LECTURE_NOTES_ON_HIGH_VOLTAGE_ENGINEERIN.pdf

LECTURE NOTES
ON
HIGH VOLTAGE ENGINEERING
A
l
l
J
N
T
U
W
o
r
l
d

UNIT-I
INTRODUCTION TO HIGH VOLTAGE TECHNOLOGY AND APPLICATIONS
INTRODUCTION
The potential at a point plays an important role in obtaining any information regarding the electrostatic
field at that point. The electric field intensity can be obtained from the potential by gradient operation
on the potential
i.e. E = – ∇ V ...(1)
which is nothing but differentiation and the electric field intensity can be used to find electric flux
density using the relation
D = εE ...(2)
The divergence of this flux density which is again a differentiation results in volume charge
density.
∇ . D = ρv ...(3)
Therefore, our objective should be to evaluate potential which of course can be found in terms
of, charge configuration. However it is not a simple job as the exact distribution of charges for a
particular potential at a point is not readily available. Writing εE = D in equation (3) we have
 εE = ρv
or – ∇  ε  ∇ V = ρv
or ε ∇
2
V = – ρv
ρ v
or ∇
2
V = – ε ...(4)
This is known as Poisson‘s equation. However, in most of the high voltage equipments, space
charges are not present and hence ρv = 0 and hence equation (4) is written as
∇
2
V = 0 ...(5)
Equation (5) is known as Laplace‘s equation
If ρv = 0, it indicates zero volume charge density but it allows point charges, line charge, ring
charge and surface charge density to exist at singular location as sources of the field.
Here ∇ is a vector operator and is termed as del operator and expressed mathematically in
cartesian coordinates as
∇ =
∂
x 
∂
y 
∂
z ...(6)
a a a
∂x ∂y ∂z
where a x , ay and az are unit vectors in the respective increasing directions.
A
l
l
J
N
T
U
W
o
r
l
d

Hence Laplace‘s equation in cartesian coordinates is given as
∇
2
V =
∂
2
V

∂
2
V

∂
2
V
∂x 2 ∂y 2 ∂z 2 = 0 ...(7)
Since ∇ . ∇ is a dot produce of two vectors, it is a scalar quantity. Following methods are
normally used for determination of the potential distribution
(i) Numerical methods
(ii) Electrolytic tank method.
Some of the numerical methods used are
(a) Finite difference method (FDM)
(b) Finite element method (FEM)
(c) Charge simulation method (CSM)
(d) Surface charge simulation method (SCSM).
FINITE DIFFERENCE METHOD
Let us assume that voltage variations is a two dimensional problem i.e. it varies in x-y plane and it
does not vary along z-co-ordinate and let us divide the interior of a cross section of the region where
the potential distribution is required into squares of length h on a side as shown in Fig. 0.1.
y
V2
V3
b
V1
c a
x
V0
d
V
4
Fig. 0.1 A portion of a region containing a two-dimensional potential field divided
into square of side h .
Assuming the region to be charge free
∇ . D = 0 or ∇ . E = 0
and for a two-dimensional situation
∂Ex

∂Ey
= 0
∂x ∂y
and from equation (7) the Laplace equation is
∂
2
V

∂
2
V
= 0 ...(8)
∂x
2
∂y
2
Approximate values for these partial derivatives may be obtained in terms of the assumed
values (Here V0 is to be obtained when V1, V2, V3 and V4 are known Fig. 1.
A
l
l
J
N
T
U
W
o
r
l
d

∂V V1 − V0 and ∂V  V0 − V3 ...(9)
∂x h ∂x h
From the gradients
a c
∂
2
V
∂V − ∂V

∂x a ∂x c
 V1 − V0 − V0  V3 ...(10)
∂x
2
0 h h
2
Similarly ∂
2
V V2 − V0 − V0  V4
∂y
2
0 h
2
Substituting in equation (8) we have
∂
2
V  ∂
2
V V1  V2  V3  V4 − 4V0 = 0
∂x
2
∂y
2
h2
1
or V0 = (V1 + V2 + V3 + V4) ...(11)
4
As mentioned earlier the potentials at four corners of the square are either known through com-
putations or at start, these correspond to boundary potentials which are known a priori. From equation
(11) it is clear that the potential at point O is the average of the potential at the four neighbouring
points. The iterative method uses equation (11) to determine the potential at the corner of every square
sub-division in turn and then the process is repeated over the entire region until the difference in values is
less than a prespecified value.
The method is found suitable only for two dimensional symmetrical field where a direct solution is
possible. In order to work for irregular three dimensional field so that these nodes are fixed upon
boundaries, becomes extremely difficult. Also to solve for such fields as very large number of V(x, y)
values of potential are required which needs very large computer memory and computation time and hence
this method is normally not recommended for a solution of such electrostatic problems.
FINITE ELEMENT METHOD
This method is not based on seeking the direct solution of Laplace equation as in case of FDM, instead
in Finite element method use is made of the fact that in an electrostatic field the total energy enclosed
in the whole field region acquires a minimum value. This means that this voltage distribution under
given conditions of electrode surface should make the enclosed energy function to be a minimum for a
given dielectric volume v.
We know that electrostatic energy stored per unit volume is given as
W = 1 ∈ E
2
...(12)
2
For a situation where electric field is not uniform, and if it can be assumed uniform for a differ-
ential volume δv, the electric energy over the complete volume is given as
W =
1
2 zV
1
2 ∈ ( − ∇V ) dv ...(13)
A
l
l
J
N
T
U
W
o
r
l
d

To obtain voltage distribution, our performance index is to minimise W as given in equation
(13).
Let us assume an isotropic dielectric medium and an electrostatic field without any space charge.
The potential V would be determined by the boundaries formed by the metal electrode surfaces.
Equation (13) can be rewritten in cartesian co-ordinates as
1
zzz
L
F ∂V I2
F ∂V I2
F∂V I 2 O
W = ∈ MG J  G J  G J P dxdydz ...(14)
2 M
H
∂x K H ∂y K H ∂z K P
N Q
Assuming that potential distribution is only two-dimensional and there is no change in potential
∂V
along z-direction, then ∂z = 0 and hence equation (14) reduces to
WA = z L
1
R ∂V I2
 F ∂V I2UO ...(15)
∈
M |F |
P dxdy
G J
zz M 2
G J V
S
∂x K P
H H ∂y K |
N
|
T WQ
Here z is constant and WA represents the energy density per unit area and the quantity within
integral sign represents differential energy per elementary area dA = dxdy.
In this method also the field between electrodes is divided into discrete elements as in FDM.
The shape of these elements is chosen to be triangular for two dimensional representation and tetrahe-
dron for three dimensional field representation Fig. 0.2 (a) and (b).
Vk
Vk
V
h Vj
V
i
V
j
Vi
Fig. (a) Triangular finite element (b) Tetrahedron finite element.
The shape and size of these finite elements is suitably chosen and these are irregularly distrib-
uted within the field. It is to be noted that wherever within the medium higher electric stresses are
expected e.g. corners and edges of electrodes, triangles of smaller size should be chosen.
Let us consider an element e1 as shown in Fig. 0.2(a) as part of the total field having nodes i, j
and k in anti-clockwise direction. There will be a large no. of such elements e1, e2 .....eN . Having
obtained the potential of the nodes of these elements, the potential distribution within each elements is
required to be obtained. For this normally a linear relations of V on x and y is assumed and hence the
first order approximation gives
V(x, y) = a1 + a2x + a3 y ...(16)
A
l
l
J
N
T
U
W
o
r
l
d

It is to be noted that for better accuracy of results higher order approximation e.g. square or
cubic would be required. Equation (16) implies that electric field intensity within the element is con-
stant and potentials at any point within the element are linearly distributed. The potentials at nodes i, j
and k are given as
Vi = a1 + a2xi + a3yi
Vj = a1 + a2xj + a3yj
Vk = a1 + a2xk + a3yk ...(17)
Equation (17) can be rewritten in matrix form as
L
V 1 xi
y a
i O L i O L 1 O
M P M x
j
P M P
M
V
j P  M1 y
j P M
a
2 P
NV
k Q N1 x
k
y
k Q Na
3 Q
By using Cramer‘s rules, the coefficient a1, a2, a3 can be obtained as follows
a = 1 (α V + α V + α V )
1
2∆ e
i i j j k k
a = 1 (β V + β V + β V )
2 k
2∆ e
i i j j k
and a3 =
1
(γi Vi + γj Vj + γk Vk)
2∆ e
...(18)
...(19)
where αi = xj yk – xk yj, αj = xk yi – xi yk , αk = xi yj – xj yi
βi = yi – yk , βj = yk – yi, βk = yi – yj
γi = xk – xj, γj = xi – xk, γk = xj – xi
and 2∆e = αi + αj + αk = βiγj – βjγi
where ∆e represents the area of the triangular element under consideration. As mentioned earlier the
nodes must be numbered anticlockwise, else ∆e may turn out to be negative.
From equation (16), the partial derivatives of V are
∂V = a
2
= f(V , V , V ) and ∂V = a
3
= f(V , V , V ) ...(20)
∂x ij k
∂y ijk
We know that for obtaining the voltage at various nodes we have to minimise the energy within
the whole system for which derivatives of energies with respect to potential distribution in each ele-
ment is required. For the element e under consideration, let We be the energy enclosed in the element,
then energy per unit length in the z-direction We /z denoted by W∆e can be obtained by using equation
(15) as follows
W 1 R∂V I 2 F ∂V I 2
U
e |F |
W =  ∈ ∆e G J 
G J V ...(21)
∆e
z 2 S
∂x K
H H ∂y K |
|
∆e = zze dxdy
T We
Here
To obtain condition for energy minimisation we differentiate partially equation (21) with
respect to Vi , Vj and Vk separately. Thus partially differentiating equation (21) with respect to Vi and
making use of equations (19) and (20).
A
l
l
J
N
T
U
W
o
r
l
d

∂W 1 F ∂a ∂a I
We have ∆e  ∈ ∆ e G 2a2
2
 3 J= 1
∈ ( a 2 β i  a3 γ i )
2 ∂Vi 2
∂Vi H ∂Vi K
= ε [(β
2
 γ
2
) V  (β β γ γ )V  (β β γ γ )V ] ...(22)
j j j k k
4∆ e
i i i i i i i k
Similarly, finding partial derivatives of equation (21) with respect to Vj and Vk and following
the procedure outlined above for partial derivative with respect to Vi and arranging all the three
equation in matrix from we have
∂W
∆e ε
L (β i
2
 γ i
2
)
M
 M(β j β i  γ j γ i )
∂Ve
4∆ e
M(β k β i  γ k γ i )
N
ε M
L(C
ii
)(C
ij
)
= M
(C ji )
C
jj
C
kj
4∆ e
M Cki
N
(β i β j  γ i γ j ) (β i β k  γ i γ k ) OLVi O
(β j
2
 γ j
2
) (β jβ k  γ j γ k )
PM
Vj P
(β β  γ γ ) (β 2
γ 2
) PM
Vk
P
k j k j k k P
N Q
Q
C
ik O LVi O
C jk
P
M
V P = [C] [V]
P M j
P ee
C
kk P NV
k Qe
Qe
...(22a)
...(23)
After considering a typical element e, the next step is to take into account all such elements in
the region under consideration and the energy associated with all the elements will then be
N
1
W =∑ We  ∈ [V
T
] [C] [V] ...(24)
2
e  1
LV
1 O
where MV2 P
[V] =
M M P
M P
NV
n Q
and n is the total number of nodes in the system and N is the no. of elements and [C] is called the
global stiffness matrix which is the sum of the individual matrices.
In general ∂W leads to
∂Vh
n
∑Vi Cik = 0 ...(25)
i  1
The solution of the above equations gives voltage distribution in the region. Of course while
seeking the final solution the boundary conditions must be satisfied and hence this would require
some iterative method for the exact solution.
The second approach could be to formulate energy function in terms of the unknown nodal voltage.
This energy function is subjected to certain constraints in terms of boundary conditions. The objective then
is to min. [W] subject to certain constraints. For this various mathematical programming techniques like,
Fletcher Powell technique, Fletcher technique, direct search techniques, self scaling variable metric
techniques can be used. A computer program can be developed and accuracy of the result can be obtained
depending upon the convergence critsion fed into the computer. A suitable initial guess for the solution can
always be made depending upon the system configuration and during every iteration the voltage can be
updated till all the boundary conditions are satisfied and the energy
A
l
l
J
N
T
U
W
o
r
l
d

function is minimised that is when the change in the energy function between two consecutive itera-
tions is less than a prespecified value.
The finite element method is useful for estimating electric fields at highly curved and thin elec-
trode surfaces with composite dielectric materials especially when the electric fields are uniform or
weakly non-uniform and can be expressed in two dimensioned geometrics. The method is normally
not recommended for three dimensional non-uniform fields.
CHARGE SIMULATION METHOD
As suggested by the name itself, in this method, the distributed charges on the surface of a conductor/
electrode or dielectric interfaces is simulated by replacing these charges by n discrete fictitious indi-
vidual charges arranged suitably inside the conductor or outside the space in which the field is to be
computed. These charges could be in the form of point, line or ring, depending upon the shape of the
electrode under consideration. It could be a suitable combination of these fictitious charges. The posi-
tion and type of simulation charges are to be determined first and then the field on the electrode
surface is determined by the potential function of these individual charges. In order to determine the
magnitude of these charges n no. of points are chosen on the surface of the conductor. These points
are known as ―contour points‘‘. The sum of the potentials due to fictitious charge distribution at any
contour points should correspond to the conductor potential Vc which is known a priori.
Suppose qi, is one of the fictitious charges and Vi is the potential of any point Pi in space which
is independent of the coordinate system chosen, the total potential Vi due to all the charges is given as
n
Vi = ∑
p
ij
q
j ...(26)
j  1
where pij are known as ‗‗potential Co-efficient‘‘ which are to be determined for different types of
charges by using Laplace‘s equation. We know that potential at a point P at a distance ‗a‘ from a point
charge q is given as
q
V = ...(27)
4π ∈a
So here the potential co-efficient p is 1
4π ∈ a
Similarly, these co-efficients for linear and ring or circular charges can also be obtained. It is
found these are also dependent upon various distance of these charges from the point under considera-
tion where potential is to be obtained and the permittivity of the medium as in case of a point charge
and hence potential co-efficients are constant number and hence the potential due to various types of
charges are a linear function of charges and this is how we get the potential at a point due to various
charges as an algebraic sum of potential due to individual charges.
A few contour points must also be taken at the electrode boundaries also and the potential due
to the simulated charge system should be obtained at these points and this should correspond to the
equipotentials or else, the type and location of charges should be changed to acquire the desired shape
and the given potential. Suppose we take ‗n‘ number of contour points and n no. of charges, the
follow-ing set of equations can be written
A
l
l
J
N
T
U
W
o
r
l
d

L p
11
p
12
M
Mp
21
p
22
M :
p
n 2
N p
n1
... p1n
... p2n
... pnn
O Lq
1 O
P Mq P
P M :
2
P
P M P
Q Nq
n Q
LV
1 O
MV P
...(28)
 M 2 P
M :
P
NV
n Q
The solution of these equations gives the magnitude of the individual charges and which corre-
sponds to electrode potential (V1 ....... Vn) at the given discrete points. Next, it is necessary to check
whether the type and location of charges as obtained from the solution of equation (28) satisfies the
actual boundary conditions every where on the electrode surfaces. It is just possible that at certain
check points the charges may not satisfy the potential at those points. This check for individual point
is carried out using equation (26). If simulation does not meet the accuracy criterion, the procedure is
repeated by changing either the number or type or location or all, of the simulation charges till
adequate charge system (simulation) is obtained. Once, this is achieved, potential or electric field
intensity at any point can be obtained.
The field intensity at a point due to various charges is obtained by vector addition of intensity
due to individual charges at that point. However, it is desirable to obtain the individual directional
components of field intensity separately. In cartesion coordinate system, the component of electric
field intensity along x-direction for n number of charges is given as
n
∂pij
n
E
x

∑
q
j

∑
(f
ij
)
x
q
j ...(29)
∂x
j  1 j  1
where (fij)x are known as field intensity co-efficients in x-direction.
In this method it is very important to select a suitable type of simulation charges and their
location for faster convergence of the solution e.g. for cylindrical electrodes finite line charges are
suitable, spherical electrodes have point charges or ring charges as suitable charges. However, for
fields with axial symmetry having projected circular structures, ring charges are found better. Experi-
ence of working on such problems certainly will play an important role for better and faster selection.
The procedure for CSM is summarised as follows :
1. Choose a suitable type and location of simulation charges within the electride system.
2. Select some contour point on the surface of the electrodes. A relatively larger no. of
contour points should be selected on the curved or corner points of the electrode.
3. Calculate the pij for different charges and locations (contour points) and assemble in the
form of a matrix.
4. Obtain inverse of this matrix and calculate the magnitude of charges (simulation).
5. Test whether the solution so obtained is feasible or not by selecting some check points on
the conductor surface. If the solution is feasible stop and calculate the electric field
intensity at requisite point. If not, repeat the procedure by either changing the type or
location of the simulation charges.
CSM has proved quite useful for estimation of electric field intensity for two and three dimen-
sional fields both with or without axial symmetry. It is a simple method and is found computationally
efficient and provides accurate results.
A
l
l
J
N
T
U
W
o
r
l
d

The simplicity with which CSM takes care of curved and rounded surfaces of electrodes or
interfaces of composite dielectric medium makes it a suitable method for field estimation. The compu-
tation time is much less as compared to FDM and FEM.
However, it is difficult to apply this methods for thin electrodes e.g. foils, plates or coatings as
some minimum gap distance between the location of a charge and electrode contours is required. Also,
it is found difficult to apply this method for electrodes with highly irregular and complicated bounda-
ries with sharp edges etc.
However, as mentioned earlier a good experience of selecting type and location of simulation
charge may solve some of these problem.
An improved version of CSM known as surface charge simulation method (SCSM) described
below is used to overcome the problem faced in CSM.
SURFACE CHARGE SIMULATION METHOD
Here a suitably distributed surface charge is used to simulate the complete equipotential surface i.e.
the electrode contour since the surface charge is located on the contour surface itself. In actual practice
the existing surface charge on the electrode configuration is simulated by integration of ring charges
placed on the electrode contour and dielectric boundaries. This results into a physically correct
reproduction of the whole electrode configuration.
The electrode contours are segmented as shown in Fig. 0.3 and to each segment ‗S’ a surface
charge density is assigned by a given function Sk(x) which could be a first degree approximation or a
polynomial as follows
n
σ(x) = ∑Sk (x).σ k ...(29)
k  0
(x)
surfacechargedens
ity
x

k–1 k

k+1
Distance along the electrode contour
Fig. . Segmented Contour path with assigned σ
A
l
l
J
N
T
U
W
o
r
l
d

The individual segments along the contour path can be represented as shown in Fig. 0.4.
Sk(x)
1
O x
k–1 xk
x
x
k+1
Fig. Representation of a segment Sk(x)
The value of Sk (x) is zero for x < xk–1 and is unity at x = xk and in between xk–1 and xk is given as
x −x
k −1 .
x
k
− x
k −1
With the representation the contour sarface is reproduced accurately and exactly and thus the
continuity of charge between the segments is assumed. Surface charges can be simulated either by line
or ring charges. Ring charge simulation is found to be more useful for fields with symmetry of
rotation. Each contour segment is assigned m no. of charges and the potential due to a charge qj, is
given by equation (26) and is rewritten here
m
V
i
=
∑
p
i
q
j
j  1
The potential co-efficient pik for a contour point i due to kth contour segment is obtained as
shown in Fig. 05.
y
x
l
x x x x
x
l = 1
m
Fig. 0.5. Concentrated charges to simulate surface charges
and is give as
p
ik σk
=
zx
σ( x ) . p
ix
dx ...(30)
Now substituting equation (29) in equation (30) we have
n
p
ikσk
=
zx ∑
S
k
(x) . σk
p
ix
dx
...(31)
k  0
Since each segment is divided into m intervals as shown in Fig. 0.5, equation (31) can be
rewritten as
xk m
(x) p dx 
∑ (y ) p σ
s s
...(32)
p
ikσk
=
zxk −1 k ix k l i l k
l  1
The potential coefficient pil are similar to the coefficients derived from a single concentrated
charge in CSM. This coefficient, therefore, can be obtained for a line charge or by solving elliptical
integral for a ring charge. The electric field intensity at any contour point i due to kth contour segment
is given as
A
l
l
J
N
T
U
W
o
r
l
d

m
E
i
=
∑σk
f
ik ...(33)
x  1
where fik are the field intensity co-efficients.
As discussed this method requires a large number of elements, normally more than 2500, inde-
pendent of the surface shape and thus require large computational efforts. Also, due to certain
practical difficulties this method is not used as frequently as other numerical methods for estimation of
electric fields.
COMPARISON OF VARIOUS TECHNIQUES
Out of the various techniques FDM is the simplest to compute and understand but the computation
effort and computer memory requirements are the highest. Also, since all difference equations are
approximation to the actual field conditions, the final solution may have considerable error.
Finite element method is a general method and has been used for almost all fields of engineer-
ing. The method is suitable for estimating fields at highly curved and thin electrode surfaces with
different dielectric materials. However, this method is more useful for uniform or weakly non-uniform
fields and which can be represented by two dimensional geometries. This method is recommended for
three dimensional complicated field configurations.
Charge Simulation Method (CSM) is considered to be one of the most superior and acceptable
method for two and three dimensional configuration with more than one dielectric and with electrode
systems of any desired shape since this method is based on minimization of the energy function which
could be subjected to any operating constraints e.g. environmental condition, it has proved to be
highly accurate method. Because of inherent features of the technique, this method also helps in
optimising electrode configuration. In this electrode configuration optimisation problems the objective
is to have field intensity as low as possible subject to the condition that a constant field intensity exists
on the complete electrode surface. With this optimisation, a higher life expectancy of high voltage
equipments can be achieved.
However, as mentioned earlier this method can not be used for thin electrodes e.g. foils, plates
or coatings due to the requirement of a minimum gap distance between the location of a charge and
electrode contour. Also, this method is not suitable for highly irregular electrode boundaries.
The surface charge simulation method even though takes into account the actual surface charge
distribution on the electrode surface, this method is not normally recommended for solution of field
problem due to some practical difficulties.
An important difference between the various method is that the FDM and FEM can be used
only for bounded field whereas CSM and SCMS can also be used for unbounded fields.
OPTIMISATION OF ELECTRODE CONFIGURATION
Various numerical techniques have been used to optimise the electrode configuration so that the
dielectric material is optimally utilised as a result a considerable improvement in dielectric behaviour
is achieved and a higher life expectancy of high voltage equipments can be anticipated. When we talk
of electrode configuration optimisation, we really mean the electric field intensity optimisation. Even
though some work has been dedicated for electrode configurations optimisation by FDM and FEM
methods, yet the inherent suitability of CSM for optimisation, lot of work has been reported in
literature using this technique.
A
l
l
J
N
T
U
W
o
r
l
d

The objective of optimisation is to determine the configuration of electrodes which may result into a
minimum and constant field intensity on the complete electrode surface. The optimisation tech-nique is
based on the partial discharge inception electric field intensity Epd which depends upon the dielectric
material, its pressure (if gas is the medium) and the electrode configuration. It is to be noted that if the
electric field is uniform or weakly non-uniform, the partial discharge or normal breakdown takes place at
the same electric field intensity. Therefore, it is only the electrode configuration which can be optimised. If
Epd is more than the electric field intensity E applied, partial discharge can not take place which means the
electrode can be said to be optimised, if at a given voltage the maximum value
of
E
on its surface is as small as possible. Since the maximum value of E/Epd depends upon three
E
pd
parametres the shape, size and position of electrodes, three different types of optimisation possibilities
exist. The optimum shape of an electrode in characterised by
Min. (E/Epd)max = constant ...(34) The optimisation methods are based on
iterative process and when equation (0.34) is satisfied, the optimum electrode configuration is
obtained. While using CSM, following strategies are used for
optomisation of electrode configuration.
(i) Displacement of contour points perpendicular to the surface
(ii) Changing the position of the ‗‗optimisation charges‘‘ and contour points
(iii) Modification of contour elements
A brief view of these methods is given below.
A
l
l
J
N
T
U
W
o
r
l
d

UNIT-II:
BREAK DOWN IN GASEOUS AND LIQUID DIELECTRICS
INTRODUCTION
With ever increasing demand of electrical energy, the power system is growing both in size and com-
plexities. The generating capacities of power plants and transmission voltage are on the increase be-
cause of their inherent advantages. If the transmission voltage is doubled, the power transfer capability
of the system becomes four times and the line losses are also relatively reduced. As a result, it
becomes a stronger and economical system. In India, we already have 400 kV lines in operation and
800 kV lines are being planned. In big cities, the conventional transmission voltages (110 kV–220 kV
etc.) are being used as distribution voltages because of increased demand. A system (transmission,
switchgear, etc.) designed for 400 kV and above using conventional insulating materials is both bulky
and expensive and, therefore, newer and newer insulating materials are being investigated to bring
down both the cost and space requirements. The electrically live conductors are supported on
insulating materials and sufficient air clearances are provided to avoid flashover or short circuits
between the live parts of the system and the grounded structures. Sometimes, a live conductor is to be
immersed in an insulating liquid to bring down the size of the container and at the same time provide
sufficient insulation between the live conductor and the grounded container. In electrical engineering
all the three media, viz. the gas, the liquid and the solid are being used and, therefore, we study here
the mechanism of breakdown of these media.
MECHANISM OF BREAKDOWN OF GASES
At normal temperature and pressure, the gases are excellent insulators. The current conduction is of
the order of 10
–10
A/cm
2
. This current conduction results from the ionisation of air by the cosmic
radiation and the radioactive substances present in the atmosphere and the earth. At higher fields,
charged parti-cles may gain sufficient energy between collision to cause ionisation on impact with
neutral molecules. It is known that during an elastic collision, an electron loses little energy and
rapidly builds up its kinetic energy which is supplied by an external electric field. On the other hand,
during elastic colli-sion, a large part of the kinetic energy is transformed into potential energy by
ionising the molecule struck by the electron. Ionisation by electron impact under strong electric field
is the most important process leading to breakdown of gases.
This ionisation by radiation or photons involves the interaction of radiation with matter.
Photoionisation occurs when the amount of radiation energy absorbed by an atom or molecule exceeds its
ionisation energy and is represented as A + hν → A
+
+ e where A represents a neutral atom or
1
A
l
l
J
N
T
U
W
o
r
l
d

molecule in the gas and hν the photon energy. Photoionization is a secondary ionization process and is
essential in the streamer breakdown mechanism and in some corona discharges. If the photon energy
is less than the ionization energy, it may still be absorbed thus raising the atom to a higher energy
level. This is known as photoexcitation.
The life time of certain elements in some of the excited electronic states extends to seconds.
These are known as metastable states and these atoms are known as metastables. Metastables have a
relatively high potential energy and are, therefore, able to ionize neutral particles. Let A be the atom to
be ionized and B
m
the metastable, when B
m
collides with A, ionization may take place according to the
reaction.
A + B
m
→ A
+
+ B + e
Ionization by metastable interactions comes into operation long after excitation and it has been
shown that these reactions are responsible for long-time lags observed in some gases.
Thermal Ionisation: The term thermal ionisation in general applies to the ionizing actions of
molecular collisions, radiation and electron collisions occurring in gases at high temperatures. When a
gas is heated to high temperature, some of the gas molecules acquire high kinetic energy and these
particles after collision with neutral particles ionize them and release electrons. These electrons and
other high-velocity molecules in turn collide with other particles and release more electrons. Thus, the
gas gets ionized. In this process, some of the electrons may recombine with positive ions resulting into
neutral molecule. Therefore, a situation is reached when under thermodynamic equilibrium condition
the rate of new ion formation must be equal to the rate of recombination. Using this assumption, Saha
derived an expression for the degree of ionization β in terms of the gas pressure and absolute tempera-
ture as follows:
β
2
1 (2πme )
3 / 2
(KT)5 / 2 e− W / KT
1 − β
2
p h
or β2
2.4  10 − 4
T
5 / 2
e−W
i
/KT
1− β
2
p
where p is the pressure in Torr, Wi the ionization energy of the gas, K the Boltzmann‘s constant, β the
ratio ni/n and ni the number of ionized particles of total n particles. Since β depends upon the tempera-
ture it is clear that the degree of ionization is negligible at room temperature. Also, if we substitute the
values of p, Wi, K and T, it can be shown that thermal ionization of gas becomes significant only if
temperature exceeds 1000° K.
TOWNSEND’S FIRST IONIZATION COEFFICIENT
Consider a parallel plate capacitor having gas as an insulat-ing
medium and separated by a distance d as shown in Fig. 1.1.
When no electric field is set up between the plates, a state of
equilibrium exists between the state of electron and positive
ion generation due to the decay processes. This state of
equilibrium will be disturbed moment a high electric field is
applied. The variation of current as a function of voltage
Fig. Parallel plate capacitor
A
l
l
J
N
T
U
W
o
r
l
d

was studied by Townsend. He found that the current at
first increased proportionally as the voltage is increased and
then remains constant, at I0 which corresponds to the
saturation current. At still higher voltages, the current in-
creases exponentially. The variation of current as
a function of voltage is shown in Fig. 1.2. The exponen-
tial increase in current is due to ionization of gas by elec-
tron collision. As the voltage increases V/d increases and
hence the electrons are accelerated more and more and
between collisions these acquire higher kinetic energy
and, therefore, knock out more and more electrons.
To explain the exponential rise in current, Townsend
introduced a coefficient α known as Townsend’s first ionization coefficient and is defined as the
number of electrons produced by an elec-tron per unit length of path in the direction of field. Let n0 be
the number of electons leaving the cathode and when these have moved through a distance x from the
cathode, these become n. Now when these n electrons move through a distance dx produce additional
dn electrons due to collision. There-fore,
or
or
or
or
dn = α n dx
dn
 α dx
n
ln n = αx + A
Now at x = 0, n = n0. Therefore,
ln n0 = A
ln n = α x + ln n0
ln
n
 α x
n0
At x = d, n = n0 eαd
. Therefore, in terms of current
I = I0 eαd
The term eαd
is called the electron avalanche and it represents the number of electrons
produced by one electron in travelling from cathode to anode.
CATHODE PROCESSES—SECONDARY EFFECTS
Cathode plays an important role in gas discharges by supplying electrons for the initiation, sustainance
and completion of a discharge. In a metal, under normal condition, electrons are not allowed to leave
the surface as they are tied together due to the electrostatic force between the electrons and the ions in
the lattice. The energy required to knock out an electron from a Fermi level is konwn as the work
function and is a characteristic of a given material. There are various ways in which this energy can be
supplied to release the electron.
Fig. Variation of current as a
function of voltage
A
l
l
J
N
T
U
W
o
r
l
d

Thermionic Emission: At room temperature, the conduction electrons of the metal do not have suffi-cient
thermal energy to leave the surface. However, if the metals are heated to temperature 1500°K and above,
the electrons will receive energy from the violent thermal lattice in vibration sufficient to cross the surface
barrier and leave the metal. After extensive investigation of electron emission from metals at high
temperature, Richardson developed an expression for the saturation current density Js as
Js =
4πme K
2
T
2
e
− Wα / KT
A/m
2
h
3
where the various terms have their usual significance.
Let A =
4πme K
2
h
3
the above expression becomes
Js = AT
2
e
–W/KT
which shows that the saturation current density increases with decrease in work function and increase
in temperature. Substituting the values of me, K and h, A is found to be 120 × 10
4
A/m
2
K
2
. However,
the experimentally obtained value of A is lower than what is predicted by the equation above. The
discrep-ancy is due to the surface imperfections and surface impurities of the metal. The gas present
between the electrode affects the thermionic emission as the gas may be absorbed by the metal and
can also damage the electrode surface due to continuous impinging of ions. Also, the work function is
observed to be lowered due to thermal expansion of crystal structure. Normally metals with low work
function are used as cathode for thermionic emission.
Field Emission: If a strong electric field is applied between the electrodes, the effective work function
of the cathode decreases and is given by
W′ = W – ε
3/2
E
1/2
and the saturation current density is then given by
Js = AT
2
e
–W′/KT
This is known as Schottky effect and holds good over a wide range of temperature and electric
fields. Calculations have shown that at room temperature the total emission is still low even when
fields of the order of 10
5
V/cm are applied. However, if the field is of the order of 10
7
V/cm, the
emission current has been observed to be much larger than the calculated thermionic value. This can
be ex-plained only through quantum mechanics at these high surface gradients, the cathode surface
barrier becomes very thin and quantum tunnelling of electrons occurs which leads to field emission
even at room temperature.
Electron Emission by Positive Ion and Excited Atom Bombardment
Electrons may be emitted by the bombardment of positive ion on the cathode surface. This is known as
secondary emission. In order to effect secondary emission, the positive ion must have energy more than
twice the work function of the metal since one electron will neutralize the bombarding positive ion and the
other electron will be released. If Wk and Wp are the kinetic and potential energies, respectively of the
positive ion then for secondary emission to take place Wk + Wp ≥ 2W. The electron emission by positive
ion is the principal secondary process in the Townsend spark discharge mechanism. Neutral
A
l
l
J
N
T
U
W
o
r
l
d

excited atoms or molecules (metastables) incident upon the cathode surface are also capable of releas-
ing electron from the surface.
TOWNSEND SECOND IONISATION COEFFICIENT
From the equation
I = I0 eαx
We have, taking log on both the sides.
Fig. Variation of gap current with electrode spacing in uniform E
ln I = ln I0 + αx
This is a straight line equation with slope α and intercept ln I0 as shown in Fig. if for a given
pressure p, E is kept constant.
Townsend in his earlier investigations had observed that the current in parallel plate gap in-
creased more rapidly with increase in voltage as compared to the one given by the above equation. To
explain this departure from linearity, Townsend suggested that a second mechanism must be affecting
the current. He postulated that the additional current must be due to the presence of positive ions and
the photons. The positive ions will liberate electrons by collision with gas molecules and by bombard-
ment against the cathode. Similarly, the photons will also release electrons after collision with gas
molecules and from the cathode after photon impact.
Let us consider the phenomenon of self-sustained discharge where the electrons are released
from the cathode by positive ion bombardment.
Let n0 be the number of electrons released from the cathode by ultraviolet radiation, n+ the
number of electrons released from the cathode due to positive ion bombardment and n the number of
electrons reaching the anode. Let ν, known as Townsend second ionization co-efficient be defined as
the number of electrons released from cathode per incident positive ion, Then
n = (n0 + n+)eαd
Now total number of electrons released from the cathode is (n0 + n+) and those reaching the
anode are n, therefore, the number of electrons released from the gas = n – (n0 + n+), and
corresponding to each electron released from the gas there will be one positive ion and assuming each
positive ion releases ν effective electrons from the cathode then
A
l
l
J
N
T
U
W
o
r
l
d

or
or
or
n+ = ν [n – (n0 + n+)]
n+ = νn – νn0 – νn+
(1 + ν) n+ = ν(n – n0)
n =
ν(n − n
0
)
+
1  ν
Substituting n+ in the previous expression for n, we have
L ν( n − n0 ) Oαd (1  ν ) n  νn − νn
n = Mn
0  Pe
= 0 0
eαd
1 1  ν
N  v Q
n
0  νn αd
= e
1 ν
or (n + νn) = n0 eαd
+ νneαd
or n + νn – νneαd
= n0eαd
or n[1+ ν – νeαd
] = n eαd
0
n eαd n eαd
or n = 0 = 0
1  νn(1 − eαd
) 1 − ν (eαd
− 1)
In terms of current
I =
I 0 eαd
1 − ν(eαd
− 1)
Earlier Townsend derived an expression for current as
I = I
(α − β) e
( α − β)d
0
α − β e( α − β )d
where β represents the number of ion pairs produced by positive ion travelling 1 cm path in the direc-
tion of field. Townsend‘s original suggestion that the positive ion after collision with gas molecule
releases electron does not hold good as ions rapidly lose energy in elastic collision and ordinarily are
unable to gain sufficient energy from the electric field to cause ionization on collision with gas
molecules or atoms.
In practice positive ions, photons and metastable, all the three may participate in the process of
ionization. It depends upon the experimental conditions. There may be more than one mechanism
producing secondary ionization in the discharge gap and, therefore, it is customary to express the net
secondary ionization effect by a single coefficient v and represent the current by the above equation
keeping in mind that ν may represent one or more of the several possible mechanism.
ν = ν1 + ν2 + ν3 + .....
It is to be noted that the value of ν depends upon the work function of the material. If the work
function of the cathode surface is low, under the same experimental conditions will produce more
emission. Also, the value of ν is relatively small at low value of E/p and will increase with increase in
E/p. This is because at higher values of E/p, there will be more number of positive ions and photons of
sufficiently large energy to cause ionization upon impact on the cathode surface. It is to be noted that
the influence of ν on breakdown mechanism is restricted to Townsend breakdown mechanism i.e., to
low-pressure breakdown only.
A
l
l
J
N
T
U
W
o
r
l
d

TOWNSEND BREAKDOWN MECHANISM
When voltage between the anode and cathode is increased, the current at the anode is given by
I =
I 0 eαd
1 − ν (eαd
− 1)
The current becomes infinite if
1 – ν(eαd
–1) = 0
or ν(eα d
– 1) = 1
or νeα d
≈ 1
Since normally eα d
 1
the current in the anode equals the current in the external cirrcuit. Theoretically the current becomes
infinitely large under the above mentioned condition but practically it is limited by the resistance of the
external circuit and partially by the voltage drop in the arc. The condition νeαd
= 1 defines the condition for
beginning of spark and is known as the Townsend criterion for spark formation or Townsend break-down
criterion. Using the above equations, the following three conditions are possible.
(1) νeαd
=1
The number of ion pairs produced in the gap by the passage of arc electron avalanche is suffi-
ciently large and the resulting positive ions on bombarding the cathode are able to relase one
secondary electron and so cause a repetition of the avalanche process. The discharge is then
said to be self-sustained as the discharge will sustain itself even if the source producing I0 is
removed. Therefore, the condition νeαd
 defines the threshold sparking condition.
(2) νeαd
> 1
Here ionization produced by successive avalanche is cumulative. The spark discharge grows
more rapidly the more νeαd
exceeds unity.
(3) νeαd
< 1
Here the current I is not self-sustained i.e., on removal of the source the current I0 ceases to flow.
A
l
l
J
N
T
U
W
o
r
l
d

electric field E0 due to the space charge field. Fig. shows the electric field around an avalanche as it
progresses along the gap and the resultant field i.e., the superposition of the space charge field and the
original field E0. Since the electrons have higher mobility, the space charge at the head of the avalanche is
considered to be negative and is assumed to be concentrated within a spherical volume. It can be seen from
Fig. 1.4 that the filed at the head of the avalanche is strengthened. The field between the two assumed
charge centres i.e., the electrons and positive ions is decreased as the field due to the charge centres
opposes the main field E0 and again the field between the positive space charge centre and the cathode is
strengthened as the space charge field aids the main field E0 in this region. It has been observed that if the
charge carrier number exceeds 10
6
, the field distortion becomes noticeable. If the distortion of field is of
1%, it would lead to a doubling of the avalanche but as the field distortion is only near the head of the
avalanche, it does not have a significance on the discharge phenomenon. However, if the charge carrier
exceeds 10
8
, the space charge field becomes almost of the same magni-tude as the main field E0 and hence
it may lead to initiation of a streamer. The space charge field, therefore, plays a very important role in the
mechanism of electric discharge in a non-uniform gap.
Townsend suggested that the electric spark discharge is due to the ionization of gas molecule
by the electron impact and release of electrons from cathode due to positive ion bombardment at the
cathode. According to this theory, the formative time lag of the spark should be at best equal to the
electron transit time tr. At pressures around atmospheric and above p.d. > 10
3
Torr-cm, the
experimen-tally determined time lags have been found to be much shorter than tr. Study of the
photographs of the avalanche development has also shown that under certain conditions, the space
charge developed in an avalanche is capable of transforming the avalanche into channels of ionization
known as streamers that lead to rapid development of breakdown. It has also been observed through
measurement that the transformation from avalanche to streamer generally takes place when the
charge within the avalanche head reaches a critical value of
n0eαx
≈ 10
8
or αxc ≈ 18 to 20
where Xc is the length of the avalanche parth in field direction when it reaches the critical size. If the
gap length d < Xc, the initiation of streamer is unlikely.
The short-time lags associated with the discharge development led Raether and independently
Meek and Meek and Loeb to the advancement of the theory of streamer of Kanal mechanism for spark
formation, in which the secondary mechanism results from photoionization of gas molecules and is
independent of the electrodes.
Raether and Meek have proposed that when the avalanche in the gap reaches a certain critical
size the combined space charge field and externally applied field E0 lead to intense ionization and
excitation of the gas particles in front of the avalanche head. There is recombination of electrons and
positive ion resulting in generation of photons and these photons in turn generate secondary electrons
by the photoionization process. These electrons under the influence of the electric field develop into
secondary avalanches as shown in Fig. Since photons travel with velocity of light, the process leads to
a rapid development of conduction channel across the gap.
A
l
l
J
N
T
U
W
o
r
l
d

Fig. Secondary avalanche formation by photoelectrons
Raether after thorough experimental investigation developed an empirical relation for the
streamer spark criterion of the form
αxc = 17.7 + ln xc + ln
Er
E0
where Er is the radial field due to space charge and E0 is the externally applied field.
Now for transformation of avalanche into a streamer Er ≈ E
Therefore, αxc = 17.7 + ln xc
For a uniform field gap, breakdown voltage through streamer mechanism is obtained on the
assumption that the transition from avalanche to streamer occurs when the avalanche has just crossed
the gap. The equation above, therefore, becomes
αd = 17.7 + ln d
When the critical length xc ≥ d minimum breakdown by streamer mechanism is brought about.
The condition Xc = d gives the smallest value of α to produce streamer breakdown.
Meek suggested that the transition from avalanche to streamer takes place when the radial field
about the positive space charge in an electron avalanche attains a value of the order of the externally
applied field. He showed that the value of the radial field can be otained by using the expression.
Er = 5.3 × 10
–7 αeαx
volts/cm.
1 / 2
(x / P)
where x is the distance in cm which the avalanche has progressed, p the gas pressure in Torr and α the
Townsend coefficient of ionization by electrons corresponding to the applied field E. The minimum
breakdown voltage is assumed to correspond to the condition when the avalanche has crossed the gap
of length d and the space charge field Er approaches the externally applied field i.e., at x = d, Er = E.
Substituting these values in the above equation, we have
αeαd
E= 5.3 × 10–7
(d/p)1/ 2
Taking ln on both the sides, we have
ln E = – 14.5 + ln α –
1
ln
d
+ αd
2 p
ln E – ln p = – 14.5 + ln α – ln p –
1
1n
d
+ αd
2 p
A
l
l
J
N
T
U
W
o
r
l
d

1n E  − 14.5  ln
α − 1 ln d  αd
p p 2 p
The experimentally determined values of α/p and the corresponding E/p are used to solve the
above equation using trial and error method. Values of α/p corresponding to E/p at a given pressure
are chosen until the equation is satisfied.
THE SPARKING POTENTIAL—PASCHEN’S LAW
The Townsend‘s Criterion
ν(eαd
– 1) = 1
enables the evaluation of breakdown voltage of the gap by the use of appropriate values of α/p and ν
corresponding to the values E/p when the current is too low to damage the cathode and also the space
charge distortions are minimum. A close agreement between the calculated and experimentally deter-
mined values is obtained when the gaps are short or long and the pressure is relatively low.
An expression for the breakdown voltage for uniform field gaps as a function of gap length and
gas pressure can be derived from the threshold equation by expressing the ionization coefficient α/p as
a function of field strength E and gas pressure p i.e.,
α F E I
p
 f G J
H p K
Substituting this, we have
ef(E/p) pd = 1  1
ν
Taking ln both the sides, we have
F E I L 1 O
f
G Jpd  ln
M
 1  K say
H p K ν
P
N Q
For uniform field E = Vb .
FVb I
d
Therefore, f G J. pd  K
H pd K
F Vb I K
or f G J 
pd
H pd K
or Vb = F (p.d)
This shows that the breakdown voltage of a uniform field gap is a unique function of the
product of gas pressure and the gap length for a particular gas and electrode material. This relation is
known as Paschen’s law. This relation does not mean that the breakdown voltage is directly
proportional to product pd even though it is found that for some region of the product pd the relation is
linear i.e., the breakdown voltage varies linearly with the product pd. The variation over a large range
is shown in Fig. 1.6.
A
l
l
J
N
T
U
W
o
r
l
d

Fig. 1.6 Paschen’s law curve
Let us now compare Paschen‘s law and the Townsend‘s criterion for spark potential. We draw the
experimentally obtained relation between the ionization coefficient α/p and the field strength f(E/p)
FEbI
for a given gas. Fig. 1.7. Here point G J represents the onset of ionization.
H p Kc
Fig. The relation between Townsend’s criterion for spark = k and Paschen’s criterion
Now the Townsend‘s criterion αd = K can be re-written as
α . V  K or α  K . E
p Ep p V P
This is equation to a straight line with slope equal to K/V depending upon the value of K. The
higher the voltage the smaller the slope and therefore, this line will intersect the ionization curve at
two points e.g., A and B in Fig. Therefore, there must exist two breakdown voltages at a constant
pressure (p = constant), one corresponding to the small value of gap length i.e., higher E (E = V/d) i.e.,
point B and the other to the longer gap length i.e., smaller E or smaller E/p i.e., the point A. At low
values of voltage V the slope of the straight line is large and, therefore, there is no intersection between
the line and the curve 1. This means no breakdown occurs with small voltages below Paschen‘s mini-
mum irrespective of the value of pd. The point C on the curve indicates the lowest breakdown voltage
or the minimum sparking potential. The spark over voltages corresponding to points A, B, C are shown
in the Paschen‘s curve in Fig.
A
l
l
J
N
T
U
W
o
r
l
d

The fact that there exists a minimum sparking potential in the relation between the sparking
potential and the gap length assuming p to be constant can be explained quantitatively by considering
the efficiency of ionization of electrons traversing the gap with different electron energies. Assuming
that the Townsend‘s second ionization coefficient ν is small for values pd > (pd)min., electrons cross-
ing the gap make more frequent collision with the gas molecules than at (pd)min. but the energy
gained between the successive collision is smaller than at (pd). Hence, the probability of ionization is
lower unless the voltage is increased. In case of (pd) < (pd) min., the electrons cross the gap without
making any collision and thus the sparking potential is higher. The point (pd)min., therefore,
corresponds to the highest ionization efficiency and hence minimum sparking potential.
An analytical expression for the minimum sparking potential can be obtained using the general
expression for α/p.
αAeBp/ E
or α  pAe
− Bpd /Vb
p
− Bpd / Vb
pA 1 e
Bpd / Vb
or e = α or α  pA
1 e
B
pd
/V
or d . 
b
αd pA
F
1
I
We know that αd = 1n H1  K
ν
d 
e
Bpd /Vb
1n
F
1
1 I
Therefore, pA H ν K
F
Assuming ν to be constant, let
1 I
1n H1  K
 k
ν
d 
e Bpd / Vb
K
Then
pA
In order to obtain minimum sparking potential, we rearrange the above expression as
Vb = f(pd)
Taking 1n on both sides, we have
Bpd  ln Apd
Vb K
or Vb =
Bpd
ln Apd / k
Differentiating Vb w.r. to pd and equating the derivative to zero
ln
Apd
. B − Bpd .
K
.
A
B ln Apd
dVb Apd K B
K

K
= −  0
d ( pd) F Apd I 2
F Apd I 2
F Apd I 2
H1n K H1n K H1n K
K K K
or
1 1
ln Apd F Apd I2
K Hln K K
A
l
l
J
N
T
U
W
o
r
l
d

or 1n Apd = 1
K
or 1n Apd = e
K
or
e K
(pd)
min
=
A
or V = B eK / A
 B . eK
b min 1 A
1 I
B F
Vbmin = 2.718 1n H1  K
A ν
If values of A, B and ν are known both the (pd) min and Vbmin can be obtained. However, in
practice these values are obtained through measurements and values of some of the gases are given in
the following Table 1.1.
Table Minimum Sparking Constant for various gases
Gas (pd)min Vb min volts
Air 0.55 352
Nitrogen 0.65 240
Hydrogen 1.05 230
SF6 0.26 507
CO2 0.57 420
O
2 0.70 450
Neon 4.0 245
Helium 4.0 155
Typical values for A, B and ν for air are A = 12, B = 365 and ν = 0.02.
Schuman has suggested a quadratic formulation between α/p and E/p under uniform field over
a wide but restricted range as
L E FE I O2
α
s  C M − G J P
p N p
H p K Q
where (E/p)c is the minimum value of E at which the effective ionization begins and p is the pressure,
C a constant.
We know that Townsend‘s spark criterion for uniform fields is αd = k where k = (1 + 1/ν).
Therefore, the equation above can be re-written as
K L E F E I O2
 C M − G J P
dp M p
H
p K
c P
N Q
F K 1 I1 / 2 E FE I
or G . J  − G J
p
H C pd K H p Kc
E F E I F K / C I1 / 2
or G J G J
p H p K c H pd K
A
l
l
J
N
T
U
W
o
r
l
d

