CHAPTER 2

GENERATION OF TESTING VOLTAGES

2.1 Introduction

HVAC and HVDC of well exceeding several million volts are required
for testing power apparatus that is rated for operation at EHV or UHV
transmission level. High impulse voltage is required for testing purposes to
simulate not only the amplitude and shapes of voltage at which the equipment
operates, but also to determine the factors of safety for which even much higher
test parameters are necessary. There are always some interactions between the
HV generating circuits used and the test results required. Power system
engineers, therefore require high voltages, generally for the following tasks:

1. Testing the insulation of power apparatus which may require test voltage
of up to 1.2 MVAC (depending on their normal operating voltage).

2. Simulation of overvoltage surges in power systems.

High voltage testing is the final step in ensuring the dielectric quality of
the developed insulation material. Similarly, the complete HV system or device
is tested to ensure its integrity and performance. Chapter 11 deals with various
aspects of high voltage testing. This chapter discusses the basic circuits used for
generating the test voltages. These include HVDC, HVAC, very low frequency
(VLF) AC voltages, and various types of HV impulses. High voltage testing
normally requires only moderate (from few mA up to a few A) current. Thus,
the main emphasis is usually on achieving the required voltage level with
minimum possible cost.

-31-
2.2 High Voltage DC Generation

HVDC can be generated using direct method by changing mechanical

energy directly to HVDC or using indirect method by changing mechanical
energy to HVAC and then converting it to HVDC. The direct generation is
obtained by Van de Graff generators, but this method suffers from low kVA
output, limiting its use to some special applications in physics. In the indirect
method, HVAC is rectified using HV diodes and capacitors. The simplest circuit
for HVAC rectification is the half wave rectifier circuit shown in Fig. (2.1)
where in this case the positive half cycle is rectified and the load is resistive.

Fig. (2.1). Halfwave rectifier: (a) circuit; (b) output voltage for resistive load.

The voltage doubler circuit shown in Fig. (2.2a) is widely used to

generate DC voltage of ±2 Vp from an AC voltage of Vpsint shown in Fig.
(2.2b). This circuit is a combination of two half-wave rectifier circuits. In the
first part ABCD, the –ve half cycle of the AC is rectified. Therefore, the voltage
across C1 (VAB) is constant at (-Vp) while the voltage across diode D1 (VBC) is
alternating between 0 and 2Vp (VBC = VAD – VAB). This alternating voltage
(VBC) is then rectified through diode D2 producing an output of 2Vp. The voltage
across capacitor C1 is shown in Fig. (2.2c) while the voltage across D1 is shown
in Fig. (2.2d). The output voltage waveshapes at no load and with load are
shown in Fig. (2.2e) and Fig. (2.2f), respectively.

-32-
 a

Fig. (2.2). Voltage doubler; (a) circuit, (b) source voltage, (c) voltage across C1,
(d) voltage across D1, (e) output voltage at no load, (f) output voltage
at load.

-33-
 Ripple voltage (V) and voltage drop (V) are the two important
parameters to be considered when dealing with rectification of HVAC to HVDC.
In reference to Fig. (2.2f), V is defined as half of the difference between the
maximum and the minimum values of output voltage. Thus,

V = 0.5( V o, max - V o, min ) (2.1)

In most applications, the percentage ripple is allowed between 3 % to

5 %. The voltage drop V and ripple factor RF of such a circuit are defined as:

V = Vn.l - Vf .l (2.2)

V  V
RF = (2.3)
(VfL )

where Vn.l and Vf.l are average values of no load and full load DC output
voltages, respectively.

The maximum voltage across the diodes as well as the capacitor is 2Vp.
Since the cost of these components increases at much higher rate than the
increase in their voltage rating, it is not economical to increase their voltage
rating for generation of high voltage. Instead, two or more doublers circuits are
connected in cascade to form a voltage multiplier circuit. Fig. (2.3) shows a
multiplier circuit consisting of three stages (n=3) with maximum output voltage
of 6Vp. For voltage multiplier circuit, V and V are given as [1]:

n n  1
I
V  (2.4)
4fC

I 2 3 n2 n 
V =  n    (2.5)
fC  3 2 6 

where I is the DC load current, f is the AC supply frequency, C is the stage

capacitance, and n is the number of stages. It can be seen from the above

-34-
equations that V is proportional to n2 whereas V is proportional to n3. Thus,
the number of stages in such a circuit is the limiting factor as it influences
adversely the values of V as well as V.

Fig. (2.3). HVDC voltage multiplier circuit with n=3.

The above limitations can be overcome by energizing each stage by a

transformer comprising of a tertiary low voltage winding which excites the
primary winding of the next stage as shown in Fig. (2.4). In this way, the
problems of V and V are solved, however, there is extra cost involved for the
additional transformer windings. The limiting factor in such circuits employing
cascaded transformers is the loading of the primary winding of the first stage of
the transformer which has to supply the current for all the n stages. In addition
to the above circuits, there are other HVDC generating arrangements like
Deltatron circuit and voltage multiplier circuit with individual transformers for
each stage [1].

2.3 High Voltage AC Generation

HVAC testing sources ranging from about 10 kV rms up to more than 1.5
MV rms are available due to the continuous development of AC transmission
voltages up to about 1200 kV. For routine testing, the voltage level for power-

-35-
frequency testing is always related to the highest rms phase-to-phase voltage
Vm of the power transmission system. This rated power-frequency short
duration withstand voltages, Vt is different for different apparatus within a
transmission system. For Vm < 300 kV, the ratio of Vt/Vm is up to about 1.9
while it decrease with higher values of Vm.

Fig. (2.4). A two-stage HVDC multiplier circuit with cascaded transformers.

Although power transmission systems are generally three-phase, the

testing sources usually produce single-phase to ground voltage. The control of
the HVAC output voltage is usually performed by the control of the LV input
to the primary winding of the voltage step-up systems. The waveshape must
be substantially sinusoidal with both half-cycles closely alike. If the ratio of

-36-
peak-to-rms value is within (2 5%), the rms value of total harmonics does
not exceed 5 % of the rms value of the fundamental.

Testing of HV apparatus or HV insulation always involves an

application of high voltage to capacitive load with low power dissipation only.
If Ct is the capacitance of the equipment or sample under test, and Vn is the
nominal rms voltage of the HV testing supply, the nominal kVA rating Pn of
the supply may be calculated from the formula:

Pn = Ks V2n Ct (2.6)

in which the safety factor, Ks  1 accounts for additional capacitance within

the test circuit such as those for HV electrodes and connections between test
object and the voltage source, which might have large diameters and
dimensions. This safety factor, Ks might range from only about 2 for very
high voltages of  1 MV, and may increase to higher values for lower nominal
voltages, as over-dimensioning is economically possible in such cases. Value
of Ct may change considerably depending upon the type of equipment. It can
be about 10 pF for simple post or suspension insulator and 1 nF to 10 nF for
metal clad SF6 insulated substation.

HVAC test supplies are normally designed to withstand sudden voltage

breakdown across the output terminals. The stress to the windings and coils
accompanied by such an occurrence of breakdown is usually related to the
short-circuit current and accompanying magnetic forces within the windings.
More frequently, it is the stray potential distribution between the windings
which causes insulation failures. One may also provide proper damping
resistors between HV test source and the test equipment to reduce the rate of
the sudden voltage drop and to avoid any overvoltage within the test circuit
that may be caused by the breakdown. Nominal values of such damping

-37-
resistor are between 10 k and 100 k and these will usually not influence
the test conditions.

There are three main methods for HVAC generation which are briefly
described next.

2.3.1 Single or ordinary testing transformer

It is similar to a single phase step-up transformer with emphasis on the

insulation of the HV winding. The windings are generally designed for low
current rating and partial discharge free operation up to the rated voltage. The
short-circuit impedance is kept less than 5% and the magnetizing current is also
kept low to minimize the harmonics in the output voltage. Furthermore, care is
also taken to ensure that the field distribution over the HV winding is uniform.
Such test transformers are economical for test voltage of  300 kV. Above this
level, a single test transformer becomes quite expensive due to the added
insulation requirements and due to the burden in transportation and erection of
such large units. Thus, for higher voltages (> 300 kV), cascaded transformers
are generally the main choice.

