Sunteți pe pagina 1din 9

Presented at 2001 PCIM Conference

Fusing IGBT-based Inverters


F. Iov, F. Abrahamsen, F. Blaabjerg,
Aalborg University,
Institute of Energy Technology
Pontopidanstraede 101
DK-9220 Aalborg East, Denmark
Tel.: +45 96 35 92 54, Fax.: + 45 98 15 14 11
e-mail: fbl@iet.auc.dk
Abstract. The number of applications based on
the dc-link voltage source converter is in a
continuous growth. Due to the fact that the power
level increases more energy is stored in the dclink. Even with an active protection, a high power
IGBT still has a risk of exhibiting a violent rupture
in the case of a fault. In order to avoid the
rupture a possible solution is to use a fuse
protection of the converter. From failure analysis
of a three-phase inverter, a dc-link location for
fuses might assure a protection against all faults.
Experiments made with 400 A, 1200 V IGBT
switches show that rupture can be avoided by
the use of high-speed fuse protection. The
problem of added inductance in existing HighSpeed Fuses is discussed, and new IGBT
Typower fuses in the dc-link circuit are treated.
Finally, some considerations about current
distribution in dc-link fuses and spectral analysis
of these currents are discussed when highfrequency components are present in the
current.
Introduction
The dc-link voltage source converter topology
covers currently a wide range of power
applications, with variable speed drives and UPS
as the main ones. Due to the significant
improvement in IGBT technology, the voltage
and current ratings have increased and IGBTbased high power applications already exist,
both parallel and series connected. A drawback
with the higher power is that the dc-link
capacitors store more energy. This causes a
higher risk of IGBT case rupturing when a circuit
failure condition occurs [1], [2]. The rupture can
be extremely destructive in some situations, and
it brings up certain problems such as damage of
converter, drive down-time, personal injury,
troubles with drive certifications only few have
been mentioned here. Most failures are caught
by an active over-current protection, which turns
off the switches when a fault is detected.
However, there are cases where the active
protection is not sufficient and where the

K. Ries, H. Rasmussen, P. Bjornaa


Cooper Bussmann International Inc.
Literbuen 5
DK-2740 Skovlunde, Denmark
Tel.: + 45 44 85 09 67, Fax.: + 45 44 85 09 07
e-mail: kr@bussmann.dk
consequences may be catastrophic for the
converter and its surroundings.
A solution to this problem could be a fuse
protection, which will not prevent the IGBT
destruction but it will prevent a case rupture.
There may naturally be an unwillingness to use
fuse protection in IGBT converters because it
takes up extra space, it is an extra cost, it
introduces extra losses and finally, a fuse will
add some extra stray inductance in the circuit.
These
drawbacks
may,
however,
be
compensated by the advantages such as no
rupture, reduced problems with certification and
no need for a special explosion chamber in the
design and thereby reduced manufacturing cost.
This paper will first present an overview
concerning the failures in a dc-link inverter and
possible location of the fuses. Then, a study of
the rupture phenomena and how a fuse protects
an IGBT is shown. Next, some experimental
measurements regarding the value of added
inductance by fuses for different types of busbars will be presented. Finally, based on
experimental results the high-frequency current
distribution and spectral analysis within dc-link
fuses will be presented.
Failure analysis in dc-link inverters
A typical three-phase voltage source dc-link
inverter is shown in Fig. 1
Inverter

C1
Z
Z
Z

C2

Load

Fig. 1. Three-phase inverter with voltage source dclink and IGBTs.

The rectifier side is high-inductive whereas the


inverter side is very low-inductive. The capacitor

bank typically consists of several individual


capacitors in series and parallel, and the same
can be true for the IGBT-modules, although only
paralleling is relevant because the low voltage
level is considered here. It is common practice to
connect the mid-point of the capacitor bank to
ground.
Several incidents can cause a huge over-current
in the IGBT devices, and different cases are
indicated in Fig. 2.

phase-to-phase faults is that the short circuit


current only appears in one of the dc-link rails.
Possible location of Fuses in IGBT Inverters
Some possible locations of the fuses in a voltage
source inverter [1], [2] are shown in Fig. 3. Only
those solutions are presented which have fuses
both in the upper and in the lower part of the
inverter, because it is the only way to give a full
protection of the inverter.

