements. The test section is fixed on a wooden table.
We obtained a reproducible injection current, -a 1250 A,

127 kHz, damped sine wave- , by discharging a 0.5 µF, 25 kV
capacitor through a spark-gap into a 3:1 pulse transformer.
The secondary of the transformer is connected to two
parallel wires, which were _ both connected to our test
object. The length of the 11 Lecher 11 -line (1. 25m) was short
compared to the wave length.
The injection current was measured with a Pearson probe.
our test loop was a single turn loop placed in a fixed
position as shown in Fig. 3. 27. The output signals were
transported by RG-214 cables surrounded by an extra copper
braid, connected to the RG-214 on both sides. The cables
were terminated by 50 Q and the signals were registered
with a Nicolet 4094C digital oscilloscope (200 MHz max.
sampling rate, 8 bit resolution) which was placed in an
EMC-cabinet.
70
EXPERIMENTAL RESULTS AND DISCUSSION
The first tests with the setup were carried out for two
simple situations, for which calculations. with the formula
of Biot and Savart are possible (see Fig.3.28a, "wire
structure 1" and "wire structure 2"} and which serve as a
check on our test method. The measured mutual inductances
between injection circuit and test loop of wire structure 1
and wire structure 2 are in good agreement; with results
obtained by the formula shown in Fig.3.29.
ar
IJ. .
M~Jl'[J rxdl
4n • 3
J• dS
·
S L r
L
Fig.3.29: Wire structure for which Ztr follows from a

Biot-Savart calculation.
These two cases show that a simple rerouting of the

interference current, by a return of all leads at one side
of an instrument (case2) already significantly reduces the
coupling compared with the situation where interference
currents flow freely into that instrument (wire structure
1, "a worst case"}.
Figure 3.28 shows the results of the described series of
measurements for a disk and a disk with cylinder. By using
a GS, - in our measurements a disk, a disk with cylinder or
a strip-, we obtained large reductions of the mutual
inductance. For example: a disk with a radius of 75mm as
grounding structure compared with wire structure 2 gives a
reduction of a factor 44. If our test loop would correspond
to a loop formed by copper strips on a printed circuit
board, this same reduction by a factor 44 is not easily
realized by a redesign of the printed circuit board.
The magnitude of M with a disk as grounding structure,
calculated with the approximate formula 3.23, deviates
considerably from the measurements (see Fig.3.28a). To
explain this deviation we note that the derived approximate
formula is based on a two-dimensional solution for an
infinite plate, while the real geometry is essentially
71
I
. : 9.7nH M: measured o
C: calculated +
"' [ 1
~ ;:: ~TWO
M: 9.4nH
WIRE
11.1 . : : / STRUCTURES
[p~] 1
-+:
M: lnH
C: 0.98nH
:Jp 2 c.~,, µ
0
a
- - 2 5 TL
'ITR
STL /
--DISK
10 20 30 60 70
DISK WITH
[p=] t
44 5
41 R=SOmm
9
22.5
21.3
R=75
10 20 30 l[mm] 60 70
Fig. 3.28: The mutual inductance of simple grounding

structures (a) a disk, and (b) a disk with cylinder.
Figure (a) also shows results for two simple wire
structures (wire structure 1 and 2) where Biot-savart
calculations were possible.
72
three-dimensional. The bending of the H-lines around the
disk implies a current density concentration at the edges
of both left- and right aide. This effect and the related
30-expansion of the field lines on the right are not
treated in our calculation. We expect that both wil reduce
the flux through the testloop ( see Fig. 3. 3 o) • Clearly,
Eq.3.23 is an approximation only; however it gives us a
useful estimate of the dependence on the parameters.
H
y
I
z
"
Fig. 3.30: Sketch of H-lines around a disk. The bending of

the H-lines around the disk and the related 30-expansion
on the right reduce the flux through the test loop.
No analytic calculations were possible for the disk with
cylinder. Figure 3.31 shows a sketch of the pattern of the
current density distribution for this case. All
measurements for disks with cylinder are plotted in
Fig.3.32. Curve fitting of these data [Sta 87], shows an
approximate (R+l)- 2 -dependence of the mutual inductance
(see Fig.3.32). The effect of the cylinder can apparently
be described by a virtual increase of the radius of the
disk with 1, the length of the cylinder.
2a
Fig. 3.31: Sketch of the current density distribution for

a disk with an attached cylinder.
73
A'
H-lines J-lines
(a)
~ H.dl=I
H-lines
J-lines
(b)
Fig 3. 33: Sketch of the surf ace current lines and the
H-field lines for a strip with (a) a vertical and (b) a
horizontal connection of the injection wires. The full
lines give J or H at the injection side (front); The dotted
lines are at the rear.
74
For a disk with a radius R much larger than the distance 2a
between the leads the disk itself gives already a large
reduction. In this case the benefit of an attached cylinder
is small. This in contrast with smaller disk (ID=:a) ; now
attaching a cylinder is quite beneficial.
,....--~~~~~~~~~~
M
M = Cl(R+l)-S
[pH]
B 1.97±0.096
0 R = 25
6. R = 50
D R = 75
40
\6Ä
Fig.3.32:
2
Measured mutual inductan-
ces M of all disks with a
cylinder fitted to a
12 single curve. The data
points are taken from
Fig.3.28 with the same
45 65 85 105 125 145 numbering.
(R+ 1) l.inmJ
Table 30 gives the experimental results of the mutual
inductance for the strips. The table also includes data for
a reference disk, to illustrate the difference between the
disk and a strip with a strip of the same height, for
different orientations of the current injection wires.
A horizontal connection of the wires gives nearly no
reduction of the mutual inductance, as compared to the
mutual inductance of the reference disk. Connections made
vertically for the same strip give a reduction by a factor
two.
No calculations have been done for these strips. Figure
3.33 gives a sketch of the.current density distribution and
H-lines for bath cases. The sketches of the current
distributions suggest two possible explanations why the
75
magnitude of the H-field, at the rear side, is lower for
case (a) than for case (b):
1" . a _..J..arger part ..of_ ...the total curre.nt flows at the
injection side;
2. the current at the rear side flows via a wider
cross-section A-A'.
Therefore the magnitude of the current density -and of the
corresponding H-field- at the rear side of the wide strip
is lower.
Mutual inductance strip (drawn) disk (dashed)

measurements M =J V dt/I lnj M
lnd
~
4on
1 (n) lil 131.15 pH

=
-- 1~--·H-m"OHOOo.HOoo-•---·--- ------·--··~··.""" ..,"_.___" _____ R 25 mm
M = 145 pH
1
tf,
\•.,' 1
63.6 pH
______ __
1
..
"
:
? ...
' ..." ________

.._,., "
\
I
1
1 ~r... ...
,,_,,_,,." "
,"_,"",
31. 7
___
pH
..." ...""."""".""....." ... ""."~" ..." .....
R = 50 mm
/ ... M = 44 pH
I • \
1 18.4 pH
1
1
'- • I
1
"
Table 30: Column 2 gives the mutual inductance M of strips

with different orientation of the current injection wires
as shown in column 1. Column 3 gives the M of two
comparable disks. Note that the vertically oriented wires
give less coupling.
The technigues to reduce the transfer impedance of simple

GS's can be applied to many practical situations on
different scales e.g.:
- Toa metal connector panel of an apparatus (Fig.3.32a).
For protection against external CM-currents all leads which
enter this panel should be correctly treated as discussed
in Ch. 2, Sec. 2.6.2. ~xtra protection is given by a rim
all around the panel, vertically oriented wires and more
metal where the current density is high.
- To a centra! cable entry plate in a concrete floor, e.g.
in a control room where many cables converge (see
76
Fig. 3. 32b) • The outer cable shields which are often used
for mechanica! protection, should be connected all around
to the entry plate. Further control of external CM-currents
is achieved by an attached cylinder, by goed ground
connections, and by the welding of the reinforcing steel in
the concrete f loor together and to the centra! cable entry
plate.
Il
concrete
floer
Fig. 3.32: Some GS's with a low transfer impedance: (a) a

connector panel with an extra rim; (b) a central cable
entry plate with cylinder for e.g. control room.
In situations with large external CM-currents and sensitive

(measuring) instruments a simple GS can be extended to an
EMC-cabinet as discussed in Ch.2, Sec 2.6.2. In this
chapter, an EMC-cabinet was used in the two setups for
measuring the transfer impedance (see Fig.3.12 and
Fig.3.27). These measurements of very small quantities
(picoHenries as shown in Fig. 3.28), only make sense when
we measure interference-free. To obtain measurements as
shown in Fig. 3. 13, the EMC-cabinet was highly essential.
Similar EMC-cabinets, in. combinatiori with a D/I-measuring
systems (Ch. 4 and 5), were designed and used to measure
transient voltages in HV-substations [Hee 87, Hee 89] and
in H.V.-lab experiments [Zha 89].
The front of the EMC-cabinet can often stay open, if the
distance between interference source and EMC-cabinet is
sufficiently large, especially if the EMC-cabinet is
located in a next door control room. A correct handling of
77
the interference currents at the points where the leads
come in, is more important than the small remaining
shiel~in9 currents near the open front. In these exampl.es
the useful EMC and signal transport properties of a
D/I-measuring system -the possibility of signal transport
over a long distance by a correctly terminated cable- are
important as will be discussed in Ch.4.
In cases where the EMC-cabinet bas to be located close to
an intense HF-interference source (see Ch.5, Fig.5.17) the
door of the EMC-cabinet must be closed to give the needed
shielding against HF-interference. Examples of interference
free recorded voltage waveforms are given in Ch.5, Fig.5.20
and Fig.5.21.
78
CHAPTER 4
ELECTROMAGNETIC COMPATIBILITY ASPECTS OF

A DIFFERENTIATING/INTEGRATING MEASURING SYSTEM
As mentioned in section 2.6.2, an EMC attractive solution

for wideband signal transport is the use of differentiated
signals. The differentiated signal should. be integrated
again at the input of the electronic equipment.
This chapter deals with the design and testing of a
high-voltage divider -based on the principl_e of consecutive
differentiation and integration- for measuring fast rising
high-voltage transients. An important advantage of such a
system is that a long measuring cable between the
high-voltage and the measuring area can be included in the
system without matching difficulties [Wol 83]. Due
attention will be paid to the EMC-aspects of this measuring
system.
The designed D/I-system has among ethers been used for
transient voltage measurements in a 150/lOkV GIS-
substation, as reported in chapter 5.
4.1 HIGH-VOLTAGE DIVIDERS

Most HV-dividers consist of two or more similar impedance
elements to achieve -at least to a first approximation- a
frequency independent dividing ratio. The divider can be
resistive, capacitive or mixed resistive/capacitive (see
Fig.4.1).
HV-arm
LV-arm
, ~ 0t ~
R mixed C L
Fig. 4.1: A HV-divider; the HV-arm and the LV-arm mostly

consist of two similar impedance elements. Of the three
possible elements a capacitor is to be preferred because of
its HF-qualities and linearity.
79
A HV-divider consists of a high-voltage arm and a
low-voltage arm.
'Phe high voltage arnt-of the divider is qenerally large in
size and therefore tends to have appreciable parasitic
capacitances and inductances. When a number of
HV-components is used in series the parasitic impedances
make it increasingly difficult to obtain a correct
frequency response. These problems are largely avoided if
the HV-arm of the divider consists of one single capacitor
[Wol 81) since this can result in a rather pure capacitive
impedance. such a capacitance can be a compressed gas
capacitor (Sch 72) or can be formed by a sensor electrode
at some distance of a HV-object [Wol 81). Gas insulated
switchgear installations (GIS, see Ch.5) have a favorable
geometry for the installation of capacitive sensors inside
the GIS (Hee 89).
The low-voltage arm of the divider can be a fairly large
capacitance (see Fig.4.2a), which together with the
high-voltage capacitor forms a g~e~g!~!Y!_9!Y!9!r·
The problems, associated with the use of undamped
capacitive dividers can be summarized as follows:
(a) Natural frequencies of the low-voltage arm.
Special care has to be taken to ensure a very low
inductance of the low-voltage arm. Usually special
capacitors (Har 79) or a number of capacitors in
parallel (Sch 71) are employed.
(b) Travelling wave oscillations on the transmission line
between the high- and the low-voltage arm.
This transmission line can be long when a compressed
gas capacitor is used [Sch 72].
(c) The matching of the signal cable which cannot be
properly terminated with its characteristic impedance.
To prevent multiple reflections a series resistor is
often employed. For long cab les, the use of a buffer
amplifier between divider and cable is a solution for
correct signal transport to the measuring equipment;
this solution is not attractive from EMC point of view.
80
(d) The signa! to interference ratio.
To ensure a sufficient signa! to interference ratio we
can use a cable with a solid outer conductor, or
transport signals with a large amplitude. If signals of
high amplitude are transported a high-quality input
attenuator is required at the receiving end (as
discussed in sec.2.6.2).
HV> (a)
HV> H~>
(b)
_t 1 ï
HV..::.aE!l_
--- ~---- - cd...
LV-arm
Fig. 4.2: (a) a capacitive divider and (b) A D/I-measuring

