The 20th International Symposium on High Voltage Engineering, Buenos Aires, Argentina, August 27 – September 01, 2017

SUPERIMPOSED IMPULSE VOLTAGE TESTING ON EXTRUDED DC-

CABLES ACCORDING TO IEC CDV 62895
1* 1
A. Voß and M. Gamlin
1
Haefely Test AG, Birsstrasse 300, 4052 Basel, Switzerland
*Email: Voss.Andreas@haefely.com

Abstract: As HVDC applications rise in the area of energy supply, there is a need for an
appropriate HVDC testing of power transmission cables with extruded insulation and their
accessories (new IEC CDV 62895 Ed. 1.0 “High Voltage Direct Current (HVDC) power
transmission cables with extruded insulation and their accessories for rated voltages up
to 320 kV for land application – Test methods and requirements”). The type and prequali-
fication tests on HVDC cable systems include the superimposed impulse voltage test,
during which several switching and lightning impulses are applied on the HVDC cable
which has been energized for multiple hours by a DC voltage. The DC voltage shall be
maintained during the application of the impulse voltage. This test condition poses a
technical challenge to the test setup, since the DC source and the impulse voltage gener-
ator have to be protected from each other. Neither must the DC voltage be present at the
impulse voltage generator nor must the impulse voltage have any influence on the DC
voltage generator. The superimposed impulse voltage test on HVDC cables is a compo-
site test according to IEC 60060-1 Ed. 3.0 which requires suitable coupling and protection
elements in combination with suitable converting devices and recording instruments. For
the impulse side, the coupling and protection is possible with either a spark gap or a ca-
pacitor. For the DC side, the coupling and protection is realized by a damping resistor.
The pros and cons as well as the design aspects of the coupling and protection devices
are introduced. A possible smart solution for the capacitive coupling by using spare ca-
pacitor stages of the impulse voltage generator is discussed. Additionally the revision of
IEC CD 60230 Ed. 2.0 “Impulse tests on cables and their accessories” is commented with
respect to the proposed test setups and the measuring system for the superimposed im-
pulse voltage test.

1 INTRODUCTION 2 SUPERIMPOSED IMPULSE VOLTAGE

TEST
The changes in the energy supply system, which
happened over the past two decades, demand The test standard for HVDC cables with extruded
technical solutions fundamentally different from insulation considers multiple tests such as the
before. There is an ever growing demand for ener- heating cycle voltage, the polarity reversal and the
gy transmission over increasing distances and with superimposed impulse voltage test [3]. Of these
increasing power. Due to its typical cost structure, tests, especially the superimposed impulse voltage
which shows higher fixed costs but lower costs per test, which superimposes a DC voltage with occa-
length, DC transmission lines become more cost sional impulses, poses technical challenges to the
efficient than AC technologies above a certain testing equipment. The test is a composite test
break even distance. The increasing power de- according to IEC 60060-1 Ed. 3.0 [4] which means
mand of these connections additionally calls for the different voltages are applied to the same test
higher transmission voltages. This altogether gives object, implicating the test equipment generating
a strong need for HVDC transmission cables, the voltages also has to withstand the superim-
which can e.g. be used for connecting offshore posed stress. In order to achieve this, additional
wind farms to the grid or to interconnect different elements in between the test equipment and the
grids. test object have to be used, which serve as a cou-
pling for the generated voltage but at the same
These needs led to technological developments of time as a blocking element for the other, superim-
HVDC cables with extruded insulation for up to posed voltage form.
600 kV and of mass impregnated cables for even
up to 700 kV [1]. However, the recent history with 2.1 Test setup
project delays due to cable problems also illus-
trates the increased technical demands of such For the test setup, IEC 60230 Ed. 2.0 gives two
cables [2]. Therefore, there exists an equally different possibilities:
strong need for accurate testing of HVDC cables in
order to ensure a safe operation which is mandato-
ry for a reliable energy supply.
 measured at the test object, as depicted in Figure
1.1.