2.3.2 Cascaded transformers

The idea of cascading transformers is based on connecting the secondary

windings of various transformers in series while their primary windings are
energized in a normal manner. Fig. (2.5) shows a three-stage cascaded
transformer. The second and the third-stage transformers are isolated from the
ground by insulators capable of withstanding a voltage of V and 2V, respectively
and are excited by tertiary windings of the first and the second stage. Here, the
total output voltage is 3V and is taken across the high voltage terminal of the
third transformer and the ground. The two main limitations of increasing the
number of stages in this scheme are the high total short circuit impedance and
the high loading of the primary of the first stage. As can be seen from Fig. (2.5),

-38-
the currents in the first, the second and the third transformer's primary windings
are 3Ip, 2Ip and Ip, respectively. The net internal impedance of this circuit (Xeq)
can be derived as follows. In this derivation, it is assumed that the leakage
reactances of the primary, the secondary and the exciting (tertiary) windings of
unit i are xpi, xhi, and xei, respectively.

Fig. (2.5). Cascaded transformer for HVAC generation.

Assuming that transformer turns ratios are such that the per unit value of
Ip = Is = Ie (i.e. Io = Ip = I). The total reactive power is given as:

Q = I2 (Xh1 + Xh 2 + Xh 3) + I2 (Xe3 + X p3) + (2I )2 (Xe2 + Xp 2) + (3I )2 X p1 (2.7)

If Xh1 = Xh2 = Xh3 = Xh, and Xe2 = Xe3 = Xe, and Xp1 = Xp2 = Xp3 = Xp, then;

Q = 3XhI2 + 5XeI2 + 14XpI2 (2.8)

Thus the cascaded transformer can be modelled by one impedance Xeq

and one current I such that,

Q = I2 Xeq = I2 [3 X h + 5 Xe + 14 X p] (2.9)

-39-
Thus,

Xeq = 3 Xh + 5 Xe + 14 Xp (2.10)

In general, for an n-stage cascaded transformer, the total impedance is

given as,

n
Xeq =  [Xhv + (n - 1) Xev + (n + 1 - v ) Xpv]
2 2
(2.11)
v=1

This relation clearly shows that as n is increased, the total impedance of

cascaded units greatly increase.

2.3.3 Resonance test systems

One of the problems associated with the HV test transformers is the

resonance of its inductance with the load capacitance. Fig. (2.6a) shows a
simplified transformer equivalent circuit with single inductance LT (it is assumed
that transformer resistance, RT = 0). If the load is purely capacitive (CL), such as
a long cable, then the load voltage is given as;

Fig. (2.6). Series LC circuit; (a) simplified HV transformer equivalent circuit

(R=0) with capacitive load, (b) resonance circuit for HVAC
generation.

-40-
 Vs
VL = (2.12)
1 -  LT CL
2

where  is the supply frequency in radians/seconds. Since the denominator is

less than 1, VL > Vs and could be several times Vs depending upon the values of
, LT and CL. This could lead to the destruction or breakdown of the load under
test. If this phenomenon is used in a controlled manner, output voltages up to 50
times Vs could be obtained from a supply of voltage Vs. This is achieved by
using an adjustable inductance L (where L >> LT) and a variable capacitance C
(C >> CL), as shown in Fig. (2.6b). This type of circuit is called a resonance
circuit and is used to generate HVAC especially when testing capacitive loads.
The main advantages of such a circuit are;

1. Less power requirement from the source (< 10% of kVA required for
testing).

2. Pure sinusoidal output waveforms due to resonance.

3. No arcing in case of a short circuit since the condition of resonance

ceases to exist with low shunt resistance of the arc.

4. Simple and compact system.

5. More than one resonance circuit can be cascaded to obtain even higher
voltages.

6. Reduction of the size and weight of AC testing facilities.

For field testing of cables, large rotating machines or gas insulated

substations, a reduction of weight and size of the testing equipment is very
desirable. This goal is achieved by the development of a different type of a
series resonant circuit, in which a choke with a constant inductance is used [1].
As the load capacitance cannot be kept within narrow limits, the supply

-41-
frequency must be continuously variable to achieve conditions of resonance.
This set-up is shown in Fig. (2.7) where an exciter supply, connected to the
LV mains, excites the series resonance circuit with a variable frequency. The
reactor can be designed for high-quality factors, which increase with
frequency up to some 100 Hz. The inductance can be single unit or any
number of units in series and/or in parallel. Here, Ct represents the test object
and other shunt capacitances. For specified testing frequency, f the resonance
is achieved when:

1
f (2.13)
2 L n C t

Fig. (2.7). Schematic diagram of series resonance test circuit with variable test
frequency [1].

The nominal inductance Ln will predominantly be designed according

to a nominal capacitance Cn = Ct which is the highest capacitance that can be
tested with the full rated voltage V = Vn of the circuit, and a nominal
frequency fn, which is the lowest frequency within this rated voltage. With the
above equation, we thus obtain:

1
Ln  (2.14)
22 f n2 C n

-42-
 The choke's maximum or nominal current I , which overheats the coil
or saturates the iron core should be checked. As the losses are very small, R
<< Ln within the whole frequency range and In may directly be derived from
the voltage drop across Ln, which is nearly the full rated voltage Vn.

Vn Cn
In   Vn (2.15)
2 f n L n Ln

Equations (2.13) to (2.15) are used to determine the normalized

operating characteristics of the circuit. For test objects with capacitance value
Ct different from Cn, the resulting testing frequency, f will also be different
from fn. The variation of the frequency is then given according to eqn (2.16):

f Cn 1
  (2.16)
fn Ct Cn / Ct

For Ct  Cn, the reactor Ln can be used up to the full rated voltage, Vn.
Although the frequency increases according to eqn (2.16), the load current will
always be lower than In. The relationship of the normalized current for Ct  Cn
is as follows:

I fn Ct
  (2.17)
In f Cn

For Ct > Cn this circuit may still be applied, if the testing voltage, V =
Vt, is reduced to keep the current at its nominal value, In. As the current, I is
always proportional to the testing voltage, then

I V Ct
 t  (2.18)
I n Vn Cn

-43-
and one can apply this equation to show the necessary reduction of the testing
voltage for Ct > Cn, if I is to be limited to In.

Vt 1
 (2.19)
Vn Cn / Ct

The normalized operating conditions given by eqns. (2.17) to (2.19) are

illustrated in Fig. (2.8). For quite small test specimens the test frequency, f
may conveniently be limited by the addition of a permanent HV capacitor such
as a capacitor voltage divider. The relatively modest variation of this
frequency for large capacitors under test will improve the flexibility of
applications, and the system can be used for the AC testing of power cables.
The actual limitations in testing of very large test specimens with lower
voltages than Vn are given by the reduction of Q for too low frequencies, and
the frequency for which the exciter transformer saturates.

Fig. (2.8). Operating characteristics of circuit according to Fig. (2.7) [1].

-44-
2.3.4 Generation of high frequency high AC voltage

High frequency, high AC voltage is required for DC power supplies. For

testing electrical apparatus, high frequency, high voltage damped oscillations are
needed which require high voltage, high frequency transformers. The advantages
of these high frequency transformers are [2]:

1. The absence of iron core in transformers results in saving in cost and size,

2. Pure sine wave output is possible due to absence of magnetic saturation,

3. Voltage builds up slowly over a few cycles and hence no damage is caused
due to switching surges, and

4. Uniform distribution of voltage across the winding coils due to subdivision of

coil stack into a number of units.

The commonly used high frequency resonant transformer is the Tesla coil,
which is a doubly tuned resonant circuit shown schematically in Fig. (2.9). The
primary voltage rating is 10 kV and the secondary may be rated to as high as
500 to 1000 kV. The primary is fed from a DC supply through the capacitor C1.
A spark gap, G connected across the primary, is triggered at a desired voltage V1
which induces a high self-excitation in the secondary. The primary and the
secondary windings (L1 and L2) are wound on an insulated former with no core
(sir-cored) and are immersed in oil. The windings are tuned to a frequency of 10
to 100 kHz by means of capacitors, C1 and C2.

The output voltage V2 is a function of the parameters L1, L2, C1, C2, and
the mutual inductance M. Usually, the winding resistance will be small which
causes damping of the oscillations. Fig. (2.10) shows waveform of typical high
voltages with high frequency damped oscillations which are used for GIS testing
[3].

-45-
Fig. (2.9). Resonant transformer for the generation of high voltage with high
frequency damped oscillation.