system. The integrator capacitors are shaded.
An alternative design of the low-voltage arm is to use a
low value resistor Rd (in fact a long characteristically
terminated measuring cable) connected to a high-ohmic
resistor R1 in series with a capacitor C1 (see Fig 4.2b).
The output signal of the divider is the voltage across
capacitor C1 • The small HV-capacitor Cd at the input of the
cable and the cable impedance Rd act as a differentiator
(t:d=CdRd) of the high-voltage. The resistor R1 in
combination with C1 act as an integrator (t: 1=R 1 C1 ) . The
total attenuation of the signa! is given by t:d/t: 1 : this can
be adapted to the requirements by a proper
.
choice of cd or
i;
1• This divider is called a g!H!!:!!!!:i!!!:i!!9L!!!:!:!9!:!!:!:!!!9
~!!!!YE!ng_!QL!l_~î!:!:!~·
The advantages of this D/I-system are:
(a) as explained in Ch.2 general purpose digital oscil-
loscopes can be used. The use of an EMC-cabinet is
essential.
(b) the components of the low-voltage arm can be of normal
size as used in electronics; because no exceptionally
81
large or small values are needed in this arm quite pure
impedances are possible. As already mentioned the
dITrer-entTäl:ing R~ can bè a long terminated cable. The
integration capacitor C1 can be a feedthrough capacitor
with excellent HF-properties.
(c) the system presents a negligible lead to the HV-source;
often quite small values of Cd are possible
(d) the system has a wide frequency response, and can also
be adapted to widely different requirements.
(e) the system is highly insensitive to interference;
normal coaxial cables are suitable for signal
transport. If necessary active integrators can be used.
(f) the system is easy to install and does not require any
optical fibers or isolation between "grounds" and is
cheap.
The principle of consecutive differentiation and
integration has been used in the past for fast pulse
measurements in plasma physics experiments [Kel 64). At the
Eindhoven University of Technology (EUT), Wolzak applied
this principle for oc, AC and impulse voltage measurements
in HV-installations [Wol 81, Wol 83). Further investiga-
tions have been done at the EUT by Van Heesch [Hee 87,
Hee 89) and Zhang [Zha 89). In the following two sections
we describe the design of a D/I-system with a step response
of less than lns and discuss the EMC-aspects of this
measuring system.
4.2 D/I-SYSTEM FOR FAST RISING VOLTAGE TRANSIENTS

The described measuring system in combination with a
Tektronix digitizer 7912AD (50Q plugin-unit 7A29) has an
overall risetime of 0.6ns (see Fig.4.13), which corresponds
to a bandwidth of 580 MHz.
In this section we deal with the design necessary to reduce
the risetime and the technica! realisation of the measuring
system. Most step response measurements presented have been
carried out by means of a mercury-wetted reed-relay pulse
generator.
82
PRINCIPLE
A idealized circuit diagram of the measuring system is
given in Fig.4.3. The high voltage capacitor is presented
by Cd; in addition a capacitance Cps of the sensor to ground
is shown. If R 1 >>Rd the following equations can be derived:
RdCd ~V
dt H ( 4 .1)
v0 (4.2)
Equations (4.1) and (4.2) are valid for the frequency

range:
1 1
R C <<w <<R (C +C ) ( 4. 3)
1 1 d d ps
1
Rici Rd(Cd+Cps)
Fig. 4.3: Diagram of a D/I HV-measuring system.
HIGH FREQUENCY PROBLEMS

To measure fast rising voltage transients (FRVT) with rise
times of about 1 ns correctly, imperfections of the
measuring system have to be carefully checked. In Fig.4.4 a
circuit diagram of the measuring system is given with
relevant imperfections represented by lumped parasitic
impedances.
1
1
...!..c
T ps
1
Fig. 4. 4: Diagram of the D/I-measuring system including

parasitic impedances important for fast measurements.
83
At high frequencies the response is primarily affected by:
-the capacitance Cp!l across integrator resistor R1
-the inductance Lpil in series with capacitor Ci
-the inductance Lpd in series with resistor Rd
-the mutual inductance M between the differentiating and
integrating circuit.
At very high frequencies also the following impedances
contribute to the distortion of the signal:
-the inductance L of the sensor and i ts capaci tance c
~ ~
to ground
-the coaxial cable
-the capacitance Cpi<! between R 1 and "ground"
-the connection of the measuring system to the oscilloscope
represented by L po and cpo .
The parasitic impedances cps' L ps and cp1a represented by
lumped impedances are in fact not localized but
distributed.
All above contributions can be mi~imized or compensated by
a correct choice of components and a proper layout of the
measuring system. Apart from the measuring system and
oscilloscope a fundamental bandwidth limitation is formed
by the geometry of the input circuit. A more detailed
discussion of the various effects leading to distortion is
given below.
INPUT CIRCUIT
Especially in high voltage measurements, where voltmeter
leads are usually long, the question: "which voltage are we
measuring" shows up as a common problem which always has to
be faced. To solve this problem three questions should be
answered:
1. which voltage do we want to know?
2. which voltage can we measure?
3. what is the input circuit?
As an example, a typical HV-measuring problem with long
voltmeter leads is shown in Fig.4.5. In this example a
HV-source is situated in a metal wal led HV-laboratory. A
voltage of interest (e.g. interesting for flashovers to the
84
wall) could be the voltage V8 between the HV-electrode and
the metal wall via the shortest distance (V s =21JE.dl). This
voltage can be measured with a capacitive sensor of a
D/I-system. The registered voltage Vreq is proportional
,
to
v s . This voltage measurement is correct because the input
circuit encloses no extra magnetic flux (Fig.4.Sb). But
when we are interested in the voltage VH of the HV-source
(VH=:JE.dl) and we use the same setup of Fig.4.5, we of
course still register the same voltage proportional to Vs •
In this situation the input circuit is very important. To
obtain VH, the time derivative of the distributed flux in
the loop 1-3-4-6-1 should be taken into account.
t---~_:L~J
s i
m
Fig. 4.5: HV-source located in a metal walled HV-lab. The

input circuit of the measuring system could be extremely
large especially in installations with single point
grounding.
Another example is the measurement of transient voltages in
a GIS-installation (Fig.4.6). lnstallation of a sensor in
the outer conductor is easy. The registered voltage is
again proportional to the local v s , which is independent of
the azimutal angle e (seè Fig.4.6b), because there is no
magnetic flux in the z-direction. In case of a radial
breakdown between the central and the outer conductor the
local voltage :JEdr is a function of e.
A correct description of the input circuit is extremely
important in physics experiments as discussed in the papers
[Wet 88) and [Dan 88). In these papers the current paths
are identified and the input circuits are constructed as
compactly as possible. Also in the measurements
85
GIS-enclosure
-
w
Ca) Rd(Cd+Cp) (b)
Fig. 4.7: (a) Sensor built in a GIS. For a correct step

response the Cps and Lps are tuned. (b) The frequency
characteristic.
TESTPOT
REED RELAIS
SOURCE Digi-
tizer ax._._
ELEC-
TRODE
RESISTANCE
BRASS SENSOR HOUSING
500
SOQ
(star with
12 branches) plug-in
unit
(a) (b)
Fig.4.8: (a) Setup for step response measurements with mercury-
wetted reed-relay pulse generator. (b) Sensor as used in Ch.5.
calculated
d/dt
~ (b)
V{t) (a)
1
~ (c) d)
measured
"
Fig. 4.9:
(a) Input pulse generated by a vibrating reed relay
(b) numerically differentiated signal a)
(c) measured signal, Lps and Cps near the sensor correctly
tuned
(d) as c, but with a too large L.
86
discussed in Ch.5 (5.1.2, also Fig.5.18) the input circuit
is extremely important.
Note that the inner conductor of a GIS is not an
equipotential during fast transients (see also discussion
in Ch.2.2).
GIS-ENCLOSURE
Fig.4.6: Measurement of voltage V in a GIS. The registered

voltage is proportional to Vs, and is independent of e.
SENSOR ELECTRODE
Two problems associated with the sensor electrode are:
-the coupling of the HV-circuit to the sensor which can be
described by capacitors, provided the wavelength remains
long compared to all dimensions;
-the transition from sensor to measuring cable.
A possible solution for the sensor to cable transition is a
conical matching transition (see e.g. [Tas 86]).
An alternative approach is (see Fig.4.7):
-continue the coaxial cable -a correct TEM structure- as
close as possible to the sensor electrode. The cable
sheath should be circumferentially connected to the sensor
housing.
-the small sensor electrode should be located in a compact
housing; tuning of the stray capacitance cps and the
(small) inductance Lps makes it possible to obtain a
linear response of the sensor up to high frequencies.
With the step response setup shown in Fig. 4. Sa we have
tuned small sensors. Figure 4.Sb shows the sensor used in a
87
RG-214/u (lm)
t [n§l • t [ls] •
q
~
~ ~
t [ns] t ~SJ
Fig. 4.10: Cable responses of the RG 58/u, the RG 214/u and

the SA 24272.
0
0 10 20 30 40
Fig. 4.11: Risetime (ns) versus cable length of the RG 58/u

and the RG 214/u.
88
150 kV GIS-substation (Ch.5.1.2). In Fig.4.9 measured and
calculated step responses are shown.
COAXIAL CABLE
To a first approximation the cable is a perfect part of the
measuring system. Due to the skin effect, mainly in the
central conductor, also the 50'1 coaxial pable presents a
bandwidth limitation (see for example [Nah 73]). This
effect has been investigated, by step response measurements
for different types and different lengths of cable [Lat
88]. Ah increase of the rise time caused by the cable can
be prevented by the choice of the type of cable, and by a
reduction of the cable length. Figure 4.10 shows the step
response of a number of cables and Fig.4.11 gives a graph
of the 10-90% risetimes vs. cable length for two different
cable types. The SA 24272 has also been used by [Mep 87].
Concerning the transfer impedance of the coaxial cable we
refer to chapter 3, Fig.3.1.
FIFTY-OHM TERMINATION RESISTOR
The 50-0hm termination resistor is designed to withstand
high transients and is located in the same cylindrical
housing as the integrator (Fig.4.12). we used carbon
composite 2W-resistors surrounded by shrink sleeve
insulation to avoid flashover across the outer resistor
surface. The resistors are compactly assembled in a star
configuration (three branches of 150'1 resistors), to obtain
a low self induction (loop 1-2-3-4-1) and a low mutual
inductance between the differentiating circuit and the
integrating pircuit. The termination resistor is tested in
combination with the passiva integrator (see Fig.4.14).
FAST PASSIVE INTEGRATOR

The integrator is designed to recover the very fast
changing signals which come out of the differentiating
circuit. The fast passive integrator, built from normal
sized components, should overcome two problems encountered
with simple RC integrators:
89
Rd parasitic capacitance
Fig 4 .12: Fast passive integrator consisting of discrete

components, together with the termination resistor located
in a cylindrical housing.
L .
pi
-.,?J
I
I
- -c~p;nsated.::_ ~--"'ë~
(C . ) ~ pi
pl.2 >...
~ ideal
Fig. 4.13: Frequency response with possible parasitic

effects and corresponding step response.
1: ideal slope (-1)
2: too much Lp1, a feedthrough helps
3 : too much Cpll
4: compensation by Cp12.
The step response has been used experimentally to optimize
the design. ·
time 1)1~ time 1)1~
Fig. 4.14: Measured step responses.
90
1) the parasitic capacitance across the integrating resis-
tor causes the output to overshoot for a fast rising
input
2) the integrating capacitor may resonate because of i ts
inductance.
A simple solution for the integrating capacitor is a feed-
through capacitor. This capacitor has almost no additional
inductance and bas therefore good HF-qualities.
The parasitic capacitance across the integrating resistor
may be minimized by the connection of several resistors in
series. Experimentally we have found .
that
.
a resistor
.
R. l
consisting of ten resistors in series, starting at Rd from
large to small (namely: lx3300, 4xlOOO, 4x680, and lxlOO,
with short interconnections) gives good results.
Computations with a circuit analysis program also shows an
improvement in bandwidth when R1 is built up from large to
small resistors in combination with the parasitic
capacitances [Lat 88].
The overshoot effect can be compensated with the stray
capacitance to the cylindrical housing (copper tube).
Experimentally we have found a optima! size of the tube for
the desired step response of the integrator. All loops in
the integrating circuit have very high resonant frequencies
and are damped. In Fig.4.13 .the frequency characteristic is
shown as expected for various parasitic effects.
Corresponding step responses are also showri in Fig.4.13.
Step response measurements have been carried out wi th a
mercury-wetted reed-relay pulse generator and with a
Tektronix 7912 AD digitizer. The 7A29 pre amp (a 500 plug
in unit) is used to exploit the rise time capabilities (600
MHz analog bandwidth) of this pre amp. Termination of the
integrator by the 500 input impedance results in a
considerable reduction of the measuring time (L=50ns). For
our purpose this is not important since we are interested
in short rise time effects. Some of these step response
measurements are plotted in Fig.4.14.
91
Notes:
-As explained in Ch.2 an EMC cabinet (Fig.2.19) is an
essent.1a~ part of thiS measuring system. The integrator
should be wel 1 bonded to the rear panel • The connection

between integrator and digitizer could be a coaxial cable
terminated at the 500 plug-in unit. When a lMO plug-in unit
is used this connection must be as short as possible to
avoid reflections. In this case the integrator may be
installed in the EMC-cabinet as depicted in Fig.2.20a.
Another possibility is to terminate the integrator with a
buffer amplifier. The connection between buffer amplifier
and digitizer can again be a coaxial cable. The buffer
amplifier must be of very high HF-quality.
-An earlier publications on fast passive integrators is
[Kel 64]. A more recent publication [Ral 84] gives step
responses of an integrator less linear than Fig.4.14.
STEP RESPONSE OF THE HEASURING SYSTEH
The overall step response of the measuring system together
with the Tektronix (7A29 pre amp) is plotted in Fig.4.15.
The overall rise time obtained is 0.6 ns, which corresponds
to a bandwidth of 580 MHz.
~T
·} I 2.Sns -j
0 6
Fig.4.15: Step response of

time fti~ the D/I-measuring system.
After the measurement wi th the mercury-wetted reed-re lay

pulse generator, fast rising voltage transients were
measured with the EHO-GIS-installation as pulse source. The
advantage of this setup is that the measuring system is
being tested at more realistic voltages and interference
levels. A measuring setup of high EMC-quality -as applied
in Ch.5.1.2 (Fig.5.16)- is necessary for this measurement.
92
Figure 4. 16 shows the GIS installation and in Fig. 4. 17 a
measured voltage front · is plotted. This steep pul se is
generated by the spontaneous breakdown of the over-volted
spark gap 2 (see Fig.17) with a 4 bar SF6 gas filing. The
very large rate of voltage , 185 kV in 3 ns, was easily
HIGH-VOLTAGE CONNECTION
bushing
central conductor
outer conductor
spacer
HV-charging section
SHAHPENit;G
GAP
Traveling wave section