2.2 Voltage waveshapes

The following composite voltage waveshapes are

given in the IEC standard for the superimposed
impulse voltage test of HVDC cables:

Figure 1.1: Superimposed impulse test arrangement with a blocking capacitor

Figure 1.2: Superimposed impulse test arrangement with a sphere gap

Figure 1: Test setups for superimposed impulse

voltage test [5].

In both setups, a current limiting resistor is used to

protect the DC voltage generator from the impulse
voltage. This resistor needs to be sufficiently large
in order to neglect the impulse voltage’s influence
on the DC Set and avoid any harm.

For the protection of the impulse voltage generator,

there exist two possible setups. In the first option,
the impulse circuit is permanently connected to the
test object via a capacitor. The capacitor presents Figure 2: Voltage waveshapes for the superim-
an open circuit to the DC voltage thereby decou- posed impulse voltage test [3].
pling it from the impulse circuit. On the other hand,
the capacitor presents a sufficiently small imped- As shown in Figure 2, there exist six different volt-
ance to the fast impulse voltages, so that it has no age waveshapes. The DC voltage is superimposed
major influence and the impulse is “pushed with a switching impulse of either the same or the
through” the capacitor to the test object. opposite polarity or with a lightning impulse of the
opposite polarity. These tests are performed for
In the second option, the impulse circuit is discon- both DC polarities.
nected from the test object by a sphere gap. In that
way, the impulse generator is only connected to The composite voltage is defined as the “maximum
the test object when the impulse is applied and the absolute value measured at the test object” [4] so
sphere gap fires. During the rest of the test proce- for the aforementioned cases, there are three dif-
dure, the impulse circuit is not connected to the ferent composite voltage levels:
test object.
 UP2,S defines the composite voltage when the
Figure 1 also shows different possibilities for con- switching impulse is applied at the same po-
necting the measuring system. For composite volt- larity as the DC voltage (U0).
age tests in general, there are three possible spots
for the connection of the measurement: At the DC  UP2,O defines the composite voltage when the
Set, at the impulse voltage generator or at the test switching impulse is applied at the opposite
object. Figure 1.1 shows the measurement at the polarity as the DC voltage (U0). UP2,O is given
test object whereas Figure 1.2 shows the voltage as 1.2 U0.
measurement at the impulse voltage generator
which is decoupled from the DC Set and the test  UP1 defines the composite voltage when the
object by a sphere gap. In accordance with IEC lightning impulse is applied at the opposite po-
60060-1 Ed. 3.0, the voltage measurements at the larity as the DC voltage (U0). UP1 is given as
sources should only be used for voltage adjust- 2.1 U0.
ments and the composite voltage should always be
3 COUPLING AND PROTECTION ELE- The exact mechanical design of the resistor de-
MENTS pends heavily on the DC voltage level of the test
object and the according composite voltages.
After the description of the framework given by the However, it is crucial, to keep the different stresses
different IEC standards, this paragraph focuses on of the composite test on the resistor in mind and
the optimal design of the coupling and protection design it accordingly.
elements as well as the analysis of the different
test setups. 3.2 Impulse coupling