V F E I F K /C I1 / 2

G J 
G J
dp H p Kc H pd K
or
V
b
F E I
pd 
F K
I 1/ 2
 G J
H c K
. pd
H p Kc
Sohst and Schröder have suggested values for Ec = 24.36 kV/cm K/C = 45.16 (kV)
2
/cm for air
at p = 1 bar and temperature 20°C.
Substituting these values in the above equation, we have
Vb = 6.72 pd + 24.36 (pd) kV
The breakdown voltages suggested in tables or obtained through the use of empirical relation
normally correspond to ambient temperature and pressure conditions, whereas the atmospheric air
provides basic insulation between various electrical equipments. Since the atmospheric conditions
(Tem-perature, pressure) vary widely from time to time and from location to location, to obtain the
actual breakdown voltage, the voltage obtained from the STP condition should be multiplied by the air
den-sity correction factor. The air density correction factor is given as
δ = 3.92 b
273  t
where b is the atmospheric pressure in cm of Hg and t the temperature in °C.
.
A
l
l
J
N
T
U
W
o
r
l
d

Breakdown in Electronegative Gases
SF6, has excellent insulating strength because of its affinity for electrons (electronegativity) i.e.,
when-ever a free electron collides with the neutral gas molecule to form negative ion, the electron is
absorbed by the neutral gas molecule. The attachment of the electron with the neutral gas molecule
may occur in two ways:
SF6 + e → SF6
–
SF6 +
e → SF5
–
+ F
The negative ions formed are relatively heavier as compared to free electrons and, therefore,
under a given electric field the ions do not attain sufficient energy to lead cumulative ionization in the
gas. Thus, these processes represent an effective way of removing electrons from the space which
otherwise would have contributed to form electron avalanche. This property, therefore, gives rise to
very high dielectric strength for SF6. The gas not only possesses a good dielectric strength but it has
the unique property of fast recombination after the source energizing the spark is removed.
The dielectric strength of SF6 at normal pressure and temperature is 2–3 times that of air and at
2 atm its strength is comparable with the transformer oil. Although SF6 is a vapour, it can be liquified
at moderate pressure and stored in steel cylinders. Even though SF6 has better insulating and arc-
quencling properties than air at an equal pressure, it has the important disadvantage that it can not be
used much above 14 kg/cm
2
unless the gas is heated to avoid liquifaction.
Application of Gases in Power System
The gases find wide application in power system to provide insulation to various equipments and
substations. The gases are also used in circuit breakers for arc interruption besides providing
insulation between breaker contacts and from contact to the enclosure used for contacts. The various
gases used are (i) air (ii) oxygen (iii) hydrogen (iv) nitrogen (v) CO2 and (vi) electronegative gases
like sulphur hexafluoride, arcton etc.
The various properties required for providing insulation and arc interruption are:
(i) High dielectric strength.
(ii) Thermal and chemical stability.
A
l
l
J
N
T
U
W
o
r
l
d

(iii) Non-inflammability.
(iv) High thermal conductivity. This assists cooling of current carrying conductors immersed in
the gas and also assists the arc-extinction process.
(v) Arc extinguishing ability. It should have a low dissociation temperature, a short thermal time
constant (ratio of energy contained in an arc column at any instant to the rate of energy dissipation at the
same instant) and should not produce conducting products such as carbon during arcing.
(vi) Commercial availability at moderate cost. Of the simple gases air is the cheapest and most
widely used for circuit breaking. Hydrogen has better arc extinguishing property but it has lower di-
electric strength as compared with air. Also if hydrogen is contaminated with air, it forms an explosive
mixture. Nitrogen has similar properties as air, CO2 has almost the same dielectric strength as air but
is a better arc extinguishing medium at moderate currents. Oxygen is a good extinguishing medium
but is chemically active. SF6 has outstanding arc-quenching properties and good dielectric strength.
Of all these gases, SF6 and air are used in commercial gas blast circuit breakers.
Air at atmospheric pressure is ‗free‘ but dry air costs a lot when stored at say 75 atmosphere. The
compressed air supply system is a vital part of an air blast C.B. Moisture from the air is removed by
refrigeration, by drying agents or by storing at several times the working pressure and then expanding it to
the working pressure for use in the C.B. The relative cost of storing the air reduces with increase in
pressure. If the air to be used by the breaker is at 35 kg/cm
2
it is common to store it at 210 kg/cm
2
.
Air has an advantage over the electronegative gases in that air can be compressed to extremely
high pressures at room temperature and then its dielectric strength even exceeds that of these gases.
The SF6 gas is toxic and its release in the form of leakage causes environmental problems.
Therefore, the electrical industry has been looking for an alternative gas or a mixture of SF6 with
some other gas as an insulating and arc interrupting medium. It has been observed that a suitable
mixture of SF6 with N2 is a good replacement for SF6. This mixture is not only finding acceptability
for providing insulation e.g., gas insulated substation and other equipments, it is able to quench high
current magni-tude arcs. The mixture is not only cost effective, it is less sensitive to find non-
uniformities present within the equipment. Electric power industry is trying to find optimum SF6 to
N2 mixture ratio for various components of the system viz., GIS, C.B., capacitors, CT, PT and cables.
A ratio 70% of SF6 and 30% of N2 is found to be optimum for circuit breaking. With this ratio, the
C.B. has higher recovery rate than pure SF6 at the same partial pressure. The future of using SF6 with
N2 or He for providing insulation and arc interruption is quite bright.
BREAKDOWN IN LIQUID DIELECTRICS
Liquid dielectrics are used for filling transformers, circuit breakers and as impregnants in high voltage
cables and capacitors. For transformer, the liquid dielectric is used both for providing insulation
between the live parts of the transformer and the grounded parts besides carrying out the heat from the
transformer to the atmosphere thus providing cooling effect. For circuit breaker, again besides
providing insulation between the live parts and the grounded parts, the liquid dielectric is used to
quench the arc developed between the breaker contacts. The liquid dielectrics mostly used are
petroleum oils. Other oils used are synthetic hydrocarbons and halogenated hydrocarbons and for very
high temperature applications sillicone oils and fluorinated hyrocarbons are also used.
The three most important properties of liquid dielectric are (i) The dielectric strength (ii) The
dielectric constant and (iii) The electrical conductivity. Other important properties are viscosity, ther-
mal stability, specific gravity, flash point etc. The most important factors which affect the dielectric
A
l
l
J
N
T
U
W
o
r
l
d

strength of oil are the, presence of fine water droplets and the fibrous impurities. The presence of even
0.01% water in oil brings down the dielectric strength to 20% of the dry oil value and the presence of
fibrous impurities brings down the dielectric strength much sharply. Therefore, whenever these oils
are used for providing electrical insulation, these should be free from moisture, products of oxidation
and other contaminants.
The main consideration in the selection of a liquid dielectric is its chemical stability. The other
considerations are the cost, the saving in space, susceptibility to environmental influences etc. The use
of liquid dielectric has brought down the size of equipment tremendously. In fact, it is practically
impossible to construct a 765 kV transformer with air as the insulating medium. Table 1.2. shows the
properties of some dielectrics commonly used in electrical equipments.
Table Dielectric properties of some liquids
S.No. Property Transformer Capacitor Cable Silicone
Oil Oil Oil Oil
1. Relative permittivity 50 Hz 2.2 – 2.3 2.1 2.3 – 2.6 2.7 – 3.0
2. Breakdown strength at 12 kV/mm 18 kV/mm 25 kV/mm 35 kV/mm
20°C 2.5 mm 1 min
3. (a) Tan δ 50 Hz 10
–3
2.5 × 10
–4
2 × 10
–3
10
–3
(b) 1 kHz 5 × 10
–4
10
–4
10
–4
10
–4
4. Resistivity ohm-cm 10
12
– 10
13
10
13
– 10
14
10
12
– 10
13
2.5 × 10
14
5. Maximum permissible water
content (ppm) 50 50 50 < 40
6. Acid value mg/gm of KOH NIL NIL NIL NIL
7. Sponification mg of KOH/gm 0.01 0.01 0.01 < 0.01
of oil
8. Specific gravity at 20°C 0.89 0.89 0.93 1.0–1.1
Liquids which are chemically pure, structurally simple and do not contain any impurity even in
traces of 1 in 10
9
, are known as pure liquids. In contrast, commercial liquids used as insulating liquids
are chemically impure and contain mixtures of complex organic molecules. In fact their behaviour is
quite erratic. No two samples of oil taken out from the same container will behave identically.
The theory of liquid insulation breakdown is less understood as of today as compared to the gas or
even solids. Many aspects of liquid breakdown have been investigated over the last decades but no general
theory has been evolved so far to explain the breakdown in liquids. Investigations carried out so far,
however, can be classified into two schools of thought. The first one tries to explain the break-down in
liquids on a model which is an extension of gaseous breakdown, based on the avalanche ionization of the
atoms caused by electon collisiron in the applied field. The electrons are assumed to be ejected from the
cathode into the liquid by either a field emission or by the field enhanced thermionic effect (Shottky‘s
effect). This breakdown mechanism explains breakdown only of highly pure liquid and does not apply to
explain the breakdown mechanism in commercially available liquids. It has been observed that conduction
in pure liquids at low electric field (1 kV/cm) is largely ionic due to dissocia-tion of impurities and
increases linearily with the field strength. At moderately high fields the conduc-tion saturates but at high
field (electric), 100 kV/cm the conduction increases more rapidly and thus breakdown takes place. Fig. 1.11
(a) shows the variation of current as a function of electric field for
A
l
l
J
N
T
U
W
o
r
l
d

hexane. This is the condition nearer to breakdown. However, if the figure is redrawn starting with low
fields, a current-electric field characteristic as shown in Fig. 1.11 (b) will be obtained. This curve has
three distinct regions as discussed above.
Conductioncurrent
High field
Saturation
Linear
(a) (b)
Fig. 1.11 Variation of current as a function of electric field
(a) High fields (b) Low fields
The second school of thought recognises that the presence of foreign particles in liquid
insulations has a marked effect on the dielectric strength of liquid dielectrics. It has been suggested
that the sus-pended particles are polarizable and are of higher permittivity than the liquid. These
particles experi-ence an electrical force directed towards the place of maximum stress. With uniform
field electrodes the movement of particles is presumed to be initiated by surface irregularities on the
electrodes, which give rise to local field gradients. The particles thus get accumulated and tend to
form a bridge across the gap which leads finally to initiation of breakdown. The impurities could also
be in the form of gaseous bubbles which obviously have lower dielectric strength than the liquid itself
and hence on breakdown of bubble the total breakdown of liquid may be triggered.
Electronic Breakdown
Once an electron is injected into the liquid, it gains energy from the electric field applied between the
electrodes. It is presumed that some electrons will gain more energy due to field than they would lose
during collision. These electrons are accelerated under the electric field and would gain sufficient
energy to knock out an electron and thus initiate the process of avalanche. The threshold condition for
the beginning of avalanche is achieved when the energy gained by the electron equals the energy lost
during ionization (electron emission) and is given by
e λ E = Chv
where λ is the mean free path, hv is the energy of ionization and C is a constant. Table 1.3 gives
typical values of dielectric strengths of some of the highly purified liquids.
Table 1.3. Dielectric strengths of pure liquids
Liquid Strength (MV/cm)
Benzene 1.1
Goodoil 1.0–4.0
Hexane 1.1–1.3
Nitrogen 1.6–1.88
Oxygen 2.4
Silicon 1.0–1.2
A
l
l
J
N
T
U
W
o
r
l
d

The electronic theory whereas predicts the relative values of dielectric strength satisfactorily,
the formative time lags observed are much longer as compared to the ones predicted by the electronic
theory.
Suspended Solid Particle Mechanism
Commercial liquids will always contain solid impurities either as fibers or as dispersed solid particles.
The permittivity of these solids (E1) will always be different from that of the liquid (E2). Let us
assume these particles to be sphere of radisus r. These particles get polarized in an electric field E and
experi-ence a force which is given as
F=r3
ε1
–ε2 E.
dE
ε 1  2ε 2 dx
and this force is directed towards a place of higher stress if ε1 > ε2 and towards a place of lower stress
if ε1 < ε2 when ε1 is the permittivity of gas bubbles. The force given above increases as the
permittivity of the suspended particles (ε1) increases. If ε1 → ∞
F=r3
1− ε2
/ε1 E
dE
1  2ε 2 / ε1 dx
Let ε1 → ∞
F = r
3
E .
dE
dx
Thus, the force will tend the particle to move towards the strongest region of the field. In a uniform
electric field which usually can be developed by a small sphere gap, the field is the strongest in the uniform
field region. Here dE/dx → 0 so that the force on the particle is zero and the particle remains in equilibrium.
Therefore, the particles will be dragged into the uniform field region. Since the permittivity of the particles
is higher than that of the liquid, the presence of particle in the uniform field region will cause flux
concentration at its surface. Other particles if present will be attracted towards the higher flux
concentration. If the particles present are large, they become aligned due to these forces and form a bridge
across the gap. The field in the liquid between the gap will increase and if it reaches critical value,
brakdown will take place. If the number of particles is not sufficient to bridge the gap, the particles will
give rise to local field enhancement and if the field exceeds the dielectric strength of liquid, local
breakdown will occur near the particles and thus will result in the formation of gas bubbles which have
much less dielectric strength and hence finally lead to the breakdown of the liquid.
The movement of the particle under the influence of electric field is oposed by the viscous
force posed by the liquid and since the particles are moving into the region of high stress, diffusion
must also be taken into account. We know that the viscous force is given by (Stoke‘s relation) FV =
6πnrν where η is the viscosity of liquid, r the raidus of the particle and v the velocity of the particle.
Equating the electrical force with the viscous force we have
6πηrν = r
3
E
dE
or ν =
r 2E dE
dx 6πη dx
However, if the diffusion process is included, the drift velocity due to diffusion will be given by
νd = –
D dN − KT dN
6πηr Ndx
N dx
A
l
l
J
N
T
U
W
o
r
l
d

where D = KT/6πηr a relation known as Stokes-Einstein relation. Here K is Boltzmann‘s constant and
T the absolute temperature. At any instant of time, the particle should have one velocity and, therefore,
equation v = vd
We have
– KT . dN r
2
E . dE
Ndx 6πη dx
6πη r
or KT dN  − r
2
E dE
r N
or KT 1n N  − r
2
E
2
r 2
It is clear that the breakdown strength E depends upon the concentration of particles N, radius r
of particle, viscosity η of liquid and temperature T of the liquid.
It has been found that liquid with solid impurities has lower dielectric strength as compared to
its pure form. Also, it has been observed that larger the size of the particles impurity the lower the
overall dielectric strength of the liquid containing the impurity.
Cavity Breakdown
It has been observed experimentally that the dielectric strength of liquid depnds upon the hydrostatic
pressure above the gap length. The higher the hydrostatic pressure, the higher the electric strength,
which suggests that a change in phase of the liquid is involved in the breakdown process. In fact,
smaller the head of liquid, the more are the chances of partially ionized gases coming out of the gap
and higher the chances of breakdown. This means a kind of vapour bubble formed is responsible for
the breakdown. The following processes might lead to formation of bubbles in the liquids:
(i) Gas pockets on the surface of electrodes.
(ii) Due to irregular surface of electrodes, point charge concentration may lead to corona dis-
charge, thus vapourizing the liquid.
(iii) Changes in temperature and pressure.
(iv) Dissociation of products by electron collisions giving rise to gaseous products.
It has been suggested that the electric field in a gas bubble which is immersed in a liquid of
permittivity ε2 is given by
Eb 
3E0
ε 2  2
Where E0 is the field in the liquid in absence of the bubble. The bubble under the influence of
the electric field E0 elongates keeping its volume constant. When the field Eb equals the gaseous ioni-
zation field, discharge takes place which will lead to decomposition of liquid and breakdown may
follow.
A more accurate expression for the bubble breakdown strength is given as
1 R 2πσ (2ε 2
ε 1
) Lπ V OU1 / 2
E
b 
|
M
b |
ε − ε
S
r 4 2rE
− 1PV
2 1 | M 0 P|
T N QW
A
l
l
J
N
T
U
W
o
r
l
d

where σ is the surface tension of the liquid, ε2 and ε1 are the permittivities of the liquid and bubble,
respectively, r the initial radius of the bubble and Vb the voltage drop in the bubble. From the expres-
sion it can be seen that the breakdown strength depends on the initial size of the bubble which of
course depends upon the hydrostatic pressure above the bubble and temperature of the liquid. Since
the above formation does not take into account the production of the initial bubble, the experimental
values of breakdown were found to be much less than the calculated values. Later on it was suggested
that only incompressible bubbles like water bubbles can elongate at constant volume according to the
simple gas law pV = RT. Such a bubble under the influence of electric field changes its shape to that of
a prolate spheroid and reaches a condition of instability when β, the ratio of the longer to the shorter
diameter of the spheroid is about 1.85 and the critical field producing the instability will be given by
Ec = 600
π σ L ε 2 O
ε
M
ε 2 − ε1
− G P H
2
r
N Q
1 Lβ cosh −1
β O
where G = M − 1P
β
2
− 1 (β
2
−
1/ 2
N 1) Q
F 1I
and H
2
= 2β1/3
G 2 β − 1 − J
β
2
H K
For transformer oil ε2 = 2.0 and water globule with r = 1 m, σ = 43 dynes/cm, the above
equation gives Ec = 226 KV/cm.
Electroconvection Breakdown
It has been recognized that the electroconvection plays an important role in breakdown of insulating fluids
subjected to high voltages. When a highly pure insulating liquid is subjected to high voltage, electrical
conduction results from charge carriers injected into the liquid from the electrode surface. The resulting
space charge gives rise to coulombic forces which under certain conditions causes hydro-dynamic
instability, yielding convecting current. It has been shown that the onset of instability is asso-ciated with a
critical voltage. As the applied voltage approaches the critical voltage, the motion at first exhibits a
structure of hexagonal cells and as the voltage is increased further the motion becomes turbulent. Thus,
interaction between the space charge and the electric field gives rise to forces creating an eddy motion of
liquid. It has been shown that when the voltage applied is near to breakdown value,
the speed of the eddy motion is given by νe = ε 2 /ρ where ρ is the density of liquid. In liquids, the
ionic drift velocity is given by
νd =
KE where K is the mobility of ions.
Let M ν e  ε 2/ KE
ν d ρ
The ratio M is usually greater than unity and sometimes much greater than unity (Table 1.4).
A
l
l
J
N
T
U
W
o
r
l
d

Table
Medium Ion ε M
Air NTP O– 1.0 2.3 × 10
–2
2
Ethanol Cl
–
2.5 26.5
Methanol H+ 33.5 4.1
Nitrobenzene Cl
–
35.5 22
Propylene Carbonate Cl
–
69 51
Transformer Oil H
+
2.3 200
Thus, in the theory of electroconvection, M plays a dominant role. The charge transport will be
largely by liquid motion rather than by ionic drift. The criterion for instability is that the local flow
velocity should be greater than drift velocity.
TREATMENT OF TRANSFORMER OIL
Even though new synthetic materials with better mechanical and thermal properties are being
developed, the use of oil/paper complex for high voltages is still finding applications. Oil, besides
being a good insulating medium, it allows better dispersion of heat. It allows transfer and absorption
of water, air and residues created by the ageing of the solid insulation. In order to achieve operational
requirements, it must be treated to attain high degree of purity.
Whatever be the nature of impurities whether solid, liquid or gaseous, these bring down the
dielectric strength of oil materially. Oil at 20°C with water contents of 44 ppm will have 25% of its
normal dielectric strength. The presence of water in paper not only increases the loss angle tan δ, it
accelerates the process of ageing. Similarly, air dissolved in oil produces a risk of forming bubble and
reduces the dielectric strength of oil.
Air Absorption: The process of air absorption can be compared to a diffusing phenomenon in
which a gaseous substance in this case air is in contact with liquid (oil here).
If the viscosity of the liquid is low, the convection movements bring about a continuous inter-
mixing whereby a uniform concentration is achieved. This phenomenon can, for example, be checked
in a tank where the air content or the water content measured both at the top and the bottom are
approximately equal.
Let G(t) = Air content of the oil after time t
Gm = Air content under saturation condition
p = Probability of absorption per unit time
S = Surface of oil
V = Volume of oil
The absorption of air by oil can be given by the equation
dG
 p .
S
[ Gm − G (
t)] dt V
with boundary condition at t = 0 G = G0
A
l
l
J
N
T
U
W
o
r
l
d

Solving the above equation
dG  p S dt
Gm − G ( t) V
or 1n {G – G (t)} = – p S t  A
m V
At t = 0 G = G0. Therefore, A = + 1n ← {Gm – G0}
or 1n Gm − G ( t)  − pS t
Gm − G0 V
or G
m
– G (t) = (G – G ) e -pSt/V
m 0
Fig. (a) shows the schematic for the measurement of air absorption by insulating oil which has
previously been degassed as a function of the absorption time. The oil is degassed and dried with the
help of the vacuum pump (1) and then introduced into the installation until the desired pressure is
reached. A part of this air is absorbed by the oil, the pressure being maintained at a constant (2) Value
by reducing the volume in absorption meter (3) Thus, air content of oil by volume can be measured.
Precision manometer (4) is used to calibrate the absorption meter. Phosphorus pentaoxide trap (5)
takes in the remainder of the water vapour.
In case of a completely degassed oil i.e., at t = 0 G = 0, we obtain
G(t) = Gm (1 – e
–pSt/V
)
To have an estimate of air absorbed by oil, let us consider a hermetically sealed bushing
impregnated under vacuum contains 20 litres of degassed oil (G0 = 0). Suppose the bushing is opened
at 25°C and remains under atmospheric pressure for 10 hours, the oil surface S = 10
3
cm
2
. Assume a
typical value of p = 0.4 cm/hr, the percentage of air absorbed in given as
G (10 hr) = 10(1 − e−10 3 / 10 4  0.4 10
)
1
2
5
4
6
3
(a)
A
l
l
J
N
T
U
W
o
r
l
d

5 6
4
3
1
2
8 7 9 10
(b)
Fig. (a), (b)
The molecules of oil are held together by their internal binding energy. In order that water
molecule takes the place of oil molecule and is dissolved in the mixture, it is necessary to provide this
molecule with a quantity of energy E in the form of heat.
Let N be the number of oil molecule n, the number of water molecules.
Pn the number of possibilities of combination for n water molecules among (N + n) molecules.
i.e.,
where
and
Pn  N  n !
N ! n!
S = Entropy of the oil
T = Absolute temperature of mixture
K = Boltzamann‘s constant
E(T) = Energy required for a water molecule to take the palce of an oil molecule
W = Water content of the oil (p.p.m.)
Wm = Maximum water content of the oil, at saturation point
Thermal equilibrium will be reached when free energy F is minimum i.e.,
∂F  0
∂n
F = E (T) – TS
S = k 1n Pn
Now ∂F  ∂E (T ) − T ∂S  0
∂n ∂n ∂n
Since P =( N  n)!
n
N ! n!
Taking 1n both sides, we have
1n Pn = 1n (N + n)! – 1n N ! – 1n n !
= 1n (N + n) + 1n (N + n – 1) + 1n (N + n – 2)
... – 1n N ! – 1n n – ln (n – 1) – 1n (n – 2) – 1n (n – 3) ...
A
l
l
J
N
T
U
W
o
r
l
d

Differentiating both sides,
1 ∂Pn  1
Pn ∂n N  n
≈
1
N  n
Since ∂S  K ∂Pn
∂n Pn ∂n
Substituting for ∂S/∂n in the equation
 1  ... −
N  n − 1
−
1
= –
N
n n (N  n)
≈ –
KN
n (N  n)
1 − 1 ...
n n − 1
We have,
or
Since
Therefore,
∂E (T ) − T ∂S  0
∂n ∂n
∂E (T )  T TKN  0
∂n n (N  n)
∂E  − TKN
∂n n (N  n)
∂E = – TK
N dn
n (N  n)
N >> N + n ≈ N
dn
∂E = – TK n
or E = – TK ln n + A
when E = 0, n = N. Therefore, 0 = – TK ln N + A or A = TK ln N
or E = TK ln N
n
or ln n − E or n ≈e− E / TK
N TK N
or n = N e
–E/TK
≈ W
m
Following impurities should be considered for purification of oil (i) solid impurities (ii) free
and dissolved water particles (iii) dissolved air. Some of the methods used to remove these impurities
have been described below.
Filtration and Treatment Under Vacuum: Different types of filters have been used. Filter press with
soft and hard filter papers is found to be more suitable for insulating oil. Due to hygroscopic
properties of the paper, oil is predried before filtering. Therefore, this oil can not be used for high
voltage insulation. The subsequent process of drying is carried out in a specially, designed tank under
vacuum. The oil is distributed over a large surface by a so-called ‗‗Rasching-ring‘‘ degassing column.
Through this process, both the complete drying and degassing are achieved simultaneously. By
suitable selection of the various components of the plant e.g., rate of flow of oil, degassing surface,
vacuum pump etc., a desired degree of purity can be obtained.
Fig. (b) shows a typical plant for oil treatment. The oil from a transformer or a storage tank is
prefiltered (1) so as to protect the feeder pump (2). In (3), the oil is heated up and is allowed to flow
A
l
l
J
N
T
U
W
o
r
l
d

through filter press (4) into degassing tank (5). The degassing tank is evacuated by means of vacuum
pump (6) whereas the second vacuum pump (7) is either connected with the degassing tank in parallel
with pump (6) or can be used for evacuating the transformer tank which is to be treated.
The operating temperature depends upon the quality and the vapour pressure of oil. In order to
prevent an excessive evaporation of the aromatics, the pressure should be greater than 0.1 Torr. The
filteration should be carried out at a suitable temperature as a higher temperature will cause certain
products of the ageing process to be dissolved again in the oil.
Centrifugal Method: This method is helpful in partially extracting solid impurities and free water. It is
totally ineffective as far as removal of water and dissolved gases is concerned and oil treated in this
manner is even over-saturated with air as air, is thoroughly mixed into it during the process. However,
if the centrifugal device is kept in a tank kept under vacuum, partial improvement can be obtained. But
the slight increase in efficiency of oil achieved is out of proportion to the additional costs involved.
Adsorption Columns: Here the oil is made to flow through one or several columns filled with an
adsorbing agent either in the form of grains or powder. Following adsorbing agents have been used:
(i) Fuller earth
(ii) Silica gel
(iii) Molecular sieves
Activated Fuller earths absorb carbonyl and hydroxyl groups which from the principal ageing
products of oil and small amount of humidity. Best results of oil treatment are obtained by a combina-
tion of Fuller earth and subsequent drying under vacuum.
Silica gel and in particular molecular sieves whose pore diameter measures 4 Å show a strong
affinity for water. Molecular sieves are capable of adsorbing water 20% of its original weight at 25°C
and water vapour pressure of 1 Torr whereas sillica gel and Fuller earth take up 6 and 4 per cent
respectively.
Molecular sieves are synthetically produced Zeolites which are activated by removal of the
crystallisation water. Their adsorption capacity remains constant upto saturation point. The
construction of an oil drying plant using molecular sieves is, therefore, simple. The plant consists of
an adsorption column containing the sieves and of an oil circulating pump.
The adsorption cycle is followed by a desorption cycle once the water content of the sieves has
exceeded 20 per cent. It has been found that the two processes adsorption and desorption are readily
reversible. In order to attain disorption of the sieves, it is sufficient to dry them in air stream of 200°C.
Electrostatic Filters: The oil to be treated is passed between the two electrodes placed in a container.
The electrostatic field charges the impurities and traces of water which are then attracted and retained
by the foam coated electrodes. This method of drying oil is found to be economical if the water
content of the oil is less than 2 ppm. It is, therefore, essential that the oil is dried before hand if the
water content is large. Also, it is desirable that the oil flow should be slow if efficient filtering is
required. Therefore, for industrial application where large quantity of oil is to be filtered, large number
of filters will have to be connected in parallel which may prove uneconomical.
TESTING OF TRANSFORMER OIL
The oil is poured in a container known as test-cell which has internal dimensions of 55 mm × 90 mm ×
100 mm high. The electrodes are polished spheres of 12.7 to 13 mm diameter, preferably of brass,
arranged horizontally with their axis not less than 40 mm above the bottom of the cell. For the test, the
A
l
l
J
N
T
U
W
o
r
l
d

distance between the spheres shall be 4 + 0.02 mm. A suitable gauge is used to adjust the gap. While
preparing the oil sample, the test-cell should be thoroughly cleaned and the moisture and suspended
particles should be avoided. Fig. 1.13 shows an experimental set-up for finding out the dielectric
strength of the given sample of oil. The voltmeter is connected on to the primary side of the high
voltage transformer but calibrated on the high voltage side.
Fig. 1.13
The gap between the spheres is adjusted to 4 mm with the help of a gauge and the spheres are
immersed in oil to a depth as mentioned earlier. The voltage is increased gradually and continuously
till a flash over of the gap is seen or the MCB operates. Note down this voltage. This voltage is known
as rapidly-applied voltage. The breakdown of the gap has taken place mainly due to field effect. The
thermal effect is minimal as the time of application is short.
Next bring the voltage back to zero and start with 40% of the rapidly applied voltage and wait
for one minute. See if the gap has broken. If not, increase the voltage everytime by 2.1/2% of the
rapidly applied voltage and wait for one minute till the flash over is seen or the MCB trips. Note down
this voltage.
Start again with zero voltage and increase the voltage to a value just obtained in the previous
step and wait for a minute. It is expected that the breakdown will take place. A few trials around this
point will give us the breakdown value of the dielectric strength. The acceptable value is 30 kV for 4
mm applied for one minute. In fact these days transformer oils with 65 kV for 4 mm 1 minute value
are available. If it is less than 30 kV, the oil should be sent for reconditioning. It is to be noted that if
the electrodes are immersed vertically in the oil, the dielectric strength measured may turn out to be
lower than what we obtained by placing the electrodes in horizontal position which is the normal
configura-tion. It is due to the fact that when oil decomposes carbon particles being lighter rise up and
if the electrodes are in vertical configuration, these will bridge the gap and the breakdown will take
place at a relatively lower value.
Application of Oil in Power Apparatus
Oil is normally used for providing insulation between the live parts of different phases and between
phases and the grounded enclosure containing the oil and the main parts of the apparatus. Also it
provides cooling effect to the apparatus placed within the enclosure. Besides providing insulation, the
oil helps the C.B. to quench the arc produced between the breaker contacts when they begin to
separate to eliminate the faulted section from the healthy section of the system.
A
l
l
J
N
T
U
W
o
r
l
d

In an oil circuit breaker, the heat of the oil decomposes the oil which boils at 658 K. The gases
liberated are approx. (i) Hydrogen, 70%, (ii) Acetylene, 20%, (iii) Methane, 5% and (iv) Ethane, 5%.
(the abbreviation for these gases could be used as HAME).
The temperature about the arc is too high for the three last-named gases to exist and the arc
itself runs into a mixture of hydrogen, carbon and copper vapour at temperature above 6000 K. The
hydrogen being a diatomic gas gets dissociated into the atomic state which changes the characteristics
of the arc on account of its associated change in its thermal conductivity. The outcome of this is that
the discharge suddenly contracts and acquires an appreciably higher core temperature. In certain cases,
the thermal ionization may be so great that the discharge runs with a lower voltage which may stop the
ionization due to the electric field strength. The transition from the field ionization to thermal
ionization is most marked in hydrogen and, therefore, in oil circuit breakers.
The separation of the C.B. contacts which are carrying current gives rise to an arc without
changing much the current wave form. Initially when the contacts just begin to separate the magnitude
of current is very large but the contact resistance being very small, a small voltage appears across
them. But the distance of separation being very very small, a large voltage gradient is set up which is
good enough to cause ionization of the particles between the contacts. Also it is known that with the
copper contacts which are generally used for the circuit breakers very little thermal ionization can
occur at temperature below the melting point. For effective field emission, the voltage gradient
required is 10
6
V/cm. From this it is clear that the arc is initiated by the field emission rather than the
thermal ioniza-tion. This high voltage gradient exists only for a fraction of a micro-second. But in this
short period, a large number of electrons would have been liberated from the cathode and these
electrons while reach-ing anode, on their way would have collided with the atoms and molecules of
the gases. Thus, each emitted electron tends to create others and these in turn derive energy from the
field and multiply. In short, the work done by the initially-emitted electrons enables the discharge to
be maintained. Finally, if the current is high, the discharge attains the form of an arc having a
temperature high enough for thermal ionization, which results in lower voltage gradient. Thus, an arc
is initiated due to field effect and then maintained due to thermal ionization.
A
l
l
J
N
T
U
W
o
r
l
d

UNIT-III
BREAKDOWN IN SOLID DIELECTRICS
Introduction:
Solid insulating materials are used almost in all electrical equipments, be it an electric heater or a 500
MW generator or a circuit breaker, solid insulation forms an integral part of all electrical equipments
especially when the operating voltages are high. The solid insulation not only provides insulation to
the live parts of the equipment from the grounded structures, it sometimes provides mechanical
support to the equipment. In general, of course, a suitable combination of solid, liquid and gaseous
insulations are used.
The processes responsible for the breakdown of gaseous dielectrics are governed by the rapid
growth of current due to emission of electrons from the cathode, ionization of the gas particles and
fast development of avalanche process. When breakdown occurs the gases regain their dielectric
strength very fast, the liquids regain partially and solid dielectrics lose their strength completely.
The breakdown of solid dielectrics not only depends upon the magnitude of voltage applied but
also it is a function of time for which the voltage is applied. Roughly speaking, the product of the
breakdown voltage and the log of the time required for breakdown is almost a constant i.e.,
Vb = 1n tb = constant
Fig. 1.14. Variation of Vb with time of application
The dielectric strength of solid materials is affected by many factors viz. ambient temperature,
humidity, duration of test, impurities or structural defects whether a.c., d.c. or impulse voltages are
being used, pressure applied to these electrodes etc. The mechanism of breakdown in solids is again
less understood. However, as is said earlier the time of application plays an important role in break-
down process, for discussion purposes, it is convenient to divide the time scale of voltage application
into regions in which different mechanisms operate. The various mechanisms are:
(i) Intrinisic Breakdown
(ii) Electromechanical Breakdown
(iii) Breakdown Due to Treeing and Tracking
(iv) Thermal Breakdown
(v) Electrochemical Breakdown
Intrinsic Breakdown
If the dielectric material is pure and homogeneous, the temperature and environmental conditions
suitably controlled and if the voltage is applied for a very short time of the order of 10
–8
second, the
dielectric strength of the specimen increases rapidly to an
A
l
l
J
N
T
U
W
o
r
l
d

upper limit known as intrinsic dielectric strength.
The intrinsic strength, therefore, depends mainly
upon the structural design of the material i.e., the
Fig. Specimen designed for intrinsic breakdown
material itself and is affected by the ambient
temperature as the structure itself might change slightly by temperature condition. In order to obtain
the intrinsic dielectric strength of a material, the samples are so prepared that there is high stress in the
centre of the specimen and much low stress at the corners as shown in Fig.
The intrinsic breakdown is obtained in times of the order of 10
–8
sec. and, therefore, has been
considered to be electronic in nature. The stresses required are of the order of one million volt/cm. The
intrinsic strength is generally assumed to have been reached when electrons in the valance band gain
sufficient energy from the electric field to cross the forbidden energy band to the conduction band. In
pure and homogenous materials, the valence and the conduction bands are separated by a large energy
gap at room temperature, no electron can jump from valance band to the conduction band.
The conductivity of pure dielectrics at room temperature is, therfore, zero. However, in
practice, no insulating material is pure and, therefore, has some impurities and/or imperfections in
their structural designs. The impurity atoms may act as traps for free electrons in energy levels that lie
just below the conduction band is small. An amorphous crystal will, therefore, always have some free
electrons in the conduction band. At room temperature some of the trapped electrons will be excited
thermally into the conduction band as the energy gap between the trapping band and the conduction
band is small. An amorphous crystal will, therefore, always have some free electrons in the
conduction band. As an electric field is applied, the electrons gain energy and due to collisions
between them the energy is shared by all electrons. In an amorphous dielectric the energy gained by
electrons from the electric field is much more than they can transfer it to the lattice. Therefore, the
temperature of electrons will exceed the lattice temperature and this will result into increase in the
number of trapped electrons reaching the conduction band and finally leading to complete breakdown.
When an electrode embeded in a solid specimen is subjected to a uniform electric field, breakdown
may occur. An electron entering the conduction band of the dielectric at the cathode will move towards the
anode under the effect of the electric field. During its movement, it gains energy and on collision it loses a
part of the energy. If the mean free path is long, the energy gained due to motion is more than lost during
collision. The process continues and finally may lead to formation of an electron avalanche similar to gases
and will lead finally to breakdown if the avalanche exceeds a certain critical size.
Electromechanical Breakdown
When a dielectric material is subjected to an electric field, charges of opposite nature are induced on
the two opposite surfaces of the material and hence a force of attraction is developed and the
speciment is subjected to electrostatic compressive forces and when these forces exceed the
mechanical withstand strength of the material, the material collapses. If the initial thickness of the
material is d0 and is compressed to a thickness d under the applied voltage V then the compressive
stress developed due to electric field is
2
F =
1
ε 0 ε r
V
2
2 d
where εr is the relative permittivity of the specimen. If γ is the Young‘s modulus, the mechanical
compressive strength is
γ 1n
d0
d
Equating the two under equilibrium condition, we have
1
ε0
εr
V
2
γ 1n
d0
2 d 2 d
2 2 2γ d 0 2 d0
A
l
l
J
N
T
U
W
o
r
l
d

or V = d . ε 0
ε r 1n d = Kd 1n d
Differentiating with respect to d, we have
dV L d0 2 d d0 O
2V  K M2d 1n − d . . P = 0
dd d d0 d
2
N Q
or 2d ln
d
0
= d
d
or ln
d
0 1
=
d 2
or
d
 0.6
d0
For any real value of voltage V, the reduction in thickness of the specimen can not be more
than 40%. If the ratio V/d at this value of V is less than the intrinsic strength of the specimen, a further
increase in V shall make the thickness unstable and the specimen collapses. The highest apparent
strength is then obtained by substituting d = 0.6 d0 in the above expressions.
V 2γ V L γ O1 / 2

ε 0ε r
1n 1.67 or  Ea  0.6 M P
d d0 ε
N 0
εr Q
The above equation is approximate only as γ depends upon the mechanical stress. The
possibility of instability occuring for lower average field is ignored i.e., the effect of stress
concentration at irregu-larities is not taken into account.
Breakdown due to Treeing and Tracking
We know that the strength of a chain is given by the strength of the weakest link in the chain. Similarly
whenever a solid material has some impurities in terms of some gas pockets or liquid pockets in it the
dielectric strength of the solid will be more or less equal to the strength of the weakest impurities. Suppose
some gas pockets are trapped in a solid material during manufacture, the gas has a relative permittivity of
unity and the solid material εr, the electric field in the gas will be εr times the field in the solid material. As
a result, the gas breaks down at a relatively lower voltage. The charge concentration here in the void will
make the field more non-uniform. The charge concentration in such voids is found to be quite large to give
fields of the order of 10 MV/cm which is higher than even the intrinsic breakdown. These charge
concentrations at the voids within the dielectric lead to breakdown step by step and finally lead to complete
rupture of the dielectric. Since the breakdown is not caused by a single discharge channel and assumes a
tree like structure as shown in Fig. 1.6, it is known as breakdown due to treeing. The treeing phenomenon
can be readily demonstrated in a laboratory by applying an impulse voltage between point plane electrodes
with the point embedded in a transparent solid dielectric such as perspex.
The treeing phenomenon can be observed in all dielectric wherever non-uniform fields prevail.
Suppose we have two electrodes separated by an insulating material and the assembly is placed in an
outdoor environment. Some contaminants in the form of moisture or dust particles will get deposited
on the surface of the insulation and leakage current starts between the electrode through the
contaminants say moisture. The current heats the moisture and causes breaks in the moisture films.
These small films then act as electrodes and sparks are drawn between the films. The sparks cause
carbonization and volatilization of the insulation and lead to formation of permanent carbontracks on
the surface of insulations. Therefore, tracking is the formation of a permanent conducting path usually
carbon across the surface of insulation. For tracking to occur, the insulating material must contain
organic substances. For this reason, for outdoor equipment, tracking severely limits the use of
insulation having organic substances. The rate of tracking can be slowed down by adding filters to the
A
l
l
J
N
T
U
W
o
r
l
d

polymers which inhibit carbonization.
Fig.
Thermal Breakdown
When an insulating material is subjected to an electric field, the material gets heated up due to
conduc-tion current and dielectric losses due to polarization. The conductivity of the material
increases with increase in termperature and a condition of instability is reached when the heat
generated exceeds the heat dissipated by the material and the material breaks down. Fig. 1.17 shows
various heating curves corresponding to different electric stresses as a function of specimen
temperature. Assuming that the temperature difference between the ambient and the specimen
temperature is small, Newton‘s law of cooling is represented by a straight line.
Fig. Thermal stability or instability of different fields
A
l
l
J
N
T
U
W
o
r
l
d

The test specimen is at thermal equilibrium corresponding to field E1 at temperature T 1 as be-
yond that heat generated is less than heat lost. Unstable equilibrium exists for field E2 at T2, and for
field E3 the state of equilibrium is never reached and hence the specimen breaks down thermally.
Fig. 1.18. Cubical speciman—Heat flow
In order to obtain basic equation for studying thermal breakdown, let us consider a small cube (Fig.
1.18) within the dielectric specimen with side ∆x and temperature difference across its faces in the direction
of heat flow (assume here flow is along x-direction) is ∆T. Therefore, the temperature gradient is
∆T ≈ dT
∆x dx
Let ∆x
2
= A. The heat flow across face 1
KA
dT
Joules
dx
Heat flow across face 2
d FdT I
dT
KA – KA H K ∆x
dx dx
dx
Here the second term indicates the heat input to the differential specimen. Therefore, the heat
absorbed by the differential cube volume
d FdT I
KA H K ∆x d F dT I

dx dx
 K
∆V dx Hdx K
The heat input to the block will be partly dissipated into the surrounding and partly it will raise
the temperature of the block. Let CV be the thermal capacity of the dielectric, σ the electrical
conductivity, E the electric field intensity. The heat generated by the electric field = σE
2
watts, and
suppose the rise in temperature of the block is ∆T, in time dt, the power required to raise the
temperature of the block by ∆T is
dT
C
watts
V
dt
dT d FdT I 2
Therefore,
C
V  K H K  σE
dt dx dx
The solution of the above equation will give us the time required to reach the critical
temperature Tc for which thermal instability will reach and the dielectric will lose its insulating
properties. However, unfortunately the equation can be solved in its present from CV, K and σ are all
functions of temperature and in fact σ may also depend on the intensity of electrical field.
Therefore, to obtain solution of the equation, we make certain practical assumptions and we
consider two extreme situations for its solution.
A
l
l
J
N
T
U
W
o
r
l
d

Case I: Assume that the heat absorbed by the block is very fast and heat generated due to the electric
field is utilized in raising the temperature of the block and no heat is dissipated into the surroundings.
We obtain, therefore, an expression for what is known as impulse thermal breakdown. The main equa-
tion reduces to
CV
dT
= σE
2
dt
The objective now is to obtain critical field strength Ec which will generate sufficient heat very
fast so that above requirement is met. Let
FE
c I
E = G J t
tc
H K
i.e., the field is a ramp function
dT dT dE
2 .
σE
= C
V dt
= C
V dEdt
and Let σ = σ e–u/KT
0
where K is Boltzamann‘s constant and σ0 is the conductivity at ambinent temperature T0.
Substituting these values in the simplified equation, we have
σ 0 e −u / KT
E
2
 CV
dE
.
dT
dE
dt
Now
dE

Ec
dt
t
c
Therefore, σ e–u/KT E
2
= C Ec dT
0 V tc dE
or σ E
2 tc dE = C eu/KTdT
0
E
c V
σ 0
t
c
E
c
T
c
u
2 KT
or z0 E dE

zT0 e dT
C
V
E
c
The integral on the left hand side
σ0
t
c
E
c 2
σ0
t
c 1 3 1
σ0 2
z0 E dE  . Ec = tc Ec
C E C E 3 3 C
V c V c V
The integral on the right hand side
Tc
u
K
2 u / KT
zT0
e
KT
dt → T e 0
0
u
when Tc >> T0
3C KT 2
e
u/KT
0
Therefore, E =
V
.
0
c
σ0
t
c u
From the above expression, it is clear that the critical condition requires a combination of criti-
cal time and critical field. However, the critical field is independent of the critical temperature due to
the fast rise in temperature.
A
l
l
J
N
T
U
W
o
r
l
d

Case II: Here we assume that the voltage applied is the minimum voltage for indefinite time so that
the thermal breakdown takes place. For this, we assume that we have a thick dielectric slab that is sub-
jected to constant ambient temperature at its surface by using sufficiently large electrodes as shown in
Fig.
Fig. Arrangement of electrode and specimen for minimum thermal B.D. voltage
Suppose that minimum voltage is applied which brings thermal breakdown. As a result after
some time, a temperature distribution will be set up within the specimen with maximum temperature
Tm at its centre and it decreases as we approach the surface.
In order to calculate maximum thermal voltage, let us consider a point inside the dielectric at a
distance x from the central axis and let the voltage and temperature at the point are Vx and Tx, respec-
tively. We further assume that all the heat generated in the dielectric will be carried away to its sur-
roundings through the electrodes. Therefore, neglecting the term
CV
dT
dt
the main equation reduces to
d
F dT I 2
HK K  σE
dx dx
Now using the relations σ E = J and E = – ∂V
∂x
d F dT I ∂V
We have K H K = – J ∂x
dx dx
Integrating both the sides w.r. to x
x d F dT I V
x ∂V
We have z0 H K
K dx = – J
z0 ∂x dx
dx dx
or K dT  −
JV
x = – σEV = – σV dv
x
dx x
dx
Let σ = σ e
–u/KT
.
0
We have
K
dT σ e−u / KT V ∂V
dx 0 x
∂x
or K E
u / KT
dT  Vx ∂V dx
σ 0 ∂x
A
l
l
J
N
T
U
W
o
r
l
d