10900
Fig.4.16: EHO-GIS-installation
~r
time [ris) - - -
Fig. 4.17: Response measurement EHO-GIS-setup
93
measured with the D/I-system. Similare measurements have
b~en_!i_2.:t:i_~
by BQggs [Bog 84], who used a special capacitive
divider, and registered a rate of rise of 80 kV in 5 ns.
Since we are comparing fast voltage dividing systems, the
highest dV/dt which can be measured, could be considered an
estimate of the quality of the system. Extra advantages of
the D/I-system are: the absence of matching problems, the
EMC-qualities, and the easy adaptation of the total
attenuation, in case very large source voltages are to be
measured.
The sensitivity of the measuring system depends on the
precise capaci tance between the sensor electrode and the
central conductor. For this reason, calibration should be
carried out on the installed sensor. In a substation the
50Hz high voltage can be used for this calibration, which
is correct if the sensitivity is known to be constant for
the frequency range of interest.
4.3 ADDITIONAL EMC-ASPECTS OF THE D/I-MEASURING SYSTEM

A D/I-system for measuring fast rising voltage transients,
as described in Sec.4.2, is suitable for DM-signal
transport without matching problems for the connecting
signal cable. An EMC-cabinet creates a protected region for
the registration equipment and the integrated signal, and
V DM, int
(a)
Fig.4.18: (a) The DM-interference represented by a series-

voltage. (b) Equivalent diagram of the D/I-system
terminated at the integrator side and (c) at both ends of
the measuring cable.
94
gives excellent protection against the CM-currents which
flow towards the EMC-cahinet. Since the cables for signal
transport have sizeable transfer impedances, DM-interferen-
ce may already leak into these cables before they enter the
EMC-cabinet. For cables shorter than a quarter wave-length
of the interference signal, the DM-interference can be
represented by one lumped series voltage source vDK,lnt'
where VDM, Int equals the product of the transfer impedance
of the cable and the CM-current (see Fig.4.18a).
Dependent on the termination of the DM-circuit, an interfe-
rence voltage V 1nt then appears at the input of the
integrator. For a D/I-system, the measuring cable can be
terminated only at the integrator side of the cable or at
both sides of the cable without bad effects .for the "legal"
DM-signal. The advantage of termination at both sides of
the cable is that in situations with imperfect termination
resistors multiple reflections are reduced and the band-
width is enlarged.
From an EMC-viewpoint, however, termination at both sides
is disadvantageous. The signal to interference ratio at the
integrator input with termination of the cable at one side
and at both sides are, respectively (see Fig.lSb,c):
Cd • VHV
( vm1v 1nt.)"1 nqle = C +C ~v-- ( 4 • 4)
" p s DM, Int
s 1 de
and (Vn/V Int) both = l+jwR d (Cd +Cps ) v DM, Int c4. 5)
si des
For the frequency range w<<l/Rd (Cd +Cps ) -see also Eq.4.3-
Eq.4.5 becomes: v
HV
(Vn/V1nt)both == jwRdCd.~V- ( 4. 6)
DM,lnt
sldes
Comparison of Eq.4.4 with Eq.4.6 shows that the

signal/interference-ratio for cables terminated at one side
is 1/{wRd (Cd +Cps ) } times better than for a cable terminated
at both ends. Since in the frequency range of interest
wRd (Cd +Cps ) is small compared to one, we prefer
single-sided termination with a good quality termination
95
resistor, as discussed in sec.4.2. Recent calculations by
Van Deursen [Deu 90] give a correction to Eq. 4. 4: in the
denominator a-term- e-c /2 -bas to be a~ to cd +cps , Where c c
is the capacitance of the length of cable used.
In situations with very high frequency signals or
interference the Eqs.4.4-4.6 should be modified because of
traveling waves on the cable and the effect of the
capacitors cd and Cps . Appropriate EMC-measures are then:
In case of very high frequency interference the passive
integrator can be of the compensated type (Fig. 4 .13) to
prevent the interference to reach the active electronics.
In case both the signal and the interference contain very
high frequencies the transfer impedance -and if possible
the length- of the connecting cable should be reduced. High
quality cables or metal conduits or cables can effectively
reduce the transfer impedance (see e.g. Fig.3.21 and
Fig.5.16).
96
CHAPTER 5
ELECTROMAGNETIC COMPATIBILITY IN GIS SUBSTATIONS
5.0 INTRODUCTION
Introduction of digital electronics in HV-substations gives
access to all advantages of fast data acquisition for
measurements and on-line control of substations. However,
due to intense EM-interference, caused by switching at the
high voltages, most of the normal digital equipment will
not function properly in HV-substations, or will even be
destroyed.
The HV-grid connects a (large) number of power stations and
consumer areàs through a meshed network. At the nodes of
this network the HV-substations are situated. A
HV-substation controls power flow and distributes the
electric energy to the consumers. Incoming and outgoing
circuits are connected to the HV-rail system via
circuit-breakers, disconnect switches and transformers (the
primary system).
There are two types of HV-substations:
-an open substation: The HV-rail system and the switch gear
are located in the open air, installed on vertical
insulating supports:
-a closed substation or GIS-substation (Gas Insulated
Switch gear): The HV-rail system and switch gear, supported
by insulating spacers, are housed in closed cylindrical
metal vessels (see Fig.5.1). The vessels are grounded and
filled with pressurized gas (e.g. SF 6 -gas). Single phase as
well as three-phase enclosed systems are possible.
Advantages of a GIS-substation compared to an open one are:
it is compact and saves space; it protects the primary
system against industrial pollution, rain, snow, and salt
deposition and it is easily accessible for maintenance work.
From an EMC-viewpoint, however, a GIS-installation is a
concentrated source of HF-interference for the secondary
system (the system for measurement and control). The
97
secondary system is in many places quite close to the
prim~ry system: e.g. in current and voltage transformers;
disconnect switch drives: sensors for pressure and
temperature and position indicators. In this chapter we
study the resulting problems, where source and victim are
the primary and the secondary system.
Examples of measuring systems protected by the use of the
concepts of Ch.2, are presented in [Hee 87] and [Hee 89].
These publications describe new wide-band current and
voltage measuring systems tested in 150 kV and 400 kV
GIS-substations. The measuring systems are based on the
differentiating/integrating measuring principle (see Ch.4).
Analog integrators and digital apparatus are used for
signal recovery and recording, respectively.
A similar system has been used for measurements on fast
rising high-voltage transients in the 150 kV GIS-substation
Eindhoven-West. The system as well as the measurements are
described in Sec.5.1.2 of this chapter. During many tests
in substations as well as in the TUE H.V. lab, the systems
have operated correctly and have shown their excellent EMC
qualities.
These examples and the applications in Sec.2.8 show, that
protection of electronic systems with our basic solution
(see Fig. 2. 22a) is straightforward and generally usable
whatever the interference levels are. In contrast to the
"open" interference sources in Sec.2.8, -a lightning
discharge and switching actions in a primary circuit of a
HV-installation-, in a GIS-substation modifications are
possible to contain the HF-interference.
The present chapter considers the GIS installation as a

source of HF-interference. Measurements of steep transient
voltages across interruptions in a GIS installation due to
switching actions are presented and modifications for
reduction of HF-interference are discussed.
98
5.1 GIS INSTALLATION AS A HF-INTERFERENCE SOURCE
A lot of locally stored capacitive energy (~Cv2) is present
in GIS installations. During switching actions with
circuit-breakers, disconnect switches and grounding
switches this energy is partially converted into high power
electromagnetic waves with fast rise times of less than
5 ns. These high power waves (v2/Z 0 ) propagate almost
loss-free in the coaxial arrangement which may have a
characteristic impedance Z of about 75 O.
. 0
No EM-waves escape from the GIS-installation when the outer
conductors of the GIS enclosure and the HV-cable form one
completely closed conductor (Fig.5.1). Of course it is
assumed here · that there are only cables connected to the
GIS: a connection to an overhead line (via a high-voltage
bushing) gives a quite open structure.
Fig. 5.1: GIS with connected cables.
Inside of a coaxial system as in Fig.5.1 the em-waves are

reflected many times. The dissipation mainly takes place in
the HV-cable (in Eindhoven West this cable is a
cross-linked polyethene cable with a Z0 of about 30 0). The
desirable reduction of the HF-waves in these HV-cables is
mainly caused by semi-conducting layers on both sides of
the insulation of the central conductor [Wol 83] • These
semi-conducting layers are meant to smooth out local field
99
enhancements caused by sharp points on the metal
conductors, but also serve an important EMC-purpose in
suppressing high-frequency transients. If, however, there
are interruptions in the GIS enclosure the HF-waves can
escape freely to the outside of the installation. The
resul ting voltages over interruptions and between other
grounded metal construction parts may lead to breakdown,
sometimes within very short times, less than l ns (see
Fig.5.20). The corresponding, very large, undesirable
14
dV/dt's and dI/dt's (l0 12-l0 V/s and 1010 -10 12 A/s, resp.)
can be responsible for interference in the secondary
system. Moreover, the low energy HF-sparking could cause
degradation of the insulation and is psychologically
unacceptable for personnel in a substation.
Interruptions are present where:
overhead transmission lines are connected to GIS
terminations (bushings);
HV-cables are connected to GIS terminations; we
distinguish:
a) the cable sheath is isolated from the GIS enclosure by
means of insulating f langes to allow cathodic protection
~f the cable sheath with a small DC voltage [FUJ 88].
b) the cable outer conductor is interrupted to permit

current measurements; this situation is discussed in
detail in Sec.5.1.2;
- Sections of GIS-enclosure are insulated from each other
(Sec.5.1.1).
In the following we discuss proposals to improve the EMC
behavior of the GIS installation, which still satisfy the
other design criteria. We will distinguish between
necessary and unnecessary interruptions.
5.1.1 UNNECESSARY INTERRUPTIONS IN A GIS-INSTALLATION

