This paper presents an ultra-compact integrated electric drive prototype. The prototype illustrates the integration of a fractional slot concentrated winding (FSCW) electric motor, a stacked polyphase bridges (SPB) converter, the control boards, and the water cooling plates into a common housing. This integrated prototype offers a high potential of compactness and cost reduction for electric and hybrid electric vehicles.

This paper presents an ultra-compact integratedelectric drive prototype. The prototype illustrates the integrationof a fractional slot concentrated winding (FSCW) electric motor,a stacked polyphase bridges (SPB) converter, the control boards,and the water cooling plates into a common housing. Thisintegrated prototype offers a high potential of compactness andcost reduction for electric and hybrid electric vehicles.

The use of conventional energy sources for electricity generation, such as fossil fuel combustion and nuclear power, is questioned because of environmental and safety issues and concerns about possible anthropogenic climate change. This has led to rapid developments in the field of renewable energy exploitation. Entire new fast-growing industries are formed to supply equipment for renewable power plants. The contribution from Swedish industry to this development has so far largely been limited to providing parts and components and in most cases no integrated power plants are offered. From a Swedish perspective an important question is therefore what measures are needed in terms of research and development to increase Swedish participation in the industries related to renewable power plants.

Three renewable primary energy sources have been investigated with respect to their energy supply potential, the industries involved, the technical state-of-the-art and finally the state of academic research. These are wind power, PV solar power and geothermal power. Currently, wind power has by far the largest installed base worldwide of these (238 GW) and also has the lowest levelised cost of electricity (LCOE) with credible figures of 7 c€/kWh presented for onshore installations. PV solar power is second in terms of installed power at 67.3 GW but shows faster growth, thanks to rapidly decreasing cost of equipment. The LCOE figures are approaching those of wind power (10 c€/kWh for Southern Europe is recently reported), which should allow for further growth. Geothermal energy production has been exploited for decades, and the installed base has reached 10.7 GW. It shows attractive electricity cost figures thanks to high capacity factor. LCOE is highly dependent on the location, but figures in the range of 4 c€-11c€ have been reported, which implies a significant potential. In terms of installations in Sweden, wind power is by far the dominant of the studied sources, with the remaining two being of marginal significance currently.

Regarding academic research to increase the industrial base within renewables in Sweden the following recommendations are given. For wind power this goal can be achieved by promoting research into generator, transformer, and power electronic converter concepts, in close cooperation with established or newly founded companies. As for PV solar power, likely the main potential for Swedish industry lies in seeking further openings in the field of balance-of-system (BOS) equipment, considering the extreme price pressure on PV cells and modules, which has effectively forced several Swedish PV module manufacturers out of business. Academic research should thus focus on BOS aspects, with the aim of reducing LCOE. Finally, for geothermal electricity generation careful monitoring of the area should be encouraged, both technically and in terms of business. This should allow for the identification of suitable areas for research. A particularly interesting aspect that applies to several kinds of renewable energy sources, due to their temporal and spatial variability, is the use of long-distance power transmission, especially high-voltage direct current (HVDC). This is a Swedish strength area with world-leading companies. Therefore, academic research should also focus on the connection of different types of large renewable energy plants to HVDC transmission links.

Modular multilevel converters, based on cascading of halfbridge converter cells, can combine low switching frequency with low harmonic interference. They can be designed for high operating voltages without direct series connection of semiconductor elements. This has led to a rapid adoption within high-power applications such as HVDC, STATCOM and railway interties. Analysing the operation of these converters in the frequency domain poses a few challenges due to the presence of significant low-order harmonic voltages in the cell capacitors. This paper presents a frequency-domain model of the MMC converter with halfbridge cells, based on a two-stage approach. First, the circuit equations are decoupled by a simple linear transformation, whereby the circuit schematic can be separated into a dc-side and an ac-side part. Second, the switching operation within the phase arms is modelled in the frequency domain by iterated convolution. The model is verified against a timedomain simulation of a converter with ratings valid for HVDC applications. It is shown that the proposed methodology, where all calculations are made in the frequency domain, can accurately reproduce the results from the simulation.

Modular multilevel converters (MMC), based on cascading of half-bridge converter cells, can combine low switching frequency with low harmonic interference. They can be designed for high operating voltages without direct series connection of semiconductor elements. This has led to a rapid adoption within high-power applications such as high voltage direct current transmission, railway interties and medium voltage industrial motor drives. Analyzing the operation of these converters in the frequency domain poses a few challenges due to the presence of significant low-order harmonic voltages in the cell capacitors. This paper treats a frequency-domain methodology for computing inner variables of the MMC with half-bridge cells, based on a two-stage approach. First, the circuit equations are decoupled by a simple linear transformation, whereby the circuit schematic can be separated into a dc-side and an ac-side part. Second, the variables of the cell strings are computed in the frequency domain by iterated convolution. It is shown that the proposed methodology, where all calculations are made in the frequency domain, can accurately reproduce the results from a PSCAD simulation. Furthermore, the model is successfully employed as part of an optimization algorithm for maximizing the power handling capability of the converter by appropriately controlling the circulating current and zero-sequence ac-side voltage. Results from this work point to significant possibilities for improvement.