K Tc
V
m/2
or zT0 eu / KT
dT 
z0 Vx dV
σ 0
This shows that the maximum thermal voltage depends upon the critical temperature Tc at the
centre of dielectric at which the specimen loses is insulating properties. However, Vm is independent
of the thickness of the insulating material but for thin specimens the thermal breakdown becomes
touch-ing asymptotically to a constant value for thick specimen. Under alternating currents the total
heat generated will be
σE
2
+ V
2
ωc tan δ
and, therefore, this being higher than what we have in d.c. circuits, the maximum thermal breakdown
voltage will be lower in a.c. supplies. In fact, higher the frequency the lower the thermal breakdown
voltage.
Table 1.5 gives for thick specimen, thermal breakdown values for some dielectric under a.c.
and d.c. voltages at 20°C.
Table 1.5. Thermal breakdown voltage
Material Maximum thermal voltage in
MV/cm
d.c. a.c.
Ceramics HV Steatite — 9.8
LF Steatite — 1.5
High grade porcelain 2.8
Organic materials Ebonite — 1.45–2.75
Polythene 3.5
Polystyrene 5.0
Polystyrene at 1 MHz 0.05
Acrylic resins 0.3–1.0
Crystals Mica muscovite 24 7–18
Rock salt 38 1.4
Quartz Perpendiculars to axis 12000 —
Paralle to axis 66 —
Impure — 2.2
Electrochemical Breakdown
Whenever cavities are formed in solid dielectrics, the dielectric strength in these solid specimen de-creases.
When the gas in the cavity breaks down, the surfaces of the specimen provide instantaneous anode and
cathode. Some of the electrons dashing against the anode with sufficient energy shall break the chemical
bonds of the insulation surface. Similarly, positive ions bombarding against the cathode may increase the
surface temperature and produce local thermal instability. Similarly, chemical degra-dation may also occur
from the active discharge products e.g., O3, NO2 etc. formed in air. The net effect of all these processes is a
slow erosion of the material and a consequent reduction in the thickness of the specimen. Normally, it is
desired that with ageing, the dielectric strength of the specimen should not decrease. However, because of
defects in manufacturing processes and/or design, the dielectric strength
A
l
l
J
N
T
U
W
o
r
l
d

decreases with time of voltage application or even without voltage application and in many cases; the
decrease in dielectric strength (Eb) with time follows the following empirical relation.
t Eb
n
 constant
where the exponent n depends upon the dielectric material, the ambient temperature humidity and the
quality of manufacture. This is the main reason why high a.c. voltage testing is not recommended. In
fact, these days very low frequency testing is being suggested (0.1 HZ) which simulates the effects of
both a.c. 50 HZ and d.c. voltages and yet the dielectric strength of the specimen is not affected much
with VLF voltage application.
The breakdown of solid dielectric due to internal discharges or partial discharges has been
elaborately explained in section 6.9 of the book.
Solid Dielectrics Used in Power Apparatus
The main requirements of the insulating materials used for power apparatus are:
1. High insulation resistance
2. High dielectric strength
3. Good mechanical properties i.e., tenacity and elasticity
4. It should not be affected by chemicals around it
5. It should be non-hygroscopic because the dielectric strength of any material goes very much
down with moisture content
Vulcanized rubber : Rubber in its natural form is highly insulating but it absorbs moisture readily and
gets oxidized into a resinous material; thereby it loses insulating properties. When it is mixed with
sulphur alongwith other carefully chosen ingredients and is subjected to a particular temperature it
changes into vulcanized rubber which does not absorb moisture and has better insulating properties
than even the pure rubber. It is elastic and resilient.
The electrical properties expected of rubber insulation are high breakdown strength and high
insulation resistance. In fact the insulation strength of the vulcanized rubber is so good that for lower
voltages the radial thickness is limited due to mechanical consideration.
The physical properties expected of rubber insulation are that the cable should withstand nor-
mal hazards of installation and it should give trouble-free service.
Vulcanized rubber insulated cables are used for wiring of houses, buildings and factories for
low-power work.
There are two main groups of synthetic rubber material : (i) general purpose synthetics which
have rubber-like properties and (ii) special purpose synthetics which have better properties than the
rubber e.g., fire resisting and oil resisting properties. The four main types are: (i) butyl rubber, (ii)
silicon rubber, (iii) neoprene, and (iv) styrene rubber.
Butyl rubber: The processing of butyl rubber is similar to that of natural rubber but it is more
difficult and its properties are comparable to those of natural rubber. The continuous temperature to which
butyl rubber can be subjected is 85°C whereas for natural rubber it is 60°C. The current rating of
A
l
l
J
N
T
U
W
o
r
l
d

butyl insulated cables is approximately same as those of paper or PVC insulated cables. Butyl rubber
compound can be so manufactured that it has low water absorption and offers interesting possibilities
for a non-metallic sheathed cable suitable for direct burial in the ground.
Silicone rubber: It is a mechanically weak material and needs external protection but it has
high heat resistant properties. It can be operated at temperatures of the order of 150°C. The raw
materials used for the silicon rubber are sand, marsh gas, salt, coke and magnesium.
Neoprene: Neoprene is a polymerized chlorobutadiene. Chlorobutadiene is a colourless liquid
which is polymerized into a solid varying from a pale yellow to a darkish brown colour. Neoprene
does not have good insulating properties and is used upto 660 V a.c. but it has very good fire resisting
properties and therefore it is more useful as a sheathing material.
Styrene rubber: Styrene is used both for insulating and sheathing of cables. It has properties
almost equal to the natural rubber.
Polyvinyl Chloride (PVC)
It is a polymer derived generally from acetylene and it can be produced in different grades
depending upon the polymerization process. For use in cable industry the polymer must be
compounded with a plasticizer which makes it plastic over a wide range of temperature. The grade of
PVC depends upon the plasticizer. PVC is inferior to vulcanized in respect of elasticity and insulation
resistance. PVC material has many grades.
General purpose type: It is used both for sheathing and as an insulating material. In this com-
pound monomeric plasticizers are used. It is to be noted that a V.R. insulated PVC sheathed cable is
not good for use.
Hard grade PVC: These are manufactured with less amount of plasticizer as compared with
general purpose type. Hard grade PVC are used for higher temperatures for short duration of time like
in soldering and are better than the general purpose type. Hard grade can not be used for low continu-
ous temperatures.
Heat resisting PVC: Because of the use of monomeric plasticizer which volatilizes at tempera-
ture 80°C–100°C, general purpose type compounds become stiff. By using polymeric plasticizers it is
possible to operate the material continuously around 100°C.
PVC compounds are normally costlier than the rubber compounds and the polymeric
plasticized compounds are more expensive than the monomeric plasticized ones. PVC is inert to
oxygen, oils, alkalis and acids and, therefore, if the environmental conditions are such that these things
are present in the atmosphere, PVC is more useful than rubber.
Polythene
This material can be used for high frequency cables. This has been used to a limited extent for
power cables also. The thermal dissipation properties are better than those of impregnated paper and
the impulse strength compares favourably with an impregnated paper-insulated device. The maximum
operating temperature of this material under short circuits is 100°C.
Cross-linked polythene: The use of polythene for cables has been limited by its low melting point.
By cross-linking the molecules, in roughly the same way as vulcanising rubber, a new material is produced
which does not melt but carbonizes at 250 to 300°C. By using chemical process it has been
A
l
l
J
N
T
U
W
o
r
l
d

made technically possible to cross-link polythene in conventional equipment for the manufacture of
rubber. This is why the product is said to be ―vulcanised‖ or ―cross-linked‖ polythene.
The polythene is inert to chemical reactions as it does not have double bonds and polar groups.
Therefore, it was thought that polythene could be cross-linked only through special condition, e.g., by
irradiating polythene with electrons, thereby it could be given properties of cross-linking such as
change of tensile strength and better temperature stability. Many irradiation processes have been
developed in the cable making industry even though large amounts of high energy radiations are
required and the procedure is expensive.
Polythene can also be irradiated with ultraviolet light, after adding to it a smal quantity of ultra-
violet sensitive material such as benzophenone. Under the influence of ultraviolet light on
benzophenone, a radical is formed of the same type as in the decomposition of peroxide by the radical
mechanism. Organic peroxides have also been used successfully to crosslink the polythene.
Impregnated paper
A suitable layer of the paper is lapped on the conductor depending upon the operating voltage.
It is then dried by the combined application of heat and vacuum. This is carried out in a hermetically
sealed steam heated chamber. The temperature is 120°–130°C before vacuum is created. After the
device is dried, an insulating compound having the same temperature as that of the chamber is forced
into the chamber. All the pores of the paper are completely filled with this compound. After impregna-
tion the device is allowed to cool under the compound so that the void formation due to compound
shrinkage is minimized.
In case of pre-impregnated type the papers are dried and impregnated before they are applied
on the conductor.
The compound used in case of impregnated paper is a semifluid and when the cables are laid
on gradients the fluid tends to move from higher to lower gradient. This reduces the compound
content at higher gradients and may result in void formation at higher gradients. This is very serious
for cables operating at voltages higher than 3.3 kV. In many cases, the failures of the cables have been
due to the void formation at the higher levels or due to the bursting of the sheath at the lower levels
because of the excessive internal pressure of the head of compound.
Insulating press boards. If the thickness of paper is 0.8 mm or more, it is called paper board.
When many layers of paper are laminated with an adhesive to get desired thickness, these are known
as press boards and are used in bushings, transformers as insulating barriers or supporting materials.
The electrical properties of press boards varies depending upon the resin content. The application of
these press boards depends upon the thickness and density of paper used. For high frequency
capacitors and cables usually low density paper (0.8 gm/cm
3
) is used where medium density paper is
used for power capacitors and high density papers are used in d.c. machines and energy storage
capacitors. The electric strength of press board is higher than that of resins or porcelain. However, it is
adversely affected by temperature above 20°C. The loss angle tan δ also decreases with increase in
temperature. The main advantage of this material is that it provides good mechanical support even at
higher temperatures upto 120°C.
Mica. Mica consists of crystalline mineral silicates of alumina and potash. It has high dielectric
strength, low dielectric losses and good mechanical strength. All these properties make it useful for
A
l
l
J
N
T
U
W
o
r
l
d

many electrical devices e.g., commutator segment separator, aremature windings, electrical heating
and cooling equipments and switchgear. Thin layers of mica are laminated with a suitable resin or
varnish to make thick sheets of mica. Mica can be mixed with the required type of resin to obtain its
application at different operating temperatures. Mica is used as a filler in insulating materials to im-
prove their dielectric strength, reduce dielectric loss and improve heat resistance property.
Ceramics. Ceramics materials are produced from clay containing aluminium oxide and other
inorganic materials. The thick parts of these substances is given the desired shape and form at room
temperature and then baked at high temperature about (1450°C) to provide a solid inelastic final struc-
ture. Ceramics also known as porcelain in one of its forms have high mechanical strength and low
permittivity (εr < 12) are widely used for insulators and bushings. These have 40% to 50% of clay, 30-
20% of aluminium oxide and 30% of fieldspar. The ceramics with higher permittivity (εr > 12) are
used in capacitors and transducers.
The specific insulation resistance of ceramics is comparatively low. The tan δ of these
materials is high and increases with increase in temperature resulting in higher dielectric loss. The
breakdown strength of porecelain compared to other insulating material is low but it remains
unaffected over a wide range of temperature variation. Porcelain is chemically insert to alkalies and
acids and, therefore, corrosion resistant and does not get contaminated. Alumina (Al2O3) has replaced
quartz because of its better thermal conductivity, insulating property and mechanical strength. It is
used for the fabrication of high current vacuum circuit breakers.
Glass. Glass is a thermoplastic inorganic material consisting of silicondioxide (SiO2), which is
available in nature in the form of quartz. Different types of metal oxides could be used for producing
different types of glasses but for use in electrical engineering only non-alkaline glasses are suitable
having alkaline content less than 0.8%.
The dielectric constant of glass varies between 3.6 and 10.0 and the density varies between
2000 kg/m
3
and 6000 kg/m
3
. The loss angle tan δ is less than 10
–3
and losses are higher for lower
frequen-cies. Its dielectric strength varies between 300 and 500 kV/mm and it decreases with increase
in tem-perature. Glass is used for X-ray equipments, electronic valves, electric bulbs etc.
Epoxy Resins. Epoxy resins are low molecular but soluble thermosetting plastics which
exhibit sufficient hardening quality in their molecules. The chemical cross-linking of epoxy resins is
normally carried out at room temperatures either by a catalytic mechanism or by bridging across
epoxy molecule through the epoxy or hydroxyl group.
Epoxy resins have high dielectric and mechanical strength. They can be cast into desired
shapes even at room temperature. They are highly elastic and it is found that when it is subjected to a
pressure of 175000 psi, it returned to its original shape after the load is removed. The dielectric
constant varies between 2.5 and 4.0. Epoxy resins basically being non-polar substances have high dc
specific insula-tion resistance and low loss tan δ compared to polar materials like PVC. However,
when the tempera-ture exceeds 100°C the specific insulation resistance begins to decrease
considerably and tan δ in-creases. Compared to porcelain the breakdown strength of epoxy resin is
almost double at temperatures upto 100°C but decreases rapidly at higher temperatures.
As filler materials, the inorganic substances like quartz powder (SiO2) are used for casting
applications. In SF6 gas insulated systems having epoxy resin spacers, aluminium oxide and also dolo-
A
l
l
J
N
T
U
W
o
r
l
d

mite are used as filler materials. These are found to be more compatible to the decomposed products
of SF6 by partial discharge and arcing discharges.
It is to be noted that the cast or encapsulation should not contain voids or humidity especially in
high voltage applications and the material is desired to be homogeneous. It is, therefore, desirable to dry
and degas the individual components of the mixture and casting is preferably carried out in vacuum.
The epoxy resins casts are inert to ether, alcohol and benzol. However, most of them are
soluble in mineral oils at about 70°C. It is for this reason that they are not found suitable for
applications in filled transformers.
There are certain application which require insulating materials to operate between a high range
of temperature e.g., –270°C to 400°C. Some of the applications are space shuttle solar arrays, capaci-
tors, transformers high speed locomotive, microprocessor chip carriers, cryogenic cables and other
applications at cryogenic temperatures. For this some thermoplastic polymer films are used which
have unique combination of electrical, mechanical and physical quantities and these materials are able
to retain these properties over a wide range of temperatures where other insulating materials may fail.
Perfluoro carbon films have high dielectric strength very low dielectric constant of 2 and low dielectric
loss of 2 × 10
–4
at 100 Hz and 7.5 × 10
–4
at 100 MHz. These films are used under extreme
conditions of temperature and environment. These films are used for insulation on high temperature
wires, cables, motor coils phase and ground insulation and for capacitors. This is also used as a
substrate for flexible printed circuits and flexible cables.
Another insulating film in which has the best thermal properties in this category of insulating
materials is polyimide film under the trade name of Kapton manufactured by DuPont of America.
These films can be used between a very wide range of temperature variation varying between –270°C
and 350°C. Its continuous temperature rating is 240°C. It has high dielectric and tensile strength. The
disadvantages of the film are
(i) high moisture absorption rate and (ii) it is affected by alkalies and strong inorganic acids.
Kepton films can be used capacitors, transformers formed coil insulation, motor state insulation
and flexible printed circuits. The film is selectively costlier and is mainly used where its unique
charac-teristics makes it the only suitable insulation. The use of this insulation for motors reduces the
overall dimensions of the motors for the same ratings. It is, therefore, used in almost all situations
whose space is a serious problem and the other nature insulation result in bigger dimension.
Another recently developed resins is poly carbonate (PC) which is good heat resistant; it is
flexible and has good dielectric characteristic. It is not affected by oils, fats and dilute acids but is
adversely affected by alkalies, esters and aromatic hydrocarbons. The film being cost effective and fast
resistant, it is used for coil insulation, slot insulation for motors and for capacitor insulation. This is
known as the lexon polymer.
General Electric Co. of USA has developed a film under the trade name Ultem which is a poly
etherimine (PEI) film which has dielectric strength comparable to that of polyimide film and has
higher thermal conductivity and lower moisture absorption and is relatively less costlier. It is used as
insula-tion for transformers and motors.
A
l
l
J
N
T
U
W
o
r
l
d

Application of Insulating Materials
Insulating fluids (gases and liquids) provide insulation between phases and between phase and
grounded parts of electrical equipments. These also carry out heat from the windings of the electrical
equipments. However, solid insulating materials are used only to provide insulation only.
International Electrotechincal Commission has categories various insulating materials depend-
ing upon the temperature of operations of the equipments under the following categories.
Class Y 90°C Natural rubber, PVC, paper cotton, silk without impregnation.
Class A 105°C Same as class Y but impregnated
Class E 120°C Polyethylene, terephthalate, cellulose tricetrate, polyvinyl acetate enamel
Class B 130°C Bakelite, bituminised asbestos, fibre glass, mica, polyester enamel
Class F 155°C As class B but with epoxy based resin
Class H 180°C As class B with silicon resin binder silicone rubber, aromatic polyamide
(nomex paper and fibre), polyimide film (enamel, varnish and film) and estermide
enamel
Class C Above 180°C, as class B but with suitable non-organic binders, teflon and other high
temperature polymers.
While describing the dielectric and other properties of various insulating materials, their appli-
cation for various electrical apparatus has also been mentioned in the previous paragraphs. However, a
reverse process i.e., what insulating materials are used for a particular apparatus depending upon its
ratings and environmental condition where the apparatus is required to operate, is also desirable and a
brief review is given here.
Power Transformers. For small rating, the coils are made of super-enamelled copper wire. For
layer to layer, coil to coil and coil to ground (iron core) craft paper is used.
However, for large size transformers paper or glass tape is rapped on the rectangular
conductors whereas for coil to coil or coil to ground, insulation is provided using thick radial spacers
made of press board or glas fibre.
In oil-filled transformers, the transformer oil is the main insulation. However between various
layers of low voltage and high voltage winding oil-impregnated press boards are placed.
SF6 gas insulated power transformers make use of sheet aluminium conductors for windings
and turn to turn insulation is provided by a polymer film. The transformer has annular cooling ducts
through which SF6 gas circulates for cooling the winding. SF6 gas provides insulations to all major
gaps in the transformer. This transformer is used where oil filled transform is not suitable e.g., in
cinema halls, high rise buildings and some especial circumstances: The end turns of a large power
transformer are provided with extra insulation to avoid damage to coil when lighting or switching
surges of high frequency are incident on the transformer winding.
The terminal bushings of large size power transformer are made of condenser type bushing. The
terminal itself consists of a brass rod or tube which is wound with alternate layers of treated paper and tin
foil, so proportioned, as to length, that the series of condensers formed by the tin foil cylinders and the
intervening insulation have equal capacitances, thereby the dielectric stress is distributed uniformly.
Circuit Breakers. The basic construction of any circuit breaker requires the separation of con-
tacts in an insulating fluid which serves two functions here:
(i) It extinguishes the arc drawn between the contacts when the CB, opens.
(ii) It provides adequate insulation between the contacts and from each contact to earth.
A
l
l
J
N
T
U
W
o
r
l
d

Many insulating fluids are used for arc extinction and the fluid chosen depends upon the rating
and type of C.B. The insulating fluids commonly used for circuit breakers are
(i) Air at atmospheric pressure: Air break circuit breaker upto 11 kV.
(ii) Compressed air (Air blast circuit breaker between 220 kV and 400 kV)
(iii) Mineral oil which produces hydrogen for arc extrictrion (transformer oil)
(a) Plain break oil, C.B. 11 kV–66 kV
(b) Controlled break oil C.B. or bulk oil C.B. between 66 kV–220 kV
(c) Minimum oil C.B. between 66 kV and 132 kV.
(iv) Ultra high vacuum C.B. upto 33 kV.
(v) SF6 circuit breakers above 220 kV.
The controlled break and minimum oil circuit breakers enclose the breaker contacts in an
arcing chamber made of insulating materials such as glassfibre reinforced synthetic resins etc.
Rotating Machines. For low voltage a.c. and d.c. machines, the winding wire are super
enamelled wire and the other insulation used are vulcanised rubber and varnished cambric and paper.
For high voltage and large power capacity machines, the space limitations demand the use of
insulating materials having substantially greater dielectric strength. Mica is considered to be a good
choice not only due to space requirements but because of its ability to withstand higher temperatures.
However, the brittleness of mica makes it necessary to build up the required thickness by using thin
flakes cemented together by varnish or bakelite generally with a backing of thin paper or cloth and
then baking it under pressure. Epoxy resin bounded mica paper is widely used for both low and high
voltage machines. Multilayer slot insulation is made of press board and polyester film. However, for
machines with high operating temperatures kapton polymide is used for slot insulation. Mica has
always been used for stator insulations. In addition to mica, conducting non-woven polyesters are used
for corona protection both inside and at the edges of the slots. Glass fibre reinforced epoxy wedge
profiles are used to provide support between the winding bars, slots and the core laminations.
Power Cables. The various insulating materials used are vulcanised rubber, PVC,
Polyethylene and impregnated papers.
Vulcanised rubber, insulated cables are used for wiring of houses, buildings and factories for
low power work.
PVC is inert to oxygen, oils, alkalies and acids and therefore, if the environmental conditions
are such that these things are present in the atmosphere, PVC is more useful than rubber.
Polyethylene is used for high frequency cables. This has been used to a limited extent for
power cables also. The thermal dissipation properties are better than those of impregnated poper. The
maxi-mum oprating temperature of this cable under short circuits is 100°C.
In case of impregnated paper, a suitable layer of the paper is lapped on the conductor
depending upon the operating voltage. It is then dried by the combined application of heat and
vacuum. The compound used in case of impregnated paper is semifluid and when the cables are laid
on gradients the fluid tends to move from higher to lower gradients which reduces the compound
content at higher gradients and may result in void formation at higher gradients. For this reason,
impregnated paper cables are used upto 3.3 kV.
Following methods are used for elimination of void formation in the cables:
(i) The use of low viscosity mineral oil for the impregnation of the dielectric and the inclusion
of oil channels so that any tendency of void formation (due to cyclic heating and cooling of
impregnate) is eliminated.
A
l
l
J
N
T
U
W
o
r
l
d

(ii) The use of inert gas at high pressure within the metal sheath and indirect contact with the
dielectric.
Because of the good thermal characteristics and high dielectric strength of the gas SF6, it is used for
insulating the cables also. SF6 gas insulated cables can be matched to overhead lines and can be operated
corresponding to their surge impedance loading. These cables can be used for transporting thousands of
MVA even at UHV, whereas the conventional cables are limited to 1000 MVA and 500 kV.
Power Capacitors. Capacitor design economics suggests the use of individual unit assembled
in appropriate series and parallel connected groups to obtain the desired bank voltage and reactive
power ratings both in shunt and series capacitor equipments. Series capacitor duty usually requires
that a unit designated for a series application be more conservatively rated than a shunt unit. However,
there is no basic difference in the construction of the two capacitors.
The most commonly used capacitor for the purpose is the impregnated paper capacitor. This
consists of a pair of aluminium foil electrodes separated by a number of Kraft paper tissues which are
impregnated with chlorinated diphenyl and has a higher permittivity and results in reduction in the
quantity of materials required for a given capacitance and the cost.
The working stress of an impregnated paper is 15 to 25 V/ and papers of thickness 6–12 are
available and hence depending upon the operating voltage of the capacitor, a suitable thickness of the
paper can be selected. Because of imperfection involved in the manufacturing process of the dielectric
paper it is desirable to use at least two layers of tissues between metal foils so that the possibility of
coincidence of weak spots is avoided.
The effective relative permittivity depends upon the paper and the impregnant. For chlorinated
diphenyl impregnant the relative permittivity lies between 5 and 6. Normally through past experience,
the area of the plate for a particular material of paper and impregnant per microfarad of capacitance is
known and hence it is possible to obtain the number of turns of paper to be wound on a given diameter
of mandrel for a specified foil width and for the particular lay-up of foil and paper.
The method of laying up the paper and metallic foil and the connection of lugs is shown in Fig.
1.20. Two layers of dielectric are used as without it rolling would short circuit the plates. As a result of
this, two capacitors in parallel are formed by the roll. The foil and the paper interleaved in this fashion
are wound on to a mandrel which is split to allow easy removal of the finished roll. If the section of
the container is same as that of the roll, minimum overall value for the capacitor is obtained. As a
result of this, quantity of free impregnant is a minimum thereby the risk of leakage of impregnant with
variation in temperature is reduced. Sometimes a high resistance (for discharge) is connected across
the termi-nals of the capacitor for safety reasons.
Terminal
tapes
Foils
Fig. Impregnated paper capacitor-terminal tape type
A
l
l
J
N
T
U
W
o
r
l
d

The replacement of linen by the Kraft paper and oil by askarel made it possible to have indi-
vidual unit ratings upto 15 kVAr by 1930. After making some costly refinements in basic
paper/askarel dielectric 100 kVAr rating capacitor were manufactured by 1960.
General Electric Company designed a 150 kVAr unit using a paper/poly propylene film/askarel
dielectric.
Further advances in the manufacture of dielectric materials led to single unit of 600 kVAr even
though the rating of a single unit based on economy ranges between 200 and 300 kVAr. Replacement
of askarel with non-PCB fluids did not have much effect on unit sizes or ratings. The newer all
polypropylene film dielectric units offer distinct advantages in reduced losses and probability of case
rapture as well as improvement in unit ratings. The large size units have made it possible to reduce the
physical equip-ment size and the site area requirements.
With further development, it has now been possible to have series and shout capacitor rating
upto 550 kV and bank rating of upto 800 MVAr. The average price of smaller units in terms of 100
kVAr is Rs. 100 per kVAr or $2 per kVAr. It is to be noted that aluminium foil are used in these
capacitors as it has high thermal and electrical conductivity, has high tensile strength, high melting
point, is light in weight, low cost and is easily available.
Capacitor Bushings. Capacitor bushing is used for the terminals of high voltage transformers
and switch gears. The power conductor is insulated from the flange by a capacitor bushing consisting
of some dielectric material with metal foils cylinderical sheaths of different lengths and radii
embedded in it as shown in Fig. 1.21 thus splitting up what essentially a capacitor having high voltage
conductor and flange as it‘s plates, into a number of capacitors in series.
Power
Conductor
Metal Foil
Cylinders
Dielectric
(Varnish Paper)
Flange
Fig. Capacitor bushing
A
l
l
J
N
T
U
W
o
r
l
d

The capacitance of the capacitors formed by the metal foil cylinders is given by
εl
C =
2 ln
R
2
R
1
where l is the axial length of the capacitor R1 and R2 are the radii of its cylinderical plates. If these
capacitor have the same capacitance, the potential difference between their plates will be equal. The
equal capacitance between different layers is made possible by choosing suitable axial length together
with ratio
R2
. With this strategy the potential gradient in the dielectric is uniform but the edges of the
R
1
foil sheets lie on a curve, thus giving unequal surfaces of dielectric between the edges of successive
sheets. This is undesirable as this would result into flashovers by ―Creeping‖ along the surface. How-
ever, if the differences between the lengths of successive sheets are made equal, the radial stress is not
uniform and hence a compromise between the two conditions is usually adopted.
There are three types of papers used as insulating materials for capacitor bushings; oil impreg-
nated paper, resin bonded paper and resin impregnated paper. The oil impregnated paper bushing is
made by wrapping untreated paper after inserting foil sheets at the appropriate position and then im-
pregnating with transformer oil after vacuum drying. Before impregnation, it is ensured that moisture
and air voids are avoided. This bushing can work at a radial stress of 40 kV/cm.
In case of resin impregnated bushing creped paper tape is wrapped round the conductor and
then dried in an autoclave under controlled heat and vacuum. Epoxyresin is then sprayed to fill the
winding. The permissible radial stress in this case is 30 kV/cm.
In case of resin bonded paper bushing, the paper is first coated with epoxyresin and wrapped
round a cylinderical form under heat and pressure after inserting foil sheets at appropriate position.
The permissible radial stress in this case in 20 kV/cm.
BREAKDOWN IN VACUUM
A vacuum system is one in which the pressure maintained is at a value below the atmospheric pressure
and is measured in terms of mm of mercury. One standard atmospheric pressure at 0°C is equal to 760
mm of mercury. One mm of Hg pressure is also known as one torr after the name of Torricelli who
was the first to obtain pressures below atmosphere, with the help of mercury barometer. Sometimes
10
–3
torr is known as one micron. It is now possible to obtain pressures as low as 10
–8
torr.
In a Townsend type of discharge, in a gas, the mean free path of the particles is small and electrons
get multiplied due to various ionization processes and an electron avalanche is formed. In a vacuum of the
order of 10
–5
torr, the mean free path is of the order of few metres and thus when the electrodes are
separated by a few mm an electron crosses the gap without any collision. Therefore, in a vacuum, the
current growth prior to breakdown can not take place due to formation of electron ava-lanches. However, if
it could be possible to liberate gas in the vacuum by some means, the discharge could take place according
to Townsend process. Thus, a vacuum arc is different from the general class of low and high pressure arcs.
In the vacuum arc, the neutral atoms, ions and electrons do not come from the medium in which the arc is
drawn but they are obtained from the electrodes themselves by
A
l
l
J
N
T
U
W
o
r
l
d

evaporating its surface material. Because of the large mean free path for the electrons, the dielectric
strength of the vacuum is a thousand times more than when the gas is used as the interrupting medium.
In this range of vacuum, the breakdown strength is independent of the gas density and depends only
on the gap length and upon the condition of electrode surface. Highly polished and thoroughly
degassed electrodes show higher breakdown strength. Electrodes get roughened after use and thus the
dielectric strength or breakdown strength decreases which can be improved by applying successive
high voltage impulses which of course does not change the roughened surface but removes the loosely
adhering metal particles from the electrodes which were deposited during arcing. It has been observed
that for a vacuum of 10
–6
torr, some of the metals like silver, bismuth-copper etc. attain their
maximum break-down strength when the gap is slightly less than 3 mm. This property of vacuum
switches permits the use of short gaps for fast operation.
Electric Discharge in Vacuum
The electric discharge in vacuum results from the neutral atoms, ions and electrons emitted from the
electrodes themselves. Cathode spots are formed depending upon the current flowing. For low currents a
highly mobile cathode spot is formed and for large currents a multiple number of cathode spots are formed.
These spots constitute the main source of vapour in the arc. The processes involved in drawing the
discharge will be due to high electric field between the contacts or resistive heating produced at the point of
operation or a combination of the two. The cathode surfaces, normally, are not perfectly smooth but have
many micro projections. Due to their small area of cross-section, the projec-tions will suffer explosive
evaporation by resistive heating and supply sufficient quantity of vapour for the arc formation. Since in
case of vacuum, the emission occurs only at the cathode spots and not from the entire surface of the
cathode, the vacuum discharge is also known as cold cathode discharge. In cold cathode the emission of
electrons could be due to any of the combinations of the following mecha-nisms: (i) Field emission; (ii)
Thermionic emission; (iii) Field and Thermionic emission; (iv) Second-ary emission by positive ion
bombardment; (v) Secondary emission by photons; and (vi) Pinch effect.
The stability of discharge in vacuum depends upon: (i) the contact material and its vapour pres-
sure, and (ii) circuit parameters such as voltage, current, inductance and capacitance. It has been ob-
served that higher the vapour pressure at low temperature the better is the stability of the discharge.
There are certain metals like Zn, Bi which show these characteristics and are better electrode materials
for vacuum breakers. Besides the vapour pressure, the thermal conductivity of the metal also affects
the current chopping level. A good heat conducting metal will cool its surface faster and hence its
elec-trode surface temperature will fall which will result into reduction in evaporation rate and arc will
be chopped because of insufficient vapour. On the other hand, a bad heat conductor will maintain its
temperature and vaporization for a longer time and the arc will be more stable.
The process of multiplication of charged particles by the process of collision is very small in
the space between the electrode in vacuum, electron avalanche is not possible. If somehow a gas cloud
could be formed in vacuum, the usual kind of breakdown process can take place. This is the line of
action adopted by the researchers to study mechanism of breakdown in vacuum. By finding the way,
gas cloud could be created in a vacuum.
Non-metallic Electron Emission Mechanism
The pre-breakdown conduction current in vacuum normally originates from a nonmetallic electrode
surface. These are present in the form of insulating/semiconducting oxide layer on the surfaces or as
impurities in the electrode material. These microinclusions present in the electrode surface can
produce strong electron emission and significantly reduce the break down strength of the gap.
A
l
l
J
N
T
U
W
o
r
l
d

Even when a vacuum system is completely sealed off, the electrode surfaces may still get con-
taminated. It has been observed that when glass is heated to ‗its‘ working temperature for sealing the
electrodes into a closed container, fluxes are vaporised from the glass which get deposited in the cool
inner surfaces in the form of spherical particles upto a m diameter . Therefore, the surface of a sealed
electrode may have on its surface contaminates e.g., sodium, potassium, boron aluminium and silicon.
When an electric field is applied across such electrodes the oxides adsorbates and dust particles, then
undergo chemical changes e.g., oxides and adsorbates undergo chemical reactions which are initiated
by photons, electrons and ions and thus these contaminants limit the maximum field intensity for the
following reasons:
(i) The adsorbates and dust enhance the field emission of electrons.
(ii) The oxides adsorbates and dust particles enhance the secondary electron emission.
(iii) The oxides adsorbates and dust particles exhibit stimulated desorption of molecules and
ions under the impact of electrons, protons or ions.
Due to these mechanism, there is increase in electron emission process and therefore, more
electric field energy is converted into kinetic energy of electron and ions which leads to an increase in
surface energy of the metal. Thus, the electric strength of the gap may reduce to a level as low as 10
kV/ cm as compared to 10
4
kV/cm which is required for the field emission process
Clump Mechanism
The vacuum breakdown mechanism based on this theory makes following assumption:
(i) A loosely bound particle known as clump exists on one of the electrode surfaces.
(ii) When a high voltage is applied between the two electrodes, this clump gets charged and
subsequently gets detached from the mother electrode and is attracted by the other electrode.
(iii) The breakdown occurs due to a discharge in the vapour or gas released by the impact to the
particle at the opposite electrode.
It has been observed that for a certain vacuum gap if frequent recurrent electric breakdowns are
carried out, the withstand voltage of the gap increases and after certain number of breakdown, it
reaches an optimum maximum value. This is known as conditioning of electrodes and is of paramount
importance from practical reasons. In this electrode conditioning, the microemission sites are
supposed to have been destroyed.
Various methods for conditioning the electrodes have been suggested. Some of these are
(i) To treat the electrodes by means of hydrogen glow discharge. This method gives more
consistent results.
(ii) Allowing the pre-breakdown currents in the gap to flow for some time or to heat the elec-
trodes in vacuum to high temperature.
(iii) Treating the electrodes with repeated spark breakdown. This method is however quite time
consuming.
The area of electrodes for breakdown of gases, liquids, solids or vacuum plays an important
role. It has been observed that if the area of electrodes is increased for the same gap distance in
uniform field, the breakdown voltages are reduced.
Effect of Pressure on Breakdown Voltage
It has been observed that in case of very small gaps of less than a mm and the gas pressure between
the gap lies in the range 10
–9
to 10
–2
Torr, there is no change in the breakdown voltage i.e., if the gap
length is small a variation of gas pressure in the range given above doesn‘t affect the breakdown
voltage. However, if the gap length is large say about 20 cm, the variation of gas pressure between the
gap adversely affects the withstand voltage and the withstand voltage lowers drastically.
A
l
l
J
N
T
U
W
o
r
l
d

UNIT-IV
GENERATION OF HIGH D.C AND A.C VOLTAGES AND CURRENTS
Introduction:
There are various applications of high d.c. voltages in industries, research medical sciences
etc. HVDC transmission over both overhead lines and underground cables is becoming more and more
popular. HVDC is used for testing HVAC cables of long lengths as these have very large capacitance
and would require very large values of currents if tested on HVAC voltages. Even though D.C. tests
on A.C. cables is convenient and economical, these suffer from the fact that the stress distribution
within the insulating material is different from the normal operating condition. In industry it is being
used for electrostatic precipitation of ashing in thermal power plants, electrostatic painting, cement
industry, communication systems etc. HVDC is also being used extensively in physics for particle
acceleration and in medical equipments (X-Rays).
The most efficient method of generating high D.C. voltages is through the process of rectifica-
tion employing voltage multiplier circuits. Electrostatic generators have also been used for generating
high D.C. voltages.
According to IEEE standards 4-1978, the value of a direct test voltage is defined by its
arithematic mean value Vd and is expressed mathematically as
Vd 
1
z0
T
υa t fdt (2.1)
T
where T is the time period of the voltage wave having a frequency f = 1/T. Test voltages generated
using rectifiers are never constant in magnitude. These deviate from the mean value periodically and
this deviation is known as ripple. The magnitude of the ripple voltage denoted by δV is defined as half
the difference between the maximum and minimum values of voltage i.e.,
δV = 1
[V
max
−V
min
]
(2.2)
2
and ripple factor is defined as the ratio of ripple magnitude to the mean value Vd i.e., δV/Vd. The test
voltages should not have ripple factor more than 5% or as specified in a specific standard for a
particu-lar equipment as the requirement on voltage shape may differ for different applications.
HALF-WAVE RECTIFIER CIRCUIT
The simplest circuit for generation of high direct voltage is the half wave rectifier shown in Fig. 2.1
Here RL is the load resistance and C the capacitance to smoothen the d.c. output voltage.
A
l
l
J
N
T
U
W
o
r
l
d

If the capacitor is not connected, pulsating d.c. voltage is obtained at the output terminals whereas
with the capacitance C, the pulsation at the output terminal are reduced. Assuming the ideal transformer
and small internal resistance of the diode during conduction the capacitor C is charged to the maximum
voltage Vmax during conduction of the diode D. Assuming that there is no load connected, the d.c. voltage
across capacitance remains constant at Vmax whereas the supply voltage oscillates between
Vmax and during negative half cycle the potential of point A becomes – Vmax and hence the diode
must be rated for 2Vmax. This would also be the case if the transformer is grounded at A instead of B
as shown in Fig. 2.1 (a). Such a circuit is known as voltage doubler due to Villard for which the
output voltage would be taken across D. This d.c. voltage, however, oscillates between zero and
2Vmax and is needed for the Cascade circuit.
A i ( t ) v
max
iL
D
v (t)
1 C RL vd
B T
(a) (b)
V
max V
min
E 2 v
F
t
1
t
(c)
Fig. (a) Single Phase rectifier (b) Output voltage without C (c) Output voltage with C
If the circuit is loaded, the output voltage does not remain constant at Vmax. After point E (Fig.
(c)), the supply voltage becomes less than the capacitor voltage, diode stops conducting. The capacitor
can not discharge back into the a.c. system because of one way action of the diode. Instead, the current
now flows out of C to furnish the current iL through the load. While giving up this energy, the
capacitor voltage also decreases at a rate depending on the time constant CR of the circuit and it
reaches the point F corresponding to Vmin. Beyond F, the supply voltage is greater than the capacitor
voltage and hence the diode D starts conducting charging the capacitor C again to Vmax and also
during this period it supplies current to the load also. This second pulse of ip(ic + il) is of shorter
duration than the initial charging pulse as it serve mainly to restore into C the energy that C meanwhile
had supplied to load. Thus, while each pulse of diode current lasts much less than a half cycle, the load
receives current more continuously from C.
Assuming the charge supplied by the transformer to the load during the conduction period t,
which is very small to be negligible, the charge supplied by the transformer to the capacitor during
conduction equals the charge supplied by the capacitor to the load. Note that ic>> iL. During one
period T = 1/f of the a.c voltage, a charge Q is transferred to the load RL and is given as
A
l
l
J
N
T
U
W
o
r
l
d

Q =
zT
i
L
a t
f dt 
zT
VRL at f I
dt = IT = f
RL
where I is the mean value of the d.c output iL(t) and VRL(t) the d.c. voltage which includes a ripple as
shown in Fig. 2.1 (c).
This charge is supplied by the capacitor over the period T when the voltage changes from Vmax
to Vmin over approximately period T neglecting the conduction period of the diode.
Suppose at any time the voltage of the capacitor is V and it decreases by an amount of dV over
the time dt then charge delivered by the capacitor during this time is
dQ = CdV
Therefore, if voltage changes from Vmax to Vmin, the charge delivered by the capacitor
z
V
min
− Vmin g
dQ 
zVmax
CdV  − C
b V
max
Or the magnitude of charge delivered by the capacitor
Q = C (Vmax– Vmin) (2.3)
Using equation (2.2)
Q = 2δVC (2.4)
Therefore, 2δVC = IT
or δV = IT  I (2.5)
2 fC
2C
Equation (2.5) shows that the ripple in a rectifier output depends upon the load current and the circuit
parameter like f and C. The product fC is, therefore, an important design factor for the rectifiers. The
higher the frequency of supply and larger the value of filtering capacitor the smaller will be the ripple
in the d.c. output.
The single phase half-wave rectifier circuits have the following disadvantages:
(i) The size of the circuits is very large if high and pure d.c. output voltages are desired.
(ii) The h.t. transformer may get saturated if the amplitude of direct current is comparable with
the nominal alternating current of the transformer.
It is to be noted that all the circuits considered here are able to supply relatively low currents
and therefore are not suitable for high current applications such as HVDC transmission.
When high d.c. voltages are to be generated, voltage doubler or cascaded voltage multiplier
circuits are used. One of the most popular doubler circuit due to Greinacher is shown in Fig. 2.2.
Suppose B is more positive with respect to A and the diode D1 conducts thus charging the
capacitor C1 to Vmax with polarity as shown in Fig. 2.2. During the next half cycle terminal A of the
capacitor C1 rises to Vmax and hence terminal M attains a potential of 2 Vmax. Thus, the capacitor C2
is charged to 2 Vmax through D2. Normally the voltage across the load will be less than 2 Vmax
depending upon the time constant of the circuit C2RL.
A
l
l
J
N
T
U
W
o
r
l
d

Fig. Greinacher voltage doubler circuit
COCKROFT-WALTON VOLTAGE MULTIPLIER CIRCUIT
In 1932, Cockroft and Walton suggested an improvement over the circuit developed by Greinacher for
producing high D.C. voltages. Fig. 2.3. shows a multistage single phase cascade circuit of the Cockroft-
Walton type.
D3
No Load Operation: The portion ABM′MA is 0 O
exactly indentical to Greinarcher voltage doubler
circuit and the voltage across C becomes 2Vmax C
3 C3
when M attains a voltage 2Vmax. D3 RL
During the next half cycle when B becomes
N
positive with respect to A, potential of M falls and, N
D2
therefore, potential of N also falls becoming less
than potential at M′ hence C2 is charged through
C
2 C2
D2. Next half cycle A becomes more positive and D2
potential of M and N rise thus charging C′2 through M M
D′2. Finally all the capacitors C′1, C′2, C′3, C, C2, D1
and C3 are charged. The voltage across the column
C1
of capacitors consisting of C , C , C , keeps on C1
1 2 3
D
1
oscillating as the supply voltage alternates. This A
column, therefore, is known as oscillating column.
However, the voltage across the capacitances C′ ,
1
C′2 , C′3 , remains constant and is known as
smoothening column. The voltages at M′, N′, and
O′ are 2 Vmax 4 Vmax and 6 Vmax. Therefore, voltage B
across all the capacitors is 2 Vmax except for C1
where it is Vmax only. The total output voltage is 2n
where n is the number of stages. Thus, the
use of multistages arranged in the manner shown enables very high voltage to be obtained. The equal
stress of the elements (both capacitors and diodes) used is very helpful and promotes a modular design
of such generators.
Generator Loaded: When the generator is loaded, the output voltage will never reach the value 2n
V
max
Fig. 2.3
A
l
l
J
N
T
U
W
o
r
l
d

Vmax. Also, the output wave will consist of ripples on the voltage. Thus, we have to deal with two
quantities, the voltage drop ∆V and the ripple δV.
Suppose a charge q is transferred to the load per cycle. This charge is q = I/f = IT. The charge
comes from the smoothening column, the series connection of C′1, C′2, C′3,. If no charge were
transferred during T from this stack via D1, D2, D3, to the oscillating column, the peak to peak ripple
would merely be
n
2δV = IT ∑
1
′ (2.6)
n  0
C
i
But in practice charges are transferred. The process is explained with the help of circuits in Fig. 2.4 (a)
and (b).
Fig. 2.4 (a) shows arrangement when point A is more positive with reference to B and charging
of smoothing column takes place and Fig. 2.4 (b) shows the arrangement when in the next half cycle B
becomes positive with reference to A and charging of oscillating column takes place. Refer to Fig. 2.4
(a). Say the potential of point O′ is now 6 Vmax. This discharges through the load resistance and say the
charge lost is q = IT over the cycle. This must be regained during the charging cycle (Fig. 2.4 (a)) for
stable operation of the generator. C3 is, therefore supplied a charge q from C3. For this C2 must
acquire a charge of 2q so that it can supply q charge to the load and q to C3, in the next half cycle
termed by cockroft and Walton as the transfer cycle (Fig. 2.4 (b)). Similarly C′1 must acquire for
stability reasons a charge 3q so that it can supply a charge q to the load and 2q to the capacitor C2 in
the next half cycle (transfer half cycle).
0
D3
O 0 D3 O
D3
q
D
q
C3 C3
3
C3
C3
R
L RL
q D2 q D2
N N N N
D2
2q
D2
2q
C2 C2 C2 C2
2q D1
M
2q D1
M M M’
3q 2q 3q
C1
D1
C1 C1
D1
C1
3q 3q
A A
1
B B
(a) (b)
Fig. (a) Charging of smoothening Column (b) Charging of oscillating column
A
l
l
J
N
T
U
W
o
r
l
d

During the transfer cycle shown in Fig. 2.4 (b), the diodes D1, D2, D3, conduct when B is
positive with reference to A. Here C′2 transfers q charge to C3, C1 transfers charge 2q to C2 and the
transformer provides change 3q.
For n-stage circuit, the total ripple will be
I F 1 2 3 n I
2δV = G   ... J
C′n − 1
C′n − 2
f HC ′n
C
′1 K
I F 1 2 3 n I
or δV = G   ... J (2.7)
C′n − 1
C′n − 2
2 f H C
′n
C
′1 K
From equation (2.7), it is clear that in a multistage circuit the lowest capacitors are responsible for most
ripple and it is, therefore, desirable to increase the capacitance in the lower stages. However, this is
objectionable from the view point of High Voltage Circuit where if the load is large and the load voltage
goes down, the smaller capacitors (within the column) would be overstressed. Therefore, capacitors of
equal value are used in practical circuits i.e., C′n = C′n – 1 = ... C′1 = C and the ripple is given as
δV =
I n an  1f 
In an  1f (2.8)
2 fC 2 4 fC
The second quantity to be evaluated is the voltage drop ∆V which is the difference between the
theoretical no load voltage 2nVmax and the onload voltage. In order to obtain the voltage drop ∆V refer
to Fig. 2.4 (a).
Here C′ is not charged upto full voltage 2V but only to 2V max
– 3q/C because of the charge
1 max
given up through C1 in one cycle which gives a voltage drop of 3q/C = 3I/fC
The voltage drop in the transformer is assumed to be negligible. Thus, C2 is charged to the
voltage
F 3 I I 3I
G 2V
max − J −
fC
H fC K
since the reduction in voltage across C′3 again is 3I/fC. Therefore, C′2 attains the voltage
F 3 I  3 I  2 I I
2V
max – G J
fC
H K
In a three stage generator
∆V = 3I
1 fC
∆V = 2  3  a3 − 1f I
fC
2 m r
∆V3 = (2 × 3 + 2 × 2 + 1)
I
fC
A
l
l
J
N
T
U
W
o
r
l
d

In general for a n-stage generator
∆V= nI
n fC
∆V
n – 1 =
I
{2n + (n – 1)}
fC
∆V
n – 2=
I
{2n + 2 (n – 1) + (n – 2)}
fC
.
.
.
∆V1 =
I
{2n + 2 (n – 1) + 2 (n – 2) + ... 2 × 3 + 2 × 2 + 1}
fC
∆V = ∆Vn + ∆Vn – 1 + ... + ∆V1
After omitting I/fC, the series can be rewritten as:
Tn = n
Tn – 1 = 2n + (n – 1)
Tn – 2 = 2n + 2 (n – 1) + (n – 2)
Tn – 3 = 2n + 2 (n – 1) + 2 (n – 2) + (n – 3)
.
.
.
T1 = 2n + 2 (n – 1) + 2 (n – 2) + ... + 2 × 3 + 2 × 2 +
1 T = Tn + Tn – 1 + Tn – 2 + ... + T1
To sum up we add the last term of all the terms (Tn through T1) and again add the last term of
the remaining term and so on, i.e.,
[n + (n – 1) + (n – 2)+ ... +2 + 1]
+ [2n + 2 (n – 1) + 2 (n – 2) + ... + 2 × 2]
+ [2n + 2 (n – 1) + ... + 2 × 4 + 2 × 3]
+ [2n + 2 (n – 1) + ... + 2 × 4]
+ [2n + 2 (n – 1) + 2 (n – 2) + ... + 2 × 5] + ... [2n ]
Rearranging the above terms we have
n + (n – 1) + (n – 2) + ... + 2 + 1
+ [2n + 2 (n – 1) + 2 (n – 2) + ... + 2 × 2 + 2 × 1] – 2 × 1
+ [2n + 2 (n – 1) + 2 (n – 2) + ... + 2 × 3 + 2 × 2 + 2 × 1] – 2 × 2 – 2 × 1
+ [2n + 2 (n – 1) + 2 (n – 2) + ... + 2 × 4 + 2 × 3 + 2 × 2 + 2 × 1]
– 2 × 3 – 2 × 2 – 2 × 1
.
.
.
[2 × n + 2 (n – 1) + ... + 2 × 2 + 2 × 1] – [2 ( n – 1)]
+ 2 (n – 2) + ... + 2 × 2 + 2 ×1]
A
l
l
J
N
T
U
W
o
r
l
d

or n + (n – 1) + (n – 2) + ... + 2 + 1
Plus (n – 1) number of terms of 2 [n + (n – 1) + ...+ 2 + 1]
minus 2 [1 + (1 + 2) + (1 + 2 + 3) + ... + ... {1 + 2 + 3 + ... (n – 1)}]
The last term (minus term) is rewritten as
2 [1 + (1 + 2) + ... + {1 + 2 + 3 + ... (n – 1)} + {1 + 2 + ... + n}]
– 2 [1 + 2 + 3 + ... + n]
The nth term of the first part of the above series is given as
tn =
2n(n

1)
(n2 n)
2
Therefore, the above terms are equal to
= ∑ (n
2
+ n) – 2 ∑ n
= ∑ (n
2
– n)
Taking once again all the term we have
T = ∑ n + 2 (n – 1) ∑ n – ∑ (n
2
– n)
= 2n ∑ n – ∑ n
2
=2n . n ( n  1) − n ( n  1) ( 2 n  1)
2 6
=
6 ( n
3
 n
2
) – n ( 2 n
2
 3n  1)
6
=
6n
3
 6n
2
– 2n
3
– 3n
2
– n
6
4 n
3
 3n
2
– n 2 3  n
2
n
= = n – (2.9)
6 3 2 6
Here again the lowest capacitors contribute most to the voltage drop ∆V and so it is
advantageous to increase their capacitance in suitable steps. However, only a doubling of C1 is
convenient as this capacitors has to withstand only half of the voltage of other capacitors. Therefore,
∆V1 decreases by an amount nI/fC which rreduces ∆V of every stage by the same amount i.e., by
nI
n .
2 fC
Hence ∆V =
I F 2
n
3
–
nI (2.10)
fC H3 6 K
If n ≥ 4 we find that the linear term can be neglected and, therefore, the voltage drop can be
approximated to
∆V ≈ I .2 n3 (2.11)
fC 3
A
l
l
J
N
T
U
W
o
r
l
d

