Lappeenranta University of Technology Faculty of Technology
Degree Program in Electrical Engineering
MASTER’S THESIS Author:
Antti SUMMANEN
Hardware design of a multiconverter for hybrid vehicle applications
1st examiner: prof. Pertti SILVENTOINEN 2nd examiner: M.Sc. Tero J ¨ARVEL ¨AINEN
ABSTRACT
Lappeenranta University of Technology Faculty of Technology
Degree Program in Electrical Engineering Antti Summanen
Hardware design of a multiconverter for hybrid vehicle applications 2015
Master’s Thesis
34 pages, 21 figures, 12 tables
Examiners: Prof. D.Sc. Pertti SILVENTOINEN M.Sc. Tero J ¨ARVEL ¨AINEN
Keywords: Inverters, DC/DC-converters, thermal modeling, silicon-carbide semiconductors
Hybrid vehicle applications often require both high-voltage system and low-voltage system. High-voltage system usually includes an energy storage, most often either a supercapacitor or a battery, diesel genera- tor or range extender, and the traction drive of the vehicle. High-voltage system also often includes some auxilliary drives such as air-compressors or hydraulics pumps. The low-voltage system usually comprises of the control units, headlamps etc. Traditionally the low-voltage system has been fed from the alterna- tor of the diesel motor, but with the high-voltage system available, the idea of removing the alternator and downsizing the battery, by the use of DC/DC-converter between HV- and LV-systems has gained more interest. The device described in this thesis is a multiconverter, which includes an high-voltage inverter, which could be used for the auxilliary drives, and an isolated DC/DC-converter for powering the low-voltage system.
This thesis describes the hardware design of the multiconverter, with the main focus on the electrical dimensioning and the thermal analysis of the high-voltage side of the unit. For the DC/DC-converter the traditional solution of using silicon IGBT transistors as the power semiconductors is compared with silicon-carbide MOSFET transistors.
The thermal model is verified with efficiency test of the prototype and the results are compared with the calculated values. Improvement possibilities of the used thermal model are also analyzed based on the measurement results.
TIIVISTELM ¨ A
Lappeenrannan teknillinen yliopisto Teknillinen tiedekunta
S¨ahk¨otekniikan koulutusohjelma Antti Summanen
Hybridiajoneuvosovelluksiin tarkoitetun monil¨aht¨oisen tehonmuokkaimen suunnittelu 2015
Diplomity¨o
34 sivua, 21 kuvaa, 12 taulukkoa
Tarkastajat: Prof. TkT. Pertti SILVENTOINEN DI Tero J ¨ARVEL ¨AINEN
Hakusanat: Invertterit, DC/DC-hakkurit, l¨amp¨omallinnus, piikarbidipuolijohteet
Hybridiajoneuvosovellukset vaativat usein sek¨a korkea- ett¨a matalaj¨annitej¨arjestelm¨an. Korkeaj¨annitej¨ar- jestelm¨a sis¨alt¨a¨a yleens¨a energiavaraston, joka on joko superkondansaattori tai korkeaj¨anniteakusto, dieselgeneraattorin tai range extenderin ja ajok¨ayt¨on. Korkeaj¨annitej¨arjestelm¨a¨an liitet¨a¨an usein my¨os erilaisia apuk¨aytt¨oj¨a kuten kompressoreita ja hydraulipumppuja. Matalaj¨annitej¨arjelm¨a koostuu yleens¨a ohjausyksik¨oist¨a, ajovaloista, yms. laitteista. Perinteisesti matalaj¨annitej¨arjestelm¨a¨a on sy¨otetty diesel- moottorin laturista, mutta korkeaj¨annitej¨arjestelmien my¨ot¨a DC/DC-hakkurin k¨aytt¨aminen korkea- ja matalaj¨annitej¨arjestelmien v¨alill¨a on her¨att¨anyt kiinnostusta, koska t¨all¨oin laturin voisi poistaa ja mata- laj¨anniteakustoa pienent¨a¨a. T¨ass¨a ty¨oss¨a kuvatun monil¨ah¨oisen tehonmuokkaimen invertterisilta soveltuu apuk¨aytt¨ojen ajamiseen, ja erotettu DC/DC-hakkuri matalaj¨annitej¨arjestelm¨an sy¨ott¨amiseen.