State-of-the-art silicon carbide switches have current ratings that are not sufficiently high to be used in high-power converters. It is, therefore, necessary to connect several switches in parallel in order to reach sufficient current capabilities. An investigation of parallel-connected normally ON silicon carbide JFETs is presented in this paper. The device parameters that play the most important role for the parallel connection are the pinch-off voltage, the gate-source reverse breakdown voltage, the spread in the on-state resistances, and the variations in static transfer characteristics of the devices. Moreover, it is experimentally shown that a fifth factor affecting the parallel connection of the devices is the parasitic inductances of the circuit layout. The temperature dependence of the gate-source reverse breakdown voltages is analyzed for two different designs of silicon carbide JFETs. If the spread in the pinch-off and gate-source reverse breakdown voltages is sufficiently large, there might be no possibility for a stable off-state operation of a pair of transistors without forcing one of the gate voltages to exceed the breakdown voltage. A solution to this problem using individual gate circuits for the JFETs is given. The switching performance of two pairs of parallel-connected devices with different combinations of parameters is compared employing two different gate-driver configurations. Three different circuit layouts are considered and the effect of the parasitic inductances is experimentally investigated. It is found that using a single gate circuit for the two mismatched JFETs may improve the switching performance and therefore the distribution of the switching losses significantly. Based on the measured switching losses, it is also clear that regardless of the design of the gate drivers, the lowest total switching losses for the devices are obtained when they are symmetrically placed.

Considering the present development of the available Silicon Carbide switches, their current ratings are so low that they cannot be used for high-power converters. It is therefore necessary to connect several switches in parallel in order to obtain sufficient current ratings. An investigation of parallel-connected normally-on Silicon Carbide Junction Field Effect Transistors is presented in this paper. The parameters that play the most important role for the parallel connection are the pinch-off and the gate-source breakdown voltages. The temperature dependency of those two voltages is analyzed based on the pnp structure of the device. If the spread in these parameters is sufficiently large there might be no possibility for a stable off-state operation of a pair of transistors without forcing one of the gate voltages to exceed the breakdown voltage, especially at high temperatures. A solution to this problem is given. The switching performance of two pairs of parallel-connected devices is compared with respect to their pinch-off voltages, and it is found that differences of approximately 25% in switching losses could result from a difference in the pinch-off voltage of 0.5 V.

Due to the low current ratings of the currently available silicon carbide (SiC) switches they cannot be employed in high-power converters. Thus, it is necessary to parallel-connect several switches in order to reach higher current ratings. This paper presents an investigation of parallel-connected normally-on SiC junction field effect transistors. There are four crucial parameters affecting the effectiveness of the parallel-connected switches. However, the pinch-off voltage and the reverse breakdown voltage of the gates seem to be the most important parameters which affect the switching performance of the devices. In particular, the spread in these two parameters might affect the stable off-state operation of the switches. The switching performance and the switching losses of a pair of parallel-connected devices having different reverse breakdown voltages of the gates is investigated by employing three different gate-driver configurations. It is experimentally shown that using a single gate-driver circuit the switching performance of the parallel-connected devices is almost identical, while the total switching losses are lower compared to the other two configurations.

The very low on-state resistance, the voltagecontrolledgate, and the relative simplicity of fabrication of thenormally-ON silicon carbide junction field effect transistor makethis device the most important player among all state-of-theartsilicon carbide transistors. However, the normally-ON naturecounts as the main factor which keeps this device far frombeing considered as an alternative to the silicon insulated-gatebipolar transistor. A self-powered gate driver without externalpower supply for normally-ON silicon carbide junction field effecttransistors is presented in this paper. The proposed circuit isable to handle the shoot-through current when the devices aresubjected to the dc-link voltage by utilizing the energy associatedwith this current. On the other hand it supplies the necessarynegative gate-source voltage during the steady-state operation. Adetailed description of the operating states of the proposed circuitalong with various design considerations are presented. Fromexperiments which were performed in a half-bridge converter, itis shown that the shoot-through current can be turned off withinapproximately 15 s. Moreover, it is shown that the proposedgate driver can properly switch the devices.

