Comparison of Planar- and Trench--Modules for resonant applications M. Helsper Christian-Albrechts-University of Kiel Faculty of Engineering Power Electronics and Electrical Drives Kaiserstr. 2 24143 Kiel Germany Tel. +49 431 88 615 Fax. +49 431 88 613 mth@tf.uni-kiel.de F. W. Fuchs Christian-Albrechts-University of Kiel Faculty of Engineering Power Electronics and Electrical Drives Kaiserstr. 2 24143 Kiel Germany Tel. +49 431 88 61 Fax. +49 431 88 613 fwf@tf.uni-kiel.de M. Münzer EUPEC GmbH + Co. KG Max Planck Str. 5 59581 Warstein Germany Tel. +49 292 764 1195 Fax. +49 3 74 Mark.Muenzer@eupec.com Abstract: For the qualification in resonant converters in the Zero Voltage Mode at high frequencies a Standard- -Module, a Fast--Module, both with s of the 2. generation with a planar gate structure and a new Trench-Gate--Module of the 3. generation are characterized and compared. Investigations show that the Trench--Module can be driven at high frequencies in a low loss manner like the Fast--Module which is optimized this mode. Indeed for the Trench- these advantage exists at a mode at high voltages and relatively low switch off currents. At high switch off currents and high frequencies the Fast--module has the best performance. At lower and medium frequencies the new Trench- works with the lowest losses. I. Introduction In resonant applications like microwave, arc welding or battery chargers -modules are increasingly used according to their good static and dynamic performance [1,2,]. The modules work in these applications usually at frequencies higher than 2 khz. Maximum frequencies between 1 khz and 2 khz are possible [3] in soft switching topologies. In this operation besides a good static ability especially the switching behaviour of the -modules is very important. To evaluate the properties of -modules for soft switching applications the datasheets give not enough information. Thus numerous publications [4, 5, 6, 7] have dealt with this topic. This paper presents an investigation and comparison of three -Modules from one manufacturer in terms of their abilities for the application in resonant inverters which work in the Zero Voltage Mode at high frequencies. W off [mws] 15 1 5 Trench standard fast 1 2 3 4 V CEsat [V] Figure 1: Trade off between saturation voltage V CEsat and switch-off losses W off for the investigated s at hard switching, cond.: I C =75 A, V CC =6V, T J =125 C
This is at first a planar Standard--Module, type BSM75GD12DN2, of the 2. generation, optimized for hard switching applications with low and medium switching frequencies. Second a planar Fast--Module of the 2. generation, type FS75R12KS4, specialized for high frequency applications is tested. Third a new Trench--Module of the 3. generation, type FS75R12KE3, with a trench gate structure and a field stop zone [8] is measured. This is first of all provided for applications in hard switching converters with lower and medium frequencies. Figure 1 shows in which manner the necessary trade off between saturation voltages and the switching losses is realized for the modules at hard switching. The general advantage of the Trench- is clearly to be observed here. An important question of this investigation is to find out if the Trench- is well suited for ZVS applications with high frequencies also without special optimization. V GE I C V CE P V Figure 3: Behaviour of an Trench Module in the ZVS-Mode, 1µs/DIV, Ch. 2: V GE 1 V/DIV, Ch. 3: I C, I D 4 A/DIV, Ch. A: V CE = 2 V/DIV, Ch. C: P V 1 kw/div II. ZVS test circuit For the investigations a test circuit, see figure 2, has been built which operates under application specific conditions. It has the topology of a voltage source series loaded resonant dc-to-ac converter [9]. If the upper S 1 switches off the diode D 2 takes over the load current for a short time. The S 2 goes in a switch on standby to this time. At the zero crossing of the load current the S 2, switches on passively at nearly zero voltage and takes over the load current. Before the end of this half sinus wave the S 2 is soft switched off. In practice the switch off is performed usually at low load currents to limit the losses, moreover snubber capacities C S are used to reduce the switching losses. = = V CC I L V CC R L L L C L V GE Figure 2: Principle test circuit S 1 D 1 C S S 2 D 2 V CE C S To realize the zero voltage switching mode of the -modules the resonant circuit must be driven at frequencies higher than the resonant frequency of the load. Figure 3 shows the typical behaviour of the S 2 and its anti-parallel diode D 2 in this mode. I C To guaranty the test at a defined chip temperature the test mode is limited to only 4 periods. By means of mounting on a controlled heat plate a control of the chip temperature T J is possible. All -modules are tested under the same test conditions to ensure a good comparability. III. ZVS switch off In the ZVS-mode the switch off of the has a superior importance. Figure 4 shows the active switch off of the Trench- in this mode.