T¨ass¨a ty¨oss¨a k¨ayd¨a¨an l¨api edell¨a mainitun tehonmuokkaimen suunnittelu, keskittyen eritoten laitteen korkeaj¨anniteosien mitoitukseen ja termiseen suunniteluun. DC/DC-hakkurin osalta perinteisi¨a piist¨a valmistettuja IGBT transistoreja vertaillaan piikarbidi MOSFET transistoreihin.
L¨amp¨omallilaskujen paikkaansapit¨avyytt¨a tutkitaan suorittamalla prototyyppilaitteelle hy¨otysuhdemit- taus, jonka tuloksia verrataan laskettuihin tuloksiin. L¨amp¨omallin parannusmahdollisuuksia k¨asitell¨a¨an my¨os hy¨otysuhdemittauksen tulosten perusteella.
Contents
1 INTRODUCTION 1
1.1 Motivation and goals . . . 1
1.2 Structure of this thesis . . . 2
2 THEORY 3 2.1 Converter and Inverter . . . 3
2.2 DC/DC-converter . . . 4
2.3 Thermal design . . . 6
3 DESIGN SPECIFICATIONS 9 3.1 Customer specifications . . . 9
3.2 Additional specifications . . . 9
4 POWER ELECTRONICS DESIGN 10 4.1 Inverter . . . 10
4.2 Brake chopper . . . 10
4.3 Half-bridge converter . . . 11
4.4 Full-bridge converter . . . 11
5 THERMAL DESIGN 12 5.1 DC/DC-converter . . . 12
5.1.1 Buck/Boost -converter with IGBT loss model . . . 12
5.1.2 Buck/Boost -converter with MOSFET loss model . . . 16
5.1.3 H-bridge -converter with IGBT loss model . . . 17
5.1.4 H-bridge -converter with MOSFET loss model . . . 18
5.2 Inverter brigde . . . 19
5.3 Whole system . . . 25
6 MECHANICS DESIGN 27
7 RESULTS 29
8 CONCLUSION 33
Abbreviations and symbols
AC Alternating current
C Capacitance
DC Direct current
I,i Current
D Duty-cycle
E Energy
f Frequency
L Inductance
IGBT Insulated-Gate Bipolar Transistor
MOSFET Metal-Oxide-Semiconductor Field-Effect Transistor
P Power
PWM Pulse width modulation
R Resistance
RMS Root Mean Square
T Temperature
U,u Voltage
ZVS Zero-voltage switching
ř Angular frequency
ˆ Peak value
A Ampere
C Celcius
F Farad
H Henry
Hz Herz
J Joule
s Second
V Volt
W Watt
1 INTRODUCTION
This master’s thesis describes the hardware design of a combined AC/DC-converter and DC/DC-converter.
The main focus of this work is the thermal design of high voltage side of the DC/DC-converter. The main parts of the device are shown in figure 1. This work is conducted for Visedo OY.
Figure 1: System overview
1.1 Motivation and goals
Inverters and DC/DC-converters are widely used for power conversions in hybrid systems. Inverters are commonly used to drive vehicle traction motors as well auxiliary drives such as pump and fan motors.
Inverters are also used to control generators in hybrid systems. DC/DC-converters are used for trans- ferring power between different DC-voltage levels. They can be used to keep DC-link voltage stable when the voltage level of energy storage (supercapacitor or battery) is decreasing. They can also be used to transfer power between high-voltage system and low-voltage system of the vehicle. In this case an isolated converter is required for safety of users. To optimize space usage and cost effectiveness the decision was made to develop a multiconverter instead of two separate devices. This project originated as an air-cooled unit for windmill applications, and was later to be developed into liquid-cooled unit for vehicle applications.