The very low on-state resistance, the voltage-controlled gate, and the relative simplicity of fabrication of the normally ON silicon carbide junction field-effect transistor (JFET) make this device the most important player among all state-of-the-art silicon carbide transistors. However, the normally ON nature counts as the main factor which keeps this device far from being considered as an alternative to the silicon insulated-gate bipolar transistor. A self-powered gate driver without external power supply for normally ON silicon carbide JFETs is presented in this paper. The proposed circuit is able to handle the short-circuit currents when the devices are subjected to the dc-link voltage by utilizing the energy associated with this current. On the other hand, it supplies the necessary negative gate-source voltage during the steady-state operation. A detailed description of the operating states in conjunction with a theoretical analysis of the proposed self-powered gate driver is presented. The first part of the experimental investigation has been performed when the proposed circuit is connected to a device which is directly subjected to the dc-link voltage. The second set of measurements were recorded when the self-powered gate-driver was employed as the driver of normally ON components in a half-bridge converter. From the experimental results, it is shown that the short-circuit current is cleared within approximately 20μs after the dc-link voltage is applied, while the power consumption when all devices are kept in the OFF state equals 0.37W. Moreover, it is experimentally shown that the proposed gate driver can properly switch when it is employed in a half-bridge converter. Finally, limitations regarding the range of the applications where the self-powered gate drive can efficiently operate are also discussed.

A circuit solution to the normally-ON property of the normally-ON silicon carbide junction field-effect transistor, namely the self-powered gate driver, has been recently proposed. This letter sheds some light on the design process of the self-powered gate driver concept as well as limitations from the system perspective. It is experimentally shown that the parameters of the self-powered gate driver must be chosen taking into account a tradeoff between a fast response and stable operation of the driver. Moreover, the influence of the shoot-through current in the fast activation of the self-powered gate driver is also presented.

An experimental performance comparison between SiC JFET and SiC BJT switches which are used as the main switch for a 2 kW dc/dc converter is presented. In order to perform a fair comparison and due to the different chip areas of these two SiC devices, they both operate under the same on-state losses. Moreover, the switching speeds of the gate and base drivers are approximately equal. It is experimentally shown that the SiC BJT is switching slightly faster than the SiC JFET under the same circuit conditions, while the driver loss for the SiC BJT is higher than for the JFET, especially at relatively low switching frequencies. Various experimental results dealing with the switching performance of the SiC devices and the power losses at different switching frequencies are presented. It is found that the BJT converter has a higher efficiency (99.0% measured at 50 kHz) that the JFET converter.

This paper studies the possibility of building a Modular Multilevel Converter (M2C) using Silicon Carbide (SiC) switches. The main focus is on a theoretical investigation of the conduction losses of such a converter and a comparison to a corresponding converter with silicon insulated gate bipolar transistors. Both SiC BJTs and JFETs are considered and compared in order to choose the most suitable technology. One of the sub-modules of a down-scaled 10 kVA prototype M2C is replaced with a sub-module with SiC JFETs without anti-parallel diodes. It is shown that diode-less operation is possible with the JFETs conducting in the negative direction, leaving the possibility to use the body diode during the switching transients. Experimental waveforms for the SiC sub-module verify the feasibility during normal steady-state operation. The loss estimation shows that a 300 MW M2C for high-voltage direct current transmission would potentially have an efficiency of approximately 99,8 % if equipped with future 3.3 kV 1.2 kA SiC JFETs.

This paper studies the possibility of building a modular multilevel converter (M2C) using silicon carbide (SiC) switches. The main focus is on a theoretical investigation of the conduction losses of such a converter and a comparison to a corresponding converter with silicon-insulated gate bipolar transistors. Both SiC BJTs and JFETs are considered and compared in order to choose the most suitable technology. One of the submodules of a down-scaled 3 kVA prototype M2C is replaced with a submodule with SiC JFETs without antiparallel diodes. It is shown that the diode-less operation is possible with the JFETs conducting in the negative direction, leaving the possibility to use the body diode during the switching transients. Experimental waveforms for the SiC submodule verify the feasibility during normal steady-state operation. The loss estimation shows that a 300 MW M2C for high-voltage direct current transmission would potentially have an efficiency of approximately 99.8% if equipped with future 3.3 kV 1.2 kA SiC JFETs.

Previous results concerning instability of the dc linkin inverter drives fed from a dc grid or via a rectifier are extended.It is shown that rectifier–inverter drives equipped with small (film)dc-link capacitors may need active stabilization. The impact oflimited bandwidth and switching frequency in the inverter–motorcurrent control loop is considered, and recommendations forselection of the dc-link capacitor, the switching frequency, andthe dc-link stabilization control law in relation to each other aregiven. This control law is incorporated in a field-weakening (toenhance voltage sag ride-through) current controller for whichdesign recommendations are presented.

The paper proposes a novel topology of a simple base drive unit for silicon carbide bipolar junction transistors (BJTs) based on the current-source principle. Energy stored in a small, air-cored inductor is employed to generate a current peak forcing the BJT to turn-on (10–20ns) very rapidly. The driver enables very high switching performance and very low switching losses of the driven BJT. Both the current source and the unit delivering the steady-state current to the base are supplied from the same low-voltage source in order to limit power consumption. Operation principles as well as selected design issues are discussed in the paper and illustrated by experiments. The 1200V/6A SiC BJT driven by the proposed circuit shows a very fast switching speed.