In some single phase enclosed GIS-installations, sections
of the enclosure are electrically insulated with respect to
each other and are grounded per section at one single
point which is connected to the grounding mesh of the
100
substation (see Fig.5.2). In these installations the 50Hz
HV-circuit consists of the three high-voltage centra!
conductors; the grounding mesh carries currents during
fault conditions only. It is (naively) assumed that no
50Hz-current will flow in the metal enclosure, which would
mean that the enclosure does not contribute to the total
dissipation of the GIS. However, even for 50Hz, induced
eddy currents flow at the inside surface and, in the
opposite direction, at the outside surface of the enclosure
(see Fig.5.2).
I
Fig. 5.2: Enclosure of a GIS grounded at a single point per

section. Induced 50 Hz currents flow in opposite directions
at the inside and outside surfaces of the enclosure.
From an EMC point of view it is advisable to make full
metallic contact all around between adjacent sections of
the enclosure. In this manner, we avoid the escape of
HF-waves via the transitions. The dissipation in the
enclosure per single phase, because of induced
50Hz-currents, depends on the interconnection of the three
single phase enclosures at the ends of the GIS (see
Fig.5.3). When these connections are made with wide metal
Fig. 5.3: Simplified sketch of three single phase

enclosures interconnected by wide metal plates at the ends.
101
plates (at both ends) the magnetic fields between each pair
of the enclosures are quite small, and correspondingly
the SOHz.-currents .in the eru::losuras. t:.end to be equal and
opposite to the currents in the central conductors.
The above described situations illustrate two limiting
cases of current flow in the enclosures, namely:
-eddy currents flow in opposite direction at both sides of
the enclosure wall ( insulated enclosures grounded at one
point):
-currents flow in one direction in the enclosure
(continuous enclosures with the ends interconnected by wide
metal plates).
To answer the question under what circumstances the
dissipation is minimal, we consider both configurations.
Note that when less metal is used f or the interconnection
at the ends a mixture of both cases results.
INSULATED ENCLOSURES
First we determine the dissipated power in the enclosure
for the case where adjacent sections are insulated from
each other. Per section the enclosure is grounded at one
point connected to the grounding mesh. Let I be the current
in the central conductor (see Fig.5.4). The dissipated
2
power P (W/m ) in the enclosure is equal to the sum of the
r y
Fig. 5.4: Interrupted GIS enclosure connected to the

grounding mesh at a single point (a). For R>>d the E-field
in the enclosure wall is equal to the E-field in a flat
large plate (b).
102
time average of the Poynting vectors directed into the
enclosure wall. The Poynting vector s is equal to the
vector product of the electric field E and the magnetic
field H. We therefore have to know E and H at both sides of
the enclosure wall.
To calculate the H-field we neglect the proximity effect
between the enclosure and the other two enclosures in the
substation. The H-field lines, at the inside as well as
close to the outside of the vessel, are considered to be
concentric circles around the central conductor. The
H-field magnitude along these circles is given by H0 = I/2nr
(Fig.5.4).
When the radius R of the enclosure is much larger than the
thickness d of the enclosure wall the E-field in the wall
can be approximated by the expression for the E-field in an
infinitely large plate with at both sides of the plate an
H-field in the y-direction with magnitude H y = I/2nR. The
expression for the internal E-field is given by
(Kad 59, pg.19]:
sinh kx d d
Ez= kp cash kd/2 HY, --2- <x< -2- (5 .1)
fi µ J
112
l+j 2
where k 6 6= ( w is the skin depth, p is the
0
resistivity and µ 0 µ I" is the permeability. Assumed is an
exp(jwt) time-dependence.
The E-field in the wall is an odd function of x, which
means that the Poynting vector at the left hand side as
well as at the right hand side of the enclosure wall can be
directed into the material. Therefore, the power
dissipation per square meter enclosure is equal to:
d
-l *
E H +E H
4 [ z y z y *] d
=
x=- x=--
2 2
sinh 6d d
e. - sin 6
pH2 IRe{k tanh k~} 6 Hy
"2
[W/m
2
] , (5.2)
cosh äd + cos ä
y d
103
CONTINUOUS ENCLOSURES
Secondly we determine the dissipated power in the enclosure

for the case where neighboring enclosures are connected to
each other at the ends by means of wide metal plates. The
H-field lines inside the vessel are concentric circles
around the central conductor with H6 =I/2nr, rc:sr:sR. The
H-field. outside the vessel is zero (see Fig.5.4a).
/
H /
y
Fig. 5.5: The continuous enclosure used as current return

path. The H-field at the outside of the vessel is zero (a).
For R»d the E-field in the plate can be considered to be
equal to the E-field in an flat large plate with an H-field
at one side only (b).
Whe~ R»d, the E-field in the wall can be approximated by
the expression for the E-field in an infinitely large plate
with thickness d, with at the left hand side of the plate
an H-field in the y-direction with magnitude Hy =I/2nR and
no field at the right hand side (see Fig.5.4b).
The expression for Ez is given by [Ka 59, pg.139]:
E = -kp cosh k(x-d) H O<x<d (5. 3)

z slnh kd y '
The Poynting vector at the left hand side of the plate is

directed into the plate, therefore the dissipation per
square meter is
p = -i [E:Hy+EZH;] = ~ pH~ ~e{k coth kd} =
x 0
sinh 2~ + sin 2~
.! e.0 tt2 0 0
(5.4)
2 Y cosh 2~ - cos 2~
0 0
104
COMPARISON OF DISSIPATION IN ENCLOSURES
In this section we compare P1 , the dissipation in an
enclosure due to a current which is equal and opposite to
the central conductor (Fig.5.5), with P2 , the dissipation
in an enclosure with only eddy currents (Fig.5.4). The
dissipations P1 and P2 are given by Eq.5.4 and Eq.5.2,
respectively. We are especially interested in a comparison
of P1 with P2 at the mains frequency as a function of the
enclosure thickness d for a given material (i.e.: p and 5
are constant) • Therefore, Fig. 5. 6 shows the graphs of P1
and P2 [W/m 2 ] as a function of d/5.
4
p, 6 is kept constant
pl :Y-4?Z?ZJ
P2 :Y;:z:t?Z\
A B
1 2 3 4 5 6 7 8
a/o
Fig.5.6: The dissipation per square meter enclosure for an
enclosure with at one side an H-field (a) and at both sides
an H-field (b) as a function of the wall thickness.
We may characterize two regions A and B, on either side of
the point d=l. sa where P1 = P2 •
-In Region A, O<d<l.55, P1 is larger than P2 •
For d~ O, P1 is very large and P2 is almost zero.
-In Region B, d>l.55, P1 is smaller than P2 •
For d>45, P2 is almost equal to 2xP1
We apply this comparison to a GIS enclosure made of iron or
aluminum; to the lead sheath of a single phase HV-cable and
to an Isolated-Phase Bussystem [Sen 83).
105
GIS enclosures
Some GIS enclosures are made of iron. The thickness of the
enclosure wan. (cnosen also to safely contain the
pressurized insulating gas) varies between 6 and Smm. The
skin depth ö Fe at 50Hz is on the order of lmm. In Fig.5.6,
we are clearly in Region B (d>3xö Fe ) which means that P1 is
smaller than P2 • In this case, both for 50Hz-dissipation as
well as for HF-EMC arguments, a continuous enclosure with
at one side an H-field is advisable. Of course less
dissipation occurs in an aluminum enclosure.
Most GIS-installations have aluminum enclosures. The
thickness of the enclosure wall varies between 8mm and
15mm. The skin depth for aluminum ö Al at SOHz is about
equal to 12.Jmm. Now we are in Region A {left of d<l.Sö).
In this case the dissipation is less for an non-continuous
enclosure. However, a continuous GIS-enclosure may be still
advisable, because:
-the dissipation in the enclosure is only a fraction of the
total dissipation in the GIS-installation;
-P1 could still be considerably less than the sum of P2 and
dis::;ipation by eddy currents in nearby structural steel
caused by the H-f ield outside the interrupted enclosure.
Note: When the ends of the GIS are interconnected by narrow
strips and not by wide metal plates, an H-field is present
outside the continuous enclosure. In this si tuation the
dissipated power in the enclosure is between P and P2 over
1
the entire region (A+B, see Fig.5.6).
HV-cables vith a lead sheath
For HV-cables it is important to keep the power dissipation
in the lead sheath low to limit the temperature rise in the
dielectric. The skin depth of lead öPb for 50Hz is about
36.Jmm. Therefore the variable d/öPb is in Region A, close
to zero. To avo id large net currents in the HV-cable
sheaths of a three phase system, "cross-bonding" techniques
(see Fig.5. 7) are applied. By means of this technique,
normally used for long lengths of cable, the induced
currents in the HV-cable sheaths are zero because the
106
Fig.5.7: Cross-bonding of lead sheaths of HV-cables.
overall induced flux in every closed loop formed by the
sheaths is zero. The left and right far ends of the cables
are interconnected and grounded to carry away capacitive
currents during abnormal conditions.
Note:
(1) The interconnections at the end of the cables in
Fig.5.7. can coincide with the interconnections at the ends
of the GIS in Fig.5.3.
(2) When the HV-cables ends in a GIS-installation and .the
cable sheaths are connected all around to the GIS-enclosure
we are in the EMC-ideal situation which is described in
sec.5.1.
Isolated-phase bussystems {Sen 83]
Isolated-phase bussystems are used in power stations to
connect the generator to the main transformer (Fig.5.8).
Each of the three-phase conductors is completely shielded
by an aluminum enclosure moisture and
Fig. 5.8: A sketch of the basic layout of an isolated-phase

bussystem.
107
accidental contact. The wall thickness is about equal to
7mm. In this case we are in Fig.5~6 to the left of point
d=l.Só ~O.A 1 (5efi%)""'12.31mn). Although P is larger than P ,
1 2
nevertheless the enclosure ends are interconnected with
wide metal plates. These plates allow net enc.losure
currents to flow, which largely cancel the magnetic fields
outside the enclosure. Consequently, the electromagnetic
forces between the enclosures and between the conductors
remain low even for dangerously large currents for instance
during faults. An additional advantage is that eddy current
losses in structural steel in the vicinity remain very
small.
5.1.2 NECESSARY INTERRUPTIONS INA GIS-INSTALLATION

In our attempt to contain HF-waves inside the GIS-enclosure
it is obvious that interruptions are to be avoided as much
as possible. Unavoidable interruptions, however, remain at
the transition from a HV-line to a GIS and in some
situations at the transition HV-cable to GIS. This section
deals with the control of HF-waves escaping via these
necessary interruptions. We briefly discuss a HV-line/GIS
transition and a HV-cable/GIS transition if cathodic
protection of the cable sheath is used. We discuss in
detail a HV-line/GIS transition where HV-cable sheaths are
interrupted to permit current measurements.
GIS/HV-LINE TRANSITION
A connection of an overhead line to a GIS-installation via
a high-voltage bushing gives a quite open structure (see
Fig.5.9). During switching actions HF-waves escape via this
transition. The escaped hf-waves propagate partly via the
HV-line circuit to the outside world and partly via the
transmission line formed by GIS-enclosure and ground in the
direction of the GIS-building. Reflected waves returning
along the HV-line partly reflect again at the transition, a
small part turns back into the GIS-installation and a
larger part travels between GIS-enclosure and ground to the
GIS-building.
108
METAL PLATE
GIS-INSTALLATION
----- ---------- -- ------- -------- ----.. .
Fig.5.9: GIS/HV-line termination with protective measures.

To avoid large currents inside the GIS building, we create
a protected region with a metal plate (analoguous to a
simple grounding structure, see Sec.3.3). The GIS-enclosure
should be connected all around to the grounding structure
(GS). To achieve a better control of the external
HF-currents this metal plate is horizontally extended in
the direction of the HV-lines (Fig.5.9). Further control is
achieved by the installation of grounding pins (also
adequate for low frequency currents, e.g. in the event of a
short circuit) and by the connection of the GS to
structural steel in the building. Useful HF-losses can be
obtained with strings of carbon composite resistors (~5000)
connected between the GIS-enclosure and the extension of
the GS.
HV-CABLE/GIS TRANSITION; CATHODIC PROTECTION

To allow cathodic protection of the cable sheath with a
small cc-voltage the cable sheath can be isolated from the
GIS enclosure (Fig.5.10). Traveling wave transients, due to
switching actions in the GIS-installation, often cause
sparking across the insulating flanges at the cable/GIS
transition. Fujimoto [Fuj 88] investigated the use of
capacitors, metal-oxide varistors (MOVs) and spark gaps, as
a means of protecting the insulating flange, both for new
installations as well as f or existing installations as
retrofits.
109
GIS
INSULATING
FLANGE
Fig.5.10: Basic cable/GIS transitlon, insulated to allow

cathodic protection [Fuj 88].
The results of this investigation were: a) spark gaps
installed circumferentially around the insulating flange
are not a suitable 11 quick fix" solution for existing
installations; b) it is better to use MOVs or capacitors to
provide a HF-current path, if the components are carefully
installed. The connection leads must be kept as short as
possible, and several components, perhaps one at each bolt
hole, are necessary to reduce the inductance.
In this context we remark that feedthrough capacitors have
excellent HF properties. Carbon composite resistors,
surrounded by shrink sleeve, and in series with the
capacitors give extra HF-losses (see Fig.5.11). Several
series-connected capacitors and resistors should be
installed circumferentially around the insulating flange. A
compact arrangement as shown in Fig. 5. 11 should keep the
inductance low.
DISK
R Set of resistors and
feedthrough capacitors
High-pressur GIS
oil-filled
__ _j
Fig. 5 .11: Feedthrough capacitor in series with a carbon
composite resistor provide a compact HF-current path and
attennuates the transient.
110
GIS/HV-CABLE TRANSITION; CURRENT TRANSFORMER ARRANGEMENT
Similar problems as above occur at a GIS/cable transition