The maximum output voltage is given by
V = 2nV – I . 2n3 (2.12)
max
0 max fC 3
From (2.12) it is clear that for a given number of stages, a given frequency and capacitance of
each stage, the output voltage decrease linearily with load current I. For a given load, however, V0 =
(V0max– V) may rise initially with the number of stages n, and reaches a maximum value but decays
beyond on optimum number of stage. The optimum number of stages assuming a constant Vmax, I, f
and C can be obtained for maximum value of V0 max by differentiating equation (2.12) with respect to
n and equating it to zero.
dV
max
 2Vmax – 2 I 3 n
2
 0
fC
dn 3
= V – I n
2
 0
max fC
V
max
fC
or
n
opt
=
I (2.13)
Substituting nopt in equation (2.12) we have
Vmax fC F2V
max –
2I Vmax fC I
(V
0 max
)
max
=
I
G
3 fC
J
H I K
V
max
fC
F 2 I
= I H2Vmax − 3
V
max
K
Vmax fC
.
4 V
max (2.14)
= I 3
It is to be noted that in general it is more economical to use high frequency and smaller value of
capacitance to reduce the ripples or the voltage drop rather than low frequency and high capacitance.
Cascaded generators of Cockroft-Walton type are used and manufactured world wide these
days. A typical circuit is shown in Fig. 2.5. In general a direct current upto 20 mA is required for high
voltages between 1 MV and 2 MV. In case where a higher value of current is required, symmetrical
cascaded rectifiers have been developed. These consist of mainly two rectifiers in cascade with a
common smoothing column. The symmetrical cascaded rectifier has a smaller voltage drop and also a
smaller voltage ripple than the simple cascade. The alternating current input to the individual circuits
must be provided at the appropriate high potential; this can be done by means of isolating transformer.
Fig. 2.6 shows a typical cascaded rectifier circuit. Each stage consists of one transformer which feeds
two half wave rectifiers.
A
l
l
J
N
T
U
W
o
r
l
d

E M G
3 
A V V mA
W
Fig. 2.5 A typical Cockroft circuit
Fig. Cascaded rectifier circuit
As the storage capacitors of these half wave rectifiers are series connected even the h.v. winding of
T1 can not be grounded. This means that the main insulation between the primary and the secondary
winding of T1 has to be insulated for a d.c. voltage of magnitude Vmax, the peak voltage of T1. The same is
required for T2 also but this time the high voltage winding is at a voltage of 3Vmax. It would be
A
l
l
J
N
T
U
W
o
r
l
d

difficult to provide the whole main insulation within this transformer, an isolating transformer T
supplies T2. The cascading of every stage would thus require an additional isolating transformer which
makes this circuit less economical for more than two stages.
ELECTROSTATIC GENERATOR
In electromagnetic generators, current carrying conductors
are moved against the electromagnetic forces acting upon
them. In contrast to the generator, electrostatic generators E dx
V d
convert mechanical energy into electric energy directly. The
electric charges are moved against the force of electric Belt
fields, thereby higher potential energy is gained at the cost _
V
of mechanical energy.
The basic principle of operation is explained with Fig.
the help of Fig. 2.7.
An insulated belt is moving with uniform velocity ν in an electric field of strength E (x). Suppose
the width of the belt is b and the charge density σ consider a length dx of the belt, the charge dq = σ bdx.
The force experienced by this charge (or the force experienced by the belt).
dF = Edq = E σ bdx
or F = σb zEdx
Normally the electric field is uniform
∴ F = σbV
The power required to move the belt
= Force × Velocity
= Fv = σbVν 
Now current I = dq σb dx = σbv (2.16)
dt dt
∴ The power required to move the belt
P = Fν = σbVν = VI (2.17)
Assuming no losses, the power output is also equal to VI.
Fig. 2.8 shows belt driven electrostatic generator developed by Van deGraaf in 1931. An insu-lating
belt is run over pulleys. The belt, the width of which may vary from a few cms to metres is driven at a
speed of about 15 to 30 m/sec, by means of a motor connected to the lower pulley. The belt near the lower
pully is charged electrostatically by an excitation arrangement. The lower charge spray unit consists of a
number of needles connected to the controllable d.c. source (10 kV–100 kV) so that the discharge between
the points and the belt is maintained. The charge is conveyed to the upper end where it is collected from the
belt by discharging points connected to the inside of an insulated metal electrode through which the belt
passes. The entire equipment is enclosed in an earthed metal tank filled with insulating gases of good
dielectric strength viz. SF6 etc. So that the potential of the electrode could be
A
l
l
J
N
T
U
W
o
r
l
d

raised to relatively higher voltage without corona discharges or for a certain voltage a smaller size of
the equipment will result. Also, the shape of the h.t., electrode should be such that the surface gradient
of electric field is made uniform to reduce again corona discharges, even though it is desirable to
avoid corona entirely. An isolated sphere is the most favourable electrode shape and will maintain a
uniform field E with a voltage of Er where r is the radius of the sphere.
+ +
H. V. terminal
+ +
– Upper spray points
+ + – +
Collector + –
Upper pulley
+ –
+ + (insulated from earth)
+ –
+ – –
+ – Insulating belt
+ –
+ –
+ –
+ –
+ –
+ –
Motor driven pulley
+ –
Lower spray points
+ –
+ –
+ – Controllable spray
voltage
Fig. Van de Graaf generator
As the h.t. electrode collects charges its potential rises. The potential at any instant is given as
V = q/C where q is the charge collected at that instant. It appears as though if the charge were
collected for a long time any amount of voltage could be generated. However, as the potential of
electrode rises, the field set up by the electrode increases and that may ionise the surrounding medium
and, therefore, this would be the limiting value of the voltage. In practice, equilibrium is established at
a terminal voltage which is such that the charging current
F dV I
HI  C K
dt
equals the discharge current which will include the load current and the leakage and corona loss currents.
The moving belt system also distorts the electric field and, therefore, it is placed within properly shaped
field grading rings. The grading is provided by resistors and additional corona discharge elements.
The collector needle system is placed near the point where the belt enters the h.t. terminal. A
second point system excited by a self-inducing arrangement enables the down going belt to be charged
to the polarity opposite to that of the terminal and thus the rate of charging of the latter, for a given
speed, is doubled. The self inducing arrangement requires insulating the upper pulley and maintaining
it at a potential higher than that of the h.t. terminal by connecting the pulley to the collector needle
A
l
l
J
N
T
U
W
o
r
l
d

system. The arrangement also consists of a row of points (shown as upper spray points in Fig. 2.8)
connected to the inside of the h.t. terminal and directed towards the pulley above its points of entry
into the terminal. As the pulley is at a higher potential (positive), the negative charges due to corona
discharge at the upper spray points are collected by the belt. This neutralises any remaining positive
charge on the belt and leaves an excess of negative charges on the down going belt to be neutralised
by the lower spray points. Since these negative charges leave the h.t. terminal, the potential of the h.t.
terminal is raised by the corresponding amount.
In order to have a rough estimate of the current supplied by the generator, let us assume that
the electric field E is normal to the belt and is homogeneous.
We know that D = ε E where D is the flux density and since the medium surrounding the h.t.
terminal is say air ε = 1 and ε = 8.854 × 10
–12
F/metre.
r 0
According to Gauss law, D = σ the surface charge density.
Therefore, D = σ = ε E (2.18)
Assuming E = 30 kV/cm or 30,000 kV/m
σ = 8.854 × 10
–12
× 3000 × 10
3
= 26.562 × 10
–6
C/m
2
Assuming for a typical system b = 1 metre and velocity of the belt ν = 10 m/sec, and using
equation (2.16), the current supplied by the generator is given as
I = σ bν
= 26.562 × 10
–6
× 1 × 10
= 26.562 × 10
–5
Amp
= 265 A
From equation (2.16) it is clear that current I depends upon σ, b and  The belt width (b) and velocity
ν being limited by mechanical reasons, the current can be increased by having higher value of  σ can
be increased by using gases of higher dielectric strength so that electric field intensity E could be
increased without the inception of ionisation of the medium surrounding the h.t. terminal. However,
with all these arrangements, the actual short circuit currents are limited only to a few mA even for
large generators.
The advantages of the generator are:
(i) Very high voltages can be easily generated
(ii) Ripple free output
(iii) Precision and flexibility of control
The disadvantages are:
(i) Low current output
(ii) Limitations on belt velocity due to its tendency for vibration. The vibrations may make it
difficult to have an accurate grading of electric fields
These generators are used in nuclear physics laboratories for particle acceleration and other processes
in research work.
A
l
l
J
N
T
U
W
o
r
l
d

GENERATION OF HIGH A.C. VOLTAGES
Most of the present day transmission and distribution networks are operating on a.c. voltages and
hence most of the testing equipments relate to high a.c. voltages. Even though most of the equipments
on the system are 3-phase systems, a single phase transformer operating at power frequency is the
most common from of HVAC testing equipment.
Test transformers normally used for the purpose have low power rating but high voltage
ratings. These transformers are mainly used for short time tests on high voltage equipments. The
currents required for these tests on various equipments are given below:
Insulators, C.B., bushings, Instrument
transformers = 0.1– 0.5 A
Power transformers, h.v. capacitors. = 0.5–1 A
Cables = 1 A and above
The design of a test transformer is similar to a potential transformer used for the measurement
of voltage and power in transmission lines. The flux density chosen is low so that it does not draw
large magnetising current which would otherwise saturate the core and produce higher harmonics.
Cascaded Transformers
For voltages higher than 400 KV, it is desired to cascade two or more transformers depending upon the
voltage requirements. With this, the weight of the whole unit is subdivided into single units and, there-fore,
transport and erection becomes easier. Also, with this, the transformer cost for a given voltage may be
reduced, since cascaded units need not individually possess the expensive and heavy insulation required in
single stage transformers for high voltages exceeding 345 kV. It is found that the cost of insulation for such
voltages for a single unit becomes proportional to square of operating voltage.
Fig. 2.9 shows a basic scheme for cascading three transformers. The primary of the first stage
transformer is connected to a low voltage supply. A voltage is available across the secondary of this
transformer. The tertiary winding (excitation winding) of first stage has the same number of turns as
the primary winding, and feeds the primary of the second stage transformer. The potential of the
tertiary is fixed to the potential V of the secondary winding as shown in Fig. 2.9. The secondary
winding of the second stage transformer is connected in series with the secondary winding of the first
stage transformer, so that a voltage of 2V is available between the ground and the terminal of
secondary of the second stage transformer. Similarly, the stage-III transformer is connected in series
with the second stage transformer. With this the output voltage between ground and the third stage
transformer, secondary is 3V. it is to be noted that the individual stages except the upper most must
have three-winding transformers. The upper most, however, will be a two winding transformer.
Fig. 2.9 shows metal tank construction of transformers and the secondary winding is not
divided. Here the low voltage terminal of the secondary winding is connected to the tank. The tank of
stage-I transformer is earthed. The tanks of stage-II and stage-III transformers have potentials of V and
2V, respectively above earth and, therefore, these must be insulated from the earth with suitable solid
insulation. Through h.t. bushings, the leads from the tertiary winding and the h.v. winding are brought
out to be connected to the next stage transformer.
A
l
l
J
N
T
U
W
o
r
l
d

III
P
II
I = —
V
p p p
I
2p 2p p 3V
2V
3p
p
V
Fig. 2.9 Basic 3 stage cascaded transformer
However, if the high voltage windings are of mid-point potential type, the tanks are held at 0.5
V, 1.5 V and 2.5 V, respectively. This connection results in a cheaper construction and the high
voltage insulation now needs to be designed for V/2 from its tank potential.
The main disadvantage of cascading the transformers is that the lower stages of the primaries of
the transformers are loaded more as compared with the upper stages.
The loading of various windings is indicated by
P in Fig. 2.9. For the three-stage transformer, the total
output VA will be 3VI = 3P and, therefore, each of the
secondary winding of the transformer would carry a cur-
rent of I = P/V. The primary winding of stage-III trans-
former is loaded with P and so also the tertiary winding
of second stage transformer. Therefore, the primary of
the second stage transformer would be loaded with 2P.
Extending the same logic, it is found that the first stage
primary would be loaded with P. Therefore, while de-
signing the primaries and tertiaries of these transformers,
this factor must be taken into consideration. Fig. Equivalent circuit of one stage
The total short circuit impedance of a cascaded
transformer from data for individual stages can be obtained. The equivalent circuit of an individual
stage is shown in Fig.
Here Zp, Zs, and Zt, are the impedances associated with each winding. The impedances are
shown in series with an ideal 3-winding transformer with corresponding number of turns Np, Ns and
Nt. The impedances are obtained either from calculated or experimentally-derived results of the three
short-circuit tests between any two windings taken at a time.
A
l
l
J
N
T
U
W
o
r
l
d

Let Zps = leakage impedance measured on primary side with secondary short circuited and ter-
tiary open.
Zpt = leakage impedance measured on primary side with tertiary short circuited and second-
ary open.
Zst = leakage impedance on secondary side with tertiary short circuited and primary open.
If these measured impedances are referred to primary side then
Zps = Zp + Zs, Zpt = Zp + Zt and Zst = Zs + Zt
Solving these equations, we have
1 1
Zp = (Zps + Zpt – Zst), Zs = (Zps + Zst – Zpt)
2 2
1
and Zt = (Zpt + Zst – Zps) (2.19)
2
Assuming negligible magnetising current, the sum of the ampere turns of all the windings must
be zero.
Np Ip – Ns Is – Nt It = 0
Assuming lossless transformer, we have,
Zp = jXp, Zs = jXs and Zt = jXt
Xt 2
XP 3
X
S
3 I
I
X
t1
X
P2
2I
X
S2
I
V
XP
1 3I
XS1
I
V1
Fig. 2.11 Equivalent circuit of 3-stage transformer
Also let Np = Nt for all stages, the equivalent
cir-cuit for a 3-stage transformer would be given as in
Fig. 2.11
Fig. 2.11 can be further reduced to a very simpli-
fied circuit as shown in Fig. 2.12. The resulting short
circuit reactance Xres is obtained from the condition that
I X
res
=
3N
s V1
V1
V
2
N
p
Fig. A simplified equivalent circuit
A
l
l
J
N
T
U
W
o
r
l
d

the power rating of the two circuits be the same. Here currents have been shown corresponding to
high voltage side.
I
2
X = (3I)
2
X + (2I)
2
X
p
+ I
2
X
p
+ I
2
X
s
+ I
2
X + I
2
X
s
+ (2I)
2
X + I
2
X
t
res p s t
Xres = 14Xp + 3Xs + 5xt (2.20)
instead of 3(Xp + Xs + Xt) as might be expected. Equation (2.20) can be generalised for an n-stage
transformer as follows:
n
Xres = ∑
[(n – i

1)2 X
pi
X
si  (i – 1)
2
Xti ]
i  1
Where Xpi, Xsi and Xti are the short-circuit reactance of the primary, secondary and tertiary windings
of ith transformer.
It has been observed that the impedance of a two-stage transformer is about 3–4 times the
impedance of one unit and a three-stage impedance is 8–9 times the impedance of one unit
transformer. Hence, in order to have a low impedance of a cascaded transformer, it is desirable that
the impedance of individual units should be as small as possible.
Reactive Power Compensation
As is mentioned earlier, the test transformers are used for testing the insulation of various electrical
equipments. This means the load connected to these transformers is highly capacitive. Therefore, if
rated voltage is available at the output terminals of the test transformer and a test piece (capacitive
load) is connected across its terminals, the voltage across the load becomes higher than the rated volt-
age as the load draws leading current. Thus, it is necessary to regulate the input voltage to the test
transformer so that the voltage across the load, which is variable, depending on the test specimen,
remains the rated voltage. Another possibility is that a variable inductor should be connected across
the supply as shown in Fig. 2.13 so that the reactive power supplied by the load is absorbed by the
inductor and thus the voltage across the test transformer is maintained within limits.
I. Regulating transformer II. Compensating reactor III. Test transformer with commutable primary
Fig. Basic principle of reactive power compensation
It should be noted that the test transformer should be able to supply the maximum value of load
current for which it has been designed at all intermediate voltages including the rated voltage. The
power voltage characteristic is, therefore, a straight line as shown by line A in Fig. 2.14. The compen-
sating reactive power absorbed by the air-cored inductor is shown on parabolas B, C and D. These will
A
l
l
J
N
T
U
W
o
r
l
d

be parabolas as the reactive power = V
2
/X. Curve B corresponds to the condition when the transformer
primary is connected in parallel and the reactor is connected at position 1 in Fig. 2.13.
Similarly Curve C—Transformer primary connected in parallel and reactor at position 1 con-
nected.
Curve D—Transformer primary connected in series and reactor at position 2.
Fig. 2.14 Reactive power compensation
When the primary series is connected, for the same supply voltage, voltage per turn of primary
becomes half its value when it is parallel connected and, therefore, the secondary voltage becomes 1/2
of the rated voltage and hence the curve starts at 50% of the rated voltage. The power of the voltage
regulator is proportional to the supply voltage and, therefore, is represented by line E in Fig. 2.14 and
the maximum power at rated voltage is 33.3% of the maximum power requirement of the transformer.
All possible operating conditions of the test transformer lie within the triangular area enclosed by the
line A, the abscissa and the 100% rated voltage line. This area has been sub-divided into different
parts, so that the permissible supply power (Here 33% of maximum transformer load) is never
exceeded. The value of the highest voltage is always taken for the evaluation of the compensation
arrangement. Since the impedance of the test transformer is usually large (about 20–25%), the range
under 25% of the rated voltage is not considered.
It is clear from the above considerations that the design of the compensating reactor depends
upon
(i) The capacitance and operating voltage of test specimen.
(ii) The power rating of the available regulator.
(iii) The possibility of different connections of the winding of test transformer.
(iv) The power rating of the test transformer.
In order that the test laboratory meets all the different requirements, every particular case must
be investigated and a suitable reactor must be designed for reactive power compensation.
In multistage transformers with large power output, it is desirable to provide reactive power
compensation at every stage, so that the voltage stability of the test transformer is greatly improved.
SERIES RESONANT CIRCUIT
The equivalent circuit of a single-stage-test transformer alongwith its capacitive load is shown in Fig. .
Here L1 represents the inductance of the voltage regulator and the transformer primary, L the
A
l
l
J
N
T
U
W
o
r
l
d

exciting inductance of the transformer, L2 the
inductance of the transformer secondary and C the
capacitance of the load. Normally inductance L is very
large as compared to L1 and L2 and hence its shunting
effect can be neglected. Usually the load capacitance
is variable and it is possible that for certain loading,
resonance may occur in the circuit suddenly and the
current will then only be limited by the resistance of Fig. Equivalent circuit of a single
the circuit and the voltage across the test specimen may stage loaded transformer
go up as high as 20 to 40 times the desired value.
Similarly, presence of harmonics due to saturation of iron core of transformer may also result
in resonance. Third harmonic frequencies have been found to be quite disastrous.
With series resonance, the resonance is controlled at fundamental frequency and hence no un-
wanted resonance occurs.
The development of series resonance circuit for testing purpose has been very widely welcome
by the cable industry as they faced resonance problem with test transformer while testing short lengths
of cables.
In the initial stages, it was difficult to manufacture continuously variable high voltage and high
value reactors to be used in the series circuit and therefore, indirect methods to achieve this objective
were employed. Fig. 2.16 shows a continuously variable reactor connected in the low voltage winding
of the step up transformer whose secondary is rated for the full test voltage. C2 represents the load
capacitance.
Fig. Single transformer/reactor series resonance circuit
If N is the transformation ratio and L is the inductance on the low voltage side of the trans-
former, then it is reflected with N
2
L value on the secondary side (load side) of the transformer. For
certain setting of the reactor, the inductive reactance may equal the capacitive reactance of the circuit,
hence resonance will take place. Thus, the reactive power requirement of the supply becomes zero and
A
l
l
J
N
T
U
W
o
r
l
d

it has to supply only the losses of the circuit. However, the transformer has to carry the full load
current on the high voltage side. This is a disadvantage of the method. The inductor are designed for
high quality factors Q = ωL / R. The feed transformer, therefore, injects the losses of the circuit only.
It has now been possible to manufacture high voltage continuously variable reactors 300 kV
per unit using a new technique with split iron core. With this, the testing step up transformer can be
omitted as shown in Fig. 2.17. The inductance of these inductors can be varied over a wide range
depending upon the capacitance of the load to produce resonance.
R
L
C 2
C 2
Feed
(b)
To transformer motor
(a)
Fig. (a) Series resonance circuit with variable h.t. reactors (b) Equivalent circuit of (a)
Fig. 2.17 (b) represents an equivalent circuit for series resonance circuit. Here R is usually of
low value. After the resonance condition is achieved, the output voltage can be increased by
increasing the input voltage. The feed transformers are rated for nominal current ratings of the reactor.
Under resonance, the output voltage will be
V 1
V
0
=
RωC2
Where V is the supply voltage.
Since at resonance
ωL =
1
ωC2
Therefore V0 =V ω L  VQ
R
where Q is the quality factor of the inductor which usually varies between 40 and 80. This means that
with Q = 40, the output voltage is 40 times the supply voltage. It also means that the reactive power
requirements of the load capacitance in kVA is 40 times the power to be provided by the feed trans-
former in KW. This results in a relatively small power rating for the feed transformer.
A
l
l
J
N
T
U
W
o
r
l
d

The following are the advantages of series resonance circuit.
(i) The power requirements in KW of the feed circuit are (kVA)/Q where kVA is the reactive
power requirements of the load and Q is the quality factor of variable reactor usually
greater than 40. Hence, the requirement is very small.
(ii) The series resonance circuit suppresses harmonics and interference to a large extent. The
near sinusoidal wave helps accurate partial discharge of measurements and is also desirable
for measuring loss angle and capacitance of insulating materials using Schering Bridge.
(iii) In case of a flashover or breakdown of a test specimen during testing on high voltage side,
the resonant circuit is detuned and the test voltage collapses immediately. The short circuit
current is limited by the reactance of the variable reactor. It has proved to be of great value
as the weak part of the isolation of the specimen does not get destroyed. In fact, since the
arc flash over has very small energy, it is easier to observe where exactly the flashover is
occurring by delaying the tripping of supply and allowing the recurrence of flashover.
(iv) No separate compensating reactors (just as we have in case of test transformers) are
required. This results in a lower overall weight.
(v) When testing SF6 switchgear, multiple breakdowns do not result in high transients. Hence,
no special protection against transients is required.
(vi) Series or parallel connections of several units is not at all a problem. Any number of units
can be connected in series without bothering for the impedance problem which is very
severely associated with a cascaded test transformer. In case the test specimen requires
large current for testing, units may be connected in parallel without any problem.
Fig. Parallel resonance system
Fig. 2.18 shows schematic of a typical parallel resonant systems. Here the variable reactor is
incorporated into the high voltage transformer by introducing a variable air gap in the core of the
transformer. With this connection, variation in load capacitance and losses cause variation in input
current only. The output voltage remains practically constant. Within the units of single stage design,
the parallel resonant method offers optimum testing performance.
In an attempt to take advantage of both the methods of connections, i.e., series and parallel
resonant systems, a third system employing series parallel connections was tried. This is basically a
modification of a series resonant system to provide most of the characteristics of the parallel system.
Fig. shows a schematic of a typical series parallel method.
A
l
l
J
N
T
U
W
o
r
l
d

Fig. Series-parallel resonant system
Here the output voltage is achieved by auto transformer action and parallel compensation is
achieved by the connection of the reactor. It has been observed that during the process of tuning for
most of the loads, there is a certain gap opening that will result in the parallel connected test system
going into uncontrolled over voltaging of the test sample and if the test set is allowed to operate for a
long time, excessive heating and damage to the reactor would result.
Also, it has been observed experimentally that complete balance of ampere turns takes place
when the system operates under parallel resonance condition. Under all other settings of the variable
reactor, an unbalance in the ampere turns will force large leakage flux into the surrounding metallic
tank and clamping structure which will cause large circulating currents resulting in hot spots which
will affect adversely the dielectric strength of oil in the tank.
In view of the above considerations, it has been recommended not to go in for series-parallel
resonant mode of operation for testing purpose. If a single stage system upto 300 kV using the
resonance test voltage is required, parallel resonant system must be adopted. For test voltage
exceeding 300 kV, the series resonant method is strongly recommended.
The specific weight of a cascaded test transformer varies between 10 and 20 kg/kVA whereas
for a series resonant circuit with variable high voltage reactors it lies between 3 and 6 kg/kVA.
With the development of static
frequency convertor, it has now been possible
to reduce the specific weight still further. In
order to obtain resonance in the circuit a choke
of constant magnitude can be used and as the
load capacitance changes the source frequency
should be changed. Fig. 2.20 shows a
schematic diagram of a series resonant circuit
with variable frequency source. Fig. Schematic diagram of series resonant
The frequency convertor supplies the
losses of the testing circuit only which are
usually of the order of 3% of the reactive
power of the load capacitor as the chokes can be designed for very high quality factors.
A word of caution is very important, here in regard to testing of test specimen having large
capacitance. With a fixed reactance, the frequency for resonance will be small as compared to normal
frequency. If the voltage applied is taken as the normal voltage the core of the feed transformer will
get saturated as V/f then becomes large and the flux in the core will be large. So, a suitable voltage
must be applied to avoid this situation.
circuit with variable frequency sources
A
l
l
J
N
T
U
W
o
r
l
d

UNIT-V
MEASURMENTS OF HIGH VOLTAGES AND CURRENTS
INTRODUCTION
Transient measurements have much in common with measurements of steady state quantities but the
short-lived nature of the transients which we are trying to record introduces special problems.
Frequently the transient quantity to be measured is not recorded directly because of its large
magnitudes e.g. when a shunt is used to measure current, we really measure the voltage across the
shunt and then we assume that the voltage is proportional to the current, a fact which should not be
taken for granted with transient currents. Often the voltage appearing across the shunt may be
insufficient to drive the measuring device; it requires amplification. On the other hand, if the voltage
to be measured is too large to be measured with the usual meters, it must be attenuated. This suggests
an idea of a measuring system rather than a measuring device.
Measurements of high voltages and currents involves much more complex problems which a
specialist, in common electrical measurement, does not have to face. The high voltage equipments
have large stray capacitances with respect to the grounded structures and hence large voltage gradients
are set up. A person handling these equipments and the measuring devices must be protected against
these over voltages. For this, large structures are required to control the electrical fields and to avoid
flash over between the equipment and the grounded structures. Sometimes, these structures are re-
quired to control heat dissipation within the circuits. Therefore, the location and layout of the
equipments is very important to avoid these problems. Electromagnetic fields create problems in the
measurements of impulse voltages and currents and should be minimised.
The chapter is devoted to describing various devices and circuits for measurement of high voltages
and currents. The application of the device to the type of voltages and currents is also discussed.
SPHERE GAP
Sphere gap is by now considered as one of the standard methods for the measurement of peak value of d.c.,
a.c. and impulse voltages and is used for checking the voltmeters and other voltage measuring devices used
in high voltage test circuits. Two identical metallic spheres separated by certain distance form a sphere gap.
The sphere gap can be used for measurement of impulse voltage of either polarity provided that the impulse
is of a standard wave form and has wave front time at least 1 micro sec. and wave tail time of 5 micro sec.
Also, the gap length between the sphere should not exceed a sphere radius. If these conditions are satisfied
and the specifications regarding the shape, mounting, clearances
110
A
l
l
J
N
T
U
W
o
r
l
d

of the spheres are met, the results obtained by the use of sphere gaps are reliable to within ±3%. It has
been suggested in standard specification that in places where the availability of ultraviolet radiation is
low, irradiation of the gap by radioactive or other ionizing media should be used when voltages of
magnitude less than 50 kV are being measured or where higher voltages with accurate results are to be
obtained.
In order to understand the importance of irradiation of sphere gap for measurement of impulse
voltages especially which are of short duration, it is necessary to understand the time-lag involved in
the development of spark process. This time lag consists of two components—(i) The statistical time-
lag caused by the need of an electron to appear in the gap during the application of the voltage. (ii)
The formative time lag which is the time required for the breakdown to develop once initiated.
The statistical time-lag depends on the irradiation level of the gap. If the gap is sufficiently
irradiated so that an electron exists in the gap to initiate the spark process and if the gap is subjected to
an impulse voltage, the breakdown will take place when the peak voltage exceeds the d.c. breakdown
value. However, if the irradiation level is low, the voltage must be maintained above the d.c. break-
down value for a longer period before an electron appears. Various methods have been used for
irradia-tion e.g. radioactive material, ultraviolet illumination as supplied by mercury arc lamp and
corona discharges.
It has been observed that large variation can occur in the statistical time-lag characteristic of a
gap when illuminated by a specified light source, unless the cathode conditions are also precisely
specified.
Irradiation by radioactive materials has the advantage in that they can form a stable source of
irradiation and that they produce an amount of ionisation in the gap which is largely independent of
the gap voltage and of the surface conditions of the electrode. The radioactive material may be placed
inside high voltage electrode close behind the sparking surface or the radioactive material may form
the sparking surface.
The influence of the light from the impulse generator spark gap on the operation of the sphere
gaps has been studied. Here the illumination is intense and occurs at the exact instant when it is re-
quired, namely, at the instant of application of the voltage wave to the sphere gap.
The formative time lag depends mainly upon the mechanism of spark growth. In case of
second-ary electron emission, it is the transit time taken by the positive ion to travel from anode to
cathode that decides that formative time lag. The formative time-lag decreases with the applied over
voltage and increase with gap length and field non-uniformity.
Specifications on Spheres and Associated Accessories
The spheres should be so made that their surfaces are smooth and their curvatures as uniform as
possible. The curvature should be measured by a spherometer at various positions over an area
enclosed by a circle of radius 0.3 D about the sparking point where D is the diameter of the sphere and
sparking points on the two spheres are those which are at minimum distances from each other.
For smaller size, the spheres are placed in horizontal configuration whereas large sizes
(diameters), the spheres are mounted with the axis of the sphere gaps vertical and the lower sphere is
grounded. In either case, it is important that the spheres should be so placed that the space between
spheres is free from external electric fields and from bodies which may affect the field between the
spheres (Figs. 4.1 and 4.2).
A
l
l
J
N
T
U
W
o
r
l
d

4 X
B
0·5D 0·2D
0·2D 0·5D
15D
3 3
2D
2
2D
S
min
2D 2D
A A
1
Fig.
1
4
0·5D
0·20
5
X 2
2
D
D
P B
S
1
5
D
0·2D A
0·5D 3
1·5D
Fig.
15D
A
l
l
J
N
T
U
W
o
r
l
d

According to BSS 358: 1939, when one sphere is grounded, the distance from the sparking
point of the high voltage sphere to the equivalent earth plane to which the earthed sphere is connected
should lie within the limits as given in Table 4.1.
Table
Height of sparking point of high voltage sphere above the equivalent earth plane.
S = Sparking point distance
Sphere Diameter S < 0.5 D S > 0.5 D
D Maxm. Min. Maxm. Min.
Height Height Height Height
Upto 25 cms. 7 D 10 S 7 D 5 D
50 cms. 6 D 8 S 6 D 4 D
75 cms. 6 D 8 S 6 D 4 D
100 cms. 5 D 7 S 5 D 3.5 D
150 cms. 4 D 6 S 4 D 3 D
200 cms. 4 D 6 S 4 D 3 D
In order to avoid corona discharge, the shanks supporting the spheres should be free from
sharp edges and corners. The distance of the sparking point from any conducting surface except the
shanks should be greater than
F
V
I
H 25  K cms
3
where V is the peak voltage is kV to be measured. When large spheres are used for the measurement
of low voltages the limiting distance should not be less than a sphere diameter.
It has been observed that the metal of which the spheres are made does not affect the accuracy
of measurements MSS 358: 1939 states that the spheres may be made of brass, bronze, steel, copper,
aluminium or light alloys. The only requirement is that the surfaces of these spheres should be clean,
free from grease films, dust or deposited moisture. Also, the gap between the spheres should be kept
free from floating dust particles, fibres etc.
For power frequency tests, a protective resistance with a value of 1Ω/V should be connected in
between the spheres and the test equipment to limit the discharge current and to prevent high
frequency oscillations in the circuit which may otherwise result in excessive pitting of the spheres. For
higher frequencies, the voltage drop would increase and it is necessary to have a smaller value of the
resistance. For impulse voltage the protective resistors are not required. If the conditions of the spheres
and its associated accessories as given above are satisfied, the spheres will spark at a peak voltage
which will be close to the nominal value shown in Table 4.2. These calibration values relate to a
temperature of 20°C and pressure of 760 mm Hg. For a.c. and impulse voltages, the tables are
considered to be accurate within 3% for gap lengths upto 0.5 D. The tables are not valid for gap
lengths less than 0.05 D and impulse voltages less than 10 kV. If the gap length is greater than 0.5 D,
the results are less accurate and are shown in brackets.
A
l
l
J
N
T
U
W
o
r
l
d

Table
Sphere gap with one sphere earthed
Peak value of disruptive discharge voltages (50% for impulse tests) are valid for (i) alternating voltages
(ii) d.c. voltage of either polarity (iii) negative lightning and switching impulse voltages
Sphere Gap Voltage KV Peak
Spacing mm Sphere dia in cm.
12.5 25 50 75 100 150 200
10 31.7
20 59.0
30 85 86
40 108 112
50 129 137 138 138 138 138
75 167 195 202 203 203 203 203
100 (195) 244 263 265 266 266 266
125 (214) 282 320 327 330 330 330
150 (314) 373 387 390 390 390
175 (342) 420 443 443 450 450
200 (366) 460 492 510 510 510
250 (400) 530 585 615 630 630
300 (585) 665 710 745 750
350 (630) 735 800 850 855
400 (670) (800) 875 955 975
450 (700) (850) 945 1050 1080
500 (730) (895) 1010 1130 1180
600 (970) (1110) 1280 1340
700 (1025) (1200) 1390 1480
800 (1260) 1490 1600
900 (1320) 1580 1720
1000 (1360) 1660 1840
1100 1730 1940
1200 1800 2020
1300 1870 2100
1400 1920 2180
1500 1960 2250
1600 2320
1700 2370
1800 2410
1900 2460
2000 2490
A
l
l
J
N
T
U
W
o
r
l
d

Due to dust and fibre present in the air, the measurement of d.c. voltages is generally subject to
larger errors. Here the accuracy is within 5% provided the spacing is less than 0.4 D and excessive
dust is not present.
The procedure for high voltage measurement using sphere gaps depends upon the type of
voltage to be measured.
Table
Sphere Gap with one sphere grounded
Peak values of disruptive discharge voltages (50% values).
Positive lightning and switching impulse voltages
Peak Voltage kV
Sphere Gap Sphere dia in cms
Spacing mm 12.5 25 50 75 100 150 200
10 31.7
20 59 59
30 85.5 86
40 110 112
50 134 138 138 138 138 138 138
75 (181) 199 203 202 203 203 203
100 (215) 254 263 265 266 266 266
125 (239) 299 323 327 330 330 330
150 (337) 380 387 390 390 390
175 (368) 432 447 450 450 450
200 (395) 480 505 510 510 510
250 (433) 555 605 620 630 630
300 (620) 695 725 745 760
350 (670) 770 815 858 820
400 (715) (835) 900 965 980
450 (745) (890) 980 1060 1090
500 (775) (940) 1040 1150 1190
600 (1020) (1150) (1310) 1380
700 (1070) (1240) (1430) 1550
750 (1090) (1280) (1480) 1620
800 (1310) (1530) 1690
900 (1370) (1630) (1820)
1000 (1410) (1720) 1930
1100 (1790) (2030)
1200 (1860) (2120)
For the measurement of a.c. or d.c. voltage, a reduced voltage is applied to begin with so that the
switching transient does not flash over the sphere gap and then the voltage is increased gradually till the
gap breaks down. Alternatively the voltage is applied across a relatively large gap and the spacing is
A
l
l
J
N
T
U
W
o
r
l
d

then gradually decreased till the gap breaks down. Corresponding to this gap the value of peak voltage
can be read out from the calibration tables. However, it is reminded that the calibration tables values
correspond to 760 mm Hg pressure and 20°C temperature. Any deviation from the value, a correction
factor will have to be used to get the correct value of the voltage being measured.
For the measurement of 50% impulse disruptive discharge voltages, the spacing of the sphere
gap or the charging voltage of the impulse generator is adjusted in steps of 3% of the expected
disruptive voltage. Six applications of the impulse should be made at each step and the interval
between two applications is 5 seconds. The value giving 50% probability to disruptive discharge is
preferably obtained by interpolation between at least two gap or voltage settings, one resulting in two
disruptive discharges or less out of six applications and the other in four disruptive discharges or more
out of again six applications.
Another method, simple though less accurate, is to adjust the settings such that four to six
disruptive discharges are obtained in a series of ten successive applications.
The breakdown voltage of a sphere gap increases with increase in pressure and decreases with
increase in temperature. The value of disruptive voltages as given in Tables 4.2 and 4.3 correspond to
760 mm Hg pressure and 20°C. For small variation in temperatures and pressures, the disruptive
voltage is closely proportional to the relative air density. The relative air density δ is given by
δ = 293b
 t)
760 (273
where b and t are the atmospheric conditions (pressure in mm of Hg and temperature in °C
respectively) during measurement. The disruptive voltage V is given V = KdV0
Where V0 is the disruptive voltage as given in the Tables 4.2 and 4.3 and Kd is a correction factor
given in Table 4.4. Kd is a slightly non-linear function of δ a result explained by Paschen's law.
Table
Air density correction factor Kd
δ 0.70 0.75 0.8 0.85 0.90 0.95 1.0 1.05 1.10
Kd 0.72 0.76 0.81 0.86 0.90 0.95 1.0 1.05 1.09
Some of the other factors which influence the breakdown value of air are discussed here.
Influence of Nearby Earthed Objects
The influence of nearby earthed object on the direct voltage breakdown of horizontal gaps was studied
by Kuffel and Husbands. They surrounded the gap by a cylindrical metal cage and found that the
breakdown voltage reduced materially especially when the gap length exceeded a sphere radius. The
experiments were conducted on 6.25 and 12.5 cm. diameter spheres when the radius of the
surrounding metal cylinder (B) was varied from 12.6 D to 4 D. The observation corresponding to 12.6
D was taken as a reference. The reduction in the breakdown voltage for a given S/D fitted closely into
an empirical relation of the form.
∆V = m ln D
B
 C
A
l
l
J
N
T
U
W
o
r
l
d

CriticalbreakdownvoltageinKV
650
S/D = 0.6
600
550
500
S/D = 0.4
450
400
0
0 2 4 6 8 10 12
A/D (diameter)
Fig. Breakdown voltage as a function of A/D
Where ∆V = per cent reduction in voltage in the breakdown voltage from the value when the
clearance was 12.6 D, and m and C are the factors dependent on the ratio S/D.
Fiegel and Keen have studied the influence of nearby ground plane on impulse breakdown
voltage of a 50 cm diameter sphere gap using 1.5/40 micro sec. negative polarity impulse wave. Fig.
4.3 shows the breakdown voltage as a function of A/D for various values of S/D. The voltage values
were corrected for relative air density.
It is observed that the voltage increases with increase in the ratio A/D. The results have been
compared with those given in Table 4.2 and represented in Fig. 4.3 by dashed lines. The results also agree
with the recommendation regarding the minimum and maximum values of A/D as given in Table 4.1.
Influence of Humidity
Kuffel has studied the effect of the humidity on the breakdown voltage by using spheres of 2 cms to
25 cms diameters and uniform field electrodes. The effect was found to be maximum in the region 0.4
mm Hg. and thereafter the change was decreased. Between 4–17 mm Hg. the relation between
breakdown voltage and humidity was practically linear for spacing less than that which gave the
maximum humidity effect. Fig. 4.4 shows the effect of humidity on the breakdown voltage of a 25 cm
diameter sphere with spacing of 1 cm when a.c. and d.c voltages are applied. It can be seen that
(i) The a.c. breakdown voltage is slightly less than d.c. voltage.
(ii) The breakdown voltage increases with the partial pressure of water vapour.
It has also been observed that
(i) The humidity effect increases with the size of spheres and is largest for uniform field elec-
trodes.
(ii) The voltage change for a given humidity change increase with gap length.
A
l
l
J
N
T
U
W
o
r
l
d

Fig. Breakdown voltage humidity relation for a.c. and d.c. for
1.0 cm gap between 25 cms diameter spheres
The increase in breakdown voltage with increase in partial pressure of water vapour and this
increase in voltage with increase in gap length is due to the relative values of ionisation and
attachment coefficients in air. The water particles readily attach free electrons, forming negative ions.
These ions therefore slow down and are unable to ionise neutral molecules under field conditions in
which electrons will readily ionise. It has been observed that within the humidity range of 4 to 17
g/m
3
(relative humidity of 25 to 95% for 20°C temperature) the relative increase of breakdown
voltage is found to be between 0.2 to 0.35% per gm/m
3
for the largest sphere of diameter 100 cms and
gap length upto 50 cms.
Influence of Dust Particles
When a dust particle is floating between the gap this results into erratic breakdown in homogeneous or
slightly inhomogenous electrode configurations. When the dust particle comes in contact with one
electrode under the application of d.c. voltage, it gets charged to the polarity of the electrode and gets
attracted by the opposite electrode due to the field forces and the breakdown is triggered shortly
before arrival. Gaps subjected to a.c. voltages are also sensitive to dust particles but the probability of
erratic breakdown is less. Under d.c. voltages erratic breakdowns occur within a few minutes even for
voltages as low as 80% of the nominal breakdown voltages. This is a major problem, with high d.c.
voltage measurements with sphere gaps.
UNIFORM FIELD SPARK GAPS
Bruce suggested the use of uniform field spark gaps for the measurements of a.c., d.c. and impulse
voltages. These gaps provide accuracy to within 0.2% for a.c. voltage measurements an appreciable
A
l
l
J
N
T
U
W
o
r
l
d

improvement as compared with the equivalent sphere gap arrangement. Fig. 4.5 shows a half-contour
of one electrode having plane sparking surfaces with edges of gradually increasing curvature.
Fig. Half contour of uniform spark gap
The portion AB is flat, the total diameter of the flat portion being greater than the maximum
spacing between the electrodes. The portion BC consists of a sine curve based on the axes OB and OC
and given by XY = CO sin (BX/BO . π/2). CD is an arc of a circle with centre at O.
Bruce showed that the breakdown voltage V of a gap of length S cms in air at 20°C and 760
mm Hg. pressure is within 0.2 per cent of the value given by the empirical relation.
V = 24.22S + 6.08 S
This equation, therefore, replaces Tables 4.2 and 4.3 which are necessary for sphere gaps. This
is a great advantage, that is, if the spacing between the spheres for breakdown is known the
breakdown voltage can be calculated.
The other advantages of uniform field spark gaps are
(i) No influence of nearby earthed objects
(ii) No polarity effect.
However, the disadvantages are
(i) Very accurate mechanical finish of the electrode is required.
(ii) Careful parallel alignment of the two electrodes.
(iii) Influence of dust brings in erratic breakdown of the gap. This is much more serious in
these gaps as compared with sphere gaps as the highly stressed electrode areas become
much larger.
Therefore, a uniform field gap is normally not used for voltage measurements.
ROD GAPS
A rod gap may be used to measure the peak value of power frequency and impulse voltages. The gap
usually consists of two 1.27 cm square rod electrodes square in section at their end and are mounted
on insulating stands so that a length of rod equal to or greater than one half of the gap spacing
overhangs the inner edge of the support. The breakdown voltages as found in American standards for
different gap lengths at 25° C, 760 mm Hg. pressure and with water vapour pressure of 15.5 mm Hg.
are reproduced here.
A
l
l
J
N
T
U
W
o
r
l
d

Gap length in Breakdown Voltage KV Gap Length in cms. Breakdown
Cms. peak Voltage KV peak
2 26 80 435
4 47 90 488
6 62 100 537
8 72 120 642
10 81 140 744
15 102 160 847
20 124 180 950
25 147 200 1054
30 172 220 1160
35 198
40 225
50 278
60 332
70 382
The breakdown voltage is a rod gap increases more or less linearly with increasing relative air
density over the normal variations in atmospheric pressure. Also, the breakdown voltage increases
with increasing relative humidity, the standard humidity being taken as 15.5 mm Hg.
Because of the large variation in breakdown voltage for the same spacing and the uncertainties
associated with the influence of humidity, rod gaps are no longer used for measurement of a.c. or impulse
voltages. However, more recent investigations have shown that these rods can be used for d.c. measurement
provided certain regulations regarding the electrode configurations are observed. The arrangement consists
of two hemispherically capped rods of about 20 mm diameter as shown in Fig. 4.6.
Fig. Electrode arrangement for a rod gap to measure HVDC
The earthed electrode must be long enough to initiate positive breakdown streamers if the high
voltage rod is the cathode. With this arrangement, the breakdown voltage will always be initiated by
positive streamers for both the polarities thus giving a very small variation and being humidity
dependent. Except for low voltages (less than 120 kV), where the accuracy is low, the breakdown
voltage can be given by the empirical relation.
A
l
l
J
N
T
U
W
o
r
l
d

V = δ (A + BS) 4 51.  10
–2
(h  8.65) kV
where h is the absolute humidity in gm/m
3
and varies between 4 and 20 gm/m
3
in the above relation.
The breakdown voltage is linearly related with the gap spacing and the slope of the relation B = 5.1
kV/cm and is found to be independent of the polarity of voltage. However constant A is polarity
dependent and has the values
A = 20 kV for positive polarity
= 15 kV for negative polarity of the high voltage electrode.
The accuracy of the above relation is better than 20% and, therefore, provides better accuracy
even as compared to a sphere gap.
ELECTROSTATIC VOLTMETER
The electric field according to Coulomb is the field of forces. The electric field is produced by voltage and,
therefore, if the field force could be measured, the voltage can also be measured. Whenever a voltage is
applied to a parallel plate electrode arrangement, an electric field is set up between the plates. It is possible
to have uniform electric field between the plates with suitable arrangement of the plates. The field is
uniform, normal to the two plates and directed towards the negative plate. If A is the area of the plate and E
is the electric field intensity between the plates ε the permittivity of the medium between the plates, we
know that the energy density of the electric field between the plates is given as,
Wd =
1
ε E
2
2
Consider a differential volume between the plates and parallel to the plates with area A and
thickness dx, the energy content in this differential volume Adx is
dW = Wd Adx =
1
εE
2
Adx
2
Now force F between the plates is defined as the derivative of stored electric energy along the
field direction i.e.,
F=
dW

1
εE2 A
dx 2
Now E = V/d where V is the voltage to be measured and d the distance of separation between
the plates. Therefore, the expression for force
1
ε
V
2
A
F =
d 2
2
Since the two plates are oppositely charged, there is always force of attraction between the
plates. If the voltage is time dependant, the force developed is also time dependant. In such a case the
mean value of force is used to measure the voltage. Thus
T
2
εA
2
F = 1 1 1 V (t) 1 1 2 1 V
rms
T z0 F(t)dt  Tz 2 ε d
2
A dt  2 d
2
. T z V
(t)dt  2 εA d
2
A
l
l
J
N
T
U
W
o
r
l
d

Electrostatic voltmeters measure the force based on the above equations and are arranged such
that one of the plates is rigidly fixed whereas the other is allowed to move. With this the electric field
gets disturbed. For this reason, the movable electrode is allowed to move by not more than a fraction
of a millimetre to a few millimetres even for high voltages so that the change in electric field is
negligibly small. As the force is proportional to square of Vrms, the meter can be used both for a.c. and
d.c. voltage measurement.
The force developed between the plates is sufficient to be used to measure the voltage. Various
designs of the voltmeter have been developed which differ in the construction of electrode
arrangement and in the use of different methods of restoring forces required to balance the
electrostatic force of attraction. Some of the methods are
(i) Suspension of moving electrode on one arm of a balance.
(ii) Suspension of the moving electrode on a spring.
(iii) Pendulous suspension of the moving electrode.
(iv) Torsional suspension of moving electrode.
The small movement is generally transmitted and amplified by electrical or optical methods. If
the electrode movement is minimised and the field distribution can exactly be calculated, the meter
can be used for absolute voltage measurement as the calibration can be made in terms of the
fundamental quantities of length and force.
From the expression for the force, it is clear that for a given voltage to be measured, the higher
the force, the greater is the precision that can be obtained with the meter. In order to achieve higher
force for a given voltage, the area of the plates should be large, the spacing between the plates (d)
should be small and some dielectric medium other than air should be used in between the plates. If
uniformity of electric field is to be maintained an increase in area A must be accompanied by an
increase in the area of the surrounding guard ring and of the opposing plate and the electrode may,
therefore, become unduly large specially for higher voltages. Similarly the gap length cannot be made
very small as this is limited by the breakdown strength of the dielectric medium between the plates. If
air is used as the medium, gradients upto 5 kV/cm have been found satisfactory. For higher gradients
vacuum or SF6 gas has been used.
The greatest advantage of the electrostatic voltmeter is its extremely low loading effect as only
electric fields are required to be set up. Because of high resistance of the medium between the plates,
the active power loss is negligibly small. The voltage source loading is, therefore, limited only to the
reactive power required to charge the instrument capacitance which can be as low as a few picofarads
for low voltage voltmeters.
The measuring system as such does not put any upper limit on the frequency of supply to be
measured. However, as the load inductance and the measuring system capacitance form a series
resonance circuit, a limit is imposed on the frequency range. For low range voltmeters, the upper
frequency is generally limited to a few MHz.
Fig. 4.7 shows a schematic diagram of an absolute electrostatic voltmeter. The hemispherical metal
dome D encloses a sensitive balance B which measures the force of attraction between the movable disc
which hangs from one of its arms and the lower plate P. The movable electrode M hangs with a clearance
of above 0.01 cm, in a central opening in the upper plate which serves as a guard ring. The
A
l
l
J
N
T
U
W
o
r
l
d