The paper discusses the switching performance of the dual gate trench SiC JFET. In applications such as dc/dc converters, when fast switching is expected the standard totem-pole driver is not sufficient. The reason for this is that both the internal resistance and the parasitic capacitances of this device are significantly higher than for other designs. Instead, the gate driver with a dynamic current source is proposed in this paper to speed-up the switching process. Performed double-pulse measurements show improved dynamic performance of the tested DGTJFET with the new driver.

This paper describes issues related to design,construction and experimental verification of a 6 kW, 200 kHzboost converter (300 V/600 V) built with four parallel-connectedSiC bipolar transistors. The main focus is on parallel-connectionof the SiC BJTs: crucial device parameters and influence of theparasitics are discussed. A special solution for the base-driveunit, based on the dual-source driver concept, is also presentedin this paper. Experimental verification of the boost converterwith special attention to power loss measurement and thermalperformance of the parallel-connected transistors is also shown.The peak efficiency measured at nominal conditions wasapproximately 98.5% where the base-drive unit causes around 10% of the total losses.

This paper describes the concept, design, construction, and experimental investigation of a 40-kVA inverter with silicon carbide junction field-effect transistors (JFETs). The inverter was designed to reach an efficiency exceeding 99.5%. The size of the heat sink is significantly reduced in comparison to silicon insulated-gate bipolar transistor designs, and the high efficiency makes it possible to use free-convection cooling. This could potentially increase reliability compared with solutions with fans. A very low conduction loss has been achieved by parallel connecting ten 85-m Omega normally-ON JFETs in each switch position. A special gate-drive solution was applied, forcing the transistors to switch very fast (approximately 10 kV/mu s), resulting in very low switching losses. As output power is almost equal to input power, special effort was done to precisely determine the amount of semiconductor power losses via comparative thermal measurements. A detailed analysis of the measurements shows that the efficiency of the inverter is close to 99.7% at 40 kVA.

This paper describes the concept, the design, the construction, and experimental investigation of a 40 kVA inverter with Silicon Carbide Junction Field Effect Transistors. The inverter was designed to have an efficiency exceeding 99.5%. Due to the low losses free convection cooling could be used. Since no fans are used the reliability can be increased compared to solutions with fans. A very low conduction loss has been achieved by parallel connecting ten 85 mΩ normally-on JFETs in each switch position. A special gate-drive solution was applied forcing the transistors to switch very fast (approx. 20 kV/μs) resulting in very low switching losses. As the output power is almost equal to the input power a special effort was done to precisely determine the amount of semiconductor power losses via comparative thermal measurements. A detailed analysis of the measurements shows that the efficiency of the inverter is approximately 99.7% at 40 kVA.

This paper describes issues related to parallel connection of SiC bipolar junction transistors (BJTs) in discrete packages. The devices are applied in a high-frequency dc/dc boost converter where the switching losses significantly exceed the conduction losses. The design and construction of the converter is discussed-with special emphasis on successful parallel-operation of the discrete BJTs. All considerations are experimentally illustrated by a 6-kW, 250-kHz boost converter (300 V/600 V). A special solution for the base-drive unit, based on the dual-source driver concept, is also shown in this paper. The performance of this driver and the current sharing of the BJTs are both presented. The power losses and thermal performance of the parallel-connected transistors have been determined experimentally for different powers and switching frequencies. An efficiency of 98.23% (+/- 0.02%) was measured using a calorimetric setup, while the maximum temperature difference among the four devices is 12 degrees C.

This chapter presents recent advances in power semiconductors technology with special attention on wide bandgap (WBG) transistors. A short introduction to the state-of-the-art Silicon power devices is given, and the characteristics of the various SiC power switches are also described. Design considerations of gate and base-drive circuits for various SiC power switches along with experimental results of their switching performance are presented in details. Moreover, a section on applications of SiC power devices is also included, where the three design directions (high-efficiency, high switching frequency and high-temperature) that might be followed using SiC technology are shown. Last but not least, a short overview of Gallium Nitride transistors is presented in the last section of this chapter.

Driving a silicon carbide bipolar junction transistor is not a trivial issue, if low drive power consumption and short-switching times are desired. A dual-source base-drive unit with a speed-up capacitor consisting of a low-and a high-voltage source is, therefore, proposed in this paper. As a significant base current is required during the conduction state, the driver power consumption is higher than for other semiconductor switches. In the presented solution, the steady-state base current is provided by a low-voltage source and is optimized for lowpower losses. On the contrary, a second source with a higher voltage and speed-up capacitor is used in order to improve the switching performance of the device. The proposed driver has experimentally been compared to other standard driver solutions by using a double-pulse circuit and a 2-kW dc/dc boost converter. Switching times of 20 ns at turn-ON and 35 ns at turn-OFF were recorded. Finally, the efficiency of the converter was determined experimentally at various switching frequencies. From power loss measurements at 100-kHz switching frequency using the proposed driver in a 2-kW dc/dc boost converter, it was found that the efficiency was approximately 99.0%. In the same operating point, the driver power consumption was only 0.08% of the rated power.