if a current-transformer is used. In this case cable
sheaths are interrupted to permit current measurements.
Induced HF-vol tages over these interruptions may lead to
breakdown. The low energy HF-sparking could cause
degradation of the insulation near the transition. We have
investigated this problem in the 150/lOkV GIS-substation
Eindhoven-West of the PNEM (Power Company Province
Noord-Brabant). Fast rising high-voltage transients across
the interruption are measured and possible solutions to
avoid HF-sparking are tested.
The substation Eindhoven-West is linked with the
substations Eindhoven-Noord and Eindhoven-Zuid via two
three-phase connections, each consisting of three 150kV
XLPE-cables. It supplies electric energy to the consumers
via two 150/lOkV transformers. Figure 5 .12 and Fig. 5 .13
depict the three-phase enclosed GIS-installation and show a
diagram of the substation, respectively. The dashed contour
in Fig.5.13 encloses the GIS part of the installation.
GIS
"111..... PROBLEM
CAB LE
Fig. 5.12: Three-phase GIS-substation Eindhoven-West
111
5 km
CABLES TO CABLES TO
EINDHOVEN L EINDHOVEN L
cp
,--- SOUTH yu
~--------~--1
NORTH
' ( ~: LINE-DISCONNECTORS
CIRCUIT
BREAK ERS
t t RAIL-DISCONNECTORS
IA J
l----.t-=-~-----L---+-~___,l-t+-)-t--l______.-~::
-+_
î ~!:~i~s ~ 1
CB
itp.~c .
f
"""-!~
GIS
1
1
1
short
~--·----------~--------~
cab~les
-
l
1
_ ~
I· 150 kV
short
2
'
_ ~ .
,
cab~es -
".....-.. 10 kV "~
Fig. 5.13: Configuration substation Eindhoven-West.
112
The GIS contains: two rail systems (A and B); five
three-phase circuit-breakers, two for each 150kV connection
(CB-North and CB-South), two for the 150/lOkV transformers
(CB-trafo-1 and CB-trafo-2) and one between rails A and B;
the necessary disconnectors (for rail, cable, and
transformer} and grounding switches.
The GIS-installation has four three-phase terminations, two
for the 150kV cables to substations North and South and two
for the 150kV cables to the transformers. At the ends of
the GIS, around each cable (total twelve), a current
transformer is placed outside the GIS-enclosure as depicted
in Fig.5.14. The lead sheath of the cable is interrupted to
permit current measurement. An insulating collar also
serves as a pressure-seal. With a grounding strip, outside
the current-transformer, the 50Hz-circuit is completed.
GIS----11 r+---~insulation
1+----<11- carbon layer
insulating
col lar
· ':-cylindrical
Grounding '· ' spark gap
strip
(50Hz-return)
current-transformator
---cable
Fig. 5 .14: Simplified sketch of the GIS/cable-transi tion

with current-transformer arrangement. Only one of the three
cables is shown.
To avoid degradation of the insulating collar and the cable
insulation, voltage limiting 0 cylindrical spark gaps" are
installed (see Fig.5.14} .The sparking of these gaps -and
at many ether places- during switching actions indicates
11 3
the escape of HF-waves. In an attempt to reduce hf-sparking
the power company had prepared a brass cylinder which could
be inst.all.ed arounà. ...~ threa-phai;e ourrent-transformer
arrangement (Fig.5.15a) to provide a closed low-inductance
return path for the HF-current. The cylinder has a length
of 115cm and a radius of 50cm. Another modification, a
proposal of our group, is to install a ring of resistors,
with a total resistance of about 300 across each
cylindrical spark gap (Fig.5.15b). In this manner an even
more compact and dissipative circuit for the HF-currents is
formed. The 50Hz-currents still flow via the grounding
strip around the current-transformer. Each resistor ring is
composed of 24 parallel branches each with two resistors in
series. We used carbon composita 2W-resistors with shrink
sleeve insulation and a resistance.
GIS
resistors
brass
cylinder
cylindrical
spark gap
trafo's lead sheath

(3x)
Fig. 5.15: A GIS-transition with (a) a brass cylinder

around the current-transformer arrangement and (b) with a
ring of resistors across each spark gap.
To study the effect of these two modifications the next
measurements have been done:
-we measured the voltage VG across the cylindrical spark
gap as a function of the gap distance (d=2, 5, 10 and 20mm)
-we measured VG when the brass cylinder is installed
around the current-transformer arrangement (d= lOmm), and
-we measured VG when resistor rings are installed across
the cylindrical spark gap (d= lOmm).
These measurements have been done during circuitbreaker as
well as disconnector switching actions.
114
SETUP
Experiments have been done at the transition of termination
south (see Fig.5.16). A Tektronix 7912 AD digitizer (600MHz
analog bandwidth, 50'2 p_lug-in unit) was used to register
the first 50ns of the voltage across the spark gap of phase
8. With two multi-channel Nicolet 4094C digital
oscilloscopes (200MHz max. sampling rate, 8 bit resolution)
we had determined beforehand that the given circuit breaker
often showed its first breakdown at phase 8. The
time-window of the Nicolets during these measurements was
l250ns ( lMQ input) • The applied measuring systems were
based on the D/I-measuring principle. The rise time of the
step response of the measuring system in combination with
the Tektronix was 0.6ns (Sec.4.2).
GIS-EINDHOVEN-WEST
SOUTH TRAFOl
'---r----....---•.---.-_:.
.. -J - - 1 ..:_-:,. -1- -
,..... "' ·y '{ .
I
1
{'j
-
, "\ \
,_, \
I ,....
1 \,)
(' \
_, 1
\ ... I \ ... I
") I \
3xRG214/u
\
' '
~-.,,,,,,.
/
I
' , ___
"....
1
,,,./
each in an extra
Cu braid
2x TEKTRONIX
NICOLET~§§§§§§êê~:.::__~~~~.....J
4094C F
7912AD
digitizer
scopes +
EMC-CABINET EMC-CABINET
Fig. 5.16: Setup for voltage measurements across the

sparkgaps of termination South.
The Tektronix, together with a PC for data-handling, and
the two Nicolets were placed in two EMC-cabinets. These
EMC-cabinets and the GIS-installation were firmly
interconnected by means of a grounding structure. The three
signal cables (type RG-214/u) of the measuring systems for
the Nicolets were each surrounded by an extra copper braid
and tied to the grounding structure. To achieve optima!
wide band signal transport to the Tektronix a short
115
(4 meter, see also Fig.4.11) RG-214 cable was used. The
cable was fed in through an "inside extension" (see Ch.2,
Fig.2.20). Inside the EMC-cabinet the cable was terrninated
by its characteristic impedance. To be absolutely safe the
cable was surrounded by a copper tube which was connected
bath to GIS and to the EMC cabinet. The EMC-cabinets were
located close to the intense HF-interference source ( see
Fig.5.17). Therefore, the EMC-cabinets had to be closed
during measurements to give the required shielding against
HF-interference.
Note that for this setup a lot of "plumbing" was necessary.
Under less cri tical circurnstances, we may omi t the extra
copper around the D/I measuring cable, and we may leave the
door of the EMC-cabinet open.
Fig. 5.17: EMC-cabinet with

Tektronix digitizer and PC
for data handling located
close to the GIS at the
left.
The rneasurements were carried out with the D/I-system

described in Ch.4. The signal registered with the Tektronix
(A~ r ~
is proportional 10 mV/kV, 1 ?Ons) to the voltage Vss
2
between the sensor and the sparkgap cylinder (Vss= J E.dl,
see Fig.5.18).
Concerning the input circuit of the rneasuring systern we

distinguish two situations:
(a) there is no are between the two cylinders of the spark
gap,
(b) there is an are.
1 16
In situation (a) the voltage Vss is equal to the voltage VG
across the gap when the measuring circuit 1-2-3-4 is
compact (see Fig.5.18a), i.e:
-the induced voltage in this loop, due to the current I, is
small compared to V0 ~
-the measuring loop contains 500, the capacitance of the
sensor, and an inductance. To keep the inductance low the
connection 3-4 is shaped as a short wide metal bracket.
In situation (b) the voltage Vss is the sum of the are
voltage Vore (JE. dl through the are) and the voltage V1nd
induced in loop 1-2-3-4-5-6 (see Fig.18b). The voltage V1nd
is far from small compared with Vare . Moreover the voltage
vss depends on the are position in the spark gap.
1 1 1 1
1 1 1 1
1
1 1l
1
1
j
1
l
I :j:
1 1I
1 1
1 1 1 1
1 1 1 1 He
SENSOR
(a) {b)
Fig. 5.18: The input circuit of the measuring system

without an are (a) and with an are between the gap (b).
11 7
EXPERIHENTAL RESULTS AND DISCUSSION
The proposed experiments were mainly carried out during

switchfiig actions-wTth circuit-breaker ~fouth (CB-South) and
rail-disconnector North (DC-North) in substation Eindhoven-
West. Because in real substations the number of switching
actions should be limited and the exact switching time with
regard to the 50Hz voltage is unknown, there is a certain
spread in the results. Preparations in our own laboratory
with a GIS-setup which generated a much more reproducible
transient,-were therefore very useful.
The initial state of the Eindhoven-West installation was:

(a) in advance of the closure of CB-South: rail A was fed
via HV-cable North; CB-North, rail- and cable-DCs North and
South were closed; HV-cable South was open at the far end
(see Fig.5.19a);
(b) in advance of the closure with rail-DC North: rail A
was fed in via HV-cable South; CB-South, rail- and cable-DC
South were closed; CB-North and cable-DC North were open
(see Fig.5.19b).
During both actions a) and b) sparks were audible and
clearly visible across the spark gaps of termination South.
In contrast to a oc-action, which is always accompanied by
a long train of sparks, a CB-action is accompanied by one
r--------- 1 r- ---------.
RAIL A I I RAIL A 1
r
1
1
-, 1 rail-DC
N
r
1
1
-1 1
1 1
1
1 r -,
1 1
1 l Jo
1
:S . N
r , 1
rail-DC ~· 11 1
, 1
-L •S
_, 0 oL--..l
1
L o
1
-'
r. ,
u r 1 1 u r 1
1
t
1h
1
CB ~ r- -: 1 1 f -1 ~
:. ~
CB ~
:1:
r --: 1
~ -1 !
~ _J 1 : 1 1
r- ,
1
~
1
:
cable-DC r
! -; cable-DC ,r-1 i
-· L JI
_J
1LJ
L
" -''
LTI IJmeasur~ng-location
- - L ?eas;;r~ng - - -
I location
-
_J
~ ~
Fig. 5 .19: Simplified diagrams of substation Eindhoven-
West. The diagram at the left hand shows the initial state
for switching actions with CB-South and at the right hand
for switching actions with rail-DC North.
118
audible spark. Only the experimental results obtained
during CB-actions are discussed here, because they provided
clearer and more reproducible signals.
The purpose of the experiments was to study the effect on
the sparking: of a brass cylinder installed around the
current-transformer arrangement, and of resistor rings
installed across the cylindrical spark gaps. Therefore we
have measured the voltage across the spark gap in bath
situations. As a reference we have measured the voltage
across the gap as a function of the gap distance in case
there is neither a brass cylinder nor resistor rings. These
measurements have been done at termination South during
actions with CB-South.
Original situation: cylindrical spark gaps only
The first series of measurements were the voltage
measurements across the spark gap of phase 8 at termination
South with different gap distances during switching actions
with CB-South.
Table SA gives the results of the measurements. The values
of the maximal top voltage and the average top voltage are
obained from waveforms as in Fig.5.20.
Figure 5.20 shows the recordings of voltage waveforms with
four different spark gap distances. The recordings show a
considerable change in waveforms due to the variation of
the gap distances between 2 and lOmm. For distances between
10 and 20mm the waveforms are identical.
For gap distances 2 and 5mm the recordings show a steep
voltage drop, corresponding with breakdown across the spark
gap. Although this breakdown limits the magnitude of the
voltages, the fast voltage drop corresponds with very large
dV/dt's and dI/dt's. These rapid changes can be responsible
for interference in the secondary system. Noticeable is:
-the 0.6ns voltage drop in air because of an overstressed
spark gap (d= 2mm);
-after a breakdown of the spark gap there is still a
considerable voltage (a few kV 1 s) because of the large
119
CB-South
maximum average
spark gap top- top- deviation number of
distance voltage voltage measurements
(mm) ! (kV) (kV)
2 21.9 16.l 30% 4
1
5 33.0 31.4 2
1
* *
10-20 : 39.7 29.0 40% 6
i 1
* ) not enougrh measurements
TABLE 5A: Amplitude as function of spark gap distance.
40
d=2 mm
r
d=5 mm
30
breakdown breakdown
20
tU 10·
> 10
0 0
-10
10 20 30 40 50 0 10 20 30 40 50
t [ns:t. t [ns]_
:,.f
40
20
0 0
-10 -10
0 20 40 60 80 100 0 20 40 60 80 100
t [ns] t [ns]
Fig. 5.20: Voltage across the cylindrical spark gap

measured with Tektronix digitizer for different gap
distances: original position, without a brass cylinder or
resistor rings: measured at termination South during
switching actions with CB-South.
120
measuring loop in combination with the large dI/dt (about
300 A/ns), already discussed in Fig. 5 .18. Calculations
with a 2D~model on this problem have been done. Uncertain
in this model are however: the unknown are radius; the
limited validity of the 20-model since the field lines
escaping through the gap expand in 30.
For gap distances larger than lOmm the waveforms were not
affected by the spark gap during the first lOOns. Although
the waveform reached . i ts top in this time interval no
breakdown occurred. On a longer time scale breakdown always
occurred. The time necessary to initiate a breakdown
increases with the gap distance and inversely with the
voltage. Therefore, for transients breakdown can occur a
certain time after the first top of the voltage.
Brass cylinder and resistor rings