C1

C1

C1

C2

C2

C2

a)
C1

a)
C1

C2

C2

c)

b)

C1

C1

C2

C2

d)

b)

C1

C1

C2

C2

c)

d)

Fig. 3. Possible fuse configurations in a voltage source


inverter: a) dc-link, b) in series with the
capacitors, c) in series with the IGBT-modules, d)
modular paralleling of half-bridges to increase
the converter power rating.
e)

Fig. 2. Fault cases in a voltage source inverter: a) dclink shoot-through, b) two phases with no
inductance, c) two phases with inductance, d)
positive rail earth fault with a very small
inductance, e) negative rail earth fault with
inductance.

The discussion is mainly focused on drives


application but it could easily be other
applications too. The first is the dc-link shoot
through as shown in Fig. 2a. This can happen in
the case of a fault in the control circuit or an
interior fault in an IGBT-module. A dangerous
case happens if a fault is detected by the overcurrent protection and the converter is put into
operation again because the user subsequently
has ignored the error message.
The next cases in Fig. 2b and Fig. 2c are shortcircuits between two phases. A reason could be
failure of the insulation in the motor windings.
The two cases are principally similar, but the
amount of inductance in the current path
determines the current rise-time and how the
fault evolves. The last cases shown in Fig. 2d
and Fig. 2e are earth faults. These could also be
caused by e.g. bad motor winding isolation. The
difference between the earth faults and the

It may seem most obvious to place two fuses in


the dc-link (a) as they can protect against all
faults. However, placement in series with the
capacitor bank (b) provides in principle the same
protection. The advantage here is that fuses in
series with the capacitor do not carry the active
power producing current but only ac-current. The
solution in (c), where one fuse is used in series
with each IGBT seems at first sight most
expensive, but the current in each of the six
fuses is smaller than in the dc-link fuses. There
is no fixed ratio between the fuse currents for
case a) and c) because the dc-link mainly carries
active power whereas a fuse in series with an
IGBT carries both active and reactive power to
the load.
Short-circuit test
Tests with short-circuit of an IGBT have been
performed in order to study the rupture
phenomena and how a fuse protects an IGBT. A
diagram of the basic test facility is shown in Fig.
4. The tests are made with a 1200 V, 400 A
standard single switch IGBT-module and a 660
V, 200 A Bussmann High-Speed Fuse. The tests
are performed as a simple short-circuit of an
isolated capacitor bank through the IGBT. The
dc-link capacitance is 4.7 mF and the parasitic

fuse

VCE

VGE

200
100
0

Vfuse
+
VCE

-100

50

100

150

200

250

VCE [V]
300

350

400

450

500

VGE [V]

time [us]
50

VCE [V]

Ic

VCE

40

600

30

400

20

200

10
50

100

150

200

250

300

350

400

450

IC [kA]

1000

Vfuse [V]

500

30

50

100

150

200

250

300

350

400

450 500

50

100

150

200

250

300

350

400

450 500

time [us]

600
400
200
0

time [us]
Power [MW]

15

1000

12

800

IGBT power

600

400

IGBT energy

200
0

50

100

150

200

250

300

350

time [us]

400

450

500 0

The conclusion of the short-circuit tests is that a


fuse can prevent a rupture of an IGBT switch and
furthermore makes it much easier to protect the
driver circuits from being destroyed by overvoltage. For this example with a 400 A single
switch IGBT and a 200 A fuse there is a security
factor of 550/200 = 2.75 against rupture in terms
2
of energy, and 6.5/1.5 = 4.3 in terms of I t, see
Fig. 7.

0
500

time [us]

600

400

800

IGBT
energy

4
x 10

without fuse

I t [A s]

IGBT
power

Energy [J]

1000

15

Power [MW]

450

Fig. 6. Short circuit test of an 1200 V, 400 A IGBT


switch with a 200 A high-speed fuse.

2
0

with fuse
0

50

200

3
0 0

400

800

100

12

350

10
0

400 A IGBT module

200

0 0

300

20

1000

800

250

time [us]

Ic

The normal dc-link voltage used with 1200 V


components is around 560-700 V, but the test is
performed at 800 V which is considered as a
worst case that occurs in the case of e.g. a fast
breaking of a motor. The tests are carried out at
20C. The current is measured with a PEM CWT
1500 Rogowski coil transducer with a bandwidth
of 7.5 MHz, having an inserted inductance of a
few pH. The voltages are measured with normal
voltage probes by an oscilloscope. The sampling
rate is 50 MHz.