The analysis of the output waveform can be performed in a simple manner

by neglecting the winding resistance. Let the capacitor, C1 be charged to a
voltage, V1 just before the spark gap is triggered. After triggering the gap, a
current, i1 flows through the primary winding Ll which produces a current i2
through L2 and C2.

Then;

1 t di di
V1  
C1 0
i1 dt  L1 1  M 2
dt dt
(2.20.a)

and,

1 t di 2 di
0 
C2 0
i 2 dt  L 2
dt
M 1
dt
(2.20.b)

Converting the above two equations into s-domain by Laplace transform, we get:

V1  1 
 L1s   I1  MsI 2 (2.21.a)
s  C1s 

and

-46-
  1 
0  (Ms)I1  L 2 s   I2 (2.21.b)
 C 2s 

Fig. (2.10). High voltage waveform with high frequency damped oscillations
used for GIS testing [3].

Where, I1 and I2 are the Lapace transformed values, of i1 and i2. The output
voltage V2 across the capacitor, C2 is

t
1
C 2 0
V2 = i 2 dt (2.22a)

or its transformed equation is:

-47-
 I2
V2 s   (2.22b)
C 2s

where V2(s) is the Laplace transform of V2.

The solution for V2 from the above equation is obtained as:

V2 
MV1 1
cos  1 t  cos  2 t  (2.23a)
L1 L 2 C1  2   12
2

where,

M2
2  1   1  K c2 (2.23b)
L1L 2

Kc = coefficient of coupling between the windings having inductances, L1 and L2,

and

2
1,2
2  22
 1
2
 2  22 
  1
 2
  12 22 1  K c2

  (2.23c)
 

1 1
1  and 2  (2.23d)
L1C1 L 2C 2

The peak amplitude of the secondary voltage V2 can be expressed as,

L2
V2 max  V1e (2.24a)
L1

where,

2 1  
e  , and (2.24b)
1  a  2
 4a

L 2 C 2 12
a  (2.24c)
L1C1 22

-48-
 A more simplified analysis for the Telsa coil may be presented by
considering that the energy stored in the primary circuit i.e. in the capacitance, C1
is transferred to C2 via the magnetic coupling. If W1 is the energy stored in C1
and W2 is the energy transferred to C2 and if the efficiency of the transformer is
, then W2 =  W1 or

1 1
 C1V12  C 2 V22 (2.25)
2 2

which gives

C1
V2  V1  (2.26)
C2

It can be shown that if the coefficient of coupling, Kc is large, the

oscillation frequency is less, and for large values of the winding resistance and
Kc, the waveform may become a unidirectional impulse. This is shown in section
2.5.3 while dealing with the generation of switching impulses.

2.3.5 Generation of very low frequency voltages

Power cables are carefully tested by manufacturers with AC and/or DC

voltages to ensure conformance with applicable specifications and industry
standards. During transport, installation, and backfilling, cables are vulnerable to
external damage. Therefore, cables are usually also tested after installation and
prior to placing them in service to confirm that installation and jointing and
terminating work has been carried out in a satisfactory manner. Moreover,
periodic testing of service-aged cables may also be performed to determine
possible system degradation and to reduce or eliminate service failures. Very low
frequency (VLF) testing describes a technique for field testing of service-aged
cable systems with extruded insulation. The VLF test normally uses a 0.1 Hz
pulse wave which changes polarity in a sinusoidal manner. The sinusoidal
transition in the power frequency range may initiate a partial discharge at a defect

-49-
while the 0.1 Hz pulse wave may develop into a breakthrough channel. Due to
sinusoidal transitions between HV pulses, traveling waves are not generated. Due
to continuous polarity change, dangerous space charge build up cannot occur.
Cables can be tested with an AC voltage up to three times conductor-to-ground
voltage with a device comparable in size, weight, and power requirements to a
DC test set.

VLF can be generated in various waveforms such as cosine-pulse, sine or

square with high slew rate. Convenient lightweight low-frequency (0.1 Hz) high-
voltage sources, having low kVA rating in comparison to equivalent 60-Hz HV
sources, have become commercially available, thus permitting partial-discharge
site location as well as dissipation factor (tan ) measurements to be made on
polymeric distribution cables in the field. Since space charge effects predominate
at low frequencies, particularly in aged cables, great care must be exercised when
interpreting the tan  measurements. In the presence of space charge effects, the
tan  value obtained at 0.1 Hz will generally exceed to that of the corresponding
60 Hz value by an appreciable amount [4].

A schematic circuit diagram of a typical VLF (0.1 Hz) test set is shown in
Fig. (2.11). The high voltage DC is converted to a high voltage having a
frequency of 0.1 Hz. A rotating rectifier in conjunction with a high voltage choke
converts the DC signal into an alternating AC signal. Fig. (2.12) shows typical
voltage wave shape produced by the VLF generator. The maximum 0.1 Hz
alternating voltage output of the very low frequency (VLF) test set is typically 36
kV rms into a maximum capacitive load of 3 F. Low frequency voltage
withstand test is also employed to assess aging of polymeric cables while in
service. They are frequently employed in lieu of the high voltage DC test, which
is believed to cause damage to aged polymeric cable insulating system, under
some conditions that may still have some remaining useful life under normal
operating AC voltage conditions [4].

-50-
Fig. (2.11). Schematic of a very low frequency (VLF) high voltage test set with
cable specimen under test [4].

time of polarity change tu 2.6 ms

with 0.5F supported condenser
and up to 4.5 F cable capacitance
Fig. (2.12). Typical voltage wave shape produced by VLF generator.

2.4 High Voltage Impulse Generation

It is well known that power system components are subjected to severe

overvoltage due to internal switching or external lightning surges.
Consequently, the integrity of the individual components, devices and
subsystems must be checked through high voltage surge testing. High voltage
surges that are simulated in laboratories are commonly known as high voltage
impulses. An impulse is defined as a unidirectional voltage (or current) rising
quickly to its peak value and then decaying slowly to zero. Thus, an impulse has
two parts i.e. a rising part which is usually realized by charging a capacitor and a
decaying part which is realized by discharging a capacitor.

-51-
2.4.1 Single stage impulse generator

Fig. (2.13a) shows a capacitor, C2 which is charged through a resistor, R1

from a much larger capacitor, C1. In Fig. (2.13b) the two capacitors are
discharged through a resistor, R2. Combining the two circuits and replacing the
mechanical switch by a HV switch, e.g. a sphere spark gap, the simple impulse
generating circuit of Fig. (2.13c) is obtained. This circuit is called high voltage
efficiency circuit since the output is taken across the parallel combination of C2
and R2. If the location of R2 is changed as shown in Fig. (2.13d), the resulting
circuit has low voltage efficiency since the output will be divided between R1
and R2 before it is taken across C2. The impulse voltage waveform can be
modelled mathematically by a double exponential wave defined by the equation:

V(t) = A [exp(-t) - exp(-t)] (2.27)

where  and  are time constants of the circuit and control the wave decay and
rise rates, respectively.

Generally  >> , where  corresponds to the charging or the rising

portion of the voltage while  corresponds to the discharging or decaying
portion of the waveform. Fig. (2.14) shows the typical impulse waveshape,
which is characterized by its front time tf and the tail time, tt. Tail time or time to
half of the voltage peak value (tt) is the time between t0=0 and the time where the
impulse decays to half of its peak value. Front time, tf is defined as peak time
(tp) for switching or [(t.9-t.3)/0.6] for lightning impulses, respectively. The front
time for a standard lightning impulse is 1.2 s while its tail time is 50 s
(1.2/50). The front time and tail time for switching impulse are not fixed since
switching surges depend on many factors such as network impedances, circuit
breaker characteristics, etc. However, switching surges are slower than lightning.
The front time is in the several hundred micro second range while its tail time is
in thousands of micro second range. There are several standard switching
impulses such as 250/2500, 140/1400, 1760/1700 and 100/3000 s. For both

-52-
impulses, the tolerance allowed in peak value is 3%. For front time, the
tolerances allowed are 30% and 20% for lightning and switching impulses
respectively. On the other hand, the allowed tolerances for tail time are 20%
for lightning and 60% for switching impulses [5].

Fig. (2.13). Principle of HV impulse generation; (a) charging a capacitor,

(b) discharging capacitors, (c) basic low voltage efficiency impulse
generating circuit and (d) high efficiency circuit.

