17th European Photovoltaic Solar Energy Conference, Munich, Germany, Oct. 22 – Oct. 26, 2001
Interference Voltages induced by Magnetic Fields of
Simulated Lightning Currents in Photovoltaic Modules and Arrays
H. Haeberlin
Berner Fachhochschule, Hochschule fuer Technik und Architektur, Jlcoweg 1, CH-3400 Burgdorf, Switzerland
Phone +41 34 / 426 68 11, Fax +41 34 / 426 68 13
e-mail: heinrich.haeberlin@fhburg.ch, internet: http://www.pvtest.ch
ABSTRACT: Already in 1990 – 1993 HTA Burgdorf’s PV laboratory had carried out tests about sensitivity of PV
modules against lightning currents flowing in or close to the frame of a PV module [1, 2]. For these tests a high impulse current generator with imax ≤ 100kA and di/dtmax ≤ 50kA/µs was used. With this device it was possible to expose
the part of the module closest to the impulse current (aera ≤ 50cm⋅40cm) to the fast changing magnetic field of this
current and measure the resulting voltage induced in the module. It was found that lightning currents flowing in the
metallic frames of PV-modules may cause a certain degradation of the I-V-characteristic and that frameless modules
are more sensitive to lightning currents. It was also shown that for modules with metallic frames an increase of the distance to the lightning current path to a few centimeters was already sufficient to avoid any damage to the module [1, 2].
In 1998 - 2000, in an EU project (PV-EMI, JOR3 CT98 0217, partners: FhG/ISE, HTA Burgdorf, KEMA) a larger
high impulse current generator could be built. With this device it is possible to expose whole PV modules and wired
models of PV arrays with an area of up to 1.25m⋅2.25m to the magnetic field of a high impulse current with
imax ≤ 120kA and di/dtmax ≤ 40kA/µs. These values are higher than those of average lightning currents. Therefore the
earlier experiments could be repeated on a much larger scale. Main results of these tests are given in this paper.
Keywords: PV modules – 1 : Induced voltages – 2 : Lightning – 3
1.
Coaxial Impulse Current Generator
Fig. 1 shows the principal layout of the coaxial impulse
current generator used for the tests. In fig. 2 there is model
of a PV array that is going to be tested in it. Fig. 3 shows a
typical impulse current waveform used for the tests. With
this generator, impulse currents up to 125kA with
maximum values of di/dt between 15kA/µs and 40kA/µs
(depending of the array size) can be produced. Such rise
velocities are representative for typical lightning currents.
Fig. 2:
Partial view of the
coaxial impulse
current generator
(through open door)
with a model of an
array consisting of 3
PV modules KC 60.
Fig. 3:
Waveform of a typical
impulse current used
for many tests.
imax ≈ 100 kA,
di/dtmax ≈ 25 kA/µs.
Scales used:
20kA/division and
2µs/division.
2.
Fig. 1:
Schematic diagram of the coaxial impulse current
generator used for the tests. A wired model of a PV
array with 3 modules is also shown.
Induced Voltages in single PV Modules
In each loop, no matter if it is an internal wiring loop in the
module or an external wiring loop, a lightning current
induces a voltage in that loop:
Induced voltage in a loop: u = M⋅di/dt .
(M = mutual inductance between current path and loop).
In a PV module with cristalline cells there are at least two
internal wiring loops. In principle, depending on the
orientation of these internal wiring loops, the voltages
induced in these loops can add or (partially) compensate.
17th European Photovoltaic Solar Energy Conference, Munich, Germany, Oct. 22 – Oct. 26, 2001
However, the situation is more complicated owing to the
bypass diodes, which are usually mounted across these
internal wiring loops, as these are either conducting or
reverse biased according to the polarity of the induced
voltage. Taking into account the effect of bypass diodes,
three different types of modules were identified:
• Additive modules with even row number (most
frequent type, example: Kyocera KC60):
Depending on the direction of the lightning current,
induced voltages in the internal loops add (bypass
diodes reverse biased) or are 0 (bypass diodes forward
biased and conducting).
• Compensating modules with even row numbers
(example: Solarex MSX 60/64):
Different polarities of the induced voltages in the loops.
Half of the loops are short circuited by their bypass
diodes. Only half of the loops are active and contribute
to the measured induced voltage at the terminals.