3.1 DC coupling Whereas the coupling resistor is mandatory for

connecting the DC Set to the test object, there are
The current limiting resistor is the coupling and two different options for the impulse voltage gener-
protection element used for connecting the DC ator connection: A blocking capacitor or a sphere
generator to the test object. It is a key part used in gap. Whereas a sphere gap only connects the
both setups given in Figure 1 and therefore a man- impulse generator during the impulse, when it is
datory part in the superimposed impulse voltage necessary, a blocking capacitor presents a steady
test setup. connection which however is only possible for the
impulse voltage to pass. It is a main incentive of
In order to entirely block the impulse voltages from this paper to present the advantages and disad-
the DC set, the resistance needs to be sufficiently vantages of the two solutions and to identify the
high so that the impulse capacitors are discharged superior.
via the test object and the waveshaping resistors
but not via the DC set. On the other hand, the re- 3.2.1 Sphere gap
sistor also determines the time constant
 = R * CDUT, by which the test object is charged The first possibility of connecting the impulse volt-
with the DC voltage, and forms a resistive voltage age generator to the test object is by using a
divider together with the measuring system, which sphere gap. It disconnects the impulse voltage
reduces the DC Set’s efficiency (s. Figure 1.1). generator from the DC voltage when no impulse is
Therefore, the resistance must not be excessively applied and it fires during the impulse thereby con-
high either to allow for a fast DC charging of the necting the impulse voltage to the test object. In
test object and a reasonable efficiency of the DC order for that test setup to work properly, it is es-
source. sential to control the sphere gap so it is conducting
during the whole impulse and insulating during all
As described above, the design of the DC coupling other times.
resistor depends heavily on the other components
of the test setup as the DC Set, the test object and First of all, the sphere diameter and the distance
the voltage divider. For the voltage levels however, between the spheres have to be chosen in that
which the analyzed IEC 62895 is referring to, a way, that the sphere gap can permanently and
resistance in the range of 10 – 100 M was found safely withstand the applied DC voltage UDC. Addi-
suitable to fulfill all the aforementioned criteria. A tionally, the voltage across the sphere gap USG has
lower resistance allows for a faster charging of the to increase during the impulse by at least an
test object and a better DC efficiency but puts amount U, so that the voltage across the gap is
more stress on the DC Set. The same applies for a higher than the breakdown voltage UBD and the
larger resistance vice versa. gap is triggered according to IEC 60052. The volt-
age drop across the sphere gap can be calculated
The design of the DC coupling resistor however is according to Figure 3 by:
not only limited to the electrical resistance but also
to the mechanical dimensions. These are strongly USG = UDC – UIMP (1)
determined by the different composite voltages
which were described in section 2.2: USG
 UP2,S and UP1 define the maximum switching
and lightning impulse voltage levels at the test
object and the resistor. They are the relevant
voltages for the field strength design of the re-
sistor.
UDC = ∧ UIMP
 UP2,O and UP1 define the maximum switching
and lightning voltage drops across the resis-
tor. They are the relevant voltages for the de-
sign of the resistor’s length.
Figure 3: Voltage drop across sphere gap
In order for the sphere gap to trigger, the absolute The sphere gap’s extinction however is just one
voltage across the sphere gap has to be bigger disadvantage of this arrangement, which has to be
than the breakdown voltage: controlled. A second problem is the cable’s dis-
charge for opposing impulses, which occurs when
|USG| = | UDC – UIMP| > UBD = UDC + U (2) the sphere gap stays conducting during the whole
impulse, which is desirable as described above.
This gives two conditions, depending on if the im- During that time, the test object is charged as usu-
pulse has the same or the opposite polarity as the al by the DC Set with a time constant DC = RCoup *
DC voltage: CDUT but the test object is also discharged via the
UIMP,O < – U (3) front and tail resistors of the impulse voltage gen-
UIMP,S > 2 UDC + U (4) erator with a second time constant
IMP = (RFront + RTail) * CDUT, which can be deduced
Equations 3 and 4 clearly show that the sphere from Figure 1.2. Since the impulse time constant
gap is far easier to trigger for opposing impulse will be in the range if the impulse’s duration, a con-
polarities, since this increases the voltage drop siderable part of the energy, which is stored in the
across the sphere gap. If the impulse has the same test object, will be discharged via the impulse re-
polarity as the DC voltage, it has to be at least sistors causing the voltage to not recover to the full
twice as high as the DC voltage plus an additional original level. At the same time, the DC time con-
increment U to safely trigger the gap. If this re- stant is far bigger than the impulse duration so that
quirement is not fulfilled, an artificial triggering, the DC Set will not be able to recharge the cable.
which is synchronized to the firing of the impulse The resulting voltage waveshape is shown in Fig-
generator, will be necessary. ure 5.