200

40

400

Fig. 4. Diagram of the IGBT short-circuit test facility


with indication of the measured values.

150

50

600

200

-100

100

VCE

800

Ic

300

50

1000

IC [kA]

400 A IGBT module, 200 A fuse

300

Energy [J]

Lstray

I t-values in the two cases are compared in


Fig.7.

VGE [V]

inductance of the circuit without the fuse is 30


nH. Both values correspond well to what is used
in practical converters together with 400 A
IGBTs.

50

100

150

200

250

300

time [us]

350

400

450

500

Fig. 5. Short circuit test of an 1200 V, 400 A IGBT


switch without a fuse.

The results for the direct IGBT short-circuit test


are shown in Fig. 5. and the results for the test
with an inserted fuse are shown on Fig. 6. The

100

150

200

250

300

time [us]

350

400

450

500

Fig. 7. I t-values for tests with and without a fuse.

Added inductance by fuses in dc-link circuit


Another major aspect is the inductance value
added in the dc-link circuit by using fuses. A test
setup for measuring the circuit inductance has
been developed. The test setup consists of a dc

IGBT fuse

considered during turn-off. So, the voltage peak


at 0.8 ms is entirely due to (L*di/dt) in the circuit.
350
300

Bus-bar current [A]

capacitor, a bus-bar with a positive and negative


rail, and an IGBT half-bridge. The setup is shown
in Fig. 8. The bus-bar is changeable, and a fuse
can be mounted onto it. This makes it easy to
test different fuses and different ways of
mounting them on the bus-bar. The copper bars
are 2 mm thick and there is a 0.2 mm insulation
between the positive and negative rails.

evaluated points

250
200
150

100
50
0
-50
0.6

0.65

0.7

0.75

0.8

time [us]

0.85

0.9

0.95

Fig. 10. Measured current during turn-off. The dots


indicate all the points in which di/dt is evaluated.
Capacitance

500

Changeable fixture

U2

400

Fig. 8. Test setup for measurement of the added


inductance using a fuse.

An electrical diagram of the test setup is shown


in Fig. 9

Voltages [V]

IGBT module

U1

300
200
100
0

calculated bus-bar voltage


U1

D1

U2
V

L1

C1

Trigger
circuit
+15V

IGBT
module

IGBT
RG
LEM

LEM
U3

fuse

U4
V

Main circuit

-15V
Gate-drive circuit

Fig. 9. Electrical diagrams of the test-circuit for


measuring added inductance using a fuse.

The operating principle of the circuit is as follows:


The capacitor C1 is initially charged to a preset
voltage (e.g. 130 V). A very short pulse is applied
on the IGBT's gate (about 60 ms). Then the IGBT
is turned off very rapidly. The fuse inductance is
calculated based on measured voltages and
currents from the circuit during turn-off. Varying
the gate-resistance RG as shown in Fig. 9 the
current gradient di/dt during turn off can be
controlled.
An example of a measured current during turnoff is shown in Fig. 10. The corresponding
voltages are shown in Fig. 11. Two measured
voltages are shown as well as the calculated
difference between them, which is the bus-bar
voltage drop. It is seen that initially, when the
di/dt is zero, the voltage drop is only a few volts.
This means that the resistive voltage drop can be
ignored when the large inductive voltage drop is

-100

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95

time [us]

Fig. 11. Measured capacitor and IGBT voltages, and


the calculated difference between them.

The current gradient can be calculated based on


the experimental data shown in Fig. 10. The busbar inductance is calculated by evaluating the
di/dt in a number of points around the voltage
peak. A simple division of the bus-bar voltage
with di/dt gives the circuit inductance.
The test-setup has been realized in the
laboratory with maximum focus on flexibility.
Thus, it can be changed and tested with different
sizes of bus-bars and fuses.
The inductance for three types of reference busbars: 50 mm, 70 mm and 120 mm width have
been measured without a fuse. Fig. 12 shows the
50 mm bus-bar used during the test. The results
are summarized in table 1.

Fig. 12. 50 mm reference bus-bar.

Table 1. Inductance values for the reference


bus-bars at di/dt = 4.3 kA/ms
Bus-bar width
Bus-bar inductance
[mm]
[nH]
50
21
70
20
120
20

Next, the added inductance for different types of


fuses, have been measured for the considered
bus-bars. Fig. 13 and Fig. 14 shows two types of
Typower IGBT fuses .