Fig.Schematic diagram of electrostatic voltmeter
diameter of each of the plates is 1 metre. Light reflected from a mirror carried by the balance beam
serves to magnify its motion and to indicate to the operator at a safe distance when a condition of
equilibrium is reached. As the spacing between the two electrodes is large (about 100 cms for a
voltage of about 300 kV), the uniformity of the electric field is maintained by the guard rings G which
surround the space between the discs M and P. The guard rings G are maintained at a constant
potential in space by a capacitance divider ensuring a uniform spatial potential distribution. When
voltages in the range 10 to 100 kV are measured, the accuracy is of the order of 0.01 per cent.
Hueter has used a pair of sphares of 100 cms diameter for the measurement of high voltages
utilising the electrostatic attractive force between them. The spheres are arranged with a vertical axis
and at a spacing slightly greater than the sparking distance for the particular voltage to be measured.
The upper high voltage sphere is supported on a spring and the extension of spring caused by the
electrostatic force is magnified by a lamp-mirror scale arrangement. An accuracy of 0.5 per cent has
been achieved by the arrangement.
Electrostatic voltmeters using compressed gas as the insulating medium have been developed.
Here for a given voltage the shorter gap length enables the required uniformity of the field to be
maintained with electrodes of smaller size and a more compact system can be evolved.
One such voltmeter using SF6 gas has been used which can measure voltages upto 1000 kV and
accuracy is of the order of 0.1%. The high voltage electrode and earthed plane provide uniform electric
field within the region of a 5 cm diameter disc set in a 65 cm diameter guard plane. A weighing balance
A
l
l
J
N
T
U
W
o
r
l
d

arrangement is used to allow a large damping mass. The gap length can be varied between 2.5, 5 and
10 cms and due to maximum working electric stress of 100 kV/cm, the voltage ranges can be selected
to 250 kV, 500 kV and 100 kV. With 100 kV/cm as gradient, the average force on the disc is found to
be 0.8681 N equivalent to 88.52 gm wt. The disc movements are kept as small as 1 m by the
weighing balance arrangement.
The voltmeters are used for the measurement of high a.c. and d.c. voltages. The measurement
of voltages lower than about 50 volt is, however, not possible, as the forces become too small.
GENERATING VOLTMETER
Whenever the source loading is not permitted or when direct connection to the high voltage source is
to be avoided, the generating principle is employed for the measurement of high voltages, A
generating voltmeter is a variable capacitor electrostatic voltage generator which generates current
proportional to the voltage to be measured. Similar to electrostatic voltmeter the generating voltmeter
provides loss free measurement of d.c. and a.c. voltages. The device is driven by an external constant
speed motor and does not absorb power or energy from the voltage measuring source. The principle of
operation is explained with the help of Fig. 4.8. H is a high voltage electrode and the earthed electrode
is subdivided into a sensing or pick up electrode P, a guard electrode G and a movable electrode M, all
of which are at the same potential. The high voltage electrode H develops an electric field between
itself and the electrodes P, G and M. The field lines are shown in Fig. 4.8. The electric field density σ
is also shown. If electrode M is fixed and the voltage V is changed, the field density σ would change
and thus a current i (t) would flow between P and the ground.
Fig. Principle of generating voltmeter
i (t) =
dq(t)

d
zσ (a)da
dt dt
Where σ (a) is the electric field density or charge density along some path and is assumed constant
over the differential area da of the pick up electrode. In this case σ (a) is a function of time also and ∫
da the area of the pick up electrode P exposed to the electric field.
However, if the voltage V to be measured is constant (d.c voltage), a current i(t) will flow only
if it is moved i.e. now σ (a) will not be function of time but the charge q is changing because the area
of the pick up electrode exposed to the electric field is changing. The current i(t) is given by
A
l
l
J
N
T
U
W
o
r
l
d

i(t) =
d
zA ( t )
σ (a)da  ε
d
zA ( t )
E(a)da
dt dt
where σ (a) = εE(a) and ε is the permittivity of the medium between the high voltage electrode and
the grounded electrode. The integral boundary denotes the time varying exposed area.
The high voltage electrode and the grounded electrode in fact constitute a capacitance system.
The capacitance is, however, a function of time as the area A varies with time and, therefore, the
charge q(t) is given as
q(t) = C(t)V(t)
and i(t) = dq C ( t) dV ( t)  V ( t ) dC ( t)
dt dt
dt
For d.c. voltages dV ( t)  0
dt
Hence i(t) = VdC ( t)
dt
If the capacitance varies linearly with time and reaches
reduces to zero linearly in time Tc /2, the capacitance is given
its peak value Cm is time Tc /2 and
again as
C(t) = 2
Cm
t Tc
For a constant speed of n rpm of synchronous motor which is varying the capacitance, time Tc
is given by Tc = 60/n.
Therefore I = 2C V n  n C V
60 30 m
m
If the capacitance C varies sinusoidally between the limits C0 and (C0 + Cm) then
C = C0 + Cm sin
wt and the current i is then given as
i(t) = im cos wt where im =
VCmω Here ω is the angular frequency of variation
of the capacitance. If ω is constant, the current
measured is proportional to the voltage being
measured. Generally the current is rectified and
measured by a moving coil meter. Generating
voltmeters can be used for a.c. voltage
measurement also provided the angular frequency
ω is the same or equal to half that of the voltage
being measured. Fig 4.9 shows the variations of C
as a function of time together with a.c. voltage,
Fig. Capacitance and voltage variation
the frequency of which is twice the frequency of
C (t).
It can be seen from Fig. 4.9 that whatever be the phase relation between voltage and the
capacitance, over one cycle variation of the voltage is same (e.g. V(t1) – V(t2)) and the rate of change of
A
l
l
J
N
T
U
W
o
r
l
d

capacitance over the period Tv is equal to Cm/Tv. Therefore, the instantaneous value of current i(t) =
Cm fvV(t) where fv = 1/Tv the frequency of voltage.
Since fv = 2fc and fc = 60/n we obtain
I(t) = n/30 CmV(t)
Fig. 4.10 shows a schematic diagram of a generating voltmeter which employs rotating vanes
for variation of capacitance. The high voltage electrode is connected to a disc electrode D3 which is
kept at a fixed distance on the axis of the other low voltage electrodes D2, D1, and D0. The rotor D0 is
driven at a constant speed by a synchronous motor at a suitable speed. The rotor vanes of D0 cause
periodic change in capacitance between the insulated disc D2 and the high voltage electrode D 3. The
number and shape of vanes are so designed that a suitable variation of capacitance (sinusodial or
linear) is achieved. The a.c. current is rectified and is measured using moving coil meters. If the
current is small an amplifier may be used before the current is measured.
Fig. Schematic diagram of generating voltmeter
Generating voltmeters are linear scale instruments and applicable over a wide range of
voltages. The sensitivity can be increased by increasing the area of the pick up electrode and by using
amplifier circuits.
The main advantages of generating voltmeters are (i) scale is linear and can be extrapolated (ii)
source loading is practically zero (iii) no direct connection to the high voltage electrode.
However, they require calibration and construction is quite cumbersome.
The breakdown of insulating materials depends upon the magnitude of voltage applied and the time
of application of voltage. However, if the peak value of voltage is large as compared to breakdown strength
of the insulating material, the disruptive discharge phenomenon is in general caused by the instantaneous
maximum field gradient stressing the material. Various methods discussed so far can measure peak
voltages but because of complex calibration procedures and limited accuracy call for
A
l
l
J
N
T
U
W
o
r
l
d

more convenient and more accurate methods. A more convenient though less accurate method would
be the use of a testing transformer wherein the output voltage is measured and recorded and the input
voltage is obtained by multiplying the output voltage by the transformation ratio. However, here the
output voltage depends upon the loading of the secondary winding and wave shape variation is caused
by the transformer impedances and hence this method is unacceptable for peak voltage measurements.
THE CHUBB-FORTESCUE METHOD
Chubb and Fortescue suggested a simple and accurate method of measuring peak value of a.c.
voltages. The basic circuit consists of a standard capacitor, two diodes and a current integrating
ammeter (MC ammeter) as shown in Fig. 4.11 (a).
v (t)
C
i c (t)
D
1
D
2
A
(a)
C
Rd
D
1
D
2
A
(b)
Fig. (a) Basic circuit (b) Modified circuit
The displacement current ic(t), Fig. 4.12 is given by the rate of change of the charge and hence
the voltage V(t) to be measured flows through the high voltage capacitor C and is subdivided into
positive and negative components by the back to back connected diodes. The voltage drop across these
diodes can be neglected (1 V for Si diodes) as compared with the voltage to be measured. The
measuring instrument (M.C. ammeter) is included in one of the branches. The ammeter reads the mean
value of the current.
1 t2 dv (t) C I
I = zt1 C . dt  . 2Vm  2Vm fC or Vm 
T dt T 2 fC
The relation is similar to the one obtained in case of generating voltmeters. An increased
current would be obtained if the current reaches zero more than once during one half cycle. This
means the wave shapes of the voltage would contain more than one maxima per half cycle. The
standard a.c. voltages for testing should not contain any harmonics and, therefore, there could be very
short and rapid voltages caused by the heavy predischarges, within the test circuit which could
introduce errors in measurements. To eliminate this problem filtering of a.c. voltage is carried out by
introducing a damping resistor in between the capacitor and the diode circuit, Fig. 4.11 (b).
A
l
l
J
N
T
U
W
o
r
l
d

Fig. 4.12
Also, if full wave rectifier is used instead of the half wave as shown in Fig. 4.11, the factor 2 in
the denominator of the above equation should be replaced by 4. Since the frequency f, the capacitance
C and current I can be measured accurately, the measurement of symmetrical a.c. voltages using
Chubb and Fortescue method is quite accurate and it can be used for calibration of other peak voltage
measuring devices.
Fig. 4.13 shows a digital peak voltage measuring circuit. In contrast to the method discussed
just now, the rectified current is not measured directly, instead a proportional analog voltage signal is
derived which is then converted into a proportional medium frequency for using a voltage to frequency
convertor (Block A in Fig. 4.13). The frequency ratio fm/f is measured with a gate circuit controlled by
the a.c. power frequency (supply frequency f) and a counter that opens for an adjustable number of
period ∆t = p/f. The number of cycles n counted during this interval is
Fig. 4.13 Digital peak voltmeter
n = ∆tfm =
p
fm
f
where p is a constant of the instrument.
Let A = fm  fm  fm . 1
R2Vm fC
Ric f 2 RVm C
Therefore, n = p 2ARVmC
where A represents the voltage to frequency conversion factor.
A
l
l
J
N
T
U
W
o
r
l
d

Thus the indicator can be calibrated to read Vm directly by selecting suitable values of A, p and
R.
The voltmeter is found to given an accuracy of 0.35%.
Peak Voltmeters with Potential Dividers
Passive circuits are not very frequently used these days for measurement of the peak value of a.c. or
impulse voltages. The development of fully integrated operational amplifiers and other electronic
circuits has made it possible to sample and hold such voltages and thus make measurements and,
therefore, have replaced the conventional passive circuits. However, it is to be noted that if the passive
circuits are designed properly, they provide simplicity and adequate accuracy and hence a small
description of these circuits is in order. Passive circuits are cheap,
reliable and have a high order of electromagnetic
compatibility. However, in contrast, the most
sophisticated electronic instruments are costlier and
their electromagnetic compatibility (EMC) is low.
The passive circuits cannot measure high
voltages directly and use potential dividers
preferably of the capacitance type.
Fig. 4.14 shows a simple peak voltmeter
circuit consisting of a capacitor voltage divider
which reduces the voltage V to be measured to a Fig. Peak voltmeter
low voltage Vm.
Suppose R2 and Rd are not present and the supply voltage is V. The voltage across the storage
capacitor Cs will be equal to the peak value of voltage across C2 assuming voltage drop across the diode to
be negligibly small. The voltage could be measured by an electrostatic voltmeter or other suitable
voltmeters with very high input impedance. If the reverse current through the diode is very small and the
discharge time constant of the storage capacitor very large, the storage capacitor will not discharge
significantly for a long time and hence it will hold the voltage to its value for a long time. If now, V is
decreased, the voltage V2 decreases proportionately and since now the voltage across C2 is smaller than the
voltage across Cs to which it is already charged, therefore, the diode does not conduct and the voltage
across Cs does not follow the voltage across C2. Hence, a discharge resistor Rd must be introduced into the
circuit so that the voltage across Cs follows the voltage across C2. From measurement point of view it is
desirable that the quantity to be measured should be indicated by the meter within a few seconds and hence
Rd is so chosen that RdCs ≈ 1 sec. As a result of this, following errors are introduced. With the connection
of Rd, the voltage across Cs will decrease continuously even when the input voltage is kept constant. Also,
it will discharge the capacitor C2 and the mean potential of V2(t) will gain a negative d.c. component.
Hence a leakage resistor R2 must be inserted in parallel with C2 to equalise these unipolar discharge
currents. The second error corresponds to the voltage shape across the storage capacitor which contains
ripple and is due to the discharge of the capacitor Cs. If the input impedance of the measuring device is
very high, the ripple is independent of the meter being used. The error is approximately proportional to the
ripple factor and is thus frequency dependent as the discharge time-constant cannot be changed. If RdCs = 1
sec, the discharge error amounts to 1% for 50 Hz and 0.33%.
A
l
l
J
N
T
U
W
o
r
l
d

for 150 Hz. The third source of error is related to this discharge error. During the conduction time (when
the voltage across Cs is lower than that across C2 because of discharge of Cs through Rd) of the diode the
storage capacitor Cs is recharged to the peak value and thus Cs becomes parallel with C2. If
discharge error is ed, recharge error er is given by
er  2ed
Cs
C  C  C
1 2 s
Hence Cs should be small as compared
with C2 to keep down the recharge error.
It has also been observed that in order to
keep the overall error to a low value, it is desirable
to have a high value of R2. The same effect can be
obtained by providing an equalising arm to the low
voltage arm of the voltage divider as shown in Fig.
4.15. This is accomplished by the addition of
a second network comprising diode, Cs and Rd for negative polarity currents to the circuit shown in
Fig. 4.14. With this, the d.c. currents in both branches are opposite in polarity and equalise each other.
The errors due to R2 are thus eliminated.
Rabus developed another circuit shown in Fig. 4.16. to reduce errors due to resistances. Two
storage capacitors are connected by a resistor Rs within every branch and both are discharged by only
one resistance Rd.
D
2 R
s
D2 D1 D1
Cs2 Cs1 R
d
R
d
Cs Cs V
m
1 2
Fig. Two-way booster circuit designed by Rabus
Here because of the presence of Rs, the discharge of the storage capacitor Cs2 is delayed and hence
the inherent discharge error ed is reduced. However, since these are two storage capacitors within one
branch, they would draw more charge from the capacitor C2 and hence the recharge error er would
increase. It is, therefore, a matter of designing various elements in the circuit so that the total sum of all the
Fig. Modified peak voltmeter circuit
A
l
l
J
N
T
U
W
o
r
l
d

errors is a minimum. It has been observed that with the commonly used circuit elements in the voltage
dividers, the error can be kept to well within about 1% even for frequencies below 20 Hz.
The capacitor C1 has to withstand high voltage to be measured and is always placed within the
test area whereas the low voltage arm C2 including the peak circuit and instrument form a measuring
unit located in the control area. Hence a coaxial cable is always required to connect the two areas. The
cable capacitance comes parallel with the capacitance C2 which is usually changed in steps if the
A
l
l
J
N
T
U
W
o
r
l
d

voltage to be measured is changed. A change of the length of the cable would, thus, also require
recalibration of the system. The sheath of the coaxial cable picks up the electrostatic fields and thus
prevents the penetration of this field to the core of the conductor. Also, even though transient
magnetic fields will penetrate into the core of the cable, no appreciable voltage (extraneous of noise)
is induced due to the symmetrical arrangement and hence a coaxial cable provides a good connection
between the two areas. Whenever, a discharge takes place at the high voltage end of capacitor C1 to
the cable connection where the current looks into a change in impedance a high voltage of short
duration may be built up at the low voltage end of the capacitor C1 which must be limited by using an
over voltage protection device (protection gap). These devices will also prevent complete damage of
the measuring circuit if the insulation of C1 fails.
IMPULSE VOLTAGE MEASUREMENTS USING VOLTAGE DIVIDERS
If the amplitudes of the impulse voltage is not high and is in the range of a few kilovolts, it is possible
to measure them even when these are of short duration by using CROS. However, if the voltages to be
measured are of high magnitude of the order of magavolts which normally is the case for testing and
research purposes, various problems arise. The voltage dividers required are of special design and
need a thorough understanding of the interaction present in these voltage dividing systems. Fig. 4.17
shows a layout of a voltage testing circuit within a high voltage testing area. The voltage generator G
is connected to a test object—T through a lead L.
Fig. 4.17 Basic voltage testing circuit
These three elements form a voltage generating system. The lead L consists of a lead wire and
a resistance to damp oscillation or to limit short-circuit currents if of the test object fails. The
measuring system starts at the terminals of the test object and consists of a connecting lead CL to the
voltage divider D. The output of the divider is fed to the measuring instrument (CRO etc.) M. The
appropriate ground return should assure low voltage drops for even highly transient phenomena and
keep the ground potential of zero as far as possible.
It is to be noted that the test object is a predominantly capacitive element and thus this forms an
oscillatory circuit with the inductance of the load. These oscillations are likely to be excited by any
steep voltage rise from the generator output, but will only partly be detected by the voltage divider. A
resistor in series with the connecting leads damps out these oscillations. The voltage divider should
always be connected outside the generator circuit towards the load circuit (Test object) for accurate
measurement. In case it is connected within the generator circuit, and the test object discharges
(chopped wave) the whole generator including voltage divider will be discharged by this short circuit
at the test object and thus the voltage divider is loaded by the voltage drop across the lead L. As a
result, the voltage measurement will be wrong.
Yet for another reason, the voltage divider should be located away from the generator circuit.
The dividers cannot be shielded against external fields. All objects in the vicinity of the divider which
A
l
l
J
N
T
U
W
o
r
l
d

may acquire transient potentials during a test will disturb the field distribution and thus the divider
performance. Therefore, the connecting lead CL is an integral part of the potential divider circuit.
In order to avoid electromagnetic interference between the measuring instrument M and C the high
voltage test area, the length of the delay cable should be adequately chosen. Very short length of the cable
can be used only if the measuring instrument has high level of electromagnetic compatibility (EMC). For
any type of voltage to be measured, the cable should be co-axial type. The outer conductor provides a
shield against the electrostatic field and thus prevents the penetration of this field to the inner conductor.
Even though, the transient magnetic fields will penetrate into the cable, no appreciable voltage is induced
due to the symmetrical arrangement. Ordinary coaxial cables with braided shields may well be used for d.c.
and a.c. voltages. However, for impulse voltage measurement double shielded cables with predominently
two insulated braided shields will be used for better accuracy.
During disruption of test object, very heavy transient current flow and hence the potential of
the ground may rise to dangerously high values if proper earthing is not provided. For this, large metal
sheets of highly conducting material such as copper or aluminium are used. Most of the modern high
voltage laboratories provide such ground return along with a Faraday Cage for a complete shielding of
the laboratory. Expanded metal sheets give similar performance. At least metal tapes of large width
should be used to reduce the impedance.
Voltage Divider
Voltages dividers for a.c., d.c. or impulse voltages may consist of resistors or capacitors or a
convenient combination of these elements. Inductors are normally not used as voltage dividing
elements as pure inductances of proper magnitudes without stray capacitance cannot be built and also
these inductances would otherwise form oscillatory circuit with the inherent capacitance of the test
object and this may lead to inaccuracy in measurement and high voltages in the measuring circuit. The
height of a voltage divider depends upon the flash over voltage and this follows from the rated
maximum voltage applied. Now, the potential distribution may not be uniform and hence the height
also depends upon the design of the high voltage electrode, the top electrode. For voltages in the
megavolt range, the height of the divider becomes large. As a thumb rule following clearances
between top electrode and ground may be assumed.
2.5 to 3 metres/MV for d.c. voltages.
2 to 2.5 m/MV for lightning impulse voltages.
More than 5 m/MV rms for a.c. voltages.
More than 4 m/MV for switching impulse voltage.
The potential divider is most simply represented by two
impedances Z1 and Z2 connected in series and the sample voltage
required for measurement is taken from across Z2, Fig. 4.18.
If the voltage to be measured is V1 and sampled voltage V2,
then
Z2
Fig. 4.18 Basic diagram of a poten-
V2 = V1 tial divider circuit

Z
1
Z
2
If the impedances are pure resistances
V
2 =
R
2 V
1
R1 R2
A
l
l
J
N
T
U
W
o
r
l
d

and in case pure capacitances are used
V2 =
C
1
V1
C1  C2
The voltage V2 is normally only a few hundred volts and hence the value of Z2 is so chosen
that V2 across it gives sufficient deflection on a CRO. Therefore, most of the voltage drop is available
across the impedance Z1 and since the voltage to be measured is in megavolt the length of Z1 is large
which result in inaccurate measurements because of the stray capacitances associated with long length
voltage dividers (especially with impulse voltage measurements) unless special precautions are taken.
On the low voltage side of the potential dividers where a screened cable of finite length has to be
employed for connection to the oscillograph other errors and distortion of wave shape can also occur.
Resistance Potential Dividers
The resistance potential dividers are the first to appear because of their simplicity of construction, less
space requirements, less weight and easy portability. These can be placed near the test object which
might not always be confined to one location.
The length of the divider depends upon two or three factors. The maximum voltage to be
measured is the first and if height is a limitation, the length can be based on a surface flash over
gradient in the order of 3–4 kV/cm irrespective of whether the resistance R1 is of liquid or wirewound
construction. The length also depends upon the resistance value but this is implicitly bound up with
the stray capacitance of the resistance column, the product of the two (RC) giving a time constant the
value of which must not exceed the duration of the wave front it is required to record.
It is to be noted with caution that the resistance of the potential divider should be matched to
the equivalent resistance of a given generator to obtain a given wave shape.
Fig. 4.19 (a) shows a common form of resistance potential divider used for testing purposes
where the wave front time of the wave is less than 1 micro sec.
R1 R1 R1
R
3 Z Z R3 Z
V1
R
2V2
R
2 R4
R
2 R4
(a) (b) (c)
Fig. 4.19 Various forms of resistance potential dividers recording circuits (a) Matching at divider end
(b) Matching at Oscillograph end (c) Matching at both ends of delay cable
Here R3, the resistance at the divider end of the delay cable is chosen such that R2 + R3 = Z
which puts an upper limit on R2 i.e., R2 < Z. In fact, sometimes the condition for matching is given as
Z = R3 +
R1 R2
R1  R2
A
l
l
J
N
T
U
W
o
r
l
d

But, since usually R1 > > R2, the above relation reduces to Z = R3 + R2. From Fig. 4.19 (a), the
voltage appearing across R2 is
V2 =
Z1 V1
Z1 R1
where Z1 is the equivalent impedance of R2 in parallel with (Z + R3), the surge impedance of the cable
being represented by an impedance Z to ground.
Now Z = (Z  R3 )R2 
(Z  R3 )R2
R2  Z  R3
1 2Z
Therefore, V2 =
(Z  R3 )R2 V1
2Z Z1  R1
However, the voltage entering the delay cable is
V =
V
2
Z  Z (Z  R3 )R2 . V1  V R2
3
1
Z  R3 Z  R3 2Z Z1  R1 2(Z1  R1 )
As this voltage wave reaches the CRO end of the delay cable, it suffers reflections as the
impedance offered by the CRO is infinite and as a result the voltage wave transmitted into the CRO is
doubled. The CRO, therefore, records a voltage
V ′ =
R2
V
1
3
Z1  R1
The reflected wave, however, as it reaches the low voltage arm of the potential divider does not
suffer any reflection as Z = R2 + R3 and is totally absorbed by (R2 + R3).
Since R2 is smaller than Z and Z1 is a parallel combination of R2 and (R3 + Z), Z1 is going to be smaller
than R2 and since R1 > > R2, R1 will be much greater than Z1 and, therefore to a first approximation
Z1 + R1 ≈ R1.
Therefore, V ′ = R2 V ≈ R2 V as R 2
< < R 1
3
R
1
1
R1 
1
R
2
Fig. 4.19 (b) and (c) are the variants of the potential divider circuit of Fig. 4.19 (a). The cable
matching is done by a pure ohmic resistance R4 = Z at the end of the delay cable and, therefore, the
voltage reflection coefficient is zero i.e. the voltage at the end of the cable is transmitted completely
into R4 and hence appears across the CRO plates without being reflected. As the input impedance of
the delay cable is R4 = Z, this resistance is a parallel to R2 and forms an integral part of the divider‘s
low voltage arm. The voltage of such a divider is, therefore, calculated as follows:
Equivalent impedance
= R1 +
R2 Z

R1 (R2  Z)  R2 Z
R2  Z (R2  Z )
Therefore, Current I =
V1 (R2  Z )
R1 (R2  Z )  R2 Z
and voltage V2 = IR2 Z  V1 (R2  Z ) R2 Z
R2  Z R1 (R2  Z )  R2 Z R2  Z
A
l
l
J
N
T
U
W
o
r
l
d

=
R2 Z
V1
R (R  Z )  R Z
1 2 2
or voltage ratio
V2

R2 Z
V1 R1 (R2  Z )  R2 Z
Due to the matching at the CRO end of the delay cable, the voltage does not suffer any reflection
at that end and the voltage recorded by the CRO is given as
V
2
=
R2 Z V1

R2 ZV1

R2 V1
R ( R  Z )  R Z ( R  R )Z  R R
( R1  R2 ) 
R1 R2
1 2 2 1 2 1 2
Z
Normally for undistorted wave shape through the cable
Z ≈ R2
Therefore,
V2 =
R2
V1
2R1  R2
For a given applied voltage V1 this arrangement will produce a smaller deflection on the CRO
plates as compared to the one in Fig. 4.19 (a).
The arrangement of Fig. 4.19 (c) provides for matching at both ends of the delay cable and is to
be recommended where it is felt necessary to reduce to the minimum irregularities produced in the
delay cable circuit. Since matching is provided at the CRO end of the delay cable, therefore, there is
no reflection of the voltage at that end and the voltage recorded will be half of that recorded in the
arrangement of Fig. 4.19 (a) viz.
V
2 =
R2
V1
2(R1  R2 )
It is desirable to enclose the low voltage resistance (s) of the potential dividers in a metal
screening box. Steel sheet is a suitable material for this box which could be provided with a detachable
close fitting lid for easy access. If there are two low voltage resistors at the divider position as in Fig.
4.19 (a) and (c), they should be contained in the screening box, as close together as possible, with a
removable metallic partition between them. The partition serves two purposes (i) it acts as an
electrostatic shield between the two resistors (ii) it facilitates the changing of the resistors. The lengths
of the leads should be short so that practically no inductance is contributed by these leads. The
screening box should be fitted with a large earthing terminal. Fig. 4.20 shows a sketched cross-section
of possible layout for the low voltage arm of voltage divider.
Circuit elements
From high voltage arm
R2, C2
Matching Metal
casing
impedance if reqd.
Delay cable
Fig. Cross-section of low voltage arm of a voltage divider
A
l
l
J
N
T
U
W
o
r
l
d

Capacitance Potential Dividers
Capacitance potential dividers are more complex than the resistance type. For measurement of impulse
voltages not exceeding 1 MV capacitance dividers can be both portable and transportable. In general, for
measurement of 1 MV and over, the capacitance divider is a laboratory fixture. The capacitance dividers
are usually made of capacitor units mounted one above the other and bolted together. It is this failure which
makes the small dividers portable. A screening box similar to that described earlier can be used for housing
both the low voltage capacitor unit C2 and the matching resistor if required.
The low voltage capacitor C2 should be non-inductive. A form of capacitor which has given
excellent results is of mica and tin foil plate, construction, each foil having connecting tags coming out at
opposite corners. This ensures that the current cannot pass from the high voltage circuit to the delay cable
without actually going through the foil electrodes. It is also important that the coupling between the high
and low voltage arms of the divider be purely capacitive. Hence, the low voltage arm should contain one
capacitor only; two or more capacitors in parallel must be avoided because of appreciable inductance that
would thus be introduced. Further, the tappings to the delay cable must be taken off as close as possible to
the terminals of C2. Fig. 4.21 shows variants of capacitance potential dividers.
C1 C1
R1
R Z, Cd R Cd C1 ( Z – R2 )Z
3
C2 C2
C4 R2
R4 C
2
(a) (b) (c)
Fig. Capacitor dividers (a) Simple matching (b) Compensated matching (c)
Damped capacitor divider simple matching
For voltage dividers in Fig. (b) and (c), the delay cable cannot be matched at its end. A low
resistor in parallel to C2 would load the low voltage arm of the divider too heavily and decrease the
output voltage with time. Since R and Z form a potential divider and R = Z, the voltage input to the
cable will be half of the voltage across the capacitor C2. This halved voltages travels towards the open
end of the cable (CRO end) and gets doubled after reflection. That is, the voltage recorded by the CRO
is equal to the voltage across the capacitor C2. The reflected wave charges the cable to its final voltage
magnitude and is absorbed by R (i.e. reflection takes place at R and since R = Z, the wave is
completely absorbed as coefficient of voltage reflection is zero) as the capacitor C2 acts as a short
circuit for high frequency waves. The transformation ratio, therefore, changes from the value:
C1  C2
C
1
for very high frequencies to the value
C1  C2  Cd
C
1
for low frequencies.
A
l
l
J
N
T
U
W
o
r
l
d

However, the capacitance of the delay cable Cd is usually small as compared with C2.
For capacitive divider an additional damping resistance is usually connected in the lead on the high
voltage side as shown in Fig. 4.21 (c). The performance of the divider can be improved if damping resistor
which corresponds to the aperiodic limiting case is inserted in series with the individual element of
capacitor divider. This kind of damped capacitive divider acts for high frequencies as a resistive divider and
for low frequencies as a capacitive divider. It can, therefore, be used over a wide range of frequencies i.e.
for impulse voltages of very different duration and also for alternating voltages.
Fig. Simplified diagram of a resistance potential divider
Fig. 4.22 shows a simplified diagram of a resistance potential divider after taking into
considerations the lead in connection as the inductance and the stray capacitance as lumped
capacitance. Here L represents the loop inductance of the lead-in connection for the high voltage arm.
The damping resistance Rd limits the transient overshoot in the circuit formed by test object, L, Rd and
C. Its value has a decided effect on the performance of the divider. In order to evaluate the voltage
transformation of the divider, the low voltage arm voltage V2 resulting from a square wave impulse V1
on the hv side must be investigaged. The voltage V2 follows curve 2 in Fig. 4.23 (a) in case of
aperiodic damping and curve 2 in Fig. 4.23 (b) in case of sub-critical damping. The total area between
curves 1 and 2 taking into consideration the polarity, is described as the response time.
1
2
1 –
–
V2 (t)
2 V2 (t) + +
+
(a) t (b) t
Fig. The response of resistance voltage divider
With subcritical damping, even though the response time is smaller, the damping should not be very
small. This is because an undesirable resonance may occur for a certain frequency within the passing
frequency band of the divider. A compromise must therefore be realised between the short rise time and the
rapid stabilization of the measuring system. According to IEC publication No. 60 a maximum overshoot of
3% is allowed for the full impulse wave, 5% for an impulse wave chopped on the front at times shorter than
1 micro sec. In order to fulfill these requirements, the response time of the divider must not exceed 0.2
micro sec. for full impulse waves 1.2/50 or 1.2/5 or impulse waves chopped on the
A
l
l
J
N
T
U
W
o
r
l
d

tail. If the impulse wave is chopped on the front at time shorter than 1 micro sec the response time
must be not greater than 5% of the time to chopping.
Klydonograph or Surge Recorder
Since lightning surges are infrequent and random in nature, it is necessary to instal a large number of
recording devices to obtain a reasonable amount of data regarding these surges produced on
transmission lines and other equipments. Some fairly simple devices have been developed for this
purpose. Klydonograph is one such device which makes use of the patterns known as Litchenberg
figures which are produced on a photographic film by surface corona discharges.
The Klydonograph (Fig. 4.24) consists of a rounded electrode resting upon the emulsion side of
a photographic film or plate which is kept on the smooth surface of an insulating material plate backed
by a plate electrode. The minimum critical voltage to produce a figure is about 2 kV and the maximum
voltage that can be recorded is about 20 kV, as at higher voltages spark overs occurs which spoils the
film. The device can be used with a potential divider to measure higher voltages and with a resistance
shunt to measure impulse current.
Locking
ring
Keramot
cap
Plate
electrode
Top plate connected to
potential divider tapping
Electrode support
Removable plug
Adjustable holder
Compression spring
Stainless steel
hemispherical electrode
Photographic film (emulsion side)
Keramot backing plate
Locking ring
Electrode support
Base plate connected to earth
Positioning device
Fig. Klydonograph
There are characteristic differences between the figures for positive and negative voltages.
However, for either polarity the radius of the figure (if it is symmetrical) or the maximum distance
from the centre of the figure to its outside edge (if it is unsymmetrical) is a function only of the
applied voltage. The oscillatory voltages produce superimposed effects for each part of the wave.
Thus it is possible to know whether the wave is unidirectional or oscillatory. Since the size of the
figure for positive polarity is larger, it is preferable to use positive polarity figures. This is particularly
desirable in case of measurement of surges on transmission lines or other such equipment which are
ordinarily operating on a.c. voltage and the alternating voltage gives a black band along the centre of
the film caused by superposition of positive and negative figures produced on each half cycle. For
each surge voltage it is possible to obtain both positive and negative polarity figures by connecting
pairs of electrodes in parallel, one pair with a high voltage point and an earthed plate and the other
pair with a high voltage plate and an earthed point.
A
l
l
J
N
T
U
W
o
r
l
d

Klydonograph being a simple and inexpensive device, a large number of elements can be used
for measurement. It has been used in the past quite extensively for providing statistical data on
magnitude, polarity and frequency of voltage surges on transmission lines even though its accuracy of
measurement is only of the order of 25 per cent.
MEASUREMENT OF HIGH D.C., AND IMPULSE CURRENTS
High currents are used in power system for testing circuit breakers, cables lightning arresters etc. and
high currents are encountered during lightning discharges, switching transients and shunt faults. These
currents require special techniques for their measurements.
High Direct Currents
Low resistance shunts are used for measurement of these currents. The voltage drop across the shunt
resistance is measured with the help of a millivoltmeter. The value of the resistance varies usually
between 10 microohm and 13 milliohm. This depends upon the heating effect and the loading
permitted in the circuit. The voltage drop is limited to a few millivolts usually less than 1 V. These
resistances are oil immersed and are made as three or four terminal resistances to provide separate
terminals for voltage measurement for better accuracy.
Hall Generators
Hall effect (Fig. 4.25) is used to measure very high direct current. Whenever electric current flows
through a metal plate placed in a magnetic field perpendicular to it, Lorenz force will deflect the
electrons in the metal structure in a direction perpendicular to the direction of both the magnetic field
and the flow of current. The charge displacement results in an e.m.f. in the perpendicular direction
called the Hall voltage. The Hall voltage is proportional to the current I, the magnetic flux density B
and inversely proportional to the plate thickness d i.e.,
BI
V
H
=R
d
where R is the Hall coefficient which depends upon the material of the plate and temperature of the
plate. For metals the Hall coefficient is very small and hence semiconductor materials are used for
which the Hall coefficient is high.
B
I B I
I
d
V
H
VH (Constant)
B
R
E
(a)
(b)
Fig. Hall generator
A
l
l
J
N
T
U
W
o
r
l
d

When large d.c. currents are to be measured the current carrying conductor is passed through
an iron cored magnetic circuit (Fig. 4.25 (b)). The magnetic field intensity produced by the conductor
in the air gap at a depth d is given by
1
H =
2 π d
The Hall element is placed in the air gap and a small constant d.c. current is passed through the
element. The voltage developed across the Hall element is measured and by using the expression for
Hall voltage the flux density B is calculated and hence the value of current I is obtained.
High Power Frequency Currents
High Power frequency currents are normally measured using current transformers as use of low
resistance shunts involves unnecessary power loss. Besides, the current transformers provide isolation
from high voltage circuits and thus it is safer to work on HV circuits Fig. 4.26 shows a scheme for
current measurements using current transformers and electro-optical technique.
Fig. Current transformers and electro-optical system for high a.c. current measurements
A voltage signal proportional to the current to be measured is produced and is transmitted to
the ground through the electro-optical device. Light pulses proportional to the voltage signal are
transmit-ted by a glass optical fibre bundle to a photodetector and converted back into an analog
voltage signal. The required power for the signal convertor and optical device are obtained from
suitable current and voltage transformers.
A
l
l
J
N
T
U
W
o
r
l
d

High Frequency and Impulse Currents
In power system the amplitude of currents may vary between a few amperes to a few hundred
kiloamperes and the rate of rise of currents can be as high as 10
10
A/sec and the rise time can vary
between a few micro seconds to a few macro seconds. Therefore, the device to be used for measuring
such currents should be capable of having a good frequency response over a very wide frequency
band. The methods normally employed are—(i) resistive shunts; (ii) elements using induction effects;
(iii) Faraday and Hall effect devices. With these methods the accuracy of measurement varies between
1 to 10%. Fig. 4.27 shows the circuit diagram of the most commonly used method for high impulse
current measurement. The voltage across the shunt resistance R due to impulse current i(t) is fed to the
oscilloscope through a delay cable D. The delay cable is terminated through an impedance Z equal to
the surge impedance of the cable to avoid reflection of the voltage to be measured and thus true
measurement of the voltage is obtained. Since the dimension of the resistive element is
large, it will have residual inductance L and stray
i(t)
Z
capacitance C. The inductance could be neglected
at low frequencies but at higher frequencies the R Z v(t)
inductive reactance would be comparable with the
resistance of the shunt. The effect of inductance
and capacitance above 1 MHz usually should be
Fig. Circuit for high impulse current
considered. The resistance values range between measurement
10 micro ohm to a few milliohms and the voltage
drop is of the order of few volts. The resistive shunts used for measurements of impulse currents of
large duration is achieved only at considerable expense for thermal reasons. The resistive shunts for
impulse current of short duration can be built with rise time of a few nano seconds of magnitude. The
resistance element can be made of parallel carbon film resistors or low inductance wire resistors of
parallel resistance wires or resistance foils.
Assuming the stray capacitance to be negligibly small the voltage drop across the shunt in
complex frequency domain may be written as
V(s) = I(s)[R + Ls]
It is to be noted that in order to have flat frequency response of the resistive element the stray
inductance and capacitance associated with the element must be made as small as possible. In order to
minimise the stray field effects following designs of the resistive elements have been suggested and
used
1. Bifilar flat strip shunt.
2. Co-axial tube or Park‘s shunt
3. Co-axial squirrel cage shunt.
The bifilar flat strip shunts suffer from stray inductance associated with the resistance element and
its potential leads are linked to a small part of the magnetic flux generated by the current that is being
measured. In order to eliminate the problems associated with the bifilar shunts, coaxial shunts were
developed (Fig. 4.28). Here the current enters the inner cylinder of the shunt element and returns through
an outer cylinder. The space between the two cylinders is occupied by air which acts like a perfacts
insulator. The voltage drop across the element is measured between the potential pick up point
A
l
l
J
N
T
U
W
o
r
l
d

and the outer case. The frequency response of this element is almost a flat characteristic upto about
1000 MHz and the response time is a few nanoseconds. The upper frequency limit is governed by the
skin effect in the sensitive element.
(i)
(ii)
Fig. (i) Bifilar flat strip; (ii) Co-axial squirrel cage
Squirrel cage shunts are high ohmic shunts which can dissipate larger energies as compared to
coaxial shunts which are unsuitable due to their limitation of heat dissipation, larger wall thickness
and the skin effect. Squirrel cage shunt consists of thick metallic rods or strips placed around the
periphery of a cylinder and the structure resembles the rotor construction of a double squirrel cage
induction motor. The step response of the element is peaky and, therefore, a compensating network is
used in conjunction with the element to improve its frequency response. Rise times less than 8 n sec
and band width of 400 MHz have been obtained with these shunts.
Elements using Induction Effects
If the current to be measured is flowing through a conductor which is surrounded by a coil as shown in
Fig. 4.29, and M is the mutual inductance between the coil and the conductor, the voltage across the
coil terminals will be:
v(t) = M
di
dt
Usually the coil is wound on a non-magnetic former in the form of a toroid and has a large number of
turns, to have sufficient voltage induced which could be recorded. The coil is wound criss-cross to
reduce the leakage inductance. If M is the number of turns of the coil, A the coil area and lm its mean
length, the mutual inductance is given by
 0 NA
M =
l
Usually an integrating circuit RC is employed as shown in Fig. 4.29 to obtain the output voltage pro-
portional to the current to be measured. The output voltage is given by
1 t 1 di M M
v0(t) = z0v ( t )dt  z M
. dt  zdi  i ( t)
RC RC dt RC RC
or v(t) = RC v 0 ( t)
M
A
l
l
J
N
T
U
W
o
r
l
d

Integration of v(t) can be carried out more elegantly by using an appropriately wired
operational amplifier. The frequency response of the Rogowski coil is flat upto 100 MHz but beyond
that it is affected by the stray electric and magnetic fields and also by the skin effect.
i(t)
v(t)
v0(t)
Fig. Rogowski coil for high impulse current measurements
Magnetic Links
These are used for the measurement of peak magnitude of the current flowing in a conductor. These
links consist of a small number of short steel strips on high retentivity. The link is mounted at a known
distance from the current carrying conductor. It has been found through experiments that the remanant
magnetism of the link after impulse current of 0.5/5 micro sec shape passes through the conductor is
same as that caused by a direct current of the same peak value. Measurement of the remanance
possessed by the link after the impulse current has passed through the conductor enables to calculate
the peak value of the current. For accurate measurements, it is usual to mount two or more links at
different distances from the same conductor. Because of its relative simplicity, the method has been
used for measurement of lightning current especially on transmission towers.
It is to be noted that the magnetic links help in recording the peak value of the impulse current
but gives no information regarding the wave shape of the current. For this purpose, an instrument
called Fulcronograph has been developed which consists of an aluminium wheel round the rim of
which are slots containing magnetic links of sufficient length to project on both sides of the wheel. As
the wheel is rotated, the links pass successively through a pair of narrow coils through which flows the
current to be measured. The current at the instant during which a particular link traverses the coil, can
be determined by a subsequent measurement of the residual flux in the link and, therefore, a curve
relating the variation of current with time can be obtained. The time scale is governed by the speed of
rotation of the wheel.
Hall Generators
The high amplitude a.c. and impulse currents can be measured by Hall Generator described earlier.
For the Hall Generator, though a constant control current flows which is permeated by the magnetic
field of the current to be measured, the Hall voltage is directly proportional to the measuring current.
This method became popular with the development of semi-conductor with sufficient high value of
Hall constant. The band width of such devices is found to be about 50 MHz with suitable
compensating devices and feedback.
Faraday Generator or Magneto Optic Method
These methods of current measurement use the rotation of the plane of polarisation in materials by the
magnetic field which is proportional to the current (Faraday effect). When a linearly polarised light
A
l
l
J
N
T
U
W
o
r
l
d

beam passes through a transparent crystal in the presence of a magnetic field, the plane of polarisation
of the light beam undergoes relation. The angle of rotation is given by
θ  α Bl
where α = A constant of the cyrstal which is a function of the wave length of the light.
B = Magnetic flux density due to the current to be measured in this case. l
= Length of the crystal.
C
i(t)
P
1
P
2
F
L
CRO
PM
Fig. Magneto-optical method
Fig. 4.30 shows a schematic diagram of Magneto-optic method. Crystal C is placed parallel to the
magnetic field produced by the current to be measured. A beam of light from a stabilised light source is
made incident on the crystal C after it is passed through the polariser P1. The light beam undergoes rotation
of its plane of polarisation. After the beam passes through the analyser P2, the beam is focussed on a
photomultiplier, the output of which is fed to a CRO. The filter F allows only the monochromatic light to
pass through it. Photoluminescent diodes too, the momentary light emission of which is proportional to the
current flowing through them, can be used for current measurement. The following are the advantages of
the method (i) It provides isolation of the measuring set up from the main current circuit. (ii) It is
insensitive to overloading. (iii) As the signal transmission is through an optical system no insulation
problem is faced. However, this device does not operate for d.c. current.
A
l
l
J
N
T
U
W
o
r
l
d

UNIT-VI
OVER VOLTAGE PHENOMENON AND INSULATION CO-ORDINATION
OVER VOLTAGE DUE TO ARCING GROUND
Fig. 7.21 shows a 3-phase system with isolated neutral. The shunt capacitances are also shown. Under
balanced conditions and complete transposed transmission lines, the potential of the neutral is near the
ground potential and the currents in various phases through the shunt capacitors are leading their cor-
responding voltages by 90°. They are displaced from each other by 120° so that the net sum of the
three currents is zero (Fig. 7.21). Say there is line-to-ground fault on one of the three phases (say phase
‗c‘). The voltage across the shunt capacitor of that phase reduces to zero whereas those of the healthy
phases become line-to-line voltages and now they are displaced by 60° rather than 120°. The net
charging current now is three times the phase current under balanced conditions (Fig. 7.21 (c)). These
currents flow through the fault and the windings of the alternator. The magnitude of this current is
often suffi-cient to sustain an arc and, therefore, we have an arcing ground. This could be due to a
flashover of a support insulator. Here this flashover acts as a switch. If the arc extinguishes when the
current is pass-ing through zero value, the capacitors in phases a and b are charged to line voltages.
The voltage across the line and the grounded points of the post insulator will be the super-position of
the capacitor voltage and the generator voltage and this voltage may be good enough to cause
flashover which is equivalent to restrike in a circuit breaker. Because of the presence of the inductance
of the generator winding, the capacitances will form an oscillatory circuit and these oscillations may
build up to still higher voltages and the arc may reignite causing further transient disturbances which
may finally lead to complete rupture of the post insulators.
a
b
c
Va
3 Ic Va
Ib
Ia
N
Vb
60°
Vc Vb
I c
E, Vc
Fig. (a) 3-phase system with isolated neutral; (b) Phasor diagram under
healthy condition; (c) Phasor diagram under faulted condition.
LIGHTNING PHENOMENON
Lightning has been a source of wonder to mankind for thousands of years. Schonland points out that
any real scientific search for the first time was made into the phenomenon of lightning by Franklin in
18th century.
A
l
l
J
N
T
U
W
o
r
l
d