The second series of measurements were the voltage
measurements across the same spark gap during switching
actions with CB-south in the case that termination South
was equipped with:
-spark gaps only;
-spark gaps combined with a brass cylinder;
-spark gaps combined with resistor rings.
To make a comparison possible between these three tests the
gap distance was chosen lOmm so that no breakdown occurred
during the first lOOns.
Table 5B gives the results of the measurements. The values
of the maximum top voltage and the average voltage are
obtained from waveforms as in Fig.5.21. Figure 5.21 shows
recordings of the Tektronix digitizer.
A brass cylinder around the current-transformer
arrangement, gives in comparison with only spark gaps,
little or no reduction of the voltage. Although the
HF-circuit is more compact, it is still too large for the
fast transients to avoid sparking (l.15m corresponds with a
back and forth transit time of 6.9ns).
121
CB-South
maximum average
DESIGN top- top- deviation number of
voltage voltage measurements
(kV) (kV)
spark gap 39.7 29.0 30% 6
-with cylinder 29.6 24.6 20% 3
-with resistors 15.9 13.3 25% 3
-
TABLE 5B: Amplitude of the voltage across the spark gap.
40 30
.d= lOmrn
~20
20
10
0
0
-10·
-2 -20
0 20 40 60 80 100 0 20 40 60 80 100
~ t (h!l]
15•
la.
,:!!,
> 5
-5·
-10
0 20 40 60 80 100
t [ns]
Fig. 5. 21: Voltage across the spark gap during swi tching
actions with CB-South. Left: only spark gaps; right: spark
gaps with brass cylinder; below: spark gaps with resistor
rings.
122
Resistor rings across the spark gaps reduced the voltages
across the gaps significantly. In contrast to the two other
test situations no sparking occurred irrespective of the
switching action (CB- o~ DC-action). Because the spark gap
is the weakest link the risk to the insulating collar is
even more reduced.
The 50Hz-current measurements are not affected by the
resistor rings (R"' 300). The induced current in the loop
formed by resistor rings and grounding strip outside the
current transformer (see Fig.5.14) is only a few milliamps
(measured by PNEM: 2-3mA while the total primary current is
400A). Also for a sheath current there are no problems
because the impedance of the grounding strip ("' 1. smn at
50Hz) is much smaller than the 300 of the resistor rings.
Dissipation is not a limiting factor in the choice of the
resistors. A resistor ring is composed of 24 parallel
branches each with two resistors in series. When we
approximate the waveform of the voltage during a CB-action
(see Fig.5.22) as an exponential curve with a maximum
voltage of 13. 3 kV and an 1/e-time of 200 ns the peak
dissipated power is about 5. 9 MW, i.e. about 122 kW per
resistor. The corresponding energy is however only 12.3 :mJ
per resistor because of the short pulse duration. Even when
15
J~1f\,~ -~- .~- -

1110
>"' 5
0 0.25 0.5 0.75 1
Fig. 5. 22: The transient voltage across the spark gap

equipped with a resistor ring falls off in about 2oons
(deviation 20%). This is obtained from 5 measurements
(recordings of the Nicolet) such as shown in this figure.
For the average top-voltage the value given in Table 5B is
used.
123
we take into acount the repetitive behavior of the
breakdowns during DC-actions the energy still remains low.
The power and energy per (single) waveform during a
DC-action is about 33 kW and 3.3 mJ per resistor,
respectively. The total dissipated energy is about
(500/lO)x3.3mJ=l65mJ when during 500ms every half-period a
breakdown occurs.
A more important question is whether the resistors can
withstand the high pulse voltages (about SkV per resistor
during ' 200ns). Therefore we used carbon composita
2w-resistors with shrink sleeve insulation to avoid flash
over across the outer resistor surface. Moreover the
resistors must be carefully installed to reduce the
inductance. The connection leads must be kept as short as
possible, and several series-connected resistors should be
installed circumferentially around the cylindrical spark
gap. A properly distributed resistor keeps the transient
B-fields inside the coaxial structure, but of course still
acts as an E-field antenna with respect to the outside
world.
124
CHAPTER 6
CONCLUSIONS
GENERAL CONCLUSIONS
-General, linear and basic design methods can be developed,
for the protection of electronics and (large) interconnec-
ted electrical systems against interference. such an
approach early in the design phase of apparatus or
experimental setups saves production costs for the
manufacturer and research time for the experimentalist.
-The design methods described in this thesis concentrate on
the reduction of dangerous voltages between critica!
points. This reduction can be achieved by the choice of a
correct layout, even under conditions of intense distur-
bances. In the emphasis on dangerous voltages and in the
possibility to create a safe local climate, we are
critical of the more vaguely formulated IEC definition of
Electro-Magnetic-Compatibility which is quoted here for
completeness: "The ability of a device, equipment or
system to function satisfactorily in its electromagnetic
environment without introducing intolerable electromagne-
tic disturbances to anything in that environment 11 •
From the results of chapter 2 through 5 the follor.ring
conclusionscan be drar.rn.
ELECTROHAGNETIC COHPATIBILITY AND GROUNDING
-The activity "grounding" should, and can indeed, reduce
dangerous voltage differences between critical points to
safe values.
-A grounding system is always a part of a number of
interlinked current loops. Design of a grounding system
involves an analysis based on the concept of closed
current paths.
-The circuits for ground currents should be designed as
compactly and locally as the circumstances allow: this
also results in a clearer design.
-A properly designed grounding structure is a structure
with a low transfer impedance. Through this structure
large external CM-currents may flow and still no dangerous
voltage differences are induced. In many electrical
installations grounding problems can be solved locally by
creating protected regions formed by grounding structures.
-The creation of a protected region does not rely on the
improvement of the overall 11 EM-environment 11 but instead,
provides an excellent local "climate" where it is needed.
-our design method keeps fluxes (self and mutual) small so
that we will have fewer deviation from the Kirchhoff 's
Voltage Law than a less compact design would give.
Moreover the compact and local approach reduces capacitive
and resistive coupling.
125
TRANSFER IHPEDANCE OF GROUNDING STRUCTURES
-The significance of the transfer impedance concept for
grounding structures is that simple measurements can
proviae' lts vaïue and thereby a meaningful criterion for
the quality of the grounding structure (including the
entire 11 layout" of the "network").
-The transfer impedance is only meaningful when primary and
secundary circuit are fully defined.
Grounding structures can be divided into tvo categories:
"ground1ng structures to protect leads" and "grounding
structures to protect instruments".
Groun~1ng structures to protect leads
-A 'GS for leads provides a controlled current path for

external CM-currents and creates a protected region for
the leads by means of its shape.
-The transfer impedance of a tube drops rapidly to zero at
higher frequencies because of its symmetrie geometry.
-Although the dc-resistance of an iron tube is higher than
that of a copper tube, the permeability of iron gives a
small skindepth, so that already at low frequencies the
transfer impedance of an iron tube falls below that of a
copper tube.
-Practical GS's with a low transfer impedance can also be
shaped as a plate or a conduit.
-In situations with intense interference, the transfer
impedance of condui ts can be further reduced wi th metal
covers.
-Due to the higher surface impedance of iron, in comparison
to copper, the external CM-current is reduced more by an
iron conduit than by a copper one.
Grounding structures to protect instruments
-For the protection of instruments we use GS' s around or
near the instruments. The useful DM-currents can enter the
protected region, whereas the external CM-currents are
diverted.
-In situations with a moderate level of disturbance, simple
GS' s -as described in this thesis- can be installed in
many practical situations, on different scales. Simple
techniques as orientation of the cable connections and
more metal at strategical places (e.g. plates, disks or
cylinders) can give significant reductions of the transfer
impedance of these GS's.
-In situations where much protection is necessary an
EMC-cabinet is an efficient solution. The front of the
EMC-cabinet can aften stay open, if the interference
source is at some distance. A correct handling of the
interference currents at the points where the leads come
in, is more important than the lack of shielding near the
open front. In cases where the EMC-cabinet has to be
located close to an intense HF-interference source the
126
door of the EMC-cabinet must be closed to also give the
required shielding.
DIFFERENTIATING/INTEGRATING HEASURING SYSTEH
-A D/I measuring system.has been developed with a overal1
risetime of 0.6ns, corresponding to a bandwidth of 580MHz.
-The advantages of a D/I-system are:
i) The benefits of a single capacitor as the high-voltage
arm in a impulse voltage divider are fully exploited
when this capacitor forms a part of a D/I-measuring
system: no traveling wave between high and low voltage
arm and an easy matching of the measuring cable.
ii) The components of the low-voltage arm can be of normal
size as used in electronics. The differentiating
resistor can be a long terminated cable. The
integrator capacitor can be a feedthrough capacitor.
-The advantage of a feedthrough capacitor as integrator
capacitor is that its symmetrie configuration reduces
the mutual inductance to the output to zero.
-The type and length of a coaxial cable should be
carefully selected when nanosecond risetime measurements
have to be made.
-Differentiating sensors combined with a~propriate
cabinets allow EMC-correct measurements in high-
interference surroundings, in high-voltage engineering,
in lightning research and in pulsed power.
ELECTROHAGNETIC COMPATIBILITY IN GIS-SUBSTATIONS
-A GIS-installation is an intense and concentrated source
of HF-interference. During switching actions locally
stored electrical energy is partially converted into high
power EM-waves with fast rise times of less than 5 ns.
-At interruptions in GIS-enclosures HF-waves can freely
escape to the outside of the installation. Resulting
voltages over interruptions and between other grounded
metal construction parts may lead to breakdown, sometimes
within very short times, less than 1 ns.
-A GIS-installation with only cables connected to the GIS
is an EMC-ideal installation when the outer conductors of
the GIS enclosure and of the HV-cables form one completely
closed conductor.
-In situations with necessary interruptions in a GIS-
enclosure (eg. GIS/HV-line transitions and in some
situations at GIS/HV-cable transitions) reduction of HF-
waves escaping via these interruption is possible. Carbon
composite resistors, surrounded by shrink sleeve, across
GIS/HV-cable interruptions reduce the voltages across
these interruptions and avoid sparking. Useful HF- losses
are caused by the resistors.
127
REFERENCES
[Ber 81] Bersier, R.