An novel control strategy of the series loaded resonant converter is presented. The main objective of the control strategy is to minimize the switching losses in the main switching elements (IGBT). The results are experimentally verified on a 60kW/25kHz prototype converter. The IGBT losses obtained with the proposed control strategy are compared with those of the commonly used frequency control and phase-shift control strategies. With the proposed control strategy it is found that the losses are reduced in the entire operating range compared to phase-shift control. The same statement is valid also for frequency control, except for the highest output voltage.

Two different dynamic effects influencing the IGBT losses in soft switching converters are demonstrated. The first one, The Dynamic tail-charge effect shows that the tail-charge is dependent not only on the absolute value of the current at turn-off but also on the dynamics of the current. This effect may have a significant impact on the optimization of zero-current-switching converters. The Dynamic conduction losses originate from the conductivity modulation lag of the IGBT. It is shown by experiments that the on-state losses depend on the operating frequency. Different methods to accurately determine the on-state losses are evaluated. It was found that the best method is an indirect measurement where the stray inductance is identified by the use of an oscillating circuit. The experiments are performed under a sinusoidal current excitation at a fixed amplitude (150 A) for different frequencies (up to 104 kHz). The switching devices used are IGBT-modules rated 300-400 A/1200 V in a bridge-leg configuration. From the experiments performed it is found that IGBTs of a modern PT design have the lowest losses in the series-loaded resonant converters studied in the present paper.

In this paper a method to adjust the AC winding capacitance of high-voltage high-frequency transformers by means of a winding-rectifier integration is described. First a theoretical background to the method is given. From the theory, an equivalent circuit describing the characteristics of the combination of the transformer and the rectifier is derived. The derived circuit introduces the concept of a DC-capacitance. Finally, the equivalent circuit and the method itself are verified by means of experiments on a transformer-rectifier system from an industrial application with the ratings 70 kV, 30 kW, 25 kHz. The results from the experiments show that it is possible to vary the AC component of the winding capacitance from a few percent up to 95 percent of the total winding capacitance. This means that it is virtually free to choose between AC or DC capacitance during the design stage. This is very important in applications such as resonant converters with transformers having secondary windings connected to rectifiers with capacitive output filters.

Soft-switching converters equipped with insulated gate bipolar transistors (IGBTs) in silicon (Si) have to be dimensioned with respect to additional losses due to the dynamic conduction losses originating from the conductivity modulation lag. Replacing the IGBTs with emerging silicon carbide (SiC) transistors could reduce not only the dynamic conduction losses but also other loss components of the IGBTs. In the present paper, therefore, several types of SiC transistors are compared to a state-of-the-art 1200-V Si IGBT. First, the conduction losses with sinusoidal current at a fixed amplitude (150 A) are investigated at different frequencies up to 200 kHz. It was found that the SiC transistors showed no signs of dynamic conduction losses in the studied frequency range. Second, the SiC transistors were compared to the Si IGBT in a realistic soft-switching converter test system. Using a calorimetric approach, it was found that all SiC transistors showed loss reductions of more than 50%. In some cases loss reductions of 65% were achieved even if the chip area of the SiC transistor was only 11% of that of the Si IGBT. It was concluded that by increasing the chip area to a third of the Si IGBT, the SiC vertical trench junction field-effect transistor could yield a loss reduction of approximately 90%. The reverse conduction capability of the channel of unipolar devices is also identified to be an important property for loss reductions. A majority of the new SiC devices are challenging from a gate/base driver point-of-view. This aspect must also be taken into consideration when making new designs of soft-switching converters using new SiC transistors.

Parasitic inductances caused by the package of semiconductor devices in power converters, are limiting the switching speed and giving rise to higher switching losses than necessary. In this study a half-bridge planar power module with Silicon Carbide (SiC) MOSFET bare dies was designed and manufactured for ultra-low parasitic inductance. The circuit structure was simulated and the parasitic inductances were extracted from ANSYS-Q3D. The values were then fed into LT-Spice to simulate the electrical behavior of the half-bridge. The experimental and simulation results were compared to each other and were used to adjust and easily extend the simulation model with additional MOSFETs for higher current capability. It was shown that the proposed planar module, with four parallel SiC MOSFETs at each position, is able to switch 600V and 400A during 40 and 17ns with E-ON and E-OFF equal to 3.1 and 1.3mJ, respectively. Moreover, unlike the commercial modules, this design allows double-sided cooling to extract the generated heat from the device, resulting in lower operating temperature.