I CS2 V GE S2 P VS2 W S2 V CES2 Woff [mws] 8 6 4 2 2 3 4 5 6 7 8 I c off [A] 25 C 6V 13.6nF Fast 25 C 6V 13.6nF Trench 25 C 6V 13.6nF Trench 125 C 6V 13.6nF Fast 125 C 6V 13.6nF Figure 6: Switch off losses of the s via I C off At reduced current at switch off it can be seen that at V CC =6V the losses of the Trench- converge to that of the Fast-. Figure 4: Soft switch off of a Trench- cond.: V CC =6V, C S =13.6nF, T J = 25 C,.2µs/DIV Ch. 2: V GE 1 V/DIV, Ch. 3: I C 2 A/DIV, Ch. A: V CE 1 V/DIV, Ch. C: P L 1 KW/DIV, Ch. D: W 2 mws/div In opposite to the hard switching at inductive load the collector emitter voltage V CE starts rising at the beginning of the collector current fall. This is caused by the snubber capacitors C S. According to the size of C S the rise of V CE is limited. It is very important to remark that at soft switch off the tail current dominates the losses. In addition the switch off behaviour of the s depends on different parameters especially temperature, supply voltage and switch off current. Woff [mws] 1 8 6 4 2 5 1 15 2 25 3 C S [nf] 25 C 6V 75A FAST 25 C 6V 75A Trench 25 C 6V 75A Trench 125 C 6V 75A 125 C 6V 75A Fast 125 C 6V 75A Figure 5: Switch off losses of the s via C S at rated current Woff [mws] 4 3 2 1 2 3 4 5 6 V CC [V] Trench 25 C 3A 13.6nF Trench 125 C 3A 13.6nF 25 C 3A 13.6nF 125 C 3A 13.6nF Fast 25 C 3A13.6nF Fast 125 C 3A 13.6nF Figure 7: Switch off losses of the s via V CC Figure 7 shows that for the standard and the fast s the losses linearly increase with the voltage V CC. The Trench- losses show for voltages more than 4V only a small increase. This is caused by the field stop layer in this [8]. For high voltages and low switch off currents the Trench- can switch off with low-loss nearly like the Fast- specialized for this aim. III. Passive switch on To test the passive switch on in general a special circuit shown in figure 8 was used. The tested S 2 is in stand by every time at V GE = 15V and switches passive on and off. The switch S 1 is used to switch actively the load current. With an increase of C S (fig. 5) it is possible to decrease the switch off losses. But the size of this decrease is not very high compared to hard switching. This is caused by the high influence of the tail current. The Fast- shows the best performance at switch off at rated current of the module at all measured C S -values.
S 1 V GE S2 = V CC P VS2 W S2 D 1 L L 15V S 2 I CS2 Figure 8: Test circuit for passive switch on According the inductive load a certain di/dt will be pressed into the switch S 2. Figure 9 and 1 show the passive switch on for the Fast- and the Trench. The voltage drop at the finish of the current rise is due to over voltages caused by the stray inductance of the module. P VS2 V GE S2 W S2 I CS2 V CES2 Figure 9: Passive switch on of a Trench-, cond.: T J =125 C, di/dt=5 A/µs, V GE =15V,.5µs/DIV, Ch. 2: V GE 1 V/DIV, Ch. 4: I C 2 A/DIV, Ch. 1: V CE 5 V/DIV, Ch. C: P V 2 W/DIV, Ch. D: W.5 mws/div The Trench- shows a high but very short voltage spike at passive switch on and additional it reaches it s low saturation voltage very fast. The process of conductivity modulation which is responsible for this behaviour is finished in this very fast. Both s of the second generation show similar lower voltage spikes at passive switch on (see fig. 1 for the fast ). V CES2 Figure 1: Passive switch on of a fast, cond.: T C =125 C, di/dt=5 A/µs, V GE =15V, 1µs/DIV, Ch. 2: V GE 1 V/DIV, Ch. 4: I C 2 A/DIV, Ch. 1: V CE 5 V/DIV, Ch. C: P V 2 W/DIV, Ch. D: W 1 mws/div On the other hand they need nearly double the time to come into saturation. Further measurements show for all s a rise of the overvoltage at increasing di/dt. Table 1: Estimation of additional losses at passive switch on Typ W on dyn / mws T J =25 C, di/dt=5 A/µs W on dyn / mws T J =125 C, di/dt=5 A/µs Fast.7.32 Trench Standard (not to be measured).16.11.44 Table 1 specifies the additional losses caused trough the passive switch on for a di/dt = 5 A/µs which corresponds to a frequency of nearly 8 khz. At low chip temperature this losses are negligible. At high temperatures the losses for the s of the second generation could be important. IV. Calculation and comparison of total and diode losses For an estimation of the whole losses of an and a diode of the modules a calculation was