ME:ASUREMENTSDr"TF!E-1F!AONITY OF TV-RECEIVERS IN TIIE 3 1U
30MHz RANGE
Proc. 4th Int. Symp. on EMC, Zürich (1981) p.289
[Ber 83] Bersier, R. and B. Sentkutt1
RATIONALE AND EXPERPERIMENTAL EVIDENCE ON THE ADEQUACY OF
CONDUCTED INSTEAD OF RADIATED SUSCEPTIBILITY TESTS
Proc. Sth Int. Symp. on EMC, Zürich (1983) p.257.
[Bog 84] Boggs, S.A. and N. Fulimoto
TECHNIQUES AND INSTRUMENTATION FOR MEASUREMENT OF TRANSIENTS
-IN GAS-INSULATED SWITCHGEAR
IEEE Trans. on Electrical Insulation Vol. EI-19 No.2, 1984
[Dan 85] Danker, B.
NEW MEASURES TO DECREASE RADIATION FROM PRINTED CIRCUIT
BOARDS
Proc. 6th Int. Symp. on EMC Zürich (1985), p. 115.
[Dan 88] Danikas, M.G. and P.C.T. Van der Laan
FAST MEASUREMENTS OF PARTIAL DISCHARGE CURRENTS IN SOLID
DIELECTRIC SAMPLES CONTAINING VOIDS
IEEE Int. Symp. on Electrical Insulation, 1988, Boston,
pp. 250-252
[Def 90] Defourny, M.M., M.A. Van Houten, T.F. Buss and E.J.M. Van
Heesch
ELECTRICAL FIELD COMPUTATION. DESCRIPTION AND EXAMPLES WITH
EXPERIMENTAL COMPARISON
Int. ELECTROSOFT Conf., 21-23 aug. 1990, Lowell,
Massachusetts, USA, (to be published).
[Den 74] Denny. H.W. and J.A. Woody
CONSIDERATIONS IN TIIE DESIGN OF A GROUNDING SYSTEM FOR A
COMPLEX ELECTRONIC FACILITY
IEEE Symp. on EMC (1974) San Francisco, p.85.
[Deu 88] Van Deursen, A.P.J., M.A. Van Houten, E.W.L. Van Engelen,
P.F.M Gulickx, P.C.T. Van der Laan, E. Zwennes, A.J. Van
Dongen.
MEASUREMENTS OF CURRENTS AROUND AND IN LARGE GROUNDED
STRUCTURES
Proc. 19th Int. Conf. on Lightning Protection (ICLP, 1988),
Graz, Austria 3.1. p.143-147.
[Deu 89] Van Deursen A.P.J., M.A. Van Houten, P.F.M. Gulickx, P.C.T.
Van der Laan, E. Zwennes, A.J. Van Dongen.
REROUTING LIGHTNING CURRENTS IN A COMMUNICATION TOWER
[Deu 89] Van Deursen, A.P.J., J.M. Wetzer, P.C.T. Van der Laan
LOCAL PROTECTION OF EQUIPMENT IN HIGH VOLTAGE SUBSTATIONS
Proc. 6th Int. Symp. on HV Eng., 1989 New Orleans, USA, paper
31. 03, 4p.
[Deu 90] Van Deursen A.P.J.: (private communication).
128
[Eic 85] Eichert, B. and C. Staeger
EXTREMELY BROADBAND RF-CONNECTOR AND CABLE SCREENING
MEASUREMENTS (UP TO 150 dB SCREENING ATTENUATION) FOR
FREQUENCIES OF 1 kHz TO 10 GHz
IEC technica! paper, committee 46A (1985).
[Fuj 88] Fulimoto, N., S.J. Croall and S.M. Foty
TECHNIQUES FOR THE PROTECTION OF GAS-INSULATED SUBSTATIONS TO
CABLE INTERFACES
IEEE Trans. on Power Delivery, Vol.3, No.4, 1988
[Ger 85] German, R.F.
USE OF A GROUND GRID TO REDUCE PRINTED CIRCUIT BOARD
RADIATION
Proc. 6th Int. Symp. on EMC Zilrich (1985), p. 123.
[Gun 86] Gunn, R.
A COMMON APPROACH TO SIGNAL SEPARATION
EMC Technology ~ no.4 (1986) p. 53.
[Har 79] Harada,T. and Y.Aoshima
DEVELOPMENT OF HIGH PERFORMANCE LOW VOLTAGE ARMS FOR
CAPACITIVE VOLTAGE DIVIDERS
Proc. 3rd Int. Symp. on High Voltage Engineering, Milan, Italy,
1979, paper 42.14.
[Hee 87] Van Heesch, E.J.M., J.N.A.M. Van Rooil, R.G. Noil, P.C.T, Van
der Laan.
A NEW CURRENT AND VOLTAGE MEASURING SYSTEM; TESTED IN A 150kV
AND 400kV GIS
Proc. 5th Int. Symp. on High-Voltage Eng. , Vol. 3, 73. 06,
Braunschwelg (1987), 4p.
[Hee 89] Van Heesch, E.J.M., A.P.J. Van Deursen, M.A. Van Houten, G.
Jacobs, J. Kersten, P.C.T. Van der Laan.
FIELD TESTS AND RESPONSE OF THE Dil H.V. MEASURING SYSTEM
Proc. 6th Int. Symp. on High Voltage Engineering, New Orleans
(1989), paper 42.23, 4p.
[Her 72] Herring, T.H.
A DESIGN PROBLEM FOR THE GROUNDING SESSION
IEEE 1972. CH0638-7EMC
[Hou 89] Van Houten, M.A., E.J.M. Van Heesch, A.P.J Van Deursen,
J.N.A.M. Van RooiJ, R.G. Noil, P.C.T. Van der Laan
GENERAL METHODS FOR PROTECTION OF ELECTRONICS AGAINST
INTERFERENCE, TESTED IN HIGH-VOLTAGE SUBSTATIONS
Proc. 8th Int. Symp. on EMC Zürich (1989), 84N2.
[Jon 80] Jones, J.W.E. and M.A. Bridgwood
THE USE OF MAGNETIC CORES IN CONTROLLING EARTH LOOP CURRENTS
IERE Conf. Proc. No.47 on EMC 1980, p.387.
[Jon 83] Jones, J.W.E.
THE CONCEPTUAL PROBLEMS OF GROUND-PLANES
[Kad 59] Kaden, H.
WIRBELSTRÖME UND SCHIRMUNG IN DER NACHRICHTEN TECHNIK
Springer Berlin 1959 2. Auflage.
129
[Kel 64] Keller, R.
WIDEBAND HIGH VOLTAGE PROBE.
Rev. Sci. Instrum., Vol.35 (1964), pp.1057-1060.
(t.aa 78) 'htn ·der i..Mtn, P. €,·'f.
VOLTAGES IN TOROIDAL PINCH EXPERIMENTS
Los Alamos Reports LA-7335-MS, 1978.
[Laa 86) Van der Laan, P.C.T. and M.A. van Houten
DESIGN PHILOSOPHY FOR GROUNDING
Proc. 5th Int. Symp. on EMC, York (1986), p. 267.
[Laa 87) Van der Laan, P.C.T., M.A. Van Houten, A.P.J. Van Deursen
GROUNDING PHILOSOPHY
.Proc. 7th Int. Symp. on EMC, ZUrlch (1987), p. 567.
[Lat 88] Lathouwers, A.G.A.
ONTWERP VAN EEN D/I-MEETSYSTEEM VOOR HET METEN VAN SNELLE
TRANSIENTE SPANNINGEN IN GIS
Master thesis TU-E1ndhoven, EH.88.A.97
(Mar 86] Marsman, H.G.
BEVEILIGING VAN HOOGSPANNINGSNETTEN: TREND OF KEERPUNT
PATO Hfd.7: Stoorspannlngen, (1986).
[Mep 87) Meppelink, J., and P. Hofer
DESIGN AND CALIBRATION OF A HIGH VOLTAGE DIVIDER FOR
MEASUREMENTS OF VERY FAST TRANSIENTS IN GIS
Proc. 5th Int. Symp. on High-Voltage Eng., Vol.3, 71.08,
Braunschweig (1987).
[Mer 72] Mertel, H. K.
EXPLICIT GROUNDING ALLUSIONS
Proc. Int. EMC Symp. Chicago, Illlnois (1976) pp.302-304.
[Nah 73] Nahman, N.S.
A NOTE ON THE TRANSMISSION (RISE)TIME VERSUS LINE LENGTH IN
COAXIAL CABLES
IEEE Trans. on circuit theory (1973), p. 165.
[Oer 89] Oerlemans, A.J.W.A
HOOGFREQUENTE SPANNINGSMETINGEN
Master thesis TU-Eindhoven, EH.89.A.103
[Ott 76] Ott, H.W.
NOISE REDUCTION TECHNIQUES IN ELECTRONIC SYSTEMS
Wiley Newyork 1976, lst edition
[Ott 79] Ott, H.W.
GROUND -A PATH FOR CURRENT FLOW
IEEE Proc. Int. Symp. on EMC, (1979) 167
[Ott 81) Ott, H.W.
DIGITAL CIRCUIT GROUNDING AND INTERCONNECTION
Proc. 4th Int. Symp. on EMC, ZUrich (1981), p. 292.
[Pau 86] Paul, C.R.
MODELLING AND PREDICTION OF GROUND SHIFT .ON PRINTED CIRCUITS
BOARDS
Fifth Int. Conf. on EMC, York, 1986, p.37
130
[Ott 88] Ott, H.W.
NOISE REDUCTION TECHNIQUES IN ELECTRONIC SYSTEMS
Wiley Newyork 1989, 2e editlon.
[Pea 62] Pearlston, C.B.
CASE AND CABLE SHIELDING, BONDING, AND GROUNDING
CONSIDERATIONS IN ELECTROMAGNETIC INTERFERENCE
IRE Transaction on RFI. Vol. RFI-4 no.3 (1962) pp.1-16
( supplements).
[Ral 84) Raleigh, M, and R. Pechacek
FAST PASSIVE INTEGRATOR
Rev. Sci. lnstrum., Vol.55 (1984), pp. 2023-2026
[Ram 841 Ramo, S., J.R. Whlnnery and T. Van Duzer

FIELDS AND WAVES IN COMMUNICATION ELECTRONICS
John Wlley & Sons, Second Edition, 1984
[Rie 88] Van Riet, M.J.M.
AARDING GESLOTEN SCHAKEL INSTALLATIES
Electrotechniek, 66, Vol.1 (1988), p. 33.
[Sch 34] Schelkunoff, S.A.
THE ELECTROMAGNETIC THEORY OF COAXIAL TRANSMISSION LINES AND
CYLINDRICAL SHIELDS
Bell Syst. Tech. J.13 (1984) p.532-579.
[Sch 71] Schwab, A.J.
HIGH VOLTAGE MEASUREMENT TECHNIQUES
MIT Press, Cambridge, 1971.
[Sch 72] Schwab, A.J. and J.E. Pagel
PRECISION CAPACITIVE VOLTAGE DIVIDER FOR IMPULSE VOLTAGES
IEEE Transactions vol. PAS 91 (1972) pp. 2376-2382.
[Sen 83] Senff, J.J., P.C.T. Van der Laan, H. Antonides and L.M.L.F
Hossel et
SHIELDING PROPERTIES OF ISOLATED-PHASE BUSSYSTEMS
IEEE Trans. on Power App. and Systems, Vol. PAS-102 (1983),
2231-2238
[Sev 86) Sevat, P.A. A.
CONSTRUCTIE ELEMENTEN VOOR ONDERDRUKKEN VAN STORINGEN
PT Electrotechniek 39 (1984) p.36.
[Tas 86] Lange, K. and K.H. böcherer
TASCHENBUCH DER HOCHFREQUENZTECHNIK
Berlin: Springer, 1986, p. L11-L12.
[Van 80] Vance, F.E.
ELECTROMAGNETIC-INTERFERENCE CONTROL
IEEE Transaction on EMC, Vol. EMC-22 (1980), pp.319-328.
[Van 87] Vance, E.F.
COUPLING TO SHIELDED CABLES
RE Krieger Publishing Company, Malabar, Florida 1987
[VDE 86] Werkgroep Stoorspanningen in Gesloten Schakelinstallaties
concept: STOORSPANNINGEN IN ONDERSTATIONS EN MAATREGELEN TER
BESTRIJDING; deel III GESLOTEN SCHAKELINSTALLATIES
VDEN, 0067-HFL/WTGS 86-3965 Arnhem, mei 1986
. 131
[Vol 82] Volland, H.
HANDBOOK OF ATMOSPHERICS
Vol.l, CRC Press, Boca Raton USA 1982
[Wig 57] Wiglngton, R.L. and N.S. Nahman .
TRANSIENT ANALYSIS OF COAXIAL CABLES CONSIDERING SKIN EFFECT
Proc. IRE, 45 (1957). p. 166.
[Wil 75] Wikinson, W.S.
A SURVEY OF PAST, PRESENT AND POSSIBLE FUTURE SYSTEMS FOR THE
TRANSMISSION OF SIGNALS FROM THE EMC VIEWPOINT
Proc. lst. Int. Symp. on EMC, Montreux, 1975.
[Wet 88] Wetzer, J.M., C. Wen and P.C.T. Van der Laan
- BANDWIDTH LIMITATIONS OF GAP CURRENT MEASUREMENTS
Proc. IEEE Int. Symp. on Electrical Insulation, 1988, Boston,
p.355-358.
[Wol 81) Wolzak, G.G, J.A.G. Bekkers, P.C.T. Van der Laan
CAPACITVE MEASUREMENT OF HIGH DC VOLTAGES
Rev. Sci. lnstrum., Vol. 52 (1981) pp. 1572-1574.
[Wol 83) Wolzak, G.G.
THE DEVELOPMENT OF HIGH-VOLTAGE MEASURING TECHNIQUES
Ph.D. Thesis Eindhoven University of Technology, 1983.
[Zha 89] Zhang, W.Y., A.P.J. Van Deursen, P.C.T. Van der Laan
DIGITAL MEASUREMENTS ON HIGH VOLTAGE SURGE GENERATORS
Sixth Int.Symp. on High-Voltage Engineering, New Orleans,
1989, 50.04, 4p.
User software.programs:
[ANS 89] ANSOFT: A Finite Element Method field calculatlon program.
ANSOFT Corporation, University Technology Development
Center/4516 Henry street Pittsburg, PA 15213 412-683-4846
ENGINEERING SOFTWARE, January, 1989.
[STA 87] STATGRAPHICS: Statistica! graphlcs system by statistlcal
graphics corporation. 1987 STSC.INC.
[VU- 87] VU-POINT: A digltal data processing system for IBM
PC/XT/AT and compatible personal computers.
Version 1.2, oct.1987, SSS-IR-87-8347
132
SAMENVATTING
Het vakgebied Electromagnetic Compatibility (EMC) binnen de