Component dimensioning for power electronics includes electrical dimensioning as well as thermal mod- elling. Electrical dimensioning means selecting components required by the electrical specifications (volt- age levels, current and power requirements) without unnecessary overdimensioning. Thermal design is an important part of cost-effective power-electronic component dimensioning. Accurate thermal modelling prevents excessive costs caused by either underdimensioned cooling components, which lead to overheat- ing units, and overdiomensioned cooling components which increase the size and manufacturing costs of the units.
The goal of this thesis is to determine the electrical specifications for selecting components for thermal calculations and to create thermal model which can be used to compare different components in terms of efficiency and cooling requirements. Thermal model will be verified with efficiency measurements of the prototype device.
1.2 Structure of this thesis
Theory section of this thesis covers the basics of converters and DC/DC-converters as well as the basics of thermal design. Design specification and power electronics design sections cover the specifications and dimensioning calculations for the power electronics of the multiconverter. Thermal design section contains the thermal models for the inverter and the DC/DC-converter. In the case of the DC/DC-converter, the thermal model includes both silicon IGBT model and silicon-carbide MOSFET model. Mechanics design section focuses on the cooling of the device. Results show comparison between calculated and measured efficiencies as well as analyzation of the thermal model.
2 THEORY
2.1 Converter and Inverter
Inverter is an electrical device that is used for converting between direct current (DC) and alternating current (AC). Inverters are used in applications where the power source is either a DC-power storage, for example a battery, or a distributed DC-link. Most common uses for inverters are motor drives (DC to AC conversion) and generator drives (AC to DC conversion).
There are multiple different inverter topologies, which are well covered in the reference literature [1], and only 2-level voltage source inverter is covered here. Figure 2 shows the main power electronic circuit of a two level inverter.
Figure 2: Inverter main schematic
The output waveform of this kind of inverter is pulse width modulated (PWM) square wave which gets filtered into three-phase sine wave in the load inductance which might be for example motor windings, or, in the case of a grid converter, an external LCL-filter. The three phases (”L1”, ”L2” and ”L3” of figure 2) consist of two pairs of IGBT and diode (IGBT + diode pairs are labeled ”Q1” .. ”Q6” in figure 2) and the output and duty cycle of each pair is similar to figure 10 in section 5.2 with a 120 degree phase-shift between phases, and 180 degree phase shift between the upper switches (”Q1”, ”Q3” and ”Q5”) and lower switches (”Q2”, ”Q4” and ”Q6”).
Most common control strategies used in inverters are Field Oriented Control (FOC), Direct Torque Control (DTC) and scalar voltage control. These vary in complexity and accuracy of control.
Scalar voltage control uses a set voltage and frequency point or a U/f curve as a reference, and does not include a feedback. The pahse duty cycles are calculated usually by using a sine-triangle modulator, with the reference sine-waves at 120 degree phase shift between phases. Although easiest to implement, the lack of feedback leads to inaccurate control especially in transient situations, and therefore scalar control is mainly used in steady state industrial applications, for example ventilation fans, water pumps, etc.