However, the reliable triggering is not the only

challenge for the sphere gap operation. The
sphere gap also has to stay conducting during the
whole impulse. In order to achieve this, a certain
minimal voltage gradient dUSG/dt is always re- Slow DC recharging

quired. For fast lightning impulses, this does not

Sphere Gap Extinction
pose too much a challenge whereas for slower Cable is completely discharged

switching impulses, this circumstance might not

always be fulfilled. Especially during the switching
impulse’s peak, the voltage gradient is rather small
and the sphere gap might extinguish. If this is hap-
pening, the cable is charged to the switching im- Figure 5: Test object discharge during opposing
pulse peak voltage but due to the disconnected impulses
impulse generator, it cannot discharge via the tail
resistors, leading to an almost constant voltage Summarizing, the advantage of the sphere gap is
plateau after the impulse’s peak. Since the impulse the relatively easy to implement setup, as long as
voltage generator discharges regularly via the tail the triggering conditions are fulfilled. The clear
resistors, the voltage drop across the sphere gap disadvantages however are the possible extinction
will increase over time and a refiring of the sphere of the sphere gap during times of low voltage gra-
gap during the impulse’s tail occurs. At that time dients and the test object’s discharge during the
however, the impulse voltage generator is already impulse. Therefore, it is not possible with this test
that far discharged, that the voltage drops strongly setup to generate a clean superposition of a DC
and the resulting waveshape is no longer a switch- voltage with an impulse voltage.
ing impulse. Figure 4 shows the resulting voltage
across the test object. 3.2.2 Coupling capacitor

The second possibility to operate this test is via a

coupling capacitor between the impulse voltage
generator and the test object. This capacitor works
as an open circuit to the DC voltage thereby de-
coupling the impulse voltage generator from the
DC source even though there is a permanent con-
nection. That requires the capacitor to being able
to withstand the full DC voltage. On the other hand,
the coupling capacitor forms a capacitive voltage
divider together with the test object during the im-
Sphere Gap Sphere Gap
Extinction Refiring pulse. That requires the coupling capacitor to have
Figure 4: Sphere gap extinction during switching a much larger capacitance than the test object in
impulse tail order to not overly decrease the impulse voltage
generator’s efficiency. However, if these criteria using the impulse capacitors as coupling elements,
are fulfilled, this solution offers the possibility for a this option clearly forms the better way of conduct-
clean superposition of the two different voltages for ing superimposed impulse voltage test for HVDC
all described composite test voltages. The results equipment.
for that are shown in the following figure.
4 MEASURING SYSTEM
After the impulse, the test object
voltage returns to full DC level Whereas the previous paragraph focused on the
test setup to perform the test properly, this section
will describe the requirements for an appropriate
measuring system. According to IEC 60060-1,
there are generally three different measurement
locations for composite voltage tests: At each of
the sources or at the test object. However, since
the coupling elements are designed in a way to
only allow the generated voltage to pass and to
block the opposing voltage, the measurements at
Figure 6: Superimposed voltage waveshapes the sources can solely be used for the voltage
using a coupling capacitor source adjustment to properly generate the desired
voltage level and shape. Therefore, an appropriate
The clear advantage of this test setup is the composite voltage measurement always has to be
achievement of the desired composite voltage taken at the test object.
waveshape, which was not possible with the
sphere gap setup. The disadvantage however is Since for the superimposed impulse voltage test, a
the practical implementation, since the coupling DC and an impulse voltage are superimposed, only
capacitor has to not only withstand the full DC a universal voltage divider is suitable for this appli-
voltage and have a high capacitance in the range cation. This divider consists of a high resistive
of a few hundred microfarads, but it also has to be column, which is suitable for DC measurements,
placed on high voltage potential on both terminals. and a damped capacitive column, which is suitable
This implementation however is easily possible by for impulse voltage measurements [6]. However,
extending the impulse voltage generator to more not only the divider type is important for an appro-
stages, than would be necessary for the impulse priate composite measurement. The different volt-
generation, and use the spare impulse capacitors ages also demand adaption by the recording sys-
as DC coupling capacitor. This possibility presents tem. During the superimposed impulse voltage test
a smart solution for realizing the test setup without cycle, the cable is first heated by the DC voltage
the need for additional elements. This configuration for at least ten hours [3]. During this period, the DC
is shown in Figure 7. voltage has to be recorded by the measuring sys-
tem with a low sampling rate such as 1 Hz in order
to not exceed the memory depth of the system.
Since there are no fast voltage changes during the
sole DC application, this is acceptable.