elektrotechniek heeft tot doel elektronische apparaten en
elektrische systemen stÖringsvrij in elkaars nabijheid te
laten functioneren.
Dit proefschrift behandelt nieuwe EMC-concepten die leiden
tot een doelgerichte en consistente aanpak van praktische
storingsproblemen. Analyse dient methodes op te leveren om
storingen te verhelpen en liefst te voorkomen. De
ontwikkelde concepten worden in dit proefschrift voorname-
lijk toegepast in hoogspanningsinstallaties. Integratie van
moderne elektronica in dit soort installaties stelt hoge
EMC-eisen. Electromagnetic Compatibility wordt in deze ge-
vallen bereikt door een correcte layout toe te passen en
door leidingen en extra metaal op strategische plaatsert te
installeren. Overigens zijn de ontwikkelde concepten in
aangepaste vorm toepasbaar binnen de gehele elektrotech-
niek.
Het proefschrift begint met een kritische analyse van het
begrip "aarde", een onderwerp van fundamenteel belang
binnen de EMC. Tengevolge van een onjuiste voorstelling van
de werking van een aardingssysteem is een verwarrende
praktijk ontstaan rond het zelfstandig naamwoord "aarde".
Deze foute voorstelling is gekoppeld aan twee basiselemen-
ten die in de gangbare definities van "aarde" te vinden
zijn, nl.:
-Een "aarde" kan stroom opnemen of afgeven zonder daardoor
in spanning te veranderen
-Een "aarde" is een punt of vlak van .gelijke potentiaal dat
kan dienen als referentie voor onze schakeling.
Door ons nadrukkelijk te distantiëren van het potentiaal-
concept en ons te concentreren op de fysisch zinvolle
stromen, op de circuits waarin deze stromen lopen en op de
magnetische fluxen, hebben we een algemene beschermings-
filosofie ontwikkeld. Centraal hierin staat het begrip
transferimpedantie van aardingsstructuren (AS-en). Met
behulp van AS-en creëren we beschermende gebieden waarin
133
gevoelige elektronica probleemloos kan werken. We maken
onderscheid tussen AS-en voor leidingen en apparaten. In
deze zin_ vJ»:lllt àfLg.e].eidende :buitenmantel van een coaxiale
kabel als simpele AS ook reeds een beschermd gebied.· Voor
de bescherming van gevoelige instrumenten is een "EMC-kast"
een on~isbaar deel van de AS.
De transferimpedantie van een AS is een belangrijk begrip
omdat het een zinvol citerium oplevert voor de kwaliteit
van een AS (samen met de hele "layout" van .het "netwerk").
De vaak niet berekenbare transferimpedantie van een
uitgebreide AS kan experimenteel worden bepaald met
"current injection test methods" waarbij de veroorzaakte
spanningsverschillen tussen kritische klemmen worden
gemeten. Zowel modellen voor als metingen van de
transferimpedantie van AS-en voor leidingen en AS-en voor
instrumenten worden behandeld.
Voor het meten van hoge spanningen in hoogspanningsinstal-
laties is het transporteren van gedifferentieërde meet-
signalen gunstig voor EMC. Naast het belangrijke voordeel
dat de meetkabel karakteristiek kan worden afgesloten is
een tweede EMC-voordeel dat het eerste deel van de
integrator passief kan worden uitgevoerd. Onderdrukking van
snelle transiënten kan dan plaats vinden voordat deze de
kwetsbare actieve elektronica bereiken. Een belangrijke rol
is hier weggelegd voor de EMC-kast die zowel de
meetapparatuur als het geïntegreerde meetsignaal moet
beschermen. Beschreven wordt het ontwerp van een
Differentiërend/Integrerend (D/I)-systeem voor het meten
van snelle spanningstransiënten. Het D/I-systeem is onde:r
andere gebruikt voor het meten van spanningstransiënten in
het "Gas Insulated switchgear" onderstation Eindhoven-West.
Een GIS-installatie is een hoogfrequente stoorbron van hoog
electromagnetisch vermogen. Middelen om deze stoorbron
gedeeltelijk te temmen zijn voorhanden en worden besproken.
134
DANKWOORD
Het in dit proefschrift beschreven onderzoek is uitgevoerd
in de vakgroep "Hoogspanningstechniek" van de Technische
Universiteit Eindhoven. In dit onderzoek trachten we .een
balans te vinden tussen theorie en praktijk. In het zoeken
van deze balans hebben velen een zinvolle bijdrage
geleverd. Hen wil ik daarvoor danken.
Prof. dr. ir P.C.T. van der Laan ben ik zeer erkentelijk
voor de vele discussies over en stimulering van het
onderzoek.
Prof. dr. M.P.H. Weenink dank ik voor het kritisch doorne-
men van dit proefschrift en voor de aanvullende opmerkin-
gen.
Dr. A.P.J. van Deursen, dr.ir. E.J.H. van Heesch en dr.ir.
J.M. Wetzer, wil ik danken voor de vele vruchtbare discus-
sies en collegiale samenwerking in de diverse projecten en
meetexpedities.
De technische staf van het hoogspanningslaboratorium wil ik
danken voor de steun bij de experimentele realisatie van
het onderzoe·k. Met name wil ik vermelden ing. F.M. van
Gompel met wie ik plezierig heb samen gewerkt en die tevens
het tekenwerk verzorgde.
Een nuttige bijdrage aan het onderzoek werd geleverd door
de afstudeerders ir. A.G.A Lathouwers en ir. A.J.W.A.
Oerlemans en meerdere stagiaires.
Verder werd belangrijke medewerking verleend door de PNEM
en Holec Distributie Systemen te Amersfoort.
Hiervoor ben ik allen zeer erkentelijk.
~
M.A. van Houten.
135
CURRICULUM VITAE
Rinus van Houten werd op 16 december 1950 geboren te Breda.
In 1957 ging hij naar de Openbare Lagere School
"Middellaan" met als hoofd de Heer Viergever. Daarna volgde
hij achtereenvolgens:
1964-1968 Lagere Technische School "Gemeentelijke
Technische School Breda".
Afdeling: Electrotechniek.
1968-1971 Middelbare Technische School "Albertus Magnus"
Breda.
Afdeling: Electrotechniek.
1971-1976 Hogere Technische School "St.Virgilius", Breda
Afdeling: Electrotechniek •
. Afstudeerrichting: Informatietechniek.
1976-1980 Technische Hogeschool Eindhoven.
1982-1985 Afdeling: Electrotechniek.
Afstudeervakgroep: Theoretische Electrotechniek.
Afstudeeronderwerp: "Stabiliteitsonderzoek van
M.H.D. Plasma".
Afstudeerhoogleraar: Prof.dr. M.P.H. Weenink.
Afstudeerdatum: 23 mei 1985
Vervolgens werkte hij van 1985 tot 1989 in het Hoogspan-
ningslaboratorium van de Technische Universiteit te
Eindhoven als wetenschappelijk assistent. Zijn werk, onder
leiding van Prof.dr.ir. P.C.T. van der Laan, leidde tot dit
proefschrift.
136
STELLINGEN
behorende bij het proefschrift
in
van
M.A. van Houten
Eindhoven, 23 oktober 1990.

I
Vance is van mening dat zelfs de beste aardingspraktijk de grote ver-
anderingen in potentiaal van een centraal aardpunt in een gebouw, ont-
staan ten gevolge van een bliksem of EMP, niet kan voorkomen. Hij ver-
klaart dit met de spanningsval Ri + L~~ over de aardingsimpedantle.
Tot nu toe treffen we deze onjuiste mening en verklaring aan in vrij-
wel elk EMC-boek.
Edward F. Vance, uElectromagnetic-Interference Control", IEEE Trans.
on EHC, -Vol. EHC-22, No.4, 1980
II .
Classificering van aardingssystemen op grond van potentialen leidt tot
een veelvoud van aardingssystemen en tot een inconsistente aardings-
praktijk. Die inconsistentie blijkt ook uit de kunstgrepen, zoals het
toepassen van vonkbruggen en potentiaalvereffeningsstrips, waarmee men
achteraf de toch optredende spanningsverschillen tussen de afzonder-
lijk geaarde structuren tracht weg te werken.
John D.H. Osburn and Donald R.J.llh1te, uGrounding a Recommendation for
the Future", IEEE Int. Symp. on EHC Atlant& (1987), pp.155-160.
111
Ott's alternatieve definitie van aarde - een pad van (hopelijk) lage
impedantie voor stromen terug naar de bron - is een stap in de goede
richting. In feite mag die impedantie echter hoog zijn; vereist is een
lage transferimpedantle naar het te beschermen.circuit.
-Henry ll. Ott, uGround -A Path For Current Flow", IEEE Proc. Int.
Symp. on EHC, (1979) pp. 167-170
-Henry W. Ott, "Nolse Reductlon Techniques In Electronic Syste111S"
1111ey New York 1988, 2nd ed1t1on.
IV
Het belang van het transferlmpedantle concept voor aardlngsstructuren
is dat de werking van aardingsstructuren aanschouwelijk wordt. Daarom
kan dan ook vaak met eenvoudige metingen de kwaliteit van de aardings-
structuur bepaald worden. In principe is de transferimpedantle niet
alleen afhankelijk van de aardlngsstructuur, maar ook van de gehele
layout van het netwerk. In de praktijk geeft een royaal uitgevoerde
aardingsstructuur (b.v. een EMC-kast) aanzienlijke vrijheid bij de
layout van de schakeling.
Dit proefschrift, hoofdstuk 3.
v
De differentiërende/integrerende meetmethode (o.a. beschreven door
Wolzak) in combinatie met een EMC-kast levert een flexibel meetsysteem
op, dat bruikbaar is in allerlei situaties met intense elektromagneti-
sche storingsbronnen zoals. in hoogspanningstechnieken, in bliksemon-
derzoek en in "Pulsed Power".
-Gerard G. Wolzak, "The Development of High-Voltage Heasuring Tech-
nlques", proefschrift TUE (1983)
-Dit proefschrift, hoofdstuk 4.
VI
Een Gas Insulated Switchgear (GIS)-installatie in combinatie met uit-
sluitend hoogspanningskabels is vanuit het gezichtspunt van EMC een
ideale installatie, indien de GIS-omhulling en de aardschermen van de
kabels metallisch een volledig gesloten geheel vormen.
Dit proefschrift, hoofdstuk 5.
VII
Het nieuwe vakgebied EMC - de kWallficatie nieuw is nog steeds van
toepassing - is ontstaan dankzij bijdragen uit diverse gevestigde
vakgebieden die alle op hun wijze iets met elektromagnetische storing-
en te maken hebben. Deze verwarrende situatie is inherent aan de aan-
loopfase; een groei naar conceptuele helderheid en consistentie is
echter noodzakelijk. Aan deze groei kan de universiteit een belang-
rijke bijdrage leveren.
VIII
Na een periode van overaccentuering in het onderwijs van de vakken
economie en informatica zou nu meer aandacht voor de niet direct nut-
tige vakken zoals muziek en sterrenkunde een meer ontspannen leefkli-
maat ten goede kunnen komen.
IX
In vlakke landen moet een fietser vooral de wrijvingskracht overwinnen
en nauwelijks de zwaartekracht. In bergachtige streken is de situatie
net andersom. Dit maakt fietsen in Nederland tot een niet-
conservatieve activiteit, doch in Peru hoofdzakelijk tot een conserva-
tieve.
x
Op de ambtsketen van de rector magnificus en op de ingenieurs- en
doctorsbullen van de Technische Universiteit Eindhoven staat de ark
van Noach als een symbool voor "goede techniek". Het zou aanbeveling
verdienen dat de voorzitters van examencommissies hieraan enige woor-
den zouden wijden tijdens de buluitreiking.

EMC For High Voltage PDF

Uploaded by

Copyright:

Available Formats

EMC For High Voltage PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

EMC For High Voltage PDF

Uploaded by

Copyright:

Available Formats

Electromagnetic compatibility in high-voltage engineering

Citation for published version (APA):

Document status and date:

Please check the document version of this publication:

Download date: 04. Feb. 2019

M.A. VAN HOUTEN

ter verkrijging van de graad van doctor

druk: wibro dissertatiedrukkerlJ. helmond.

Houten, Marinus Albertus van

Electromagnetic compatibility in high-voltage engineering/

prof.dr.ir. P.C.T. van der Laan

ELECTROMAGNETIC COMPATIBILITY AND GROUNDING 13

THE TRANSFER IMPEDANCE or GROUNDING STRUCTURES 44

ELECTROMAGNETIC COMPATIBILITY IN GIS-SUBST ATIONS 97

Electromagnettc --compatibility, within the field of

The thesis starts with a critica! analysis of "Grounding",

High-voltage engineering and EHC

For the design and production of electrical equipment much

By grounding correctly we want to achieve the following:

2. 2 CRITICISM ON STANDARD DEFINITIONS OF 11

GROUND, A PERFECT 5INK OR SOURCE ?

LARGE IONOSPHERE OUR

Fig.2.1: Electric fields on different scales. For our

To arrive at a correct picture we rewrite Eq. 2 .1 with

Fig.2.2: Incorrect (a) and correct (b) picture of what a

We make three observations. First of all, Fig. 2.2b shows

GRDUND> A POINT OF EQUAL POTENTIAL ?

Fig. 2. 3: The voltage between the points A and B of an

For our argument here, it is important to realize that in

Fig. 2.4: Voltmeter readings as in Fig. 2.3 as a function

2.3 MISCONCEPTIONS DUE TO AN INCORRECT PHILOSOPHY

Since the analysis of grounding is commonly based on

(a) (b) (c) (d)

Fig. 2.5: The two voltmeters 1 and 2 (a) connected to the

leads · different magnetic fluxes and show therefore

SINGLE POINT GROUNDING

(a) (b) (c)

Fig. 2. 6: Single-point grounding is based on the sewer

MULTIPLE GROUNDING SYSTEHS

Fig. 2.7: Multiple grounding systems based on single point

As a special case of multiple grounding systems we mention

Ground loops have a bad reputation because:

Fig. 2.8: A circuit above a plane, with two connections to

2.4 DISTRIBUTED INDUCTANCE AND GROUNDING

depends on the actual integration path (compare Eq.2.3).

This design method keeps magnetic fluxes (self and mutual)

More grounding problems show up in the following cases;

Fig. 2.9: a) A network consisting of two subsystems

Fig. 2.10: A conduit-shaped grounding structure which

On the outside is formed a new external CM-circuit, now

2.6 PROTECTED REGIONS FORMED BY GROUNDING STRUCTURES

Fig 2.12: With grounding structures we decouple regions in

2.6.1 PROTECTED REGION FOR LEADS

(a) (b) (c) (d)

Fig. 2.14: Grounding structures (GS) for leads with their

2.6.2 PROTECTED REGION FOR INSTRUMENTS

grounding mains power

Fig. 2.17: Leads acting as large antennas. All leads carry

Fig. 2.18: Grounding structure werking as a 11 current-