Before going into the various theories explaining the charge formation in a thunder cloud and
the mechanism of lightning, it is desirable to review some of the accepted facts concerning the thunder
cloud and the associated phenomenon.
1. The height of the cloud base above the surrounding ground level may vary from 160 to
9,500 m. The charged centres which are responsible for lightning are in the range of 300 to 1500 m.
2. The maximum charge on a cloud is of the order of 10 coulombs which is built up
exponentially over a period of perhaps many seconds or even minutes.
3. The maximum potential of a cloud lies approximately within the range of 10 MV to 100 MV.
4. The energy in a lightning stroke may be of the order of 250 kWhr.
5. Raindrops:
(a) Raindrops elongate and become unstable under an electric field, the limiting diameter
being 0.3 cm in a field of 100 kV/cm.
(b) A free falling raindrop attains a constant velocity with respect to the air depending upon its
size. This velocity is 800 cms/sec. for drops of the size 0.25 cm dia. and is zero for spray.
This means that in case the air currents are moving upwards with a velocity greater than
800 cm/sec, no rain drop can fall.
(c) Falling raindrops greater than 0.5 cm in dia become unstable and break up into smaller
drops.
(d) When a drop is broken up by air currents, the water particles become positively charged
and the air negatively charged.
(e) When an ice crystal strikes with air currents, the ice crystal is negatively charged and the
air positively charged.
Wilson’s Theory of Charge Separation
Wilson‘s theory is based on the assumption that a large number of ions are present in the atmosphere. Many
of these ions attach themselves to small dust particles and water particles. It also assumes that an electric
field exists in the earth‘s atmosphere during fair weather which is directed downwards towards the earth
(Fig. 7.22(a)). The intensity of the field is approximately 1 volt/cm at the surface of the earth and decreases
gradually with height so that at 9,500 m it is only about 0.02 V/cm. A relatively large raindrop (0.1 cm
radius) falling in this field becomes polarized, the upper side acquires a negative
Electric field
+ + + +
–
– Water drop
Negative
–
+
+ + + +
+
+
+ + +
+ +
+
+
ion + +
+ + +
+
+
+
+ + + +
+ + + + + + +
–
– –
–
–
–
–
–
+
– – – – –
– – – – –
–
– ––
– –
– – – –
–
– –– – – –
– – –
– – – – – –
–
–
– – – – –
– – – – – – –
–
– – – –
(a) (b)
Fig. (a) Capture of negative ions by large falling drop; (b) Charge
separation in a thunder cloud according to Wilson’s theory.
A
l
l
J
N
T
U
W
o
r
l
d

charge and the lower side a positive charge. Subsequently, the lower part of the drop attracts –ve
charges from the atmosphere which are available in abundance in the atmosphere leaving a preponder-
ance of positive charges in the air. The upwards motion of air currents tends to carry up the top of the
cloud, the +ve air and smaller drops that the wind can blow against gravity. Meanwhile the falling
heavier raindrops which are negatively charged settle on the base of the cloud. It is to be noted that the
selective action of capturing –ve charges from the atmosphere by the lower surface of the drop is
possible. No such selective action occurs at the upper surface. Thus in the original system, both the
positive and negative charges which were mixed up, producing essentially a neutral space charge, are
now separated. Thus according to Wilson‘s theory since larger negatively charged drops settle on the
base of the cloud and smaller positively charged drops settle on the uper positions of the cloud, the
lower base of the cloud is negatively charged and the upper region is positively charged (Fig. 7.22(b)).
Simpson’s and Scarse Theory
Simpson‘s theory is based on the temperature variations in the various regions of the cloud. When
water droplets are broken due to air currents, water droplets acquire positive charges whereas the air is
negatively charged. Also when ice crystals strike with air, the air is positively charged and the crystals
negatively charged. The theory is explained with the help of Fig. 7.23.
+
+
+
+
+
+ + + +
(– 20° C)
+
+
+
+ +
+
+
+
+ +
+ +
+ + + + +
– + (– 10° C)
+ +
– – + + + +
+ + +
– –
–
– – – –
–
– – – – – – –
+
––
– – – –
–
– – – – –
–
– – – –
–
–
– – – – – – – – – –+ + + + –
– – (0° C) – – –
+ –
– – –
–
– – – – – – – –
+
+
– – – ––– – –
– –
– – – + + + + – –
Air currents
10 m/sec
Fig. Charge generation and separation in a thunder cloud
according to Simpson’s theory
Let the cloud move in the direction from left to right as shown by the arrow. The air currents
are also shown in the diagram. If the velocity of the air currents is about 10 m/sec in the base of the
cloud, these air currents when collide with the water particles in the base of the cloud, the water drops
are broken and carried upwards unless they combine together and fall down in a pocket as shown by a
pocket of positive charges (right bottom region in Fig. 7.23). With the collision of water particles we
know the air is negatively charged and the water particles positively charged. These negative charges
in the air are immediately absorbed by the cloud particles which are carried away upwards with the air
currents. The air currents go still higher in the cloud where the moisture freezes into ice crystals. The
air currents when collide with ice crystals the air is positively charged and it goes in the upper region
of cloud whereas the negatively charged ice crystals drift gently down in the lower region of the
cloud. This is how the charge is separated in a thundercloud. Once the charge separation is complete,
the conditions are now set for a lightning stroke.
A
l
l
J
N
T
U
W
o
r
l
d

Mechanism of Lightning Stroke
Lightning phenomenon is the discharge of the cloud to the ground. The cloud and the ground form
two plates of a gigantic capacitor and the dielectric medium is air. Since the lower part of the cloud is
negatively charged, the earth is positively charged by induction. Lightning discharge will require the
puncture of the air between the cloud and the earth. For breakdown of air at STP condition the electric
field required is 30 kV/cm peak. But in a cloud where the moisture content in the air is large and also
because of the high altitude (lower pressure) it is seen that for breakdown of air the electric field
required is only 10 kV/cm. The mechanism of lightning discharge is best explained with the help of
Fig.
––
–
– –– ––
– – – –
–
–
– –
–
+ + +
–
+ + + + +
(a)
+
+
+
+
+ + + + + + + + +
(c)
–– – –
– – – –
– –
–– –
–
– –– – –
– –
+
+
+
+
+
+ + + + + +
(e)
(b)
– –
–
–
–
–
+
+
+
+
+
+ + + + + + + +
(d)
– –
–
–
–
–
+
+
+
+
+
+ + + + + + +
(f)
Fig. Lightning mechanism
A
l
l
J
N
T
U
W
o
r
l
d

After a gradient of approximately 10 kV/cm is set up in the cloud, the air surrounding gets ionized.
At this a streamer (Fig. (a)) starts from the cloud towards the earth which cannot be detected with the naked
eye; only a spot travelling is detected. The current in the streamer is of the order of 100 amperes and the
speed of the streamer is 0.16 m/ sec. This streamer is known as pilot streamer because this leads to the
lightning phenomenon. Depending upon the state of ionization of the air surrounding the streamer, it is
branched to several paths and this is known as stepped leader (Fig. 7.24(b)). The leader steps are of the
order of 50 m in length and are accomplished in about a microsecond. The charge is brought from the cloud
through the already ionized path to these pauses. The air sur-rounding these pauses is again ionized and the
leader in this way reaches the earth (Fig. (c)).
Once the stepped leader has made contact with the earth it is believed that a power return stroke
(Fig. (c)) moves very fast up towards the cloud through the already ionized path by the leader. This
streamer is very intense where the current varies between 1000 amps and 200,000 amps and the speed
is about 10% that of light. It is here where the –ve charge of the cloud is being neutralized by the
positive induced charge on the earth (Fig. (d)). It is this instant which gives rise to lightning flash
which we observe with our naked eye. There may be another cell of charges in the cloud near the
neutralized charged cell. This charged cell will try to neutralize through this ionised path. This
streamer is known as dart leader (Fig. (e)). The velocity of the dart leader is about 3% of the velocity
of light. The effect of the dart leader is much more severe than that of the return stroke.
The discharge current in the return streamer is relatively very large but as it lasts only for a few
microseconds the energy contained in the streamer is small and hence this streamer is known as cold
lightning stroke whereas the dart leader is known as hot lightning stroke because even though the
current in this leader is relatively smaller but it lasts for some milliseconds and therefore the energy
contained in this leader is relatively larger.
It is found that each thunder cloud may contain as many as 40 charged cells and a heavy light-
ning stroke may occur. This is known as multiple stroke.
LINE DESIGN BASED ON LIGHTNING
The severity of switching surges for voltage 400 kV and above is much more than that due to
lightning voltages. All the same it is desired to protect the transmission lines against direct lightning
strokes. The object of good line design is to reduce the number of outages caused by lightning. To
achieve this the following actions are required.
(i) The incidence of stroke on to power conductor should be minimised.
(ii) The effect of those strokes which are incident on the system should be minimized.
To achieve (i) we know that lightning normally falls on tall objects; thus tall towers are more
vulnerable to lightning than the smaller towers. In order to keep smaller tower height for a particular
ground clearance, the span lengths will decrease which requires more number of towers and hence the
associated accessories like insulators etc. The cost will go up very high. Therefore, a compromise has
to be made so that adequate clearance is provided, at the same time keeping longer span and hence
lesser number of towers.
With a particular number of towers the chances of incidence of lightning on power conductors
can be minimized by placing a ground wire at the top of the tower structure. Refer to article 7.11 for
ground wires.
A
l
l
J
N
T
U
W
o
r
l
d

Once the stroke is incident on the ground wire, the lightning current propagates in both the
directions along the ground wire. The tower presents a discontinuity to the travelling waves; therefore
they suffer reflections and refraction. The system is, then, equivalent to a line bifurcated at the tower
point.
We know that the voltage and current transmitted into the tower will depend upon the surge
impedance of the tower and the ground impedance (tower footing resistance) of the tower. If it is low,
the wave reflected back up the tower will largely remove the potential existing due to the incident
wave. In this way the chance of flashover is eliminated. If, on the other hand, the incident wave
encoun-ters a high ground impedance, positive reflection will take place and the potential on the top of
the tower structure will be raised rather than lowered. It is, therefore, desired that for good line design
high surge impedances in the ground wire circuits, the tower structures and the tower footing should
be avoided. Various methods for lowering the tower footing resistances have been discussed in article
SWITCHING SURGE TEST VOLTAGE CHARACTERISTICS
Switching surges assume great importance for designing insulation of overhead lines operating
at voltages more than 345 kV. It has been observed that the flashover voltage for various geometircal
arrangements under unidirectional switching surge voltages decreases with increasing the front dura-
tion of the surge and the minimum switching surge corresponds to the range beetween 100 and 500 
sec. However, time to half the value has no effect as flashover takes place either at the crest or before
the crest of the switching surge. Fig. 7.25 gives the relationship between the critical flashover voltage
per metre as a function of time to flashover for on a 3 m rod-rod gap and a conductor-plane gap.
0.6
3 Peak
frequency
0.4
MV/m
0.2
0
1 10 100 1000 Time to flashover (s)
Fig. Variation of F.O. V/m as a function of time to flashover
It can be seen that the standard impulse voltage (1/50  sec) gives highest flashover voltage and
switching surge voltage with front time varyling between 100 to 500  sec has lower flashover voltage as
compared to power frequency voltage. The flashover voltage not only depends upon the crest time but
upon the gap spacing and humidity for the same crest time surges. It has been observed that the
A
l
l
J
N
T
U
W
o
r
l
d

switching surge voltage per meter gap length decreases drastically with increase in gap length and,
therefore, for ultra high voltage system, costly design clearances are required. Therefore, it is
important to know the behaviour of external insulation with different configuration under positive
switching surges as it has been found that for nearly all gap configurations which are of paractical
interest posi-tive switching impulse is lower then the negative polarity switching impulse. It has also
been observed that if the humidity varies between 3 to 16 gm/m
3
, the breakdown voltage of positive
and gaps in-creases approximately 1.7% for 1 gm/m
3
increase in absolute humidity.
For testing purposes the switching surge has been standardized with wave front time 250  sec
 20% and wave tail time 2500 to  60%  sec.
It is known that the shape of the electrode has a decided effect on the flashover voltage of the
insulation. Lot of experimental work has been carried on the switching surge flash over voltage for
long gaps using rod-plane gap and it has been attempeted to correlate these voltages with switching
surge flash over voltage of other configuration electrodes. Several investigators have shown that if the
gap length varies between 2 to 8 m, the 50% positive switching surge flash over for any configuration
is given by the expression
V50 = 500 kd
0.6
kV
where d is the gap length in metres, k is the gap factor which is a function of electrode geometry. For
rod-plane gaps K = 1.0. Thus K represents a propertionality contant and is equal to 50% flash over
voltage of any gap geometry to that of a rod-plane gap for the same gap spacing i.e.,
k=
V
50
V50 rod − plane gap
The expression for V50 applies to switching impulse of constant crest time. A more general
expression which applies to longer times to crest has been proposed as follows :
V50 =
3450 K
kV
1  8
d
here K and d have the same meaning as in the equation above. The gap factor K depends mainly on the
gap geomatry and hence on the field distribution in the gap. Table 7.1 gives values of K for different
gap configurations.
Table
Gap factor k for different configurations
Configuration Figure d = 2m 3m 4m 6m
K K K K
Rod-plane a 1 1 1 1
Conductor-plane b 1.08 1.1 1.14 1.15
Rod-rod gap c 1.27 1.26 1.21 1.14
Conductor cross arm d 1.57 1.68 1.65 1.54
Rod-structure e 1.08 — 1.07 1.06
A
l
l
J
N
T
U
W
o
r
l
d

6 cm 60 mm
dia 6 m 6 m
dia
60 mm
Semi- d
dia
spherical d
2 m
(a) (b) (c)
O 25m
25m
(d) (e)
Fig. Different gap geometries
Above two expressions for V50 and the Table 7.1 can be used to evaluate clearances in
designing extra and ultra-high voltage lines and structures.
INSULATION COORDINATION AND OVERVOLTAGE PROTECTION
Insulation coordination means the correlation of the insulation of the various equipments in a power
system to the insulation of the protective devices used for the protection of those equipments against
overvoltages. In a power system various equipments like trans-
formers, circuit breakers, bus supports etc. have different break-
down voltages and hence the volt-time characteristics. In order V
that all the equipments should be properly protected it is de-
sired that the insulation of the various protective devices must
be properly coordinated. The basic concept of insulation coor-
dination is illustrated in Fig. 7.27. Curve A is the volt-time curve B
of the protective device and B the volt-time curve of the equip- A
ment to be protected. Fig. 7.27 shows the desired positions of t
the volt-time curves of the protecting device and the equipment
Fig. Volt-time curve A
to be protected. Thus, any insulation having a withstand volt-
age strength in excess of the insulation strength of curve B is
protected by the protective device of curve A.
The ‗volt-time curve‘ expression will be used very frequently in this chapter. It is, therefore,
necessary to understand the meaning of this expression.
Volt-Time Curve
The breakdown voltage for a particular insulation of flashover voltage for a gap is a function of both the
magnitude of voltage and the time of application of the voltage. The volt-time curve is a graph showing the
relation between the crest flashover voltages and the time to flashover for a series of impulse applications
of a given wave shape. For the construction of volt-time curve the following procedure is adopted. Waves
of the same shape but of different peak values are applied to the insulation whose volt-time curve is
required. If flashover occurs on the front of the wave, the flashover point gives one point on the volt-time
curve. The other possibility is that the flashover occurs just at the peak value
curve B (device to be protected)
(protecting device and) volt-time
A
l
l
J
N
T
U
W
o
r
l
d

of the wave; this gives another point on the V-T curve. The third possibility is that the flashover occurs
on the tail side of the wave. In this case to find the point on the V-T curve, draw a horizontal line from
the peak value of this wave and also draw a vertical line passing through the point where the flashover
takes place. The intersection of the horizontal and vertical lines gives the point on the V-T curve. This
procedure is nicely shown in Fig.
Front flashover
Wave front flashover
V
o
l
t
a
g
e
Crest flashover voltage range
Volt time curve
Tail flashover
Critical flashover Critical flashover
Critical withstand
Wave tail flashover 50% of applications
voltage range 50% of applications
Rated withstand
Time of
crest
Time of critical
flashover
flashover
Time range Time range
wavefront Time range wave tail flashover no impulse
flashover Time in microseconds flashover
Fig. Volt-time curve (construction)
The overvoltages against which coordination is required could be caused on the system due to
system faults, switching operation or lightning surges. For lower voltages, normally upto about 345
kV, over voltages caused by system faults or switching operations do not cause damage to equipment
insulation although they may be detrimental to protective devices. Overvoltages caused by lightning
are of sufficient magnitude to affect the equipment insulation whereas for voltages above 345 kV it is
these switching surges which are more dangerous for the equipments than the lightning surges.
The problem of coordinating the insulation of the protective equipment involves not only guarding
the equipment insulation but also it is desired that the protecting equipment should not be damaged.
To assist in the process of insulation coordination, standard insulation levels have been
recommended. These insulation levels are defined as follows.
Basic impulse insulation levels (BIL) are reference levels expressed in impulse crest voltage
with a standard wave not longer than 1.2/50  sec wave. Apparatus insulation as demonstrated by
suitable tests shall be equal to or greater than the basic insulation level.
The problem of insulation coordination can be studied under three steps:
1. Selection of a suitable insulation which is a function of reference class voltage (i.e., 1.05
× operating voltage of the system). Table 7.2 gives the BIL for various reference class voltages.
A
l
l
J
N
T
U
W
o
r
l
d

Table Basic impulse insulation levels
Reference Standard Basic Reduced insulation
class impulse level levels
kV kV
23 150
34.5 200
46 250
69 350
92 450
115 550 450
138 650 550
161 750 650
196 900
230 1050 900
287 1300 1050
345 1550 1300
2. The design of the various equipments such that the breakdown or flashover strength of all
insulation in the station equals or exceeds the selected level as in (1).
3. Selection of protective devices that will give the apparatus as good protection as can be
justified economically.
The above procedure requires that the apparatus to be protected shall have a withstand test
value not less than the kV magnitude given in the second column of Table 7.2, irrespective of the
polarity of the wave positive or negative and irrespective of how the system was grounded.
The third column of the table gives the reduced insulation levels which are used for selecting
insulation levels of solidly grounded systems and for systems operating above 345 kV where switching
surges are of more importance than the lightning surges. At 345 kV, the switching voltage is considered to
be 2.7 p.u., i.e., 345 × 2.7 = 931.5 kV which corresponds to the lightning level. At 500 kV, however, 2.7
p.u. will mean 2.7 × 500 = 1350 kV switching voltage which exceeds the lightning voltage level. Therefore,
the ratio of switching voltage to operating voltage is reduced by using the switching resistances between
the C.B. contacts. For 500 kV it is has been possible to obtain this ratio as 2.0 and for 765 kV it is 1.7. With
further increase in operating voltages, it is hoped that the ratio could be brought to 1.5.
So, for switching voltages the reduced levels in third column
are used i.e., for 345 kV, the standard BIL is 1550 kV but if
the equipment can withstand even 1425 kV or 1300 kV it
toF.O.
Bus bar
insulation
will serve the purpose.
Fig. gives the relative position of the volt-time
pe
ak
Line
insulation
d
curves of the various equipments in a substation for proper
c
k
V
bTransformer
coordination. To illustrate the selection of the BIL of a trans- a L.A.
former to be operated on a 138 kV system assume that the Time
transformer is of large capacity and its star point is solidly
Fig. Volt-time curves
A
l
l
J
N
T
U
W
o
r
l
d

grounded. The grounding is such that the line-to-ground voltage of the healthy phase during a ground
fault on one of the phases is say 74% of the normal L-L voltage. Allowing for 5% overvoltage during
operating conditions, the arrester rms operating voltage will be 1.05 × 0.74 × 138 = 107.2 kV. The
nearest standard rating is 109 kV. The characteristic of such a L.A. is shown in Fig. 7.30. From the
figure the breakdown value of the arrester is 400 kV. Assuming a 15% margin plus 35 kV between the
insulation levels of L.A. and the transformer, the insulation level of transformer should be at least
equal to 400 + 0.15 × 400 + 35 = 495 kV. From Fig. 7.30 (or from the table the reduced level of
transformer for 138 kV is 550 kV) the insulation level of transformer is 550 kV; therefore a lightning
arrester of 109 kV rating can be applied.
kVpeak
800
700
600
500 B
400
A
300
200
100
0 0 1 2 3 4 5 6 7 8 9 10
Time  sec
Fig. Coordination of transformer insulation with lightning arrester:
A–Lightning arrester 109 kV, B–Transformer insulation withstand characteristic
It is to be noted that low voltage lines are not as highly insulated as higher voltage lines so that
lightning surges coming into the station would normally be much less than in a higher voltage station
because the high voltage surges will flashover the line insulation of low voltage line and not reach the
station.
The traditional approach to insulation coordination requires the evaluation of the highest
overvoltages to which an equipment may be subjected during operation and selection of standardized
value of withstand impulse voltage with suitable safety margin. However, it is realized that
overvoltages are a random phenomenon and it is uneconomical to design plant with such a high
degree of safety that they sustain the infrequent ones. It is also known that insulation designed on this
basis does not give 100% protection and insulation failure may occur even in well designed plants
and, therefore, it is desired to limit the frequency of insulation failures to the most economical value
taking into account equipment cost and service continuity. Insulation coordination, therefore, should
be based on evaluation and limitation of the risk of failure than on the prior choice of a safety margin.
A
l
l
J
N
T
U
W
o
r
l
d

The modern practice, therefore, is to make use of probabilistic concepts and statistical proce-
dures especially for very high voltage equipments which might later on be extended to all cases where
a close adjustment of insulation to system conditions proves economical. The statistical methods even
though laborious are quite useful.
Statistical Methods for Insulation Coordination
Both the over voltages due to lightning or switching and the breakdown strength of the insulating
media are of statistical nature. Not all lightning or switching surges are dangerous to the insulation and
particular specimen need not necessarily flashover or puncture at a particular voltage. Therefore, it is
important to design the insulation of the various equipments to be protected and the devices used for
protection not for worst possible condition but for worst probable condition as the cost of insulation
for system of the voltage more than 380 kV are proportional to square of the voltage and, therefore
any small saving in insulation will result in a large sums when considered for such large modern
power system. This, however, would involve some level of risk failure. It is desired to accept some
level of risk of failure than to design a risk-free but a very costly system.
1
Insulation
breakdown
Overvoltage probability
p0(V) distribution
p0(Vk) dVk
p0(Vk)
Vk
Fig.Overvoltage distribution and Insulation breakdown probability
The statistical methods, however, call for a very rigorous experimentation and analysis work so
as to find probability of occurence of overvoltages and probability of failure of insulation. It is found
that the distribution of breakdown for a given gap follows with some exceptions approximately
normal or Gaussian distribution. Similarly the distribution of over voltages on the system also follows
the Gaussian distribution. In order to coordinate electrical stresses due to overvoltages with the
electrical strengths of the dielectric media, it has been found convenient to represent overvoltage
distribution in the form of probability density function and the insulation breakdown probability by
the cumulative distribution function as shown in Fig. 7.31.
Suppose P0(Vk) is the probability density of an overvoltage Vk and P0 (Vk) dVk the probability
of occurence of the over voltages having a peak value Vk. To obtain the probability to disruptive dis-
charges due to these overvoltages having a value between Vk and Vk + dVk , their probability of occur-
rence P0(Vk) dVk, shall be multiplied by Pb(Vk) that an impulse of the given type and of value Vk will
produce a discharge. The resultant probability or risk of failure for overvoltage between Vk and Vk +
dVk is thus,
dR = Pb(Vk) P0(Vk) dVk
For the total voltage range we obtain the total probability of failure or risk of failure.
∞
R =
z0
P
b
(V
k
) P
0
(V
k
)dV
k
A
l
l
J
N
T
U
W
o
r
l
d

The risk of failure will thus be given by the shaded area under the curve. In actual practice,
however, it is uneconomical to use the complete distribution functions for the occurrence of
overvoltage and for the withstand of insulation and, thereforem, a compromise solution is adopted as
shown in Fig. 7.32 (a) and (b). Fig. 7.32 shows probability of occurrence of overvoltages which will
result into breakdown, by the shaded area for voltage greater than Vs known as statistical overvoltage.
1.0
p0(V)
pb(V)
0.1
Vs V
V
w
Fig. Reference probabilities for overvoltage and for insulation withstand voltage
In Fig. (b) Vw is the withstand voltage which results in flashover only in 10% of applications
and for remaining 90% of applied impulsed, no breakdown of insulation occurs. This voltage is known
as statistical withstand voltage Vw.
Fig. (a) to (c) show the functions Pb(V) and P0(V) plotted for three different cases of insulation
strength, keeping the overvoltage distribution the same. The density function P0(Vk) is the same in (a)
to (c) whereas the cumulative function giving the undetermined withstand voltage is gradu-ally shifted
along the V-axis towards high values of V. The shifting of the cumulative distribution curve to right is
equivalent to increasing the insulation strength by either using thicker insulation or material of high
dielectric strength.
p (V) p0(V)
0
1 =
vw
2 =
vw
vs
v
s
V
s
V
w R1 Vs Vw R2
(a) 1 < 2 < 3 (b)
p (V) R1
0

3 =
vw
R
2
vs
R
3
Vs Vw
R
3
1

2

3 
(c) (d)
Fig. Risk of failure as a function of statistical safety factor
A
l
l
J
N
T
U
W
o
r
l
d

Let the statistical factor of safety be defined as γ = Vw/Vs and as the withstand characteristic is
shifted towards right, the statistical factor of safety increases and hence the risk of failure decreases as
shown in Fig. 7.33 (d). However, the cost of insulation goes up as the factor of safety is increased.
OVERVOLTAGE PROTECTION
The causes of overvoltages in the system have been studied extensively in previous sections.
Basically, there are two sources: (i) external overvoltages due to mainly lightning, and (ii) internal
overvoltages mainly due to switching operation. The system can be protected against external
overvoltages using what are known as shielding methods which do not allow an arc path to form
between the line conduc-tors and ground, thereby giving inherent protection in the line design. For
protection against internal voltages normally non-shielding methods are used which allow an arc path
between the ground struc-ture and the line conductor but means are provided to quench the arc. The
use of ground wire is a shielding method whereas the use of spark gaps, and lightning arresters are the
non-shielding methods. We will study first the non-shielding methods and then the shielding methods.
However, the non-shielding methods can also be used for external over voltages.
kVpeak
1200
1050
900
750
600 Negative
450
300
Positive
150
0
0 1 2 3 4 5 6 7
Micro seconds
Fig. Volt-time curves of gaps for positive and negative polarity
The non-shielding methods are based upon the principle of insulation breakdown as the overvoltage
is incident on the protective device; thereby a part of the energy content in the overvoltage is discharged to
the ground through the protective device. The insulation breakdown is not only a function of voltage but it
depends upon the time for which it is applied and also it depends upon the shape and size of the electrodes
used. The steeper the shape of the voltage wave, the larger will be the magnitude of voltage required for
breakdown; this is because an expenditure of energy is required for the rupture of any dielectric, whether
gaseous, liquid or solid, and energy involves time. The energy
A
l
l
J
N
T
U
W
o
r
l
d

criterion for various insulations can be compared in terms of a common term known as Impulse Ratio
which is defined as the ratio of breakdown voltage due to an impulse of specified shape to the break-
down voltage at power frequency. The impulse ratio for sphere gap is unity because this gap has a
fairly uniform field and the breakdown takes place on the field ionization phenomenon mainly
whereas for a needle gap it varies between 1.5 to 2.3 depending upon the frequency and gap length.
This ratio is higher than unity because of the non-uniform field between the electrodes. The impulse
ratio of a gap of given geometry and dimension is greater with solid than with air dielectric. The
insulators should have a high impulse ratio for an economic design whereas the lightning arresters
should have a low impulse ratio so that a surge incident on the lightning arrester may be-by passed to
the ground instead of passing it on to the apparatus.
The volt-time characteristics of gaps having one electrode grounded depend upon the polarity
of the voltage wave. From Fig. 7.34 it is seen that the volt-time characteristic for positive polarity is
lower than the negative polarity, i.e. the breakdown voltage for a negative impulse is greater than for a
posi-tive because of the nearness of earthed metal or of current carrying conductors. For post
insulators the negative polarity wave has a high breakdown value whereas for suspension insulators
the reverse is true.
Horn Gap
The horn gap consists of two horn-shaped rods separated by a small distance. One end of this is con-
nected to be line and the other to the earth as shown in Fig. 7.35, with or without a series resistance.
The choke connected between the equipment to be protected and the horn gap serves two purposes: (i)
The steepness of the wave incident on the equipment to be protected is reduced. (ii) It reflects the
voltage surge back on to the horn.
Series inductance
Line
Force
Equipment
to be
protected
Fig. Horn gap connected in the system for protection
Whenever a surge voltage exceeds the breakdown value of the gap a discharge takes place and
the energy content in the rest part of the wave is by-passed to the ground. An arc is set up between the
gap, which acts like a flexible conductor and rises upwards under the influence of the electro-magnetic
forces, thus increasing the length of the arc which eventually blows out.
There are two major drawbacks of the horn gap; (i) The time of operation of the gap is quite
large as compared to the modern protective gear. (ii) If used on isolated neutral the horn gap may
constitute a vicious kind of arcing ground. For these reasons, the horn gap has almost vanished from
important power lines.
A
l
l
J
N
T
U
W
o
r
l
d

Surge Diverters
The following are the basic requirements of a surge diverter:
(i) It should not pass any current at normal or abnormal (normally 5% more than the normal
voltage) power frequency voltage.
(ii) It should breakdown as quickly as possible after the abnormal high frequency voltage arrives.
(iii) It should not only protect the equipment for which it is used but should discharge the surge
current without damaging itself.
(iv) It should interrupt the power frequency follow current after the surge is discharged to ground.
There are mainly three types of surge diverters: (i) Rod gap, (ii) Protector tube or expulsion type
of lightning arrester, (iii) Valve type of lightning arrester.
Rod gap
This type of surge diverter is perhaps the simplest, cheapest and most rugged one. Fig. 7.36 shows one
such gap for a breaker bushing. This may take the form of arcing ring. Fig. 7.37 shows the breakdown
characteristics (volt-time) of a rod gap. For a given gap and wave shape of the voltage, the time for
breakdown varies approximately inversely with the applied voltage.
1200
Conductor electrode 1000
800
kV
600
400
200
Earthed 0
0 1 2 3 4 5 6 7
Micro seconds
Fig. A rod gap Fig. Volt-time characteristic of rod gap
The time to flashover for positive polarity are lower than for negative polarities. Also it is
found that the flashover voltage depends to some extent on the length of the lower (grounded) rod. For
low values of this length there is a reasonable difference between positive (lower value) and negative
flashover voltages. Usually a length of 1.5 to 2.0 times the gap spacing is good enough to reduce this
difference to a reasonable amount. The gap setting normally chosen is such that its breakdown voltage
is not less than 30% below the voltage withstand level of the equipment to be protected.
Even though rod gap is the cheapest form of protection, it suffers from the major disadvantage that
it does not satisfy one of the basic requirements of a lightning arrester listed at no. (iv) i.e., it does not
interrupt the power frequency follow current. This means that every operation of the rod gap results
A
l
l
J
N
T
U
W
o
r
l
d

in a L-G fault and the breakers must operate to de-energize the circuit to clear the flashover. The rod
gap, therefore, is generally used as back up protection.
Expulsion type of lightning arrester
An improvement of the rod gap is the expulsion tube which consists of (i) a series gap (1) external to
the tube which is good enough to withstand normal system voltage, thereby there is no possibility of
corona or leakage current across the tube; (ii) a tube which has a fibre lining on the inner side which is
a highly gas evolving material; (iii) a spark gap (2) in the tube; and (iv) an open vent at the lower end
for the gases to be expelled (Fig. 7.38). It is desired that the breakdown
voltage of a tube must be lower than that of the insulation for which
Line
it is used. When a surge voltage is incident on the expulsion tube the
1 Series gap
series gap is spanned and an arc is formed between the electrodes
within the tube.
The heat of the arc vaporizes some of the organic material of
the tube wall causing a high gas pressure to build up in the tube. The
resulting neutral gas creates lot of turbulence within the tube and is
expelled out from the open bottom vent of the tube and it extinguishes Fibre tube
Gap
the arc at the first current zero. At this instant the rate of build up of
insulation strength is greater than the RRRV. Very high currents have
been interrupted using these tubes. The breakdown voltage of Bottom metal
expulsion tubes is slightly lower than for plain rod gaps for the same electrode
spacing. With each operation of the tube the diameter of the tube Vent for gases
(fibre lining) increases; thereby the insulation characteristics of the
Fig. Expulsion type
tube will lower down even though not materially. The volt-time
characteristics (Fig. 7.39) of the expulsion tube are somewhat better lightning arrester
than the rod gap and have the ability to interrupt power voltage after flashover.
300
34.5 kV
200
23 kV
13.8 kV
100
0
0 2 4 6 8 10 12 14 Micro seconds
Fig. Volt-time characteristic of expulsion gaps
Valve type lightning arresters
An improved but more expensive surge diverter is the valve type of lightning arrester or a non-linear
surge diverter. A porcelain bushing (Fig. 7.40) contains a number of series gaps, coil units and the
A
l
l
J
N
T
U
W
o
r
l
d

valve elements of the non-linear resistance material usually made of silicon carbide disc, the latter
possessing low resistance to high currents and high resistance to low currents. The characteristic is
usually expressed as I = KV
n
where n lies between 2 and 6 and K is constant, a function of the
geometry and dimension of the resistor. The non-linear characteristic is attributed to the properties of
the electri-cal contacts between the grains of silicon carbide. The discs are 90 mm in dia and 25 mm
thick. A grading ring or a high resistance is connected across the disc so that the system voltage is
evenly distributed over the discs. The high resistance keeps the inner assembly dry due to some heat
gener-ated.
Water-tight joint
Wet process
Porcelain housing
Solder-sealed
(metal to porcelain) wet
process porce-
lain tube
Porous blocks
making up
valve element
Ground terminal
connection
(in back not
visible)
Connector for
line conductor
Cover
Series gaps
Making up
gap element
Foundation base
Fig. Valve-type lightning arrester
Figure shows the volt-ampere characteristics of a non-linear resistance of the required type. The
closed curve represents the dynamic characteristic corresponding to the application of a voltage surge
whereas the dotted line represents the static characteristic. The voltage corresponding to the
horizontal tangent to the dynamic characteristic is known as
the residual voltage (IR drop) and is the peak value of the volt-
V
age during the discharge of the surge current. This voltage var-
ies from 3 kV to 6 kV depending upon the type of arrester i.e.,
whether station or line type, the magnitude and wave shape of
the discharge current. The spark gaps are so designed that they
give an impulse ratio of unity to the surge diverter and as a
result they are unable to interrupt high values of current and
the follow up currents are limited to 20 to 30 A. The impulse
breakdown strength of a diverter is smaller than the residual I
voltage, and therefore, from the point of view of insulation co- Fig. 7.41 Volt-ampere characteristic
ordination residual voltage decides the protection level. of valve-type LA
A
l
l
J
N
T
U
W
o
r
l
d

The operation of the arrester can be easily understood with the help of Fig. 7.42 (a) and (b).
When a surge voltage is incident at the terminal of the arrester it causes the two gap units to flashover,
thereby a path is provided to the surge to the ground through the coil element and the non-linear
resistor element. Because of the high frequency of the surge, the coil develops sufficient voltage
across its terminals to cause the by-pass gap to flashover. With this the coil is removed from the
circuit and the voltage across the LA is the IR drop due to the non-linear element. This condition
continues till power frequency currents follow the preionized path. For power frequency the
impedance of the coil is very low and, therefore, the arc will become unstable and the current will be
transferred to the coil (Fig. 7.42 (b)). The magnetic field developed by the follow current in the coil
reacts with this current in the arcs of the gap assemblies, causing them to be driven into arc quenching
chambers which are an integral part of the gap unit. The arc is extinguished at the first current zero by
cooling and lengthening the arc and also because the current and voltage are almost in phase. Thus the
diverter comes back to normal state after discharging the surge to the ground successfully.
Line Line
Gap
unit
Magnetic
coil
Gap
unit
Thyrite
valve
elements
Impulse Gap unit Power
curent
follow
current
Pre-ionizing Pre-ionizing
tip tip
By-pass
Magnetic
By-pass
Thyrite coil Thyrite
gap shunting gap shunting
resistors resistors
Pre-ionizing Gap Pre-ionizing
tip unit tip
Thyrite
valve
elements
(a) (b)
Fig. Schematic diagram of valve-type arrester indicating path of (a)
Surge current, (b) Follow current.
Location of lightning arresters
The normal practice is to locate the lightning arrester as close as possible to the equipment to be
protected. The following are the reasons for the practice: (i) This reduces the chances of surges enter-
ing the circuit between the protective equipment and the equipment to be protected. (ii) If there is a
distance between the two, a steep fronted wave after being incident on the lightning arrester, which
sparks over corresponding to its spark over voltage, enters the transformer after travelling over the
lead between the two. The wave suffers reflection at the terminal and, therefore, the total voltage at the
terminal of the transformer is the sum of reflected and the incident voltage which approaches nearly
twice the incident voltage i.e., the transformer may experience a surge twice as high as that of the
A
l
l
J
N
T
U
W
o
r
l
d

lightning arrester. If the lightning arrester is right at the terminals this could not occur. (iii) If L is the
inductance of the lead between the two, and IR the residual voltage of the lightning arrester, the
voltage incident at the transformer terminal will be
V = IR + L
di
dt
where di/dt is the rate of change of the surge current.
It is possible to provide some separation between the two, where a capacitor is connected at the
terminals of the equipment to be protected. This reduces the steepness of the wave and hence the rate
di/dt and this also reduces the stress distribution over the winding of the equipment.
There are three classes of lightning arresters available:
(i) Station type: The voltage ratings of such arresters vary from 3 kV to 312 kV and are de-
signed to discharge currents not less than 100,000 amps. They are used for the protection
of substation and power transformers.
(ii) Line type: The voltage ratings vary from 20 kV to 73 kV and can discharge currents be-
tween 65,000 amps and 100,000 amps. They are used for the protection of distribution
transformers, small power transformers and sometimes small substations.
(iii) Distribution type: The voltage ratings vary from 8 kV to 15 kV and can discharge currents
upto 65,000 amperes. They are used mainly for pole mounted substation for the protection
of distribution transformers upto and including the 15 kV classification.
Rating of lightning arrester
A lightning arrester is expected to discharge surge currents of very large magnitude, thousands of amperes,
but since the time is very short in terms of microseconds, the energy that is dissipated through the lightning
arrester is small compared with what it would have been if a few amperes of power frequency current had
been flown for a few cycles. Therefore, the main considerations in selecting the rating of a lightning arrester
is the line-to-ground dynamic voltage to which the arrester may be sub-jected for any condition of system
operation. An allowance of 5% is normally assumed, to take into account the light operating condition
under no load at the far end of the line due to Ferranti effect and the sudden loss of load on water wheel
generators. This means an arrester of 105% is used on a system where the line to ground voltage may reach
line-to-line value during line-to-ground fault condition.
The overvoltages on a system as is discussed earlier depend upon the neutral grounding condi-
tion which is determined by the parameters of the system. We recall that a system is said to be solidly
grounded only if
R0 ≤ 1
X1
and
X0
≤ 3
X1
and under this condition the line to ground voltage during a L-G fault does not exceed 80% of the L-L
voltage and, therefore, an arrester of (80% + 0.05 × 80%) = 1.05 × 80% = 84% is required. This is the
extreme situation in case of solidly grounded system. In the same system the voltage may be less than
80%; say it may be 75%. In that case the rating of the lightning arrester will be 1.05 × 75% = 78.75%.
A
l
l
J
N
T
U
W
o
r
l
d

The overvoltages can actually be obtained with the help of precalculated curves. One set of curves
corresponding to a particular system is given in Fig.
1
/X
0
R
7
95
100
6
5 90
4
3 85
2
80
1
75
70
65
(0, 0)
1 2 3 4 5 6 X 0 /X1
Voltage condition for R 1 = R 2 = 0.2 X1
Fig. Maximum line-to-ground voltage at fault location for
grounded neutral system under any fault condition
For system grounded through Peterson coil, the overvoltages may be 100% if it is properly
tuned and, therefore, it is customary to apply an arrester of 105% for such systems. Even though there
is a risk of overvoltage becoming more than 100% if it is not properly tuned, but it is generally not
feasible to select arresters of sufficiently high rating to eliminate all risks of arrester damage due to
these reasons. The voltage rating of the arrester, therefore, ranges between 75% to 105% depending
upon the neutral grounding condition.
So far we have discussed the non-shielding method. We now discuss the shielding method i.e.,
the use of ground wires for the protection of transmission lines against direct lightning strokes.
GROUND WIRES
The ground wire is a conductor running parallel to the power conductors of the transmission line and is
placed at the top of the tower structure supporting the power conductors (Fig. 7.44 (a)). For horizontal
configuration of the power line conductors, there are two ground wires to provide effective shielding to
power conductors from direct lightning stroke whereas in vertical configuration there is one ground wire.
The ground wire is made of galvanized steel wire or in the modern high voltage transmission lines ACSR
conductor of the same size as the power conductor is used. The material and size of the conduc-tor are
more from mechanical consideration rather than electrical. A reduction in the effective ground resistance
can be achieved by other relatively simpler and cheaper means. The ground wire serves the following
purposes: (i) It shields the power conductors from direct lightning strokes. (ii) Whenever a
A
l
l
J
N
T
U
W
o
r
l
d

lightning stroke falls on the tower, the ground wires on both sides of the tower provide parallel paths
for the stroke, thereby the effective impedance (surge impedance) is reduced and the tower top poten-
tial is relatively less. (iii) There is electric and magnetic coupling between the ground wire and the
power conductors, thereby the changes of insulation failure are reduced.
Ground
wire
Ground
wire

Power
conductor
(a) (b)
(a) Protective angle; (b) Protection afforded by two ground wires
Protective angle of the ground wire is defined as the angle between the vertical line passing
through the ground wire and the line passing through the outermost power conductor (Fig. 7.44 (a))
and the protective zone is the zone which is a cone with apex at the location of the ground wire and
surface generated by line passing through the outermost conductor. According to Lacey, a ground wire
provides adequate shielding to any power conductor that lies below a quarter circle drawn with its
centre at the height of ground wire and with its radius equal to the height of the ground wire above the
ground. If two or more ground wires are used, the protective zone between the two adjacent wires can
be taken as a semi-circle having as its diameter a line connecting the two ground wires (Fig. 7.44 (b)).
The field experience alongwith laboratory investigation has proved that the protective angle should be
almost 30° on plain areas whereas the angle decreases on hilly areas by an amount equal to the slope
of the hill.
The voltage to which a transmission tower is raised when a lightning strikes the tower, is
independent of the operating voltage of the system and hence the design of transmission line against
lightning for a desired performance is independent of the operating voltage. The basic requirement for the
design of a line based on direct stroke are: (i) The ground wires used for shielding the line should be
mechanically strong and should be so located that they provide sufficient shield. (ii) There should be
A
l
l
J
N
T
U
W
o
r
l
d

sufficient clearance between the power conductors themselves and between the power conductors and
the ground or the tower structure for a particular operating voltage. (iii) The tower footing resistance
should be as low as can be justified economically.
To meet the first point the ground wire as is said earlier is made of galvanized steel wire or ACSR
wire and the protective angle decides the location of the ground wire for effective shielding. The second
factor, i.e., adequate clearance between conductor and tower structure is obtained by designing a suitable
length of cross arm such that when a string is given a swing of 30° towards the tower struc-
ture the air gap between the power conductor and tower
structure should be good enough to withstand the
switching voltage expected on the system, normally
four times the line-to-ground voltage (Fig. 7.45).
The clearances between the conductors also
should be adjusted by adjusting the sag so that the
mid span flashovers are avoided.
The third requirement is to have a low tower
footing resistance economically feasible. The stand-ard
value of this resistance acceptable is approximately 10
ohms for 66 kV lines and increases with the oper-ating
voltage. For 400 kV it is approx. 80 ohms. The tower
footing resistance is the value of the footing resistance
when measured at 50 Hz. The line perform-ance with
regard to lightning depends upon the im-pulse value of
the resistance which is a function of the soil resistivity,
critical breakdown gradient of the soil, length and type
of driven grounds or counterpoises and the magnitude
of the surge current. If the con-
struction of the tower does not give a suitable value of
the footing resistance, following methods are adopted.
One possibility could be the chemical treatment
of the soil. This method is not practically possible because of the
long length of the lines and because this method needs regular
check up about the soil conditions. It is not possible to check up the
soil conditions at each and every tower of the line which runs in several miles. Therefore, this method
is used more for improving the grounds of the substation.
The methods normally used for improving the grounds of transmission towers are the use of (i)
ground rods, and (ii) counterpoises.
Ground Rods
Ground rods are used to reduce the tower footing resistance. These are put into the ground
surrounding the tower structure. Fig. 7.46 shows the variation of ground resistance with the length and
thickness of the ground rods used. It is seen that the size (thickness) of the rod does not play a major
role in reducing the ground resistance as does the length of the rod. Therefore, it is better to use thin
but long rods or many small rods.
30°
Clearance
required
Clearance determination or cross
arm length determination
A
l
l
J
N
T
U
W
o
r
l
d

250
200
150
100
50
1.25 cm
1.9 cm
2.54 cm
0 1.8 3.0 4.2 5.4 6.6 7.8 9.0
Driven depth in m
Ground rod resistance as a function of rod length
Counterpoise
A counterpoise is galvanized steel wire run in parallel or radial or a combination of the two, with
respect to the overhead line. The various configurations used are shown in Fig. 7.47.
The corners of the squares indicate the location of the tower legs. The lightning stroke as is
incident on the tower, discharges to the ground through the tower and then through the counterpoises.
It is the surge impedance of the counterpoises which is important initially and once the surge has
travelled over the counterpoise it is the leakage resistance of the counterpoise that is effective. While
selecting a suitable counterpoise it is necessary to see that the leakage resistance of the counterpoise
should always be smaller than the surge impedance; otherwise, positive reflections of the surge will
take place and hence instead of lowering the potential of the tower (by the use of counterpoise) is will
be raised.
The leakage resistance of the counterpoise depends upon the surface area, i.e., whether we have
one long continuous counterpoise say 1000 m or four smaller counterpoises of 250 m each, as far as
the leakage resistance is concerned it is same, whereas the surge impedance of say 1000 m if it is 200
ohms, then it will be 200/4, if there are four counterpoises of 250 m. each, as these four wires will
now be connected in parallel. Also if the surge takes say 6 micro-seconds to travel a distance of 1000
m to reduce the surge impedance to leakage impedance, with four of 250 m it will take 1.5.  sec, that
is, the surge will be discharged to ground faster, the shorter the length of the ground wire. It is,
therefore, desirable to have many short counterpoises instead of one long counterpoise. But we should
not have too many short counterpoises, otherwise the surge impedance will become smaller than the
leakage resistance (which is fixed for a counterpoise) and positive reflections will occur.
A
l
l
J
N
T
U
W
o
r
l
d

Single parallel continuous
Double parallel continuous
Radial
Radial and continuous
Fig. Arrangement of counterpoise
The question arises as to why we should have a low value of tower footing resistance. It is
clear that, whenever a lightning strikes a power line, a current is injected into the power system. The
voltage to which the system will be raised depends upon what impedances the current encounters. Say
if the lightning stroke strikes a tower, the potential of the tower will depend upon the impedance of the
tower. If it is high, the potential of the tower will also be high which will result in flashover of the
insulator discs and result in a line-to-ground fault. The flashover will take place from the tower
structure to the power conductor and, therefore, it is known as back flashover,
Surge Absorbers
A surge absorber is a device which absorbs energy contained in a travelling wave. Corona is a means
of absorbing energy in the form of corona loss. A short length
of cable between the equipment and the overhead line ab-
sorbs energy in the travelling wave because of its high ca-
pacitance and low inductance. Another method of absorb-
ing energy is the use of Ferranti surge absorber which con-
sists of an air core inductor connected in series with the
line and surrounded by an earthed metallic sheet called a Ferranti surge absorber
dissipator. The dissipator is insulated from the inductor by
the air as shown in Fig. 7.48.
The surge absorber acts like an air cored transformer whose primary is the low inductance
inductor and the dissipator acts as the single turn short circuit secondary. Whenever the travelling
wave is incident on the surge absorber a part of the energy contained in the wave is dissipated as heat
due to transformer action and by eddy currents. Because of the series inductance, the steepness of the
wave also is reduced. It is claimed that the stress in the end turns is reduced by 15% with the help of
surge absorber.
A
l
l
J
N
T
U
W
o
r
l
d

UNIT-VII
NON DESTRUCTIVE TESTING OF MATERIAL AND ELECTRICAL
APPARATUS
Introduction
All electrical appliances are insulated with gaseous or liquid or solid or a suitable combination of
these materials. The insulation is provided between live parts or between live part and grounded part
of the appliance. The materials may be subjected to varying degrees of voltages, temperatures and
frequen-cies and it is expected of these materials to work satisfactorily over these ranges which may
occur occasionally in the system. The dielectric losses must be low and the insulation resistance high
in order to prevent thermal breakdown of these materials. The void formation within the insulating
materials must be avoided as these deteriorate the dielectric materials.
When an insulating material is subjected to a voltage for investigation, it is usually not possible
to draw conclusion regarding the cause of breakdown from the knowledge of the breakdown voltage
particularly in solid materials. Earlier, the quality of insulation was judged, mainly by the insulation
resistance and its dielectric strength. However, these days high voltage equipments and installations
are subjected to various tests. These tests should also yield information regarding the life expectancy
and the long term stability of the insulating materials.
One of the possible testing procedure is to over-stress insulation with high a.c. and/ or d.c. or
surge voltages. However, the disadvantage of the technique is that during the process of testing the
equipment may be damaged if the insulation is faulty. For this reason, following non-destructive
testing methods that permit early detection for insulation faults are used:
(i) Measurement of the insulation resistance under d.c. voltages.
(ii) Determination of loss factor tan δ and the capacitance C.
(iii) Measurement of partial discharges.
LOSS IN A DIELECTRIC
An ideal dielectric is loss-free and if its relative permittivity is εr, its permittivity is given by
ε = ε0εr
and ε also known as the dielectric constant is a real number. A real dielectric is always associated with
loss. The following are the mechanisms which lead to the loss:
(i) Conduction loss Pc by ionic or electronic conduction. The dielectric, has σ as conductivity.
(ii) Polarization loss Pp by orientation boundary layer or deformation polarization.
167
A
l
l
J
N
T
U
W
o
r
l
d