After the heating of the cable, the impulses are

Coupling Capacitor applied and during this period, the sampling rate of
the recording system needs to change. During the
impulses, the sampling rate needs to increase to at
least a couple of 10 MHz, to properly record the
exact impulse shapes. In between the impulses,
the sampling rates needs to reduce to the DC level
Impulse Voltage though, in order to not exceed the memory depth.
Generator This adaption is called a dynamic sampling rate
and is a key point for the successful measurement
of composite voltages [7].

5 CONCLUSIONS

With the increasing demand for HVDC transmis-

sion cables and the recent technical problems
Figure 7: Superimposed impulse voltage genera- herewith, there is a strong need for the reliable
tor using spare stages as capacitive coupling testing of these components. One part of this test-
ing is the superimposed impulse voltage test ac-
Due to the superior results and the possibility, to cording to recent IEC standards 60230 and 62895.
avoid the disadvantages of this solution by smartly These drafts consider two possible test setups. A
thorough analysis has shown that only one of [5] IEC CDV 60230 Ed.2.0, “Impulse tests on ca-
these setups presents a suitable solution for this bles and their accessories”, 2014.
test in order to be performed properly. It should be [6] M. Modrusan, “Hochspannungsteiler: Typen,
strongly preferred to connect the impulse generator Messeigenschaften und Einsatz”, Bull.
by a coupling capacitor to the test object. A possi- ASE/UCD 74 (1983) S. 1030-1037.
ble smart solution for this is to extend the impulse [7] Haefely Hipotronics. Hias 744 Product Bro-
voltage generator by additional stages and use the chure, April 2017. URL: http://www.haefely-
spare impulse capacitors for this coupling. The hipotronics.com/wp-
recommended setup is shown in Figure 8. content/uploads/2017/02/LL_HiAS744.pdf, last
accessed: April 13, 2017.
Coupling Universal Impulse Voltage
Resistor Voltage Generator
Divider

Coupling
Capacitor
Composite
Test
∧
∧
UDC = Voltage UIMP
Object
Measurement

Figure 8: Recommended test setup for superim-

posed impulse voltage test

This test setup however is only suitable for the

superposition of DC and impulse voltages. The
coupling capacitor does not work for the superposi-
tion of AC with impulse voltages so for other super-
imposed application, a coupling via a sphere gap
might be the only technical choice.

The analysis has furthermore shown the need for a

composite measuring system. This system con-
sists of a universal voltage divider to measure the
different voltages at the test object and a recording
system, which has the ability to adapt the sampling
rate according to the test voltage. This also poses
a key requirement for an appropriate superim-
posed testing.

REFERENCES
[1] Prysmian Group. Double technology milestone
in the field of power transmission. URL:
http://www.prysmiangroup.com/en/corporate/m
edia/news/Double-technology-milestone-in-the-
field-of-power-transmission-00001, last ac-
cessed: March 22, 2017.
[2] S. Emrich. ABB hat Probleme mit Windpark,
July 2015. URL:
http://www.srf.ch/news/wirtschaft/abb-hat-
probleme-mit-windpark, last accessed: March
22, 2017.
[3] IEC CDV 62895 Ed.1.0, “High Voltage Direct
Current (HVDC) power transmission cables
with extruded insulation and their accessories
for rated voltages up to 320 kV for land appli-
cations – Test methods and requirements”,
2016.
[4] IEC 60060-1:2010, “High-voltage test tech-
niques - Part 1: General definitions and test re-
quirements”, 2010.