• Compensating modules with three rows (example:
Siemens SM46 / SM55):
Different polarities of the induced voltages in the loops.
One of the two loops is short circuited by its bypass
diode. Only the other loop is active and determines the
measured induced voltage at the terminals.
The highest induced voltages occur in the beginning of the
lightning current. In the first moment, all solar cells of a
module are shunted by a large capacitance (a few µF per
cell). Therefore a good approximation for the maximum
induced voltage can be obtained by calculating the voltage
in a wire going through the centres of the solar cells which
are part of the internal wiring loop.
2.1 Induced Voltages in an additive Module
(Kyocera KC60) in parallel Position
This module is an additive module with four cell rows. It is
rated for 60Wp at STC, 0.751 m long and 0.652 m wide.
Fig. 4 shows the position of the module and the current
conductor during the test.
Fig. 5: Induced voltages in KC60 without frame in
parallel position according to fig. 4 .
di/dtmax = 25 kA/µs. As voltage dividers 100:1
are used, actual voltages are 100 times higher.
Fig. 6: Induced voltages in KC60 with metal (Al) frame
in parallel position according to fig. 4. di/dtmax =
25 kA/µs, actual voltages 100 times higher.
A comparison between fig. 5 and fig. 6 shows, that if there
is no metal frame, the induced voltages in both loops are
much higher, as there is no compensating current circulating in the frame that reduces the magnetic field in loop 1
and 2 and therefore also the induced voltage. The stress of
the internal insulation is much higher for modules without
a frame. Measured maximum induced voltages depend on
distance from the lightning current (see fig. 7).
dc
Induced voltage in module KC60
∆V
caused by lightning current with di/dtmax = 25kA/µs (module parallel to current)
10000
+
Black full lines : Total voltage at module terminals
Gray dotted lines: Voltage in inner loop
Triangles + squares: Measured values
V2 CH2
-
dm1
i
Induced voltage in V
V1
CH1
KC60 without frame
1000
KC60 with metal frame
dm2
100
100
1000
Distance d from lightning current in mm
ground plate
Fig. 4: PV module KC60 mounted in parallel position,
both loops open circuit
(dm1 = 450mm, dm2 = 900mm)
Fig. 5 displays the induced voltages in a module without
frame, fig. 6 the induced voltages under the very same
conditions in a similar module with metal frame.
Fig. 7: Induced voltages in additive PV module KC 60
in parallel position with and without metal
frame for different distances d = dm1 from
simulated lightning current. All measured
values were normalised to di/dtmax = 25kA/µs.
If the module is relatively close to the lightning current,
even an overcompensation may occur (see fig. 6). In a
practical application this would mean that like with
17th European Photovoltaic Solar Energy Conference, Munich, Germany, Oct. 22 – Oct. 26, 2001
2.2 Induced Voltages in an additive Module
(Kyocera KC60) in normal Position
Similar tests were also carried out with the module in
normal position (long side of module normal to lightning
current) in the test layout shown in fig. 8. The resulting
induced voltages are indicated in fig. 9.
Frame-Reduction-Factor RF
for induced voltages in PV modules with metal frame
6
Frame-Reduction Factor RF
compensating modules only the voltage of one loop
appears at the terminals. In order to avoid unnecessary
bypass diode defects, the measurements shown were
carried out without bypass diodes (conducting diodes were
replaced by short-circuits).
5.5
5
KC60p
KC60n
4.5
MSX60p
4
MSX60n
SM46p
3.5
SM46n
3
2.5
0.1
1
dc
100
Fig. 10: Frame reduction factors RF for induced voltages measured at different modules. For greater
distances a mathematical model was used.
i
CH1
+
dm1
V1
∆V
-
V2
CH2
dm2
ground plate
Fig. 8: PV module KC60 mounted in normal position,
both loops open circuit
(dm1 = 450mm, dm2 = 1000mm)
Induced voltage in module KC60
caused by lightning current with di/dtmax = 25kA/µs (module normal to current)
10000
Black full lines : Total voltage at module terminals
Gray dotted lines: Voltage in one loop
Triangles + squares: Measured values
Induced voltage in V
10
Distance d from lightning current in m
KC60 without frame
1000
KC60 with metal frame
Similar tests performed with a SOLRIF module resulted in
comparable values of RF . The results of the tests with
single modules showed a considerable reduction of induced
voltages in modules with metal frames (typically by a
factor of 3 to 5) owing to a circulating current in the frame
which reduces the magnetic field in the module. Therefore
the use of modules with a metal frame is preferable in PV
plants exposed to lightning strokes. With compensating
modules (e.g. SM55 or MSX 64), under the same
conditions the induced voltage referred to module area is
up to a factor of 2 lower than the (worst-case) induced
voltage of additive modules (e.g. KC60) of the same area.