[1, 2]
The idea of field oriented control was introduced in 1971 by F. Blaschke in his paper ”A new method for the structural decoupling of A.C. induction machines” and it differs from scalar control by having
feedback of motor operating point. In FOC the torque and flux references are fed to current controller along with the estimated torque and flux values. To ease calculations a co-ordinate transformation to the rotor reference frame is used, reducing required current and voltage calculations to two axis (direct axis and quadrature axis) instead of the three phases. The current controller calculates the needed voltage references for the Space Vector Modulator, which reverses the co-ordinate transformation and calculates the needed phase duty cycles. [2]
Direct torque control (DTC) was developed in 1984 by I. Takahashi and T. Noguchi to improve control response of FOC-drives and provide a possibility for sensorless control, which at that time was deemed unfeasible for FOC. The basic idea of DTC is to use similar torque and flux estimations as FOC, but replace the current controller of FOC with an optimized switching table. In DTC control the torque and flux references have hysteresis comparators, that are used to select the inverter switching pattern based on torque and flux errors and an optimized switching table. Since the selection is done by comparators, the torque and flux output cannot be controlled to stay exactly at the reference, but they have a constant ripple around the reference. Microcontrollers of that era had little in the way of processing power so using a switching table instead of the heavy calculation required by FOC provided a noticeable decrease in the time needed to calculate references for a switching period. This lead to the possibility of increasing switching frequencies, and escpecially the control frequency of inverters. DTC became very popular in AC motor drives and is still widely used today. [2]
2.2 DC/DC-converter
A DC/DC converter is an electrical power conversion device used to convert between multiple DC- voltage levels. The operating principle differs between different converter topologies, but common to all DC/DC converters is the use of PWM as a way to control the output voltage and current. There are multiple qualities by which DC/DC-converters can be classified and in this case the classification is done by isolation and possibility of bi-directional energy flow. Figure 3 presents the most common DC/DC-converter topologies [1, 3, 4].
Figure 3: Most significant DC/DC-converter topologies
The basic operating principle of a Boost converter (noted ”a)” in figure 3) is to charge energy to magnetic
field of the inductor ”L1” when the switch ”Q1” is conducting. When ”Q1” is switched off, the energy in the inductor magnetic field tries to keep the current flowing through the inductor. Since the switch ”Q1”
is closed the current flows through the diode ”D1”, charging the output capacitor (”C2”) to a voltage higher than the input voltage.[1]
In the Buck converter (noted ”b)” in figure 3) the combination of voltage drop caused by current flowing through inductor ”L1” and PWM is used to decrease the output voltage. When the switch ”Q1” is conducting, the current flows through inductor ”L1” and charges the output capacitor,”C2”. When the switch is turned off, energy stored in the magnetic field of the inductor keeps the current flowing through the output capacitor and diode ”D1”.[1]
The half-bridge buck-boost converter (noted ”c)” in figure 3) is a combination of a boost converter and a buck converter. When it is used as boost converter, switch ”Q2” is modulated and switch ”Q1” is used as a diode. The energy flow will be from capacitor ”C1” to capacitor ”C2”, and the voltage level of capacitor ”C2” will be higher than the voltage of ”C1”. In the buck-mode switch ”Q1” is modulated, and switch ”Q2” will be used as a diode. The energy flow will be from capacitor ”C2” to capacitor ”C1”, and voltage level of ”C1” will be lower than voltage of ”C2”.[1]
In the full-bridge converter (noted ”d)” in figure 3) the transformer primary-side switches (”Q1” .. ”Q4”) are modulated to create a square-wave AC input voltage to the transformer ”TF1”. In this case the energy will be transferred from capacitor ”C1” to capacitor ”C2”. Transformer input voltage waveform is shown in figure 4. Full-bridge converter requires two control PWM’s, one for switches ”Q1” and ”Q4”, and another one for ”Q2” and ”Q3”. The control PWM’s have a 180 degree phase shift and they should have the same duty cycle to avoid DC-offset in the transformer input current. The transformer turns ratio determines the maximum output voltage, the PWM duty cycle can be used to control the output between zero volts and the maximum. The secondary side switches (”Q5” .. ”Q8”) can be be used as an active rectifier to increase efficiency. The direction of energy flow can be reversed by modulating the secondary side switches instead of the primary side switches.[1]
Figure 4: Full-bridge converter transformer input voltage
The flyback converter (noted ”e)” in figure 3) also uses a transformer to transmit power between the input and output of the converter. In the case when energy is transferred from capacitor ”C1” to capacitor
”C2”, the flyback converter stores energy into the leakage inductance of the transfromer primary when the switch ”Q1” is conducting. When the switch is turned off the energy is transferred to the secondary side through the transformer and the body diode of ”Q2” is used as a rectifier. Transistor ”Q2” can also be used as an active rectifier. Theoretically a flyback converter has an infinite voltage gain, therefore being able to produce any output voltage with any transformer turns ratio, but in real-life applications the ratio has to be set to match the input and output voltage specification to achieve the best possible efficiency. Modulating the switch ”Q2” instead of switch ”Q1” reverses the energy transfer direction.[1, 4]
Table 1 shows a comparison of beforementioned DC/DC converter topologies.