From the above measurement results it can be


concluded that by introducing Typower IGBT
Fuses into the dc-link the level of inductance will
increase with only very small values. The
Standard High-Speed Fuses mounted on the top
of the bus-bar will add around 20 nH of
inductance to the inverter circuit (see Table 3),
whereas by introducing a Typower IGBT Fuse
the added inductance can be reduced to around
12 nH (see Table 2). It is believed that if the
distance between the fuse / fuse element and the
return conductor can be reduced; the value of
the added inductance will be even smaller.
Variation on the width of the bus-bar seems to
have only negligible effect on the inductance
(see Table 1).
Current distribution in fuse

Fig. 13. Typower IGBT Fuse, no. 2828.

Fig. 14. Typower IGBT Fuse, no. 2805.

In Table 2 are presented the values of added


inductance for Typower IGBT Fuses mounted on
70 mm bus-bar.
Table 2. Added inductance for Typower IGBT
Fuses mounted on 70 mm bus-bar
Fus
Rated
Rated
Added inductance
e
current [A]
voltage
DL [nH]
no.
[Vdc]
280
350
800
12
5
350
800
12
282
8

Table 3 shows the added inductance for


Standard High-Speed Fuses mounted on 50 mm
bus-bar.
Table 3. Added inductance for Standard High
Speed Fuses mounted on 50 mm
bus-bar
Fus
Rated
Rated
Added
e
current
voltage
inductance
no.
[A]
[Vdc]
DL [nH]
280
180
800
19
6
380
800
21
280

The current distribution in a fuse placed in the


dc-link circuit of a three-phase voltage source
inverter is affected by eddy currents as a result
of time variation of the electromagnetic field [6],
[7]. The high-frequency currents generate two
phenomena in conductors: skin effect and
proximity effect. Due to the skin effect the
current, which flow through the conductor, is
pushed towards the surface. When the current is
divided among a group of parallel conductors the
sharing of the total current between them is
generally unequal. As this phenomenon is
dependent on the distance between conductors,
it is called direct proximity effect. Furthermore, if
a current carrying conductor is brought near the
parallel conductors (e.g. return bus-bar), the
distribution of the current is affected. This
phenomenon is called inverse proximity effect
since opposite currents flow in the return
conductor and the group of parallel conductors.
A test setup for studying current distribution in
fuse-strips has been developed. The test setup
consists of an auto-transformer (AT), a threephase rectifier bridge (RB), a dc capacitor bank,
a bus-bar with a positive and negative rail, an
IGBT half-bridge and an inductive load (L = 0.47
mH) as shown in Fig. 15.

Fig. 17. Open model of a Typower IGBT Fuse used for


tests of current distribution.
D1

T1

RB

R-L

Load
D2

T2

Fig. 15. Electrical diagram of the test setup for load


test of fuses.

The bus-bar is changeable, and a fuse can be


mounted onto it. This makes it easy to test
different fuses and different ways of mounting
them on the bus-bar. The copper bars are 2 mm
thick and there is 0.2 mm insulation between the
positive and negative rails.
The current distribution through different strips of
the fuse can be measured with a Rogowski
Current Transducer as shown in Fig. 16. During
the tests of current distribution, two types of
fuses have been used. The first one is a
Standard High-Speed Fuse, which has four
parallel strips placed vertical, 16 mm width, 70
mm length and 0.1 mm thickness, as shown in
Fig. 16. The distance between bus-bar / strip and
strips is 10 mm.
strip 1
Rogowski coil

strip 2

The measurements have been performed, for


both considered fuses, with a square wave
shape of the dc-link current at different switching
frequencies and at constant current, IDC = 100 A.
The RMS values of the currents in each strip of
the fuse were measured with the Rogowski
current transducer connected to a LeCroy
oscilloscope.
Fig. 18 shows the dc-link current waveform for
both considered fuses at 5 kHz switching
frequency.

200
150
100
50
0
-50
0

The second one is a model of a Typower IGBT


Fuse, which has five parallel strips placed
horizontal, 7 mm width, 50 mm length and 0.2
mm thickness as shown in Fig. 17. The distance
between bus-bar and strips is 17 mm, and there
is 5 mm between the strips.
strip 1
strip 2
strip 3
strip 4
strip 5

0.2

0.3

Time [ms]

0.4

0.5

The current waveform for each strip of the open


Standard High-Speed Fuse is shown in Fig. 19
150
100
50
0
-50

Strip Currents for


Standard High-Speed Open Fuse

0.1

0.2

0.3

0.4

0.5

0
150
100
50
0
-50

0.1

0.2

0.3

0.4

0.5

0
150
100
50
0
-50

0.1

0.2

0.3

0.4

0.5

0.1

0.2

0.3

0.4

0.5

Istrip 3 [A]

Istrip 2 [A]

150
100
50
0
-50

Istrip 4 [A]

Fig. 16. Open Standard High-Speed Fuse used for


tests of current distribution.