-53-
 Equation (2.27) can be derived for the circuit of Fig. (2.13c) using either
time or frequency domain circuit analysis and can be written as:

V
V(t) = [exp(-t) - exp(-t)] (2.28)
k( - )

where

V = initial voltage across C1,

a a
,  =  ( )2 - b (2.29)
2 2

where for circuits of Fig. (2.13c) and (2.13d), the constants b and k are the same
for both circuits and are given as:

1
b= , k = R1C 2 (2.30.a)
R 1 R 2 C1 C 2

However, for high voltage efficiency circuit of Fig. 2.13d, a, is given as:

 1 1 1 
a =  + +  (2.30.b)
 1 1
R C R C
1 2 R 2 1
C

and for low voltage efficiency circuit of Fig. 2.13c, a, is given as:

 1 1 1 
a =  + +  (2.30.c)
 R 1 C1 R1 C 2 R 2 C 2 

The impulse generator circuit can also be analyzed using computer

software such as P-SPICE, where the adjustment of parameters becomes simpler.
If the circuit components are known, tf and tt can be found by sketching V(t)
versus t or by solving equation (2.28) using the definitions of tf and tt. If R2 >>
R1, and C1 >> C2, then tf and tt are approximately given by the following
formulae [2].

-54-
Fig. (2.14). Impulse waveshape parameters, (tf)L = front time for lightning
impulses, (tf)s = front time for switching impulses and tf = time to
half value for both cases and V = initial voltage across C1.

C1 C2
t f = 3 R1 (2.31)
C1 + C2

t t = 0.7(R1 + R 2) (C1 + C2) (2.32)

These simplified expressions are based on the assumption that a capacitor

can be fully charged in about 3C, where C is the charging time constant, i.e. c
= R1C1C2/(C1+C2). Furthermore, it is assumed that the capacitor is discharged to
its half voltage when the elapsed time is .7d, where d is the discharge time
constant and is given as d = (R1+R2) (C1+C2). Equations (2.31) and (2.32) are
approximate and are used only for rough estimations of tf and tt.

When tf and tt are known and it is required to design the generating

circuit elements, we need to find first , ß, and then R1,R2,C1, and C2. Both the
exact as well as approximate methods are available for this purpose. For either
case, it is a normal practice to specify the values of the two circuit elements and
then calculate the values of the remaining two elements. Usually C1 and C2 are

-55-
specified such that C2 << C1. C2 can also be used as the voltage divider for
measuring and recording the output voltage. The values of R1 and R2 can be
found accurately by substituting the values of tf and tt in equation (2.28) and
finding  and ß using numerical techniques. Then the values of R1 and R2 can
be evaluated by using equations (2.29) and (2.30) and the assumed values of C1
and C2. The values of  and ß and consequently R1 and R2 can also be estimate
by using the following approximate formulae [2].

1 1
= , = (2.33)
R 2 C1 C 2 R1

Furthermore, Table (2.1) can be used to find approximate waveshape or

circuit parameters. When using Table (2.1) or equation (2.34) to find the circuit
parameters, the calculated values can be verified by evaluating tf and tt and
comparing them with specified values.

The single stage impulse generator circuit shown in Fig. (2.13) is one of
the several circuits in which either the position of R2 is changed, or it includes
some additional resistors or inductors [2]. Figure (2.15) shows examples of
these circuits.

Table (2.1) Impulse parameters for various standard waveshapes.

tf/tt
(s/s)
1
s 1 1
s 1 
Vp
  Vo
1.2/5 3.48 0.8 0.77
1.2/50 68 0.4 0.96
1.2/200 284 0.38 0.99
170/1700 2200 43.5 0.91
250/2500 3160 62.5 0.9
650/2600 2500 250 0.75
400/3000 3590 112 0.87
500/3000 3380 155 0.82

-56-
2.4.2 Multistage impulse generator

When very high impulse voltages are needed, multistage impulse

generators are used since the cost of a single stage circuit's elements become
prohibitive as the voltage rating becomes high. The basic idea of a multistage
impulse generator circuit is to charge several stage capacitors in parallel and then
discharge them in series. Fig.(2.16a) shows such a circuit, whereas Fig.(2.16b)

Fig. (2.15). Circuits for producing impulse voltages.

-57-
Fig. (2.16). Multistage impulse generation; (a) main circuit, (b) stage capacitor
charging period, (c) capacitor discharging period, (d) equivalent
single stage circuit.

-58-
shows the same circuit during the stage capacitor charging period, while
Fig.(2.16c) shows the same circuit during the discharge period. Fig.(2.16d)
shows the equivalent single-stage circuit which can be used to analyze the
multistage circuit. In this circuit, only C1 is distributed throughout all the stages.
Generally in multistage impulse circuits, R1 and R2 can also be distributed
throughout the different stages. However, C2 is normally left as a single unit
which also works as a voltage divider or represents the load capacitance. Figure
(2.17) shows a multistage generator circuit where C1, R1 and R2 are distributed
throughout the stages.

The peak output impulse value Vp can not exceed nV since the initially
available charge Vo(C1)eq will be distributed between (C1)eq and C2. The voltage
efficiency of such generators is therefore, given by:
Vp
 (2.34)
nV

where n is the number of stages.

A multistage impulse generator requires several component parts for

flexibility and for the production of the required waveshape. These may be
grouped as follows:

(a) DC charging set

The charging unit should produce a variable DC voltage of either polarity

to charge the generator capacitors to the required value. If maximum charging
voltage is V, then the impulse output from n stage generator will be  nV.

(b) Charging resistors

These will be non-inductive high value resistors of about 10 to 100 k.

Each resistor will be designed to have a maximum voltage rating to withstand
DC charging voltage level.

-59-
Fig. (2.17). Multistage impulse generator circuit incorporating the wave-shaping
resistors within the generator stages.

-60-
(c) Generator capacitors and spark gaps

These capacitors should be designed to operate under impulse voltage.

These are arranged vertically one over the other and with all the spark gaps
aligned. The capacitors are designed for numerous charging and discharging
operations. On dead short circuit, the capacitors will be capable of giving 10 kA
of current. The spark gaps will be usually spheres or hemispheres of 10 to 25 cm
diameter. Sometime, spherical ended cylinders with a central support may also
be used. The impulse energy is given by:

1
W C1 Vp2 (kJ/stage) (2.35)
2

This is an important characteristics parameter of impulse generator.

(d) Wave-shaping resistors and capacitors

Resistors should be non-inductive wound type and should be capable of

discharging impulse currents of 1000 A or more. Each resistor is to be designed
for a maximum voltage of around 100 kV. In some cases, they are wound on
thin cylindrical formers and are completely enclosed. These capacitors may be
of compressed gas or oil filled type with a capacitance of 1 to 10 F.

Modern impulse generators have their wave-shaping resistors included

internally with a flexibility to add additional resistors outside, when the
generator capacitance is changed (with series parallel connection to get the
desired energy rating at a given test voltage). Such generators optimize the set of
resistors. A commercial impulse voltage generator uses six sets of resistors
ranging from 1.0  to about 160  with different combinations (with a
maximum of two resistors at a time) such that a resistance value varying from
0.7  to 235  per stage is obtained, covering a very large range of energy and
test voltages. The resistors used are usually resin cast with voltage and energy
ratings of 200 to 250 kV and 2.0 to 5.0 kJ, respectively. The entire range of

-61-
lightning and switching impulse voltages can be covered using these resistors
either in series or in parallel combinations [6].

The shunting of load must be taken into account. Capacitive load, such as
cables will effectively increase C2 and hence increase tf. On the other hand,
inductive load such as reactors and transformers may distort the wave shape.
The influence of the load inductance on the impulse output voltage is illustrated
in Fig. (2.18).

Vo V(t)

V(t)

Vmax

0.5 Vmax

Fig. (2.18). Effect of varying the inductance across the output terminals on the
generated impulse voltage wave shape V(t).

-62-
(e) Voltage dividers

Voltage dividers are of either damped capacitive or resistive type. An

oscilloscope or Digital Impulse Measuring System (DIMS) with recording
arrangements are provided to measure the voltage across the test object.
Sometimes a sphere gap is also provided for calibration purposes (see Chapter 3
for details).