Table 1: Comparison of DC/DC-converter topologies Topology Isolated Bi-directional
Buck No No
Boost No No
Half-bridge buck-boost No Yes
Full-Bridge with transformer Yes Yes
Flyback Yes Yes
It can seen from table 1 that either isolated flyback converter or full-brigde converter with transformer, also known as H-bridge converter, is necessary to provide the required isolation between the high-voltage and low-voltage sides of the device. Both of these topologies need additional switches on the secondary side to allow bi-directional power flow, but the secondary-side switches can also be used as active rectifiers to increase efficiency.
The flyback topology was rejected because of more demanding component requirements, especially the low-side diode current capability.[3, 4]
Previous experience from DC/DC-converter design has shown that the efficiency of transformer based DC/DC-converters is considerably lower if the input/output voltage ratio is not fixed. The reason for the low efficiency in the case of a step-down converter comes from the fact that transformer ratio must be set so that the required output voltage can be created with the lowest possible input voltage. Since the output current is also a function on the transformer ratio, it causes the need to circulate lot of free- wheeling current in the primary side of the transformer to produce the required output current. This lead to decision of making the DC/DC-converter as a cascade of half-bridge buck-boost-converter and H-bridge-converter. With this solution the input voltage of the transformer can be kept at a fixed value with the boost-stage of the half-bridge converter.
2.3 Thermal design
Thermal design is a key part in the design process of any equipment. If thermal design is not well understood the resulting design might become unreliable due to insufficient cooling or expensive to manufacture due to overdimensioning of cooling components. Figure 5 shows the flowchart of the thermal design process.
Figure 5: Thermal design flowchart
As shown in figure 5 the thermal design of an electronic device starts with determining the electrical specifications for the device, in this case it means determining input and output voltage levels and the required output power. The required input and output currents are calculated based on the voltage and power specifications.
The cooling specification in figure 5 defines the type of cooling, in heavy duty work machines either liquid cooling or forced air cooling is commonly used. In this case the forced air cooling is specified.
The method for calculating losses depends on the current and voltage waveforms, see sections 5.1, 5.2 for example calculations with different waveforms.
The environment conditions specify for example the ambient temperature, water resistance and user protection requirements for the device. This can set limits for the cooling of the device. This part can sometimes be combined to the cooling specification section.
The current and voltage levels as well their waveforms give a set of requirements which the components have to meet. Components are selected based on these requirements, and their losses are calculated.
Heatsink selection is made based on the result of the loss calculations, and when heatsink has been selected the component temperatures are calculated based on the thermal resistances of the components
and the heatsinks.
If the component temperatures are calculated to stay within the specification the thermal design is finished, otherwise either component selection or heatsink selection needs to be revised to achieve required thermal performance. If the cooling is insufficient a selection between improving the heatsink or using different semiconductors is needed. Based on the authors experience, this is often a cost optimization task.
3 DESIGN SPECIFICATIONS
This device is made per specifications given by Darrox, but the frame of the device is made to be fitted in different configurations and there are additional specifications to those given by Darrox.
3.1 Customer specifications
The device specified by Darrox is a DC/DC-converter for charging batteries in cell phone towers. This device must have a converter connected to a windmill generator which provides the power for the DC/DC- converter, that charges the batteries. The Darrox windmill blades are designed so that they do not start automatically from standstill so this device must be able to spin the windmill up to speed with the energy from the battery. Electrical specifications given by Darrox are listed in table 2.