2.4 Influence of Aluminium Foil on Backside
In order to determine the influence of a metal foil on the
back side of a PV-module on induced voltages, some tests
were carried out with an additive module, the most common type of module. Tests were carried out with a KC60
without and with aluminium frame. To eliminate any differences between different module types, for this purpose an
additional metal foil was stuck right on the glass surface in
front of a module previously measured without foil. At the
module with frame, a distance of about 5 mm was
maintained to the edges of the metal frame in order to
avoid flashover. All tests were done with the module in
parallel position according to fig. 4.
Induced voltage in module KC60
caused by lightning current with di/dt max = 25kA/µs (module parallel to current)
10000
100
100
Black full lines : Total voltage without aluminium foil
Gray dotted lines: Total voltage with aluminium foil
Triangles + squares: Measured values
1000
Fig. 9: Induced voltages in additive PV module KC 60
in normal position with and without metal
frame for different distances d = dm1 from
simulated lightning current. All measured
values were normalised to di/dtmax = 25kA/µs.
Induced voltage in V
Distance d from lightning current in mm
KC60 without frame
1000
KC60 with metal frame
100
2.3 Frame Reduction Factor RF
Similar tests like those described in section 2.1 and 2.2
were also carried out with other types of modules
(MSX60/64, a compensating module with 4 cell rows, and
SM46, a compensating module with 3 cell rows).
In order to compare different modules with and without
frame, a frame reduction factor RF can be defined (ratio
between the maximum induced voltage without and with
frame). RF depends on the module type and on the distance
between the module to the lightning current (see fig. 10).
10
100
Distance d from lightning current in mm
1000
Fig. 11: Total induced voltages in additive PV module
KC 60 in parallel position with and without
metal frame and with and without an aluminium foil for different distances d = dm1 from
simulated lightning current. All measured
values were normalised to di/dtmax = 25kA/µs.
17th European Photovoltaic Solar Energy Conference, Munich, Germany, Oct. 22 – Oct. 26, 2001
With an aluminium foil, measured induced voltages were
between a factor of 7 to 10 lower than without foil under
the same conditions. An aluminium foil is therefore an
effective means to reduce the voltage induced by nearby
lightning currents considerably. Earlier experiences a few
years ago have shown that aluminium foils may make it
difficult to achieve a sufficiently high impulse voltage
withstand capability (required for instance for protection
class II modules). However, with a sufficient backside
insulation and a distance of a few mm to a metal frame, this
problem could probably be solved.
2.5 Bypass Diodes
A weak point discovered in many modules were the bypass
diodes, which were often destroyed by excessive voltage or
current transients. After excessive reverse voltage stresses
these diodes mostly were short circuited. This may be very
dangerous in PV arrays with many strings in parallel without any overcurrent protection in the strings (as proposed in
some drafts for standards). In PV arrays that are designed
to survive nearby lightning strokes, bypass diodes with a
reverse voltage rating of 1000V or even higher should be
used. In addition in large PV arrays with many strings in
parallel overcurrent protection devices in the strings should
be used in this case.
ning protection system (e.g. to avoid flashover due to
insufficient distance), the best method is to use a shielded
DC main line, whose shield is connected to the metallic
frames on the PV generator side and grounded on the other
side, with surge arresters from + and – to ground on both
sides of the cable. A simpler, but less efficient method is
the use of a ground conductor of sufficient cross section
very close to the DC main line, but in this case a significant
part of the lightning current will flow through the surge
arresters. Unfortunately is very difficult to find surge
arresters which are capable of handling significant parts of
lightning currents on the market.
5 Conclusion
Due to space limitations, only a short overview could be
given in this paper. An extended series, dealing thoroughly
with the problem of lightning protection of PV systems and
giving much more details about the tests performed, was
published in [6].