Silicon Carbide (SiC) MOSFETs offer excellent properties as switches in power converters. However, the package of the device is an issue that prevents utilizing the advantages of SiC, as for instance fast switching speed. The packages of currently available SiC devices are the same as those previously used for silicon devices with moderate electrical and thermal characteristics resulting in accelerated aging and reliability issues. Moreover, the parasitic inductance caused by the package, limits the switching time and operating frequency. By excluding the package, the parasitic inductances will be eliminated to a large extent. In this study, the procedure of manufacturing a half-bridge planar power module, using four SiC MOSFET bare dies and PCB, is described. According to simulations, the parasitic inductance Lstray of the structure is approximately 96 % lower than most commercial half-bridge modules. It is also shown that double-side cooling can be employed for the proposed module if substrates with low thermal resistance are employed.

Today's combustion engines have low efficiency and a large amount of useful energy converts to heat as waste in different type of vehicles. Improving the dynamics of the car body, injection system, the shape of the internal engine components and manipulating the fuel compositions have had influence on fuel economy, but still less than 50% of energy in the fuel is converted to useful mechanical power. Since the lost energy escapes through the exhaust system as heat, taking advantage of thermoelectricity, part of that energy can be converted to useful electrical energy, improving the overall efficiency. However, the output voltage from a thermoelectric generator is a function of hot and cold side temperature and since, the electrical system of the vehicle operates with constant voltage, the use of a power converter is necessary. In this paper, simulation and experimental results of such a high-efficiency converter(94-96%), designed for thermoelectric generators on heavy duty vehicles is presented and discussed.

The European Union’s 2020 target aims to be producing 20 % of its energy from renewable sources by 2020, to achieve a 20 % reductionin greenhouse gas emissions and a 20 % improvement in energy efficiency compared to 1990 levels. To reach these goals, the energyconsumption has to decrease which results in reduction of the emissions. The transport sector is the second largest energy consumer in theEU, responsible for 25 % of the emissions of greenhouse gases caused by the low efficiency (<40 %) of combustion engines. Much workhas been done to improve that efficiency but there is still a large amount of fuel energy that converts to heat and escapes to the ambientatmosphere through the exhaust system. Taking advantage of thermoelectricity, the heat can be recovered, improving the fuel economy. Athermoelectric generator (TEG) consists of a number of thermoelectric elements, which advantageously can be built into modules,arranged thermally and electrically, in a way such that the highest possible thermal power can be converted into electrical power. In aunique waste heat recovery (WHR) project, five international companies and research institutes cooperated and equipped a fully drivableScania prototype truck with two TEGs. The entire system, from the heat transfer in the exchangers to the electrical power system, wassimulated, built and evaluated. The primary experimental results showed that approximately 1 kW electrical power could be generatedfrom the heat energy. In this paper the entire system from design to experimental results is presented.

The efficiency of internal combustion engines in trucks and passenger cars are low (<40%). Much work has been done to make the engines more efficient internally, by improving the mechanical and electrical components. However, there is still a large amount of fuel power, which gets converted into heat and escapes through exhaust gases as waste heat. Taking advantages of thermoelectricity, part of that heat power can be converted into electrical power. In this paper, the most suitable DC/DC converter for Thermoelectric Generator in Heavy Duty Vehicles is proposed and based on the simulation results, different aspects of designing high-efficient DC/DC converters are discussed.

A considerable part of the fuel energy in vehicles never reaches the wheels and entirely converts to waste heat. In a heavy duty vehicle (HDV) the heat power that escapes from the exhaust system may reach 170 kW. The waste heat can be converted into useful electrical power using thermoelectric generator (TEG). During the last decades, many studies on the electrical power conditioning system of TEGs have been conducted. However, there is a lack of studies evaluating the electrical instrumentation, the impact of the converter-efficiency, and the TEG arrangement on a real large-scale TEG on-boarda drivable vehicle. In this study, the most important parameters for designing electrical power conditioning systems for two TEGs, developed for a real-scale HDV as well as experimental results demonstrating the recovered electrical power, are presented. Eight synchronous inter-leaved step-down converters with 98 % efficiency with perturb and observe maximum power point tracker was developed and tested for this purpose. The power conditioning system was communicating with the on-board computers through the controller area network and reported the status of the TEGs and the recovered electrical power. The maximum recovered electrical power from the TEGs reached 1 kW which was transmitted to the electrical system of the vehicle, relieving the internal combustion engine.