Table 2: Customer’s electrical specifications POUT 3,5 kW
UOUT 48 V
Additional specs from Darrox are that the device must be air cooled and weatherproof, which means IP class 54 or better.
3.2 Additional specifications
This device will be used by Visedo for mobile applications and therefore there are additional requirements to those of Darrox. Additional specifications given by Visedo are listed in table 3.
Table 3: Visedo’s additional specifications
POUT,DCDC 5 kW
POUT,Inverter 50 kW (liquid) or 5 kW (air)
UOUT,DCDC 12, 24 or 48 V
Intermediate DC-link voltage 60 ... 800 V (full power between 200 ... 800V)
Cooling Liquid or air
IP-class 67 (liquid) or 54 (air) [5]
Vibration T [6]
Configuration Dual inverter or Converter and DC-DC
Isolation Class I [7]
Operating temperature range -40 ... +100°C (liquid) or -40 ... +40°C (air) DC/DC-converter switching frequency 40 kHz
4 POWER ELECTRONICS DESIGN
The isolation requirements combined with the transformer considerations specified in section 2.2 lead to decision to design the DC/DC-converter part of the device as a cascade of half-bridge buck/boost converter and a full-brigde converter. The schematic for the main circuit of the device is shown in figure 6.
Figure 6: Main circuit schematic
4.1 Inverter
The inverter part of this device is a scaled down version of the Visedo PowerMASTER 250 heavy duty inverter. The IGBT-bridge is Semikron SKiiP 39AC12T4V1. The thermal calculations are done using the 5000 W output power with 95 % efficiency and 200 V DC-link voltage. The required input power is calculated in calculation 1.
Pgen,req= Pout,DCDC
ηnom
=5000W
0,95 = 5263,2W (1)
The maximum line-to-line RMS-voltage in the linear modulation region with 200V DC-link voltage is calculated in calculation 2.
ULL=200V
√2 = 141,421V (2)
The required generator current is calculated using the line-to-line voltage, power requirement and nominal power factor of the customer generator in calculation 3.
Igen,req= Pgen,req
√3×ULL×cos(ϕ) = 5263,2W
√3×141,421V×0,9722 = 22,101A (3) Current used in the thermal calculations is 23 A.
4.2 Brake chopper
The brake chopper is a safety function and therefore it is not intended to be run full-time, but only in the case of possible overvoltage. The specification for the brake chopper is 50 kW braking power with 800V DC-link voltage for short periods of time. The required current rating for the brake chopper IGBT is calculated in calculation 4.
Ibrake,max=Pbrake,max
UDC,max = 50kW
800V = 62,5A (4)
Since the brake chopper is not intended to be used continuously under normal operation of the device, the thermal calculations for the brake chopper IGBT are not included in the scope of this thesis.
4.3 Half-bridge converter
The Half-bridge converter is designed to transfer the full power (5 kW) of the DC/DC converter from 200 V input voltage for the Half-bridge converter to 800 V input for the Full-bridge converter. The nominal RMS value of the current is calculated in equation 5.
Iout,HB=Pnom,DCDC×ηnom1
Uin,HB = 5000W×0,951
200V = 26,316A (5)
Iripple= 0,6×Iout,HB= 0,6×26,32A = 15,789A (6) Dimensioning the inductor so that the ripple current stays below 60% requires calculating the nominal dutycycle of the DC/DC converter.
Dnom= 1− Uin,nom
Uout,nom
= 1−200V
800V= 0,75 (7)
Switching frequency for the DC/DC converter is specified to be 40 kHz, and the corresponding period is 25µs.
Lmin=Uin,nom× Dnom
Iripple×fsw
= 200V× 0,75
15,789A×40kHz = 238µH (8) The equation 8 shows that minimun value for the inductor is 238µH and for the calculations value of 250
µH is chosen.