0.1

Fig. 18. DC-link and thereby fuse current waveform.

strip 3
strip 4

DC-link Current waveform

250

Idc [A]

AT

Istrip 1 [A]

Test fuse

Time [ms]

Fig. 19. Currents waveform in each strip for the


Standard High-Speed Open Fuse with 5 kHz
switching frequency.

Because each strip has different value of mutual


inductance, big differences between the current
waveform and RMS current in the strips can be
observed. The current distribution in each strip of
this type of fuse, for the tested frequencies (fsw =
2.7-14 kHz), is summarized in Fig. 20.

0.6

0.55
0.5

strip 4

Istrip k / Idc

0.45
0.4
0.35
0.3

strip 3

0.25

strip 2

0.2
0.15
0.1
0.05

strip 1
2

10 11 12 13 14 15

Switching Frequency [kHz]

Fig. 20. Current distribution in Standard High-Speed


Open Fuse for different switching frequencies.

The current waveform for each strip of the


Typower IGBT Fuse is shown in Fig. 21.
Strip Currents in
open model of Typower IGBT Fuse

Istrip 1 [A]

60
40
20
0
-20

0.1

0.2

0.3

0.4

0.5

0.1

0.2

0.3

0.4

0.5

0.1

0.2

0.3

0.4

0.5

0.1

0.2

0.3

0.4

0.5

0.1

0.2

0.3

0.4

0.5

Istrip 2 [A]

60
40
20
0
-20

Istrip 3 [A]

60
40
20
0
-20

Istrip 4 [A]

60
40
20
0
-20

Istrip 5 [A]

60
40
20
0
-20

Time [ms]

Fig. 21. Currents waveform in each strip of Typower


IGBT Fuse with 5 kHz switching frequency.

The current distribution in each strip, for the


same switching frequencies as in precedent
case, is summarized in Fig.22.
Current Distribution in open model of Typower IGBT Fuse
0.25

Fig. 22. Current distribution in Typower IGBT Fuse for


considered switching frequencies.

Due to the inverse proximity effect the current


distribution in a Standard High-Speed Fuse is
affected even for lower frequencies and the strip
located near the bus-bar carry a significant
amount of the total current 40-50 %, whereas the
remote strip carries just 10-20 % of it. For a
Typower IGBT Fuse the current is distributed
almost equal between the strips. The both
sideways strips are loaded with a higher current
than the other ones, but the difference is only
around 5%. Result: the low-inductive IGBT-fuse
has superior capability under high frequencies.
Spectral Analysis of Currents
Based on the dc-link current and the strip
currents for both considered open fuses a
spectral analysis has been performed. The
harmonic currents magnitude are related to the
dc component of the dc-link current spectra as
follows:
Harmonic spectra of the dc-link current:
Ih dc / I dc , where: Idc is the dc component of

the dc-link current spectra and Ih dc is the


magnitude of the harmonic components.
Harmonic spectra of strips currents:
Ih strip k / I dc , where Ih strip k is the magnitude

th

of
the
harmonic
currents
for
k
strip, k = 1,..., n .
Fig. 23 shows the harmonic spectra of the dc-link
current and Fig. 24 the harmonic spectra for
each strip current in the open Standard HighSpeed Fuse.
1

Harmonic Spectra of DC-link Current

0.9

Ih dc / Idc

Current Distribution in Standard High-Speed Open Fuse

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120130 140150

Frequency [kHz]
strip 1

Fig. 23. Harmonic spectra of the dc-link current with 5


kHz switching frequency.