(f) Triggering system

The generator spheres can be triggered (i.e. to start to conduct) through

various means. Trigatron is the most commonly used method of triggering. The
trigatron gap comprises of a high-voltage spherical electrode and an earthed
main hemispherical electrode with the main gap between. A trigger electrode
with the shape of a metal rod is located inside and insulated from the main
electrode by an annular clearance. The trigatron is provided with a pulse circuit.
Fig. (2.19) shows a simplified triggering circuit which contains the basic
elements. Closing the switch, S in Fig. (2.19) produces a voltage pulse between
the trigger and the main electrode, sufficient for the electric break down of the
annular clearance. The ionization of this discharge and the disturbance of the
field in the main gap causes its complete breakdown. This breakdown initiates
the breakdown of gaps G of the other stages of the multistage impulse generator.
Thus, the instant of generating the impulse can be controlled. The switch, S
could be replaced by a synchronizing circuit so as to generate the impulse at the
required instant with respect to the power-frequency voltage cycle, thus enabling
AC and impulse high voltage to be superimposed on a test specimen as required.
A trigatron gap connected across the load could be similarly triggered at the
appropriate subsequent moment to produce chopped impulse waves [7].

-63-
 Fig. (2.19). Impulse generator triggering circuit using a Trigatron spark gap.

Fig. (2.20). Block diagram of triggering unit with its interface connections with
impulse generator system.

Recently the triggering has become faster and easier to control using
thyristors and transistors for switching where the delay times are only in the
range of nanoseconds. Also these new systems can be easily linked to digital

-64-
measuring system for better coordination with other measurements and operation
of the impulse generator [6]. Fig. (2.20) shows the interfacing of triggering unit
to the impulse generator system.

2.4.3 Generation of special switching impulses

The switching waveforms experienced in power systems are not unique.

They have front rise times of up to several milliseconds and considerably longer
decay times. They may contain oscillatory components with a frequency
ranging from a few hertz to a few kilohertz. Switching surges are accompanied
by energies much larger than lightning surges [8].

Single and multistage impulse generation circuits discussed earlier are

used to generate lightning and switching impulses with the proper choices of
elements. For some applications, special types of switching impulses such as;
impulses with very long decay times in ms range or impulses with some
controlled oscillations on their tail, are required. Table (2.2) shows a few circuits
along with their output voltage waveforms and mathematical representations of
such waveforms [9]. These circuits are briefly explained below.

Circuit # 1

Impulse generator circuits discussed earlier are modified to give longer

duration wave shapes by a proper choice of the front and tail resistors. In order
to produce unidirectional damped oscillations, an inductance L is connected in
series with the conventional impulse generator circuit. These oscillations may
have a frequency of 1 to 10 kHz, depending on the circuit parameters.

Circuit # 2

A testing transformer is excited by a voltage impulse to give long-

duration or oscillatory wave in its secondary. The capacitor C1 is charged to a
moderate DC voltage (20 to 25 kV) and then discharged into the low-voltage
winding of a power transformer. The high-voltage winding is connected in

-65-
parallel to a load capacitance, C2 of the test object. Through autotransformer
action, switching impulses of proper waveform can be generated across the test
object. The disadvantage of this technique is the considerable amount of high-
frequency distortion caused in the output waveform. Furthermore, the size of the
capacitor, C1 for producing a reasonable output voltage may be large. Variations
of the front of the generated impulse are achieved by adding inductances of
different magnitudes in series with C1.

A preferable method is to energize the transformer from an AC supply for

less than half a cycle. The energization of the transformer is initiated via a
thyristor or trigratron to be triggered at the required phase angle. The
transformer remains energized until the supply current passes through a zero
value. Then the energy stored in the transformer is discharged through the test
object. The advantages inherent in this method are mainly its small high-
frequency distortion in the output voltage and the elimination of the energy
storage capacitor C1.

Circuit # 3

In this circuit, the wave front and the wave tail times are controlled by
changing the values of R and L simultaneously. With a given generator
capacitance, C and upon choosing a suitable value for L, the wave front time or
the value of  is determined. The wave tail time or the value of  is controlled
by the value of R in the circuit. The advantage of this circuit is its simplicity.
But the waveshape control is not flexible and independent. Another
disadvantage is that the basic circuit is altered when a test object which will be
mainly capacitive or inductive, is connected across the output. Hence, the
waveshape gets changed with the change of the test object.

In table (2.2) the output voltages are expressed in s-domain. To get the
voltage in time domain, numerical solutions should be used for circuits (1) and
(2). For circuit (3) an analytical solution can be obtained.

-66-
Table (2.2) Special impulse generating circuits and their output voltage
waveforms [9].

2.4.4 Generation of chopped impulse

When a full wave surge occurs on a power network and a flashover takes
place across a bushing or an insulator etc., the voltage instantaneously falls to
zero resulting in a chopped impulse wave. The voltage chopping can take place
either on the front, at the peak or on the tail of a surge. To simulate such a
chopped surge wave, a rod-rod chopping gap is normally placed in parallel with
the test object across the impulse generator. The distance of the chopping gaps
can be adjusted to control the width of the applied chopped wave during the

-67-
chopped impulse testing. Triggered chopping gaps are often used to control the
chopping time. Chopped impulse testing is required in some applications as
discussed in Chapter 11.

2.5 Generation of Impulse Currents

Lightning strokes are associated with high-current impulses which in

turn produce high voltage impulses on transmission systems. Therefore,
generation of impulse current of the order of several hundreds of kilo-amperes
is needed for testing lightning arresters. They are also used in some other
applications [10]. The waveshapes that are in common use are the double
exponential and the rectangular waves shown in Fig. (2.21).

Fig. (2.21). Impulse current waveforms (a) exponential, (b) rectangular.

2.5.1 Double-exponential current impulses

The impulse waveshape is defined according to the IEC

recommendations. T1 is the front duration and T2 is the time to half the peak.
As per IEC, standard impulse current waves are defined as T1/T2 given by,
4/10 s and 8/20 s [11].

The circuit of Fig. (2.22) can be used to produce one of the standard
current waveforms. The basic difference between this circuit and those
generating impulse voltages is the location of sphere gaps and the capacitors.

-68-
In addition, the capacitances are much larger and the resistors are much
smaller in magnitude in the current generating circuit.

Fig. (2.22). Circuit for producing double exponential current impulses.

If the capacitor, C is charged to a voltage V and discharged when the

gap G is triggered, the current i can be shown to vary with time according to
the relation.

V
i(t) = exp (-t) sin t (2.36)
L

where  = R/2L (2.37)

and   1 / LC  R 2 / 4 L2 . (2.38)

The condition for under-damped oscillations in the current wave is

R  2 L / C . The time taken for the current i to rise from zero to the first peak
is given by

1
T1  sin 1  LC (2.39)


If the test object is an ideal surge arrester, the following approximate expressions
are used for determining the waveforms of the impulse current [10].

-69-
 V  Vr
Peak current = (2.40)
L/C

where Vr is the residual voltage of the surge arrester.

Front time T1  1.25 LC (2.41)

Time to half-peak value T2  2.5 LC (2.42)

2.5.2 Rectangular current impulses

The rectangular impulse wave as defined by the IEC recommendations is

shown in Fig. (2.21b) with the duration T 10% <1.5T90% [11]. The rectangular
current impulses are generated by discharging an artificial transmission line
lumped L and C elements into the test object through a sphere gap (Fig. 2.23). If
the line is charged to a DC voltage V and discharged through the test object of
resistance R, the current pulse is given by I = V/(R + Zo), where Zo is the surge
impedance of the line. A pulse voltage of RV/(R + Zo) is developed across the
test object. The duration T90% of the current wave is estimated as

n 1
T90%  2 L/C (2.43)
n

where n is equal to the number of L-C stages used [10].

-70-
 Fig. (2.23). Circuit for producing rectangular current waveform.

2.6 Nanosecond Pulse Generation

The mechanism of gas breakdown under high voltage pulses of

nanosecond duration is somehow different from the classical gas breakdown
mechanism [12]. This topic is discussed briefly in chapter 6. Due to an
increasing number of applications of discharges under nanosecond voltage
pulses, several electric circuits have been developed to produce such voltage
pulses. The basic idea is to convert long time low input voltage to short time
high voltage output. The long time part of the generator is the storing
component which is either capacitive or inductive. To get the high voltage pulse
a very fast switch (electric or spark) is closed to connect the load with the storage
part of the system. To get rectangular pulsed, a short transmission line is placed
with length () between the storage part of the circuit and the load, the pulse
duration (sec.) is given by:

2  r
t  (2.44)
c

where r is the relative permittivity of line insulant and c is the speed of light.