IMPORTANT NOTICE
Information contained in this paper is believed to be accurate. However, errors can never be completely excluded.
Therefore any liability in a legal sense for correctness and
completeness of the information or from any damage that
might result from its use is formally disclaimed.
2.6 Influence of framed Neighbour Modules
ACKNOWLEDGEMENTS
Under otherwise identical conditions, voltages induced in a
certain module are influenced by the presence of neighbouring modules with metallic frames, in which a circulating current can be induced. If there are neighbouring
modules closer to the lightning current, the induced voltage
in the module is increased. In one case, measured induced
voltage nearly doubled due to a similar module closer to
the lightning current. On the other hand, neighbouring
modules at greater distances reduce induced voltages.
Initial experiments about sensitivity of PV modules against
lightning currents were carried out already in 1990-1993 by
several students working towards their diploma thesis.
Most of the work described in this paper was done during
the EU project PV-EMI, JOR3 CT98 0217 (Partners:
FhG/ISE (BRD), HTA Burgdorf (CH) and KEMA (NL)).
This project was funded by Bundesamt für Bildung und
Wissenschaft (BBW) in Berne. My thanks go to all persons
involved in these projects for their valuable contributions.
Special thanks go to my former colleague, Dr. R. Minkner,
who participated not only in the early work with students,
but also in the EU project, and to Mr. R. Fischer, who was
project assistant at HTA Burgdorf during this EU project.
3
Induced Voltages in PV Arrays
Despite the non-linear characteristic of solar cells, for
maximum values of induced voltages occurring in the beginning of lightning currents, the superposition law proved
to be still valid. Therefore maximum voltages in wired PV
arrays can be calculated by adding the voltages induced in
the modules of a string and the voltage induced in the
wiring. Low area of the wiring loop can keep this wiring
voltage low. This can not always be realised in practical
PV generators. Tests showed that metallic frames of modules reduce the voltage induced in the wiring by a similar
frame reduction factor RF (see fig. 10) like the voltages
induced in the modules, if the wiring is in (and within the
boundaries of) the module plane.
4
Methods of Grounding and Lightning
Protection of PV Arrays
Extended experiments were also carried out with models of
PV arrays and different kinds of DC main cables to
determine the need and the optimum method of grounding
and the proper use of surge arresters.
To protect a PV generator against direct lightning strokes,
the best possibility is to place it into the protected area of a
lightning protection system according to EN61024. If
modules with metallic frames or a metallic support
structure are used, these elements should be grounded by a
ground conductor placed as close to the DC main line as
possible. If there is an additional connection to the light-
REFERENCES
[1]
H. Häberlin and R. Minkner: "Tests of Lightning Withstand
Capability and Measurements of Induced Voltages at a
Model of a PV-System with ZnO-Surge-Arresters". Proc.
11th EC-PV-conf., Montreux, 1992.
[2] H. Häberlin and R. Minkner: "A simple Method for
Lightning Protection of PV-Systems". Proc. 12th EU PV
Conference, Amsterdam 1994.
[3] S.A.M. Verhoeven: "Short-circuit behaviour and lightning
induced voltages in solar modules".
Proc. 12th EU PV Conference, Amsterdam 1994.
[4] S.A.M. Verhoeven, J.A.J Pettinga: "Lightning induced
voltages in a solar array to improve the grounding structure".
Proc. 13th EU PV Conference, Nice 1995.
[5] European Commission: "Lightning and Overvoltage Protection in Photovoltaic and Solar Thermal Systems", General
Information, Thermie-B Programme Action No. SME-166298-DE. Authors: H. Becker, W. Vaassen, F. Vassen, M.
Bosanac and I. Katic. TÜV Rheinland, D-51105 Köln, 2000.
[6] H. Häberlin: "Blitzschutz von Photovoltaikanlagen".
Elektrotechnik 4/2001 bis 10/2001.
(extended series of 6 parts in German)
[7] Publishable Final Report: „Development of standard test
procedures for electromagnetic interference (EMI) tests and
evaluations on photovoltaic components and plants (PV-EMI
Project)“. Contract JOR3 CT98 0217, Aug. 2000.
Further information about the research activities of HTA Burgdorf’s PV laboratory on the internet: http://www.pvtest.ch