4.4 Full-bridge converter
The transformer ratio is decided based on the required input and output levels so that the full power of the converter can be used throughout the full working voltage range. The maximum ratio for the transformer for the 48 V version is calculated in calculation 9.
T T Rmax= Uin,min
Uout,max
= 780V
60V = 13 (9)
If the transformer ratio is more than the maximum the converter will not be able to supply the specified output voltage with the specified minimum voltage. Therefore the ratio is decided to the value of 10.
The required inductance to achieve zero voltage switching cannot be reached with only the leakage inductance of the transfromer. A 50µH inductor (”L2” in figure 6) is added in series with the primary of the transformer to help reach ZVS with lower currents. ZVS can be achieved if the energy stored in the inductor is greater than then energy required to charge the output capacitance of the switching MOSFET. The energy of the input capacitance is calculated in calculation 10.
ECoss=1
2 ×Coss×UDC2 =1
2 ×120×10−12F×8002V = 3,84×10−5J (10) The amount of current above which the ZVS can be reached can be calculated by setting the energy of the indcutor equal to the energy of the MOSFET output capacitance and solving for current, this is done in calculation 11.
IZVS,min= s
Coss×UDC2 Lσ
= s
120×10−12F×8002V
50µH = 1,239A (11)
The output power corresponding toIZVS,minis calculated in equation 12.
PZVS,min=UDC×IZVS,min= 800V×1,239A = 991,5W (12)
5 THERMAL DESIGN
The thermal calculations are done individually for the inverter, and the different DC/DC-converter sec- tions. The results are then combined in the section 5.3.
5.1 DC/DC-converter
The DC/DC -converter is a cascade of Buck/Boost -converter and H-bridge converter. To ease calculations the two converter sections are calculated individually. The calculations present two thermal models. The IGBT thermal model is used for the original idea of using the same Semikron IGBT-module as in the inverter brigde and calculations of losses with discrete IGBTs from other brands. The MOSFET model is used to calculate the losses with SiC MOSFETs. Calculations include only the semiconductor losses for the high-voltage side.
5.1.1 Buck/Boost -converter with IGBT loss model
Previous experience has shown that the Buck/Boost -converter has higher requirements than the H-bridge and it was calculated first. This section covers the loss calculation model for IGBTs and section 5.1.2 covers the loss calculation model for silicon-carbide MOSFET transistors. The IGBT losses are calculated here for two IKW40N120H3 in parallel configuration. The losses are calculated for single IGBT and the multiplied with the number of components (2) to get the full losses. The IGBT loss model requires calculating the RMS- and average values for current. The RMS-value for per-IGBT current is calculated in calculation 13.
IIGBT,RMS= v
u u
tDnom×
"
ISingle+IRipple,Single
2
2
+IRipple,Single2
3 −(ISingle+IRipple,Single
2 )×IRipple,Single
#
= v
u u t0,75×
"
13,158A + 7,5A 2
2
+(7,5A)2
3 −(13,158A +7,5A
2 )×7,5A
#
= 11,548A
(13)
The RMS-value for the Diode current is calculated in calculation 14 IDiode,RMS=
v u u t
p1−Dnom×
"
ISingle+IRipple,Single
2
2
+IRipple,Single2
3 −(ISingle+IRipple,Single
2 )×IRipple,Single
#
= v u u t
p1−0,75×
"
13,158A +7,5A 2
2
+(7,5A)2
3 −(13,158A +7,5A
2 )×7,5A
#
= 9,429A
(14) The average value for IGBT current is calculated in calculation 15.
IIGBT,average =Dnom×ISingle= 0,75×13,158A = 9,868A (15) The average value for diode current is calculated in calculation 16.
IDiode,average= (1−Dnom)×ISingle= (1−0,75)×13,158A = 3,289A (16) Calculating the conduction losses of the IGBT requires estimating the internal lead resistance of the component. The lead resistance can be estimated by fitting a curve to Collector-Emitter voltage loss figure in the component datasheet[8]. Figure 7 shows the curvefitting with equation 17.