performed using the program Mathcad. Measurements show that under the test conditions the switching losses of the diode are negligible in the investigated frequency range. inductance caused through the nearly sinusoidal load current was noted at this calculations. The calculations without the passive switch on were performed in the same manner as for T J =25 C. The evaluation was performed first for 25 C. Here the passive switch on losses of the are neglected (see table 1). Measured values of the switch off energy and the switching times are used. The conduction losses were calculated according the static characteristics of the s and Diodes whereas the load current was assumed to be sinusoidal. The diagram in figure 11 shows the calculated loss split of the modules at f=5 khz under this conditions P [W] 45 4 35 3 25 2 15 1 5 PswT PcondT PonT real PtotT PtotT real 21 52 94 21 52 94 21 52 94 f [khz] Fast Trench P [W] 25 2 15 1 5 PswT PcondT PcondD Ptot 22 3 5 22 3 5 3 5 Icoff [A] Fast Trench Figure 11: Loss split of the -modules at different switch off currents, cond.: I Lmax =75A, f=5khz, V CC =6V, T J =25 C, C S =13.6nF The Fast- shows the lowest switching losses but the conduction losses are relatively high. The Trench- has very low conduction losses and at low switch off currents nearly the same low switching losses like the Fast-. This leads in this range to the lowest total losses. Additional calculations at 1 khz show the same results. Simulations at 2 khz show that at this frequency the conduction losses dominate. The Trench- has here the lowest losses. The s of the Fast- and the Standard-- Modules have nearly the same losses. According to the results in table 1 the influence of the passive switch on of the s at a temperature of T J =125 C is considered and compared to a calculation without the influence of the passive switch on. For this aim the passive switch on and the conduction phase were measured directly in the resonant converter for some few working points and regarded together as P ont real. Further more the influence of the voltage drop at the stray Figure 12: Loss split of the s at different frequencies with and without a consideration of the additional losses at passive switch on, cond.: I Lmax =75A, I Coff =3 A, V CC =6V, T J =125 C, C S =13.6nF Figure 12 shows that at T J =125 C a rise of the frequency leads for all investigated s to an increase of the difference between the calculated losses with and without a consideration of the passive switch on. Measurements show that with increasing frequencies the s ever less reach their static working point at the conduction phase. The Trench- shows only for very high frequencies a remarkable influence of the passive switch on to the total losses. For the Fast- and the especially the Standard- the passive switch on is remarkable for frequencies of approx. 5 khz and more. At relatively low frequencies the passive switch on is negligible for T J =125 C also. P [W] 35 3 25 2 15 1 5 PswT PonT real PtotT real 22 3 5 22 3 5 3 5 Icoff [A] Fast Trench- Figure 13: Loss split of the -Modules at different switch off currents, cond.: I Lmax =75A, f=52khz, V CC =6V, T J =125 C, C S =13.6nF
The losses of the investigated s at T J =125 C and f=52khz are presented in Figure 13 for different load currents. Expectedly the losses are higher then those at T J =25 C for all modules. Also at T J =125 C the Trench- has an advantage against the Fast- especially at low switch off currents. V. Conclusion Three -Modules are tested in terms of their properties for applications in resonant inverters which are working in the Zero Voltage Mode up to high frequencies. These are a planar Standard--Module of the 2. generation, a planar Fast--Module of the 2. generation specialized for resonant applications and a new Trench--Module of the 3. generation. To meet the investigation goals a test stand was established which works in the ZVS-mode. The conduction and especially the switch off losses of the are the most important parts of the total losses in this mode. The passive switch on losses are to note especially at the Standard- and the Fast-. For the Trench- these losses are negligible up to high frequencies. The results show that it is possible to drive the Trench- without special optimization up to high frequencies on a low loss level like the Fast- specialized for this application. Indeed for the Trench- these advantages exists at a mode at high voltages and relatively low switch off currents. commutation including parameter variation, PESC 21 [5] T.Reimann, Verhalten abschaltbarer Leistungshalbleiterbauelemente im ZVS-Mode, German Diss., Technical University Ilmenau, 1994 [6] S. Pendharkar, K. Shenai, Zero voltage switching behaviour of punchthrough and nonpunchthrough s!, IEEE Transactions on Electron Devices, Vol. 45, No. 8, 1998, pp. 1826-1835 [7] A. Claudio, J. Aguayo, M. Cotorogea, Special test circuit for the analysis of behaviour in ZCS Commutation, EPE-Journal, Vol. 11, No. 1, February 21, pp. 25-32 [8] L. Lorenz: s for next decade s motor drive systems, PCIM-Europe, January/February 21, pp. 2-22 [9] N. Mohan, T. Undeland, W. Robbins: Power Electronics, 2. edition, John Wiley & Sons, New York, 1995, ISBN -471-5848-8 References [1] H. Mecke, W. Fischer, F. Werther: Soft switching inverter power source for ARC welding, EPE 1997, pp. 4.333-4.337 [2] U. Kirchenberger, K. Fischer, D. Schröder: Analysis of a constant frequency series-parallel multiresonant converter, EPE 1993, pp. 76-82 [3] C. Keller, Einsatzkriterien schneller abschaltbarer Leistungshalbleiter in Quasiresonanz-Umrichtern, German Diss., Technical University Berlin, 1991 [4] A. Claudio, J. Aguayo, M. Cotorogea, Characterization of different s in ZVS