The demand for high-efficiency power converters is increasing continuously. The switching losses are typically significant in power converters. During the switching time, the component is exposed to a considerable voltage and current causing power loss. The switching time is limited by parasitic inductance produced by traces and interconnections inside and outside the package of a device. Moreover, the parasitic inductances at the input-terminal together with the Miller capacitance generate oscillations causing instability and additional losses. In order to eliminate the package parasitic inductance, four 1.2kV SiC-MOSFET bare dies, two in parallel in each position, were directly attached to a PCB sandwich designed as a half bridge. The obtained structure forms a planar power module. From ANSYS Q3D simulations it was found that the parasitic inductance between drain and source for each transistor in the proposed planar module could be reduced 92 % compared to a TO247 package. The planar module was also tested as a dc-dc converter. Switching waveforms from these experiments are also presented.

Parasitic inductances, caused by the package of semiconductor devices in power converters, are limiting theswitching speed and giving rise to higher switching losses than necessary. In this study a half-bridge planar power module with Silicon Carbide (SiC) MOSFET bare dies was designed and manufactured for ultra-low parasitic inductance. The circuit structure was simulated and the parasitic inductances were extracted from ANSYS-Q3D. The values were then fed into LT-Spice to simulate the electrical behavior of the half-bridge.The experimental and simulation results were compared to each other and were used to adjust and easily extend the simulation model with additional MOSFETs for higher current capability. It was shown that the proposed planar module, with four parallel SiC MOSFETs at each position, is able to switch 600V and 400A during 40 and 17ns with EON and EOFF equal to 3.1 and 1.3 mJ, respectively. Moreover, unlike the commercial modules, this design allows double-sided cooling to extract the generated heat from the device, resulting in lower operating temperature.

Silicon Carbide (SiC) MOSFETs offer excellent properties as switches in power converters. However, the package of the device is an issue that prevents utilizing the advantages of SiC, as for instance fast switching speed. The packages of currently available SiC devices are the same as those previously used for silicon devices with moderate electrical and thermal characteristics resulting in accelerated aging and reliability issues. Furthermore, the parasitic inductance caused by the package, limits the switching time and operating frequency. By excluding the package, the parasitic inductances will be eliminated to a large extent. In this study, the procedure of manufacturing a half-bridge planar power module, using four SiC MOSFET bare dies and PCB, is described. According to simulations in ANSYS-Q3D, the parasitic inductance Lstray of the structureis approximately 96% lower than most commercial half-bridge modules. It is also shown that double-side cooling can bee mployed for the proposed module if substrates with low thermal resistance are used.

Wide Bandgap power semiconductors such as SiC MOSFETs, have enabled compact and highly efficient power converters operated at higher frequencies. However, parasitic inductance of the package may significantly increase power losses and limit the operation. This paper aims to quantify experimentally these losses in a soft-switching converter. A 'removable' stray inductance is implemented in a setup consisting of discrete SiC MOSFET units. Thus, the power loss of the transistors with and without stray inductance can be compared. Similarly slower switching speeds are also implemented to fully emulate a 62-mm module. The power loss induced by the package can thus be evaluated.

Wide Bandgap power semiconductors such as SiC MOSFETs, have enabled compact and highly efficient power converters operated at higher frequencies. However, parasitic inductance of the package may significantly increase power losses and limit the operation. This paper aims to quantify experimentally these losses in a soft-switching converter. A "removable" stray inductance is implemented in a setup consisting of discrete SiC MOSFET units. Thus, the power loss of the transistors with and without stray inductance can be compared. Similarly slower switching speeds are also implemented to fully emulate a 62-mm module. The power loss induced by the package can thus be evaluated.

Humidity and outdoor application are a challenge for Silicon (Si) and Silicon Carbide (SiC) applications. This paper investigates the effect of humidity on SiC power MOSFET modules in a real application where no acceleration factors such as pressure or high temperature are applied. Since SiC devices can operate at higher temperature than Si, the high-temperature acceleration factor may be obsolete. Moreover, the humidity might be more critical when the temperature inside the converter enclosure and modules housing is varying with daily temperature variations and weather constraints in harsh environments. The breakdown voltages of the humidity-exposed modules are monitored regularly over a extended period of time in order to detect any increase of leakage current which indicates humidity-induced degradation. After 630 hours, the modules operated outdoor presented an increased leakage current at 1.2 kV and over the whole range of applied voltage.

Silicon Carbide (SiC) Metal-Oxide-Semiconductor Field-Effect Transistors (MOSFETs) have the potential to increase the power density in power electronics converters compared to the currently used silicon (Si). Their benefits are higher efficiency, higher switching speeds, and higher operating temperatures. Moreover, SiC MOSFETs, which are normally-off, offer the possibility to directly replace Si Isolated-Gate-BipolarTransistors (IGBTs) in already existing converter designs with minimal circuit changes. Nevertheless, as an emerging technology, the reliability performance remains to be investigated. A reliability analysis has been performed based on a full-bridge resonant converter rated at 60 kW for modern Electrostatic Precipitator (ESP) power supplies. This analysis shows that introducing SiC devices will increase the lifetime of the converter while reducing the losses. The investment costs of replacing the Si IGBTs with SiC MOSFETs can thus be covered with the reduction of the losses over the economical operational lifetime. Furthermore, a theoretical analysis on how introducing SiC MOSFETs could increase the power density of the converter while maintaining the efficiency and the reliability. Finally, an analysis on introducing redundancy as a way to improve the reliability of the system has been performed.