Several modifications have been incorporated in the above basic idea by

several investigators for achieving a better performance. Masuada [13] reported

-71-
a simple circuit to generate ns voltage pulses. A capacitor is rapidly charged and
discharged utilizing a rotary spark gap switch. Rea [14] analyzed the
characteristics of ns pulse generator circuit of Fig. (2.24) which is based on
Masuada's idea. In this case, the pulse rise-rate can be controlled by R and Li
while the pulse repetition rate is controlled by the rotation speed of the spark
gap. To improve the circuit performance, a DC supply, VDC, is added as a bias
and coupled to the circuit through C  10 Cp. It is found that the intensity of the
corona discharge produced by voltage pulses of this circuit depends on total
voltage magnitude, pulse voltage rise-rate, value of Cp and the shape of corona
electrode. Bajew et al [15] used a four stage Marx generator with rotating spark
gaps to generate ns voltage pulses. The main problems encountered in this
circuit are relatively longer rise-time (tens to hundreds of ns), the energy loss in
the sparking of the rotary gaps and the erosion of the spark gap. Alternatively,
HV semiconductor switching arrangement is introduced to overcome the
problems of mechanical switching where the pulse repetition rate can be
increased to very high values [16]. A typical waveshape produced by this circuit
is shown in Fig. (2.25) where the rise-time is around 35 ns.

Fig. (2.24). Nanosecond pulse generator circuit; dimension are in mm [16].

-72-
Fig. (2.25). Voltage pulse generated by ns pulse circuit using semiconductor
switching. (Y-axis=5 kV/div and X-axis=20 ns/div).

There are several commercially available pulse generators which can give
pulses with rise time as low as 50 ps and output voltage as high as 200 kV [17].
Generators are generally categorized according to primary switch type. Some of
these switches are spark, Reed, semiconductor opening switch (SOS) drift step
recovery device (DSRD), GaAs thyristor and inverse recovery diode (IRD).

The main applications of nanosecond pulsed corona are the removal of

harmful impurities from air [13,14], purification of water from volatile materials
[16] and the destruction of weeds and some other harmful insects and bacteria in
the soil [18,19].

2.7 Solved Examples

Example # 2.1

A HVDC doubler circuit has 60 Hz supply voltage of Vp = 100 kV. For

circuit capacitance of 0.02 F and load current of 10 mA find: i) ripple
voltage ii) voltage drop iii) ripple factor and iv) average output voltage.

Solution

i) Ripple voltage

-73-
 From equation (2.4) and for n = 1 (doubler circuit)

10 2 * 2
n n  1 
I
V 
4fC 4 * 60 * 2 *10 8
 4.17 kV

ii) Voltage drop

From equation (2.5) and for n = 1

I  2 3 n2 n  10 2 2 1 1
V   n        
fC 3 2 6  60 * 2 * 10  8 3 2 6

 8.33 kV

iii) Ripple Factor

V  V 12.5
   6.25%
2V 200

iv) Average output

Vav = 2Vp - (V+ V) = 200 – 12.5 = 187.5 kV

Example # 2.2

HVDC multiplier circuit has ten stages (n=10) with C = 0.12 F at each
stage. Supply voltage = 150 kV (peak), f = 60 Hz. When the load current is =
6 mA find (i) ripple voltage ii) voltage drop iii) optimum number of stages
from output voltage point of view; and iv) the ripple factor.

Solution

i) Ripple voltage

From equation (2.4)

6 *10 3 *10 *11

V   22.92 kV
4 * 60 *12 *10 8

-74-
ii) Voltage drop

From equation (2.5)

6 *10 3  2000 100 10 

V       595 kV
60 *12 *10 8  3 2 6

iii) Optimum number of stages

VDC = 2Vp - (V+ V)

Differentiate the above equation with respect to n and equate it to zero

to get optimum n. As a result

Vp .f .C 150 *103 * 60 *12 *10 8

n op    13.4  13 stages
I 6 *10 3

iv) Ripple factor

595  22.92
  20.6%
2 *150 *10

Example # 2.3

A 220 V/345 kV, 200 kVA testing transformer to be used for cable
testing has 10% leakage reactance and 2% resistance. The cable has to be
tested at 600 kV, 60 Hz while charging current is 0.5A. Find the required
value of the series inductor to be connected with the transformer to achieve
resonance. Assume (Rind = 1.8% of Xind). Find the primary voltage of the
transformer during this test. Also calculate rated current of this transformer
and the input power.

Solution

Assume no-load losses of transformer and the dielectric loss of cable

are zero.

-75-
Transformer reactance Xt is given as

X L kV 2 10 3452
Xt  * *103  * *103  59.5 k
100 kVA 100 200

2
Transformer resistance Rt =   (59.5) = 11.90 k
 10 

Cable reactance Xc is given as

Vc 600
Xc    1200 k
Ic 0.5

Since at resonance XL = Xc, the reactance of additional series

inductance must be 1200 - 59.5 = 1140.5 k. Thus, the required additional
inductance is

1140.5 * 103
L  3025 H
2 * 60

1.8
Inductor resistance *1140.5 = 20.5 k
100

Thus, total resistance in the current = 11.9 k + 20.5 k = 32.5 k

200
Transformer rated current =  0.58 A
345

Transformer secondary voltage = IRT since in resonance, total X = 0.

Vs = 0.58 * 32.4 k = 18.8 kV

220 * 103
Thus, the input primary voltage = * 18.8 = 12 V
345 * 103

The input power = output power = 18.87 * 103 * 0.5 = 9.4 kW

-76-
Example # 2.4

Refer to the impulse generating circuit shown in Fig. (2.13c). The

values of circuit elements are C1 = 1.12 F, R1 = 200 , C2 = 2 nF and R2 =
600 . Find the output voltage as function of time, the voltage waveform, and
the the circuit efficiency. The DC supply voltage is + 100 kV.

Solution

It is first required to find the constants a, b, and k.

 1 1 1 
a     
 R 1C1 R 1C 2 C1R 2 
 1 1 1 
 6
 9
 6   2.556 *10
6
 0.12 *10 * 200 2 *10 * 200 0.12 *10 * 600 

1
b  3.47 *1010
R1R 2 C1C 2

k = R1C2 = 4 * 10-7

2
a a
,       b
2 2

2
2.556  2.556 *103 
 *103     3.47 *1010
2  2 
 

  1.278 *10 6  1.6331012  3.47 *1010

 = 1.278 * 106 + 1.264 * 106 = 2.54 * 106

 = 1.278 * 106 - 1.264 * 106 = 1.4 * 104

From eq. (2.28)

-77-
 vt  
Vo
exp  t   exp  t 
k    

vt  
4 *10
7
100
2.54 *10 6
 1.4 *10 
4
exp  1.4 *10 t  exp  2.54 *10 
4 6

v(t) = 98.96 [exp (-0.014t) - exp (-2.54t)] kV

where t is in s.

Fig. (e2.4) shows the plot of v(t) against t. From this figure

t 0.9  t 0.3
tf = = 1.17 s,
0.6

tp = 2.0 s, and

tt  51.8 s

and Vp = 96.63 kV

Fig. (e2.4)

This is lightning impulse.

-78-
Another method to find tpeak is to differentiate v(t) with respect to t and
equate it to zero.

0 = 99.88 [-0.027 exp (-0.014 tf) + 2.53 exp (-2.53 tf)]

2.54
= exp [tp (2.54 - 0.014)]
0.014

185.4 = exp (2.5263 tp)

Take ln of both sides gives

5.222 = 2.5263 tp

tp = 2.07 s.


n  

Note tp =  


To find tt find Vp

Vp = V(tp) = 96.6 kV

Vp
Then when v(t) is , t = tt
2

Thus tt is found to be 52 s. For the circuit voltage efficiency  = Vp/Vo =

98.4%. Also efficiency is given as

C1
 (for high efficiency circuit)
C1  C 2

0.12 *10 6 0.12

 6 9
  98.4%
0.12 *10  2 *10 0.122

-79-
Example # 2.5

Using the same data in example # 2.4 repeat the calculation for the
circuit of Fig. (2.15.d).

