* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download IEEE white paper
Electrical ballast wikipedia , lookup
Ground (electricity) wikipedia , lookup
Solar micro-inverter wikipedia , lookup
Electric power system wikipedia , lookup
Power engineering wikipedia , lookup
Mercury-arc valve wikipedia , lookup
Stepper motor wikipedia , lookup
History of electric power transmission wikipedia , lookup
Voltage optimisation wikipedia , lookup
Three-phase electric power wikipedia , lookup
Resistive opto-isolator wikipedia , lookup
Power MOSFET wikipedia , lookup
Current source wikipedia , lookup
Stray voltage wikipedia , lookup
Opto-isolator wikipedia , lookup
Electrical substation wikipedia , lookup
Power inverter wikipedia , lookup
Skin effect wikipedia , lookup
Switched-mode power supply wikipedia , lookup
Distribution management system wikipedia , lookup
Power electronics wikipedia , lookup
Variable-frequency drive wikipedia , lookup
Buck converter wikipedia , lookup
Mains electricity wikipedia , lookup
Surge protector wikipedia , lookup
Earthing system wikipedia , lookup
Electrical wiring in the United Kingdom wikipedia , lookup
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: [email protected] 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: [email protected] 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 IGBT’s. 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 0 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 0 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 9 600 6 400 IGBT energy 3 200 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 8 time [us] 9 600 6 400 2 800 4 2 IGBT energy 4 x 10 without fuse 6 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. 0 2 0 with fuse 0 50 200 3 0 0 400 800 0 100 12 350 10 0 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 20°C. 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. 0 200 40 400 Fig. 4. Diagram of the IGBT short-circuit test facility with indication of the measured values. 0 150 50 600 200 -100 100 VCE 800 Ic 300 50 1000 IC [kA] C 400 A IGBT module, 200 A fuse 300 Energy [J] Lstray 2 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 IGBT’s. 50 100 150 200 250 300 time [us] 350 400 450 0 500 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 2 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 1 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 V C1 Trigger circuit +15V IGBT module IGBT RG LEM LEM U3 fuse V 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 1 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 . 3 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 C 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 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 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 3 4 5 6 7 8 9 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 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5 0 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 2 3 4 strip 4 5 6 7 8 9 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;