The temperature evolution during a short-circuit in the die of three different Silicon Carbide1200-V power devices is presented. A transient thermal simulation was performed based on the reconstructedstructure of commercially available devices. The location of the hottest point in the device iscompared. Finally, the analysis supports the necessity to turn off short-circuit events rapidly in orderto protect the device after a fault.

Commercially available Silicon Carbide (SiC) MOSFET power modules often have a design based on existing packages previously used for silicon insulated-gate bipolar transistors. However, these packages are not optimized to take advantage of the SiC benefits, such as, high switching speeds and high-temperature operation. The package of a half-bridge SiC MOSFET module has been modeled and the parasitic elements have been extracted. The model is validated through experiments. An analysis of the impact of these parasitic elements on the gate-source voltage on the chip has been performed for both low switching speeds and high switching speeds. These results reveal potential reliability issues for the gate-oxide if higher switching speeds are targeted.

An Experimental performance analysis of a parallel connection of two 1200/80 MΩ silicon carbide SiC MOSFETs is presented. Static parallel connection was found to be unproblematic. The switching performance of several pairs of parallel-connected MOSFETs is shown employing a common simple totem-pole driver. Good transient current sharing and high-speed switching waveforms with small oscillations are presented. To conclude this analysis, a dc/dc boost converter using parallel-connected SiC MOSFETs is designed for stepping up a voltage from 50 V to 560 V. It has been found that at high frequencies, a mismatch in switching losses results in thermal unbalance between the devices.

An experimental analysis of the behavior under short-circuit conditions of three different Silicon Carbide (SiC) 1200 V power devices is presented. It is found that all devices take up a substantial voltage, which is favorable for detection of short-circuits. A suitable method for short-circuit detection without any comparator is demonstrated. A SiC JFET driver with an integrated short-circuit protection (SCP) is presented where a short-circuit detection is added to a conventional driver design in a simple way. Experimental tests of the SCP driver operating under short-circuit condition and under normal operation are performed successfully.

An experimental analysis of the behavior under short-circuit conditions of three different siliconcarbide (SiC) 1200-V power devices is presented. It is found that all devices take up a substantial voltage, which is favorable for detection of short circuits. A transient thermal device simulation was performed to determine the temperature stress on the die during a short-circuit event, for the SiC MOSFET. It was found that, for reliability reasons, the short-circuit time should be limited to values well below Si IGBT tolerances. Guidelines toward a rugged design for short-circuit protection (SCP) are presented with an emphasis on improving the reliability and availability of the overall system. A SiC device driver with an integrated SCP is presented for each device-type, respectively, where a shortcircuit detection is added to a conventional driver design in a simple way. The SCP driver was experimentally evaluated with a detection time of 180 ns. For all devices, short-circuit times well below 1 µs were achieved.

The effect of humidity on SiC Power MOSFET modules is investigated in a real application. Four modules are operated outdoor and four modules are operated indoor in identical setups, while their breakdown voltages are monitored regularly. The evolution of the leakage current, indicating humidity-induced degradation is observed.

The surge current capability of the body-diode of SiC MOSFETs is experimentally analyzed in order to investigate the possibility of using SiC MOSFETs for HVDC applications. SiC MOSFET discrete devices and modules have been tested with surge currents up to 10 times the rated current and for durations up to 2 ms. Although the presence of stacking faults cannot be excluded, the experiments reveal that the failure may occur due to the latch-up of the parasitic n-p-n transistor located in the SiC MOSFET.

Commercially available silicon carbide (SiC) MOSFET power modules often have a design based on existing packages previously used for silicon insulated-gate bipolar transistors. However, these packages are not optimized to take advantage of the SiC benefits, such as high switching speeds and high-temperature operation. The package of a half-bridge SiC MOSFET module has been modeled and the parasitic elements have been extracted. The model is validated through experiments. An analysis of the impact of these parasitic elements on the gate-source voltage on the chip has been performed for both low switching speeds and high switching speeds. These results reveal potential reliability issues for the gate oxide if higher switching speeds are targeted.

Experimental investigations on the gate-oxide and body-diode reliability of commercially available Silicon Carbide (SiC) MOSFETs from the second generation are performed. The body-diode conduction test is performed with a current density of 50 A/cm2 in order to determine if the body-diode of the MOSFETs is free from bipolar degradation. The second test is stressing the gate-oxide. A negative bias is applied on the gate oxide in order to detect and quantify potential drifts.