Solution

For this low efficiency circuit,  and  are calculated as in example

(2.4) where the constants b and k are the same as in example # 2.4. However,
the constant "a" is given by equation (2.31). Thus

1 1 1
a  
R 1C1 R 1C 2 R 2 C 2

1 1 1
  
6 9
0.12 *10 * 200 2 *10 * 200 600 * 2 *10 9
 3.375 *10 6

2
3.375 *10 6  3.75 *10 6 
     3.47 *1010

2  2 

 3.364 *10 6

 = 0.01 * 106 = 1 * 104

v t  
100
4 *10 7 
3.364 *10 6  1*10 4 
[exp (-1 * 104 t) - exp (3.364 * 106 t)]

V(t) = 74.54 [e-0.01t - e-3.364*t)

where t is in s

ln  /   5.8
tp    1.73 .s
 3.354 * 106

-80-
Fig. (e2.5) shows the plot of V(t). From this figure

tt = 72 s tf = 1 s, tp = 1.5 s, and Vp = 76 kV

 can be found to be

C1 R2 600
 .  0.983 *  73.73%
C1  C 2 R1  R 2 200  600

As can be seen that the main difference between the low  and high 
circuit is the value of Vp. There is a small difference in the value of tf and
some differences in the values of tt and tp.

Fig. (e2.5)

Example # 2.6

Refer to the circuit of Fig. (2.15c) which is used to generate a 1.2/200

s impulse. Find the wave shaping resistors R1 and R2 when C1 = 0.125 F
and C2 = 1 nF. Use (a) approximate method; (b) using table (2.1); (c) exact
method.

-81-
Solution

(a) Approximate method

Using equation (2.31) and (2.32)

C1C 2
t f  3R 1
C1  C 2

 0.125 *10 6 *10 9 

1.2 * 10 = 3 R1 
-6
6

9 
 0.125 *10  10 

R1  400 

tt = 0.7 (R1 + R2) (C1 + C2)

200 * 10-6 = 0.7 (R1 + R2) (0.125 * 10-6 + 10-9)

R1 + R2 = 2268,

Thus R2  1868  1.87 k

(b) From Table (2.1) for 1.2/200 impulse

1
 284 *10 6 and

1
 0.38 *10 6


and since tf << tt equation (2.33) can be used and

1
 ,
R 2 C1

1
 ,
C 2 R1

-82-
Thus

284 *10 6
R2   2272  2.8 k
0.125 *10 6

0.38 *10 6
R1   380
1*10 6 9

(b) Exact solution

This is solved by iteration (or do loops). In this problem R1 is varied

between 160  - 400  in a 20  step, while R1 is changed from 1.4 k - 3
k in 0.2 k step. Then, a,b,k, and  are calculated.

V t  
Vo
exp  1t   exp t 
k    

Then t0.3 and t0.9 are found and consequently

tf (t0.9 - t0.3) / 0.6

Also tt is found when V(t) = Vp/2

The programme is terminated when

tf (1.2 - tf) is less than 0.2 s

tt (200 - tt) is less than 1.0 s

By doing the above the values of R1 and R2 are calculated to be 300 

and 2.5 k, respectively.

Although R2 in this exact solution is similar to those found in the

previous two methods, R1 have some difference since the wave shape is very
sensitive to R1 values.

-83-
Example # 2.7

A ten stage impulse generator have C1 = 0.2 F per stage and external
load capacitor C2 = 1 nF. Using two methods, find the values of shaping
resistors at each stage to generate standard lightning impulse (1.2/50 s). Also
find the circuit efficiency. When DC charging voltage is 100 kV what is
generator energy rating.

Solution

The equivalent single stage high efficiency generator circuit is shown in

Fig. (e2.7).

Fig. (e2.7).

n = 10,

Thus, nVo = 1000 kV, C1/n = 0.02 F, C2 = 1 nF and the front and tail
resistors are 10R1 and 10R2.

Using approximate method [equations (2.31) and (2.32)]

 0.2 
 *10 6 *10 9 
tf = 1.2 *10 6  3 10R 1   10 
 0.2 *10 6  10 9 
 
 10 

-84-
 R1 = 42 . Also

 0.2 
tt = 50 * 10-6 = 0.7 * 10 (R1 + R2)  *10 6  10 9 
 10 

R2 = 300 

Using Table (2.1) for 1.2/50 s impulse

1 1
 68 *10 6 and  0.4 *10 6
 

10 R2 C1 = 68 * 10-6

68 *10 6
R2 = = 340 
0.2 *10 6

10 R1 C2 = 0.4 * 10-6

0.4 *10 6
R1 = = 40 
10 *10 6

Vp C1 0.02 *10 6 0.02

     95.2%
nVs C1  C 2 0.02 *10 6  1*10 9 0.021

Thus, Vp = 95.2 * 100 * 10 = 952 kV

1  C1  2
The generator energy rating =   Vp
2  10 

1  0.2 *10 6 
 
2 10
 
 (952) 2 * 103

2
= 9.06 kJ


Example # 8

Design a circuit which will produce the following impulse voltage.

-85-
 V(t) = 150 (e-316t - e-16000t) kV

Solution

We first find tf & tt. For this purpose plot V(t) as function of t. The
plot is shown in Fig. (e2.8). From such a plot we can find tf and t0.5

Fig. (e2.8).

Here  = 316 (s)-1, and  = 1600 (s)-1. Therefore

16000
n  /   316
tf = tp =   250 s
 16000  316

Also, from the curve tf  250 s and tt = 2500 s

Alternatively

Vp = V(tf) 
= 150 e 316*250*10
6
  e 16000*250*106  
= 131.7 kV

-86-
Thus,
Vp
2

= 65.85 = 150 e 316 t f   e 16000 t f  
From the above equation tf is found to be  2500 s

Example # 9

An impulse current generator generate double exponential waveform

with C = 10 F and L = 10 H. Draw the output current if the charging
voltage is 20 kV and the time to the first peak is 10 s.

From Equation (2.36)

V
I(t) = exp. (-t) sin t
L

 and  from the above equation must be found.

From Equation (2.39)

1
T1 = sin-1  / LC


1
10-5 = sin-1 10-5 


Thus  = 0.2 * 104

From Equation (2.38)

1 R2
 
LC 4L2

R = 1.99 k  2 k

-87-
Thus  from Equation (2.37)

 = R/2L = 99.5 * 103

Thus i(t) = 106 * exp (-99.5 *103t) sin (0.2 * 104)t

The output current is shown in Fig. (e2.9).

Fig. (e2.9).

-88-
Questions / Problems

1. Derive the equation giving ripple voltage in a HVDC voltage multiplier

circuit.

2. Derive the equation for voltage drop of a HVDC multiplier circuit.

3. Suggest ways of reducing voltage ripple and voltage drop in HVDC

multiplier circuits.

4. Derive the formula for the optimal number of stages from output
voltage point of view in a HVDC voltage multiplier circuit.

5. Explain with a neat sketch the working principle of HVDC multiplier

circuit.

6. Explain the working principle of cascaded transformer used for HVDC

generation and discuss its advantages and drawbacks.

7. Calculate the total impedance of 3-stage cascaded transformers. State

assumptions made.

8. If the power needed from a cascaded transformer is high, each stage is

powered separately from the source. Why will such an arrangement
work?

9. Describe the working principle of a resonance circuit for HVAC

voltage generation. Discuss its advantages and disadvantages.

10. Explain how high frequency oscillations are generated using an AC

supply and Tesla coil.

11. Define the meaning of the front time and time to half value for standard
lightning and switching impulses. Also specify values of these times for
standard lightning and standard switching impulses. Write the

-89-
 mathematical equation for an impulse voltage and explain how this
equation can be used to define a standard lightning impulse and a
standard switching impulse.

12. Why are impulse voltages used for testing high voltage equipment?
What are the reasons for the use of standard voltage waveforms?

13. Discuss the two types of high voltage impulses that are generated in the
laboratory for the purpose of testing.

14. What is the difference between a surge and an impulse?

15. Define the front and tail times for lightning and switching impulses.
Give the tolerances as given by IEC specifications for standard
impulses.

16. Sketch two circuits used to generate HV impulses. Explain the basic
difference between these circuits.

17. Derive the impulse output voltage in time domain (equation 2.28).

18. How can we control the front-time and tail-time in an impulse circuit?
How is the peak amplitude of an impulse controlled?