(iii) Ionisation loss Pi by partial discharges
internal or external zones.
Fig. 6.1 shows an equivalent circuit of
R2 C2
a dielectric with loss due to conduction,
polarization and partial discharges. An R0 () R1 C1
ideal dielectric can be represented by a
C3
pure capacitor C1, conduction losses S
can be taken into account by a resistor
R0 (σ) in parallel. Polarization losses
produce a real component of the dis-
Fig. Equivalent circuit of a dielectric
placement current which is simulated
by resistor R1. Pulse partial discharges
are simulated by right hand branch. C3 is the capacitance of the void and S is the spark gap
which fires during PD discharge and the repeated recharging of C3 is effected either by a
resistor R2 or a capacitor C2.
MEASUREMENT OF RESISTIVITY
When a dielectric is subjected to a steady state static electric field E the current density Jc is given by
Jc = σE
Assuming a cuboid of the insulating material with thickness d and area A, then
Current I = Jc A and power loss = VI
= VJc A = V σ EA = Vσ V A. d =V σ V A. d  σ E
2
. Volume.
d
d dd
Therefore, specific dielectric loss = σ E
2
Watts/m
3
.
The conductivity of the insulating materials viz liquid and solid depends upon the temperature
and the moisture contents. The leakage resistance R0 (σ) of an insulating material is determined by
measuring the current when a constant d.c. voltage is applied. Since the current is a function of time as
different mechanisms are operating simultaneously. So to measure only the conduction current it is
better to measure the current about 1 min after the voltage is switched on. For simple geometries of
the specimen (cuboid or cube) specific resistivity (p = 1/σ) can be calculated from the leakage
resistance measured. If I is the conduction current measure and V the voltage applied, the leakage
resistance is given by
R =
V
 ρ
d
I A
where d is the thickness of the specimen and A is the area of section.
Fig. 6.2. shows a simple arrangement for measurement of resistivity of the insulating material. The
d.c. voltage of 100 volt or 1000 volt is applied between electrode 1 and the earth. The measuring electrode
2 is earthed through a sensitive ammeter. The third electrode known as guard ring electrode surrounds the
measuring electrode and is directly connected to ground so as to eliminate boundary field effects and
surface currents. The width of the guard electrode should be at least twice the thickness of the specimen
and the unguarded electrode (1) must extend to the outer edge of the guard electrode. The
A
l
l
J
N
T
U
W
o
r
l
d

gap between electrode 2 and 3 should be as small as possible. A thin metallic foil usually of aluminium or
lead of about 20 m thickness is placed between the electrodes and specimen for better contact. The
specific conductivity for most of the insulating material lies in the range of 10
–16
to 10
–10
S/cm, which
gives currents to be measured of these speciemen to be of the order of picoampers or nanoamperes.
High-voltage
electrode
1
2
Measuring
Guard-ring A electrode electrode
and screening
Fig. Electrode arrangement to measure the specific
resistivity of an insulation specimen
The measuring leads should be appropriately and carefully screened. The measurement of con-
duction current using d.c. voltage not only provides information regarding specific resistivity of the
material but it gives an idea of health of the insulating material. If conduction currents are large, the
insulating properties of the material are lost. This method, therefore, has proved very good in the
insulation control of large electrical machines during their period of operation.
MEASUREMENT OF DIELECTRIC CONSTANT AND LOSS FACTOR
Dielectric loss and equivalent circuit
In case of time varying electric fields, the current density Jc using Amperes law is given by
Jc = σ E + ∂ D
= σ E + ε ∂ E
∂ t ∂ t
For harmonically varying fields
E = Em ejωt
∂ E
∂ t = jEmωe
jωt
= j ω E Ir I
Therefore, Jc = σ E + j ω ε E

= (σ + j ω ε)E
In general, in addition to conduction losses, ionization 
and polarization losses also occur and, therefore, the dielec- I

tric constant ε = ε0 εr is no longer a real quantity rather it is a Fig Phasor diagram for a real
dielectric material
A
l
l
J
N
T
U
W
o
r
l
d

complex quantity. By definition, the dissipation factor tan δ is the ratio of real component of current
Iω to the reactive component Ir (Fig. 6.3).
tan δ = Iω =
p
did
P
I
r r
Here δ is the angle between the reactive component of current and the total current flowing
through the dielectric at fundamental frequency. When δ is very small tan δ = δ when δ is expressed
in radians and tan δ = sin δ = sin (90 – φ) = cos φ i.e., tan δ then equals the power factor of the
dielectric material.
As mentioned earlier, the dielectric loss consists of three components corresponding to the
three loss mechanism.
Pdiel = Pc + Pp + Pi
and for each of these an individual dissipation factor can be given such that
tan δ = tan δc + tan δp + tan
δi If only conduction losses occur then
V
2
ω ε 0
εr
A
P = P = σE
2
Ad = V
2
ω C tan δ = tan δ
diel c d
or σ E
2
= V2
ωε ε tan δ = E
2
ωε ε tan δ
d 2 0 r 0 r
or tan δ 
σ
ωε ε
0 r
This shows that the dissipation factor due to conduction loss alone is inversely proportional to
the frequency and can, therefore, be neglected at higher frequencies. However, for supply frequency
each loss component will have considerable magnitude.
In order to include all losses, it is customary to refer the existence of a loss current in addition
to the charging current by introducing complex permittivity.
ε
*
= ε′ – j ε″
and the total current I is expressed as
C0
I = (j ωε′ + ωε″) ε 0
V
where C0 is the capacitance without dielectric material.
or I = j ω C ε* . V
0 r
ε* =
(ε ′ − j ε ″ )
where = ε′ – j ε ″
ε r
r 0
r
ε* is called the complex relative permittivity or complex dielectric constant, ε′ and ε ′ are called
r r
the permittivity and relative permittivity and ε″ and εr″ are called the loss factor and relative loss
factor respectively.
The loss tangent
tan δ =
ε ″

ε r ″
ε′ ε r ′
A
l
l
J
N
T
U
W
o
r
l
d

The product of the angular frequency and ε″ is equivalent to the dielectric conductivity σ″ i.e., σ″ 
ωε″. The dielectric conductivity takes into account all the three power dissipative processes includ-ing
the one which is frequency dependent. Fig. 6.4 shows two equivalent circuits representing the
electrical behaviour of insulating materials under a.c. voltages, losses have been simulated by resistances.
I RS
I
I
RS

VC
R
P
I
CP
CS 
 CS
V/R V
(b)
(a)
Fig. Equivalent circuits for an insulating material
Normally the angle between V and the total current in a pure capacitor is 90°. Due to losses,
this angle is less than 90°. Therefore, δ is the angle by which the voltage and charging current fall
short of the 90° displacement.
For the parallel circuit the dissipation factor is given by
1
tan δ =
ωC p Rp
and for the series circuit
tan δ = ω CsRs
For a fixed frequency, both the equivalents hold good and one can be obtained from the other.
However, the frequency dependence is just the opposite in the two cases and this shows the limited
validity of these equivalent circuits.
The information obtained from the measurement of tan δ and complex permittivity is an
indica-tion of the quality of the insulating material.
(i) If tan δ varies and changes abruptly with the application of high voltage, it shows inception
of internal partial discharge.
(ii) The effect to frequency on the dielectric properties can be studied and the band of frequen-
cies where dispersion occurs i.e., where that permittivity reduces with rise in frequency can
be obtained.
HIGH VOLTAGE SCHERING BRIDGE
The bridge is widely used for capacity and dielectric loss measurement of all kinds of capacitances,
for instance cables, insulators and liquid insulating materials. We know that most of the high voltage
equipments have low capacitance and low loss factor. Typical values of these equipments are given in
Chapter 3. This bridge is then more suitable for measurement of such small capacitance equipments as
the bridge uses either high voltage or high frequency supply. If measurements for such low capacity
equipments is carried out at low voltage, the results so obtained are not accurate.
A
l
l
J
N
T
U
W
o
r
l
d

Fig. shows a high voltage schering bridge where the specimen has been represented by a
parallel combination of Rp and Cp.
Fig. Basic high voltage schering bridge
The special features of the bridge are:
1. High voltage supply, consists of a high voltage transformer with regulation, protective cir-
cuitry and special screening. The input voltage is 220 volt and output continuously variable
between 0 and 10 kV. The maximum current is 100 mA and it is of 1 kVA capacity.
2. Screened standard capacitor Cs of 100 pF  5%, 10 kV max and dissipation factor tan δ = 10
–5
.
It is a gas-filled capacitor having negligible loss factor over a wide range of frequency.
3. The impedances of arms I and II are very large and, therefore, current drawn by these arms
is small from the source and a sensitive detector is required for obtaining balance. Also,
since the impedance of arm I and II are very large as compared to III and IV, the detector
and the impedances in arm III and IV are at a potential of only a few volts (10 to 20 volts)
above earth even when the supply voltage is 10 kV, except of course, in case of breakdown
of one of the capacitors of arm I or II in which case the potential will be that of supply
voltage. Spark gaps are, therefore, provided to spark over whenever the voltage across arm
III or IV exceeds 100 volt so as to provide personnel safety and safety for the null detector.
4. Null Detector: An oscilloscope is used as a null detector. The γ–plates are supplied with
the bridge voltage Vab and the x-plates with the supply voltage V. If Vab has phase
difference with respect to V, an ellipse will appear on the screen (Fig. 6.6). However, if
magnitude balance is not reached, an inclined straight line will be observed on the screen.
The information about the phase is obtained from the area of the eclipse and the one about
the magnitude from the inclination angle. Fig. 6.6a shows that both magnitude and phase
are balanced and this represents the null point condition. Fig. (6.6c) and (d) shows that
only phase and amplitude respectively are balanced.
(a) (b) (c) (d)
Fig. Indications on null detector
A
l
l
J
N
T
U
W
o
r
l
d

The handling of bridge keys allows to meet directly both the phase and the magnitude
conditions in a single attempt. A time consuming iterative procedure being used earlier is
thus avoided and also with this a very high order of accuracy in the measurement is
achieved. The high accuracy is obtained as these null oscilloscopes are equipped with a γ –
amplifier of automatically controlled gain. If the impedances are far away from the balance
point, the whole screen is used. For nearly obtained balance, it is still almost fully used. As
Vab becomes smaller, by approaching the balance point, the gain increases automatically
only for deviations very close to balance, the ellipse area shrinks to a horizontal line.
5. Automatic Guard Potential Regulator: While measuring capacitance and loss factors using
a.c. bridges, the detrimental stray capacitances between bridge junctions and the ground
adversely affect the measurements and are the source of error. Therefore, arrangements
should be made to shield the measuring system so that these stray capacitances are either
neutralised, balanced or eliminated by precise and rigorous calculations. Fig. 6.7 shows
various stray capacitance associated with High Voltage Schering Bridge.
Fig. Schering bridge with stray capacitances
Ca, Cb, Cc and Cd are the stray capacitances at the junctions A, B, C and D of the bridge. If
point D is earthed during measurement capacitance Cd is thus eliminated. Since Cc comes across the
power supply for earthed bridge, has no influence on the measurement. The effect of other stray
capacitances Ca and Cb can be eliminated by use of auxiliary arms, either guard potential regulator or
auxiliary branch as suggested by Wagner.
Fig. 6.8 shows the basic principle of Wagner earth to eliminate the effect of stray capacitances
Ca and Cb. In this arrangement an additional arm Z is connected between the low voltage terminal of
the four arm bridge and earth. The stray capacitance C between the high voltage terminal of the bridge
and the grounded shield and the impedance Z together constitute a six arm bridge and a double
balancing procedure is required.
Switch S is first connected to the bridge point b and balance is obtained. At this point a and b
are at the same potential but not necessarily at the ground potential. Switch S is now connected to
point C and by adjusting impedance Z balance is again obtained. Under this condition point ‗a‘ must
be at the same potential as earth although it is not permanently at earth potential. Switch S is again
connected to point b and balance is obtained by adjusting bridge parameters. The procedure is
repeated till all the three points a, b and c are at the earth potential and thus Ca and Cb are eliminated.
A
l
l
J
N
T
U
W
o
r
l
d

Cs C
a
b c
D
S
Z
Fig. Bridge incorporating Wagner earth
This method is, however, now rarely
used. Instead an auxiliary arm using
automatic guard potential regulator is used.
The basic circuit is shown in Fig. 6.9.
The guard potential regulator keeps the
shield potential at the same value as that of
the detector diagonal terminals a and b for the
bridge balance considered. Since potentials of
a, b and shield are held at the same value the
stray capacitances are eliminated. During the
process of balancing the bridge the points a Fig. Automatic Wagner earth or automatic
and b attain different values of potential in guard potential regulator
magnitude and phase with respect to ground. As a result, the guard potential regulator should be able
to adjust the voltage both in magnitude and phase. This is achieved with a voltage divider arrangement
provided with coarse and fine controls, one of them fed with in-phase and the other quadrature compo-
nent of voltage. The control voltage is then the resultant of both components which can be adjusted
either in positive or in negative polarity as desired. The comparison between the shielding potential
adjusted by means of the Guard potential regulator and the bridge voltage is made in the null indicator
oscilloscope as mentioned earlier. Modifying the potential, it is easy to bring the reading of the null
detector to a horizontal straight line which shows a balance between the two voltages both in magni-
tude and phase.
The automatic guard potential regulator adjusts automatically the guard potential of the bridge
making this equal in magnitude and phase to the potential of the point a or b with respect to ground.
The regulator does not use any external source of voltage to achieve this objective. It is rather con-
nected to the bridge corner point between a or b and c and is taken as a reference voltage and this is
then transmitted to the guard circuit with unity gain both in magnitude and phase. The shields of the
leads to Cs and Cp are not grounded but connected to the output of the regulator which, in fact, is an
operational amplifier. The input impedance of the amplifier is more than 1000 Megaohms and the
output impedance is less than 0.5 ohm. The high input impedance and very low output impedance of
the amplifier does not load the detector and keeps the shield potential at any instant at an artificial
ground.
A
l
l
J
N
T
U
W
o
r
l
d

Balancing the Bridge
For ready reference Fig. 6.5 is reproduced here and its phasor diagram under balanced
condition is drawn in Fig. 6.10 (b)
V1 Cs
I1
V2/R2

I 1
R
1
=
V2
V2
C2
90° V1/ Rp V1
Fig. (a) Schering bridge (b) Phasor diagram
The bridge is balanced by successive variation of R1 and C2 until on the oscilloscope
(Detector) a horizontal straight line is observed:
At balance
ZI

Z
III
Z
II
Z
IV
Now
Z
I 
Rp
1  j ω C p Rp
Z
II =
1
j ω Cs
Z
III = RI and ZIV =
R2
 j ω C2 R2
1
From balance equation we have
Rp

1/ j ω C s (1  j ω C2 R2 )
R1 (1  i ω C p Rp )
R
2
R p (1 − j ω C p Rp )

1  j ω C R
or
2 2
R1 d1  ω
2
C p
2
Rp
2
i j ω Cs R2
or Rp
−
j ω C p Rp
2
−
j

C2
R1 d1  ω
2
C p
2
Rp
2
i R1 d1  ω
2
C p
2
Rp
2
i ωCs R2 Cs
Equating real part, we have
R
p C

2
R1 d1  ω
2
C p
2
Rp
2
i Cs
and equating imaginary part, we have
ω C p Rp
2

1
R1 d1  ω
2
C p
2
Rp
2
i ωCs R2
A
l
l
J
N
T
U
W
o
r
l
d

Now tan δ from the phasor diagram
V1 /Rp

1

V ω C
 ωC 2 R2
tan δ =
2 2
V1ω C p ωC p Rp V2 /R2
Also cos δ = V1ωCp
V1 d1 / R p
2
i ω
2
Cp
2
or cos
2
δ = ω
2
C p
2
Rp
2
1  ω
2
C p
2
Rp
2
or
ωC p Rp
 cos
2
δ .
1

R1
1  ω
2
C
2
R
2
ω C R ω C R R
p p p
p p s 2
or C = cos
2
δ C R2
R
p s
1
Since δ is usually very small cos δ = 1
Therefore Cp ≈ Cs
R2
R
1
and tan δp = ω C2R2
and since
1
= ω C2R2
ω Cp Rp
or ω
2
C R C R = 1
p p 2 2
1
≈
R1
≈
R1
or Rp = ω
2
R C C
p
ω
2
R C C R ω
2
C C R
2
2 2 2 2 s 2 2 s 2
If, however, the specimen is replaced by a series equivalent circuit, then at balance
j
Z
I
=R
s
–
ωCs
and the equation becomes
Rs − j / ω Cs  1  j ω C 2 R2
R1 j ω C s′ R2
or Rs − j − j  C2
ω C R ωC ′ R C ′
R
1 s 1 s 2 s
Equating real parts, we have
R
s  C2
R1 Cs ′
or Rs = R1
C
2
Cs ′
Fig. Phasor diagram of S.B. for series
equivalent of specimen
A
l
l
J
N
T
U
W
o
r
l
d

Similarly equating imaginary part, we have
ωCsR1 = ωCs′ R2
or Cs = Cs′
R2
R1
To find out tan δ, we draw the phasor diagram of the bridge circuit (Fig. 6.11).
tan δs =
I
1
R
s
= ωCsRs
I1 /ωCs
MEASUREMENT OF LARGE CAPACITANCE
In order to measure a large capacitance, the resistance R1 should be
able to carry large value of current and resistance R1 should be of low
value. To achieve this, a shunt of S ohm is connected across R1 as
shown in Fig. 6.12. It is desirable to connect a fixed resistance R in
series with variable resistance R1 so as to protect R1 from excessive
current, should it accidentally be set to a very low value.
We know that under balanced condition for series equivalent
representation of specimen
Cs  Cs′
R2
R
1
But here R1 is to be replaced by the equivalent of (R + R1) || S.
(R  R1 ) S
or
 S
R  R1
Fig. Shunt arrangement for
measurement of large
capacitance
or
1

R  R1  S
R
eq (R  R1 ) S
Therefore, C = C ′R . R  R1  S .R1 = Cs ′ R2 . R1 (R  R1  S)
2
s s
(R  R1 )S R1
R
1 (R  R1 ) S
usually R < < R1
L R
O
R
2 R1 ( R  R1  S) R2 R1
Therefore C
s  Cs′  Cs ′ M   1P
R1 R1 S R1 N SS Q
and tan δ = ω Cs′ R2. R/R1
If circuit elements of Schering bridge are suitably designed, the bridge principle can be used upto to
100 kHz of frequency. However, common schering bridge can be used upto about 10 kHz only.
SCHERING BRIDGE METHOD FOR GROUNDED TEST SPECIMEN
A dielectric material which is to be tested, one side of this is usually grounded e.g. underground cables
or bushings with flanges grounded to the tank of a transformer etc. There are two well known methods
A
l
l
J
N
T
U
W
o
r
l
d

used for such measurement. One is the inversion
of a Schering bridge shown in Fig. 6.13 with the
operator, ratio arms and null detector inside a
Faraday Cage at high potential. The system
requires the cage to be insulated for the full test
C
s
R
s
Cs
voltage and with suitable design may be used
upto maximum voltage available. However, there
are difficulties in inverting physically the stan-
dard capacitor and it becomes necessary to mount
it on a platform insulated for full voltage.
Second method requires grounding of
detector as shown in Fig. 6.14. In this
arrangement, stray capacitances of the high
voltage terminal Cq and of the source, leads
etc. come in parallel with the test object.
Hence, balancing is carried out in two steps:
First step: The test specimen is disconnected
and the capacitance Cq and loss factor tan δq
are measured. Second step: The test object is
connected in the bridge and new balance is
obtained. The second balance gives net
capacitance of the parallel combination i.e.,
Cs′ = Cs + Cq
D
R
2
R
1
C
2
Fig. Inverted Schering bridge
Fig. Grounded specimen
and tan δq′ =
Cs tan δ s  Cq tan δ q
Cs  Cq
Hence the capacitance and loss factor of the specimen are
Cs = Cs″ – Cq
and tan δs =
Cs ″ tan δ ″ − Cq tan δ q
Cs
If the stray capacitances are large as compared to the capacitance of the specimen, the accuracy
of measurement is poor.
SCHERING BRIDGE FOR MEASUREMENT OF HIGH LOSS FACTOR
If the loss factor tan δ of a specimen is large, the bridge arm containing resistance R2 is modified.
Resistance R2 is made as a slide wire alongwith a decade resistor and a fixed capacitance C2 is con-
nected across the resistance R2 as shown in Fig. 6.15 (a). This modification can be used for test speci-
men having loss factor of the order of 1.0. If it is more than one and upto 10 or greater C2, R2 arm is
not made a parallel combination, rather it is made a series combination as shown in Fig. 6.15 (b) Here,
of course R2 is made variable.
A
l
l
J
N
T
U
W
o
r
l
d

Cs
Cs
Cs
Cs
R
s Rs
D D
R2
R
1
C2
R
1 R
2
(a) (b)
Fig. Schering bridges for large loss factor
TRANSFORMER RATIO ARM BRIDGE
For measurement of various parameters like resistance, in-
ductance, capacitance, usually four arm bridges are used.
For high frequency measurements, the arm with high
resistances leads to difficulties due to their residual induct-
ance, capacitance and skin effect. Also if length of the leads
is large, shielding is difficult. Hence at high frequencies the
transformer ratio arm bridge which eliminates at least two
arms, are preferred. These bridges provide more accurate
results for small capacitance measurements. There are two
types of transformer ratio arm bridges (i) Voltage ratio; (ii)
Current ratio. The voltage ratio type is used for high fre-
quency low voltage application. Fig. 6.16 shows schematic
diagram of a voltage ratio arm bridge. Assuming ideal
trans-former, under balance condition:
Cs
NbRa
Na
D
Ns Rs Cs
Fig.Transformer voltage ratio arm
bridge
Vb  Nb  Cs and Rs  Ns
Ra Na
VsNsCs ′
However, in practical situation due to the presence of magnetising current and the load currents, the
voltage ratio slightly differs from the turns ratio and therefore, the method involves certain errors. The
errors are classified as ratio error and load error which can be calculated before hand for a transformer. A
typical bridge has a useful range from a fraction of a pF to about 100 F and is accurate over a wide range
of frequency from 100 Hz to 100 kHz, the accuracy being better than  0.5%.
The current ratio arm bridge is used for high voltage low frequency applications. The main
advantage of the method is that the test specimen is subjected to full system voltage. Fig. 6.17 shows
schematic diagram of the bridge. The main component of the bridge is a three winding current trans-
former with very low losses and leakage (core of high permeability). The transformer is carefully
shielded against stray magnetic fields and protected against mechanical vibrations.
A
l
l
J
N
T
U
W
o
r
l
d

The main feature of the arrangement is that HT
under balance condition, there is no net mmf across
s
the windings 1 and 2. Also the stray capacitances I s
Cs
between the windings and the screened low volt-
Rc I
age leads does not enter in the balance expression 2
C
as there is no voltage developed between them. This R
feature makes this bridge possible to use long leads Cs
without using Wagner‘s earth. The sensitivity of the 1 2 N2
bridge is higher than that of the Schering bridge. N
1 Detector
The balance is obtained by varying N1, N2 and R. 3
Under balance condition, the voltage across
the detector coil is zero and Fig.Transformer current ratio arm bridge
IsN1 = I2N2
Voltage across the R-C arm is
V = Is ′ R
1  jωCR
Current I2 through coil 2 is
Is ′
1
Is ′
I2  j ωC

1
R 
1  Jω CR
j ωc
Now total impedance of the branch consisting of Cs′ and R and C.
L 1 R O
Z = M  P
N j ωCs ′ 1  jω CR Q
Therefore, if unity voltage is applied, the current through this branch.
Is′ = 1
=
j ω C s ′ (1  j ω CR)
1 R  j ω CR j ω C s ′ R

1
j ω Cs ′ 1  j ω CR
j ω C s ′ (1  j ω CR) 1 j ω Cs ′
or I2 = . =
 j ω R(C  Cs ′ )
1  j ω R ( C  C s ′ ) 1  j ω CR1
Now again with unity voltage
Is =
1

j ω Cs ′
1 1  j ω C s Rs
Rs 
j ω C
Is [1  j ω R(C  Cs ′ )]Cs  N2
I2 (1  j ω Cs Rs ) Cs ′ N1
A
l
l
J
N
T
U
W
o
r
l
d

Therefore,
Cs

N
2
C ′ N
s 1
or Cs = Cs′
N2
N1
and tan δs = ω R (Cs′ + C)
PARTIAL DISCHARGES
Partial discharge is defined as localised discharge process in which the distance between two elec-
trodes is only partially bridged i.e., the insulation between the electrodes is partially punctured. Partial
discharges may originate directly at one of the electrodes or occur in a cavity in the dielectric. Some of
the typical partial discharges are: (i) Corona or gas discharge. These occur due to non-uniform field on
sharp edges of the conductor subjected to high voltage especially when the insulation provided is air
or gas or liquid Fig. (a). (ii) Surface discharges and discharges in laminated materials on the inter-
faces of different dielectric material such as gas/solid interface as gas gets over stressed εr times the
stress on the solid material (where εr is the relative permittivity of solid material) and ionization of gas
results Fig. (b) and (c). (iii) Cavity discharges: When cavities are formed in solid or liquid insulat-ing
materials the gas in the cavity is over stressed and discharges are formed Fig. 6.18 (d) (iv). Treeing
Channels: High intensity fields are produced in an insulating material at its sharp edges and this
deteriorates the insulating material. The continuous partial discharges so produced are known as
Treeing Channels Fig. (e).
External Partial Discharge
External partial discharge is the process which occurs external to the equipment e.g. on overhead
lines, on armature etc.
Internal Partial Discharge
Internal paratial discharge is a process of electrical discharge which occurs inside a closed system
(discharge in voids, treeing etc). This kind of classification is essential for the PD measuring system as
external discharges can be nicely distinguished from internal discharges. Partial discharge measure-
ment have been used to assess the life expectancy of insulating materials. Even though there is no well
defined relationship, yet it gives sufficient idea of the insulating properties of the material. Partial
discharges on insulation can be measured not only by electrical methods but by optical, acoustic and
chemical method also. The measuring principles are based on energy conversion process associated
with electrical discharges such as emission of electromagnetic waves, light, noise or formation of
chemical compounds. The oldest and simplest but less sensitive is the method of listening to hissing
sound coming out of partial discharge. A high value of loss factor tan δ is an indication of occurrence
of partial discharge in the material. This is also not a reliable measurement as the additional losses
gener-ated due to application of high voltage are localised and can be very small in comparison to the
volume losses resulting from polarization process. Optical methods are used only for those materials
which are transparent and thus not applicable for all materials. Acoustic detection methods using
ultrasonic trans-ducers have, however, been used with some success. The most modern and the most
accurate methods are the electrical methods. The main objective here is to separate impulse currents
associated with PD from any other phenomenon.
A
l
l
J
N
T
U
W
o
r
l
d

(a)
(b) (c)
(d ) (e)
Fig. Various partial discharges
The Partial Discharge Equivalent Circuit
If there are any partial discharges in a dielectric material, these can be measured only across its
terminal. Fig. 6.19 shows a simple capacitor arrangement in which a gas filled void is present. The partial
discharge in the void will take place as the electric stress in the void is εr times the stress in the rest of the
material where εr is the relative permittivity of the material. Due to geometry of the material, various
capacitances are formed as shown in Fig. 6.19 (a). Flux lines starting from electrode and termi-nating at the
void will form one capacitance Cb1 and similarly Cb2 between electrode B and the cavity. Cc is the
capacitance of the void. Similarly Ca1 and Ca2 are the capacitance of healthy portions of the dielectric on
the two sides of the void. Fig. 6.19 (b) shows the equivalent of 6.19 (a) where Ca = Ca1 + Ca2, and Cb =
Cb1Cb2/(Cb1 + Cb2) and Cc is the cavity capacitance. In general Ca >> Cb >> Cc.
Electrode A
C
b1 Cb
C
a1 Cc
C
a2 V Ca S
Insulating Cc ic
material
C
b2
Rc
(a)
B
(b)
Fig. (a) Dielectric material with a cavity (b) Equivalent circuit
Closing of switch S is equivalent to simulating partial discharge in the void as the voltage Vc
across the void reaches breakdown voltage. The discharge results into a current ic(t) to flow. Resistor
Rc simulates the finite value of current ic(t).
A
l
l
J
N
T
U
W
o
r
l
d

Suppose voltage V is applied across the electrode A and B and the sample is charged to this voltage
and source is removed. The voltage Vc across the void is sufficient to breakdown the void. It is equivalent
to closing switch S in Fig. 6.19 (b). As a result, the current ic(t) flows which releases a charge ∆qc = ∆VcCc
which is dispersed in the dielectric material across the capacitance Cb and Ca. Here
∆Vc is the drop in the voltage Vc as a result of discharge. The equivalent circuit during redistribution
of charge ∆qc is shown in Fig. 6.20.
C
b
Vc
A
C
a V
B
Fig. after discharge
The voltage as measured across AB will be
∆V =
Cb
∆Vc 
Cb ∆ qc
C
a  Cb
C
a  Cb
C
c
Ordinarily ∆ Vc is in kV whereas ∆V is a few volts since the ratio Cb/Ca is of the order of 10
–4
to
10
–3
. The voltage drop ∆V even though can be measured but as Cb and Cc are normally not known neither
∆Vc nor ∆qc can be obtained. Also since V is in kV and ∆V is in volts the ratio ∆V/V is very small
≈ 10
–3
, therefore the detection of ∆V/V is a tedious task.
Suppose, that the test object remains connected to the voltage source Fig. 6.21. Here Ck is the
coupling capacitor. Z is the impedance consisting either only of the lead impedance of or lead impe-dance
and PD-free inductor or filter which decouples the coupling capacitor and the test object from the source
during discharge period only, when very high frequency current pulse ic (t) circulate between
Ck and Ct. Ct is the total equipment capacitance of the test specimen.
Fig.
It is to be noted that Z offers high impedance to circular current (impulse currents) and, there-
fore, these are limited only to Ck and Ct. However, supply frequency displacement currents continue
to flow through Ck and Ct and wave shapes of currents through Ck and Ct are shown in Fig. 6.22.
A
l
l
J
N
T
U
W
o
r
l
d

Fig. Current wave forms in Ck and Ct.
It is interesting to find that pulse currents in Ck and Ct have exactly same location but opposite
polarities and these are of the same magnitude. Therefore, one can say that these pulse currents are not
supplied by the source but are due to local partial discharges. The amplitude of pulses depends upon
the voltage applied and the number of pulses depends upon the number of voids. The larger the
number of faults the higher the number of pulses over a half cycle.
During discharge, the voltage across the test object Ct falls by an amount ∆V and during this
period Ck stores the energy and release the charge between Ck and Ct thus compensating the drop ∆V.
The equivalent capacitance of the test specimen is Ct ≈ Ca + Cb assuming Cc to be negligibly small. If
Ck >> Ct, the charge transfer is given by
q =
zi (t) dt ≈ (Ca + Cb) ∆V
C
b
Now ∆V = ∆Vc
C
a  Cb
and ∆V =
q
Ca  Cb
Therefore, q  Cb ∆Vc
 Cb
C
a  Cb
C
a
or q = Cb ∆VC
Here q is known as apparent charge as it is not equal to the charge locally involved i.e. Cc ∆Vc .
This charge q is, however, more realistic than calculating ∆V, as q is independent of Ca whereas ∆V
depends upon Ca.
A
l
l
J
N
T
U
W
o
r
l
d

In practice the condition Ck >> Ct is never satisfied as the Ck will over load the supply and also
it will be uneconomical. However, if Ck is slightly greater than Ct, the sensitivity of measurement is
reduced as the compensating current ic (t) becomes small. If Ct is comparable to Ck and ∆V is the drop
in voltage of Ct as a result of discharge, the transfer of charge between Ct and Ck will result into
common voltage ∆V ′.
∆V ′ = Ct ∆ V  Ck .O = Ct ∆ V  q
Ct  Ck
Ct  Ck Ct  Ck
∆V ′ is the net rise in voltage of the parallel combination of Ck and Ct and, therefore, the charge
qm transferred to Ct from Ck will be
qm = Ck ∆ V ′
The charge qm is known as measurable charge. The ratio of measurable charge to apparent
charge is, therefore, given as
qm  Ck
q Ct  Ck
In order to have high sensitivity of measurements i.e., high qm/q it is clear that Ck should be
large compared to Ct. But we know that there are disadvantages in having large value of Ck.
Therefore, this method of measurement of PD has limited applications.
The measurement of PD current pulses provides important information concerning the
discharge processes in a test specimen. The time response of an electric discharge depends mainly on
the nature of fault and design of insulating material. The shape of the circular current is an indication
of the physical discharge process at the fault location in the test object. The principle of measurement
of PD current is shown in Fig. 6.23.
Fig. Principle of pulse current measurement
Here C indicates the stray capacitance between the lead of Ct and the earth, the input capaci-
tance of the amplifier and other stray capacitances. The function of the high pass amplifier is to sup-
press the power frequency displacement current ik(t) and Ic(t) and to further amplify the short duration
current pulses. Thus the delay cable is electrically disconnected from the resistance R. Suppose during
a partial discharge a short duration pulse current δ (t) is produced and results in apparent charge q on
Ct which will be redistributed between Ct, C and Ck . The circuit for the same is given in Fig. 6.24.
A
l
l
J
N
T
U
W
o
r
l
d

Potential across Ct =
q
Ct 
CCk
C  Ck
Therefore, voltage across C will be
q Ck q
v = . = . Ck
CCk
Ct  C  Ck Ct (C Ck )  CCk
C  Ck
= qCk = q
CCt  Ck Ct F C I
 CCk
C  Ct G1  J
H Ck K
and because of resistance R the expression for voltage across R will be
Fig. after discharge
v(t) =
q
e− t / τ
F C I
C  Ct G1  J
H Ck K
F Ct Ck I
where τ =
GC  J R
H Ct  Ck K
The wave shape of the voltage pulse is shown in
Fig. 6.25.
The voltage across the resistance R indicates a
fast rise and is followed by an exponential decay with
time constant τ. The circuit elements have thus
deferred the original current wave shape especially the
wave tail side of the wave and therefore, the
measurement of the pulse current i(t) is a difficult task.
Also, the PD pulse currents get corrupted due to
various interferences present in the system. The power
frequency displacement current ik (t) and it (t) are major
sources of interference as these currents vary from a few mA to several amperes. Higher harmonics in
the supply and pulse current in the thyristorized control circuits are always present which will interfere
with the PD currents. On load taps in a trans-former, carbon brushes in a generator are yet other
sources of noise in the circuits. Mainly interferences can be classified as follows:–
(i) Pulse shaped noise signals: These are due to impulse phenomenon similar to PD currents.
(ii) Harmonic signals: These are mainly due to power supply and thyristorised controllers. We
are taking apparent charge as the index level of the partial discharges which is integration of
PD pulse currents. Therefore, continuous alternating current of any frequency would disturb the integration
process of the measuring circuit and hence it is important that these currents (other than PD
Fig. Wave shape of voltage across R
A
l
l
J
N
T
U
W
o
r
l
d

currents) must be suppressed before the mixture of currents is passed through the integrating circuit.
The solution to the problem is obtained by using filter circuits which may be completely independent
of integrating circuits.
Fig. shows two different ways in which the measuring impedance Zm can be connected in the
circuit.
Z
Ct
V Ck
M
Z
m
(a)
Z
Ck
Ct
Zm
M
(b)
Fig.
In Fig. 6.26 (a) Zm is connected in series with Ct and provides better sensitivity as the PD
currents excited from Ct would be better picked up by measuring circuit Zm. However, the
disadvantage is that in case of puncture of the test specimen the measuring circuit would also be
damaged. Specifically for this reason, the second arrangement in which Zm is connected between the
ground terminal of Ck and the ground and is the circuit most commonly used.
As is mentioned earlier, according to international standards the level of partial discharges is
judged by quantity of apparent charge measured. The apparent charge is obtained by integration of the
circular current ic(t). This operation is carried out on the PD pulses using ‗wide band‘ and ‗narrow
band‘, measuring systems. These are basically band pass filters with amplifying action. If we examine
the frequency spectrum of the pulse current, it will be clear why band pass filters are suitable for
integrating PD pulse currents.
We know that for a non-periodic pulse current i(t), the complex frequency spectrum of the
current is given by Fourier transform as
I (j ω) =
z−
∞
∞i ( t ) e − jωt
dt
Let i (t) = I0e
-t/τ as an approximation to actual PD pulse current
∞
I 0 e − t / τ e − j ωt
dt = I0
∞
I (j ω) =
z0
e− (1/ τ  j ω )t dt
z0
A
l
l
J
N
T
U
W
o
r
l
d

I0 L− ( 1/ τ  jω )t O∞
= – Me
P
1 / τ  jω N Q0
= –
I0 τ
[ 0 − 1] 
I0 τ
1  jωτ 1  jωτ
The amplitude | I (j ω) | = I0 τ
ω
2
τ
2
1
and phase angle – tan
–1 ωτ 1
Fig. (a) shows the PD current pulse and 6.27 (b) shows the amplitude I (j ω) vs. frequency
plot.
| (j)|
Io
q = Io
 t 
Fig. (a) PD current pulse (b) Frequency spectrum of the pulse current
Here the current is approximated by an exponentially decaying curve and, therefore, neither i
(t) nor I (j ω) vanish and so a new measure of ―pulse width‖ is required. The time constant τ is a
measure of the width of i(t), a line tangent to i(t), a line tangent to i(t) at t = 0 intersects the line i = 0 at
t = τ as shown in Fig. (a).
From the above expression and Fig. (b) it is clear that as ω → 0 I (j ω) → I0 τ which means that
the d.c. content of the frequency spectrum equals the apparent charge in the pulse current. There-fore,
the frequency spectrum of PD pulses current contains complete information concerning the ap-parent
charge in the low frequency range. In order to have proper integration of the pulse current, the time
constant τ of the pulse should be greater than the time constant of the measuring circuit or the band
width (upper cut off frequency) of the measuring system should be much lower than that of the spec-
trum of the pulse currents to be measured.
Wide-Band Circuit
Fig. shows the principle of wide band circuits.
The coupling impedance Zm is a parallel combination of R, L and C whose quality factor is low.
The complex impedance Zm is given as
1  1  1  j ω C
Zm R j ω L
A
l
l
J
N
T
U
W
o
r
l
d

o  (t)
R L C Vo (t)
BP
Amp.
CRO
Fig. 6.28 PD measuring circuitc wide band
or Zm =
j ω RL
=
R
j ω L  R − RLC ω
2
1 − j
R
 jRCω
ω L
| Zm | =
R
=
R
L 2 O
1 / 2
 F R − RCω
I L
R
2
F
2
I2 O
1 / 2
1 P M1  1 − ω
P
M G J
ω 2
L G ω 2
J
M H ω L K P M 0 P
H K
N Q N Q
=
R
=
R
L F ω
2
I2
O1 / 2
2 C1 L 2 ω 0
2
ω
2
F ω
2
I2
O
1 / 2
M1  R
L
.
CL ω
2 G1 − 2 J P M1  Q
ω
2
ω
2 G1
−
ω
2 J P
M H ω 0 K P M 0 H 0 K P
N Q N Q
=
R
=
R
L F ω 0 I 2
O1 / 2
2 ω L 2 F f0 f I2 O
1 / 2
M1  Q G
ω
−
ω 0
J P M1  Q G
f
− f
0
J P
M H K P M H K P
N Q N Q
R 1
C
t
C
k
Where Q = and f0 = Where C = Cs + Cc +
2π LC Ct  Ck
L / C
Cs is the stray capacitance, Cc the capacitance of the delay cable.
The measuring impedance Zm is the impedance of a band pass filter which suppresses harmonic
currents depending upon the selected circuit quality factor Q, below and above the resonance frequency f0
i.e., Zm will suppress all frequency currents below and above its resonance frequency. The alternate is
– 20 dB per decade if Q = 1 and can be greatly increased.
Also, the measuring circuit Zm performs integration of the PD pulse currents i (t) = I0 δ (t).
The voltage v0(t) as shown in Fig. 6.28 can be obtained by writing nodal equation
V0 (s)

V0 (s)
 V0 (s) Cs  I0
R sL
V0 (s) 
RsLI0
=
I0 s/C
RLCs
2
 sL  R s
2
 s  1
LC
RC
Dharm
N-HIGHHG6-2.PM5 189
= I 0 s / C  I 0 s / C F
H
A
l
l
J
N
T
U
W
o
r
l
d

2 RC LC ( 4 R
2
C
2
)
where α = 1 and β  1 − α
2
2RC LC
q L s O
Therefore, V0(s) = . M P
C
2
β
2
N( s  α ) Q
L
q s q s  α α
V
0 (s) =  M −
C ( s  α )
2
β
2
 α )
2
β
2
( s  α )
2
β
2
C N ( s
q L s  α α β O
= M − . P
2
β
2
β ( s  α )
2
β
2
C N( s  α ) Q
q L − αt α − αt O
= Me
cos βt −
β
e sin β t
P
C N Q
O
P
Q
q − αt L α O
= e Mcos β t −
β
sin β t
P
C N Q
The above equation shows a damped oscillatory output voltage where amplitude is proportional to
the charge q. The charge due to the pulse i(t) is actually stored by the capacitor C instantaneously but due to
the presence of inductance and resistance, Oscillations are produced. If these oscillations are not damped,
the resolution time of the filter will be large and proper integration will not take place espe-cially of the
subsequent current pulses. There is a possibility of over lapping and the results obtained will be erroneous.
The resolution time as is said earlier should be smaller than the time constant τ of the current pulse [i (t)
I0e
-t/τ]. The resolution time or decay time depends upon the Q-factor and resonance
frequency f0 of the measuring impedance Zm. Let Q = 1 = R/ LC . Therefore R = L / C and
α =
2
1 
ω0
 πf0
(L / C) . C2
Suppose the resolution time is
T =
1
and f0  100 kHz
f0
The resolution time is about 10  sec and for higher values of Q, T will be still larger. The
resonance frequency is also affected by the coupling capacitance Ck and the capacitance Ct of the test
specimen as these contribute to the formation of C. Therefore, the R L C circuit should be chosen or
selected according to the test specimen so that a desired resonance frequency is obtained. The desired
central frequency f0 or a band width around f0 is decided by the band pass amplifier connected to this
resonant circuit. These amplifiers are designed for typically lower and upper cut off frequencies
A
l
l
J
N
T
U
W
o
r
l
d

(– 3 dB) between 150 kHz and 100 kHz. This band of frequency is chosen as it is much higher than the
power supply frequency and also the frequency which are not used by broadcasting stations. The
resolution time becomes less than 10  sec. and hence proper integration of the current pulse is made
possible. However, the main job of the amplifier is to increase the sensitivity of the whole measuring
system. The time dependency of the output voltage v0 (t) can be seen on the oscilloscope. In the usual
ellipse representation, the individual pulse v0 (t) are practically only recognizable on vertical lines of
different heights as one rotation of the ellipse corresponds to one period of the supply system 20 m
sec. for 50 Hz and 16.7 m sec. for 60 Hz supplies. Fig. 6.29.
The magnitude of the individual discharge is quantified by comparing the pulse crest value
with the one obtained from the calibration circuit as shown in Fig. 6.30. The calibration circuit
consists of a voltage step generator V0 and a series capacitor C0. The charge q is simulated with no
normal voltage applied to the PD testing circuit. It is possible to suggest the location of the partial
discharges in an insulating material by looking at the display on the CRO screen.
Fig.
Fig. Circuit for calibration of the oscilloscope
Narrow Band PD-Detection Circuit
A narrow band PD detection circuit is basically a very sensitive measurement receiver circuit with a
continuously variable measuring or centre frequency fm in the range of approximately 50 kHZ to sev-
eral MHz. The nomenclature to narrow-band is justified as the band width of the filter amplifier is
typically only 9 kHz. However, if special circumstances demand, the band width may be slightly made
wider or narrower than 9 kHz.
A
l
l
J
N
T
U
W
o
r
l
d

Z
V0 (t)
NB
Peak
V1 (t) V2
level P C
V
2
(t)
ampl.
R
indi- meter
L cator
Fig. Basic narrow band PD measuring circuit
Parallel combination of R and L constitute the measuring impedance Zm. The measuring impe-dance
acts as a high pass and high frequency PD currents pulses i (t) are decoupled from the test circuit. Whereas
in wide band circuits the measuring impedance Zm (R || L || C) performs integration operation on the input
Dirac delta current i(t), no integration is carried out by Zm in the narrow band circuit. A low resistance
rating of the measuring impedance Zm prevents that the series connection of Ck and Ct at-tenuates high
frequency components of PD signals. Since the delay cable is terminated with Z0 which is the surge
impedance of the cable itself the capacitance Cc of the cable does not play any role.
Assuming that the parallel combination of R and L is so chosen that L does not perform
integrating operation on the input signal i(t) = I0 δ (t), the voltage v1(t) at the input of the narrow band
amplifier is proportional to the PD impulse current i(t) i.e.,
v1 (t) = I0 δ (t) Rm
Again, assuming that i (t) = I0 e
–t/τ as in Fig. 6.27, we have
v1 (t) = I0e
–t/τ Rm
V1 (j ω) =
I0 τ Rm

V0 τ
1  j ω τ 1  j ω τ
where Rm 
RZ0
R  Z0
The time constant of the circuit T = RmC
where C =
C
t
C
k
Ct  Ck
Let S0 = V0 τ
The quantity S0 contains the information concerning the individual pulse charge q and is referred to
as the integral signal amplitude and is represented in Fig. 6.32.
V0 V 1 ( j)
S0 = V0
V1 (t) V0 
 t n
(a) (b)
Fig. (a) Approximate voltage impulse (b) Its frequency response
A
l
l
J
N
T
U
W
o
r
l
d

The input voltage to the narrow band amplifier is, therefore, represented as
v1 (t) = V0e
–t/τ
Fig. 6.32 (b) shows impulse response of the measuring impedance circuit. This voltage impulse
now is the input to the narrow band amplifier. So our objective now is to find the impulse response of
the amplifier. However, the impulse response of any network is the transfer function of the network
itself.
Assume an idealised transfer function of a narrow band amplifier as shown in Fig. 6.33 (a)
with constant magnitude G0, mid band angular frequency ωm and an angular band width ∆ω. For such
an ideal narrow band pass amplifier, the phase shift can be assumed to be linear function of angular
frequency especially within the band pass response.
We also know that there is an inherent time delay between the impulse excitation and the response
output of the network. Let this delay be t0. The output response of the narrow band receiver is given by
G ( j )
V
2 max
G0
 = t0

t

m  t0
Fig. (a) Idealised transfer function of narrow band receiver (b)
Impulse response of (a)
V2 (j ω) = G (j ω) V1 (j ω)
and since v1(t) = S0 δ (t); V1 (j ω) = S0
G(jω) =G e
− jωt
0
0
1 ω m  ∆ ω
Therefore v2(t) = zω m −
∆ ω
2
S0G0e− jωt
0 e jωt dω
π
2
S0 G0 Le jω ( t − t0 )
Oω m  ∆ω / 2
= M P
π
( t − t 0
)
N j Qω m − ∆ω / 2
ej ( ω m − ∆ ω / 2 ) ( t − t0 )
O
S G Le j ( ω m  ∆ ω / 2 ) ( t − t 0 )
−
v2(t) = 0 0
M P
π j(t − t0 )
N Q
S G jω ( t − t )L e
j ∆ ω ( t − t
0
)/ 2
− e − j ∆ ω ( t − t0 )/ 2
O
= 0 0
. ∆ ωe m 0 M P
[2 j (t − t0 )/2] ∆ ω
π N Q
=
S
0
G
0
∆ ωejωm ( t − t0 ) Si
∆ ω (t−t
0
)
π 2
Neglecting the imaginary component, we have
S0 G0 L ∆ ω (t − t0 ) O
v2 (t)  ∆ ωSi M P cos ω m (t − t0 )
π 2
N Q
A
l
l
J
N
T
U
W
o
r
l
d

= 2S G ∆fSi L ∆ ω (t − t
0
)
O cos ω
m
(t − t
0
)
0 0 M
2 P
N Q
=
V Si L ∆ ω (t − t
0
)
O cos ω
m
(t − t
0
)
2 max M
2
P
N Q
Fig. 6.33 (b) shows the plot of the above equation. It is clear that the impulse response of the
narrow band pass receiver is an oscillatory one whose main frequency is given by fm and the
amplitudes are given by Si function which is the envelope of the oscillations. The maximum value
V2max = 2S0G0 ∆f where ∆f is the idealised band width of the amplifier. There are two main
disadvantages of the narrow band amplifier:
(i) If ∆ω << ωm, the positive and negative peak values of the response are equal and hence
the polarity of the input pulse can‘t be detected.
(ii) The duration of the response is quite long.
Ideally, after the first current zero of the response
sin x
x
the amplitude should decrease to very low value for proper measurements. If we redraw Fig. 6.33 (b)
with normalised value of v2(t) along ordinate and abscissa ∆f (t – t0) we obtain the varia-tion as shown
in Fig. 6.34.
We know the response as shown in Fig. 6.32 or 6.33 is that of an ideal distortion free system.
However, the pulse response of a real filter does not show such pronounced oscillation outside of ∆
f(t– t0), =  1 i.e., in a real system after the first current zero of the response
sin x
x
the oscillations reduce to negligibly small values.
–3 –2 –1 1 2 3  f ( t – t )

r
Fig.
The response time, therefore, is defined as
τ r 
2
∆ f
For a typical value ∆f = 9 kHz, the response time τr = 220  sec. which is a much longer period
as compared to in case of a wide band amplifier circuit. τr is the resolution time of the circuit and
should be of the same order as that of the time constant of the input pulse for accurate measurement
otherwise it leads to overlapping of pulses and the measurement becomes erroneous.
The first peak of the response is an indication of the charge q of the PD current pulses and can
be detected by a peak level detector (pc meter). Fig.
A
l
l
J
N
T
U
W
o
r
l
d