Istrip k / Idc

0.225

strip 5

0.2

strip 2

0.175

0.15

strip 3

strip 4
5

10 11 12 13 14 15

Switching Frequency [kHz]

strip 1

0.2
0.1
0

0 10 20 30 40 50 60 70 80 90 100 110120 130 140150

0.3

0.1

Ih strip 3 / Idc

0
0.3

0.3

0 10 20 30 40 50 60 70 80 90 100 110120 130 140150

strip 3

0.2
0.1
0

0 10 20 30 40 50 60 70 80 90 100 110120 130 140150

strip 4

0.2
0.1
0

0 10 20 30 40 50 60 70 80 90 100 110120 130 140150

Frequency [kHz]

Fig. 24. Harmonic spectra of the strip currents in open


Standard High-Speed Fuse.

Ih strip 5 / Idc

Ih strip 4 / Idc

Ih strip 3 / Idc

Ih strip 2 / Idc

Ih strip 1 / Idc

The harmonic spectra of the strip currents, for


open model of Typower IGBT Fuse, are shown in
Fig. 25.
0.2
0.15

strip 1

0.1
0.05
0

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150

0.2
0.15

strip 2

0.1
0.05
0

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150

0.2
0.15

strip 3

0.1
0.05
0

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150

0.2
0.15
0.05
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150

0.2
0.15

strip 5

0.1
0.05
0

The spectral analysis of the strip currents offers


a foundation in studying the influence of skin
effect on the current distribution in each strip.
Conclusion

It is explained how an IGBT-based inverter circuit


can be protected against IGBT case rupture by
introducing fuses in the circuit. Different types of
faults and possible protection methods are
discussed. Furthermore, this paper shows that
by introducing Typower IGBT Fuses instead of
Standard High Speed Fuses in the dc-link the
level of inductance will increase with only very
small values. A method to measure realistic
values of added inductance is given and values
for different fuse types are published.
It is shown how inverse proximity effect affects
the current distribution in a Standard High-Speed
Fuse and a Typower IGBT Fuse.
Proximity effect has a great impact on the current
distribution among the fuse strips. Some of the
strips may carry more than the normal share of
the rated current. As a consequence the fuse
rated current must be de-rated accordingly. As
this effect depend on the distances between busbar and strips, and switching frequency, it can be
minimized only by modifying the fuse geometry.
From the test shown in this paper it is obvious
that using a fuse with paralleled strips placed
horizontally and small distance between them
and the return conductor (a Typower IGBT fuse),
the value of the added inductance decreases
and the current distribution is only slightly
affected by the inverse proximity effect.

strip 4

0.1
0

th

tested fuses only 1 to 9 odd harmonics have a


significant magnitude, the rest can be neglected.

strip 2

0.2

Ih strip 4 / Idc

Ih strip 2 / Idc

Ih strip 1 / Idc

st

0.3

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150

Frequency [kHz]

Fig. 25. Harmonic spectra of the strip currents for open


model of Typower IGBT Fuse.

From the above investigation it can be concluded


that, in the harmonic spectra only odd harmonics
order have to be taken into account. The even
harmonics have a small magnitude because of
the dc-link current shape (see Fig. 23). For both

References

[1]. F. Abrahamsen, C. Klumpner, F.


Blaabjerg, K. Ries, H. Rasmussen - Lowinductive
fuses
in
Dc-link
Inverter
Applications. Proceed. of PCIM '2000, pp.
523-528;
[2]. F. Abrahamsen, C. Klumpner, F.
Blaabjerg, K. Ries, H. Rasmussen - Fuse
protection of IGBT's against rupture,
Proceed. of NORPIE 2000, pp.64-68;
[3]. J. F. de Palma, B. Turse - IGBT
Protection: Fast Fuses can Protect IGBT's,
PCIM Magazine, vol. 24, no. 10, Oct. 1998,
pp. 64 - 71;

[4]. D. Braun, D. Pixler, P. LeMay - IGBT


Module Rupture Categorization and Testing,
Proceed. of IAS ' 97, New Orleans, Oct.
1997, pp. 1259-1266;
[5]. S. Gekenidis, E. Ramezani, H. Zeller, Explosion Tests on IGBT High Voltage
Modules, Proceed. of Power Semiconductor
Devices and ICs, ISPSD ' 99, pp. 129-132;
[6]. S. Duong, C. Schaeffer, R. Deshayes, J.L.
Gelet - Distribution of High-Frequency
Currents through the Elements of a Fuse,
IEE Proc, vol. 129, no. 3, May 1982, pp.229235;
[7]. A.F. Howe, C.M. Jordan - Derating of
semiconductor fuselinks for use in Highfrequency Application, IEE Proc, vol.129,
May 1982, pp.111-116;

S-ar putea să vă placă și