19. Define the voltage efficiency of an impulse generator. Derive the

efficiency for the circuit shown in Fig. (2.15c).

20. Explain how a multistage impulse generator circuit can be analyzed by

using the equations of a single stage circuit. Summarize the equations
and the analysis procedure.

21. Draw waveforms of chopped impulses with chopping on front, on peak

and on tail. What are the uses of such impulses? How is such
chopping achieved in practical systems?

-90-
22. Sketch various circuits used to generate switching impulse voltage.

23. Discuss the effects of inductance of the impulse circuit on the output
voltage waveform.

24. Explain how an impulse current is generated. Sketch the circuit, and
derive the output formula. How is the current wave shape controlled?

25. Repeat Q. 24 for rectangular current impulses.

26. Why do we need the triggering of an impulse generator. Explain a

circuit which achieves such triggering.

27. Where do HV pulses with nanoseconds duration are needed? Sketch

and explain a circuit which can produce such pulses.

28. For HVDC voltage multiplier, draw the voltage drop and ripple voltage
as functions of:

a) Number of stages for a fixed load current.

b) Load current for a fixed number of stage.

Given C = 0.06 F per stage.

I ranges between 1 mA to 10 mA.

n ranges between 2 and 14.

Give your comments on the results.

29. Given the load of a DC voltage doubler circuit as 4 M resistor, and

the output capacitance as 1.2 nF for a source voltage of 100 kV (rms),
60 Hz draw the output voltage and calculate the voltage ripple and
voltage drop.

30. A 60 Hz AC source of 250 kV (rms) voltage is supplying a DC doubler

circuit. The capacitance of the doubler circuit is 20 nF. If the load can

-91-
 be represented by a resistance of 15 M, what are the values of voltage
ripple and voltage drop? Calculate the output DC voltage for this
system.

31. A 25-M resistor is the load of a HVDC voltage doubler circuit. The
output capacitance is 1500 nF and the source voltage is 250 kV (rms),
60 Hz. Calculate the output voltage, the voltage ripple and the voltage
drop.

32. Repeat problem 30 above by assuming that the circuit has 4 stages and
the external load resistance is 100 M.

33. A 3-stage voltage multiplier circuit is constructed using 0.05-F

capacitors and a transformer with a 60 Hz, 50 kV (rms) output. A 50
M divider is connected across the output terminals for voltage
measurement. Calculate the voltage drop, the output voltage and the
peak-to-peak ripple for the following external loads:

a) No load. b) 50 M load.

34. Establish a table relating the reactive loading of the primary and the
total pu reactance of a cascaded transformer and the number of stages
n. Let Xp = 0.02 pu, Xh = 0.04 pu, and Xe = 0.05 pu. Comment on the
results. How many stages will you recommend in this case?

35. A 250 kVA, 400 V/345 kV testing transformer has 10% leakage
reactance and 2% resistance on 250 kVA base. A cable has to be tested
at 500 kV using the above transformer as a resonant transformer at 60
Hz. If the charging current of the cable at 500 kV is 0.45A, find the
series inductance required. Assume that the inductor used has a
resistance equal to 2% of its reactance. What will be the primary
voltage of the transformer?

-92-
36. Sketch the output voltage for the circuits shown in Fig. (p2.36). In each
case first find the charging and/or discharging time constants (s).

Fig. (p2.36).

37. Derive the voltage efficiency for the impulse generating circuits shown
in Fig. (p2.37).

-93-
 Fig. (p2.37).

38. Design a circuit to produce the following switching impulse.

V(t) = 500 (-300t - e-50t) kV

Hint, first find tf, tt and Vp. Then assume C1 and C2 and calculate the
required R1 and R2.

39. A single stage impulse generator has the following parameters: R1 =

250 , R2 = 25000 , C1 = 1.5 F, C2 = 0.1 F, Vo = 200 kV. Find
equations for the voltage waveform using two type of single stage
impulse circuits. Compare the values of tp, tt and Vp and voltage
efficiency for the two circuits.

40. Draw the output voltage for a multistage generator circuit having the
following data: DC charging voltage = 125 kV, C1 = 30 F/stage, R1 =
40 /stage, R2 = 4.5 k/stage, C2 = 0.2 nF and n = 12 stages. From the
voltage wave find tf and tt. Also find the generator energy, and its
voltage efficiency.

41. A multistage impulse generator circuit has the following parameters:

Charging voltage per stage = 100 kV, C1 = 25 F/stage, R1 = 20

-94-
 /stage, R2 = 5 k/stage, C2 = 0.5 nF and n = 6 stages. Assume
external front resistor = 50 ohms. Find tf and tt, Vp and .

42. Design a single stage impulse circuit to generate standard lightning

impulse. Assume that capacitors of 0.2 Fand 20 nF are available.

43. The charging voltage for impulse current generator is 100 kV. Find the
value of a capacitance which will give t2 (time to half value)  1300 s
when L = 2 mH and R = 1 . Find the output current formula and draw
the current waveform.

-95-
References

[1] E. Kuffel, W. Zaengle, and J. Kuffel, "High Voltage Engineering

Fundamentals" Newnes, Oxford, U.K., 2002.

[2] M. Naidu and U. Kamaraju, "High Voltage Engineering" Tata McGraw

Hill, New Delhi, India, 1995.

[3] A. Alessa and M. Nuaena, "High Voltage Tests for GIS at Site", 1st
Applied Symp an O & M of Power Networks, March 1999, Riyadh,
Saudi Arabia.

[4] R. Bartnikas and K.D. Srivastava, "Power and Communication Cables"

IEEE Press, New York, USA, 2000.

[5] IEC 60060-1, "High Voltage Test Techniques – Part 1, General

Definitions and Test Requirements", IEC Geneva, Switzerland, 1989.

[6] Haefely, Publication BE-701, Basel, Switzerland, 1982.

[7] Haefely Products Broucher GC223, Basel, Switzerland, 1997.

[8] M. Abdel Salam, H. Anis, A. El-Morshedy and R. Radwan, "High

Voltage Engineering Theory and Practice", Marcel Dekker Inc., New
York, USA, 2000.

[9] N.H. Malik, A.A. Al-Arainy, and M.I. Qureshi, "Electrical Insulation in
Power Systems", Marcel Dekker Inc., New York, USA, 1998.

[10] IEC 60099-1, "Non-Linear Resistor Type Gapped Surge Arrestors for AC
Systems. IEC, Geneva, Switzerland, 1999.

[11] IEC 60060-2, "High Voltage Test Techniques - Part II – Measuring

System", IEC, Geneva, Switzerland, 1994.

-96-
[12] J.M. Meek and J.D. Cruggs, "Electrical Breakdown of Gases", John
Wiley & Sons, New York, USA, 1978.

[13] S. Masuada and S. Husakawa, "Submicrosecond Pulse Energization for

Retrofitting Applications", Proc. 2nd Int. Conf. on Electric Precipitation,
Kyoto, Japan, p.613, 1984.

[14] M. Rea and K. Yan, "Evaluation of Pulse Voltage Generators", IEEE

Trans. on Industry Applications, Vol. 31, No. 3, pp. 507-512, 1995.

[15] G. Bajeu, I. Mustata, C. Lungu and G. Musa, "Narrow Pulsed Corona

Discharges for DC pollution", Proc. IEEE Industry Application Society’s
Annual Meeting, Vol. 3, pp. 1583-1586, 1994.

[16] A. Al-Arainy, S. Jayaram and J.D. Cross, "Pulsed Corona for Removing
Volatile Impurities from Drinking Water", Proc. of 12th Int. Conf. on
Cond. and Breakdown in Diel. Liquids (12-ICDL 96), Rome, Italy, pp.
427-431, July 1996.

[17] J. Mandowski, and M. Kristiansen, "A Review of Short Pulse Generator

Technology", IEEE Trans. on Plasma Science, Vol. 28, No. 1, pp. 102-
108, Feb. 2000.

[18] A. Mizuno and Y. Hori, "Destruction of Living Cells by Pulsed High

Voltage Applications", IEEE Trans. on Industry Applications, Vol. 24,
No. 3, pp. 387-394, 1988.

[19] A. Mizuno, T. Inoue, S. Yamaguchi, and K. Sadamoto, "Investigation of

Viruses Using Pulsed High Electric Field", IEEE Industry Application
Society’s Annual Meeting, pp. 713-719, 1990.

-97-