The narrow band PD detectors use radio interference voltage (RIV) meters for measurement of
apparent charge. The main component of an RIV meter is a selective voltmeter of high sensitivity
which can be tuned within the frequency range of interest. The selectivity may be reached by a narrow
band pass filter characteristic and thus RIV is primarily a super-heterodyne type receiver, though a
straight type narrow band receiver may also be considered as a high quality linear amplifier with a
band pass filter characteristic and of sufficient amplification to give high sensitivity. The mid band
fre-quency fm should be continuously variable and fm should be treated as a resonance frequency f0 as
this is suggested in the IEC recommendations for PD measurements.
Table 6.1 below gives comparison between the wide band and narrow band receiver circuits for
measurement of PD current pulses.
Table
Comparison between wide band and narrow band PD measuring circuits
Wide Band Narrow Band
1. Bandwidth f2–f1 = 150 to 200 kHz ∆f = 9 kHz
2. Centre frequency Fixed f0 = 80 – 150 kHz Variable fm = 50 kHz to 2 MHz
3. Pulse resolution time Small about 15  sec. Large about 220  sec.
4. Pulse polarity Detectable Not detectable
5. Noise susceptibility Relatively high as no. of Low due to selective measurements
interference sources through variable centre frequency.
increases with band width
6. Maximum admissible PD Approx 1  sec. Depends upon fm in
pulse width
7. Indication of measured Directly in pC Directly in pC
value
Table above shows relative merits and demerits of the two circuits. However, in practical situa-
tions, a system that can be switched over between wide band and narrow band should prove to be
more versatile and useful.
BRIDGE CIRCUIT
Fig shows a bridge circuit used to suppress parasitic interference signals effectively. The circuit
mainly consists of two individually balanceable measuring conductances GA and GB which are in-
serted into the ground conductors of capacitors Ck and Ct. The bridge can be balanced by a calibration
generator when it is not energized. The calibration generator is connected between ground, the high
voltage common terminal of the Ck and Ct in order to simulate external interference signals.
A
l
l
J
N
T
U
W
o
r
l
d

Fig. 6.35 Basic principle of bridge circuit
The setting of the conductances GA and GB for balance depends upon the capacitances Ck and
Ct. However, in order to have high sensitivity of the bridge, the conductances should be as low as
possible. The practical limit is given by the current carrying capacity of the individual components in
the bridge arms GA and GB. If the conductances are not of proper value, the current pulses are so long
(τ = Ck/GA = Ct/GB) that these would involve errors in measurements. Sometimes the instruction
manual suggests suitable values of GA and GB for the purpose.
If VA and VB are of the same magnitude and having the same polarity, the measuring
instrument shows zero deflection. When an external signal occurs, this will induce same current-I in
both the circuit branches which in turn would cause same voltage drops across GA and GB and hence
the measuring instruments would read zero. However, if there is partial discharge in Ct, circular
current i(t) will flow and the polarity of VA and VB will now be opposed and the differential voltage
VAB will be indicated by the meter. The measuring instrument can be of either wide band or narrow
band type. With this circuit, it is possible to obtain interference suppression factors varying from 1:20
to 1:1000 depending upon the construction of capacitances Ct and Ck and the type of source of
interference. If the capacitances Ct and Ck are identical, interference suppression factors of 1:1000 can
be achieved as the entire arrangement is then symmetrical. The test object should be placed close to
each other so that the interference effect is identical. Under highly unfavourable condition e.g. PD
testing of outdoor switch gear where an ambient noise level exists because of corona discharges and
radio interferences, the effective interference suppression factors may be only around 1: 20. Also if the
values of Ct and Ck differ materially, lower interference suppression factor is then expected as the
bridge will have to be balanced over a large frequency range and the bridge has then less accuracy of
measurement. However, the narrow band measuring system makes it possible to measure with a
variable centre frequency at which the bridge circuit achieves maximum interference suppression.
OSCILLOSCOPE AS PD MEASURING DEVICE
Oscilloscope is an integral and indispensable component of a PD measuring system. An indicating meter
e.g. a pC meter and RIV meter can give quantity of charge, whether the charge is as a result of partial
discharge or due to external interferences, cannot be estimated. This problem can be solved only if the
output wave form is studied on the Oscilloscope. Whether the origin of the discharges is from within the
test object or not, can frequently be determined based on the typical patterns. If it is ascer-tained from the
patterns that the discharge is from the test object, the magnitude of the apparent charge should be measured
with pC meters or RIV meters. The peak value of the integrated pulse current is the
A
l
l
J
N
T
U
W
o
r
l
d

desired apparent charge q. These signals are normally superposed on the a.c. test voltage for observa-
tion on the Oscilloscope. Depending upon the preferences either sine or elliptical shapes can be se-
lected. One complete rotation of the ellipse or one complete cycle of sine wave equals 20 m sec. of
duration. Since the duration of these current pulses to be measured is a few microsecond, these pulses
when seen on the power frequency wave, look like vertical lines of varying heights superimposed on
the power frequency waves.
Whenever calibration facility exists in the PD test circuit, the calibration curve of known
charge appears on the screen. The calibration pulse can be shifted entirely along the ellipse or sine
curve of the power supply and the signal to be measured can be compared with the calibration pulse.
RECURRENT SURGE GENERATOR
The power transformers, the power transmission lines and rotating machines are exposed to lightning
surges and, therefore, these should be impulse tested during their design and development stages. The
surge voltage distribution along the transformer winding is very important to know especially for the
design of its high voltage insulation. For such purposes it is not desirable to subject the winding to its
full withstand voltage rather low impulse voltage should be used to avoid risk of damage to the
winding during test and to reduce the cost of test apparatus. Therefore, low voltage test impulses
having the same wave form as the standard high voltage surges and to which transformers respond in
more or less a linear fashion have been suggested. The low voltage impulses are synchronised with the
recurrent time base of a CRO so that what is observed on the screen of the CRT is an apparently
steady state and the distribution of voltage along the transformer winding can be studied.
Fig. 6.36 shows a schematic diagram of recurrent generator developed by Rohats. The arrange-
ment of elements R1, R2 and C2 is similar to that of circuit ‗a‘ of Fig. 3.4 R1 controls the wave front
time and R2 the wave tail time.
Fig. Basic circuit of recurrent surge generator
A
l
l
J
N
T
U
W
o
r
l
d

When T1 conducts, it charges the capacitance C4 through R4 which gives deflection on the
time axis of the CRO. At the same time a positive impulse is fed through C5 to the grid of T2 which
conducts. C1 which was charged in the previous half cycle, starts discharging through L, R1, R2 and
C2. A part of the output across R3 is fed into the deflection plates of CRO. The time base is adjusted
with the resist-ance R4.
During the past many years, many circuits have been developed. One of the most modern
recur-rent surge generators is shown in Fig. 6.37.
Fig. Recurrent surge generator circuit
In this circuit both the full impulse wave and chopped impulse waves can be obtained by
controlling the firing of thyristor Th1 and Th2 through surge trigger circuit and chop trigger circuits
respectively. The wave front time and wave till timings are controlled using R1 and R2 respectively.
C1 corresponds to the total impulse capacitance and C2 the load capacitance. The impulse generation
sequence is repeated and synchronised with the mains frequency. The impulse can also be trigged
manually in a non-recurrent manner. In case the impulse circuit elements R1, R2 L are insufficient,
additional external resistance can be connected.
Fig. shows arrangement of recurrent surge generator for measurement of inter turn potential of the
transformer winding. Similarly the potential at any point of the winding can also be studied.

A
l
l
J
N
T
U
W
o
r
l
d

UNIT-VIII
HIGH VOLTAGE TESTING OF ELECTRICAL APPARATUS
Introduction
Man has been power hungry since time-immemorial. In modern times the world has seen phenomenal
increase in demand for energy, of which an important component is that of electrical energy. The
production of electrical energy in big plants under the most economic condition makes it necessary that
more and more energy be transported over longer and longer distances. Therefore, transmission at extra
high voltages and the erection of systems which may extend over whole continents has become the most
urgent problems to be solved in the near future. The very fast development of systems is followed by
studies of equipment and the service conditions they have to fulfill. These conditions will also determine
the values for testing at alternating, impulse and d.c. voltages under specific conditions.
As we go for higher and higher operating voltages (say above 1000 kV) certain problems are
associated with the testing techniques. Some of these are:
(i) Dimension of high voltage test laboratories.
(ii) Characteristics of equipment for such laboratories.
(iii) Some special aspects of the test techniques at extra high voltages.
The dimensions of laboratories for test equipments of 750 kV and above are fixed by the
following main considerations:
(i) Figures (values) of test voltages under different conditions.
(ii) Sizes of the test of equipments in a.c., d.c. and impulse voltages.
(iii) Distances between the objects under high voltage during the test period and the earthed
surroundings such as floors, walls and roofs of the buildings. The problems associated with
the characteristics of the equipments used for testing are summarised here.
In alternating voltage system, a careful choice of the characteristics of the testing transformer is
essential. It is known that the flash over voltage of the insulator in air or in any insulating fluid
depends upon the capacitance of the supply system. This is due to the fact that a voltage drop may not
maintain preliminary discharges or breakdown. It is, therefore, suggested that a capacitance of at least
1000 pF must be connected across the insulator to obtain the correct flash over or puncture voltage
and also under breakdown condition (a virtual short circuit) the supply system should be able to
supply at least 1 amp for clean and 5 amp for polluted insulators at the test voltage.
There are some difficult problems with impulse testing equipments also especially when testing
large power transformers or large reactors or large cables operating at very high voltages. The equiva-
lent capacitance of the impulse generator is usually about 40 nano farads independent of the operating
voltage which gives a stored energy of about 1/2 × 40 10
–9
× 36 × 10
9
= 720 KJ for 6 MV generators
A
l
l
J
N
T
U
W
o
r
l
d

which is required for testing equipments operating at 150 kV. It is not at all difficult to pile up a large
number of capacitances to charge them in parallel and then discharge in series to obtain a desired
impulse wave. But the difficulty exists in reducing the internal reactance of the circuit so that a short
wave front with minimum oscillation can be obtained. For example for a 4 MV circuit the inductance
of the circuit is about 140 H and it is impossible to test an equipment with a capacitance of 5000 pF
with a front time of 1.2  sec. and less than 5% overshoot on the wave front.
Cascaded rectifiers are used for high voltage d.c. testing. A careful consideration is necessary
when test on polluted insulation is to be performed which requires currents of 50 to 200 mA but ex-
tremely predischarge streamer of 0.5 to 1 amp during milliseconds occur. The generator must have an
internal reactance in order to maintain the test voltage without too high a voltage drop.
TESTING OF OVERHEAD LINE INSULATORS
Various types of overhead line insulators are (i) Pin type (ii) Post type (iii) String insulator unit (iv)
Suspension insulator string (v) Tension insulator.
Arrangement of Insulators for Test
String insulator unit should be hung by a suspension eye from an earthed metal cross arm. The test
voltage is applied between the cross arm and the conductor hung vertically down from the metal part
on the lower side of the insulator unit.
Suspension string with all its accessories as in service should be hung from an earthed metal
cross arm. The length of the cross arm should be at least 1.5 times the length of the string being tested
and should be at least equal to 0.9 m on either side of the axis of the string. No other earthed object
should be nearer to the insulator string then 0.9 m or 1.5 times the length of the string whichever is
greater. A conductor of actual size to be used in service or of diameter not less than 1 cm and length
1.5 times the length of the string is secured in the suspension clamp and should lie in a horizontal
plane. The test voltage is applied between the conductor and the cross arm and connection from the
impulse generator is made with a length of wire to one end of the conductor. For higher operating
voltages where the length of the string is large, it is advisable to sacrifice the length of the conductor
as stipu-lated above. Instead, it is desirable to bend the ends of the conductor over in a large radius.
For tension insulators the arrangement is more or less same as in suspension insulator except that it
should be held in an approximately horizontal position under a suitable tension (about 1000 Kg.).
For testing pin insulators or line post insulators, these should be mounted on the insulator pin
or line post shank with which they are to be used in service. The pin or the shank should be fixed in a
vertical position to a horizontal earthed metal cross arm situated 0.9 m above the floor of the
laboratory. A conductor of 1 cm diameter is to be laid horizontally in the top groove of the insulator
and secured by at least one turn of tie-wire, not less than 0.3 cm diameter in the tie-wire groove. The
length of the wire should be at least 1.5 times the length of the insulator and should over hang the
insulator at least 0.9 m on either side in a direction at right angles to the cross arm. The test voltage is
applied to one end of the conductor.
High voltage testing of electrical equipment requires two types of tests: (i) Type tests, and (ii)
Routine test. Type tests involves quality testing of equipment at the design and development level i.e.
samples of the product are taken and are tested when a new product is being developed and designed or
A
l
l
J
N
T
U
W
o
r
l
d

an old product is to be redesigned and developed whereas the routine tests are meant to check the
quality of the individual test piece. This is carried out to ensure quality and reliability of individual
test objects.
High voltage tests include (i) Power frequency tests and (ii) Impulse tests. These tests are car-
ried out on all insulators.
(i) 50% dry impulse flash over test.
(ii) Impulse withstand test.
(iii) Dry flash over and dry one minute test.
(iv) Wet flash over and one minute rain test.
(v) Temperature cycle test.
(vi) Electro-mechanical test.
(vii) Mechanical test.
(viii) Porosity test.
(ix) Puncture test.
(x) Mechanical routine test.
The tests mentioned above are briefly described here.
(i) The test is carried out on a clean insulator mounted as in a normal working condition. An
impulse voltage of 1/50  sec. wave shape and of an amplitude which can cause 50% flash over of the
insulator, is applied, i.e. of the impulses applied 50% of the impulses should cause flash over. The
polarity of the impulse is then reversed and procedure repeated. There must be at least 20 applications
of the impulse in each case and the insulator must not be damaged. The magnitude of the impulse
voltage should not be less than that specified in standard specifications.
(ii) The insulator is subjected to standard impulse of 1/50  sec. wave of specified value under
dry conditions with both positive and negative polarities. If five consecutive applications do not cause
any flash over or puncture, the insulator is deemed to have passed the impulse withstand test. If out of
five, two applications cause flash over, the insulator is deemed to have filed the test.
(iii) Power frequency voltage is applied to the insulator and the voltage increased to the speci-
fied value and maintained for one minute. The voltage is then increased gradually until flash over
occurs. The insulator is then flashed over at least four more times, the voltage is raised gradually to
reach flash over in about 10 seconds. The mean of at least five consecutive flash over voltages must
not be less than the value specified in specifications.
(iv) If the test is carried out under artificial rain, it is called wet flash over test. The insulator is
subjected to spray of water of following characteristics:
Precipitation rate 3  10% mm/min.
Direction 45° to the vertical
Conductivity of water 100 micro siemens  10%
Temperature of water Ambient +15°C
The insulator with 50% of the one-min. rain test voltage applied to it, is then sprayed for two
minutes, the voltage raised to the one minute test voltage in approximately 10 sec. and maintained there for
one minute. The voltage is then increased gradually till flash over occurs and the insulator is then
A
l
l
J
N
T
U
W
o
r
l
d

flashed at least four more times, the time taken to reach flash over voltage being in each case about 10
sec. The flash over voltage must not be less than the value specified in specifications.
(v) The insulator is immersed in a hot water bath whose temperature is 70° higher than normal
water bath for T minutes. It is then taken out and immediately immersed in normal water bath for T
minutes. After T minutes the insulator is again immersed in hot water bath for T minutes. The cycle is
repeated three times and it is expected that the insulator should withstand the test without damage to
the insulator or glaze. Here T = (15 + W/1.36) where W is the weight of the insulator in kgs.
(vi) The test is carried out only on suspension or tension type of insulator. The insulator is
subjected to a 2½ times the specified maximum working tension maintained for one minute. Also,
simultaneously 75% of the dry flash over voltage is applied. The insulator should withstand this test
without any damage.
(vii) This is a bending test applicable to pin type and line-post insulators. The insulator is sub-
jected to a load three times the specified maximum breaking load for one minute. There should be no
damage to the insulator and in case of post insulator the permanent set must be less than 1%.
However, in case of post insulator, the load is then raised to three times and there should not be any
damage to the insulator and its pin.
(viii) The insulator is broken and immersed in a 0.5% alcohol solution of fuchsin under a pres-
sure of 13800 kN/m
2
for 24 hours. The broken insulator is taken out and further broken. It should not
show any sign of impregnation.
(ix) An impulse over voltage is applied between the pin and the lead foil bound over the top and side
grooves in case of pin type and post insulator and between the metal fittings in case of suspension type
insulators. The voltage is 1/50  sec. wave with amplitude twice the 50% impulse flash over voltage and
negative polarity. Twenty such applications are applied. The procedure is repeated for 2.5, 3, 3.5 times the
50% impulse flash over voltage and continued till the insulator is punctured. The insulator must not
puncture if the voltage applied is equal to the one specified in the specification.
(x) The string in insulator is suspended vertically or horizontally and a tensile load 20% in
excess of the maximum specified working load is applied for one minute and no damage to the string
should occur.
TESTING OF CABLES
High voltage power cables have proved quite useful especially in case of HV d.c. transmission.
Under-ground distribution using cables not only adds to the aesthetic looks of a metropolitan city but
it pro-vides better environments and more reliable supply to the consumers.
Preparation of Cable Sample
The cable sample has to be carefully prepared for performing various tests especially electrical tests.
This is essential to avoid any excessive leakage or end flash overs which otherwise may occur during
testing and hence may give wrong information regarding the quality of cables. The length of the
sample cable varies between 50 cms to 10 m. The terminations are usually made by shielding the ends
of the cable with stress shields so as to relieve the ends from excessive high electrical stresses.
A cable is subjected to following tests:
(i) Bending tests.
(ii) Loading cycle test.
A
l
l
J
N
T
U
W
o
r
l
d

(iii) Thermal stability test.
(iv) Dielectric thermal resistance test.
(v) Life expectancy test.
(vi) Dielectric power factor test.
(vii) Power frequency withstand voltage test.
(viii) Impulse withstand voltage test.
(ix) Partial discharge test.
(i) It is to be noted that a voltage test should be made before and after a bending test. The
cable is bent round a cylinder of specified diameter to make one complete turn. It is then unwound and
rewound in the opposite direction. The cycle is to be repeated three times.
(ii) A test loop, consisting of cable and its accessories is subjected to 20 load cycles with a
minimum conductor temperature 5°C in excess of the design value and the cable is energized to 1.5
times the working voltage. The cable should not show any sign of damage.
(iii) After test as at (ii), the cable is energized with a voltage 1.5 times the working voltage for a
cable of 132 kV rating (the multiplying factor decreases with increases in operating voltage) and the
loading current is so adjusted that the temperature of the core of the cable is 5°C higher than its speci-
fied permissible temperature. The current should be maintained at this value for six hours.
(iv) The ratio of the temperature difference between the core and sheath of the cable and the
heat flow from the cable gives the thermal resistance of the sample of the cable. It should be within
the limits specified in the specifications.
(v) In order to estimate life of a cable, an accelerated life test is carried out by subjecting the
cable to a voltage stress higher than the normal working stress. It has been observed that the relation
between the expected life of the cable in hours and the voltage stress is given by
g 
K
n
t
where K is a constant which depends on material and n is the life index depending again on the material.
(vi) High Voltage Schering Bridge is used to perform dielectric power factor test on the cable sample. The
power factor is measured for different values of voltages e.g. 0.5, 1.0, 1.5 and 2.0 times the rated operating
voltages. The maximum value of power factor at normal working voltage does not exceed a specified value
(usually 0.01) at a series of temperatures ranging from 15°C to 65°C. The difference in the power factor
between rated voltage and 1.5 times the rated voltage and the rated voltage and twice the rated voltage does
not exceed a specified value. Sometimes the source is not able to supply charging current required by the
test cable, a suitable choke in series with the test cable helps
in tiding over the situation.
(vii) Cables are tested for power frequency a.c. and d.c. voltages. During manufacture the entire
cable is passed through a higher voltage test and the rated voltage to check the continuity of the cable. As a
routine test the cable is subjected to a voltage 2.5 times the working voltage for 10 min without damaging
the insulation of the cable. HV d.c. of 1.8 times the rated d.c. voltage of negative polarity for 30 min. is
applied and the cable is said to have withstood the test if no insulation failure takes place.
(viii) The test cable is subjected to 10 positive and 10 negative impulse voltage of magnitude as
specified in specification, the cable should withstand 5 applications without any damage. Usually, after
A
l
l
J
N
T
U
W
o
r
l
d

the impulse test, the power frequency dielectric power factor test is carried out to ensure that no
failure occurred during the impulse test.
(ix) Partial discharge measurement of cables is very important as it gives an indication of ex-
pected life of the cable and it gives location of fault, if any, in the cable.
When a cable is subjected to high voltage and if there is a void in the cable, the void breaks
down and a discharge takes place. As a result, there is a sudden dip in voltage in the form of an
impulse. This impulse travels along the cable as explained in detail in Chapter VI. The duration
between the normal pulse and the discharge pulse is measured on the oscilloscope and this distance
gives the loca-tion of the void from the test end of the cable. However, the shape of the pulse gives the
nature and intensity of the discharge.
In order to scan the entire length of the cable against voids or other imperfections, it is passed
through a tube of insulating material filled with distilled water. Four electrodes, two at the end and two in
the middle of the tube are arranged. The middle electrodes are located at a stipulated distance and these are
energized with high voltage. The two end electrodes and cable conductor are grounded. As the cable is
passed between the middle electrode, if a discharge is seen on the oscilloscope, a defect in this part of the
cable is stipulated and hence this part of the cable is removed from the rest of the cable.
TESTING OF BUSHINGS
Bushings are an integral component of high voltage machines. A bushing is used to bring high voltage
conductors through the grounded tank or body of the electrical equipment without excessive potential
gradients between the conductor and the edge of the hole in the body. The bushing extends into the
surface of the oil at one end and the other end is carried above the tank to a height sufficient to prevent
breakdown due to surface leakage.
Following tests are carried out on bushings:
(i) Power Factor Test
The bushing is installed as in service or immersed in oil. The high voltage terminal of the bushing is
connected to high voltage terminal of the Schering Bridge and the tank or earth portion of the bushing
is connected to the detector of the bridge. The capacitance and p.f. of the bushing is measured at
different voltages as specified in the relevant specification and the capacitance and p.f. should be
within the range specified.
(ii) Impulse Withstand Test
The bushing is subjected to impulse waves of either polarity and magnitude as specified in the
standard specification. Five consecutive full waves of standard wave form (1/50  sec.) are applied
and if two of them cause flash over, the bushing is said to be defective. If only one flash over occurs,
ten additional applications are made. If no flash over occurs, bushing is said to have passed the test.
(iii) Chopped Wave and Switching Surge Test
Chopped wave and switching surge of appropriate duration tests are carried out on high voltage bush-
ings. The procedure is identical to the one given in (ii) above.
(iv) Partial Discharge Test
In order to determine whether there is deterioration or not of the insulation used in the bushing, this
test is carried out. The procedure is explained in detail in Chapter-VI. The shape of the discharge is an
A
l
l
J
N
T
U
W
o
r
l
d

indication of nature and severity of the defect in the bushing. This is considered to be a routine test for
high voltage bushings.
(v) Visible Discharge Test at Power Frequency
The test is carried out to ascertain whether the given bushing will give rise to ratio interference or not
during operation. The test is carried out in a dark room. The voltage as specified is applied to the
bushing (IS 2099). No discharge other than that from the grading rings or arcing horns should be
visible.
(vi) Power Frequency Flash Over or Puncture Test
(Under Oil): The bushing is either immersed fully in oil or is installed as in service condition. This test
is carried out to ascertain that the internal breakdown strength of the bushing is 15% more than the
power frequency momentary dry withstand test value.
TESTING OF POWER CAPACITORS
Power capacitors is an integral part of the modern power system. These are used to control the voltage
profile of the system. Following tests are carried out on shunt power capacitors (IS 2834):
(i) Routine Tests
Routine tests are carried out on all capacitors at the manufacturer‘s premises. During testing, the ca-
pacitor should not breakdown or behave abnormally or show any visible deterioration.
(ii) Test for Output
Ammeter and Voltmeter can be used to measure the kVAr and capacitance of the capacitor. The kVAr
calculated should not differ by more than –5 to +10% of the specified value for capacitor units and 0
to 10% for capacitors banks.
The a.c. supply used for testing capacitor should have frequency between 40 Hz to 60 Hz, preferably
as near as possible to the rated frequency and the harmonics should be minimum.
(iii) Test between Terminals
Every capacitor is subjected to one of the following two tests for 10 secs:
(a) D.C. test; the test voltage being Vt = 4.3 V0
(b) A.C. test Vt = 2.15 V0, where V0 is the rms value of the voltage between terminals which in
the test connection gives the same dielectric stress in the capacitor element as the rated voltage Vn
gives in normal service.
(iv) Test between Line Terminals and Container (For capacitor units)
An a.c. voltage of value specified in column 2 of Table 5.1 is applied between the terminals (short
circuited) of the capacitor unit and its container and is maintained for one minute, no damage to the
capacitor should be observed.
Figures with single star represent values corresponding to reduced insulation level (Effectively
grounded system) and with double star full insulation level (non-effectively grounded system).
A
l
l
J
N
T
U
W
o
r
l
d

Table
Power frequency and impulse test voltages
(Between terminals and the container)
System Voltage Power Frequency Test Voltage Impulse Test
kV (rms) kV (rms) Voltage kV (peak)
12 28 75
24 50 125
36 70 170
72.5 140 325
145 230* 550*
275** 650**
245 395* 900*
460** 1050**
(v) IR Test
The insulation resistance of the test capacitor is measured with the help of a megger. The megger is
connected between one terminal of the capacitor and the container. The test voltage shall be d.c. volt-
age not less than 500 volts and the acceptable value of IR is more than 50 megohms.
(vi) Test for efficiency of Discharge Device
In order to provide safety to personnel who would be working on the capacitors, it is desirable to
connect very high resistance across the terminals of the capacitor so that they get discharged in about a
few seconds after the supply is switched off. The residual capacitor voltage after the supply voltage is
switched off should reduce to 50 volts in less than one minute of the capacitor is rated upto 650 volts
and 5 minutes if the capacitor is rated for voltage more than 650 volts.
A d.c. voltage 2 × rms rated voltage of the capacitor is applied across the parallel combination
of R and C where C is the capacitance of the capacitor under test and R is the high resistance
connected across the capacitor. The supply is switched off and the fall in voltage across the capacitor
as a function of time is recorded. If C is in microfarads and R in ohms, the time to discharge to 50
volts can be calculated from the formula
t = 2.3 × 10
–6
CR (log10 V – 1.7)
secs where V is the rated rms voltage of the capacitor in volts.
Type Tests
The type tests are carried out only once by the manufacturer to prove that the design of capacitor
complies with the design requirements:
(i) Dielectric Loss Angle Test (p.f. test)
High voltage schering bridge is used to measure dielectric power factor. The voltage applied is the
rated voltage and at temperatures 27°C  2°C. The value of the loss angle tan δ should not be more
than 10% the value agreed to between the manufacturer and the purchaser and it should not exceed
0.0035 for mineral oil impregnants and 0.005 for chlorinated impregnants.
A
l
l
J
N
T
U
W
o
r
l
d

(ii) Test for Capacitor Loss
The capacitor loss includes the dielectric loss of the capacitor and the V
2
/R loss in the discharge resist-
ance which is permanently connected. The dielectric loss can be evaluated from the loss angle as
obtained in the previous test and V
2
/R loss can also be calculated. The total power loss should not be
more than 10% of the value agreed to between the manufacturer and consumer.
(iii) Stability Test
The capacitor is placed in an enclosure whose temperature is maintained at 2°C above the maximum
working temperature for 48 hours. The loss angle is measured after 16 hours, 24 hours and 48 hours
using High voltage Schering Bridge at rated frequency and at voltage 1.2 times the rated voltage. If the
respective values of loss angle are tan δ1, tan δ2 and tan δ3, these values should satisfy the following
relations (anyone of them):
(a) tan δ1 + tan δ2 ≤ 2 tan δ2 < 2.1 tan δ1
or (b) tan δ1 ≥ tan δ2 ≥ tan δ3
(iv) Impulse voltage test between terminal and container
The capacitor is subjected to impulse voltage of 1/50  sec. Wave and magnitude as stipulated in
column 3 of Table 5.1. Five impulses of either polarity should be applied between the terminals
(joined together) and the container. It should withstand this voltage without causing any flash overs.
TESTING OF POWER TRANSFORMERS
Transformer is one of the most expensive and important equipment in power system. If it is not
suitably designed its failure may cause a lengthy and costly outage. Therefore, it is very important to
be cautious while designing its insulation, so that it can withstand transient over voltage both due to
switching and lightning. The high voltage testing of transformers is, therefore, very important and
would be discussed here. Other tests like temperature rise, short circuit, open circuit etc. are not
considered here. However, these can be found in the relevant standard specification.
Partial Discharge Test
The test is carried out on the windings of the transformer to assess the magnitude of discharges. The
transformer is connected as a test specimen similar to any other equipment as discussed in Chapter-VI and
the discharge measurements are made. The location and severity of fault is ascertained using the travelling
wave theory technique as explained in Chapter VI. The measurements are to be made at all the terminals of
the transformer and it is estimated that if the apparent measured charge exceeds 10
4
picocoulombs, the
discharge magnitude is considered to be severe and the transformer insulation should be so designed that
the discharge measurement should be much below the value of 10
4
pico-coulombs.
Impulse Testing of Transformer
The impulse level of a transformer is determined by the breakdown voltage of its minor insulation
(Insulation between turn and between windings), breakdown voltage of its major insulation (insulation
between windings and tank) and the flash over voltage of its bushings or a combination of these. The
impulse characteristics of internal insulation in a transformer differs from flash over in air in two main
respects. Firstly the impulse ratio of the transformer insulation is higher (varies from 2.1 to 2.2) than
that of bushing (1.5 for bushings, insulators etc.). Secondly, the impulse breakdown of transformer
A
l
l
J
N
T
U
W
o
r
l
d

KV
1 2 3 4 5 t
Fig. Volt time curve of typical major insulation in transformer
insulation in practically constant and is independent of time of application of impulse voltage. Fig. 5.1
shows that after three micro seconds the flash over voltage is substantially constant. The voltage stress
between the turns of the same winding and between different windings of the transformer depends
upon the steepness of the surge wave front. The voltage stress may further get aggravated by the piling
up action of the wave if the length of the surge wave is large. In fact, due to high steepness of the
surge waves, the first few turns of the winding are overstressed and that is why the modern practice is
to provide extra insulation to the first few turns of the winding. Fig. 5.2 shows the equivalent circuit of
a transformer winding for impulse voltage.
Fig. Equivalent circuit of a transformer for impulse voltage
Here C1 represents inter-turn capacitance and C2 capacitance between winding and the ground
(tank). In order that the minor insulation will be able to withstand the impulse voltage, the winding is
subjected to chopped impulse wave of higher peak voltage than the full wave. This chopped wave is
produced by flash over of a rod gap or bushing in parallel with the transformer insulation. The chop-ping
time is usually 3 to 6 micro seconds. While impulse voltage is applied between one phase and ground, high
voltages would be induced in the secondary of the transformer. To avoid this, the second-ary windings are
short-circuited and finally connected to ground. The short circuiting, however, de-creases the impedance of
the transformer and hence poses problem in adjusting the wave front and wave tail timings of wave. Also,
the minimum value of the impulse capacitance required is given by
C0 =
P  10
8
F
Z  V
2
where P = rated MVA of the transformer Z = per cent impedance of transformer. V = rated voltage of
transformer.
Fig. 5.3 shows the arrangement of the transformer for impulse testing. CRO forms an integral
part of the transformer impulse testing circuit. It is required to record to wave forms of the applied
voltage and current through the winding under test.
A
l
l
J
N
T
U
W
o
r
l
d

Fig. Arrangement for impulse testing of transformer
Impulse testing consists of the following steps:
(i) Application of impulse of magnitude 75% of the Basic Impulse Level (BIL) of the
transformer under test.
(ii) One full wave of 100% of BIL.
(iii) Two chopped wave of 115% of BIL.
(iv) One full wave of 100% BIL and
(v) One full wave of 75% of BIL.
During impulse testing the fault can be located by general observation like noise in the tank or
smoke or bubble in the breather.
If there is a fault, it appears on the Oscilloscope as a partial of complete collapse of the applied
voltage.
Study of the wave form of the neutral current also indicated the type of fault. If an arc occurs
between the turns or form turn to the ground, a train of high frequency pulses are seen on the oscillo-
scope and wave shape of impulse changes. If it is a partial discharge only, high frequency oscillations
are observed but no change in wave shape occurs.
The bushing forms an important and integral part of transformer insulation. Therefore, its im-
pulse flash over must be carefully investigated. The impulse strength of the transformer winding is
same for either polarity of wave whereas the flash over voltage for bushing is different for different
polarity. The manufacturer, however, while specifying the impulse strength of the transformer takes
into consideration the overall impulse characteristic of the transformer.
5.6 TESTING OF CIRCUIT BREAKERS
An equipment when designed to certain specification and is fabricated, needs testing for its perform-
ance. The general design is tried and the results of such tests conducted on one selected breaker and
are thus applicable to all others of identical construction. These tests are called the type tests. These
tests are classified as follows:
1. Short circuit tests:
(i) Making capacity test.
(ii) Breaking capacity test.
(iii) Short time current test.
(iv) Operating duty test
A
l
l
J
N
T
U
W
o
r
l
d

2. Dielectric tests:
(i) Power frequency test:
(a) One minute dry withstand test.
(b) One minute wet withstand test.
(ii) Impulse voltage dry withstand test.
3. Thermal test.
4. Mechanical test
Once a particular design is found satisfactory, a large number of similar C.Bs. are
manufactured for marketing. Every piece of C.B. is then tested before putting into service. These tests
are known as routine tests. With these tests it is possible to find out if incorrect assembly or inferior
quality material has been used for a proven design equipment. These tests are classified as (i)
operation tests, (ii) millivoltdrop tests, (iii) power frequency voltage tests at manufacturer‘s premises,
and (iv) power frequency voltage tests after erection on site.
We will discuss first the type tests. In that also we will discuss the short circuit tests after the
other three tests.
Dielectric Tests
The general dielectric characteristics of any circuit breaker or switchgear unit depend upon the basic
design i.e. clearances, bushing materials, etc. upon correctness and accuracy in assembly and upon the
quality of materials used. For a C.B. these factors are checked from the viewpoint of their ability to
withstand over voltages at the normal service voltage and abnormal voltages during lightning or other
phenomenon.
The test voltage is applied for a period of one minute between (i) phases with the breaker
closed, (ii) phases and earth with C.B. open, and (iii) across the terminals with breaker open. With this
the breaker must not flash over or puncture. These tests are normally made on indoor switchgear. For
such C.Bs the impulse tests generally are unnecessary because it is not exposed to impulse voltage of a
very high order. The high frequency switching surges do occur but the effect of these in cable systems
used for indoor switchgear are found to be safely withstood by the switchgear if it has withstood the
normal frequency test.
Since the outdoor switchgear is electrically exposed, they will be subjected to over voltages
caused by lightning. The effect of these voltages is much more serious than the power frequency
voltages in service. Therefore, this class of switchgear is subjected in addition to power frequency
tests, the impulse voltage tests.
The test voltage should be a standard 1/50  sec wave, the peak value of which is specified
according to the rated voltage of the breaker. A higher impulse voltage is specified for non-effectively
grounded system than those for solidly grounded system. The test voltages are applied between (i)
each pole and earth in turn with the breaker closed and remaining phases earthed, and (ii) between all
termi-nals on one side of the breaker and all the other terminals earthed, with the breaker open. The
specified voltages are withstand values i.e. the breaker should not flash over for 10 applications of the
wave. Normally this test is carried out with waves of both the polarities.
The wet dielectric test is used for outdoor switchgear. In this, the external insulation is sprayed
for two minutes while the rated service voltage is applied; the test overvoltage is then maintained for
30 seconds during which no flash over should occur. The effect of rain on external insulation is partly
beneficial, insofar as the surface is thereby cleaned, but is also harmful if the rain contains impurities.
A
l
l
J
N
T
U
W
o
r
l
d

Thermal Tests
These tests are made to check the thermal behaviour of the breakers. In this test the rated current
through all three phases of the switchgear is passed continuously for a period long enough to achieve
steady state conditions. Temperature readings are obtained by means of thermocouples whose hot
junc-tions are placed in appropriate positions. The temperature rise above ambient, of conductors,
must normally not exceed 40°C when the rated normal current is less than 800 amps and 50°C if it is
800 amps and above.
An additional requirement in the type test is the measurement of the contact resistances
between the isolating contacts and between the moving and fixed contacts. These points are generally
the main sources of excessive heat generation. The voltage drop across the breaker pole is measured
for different values of d.c. current which is a measure of the resistance of current carrying parts and
hence that of contacts.
Mechanical Tests
A C.B. must open and close at the correct speed and perform such operations without mechanical
failure. The breaker mechanism is, therefore, subjected to a mechanical endurance type test involving
repeated opening and closing of the breaker. B.S. 116: 1952 requires 500 such operations without
failure and with no adjustment of the mechanism. Some manufacture feel that as many as 20,000
operations may be reached before any useful information regarding the possible causes of failure may
be obtained. A resulting change in the material or dimensions of a particular component may consider-
ably improve the life and efficiency of the mechanism.
Short Circuit Tests
These tests are carried out in short circuit testing stations to prove the ratings of the C.Bs. Before
discussing the tests it is proper to discuss about the short circuit testing stations.
There are two types of testing stations; (i) field type, and (ii) laboratory type.
In case of field type stations the power required for testing is directly taken from a large power
system. The breaker to be tested is connected to the system. Whereas this method of testing is
economi-cal for high voltage C.Bs. it suffers from the following drawbacks:
1. The tests cannot be repeatedly carried out for research and development as it disturbs the
whole network.
2. The power available depends upon the location of the testing stations, loading conditions,
installed capacity, etc.
3. Test conditions like the desired recovery voltage, the RRRV etc. cannot be achieved con-
veniently.
In case of laboratory testing the power required for testing is provided by specially designed
generators. This method has the following advantages:
1. Test conditions such as current, voltage, power factor, restriking voltages can be controlled
accurately.
2. Several indirect testing methods can be used.
3. Tests can be repeated and hence research and development over the design is possible.
The limitations of this method are the cost and the limited power availability for testing the
breakers.
A
l
l
J
N
T
U
W
o
r
l
d

Short Circuit Test Plants
The essential components of a typical test plant are represented in Fig. 5.4. The short-circuit power is
supplied by specially designed short-circuit generators driven by induction motors. The magnitude of
voltage can be varied by adjusting excitation of the generator or the transformer ratio. A plant master-
breaker is available to interrupt the test short circuit current if the test breaker should fail. Initiation of
the short circuit may be by the master breaker, but is always done by a making switch which is
specially designed for closing on very heavy currents but never called upon to break currents. The
generator winding may be arranged for either star or delta connection according to the voltage
required; by further dividing the winding into two sections which may be connected in series or
parallel, a choice of four voltages is available. In addition to this the use of resistors and reactors in
series gives a wide range of current and power factors. The generator, transformer and reactors are
housed together, usu-ally in the building accommodating the test cells.
Fig. Schematic diagram of a typical test plant
Generator
The short circuit generator is different in design from the conventional power station. The capacity of
these generators may be of the order of 2000 MVA and very rigid bracing of the conductors and coil
ends is necessary in view of the high electromagnetic forces possible. The limiting factor for the maxi-
mum output current is the electromagnetic force. Since the operation of the generator is intermittent,
this need not be very efficient. The reduction of ventilation enables the main flux to be increased and
permits the inclusion of extra coil end supports. The machine reactance is reduced to a minimum.
Immediately before the actual closing of the making switch the generator driving motor is
switched out and the short circuit energy is taken from the kinetic energy of the generator set. This is
done to avoid any disturbance to the system during short circuit. However, in this case it is necessary
to com-pensate for the decrement in generator voltage corresponding to the diminishing generator
speed dur-ing the test. This is achieved by adjusting the generator field excitation to increase at a
suitable rate during the short circuit period.
Resistors and Reactors
The resistors are used to control the p.f. of the current and to control the rate of decay of d.c.
component of current. There are a number of coils per phase and by combinations of series and
parallel connec-tions, desired value of resistance and/or reactance can be obtained.
Capacitors
These are used for breaking line charging currents and for controlling the rate of re-striking voltage.
Short Circuit Transformers
The leakage reactance of the transformer is low so as to withstand repeated short circuits. Since they
are in use intermittently, they do not pose any cooling problem. For voltage higher than the generated
A
l
l
J
N
T
U
W
o
r
l
d

voltages, usually banks of single phase transformers are employed. In the short circuit station at
Bhopal there are three single phase units each of 11 kV/76 kV. The normal rating is 30 MVA but their
short circuit capacity is 475 MVA.
Master C.Bs.
These breakers are provided as back up which will operate, should the breaker under test fail to oper-
ate. This breaker is normally air blast type and the capacity is more than the breaker under test. After
every test, it isolates the test breaker from the supply and can handle the full short circuit of the test
circuit.
Make Switch
The make switch is closed after other switches are closed. The closing of the switch is fast, sure and
without chatter. In order to avoid bouncing and hence welding of contacts, a high air pressure is main-
tained in the chamber. The closing speed is high so that the contacts are fully closed before the short
circuit current reaches its peak value.
Test Procedure
Before the test is performed all the components are adjusted to suitable values so as to obtain desired
values of voltage, current, rate of rise of restriking voltage, p.f. etc. The measuring circuits are con-
nected and oscillograph loops are calibrated.
During the test several operations are performed in a sequence in a short time of the order of
0.2 sec. This is done with the help of a drum switch with several pairs of contacts which is rotated
with a motor. This drum when rotated closes and opens several control circuits according to a certain
se-quence. In one of the breaking capacity tests the following sequence was observed:
(i) After running the motor to a speed the supply is switched off.
(ii) Impulse excitation is switched on.
(iii) Master C.B. is closed.
(iv) Oscillograph is switched on.
(v) Make switch is closed.
(vi) C.B. under test is opened.
(vii) Master C.B. is opened.
(viii) Exciter circuit is switched off.
The circuit for direct test is shown in Fig. 5.5
Fig. Circuit for direct testing
A
l
l
J
N
T
U
W
o
r
l
d

Here XG = generator reactance, S1 and S2 are master and make switches respectively. R and X
are the resistance and reactance for limiting the current and control of p.f., T is the transformer, C, R1
and R2 is the circuit for adjusting the restriking voltage.
For testing, breaking capacity of the breaker under test, master and breaker under test are
closed first. Short circuit is applied by closing the making switch. The breaker under test is opened at
the desired moment and breaking current is determined from the oscillograph as explained earlier.
For making capacity test the master and the make switches are closed first and short circuit is
applied by closing the breaker under test. The making current is determined from the oscillograph as
explained earlier.
For short-time current test, the current is passed through the breaker for a short-time say 1
second and the oscillograph is taken as shown in Fig. 5.6.
From the oscillogram the equivalent r.m.s. value of short-time current is obtained as follows:
The time interval 0 to T is divided into 10 equal parts marked as 0, 1, 2 ..., 9, 10. Let the r.m.s.
value of currents at these instants be I0, I1, I2, ..., I9, I10 (asymmetrical values). From these values, the
r.m.s. value of short-time current is calculated using Simpson formula.
Current
0
t
1
3 5
7 9 10
Fig. Determination of short-time current
I =
1
[K2  4 (I
2
 I
2
 I
2
 ...  I
2
)  2 (I
2
 I
2
 ...  I
2
)]
3 0 1 3 5 9 2 4 10
Operating duty tests are performed according to standard specification unless the duty is
marked on the rating plate of the breaker. The tests according to specifications are:
(i) B—3′—B—3′—B at 10% of rated symmetrical breaking capacity;
(ii) B—3′—B—3′—B at 30% of rated symmetrical breaking capacity;
(iii) B—3′—B—3′—B at 60% of rated symmetrical breaking capacity;
(iv) B—3′—MB—3′—MB at not less than 100% of rated symmetrical breaking capacity and
not less than 100% of rated making capacity. Test duty (iv) may be performed as two sepa-
rate duties as follows:
(a) M—3′—M (Make test);
(b) B—3′—B—3′—B (Break test)
(v) B—3′—B—3′—B at not less than 100% rated asymmetrical breaking capacity.
A
l
l
J
N
T
U
W
o
r
l
d

Here B and M represent breaking and making operations respectively. MB denotes the making
operation followed by breaking operation without any international time-lag.3′ denotes the time in
minutes between successive operations of an operating duty.
TEST VOLTAGE
For different transmission voltages, the test voltages required are given in the following Tables:
Table
Test voltages for a.c. equipments
System nominal Power frequency Impulse withstand Switching surge
voltage (rms) withstand voltage voltage withstand voltage
(rms)
400 520 1425 875
525 670 1800 1100
765 960 2300 1350
1100 1416 2800 1800
1500 1920 3500 2200
Table
Test voltages for d.c. equipments
Normal voltage D.C. withstand Impulse withstand Switching surge
voltage kV voltage kV withstand voltage kV
 400 KV 800 1350 1000
 600 KV 1200 1900 1500
 800 KV 1600 2300 2000
Table
Test voltages required for different system voltage (a.c. system)
Nominal voltage KV Power frequency Impulse withstand Switching surge
(rms) voltage kV (rms) voltage kV voltage kV
400 800 2400 1150
765 1000 3000 1750
1100 1400 3700 2300
1500 1900 4600 2800
If the insulation requirements for a particular operating voltage are required to be studied in a
research and development laboratory, the voltage levels required in the laboratory are given in Tables
5.4 and 5.5.
A
l
l
J
N
T
U
W
o
r
l
d

Table
Test voltages required for different d.c. system voltages
Nominal voltage kV D.C. voltage kV Impulse withstand Switching surge
voltage kV voltage kV
 400 800 1750 1300
 600 1200 2500 2000
 800 1600 3000 2600
Table 5.6 shows approximate dimension of the testing equipment and the equipment to be tested.
Table
Approximate dimensions of the testing equipment
and the equipment to be tested
Nominal voltage of A.C. transformer Impulse generator Dimension of Test-
the equipment kV height (m) height (m) object
(rms)
400 10 6 7 × 2 × 11
765 15 8 11 × 2 × 17
1100 18 12 17 × 2 × 24
1500 22 15 28 × 2 × 38
Table shows some of the very important high voltage laboratories in the world.
Table
High voltage laboratories in the world
Location Dimension Power frequency Impulse Test Switching
Test Voltage MV Voltage MV Surge Voltage
MV
Australia 1.5 8.0 –
Bharat Heavy 67 × 35 × 35 1.5 3.6 2.0
Elect. Bhopal, India
CESI-Milan, Italy 150 × 75 × 55 2.3 4.8 3.0
(200 KJ)
College of 25 × 15 × 15 0.3 1.2 –
Engineering,
Guindy, Madras
College of
Engineering, 33 × 26 × 30 0.5 1.4 –
Jabalpur (MP) (16 KJ)
College of 20 × 12 × 8 0.5 1.4 –
Engineering, (16 KJ)
Kakinada, AP
A
l
l
J
N
T
U
W
o
r
l
d

Electricity DC, 65 × 65 × 45 2.5 7.2 6.0
France (1010 KJ)
Hydro-Quebec 82 × 68 × 50 2.5 6.4 6.0
Montreal, Canada (400 KJ)
Indian Instt. 37.5 × 25 × 19 1.05 3.0 1.6
of Science, (50 KJ)
Bangalore
Indian Instt. of 28 × 10 × 9.7 0.80 1.5 –
Technology, (36 KJ)
Madras
Russia 115 × 80 × 60 3.0 8.0 –
Technical 32 × 25 × 21 1.2 3.2 –
University
Darmstadt, W.
Germany
Technical University 34 × 23 × 19 1.2 3.0 –
Munich, W. Germany
A
l
l
J
N
T
U
W
o
r
l
d

LECTURE_NOTES_ON_HIGH_VOLTAGE_ENGINEERIN.pdf

More Related Content

Similar to LECTURE_NOTES_ON_HIGH_VOLTAGE_ENGINEERIN.pdf (20)

Recently uploaded (20)

LECTURE_NOTES_ON_HIGH_VOLTAGE_ENGINEERIN.pdf