Vce(Ic) =Vce,0+Rlead,IGBT×Ic (17)
Figure 7: Collector-emitter voltage loss vs. current. Left-side is curvefitting, right-side from datasheet The conduction loss of the IGBT is calculated in equation 18.
Pcond loss, IGBT=Vce(IIGBT,average)×IIGBT,average+Rlead,IGBT×IIGBT,RMS,HB2
= 1,156V×9,868A + 31×10−3W×(11,548A)2
= 15,541W
(18)
The switching losses are calculated using the switching energy losses. These losses need to be estimated from the datasheet [8] curve. The curvefitting is shown in figure 8 and is calculated with equations 19 and 20.
Eon(Rg) = 0,0035 + 0,000098×Rg (19)
Eon(Rg) = 0,0022 + 0,00003×Rg (20)
Figure 8: Turn-on and turn-off energies vs. gate resistor. Left-side is curvefitting, right-side from datasheet
A 10 Ω gate resistor is selected and the turn-on energy loss is calculated in calculation 21 and turn-off energy is calculated in calculation 21.
Eon(Rg)[J] = 0,0035 + 0,000098×10J = 4,48×10−3J (21) Eoff(Rg)[J] = 0,0022 + 0,00003×10J = 2,5×10−3J (22) The switching energies are scaled from the datasheet voltage and current values to the values selected for this design, the linear scaling is shown in calculations 23 and 24.
Eon[J] =Eon(Rg)× UDC,actual
UDC,datasheet×
ISingle−IRipple,Single
2
Idatasheet
= 4,48×10−3J×800V 600V×
13,158A−7,5A2 40A
= 1,405×10−3J
(23)
Eoff[J] =Eoff(Rg)× UDC,actual
UDC,datasheet ×
ISingle+IRipple,Single
2
Idatasheet
= 2,5×10−3J×800V 600V×
13,158A + 7,5A2 40A
= 1,409×10−3J
(24)
The switching power losses are calculated from the switching energies in calculation 25.
Psw loss, IGBT= (Eon+Eoff)×fsw= (1,405×10−3J + 1,409×10−3J)×40000Hz = 112,556W (25) Total losses per IGBT are calculated in calculation 26
Ptotal,IGBT =Pcond loss, IGBT+Psw loss, IGBT= 15,541W + 112,556W = 128,098W (26)
Calculating the condution losses for the internal diode of the component requires determining the forward voltage (Vf) and lead resistance of the diode. This is done by fitting a linear curve to theVf -curve in the datasheet [8]. The curve in figure 9 is calculated with equation 27.
Vf(If) =Vf,0+Rlead×If (27)
Figure 9: Diofe forward voltage vs. current. Left-side is curvefitting, right-side from datasheet The condution losses for the diode are calculated similarly to the IGBT losses in calculation 18, the result is shown in calculation 28
Pcond loss,diode=Vf(IDiode,average)×IDiode,average+Rlead×IDiode,RMS,HB2
= 1,022V×3,289A + 0,037W×(9,429A)2
= 6,651W
(28)
The switching energy for the diode is estimated from the reverse recovery charge given in the datasheet [8]. It is shown in calculation 29.
Esw, diode=1
2 ×Qrr×UDC=1
2 ×4,3×10−6C×800V = 1,72×10−3J (29) The switching losses for the diode are calculated in calculation 30.
Psw loss,diode=Esw,diode×fsw= 1,72×10−3J×40000Hz = 68,8W (30) The total losses for the diode are calculated in calculation 31.
Ptotal,diode =Pcond loss,diode+Psw loss,diode= 6,651W + 68,8W = 75,451W (31) The total losses for the semiconductors in the boost converter are calculated in the calculation 32.
Ptotal,boostIGBT=nIGBT×(Ptotal,IGBT+Ptotal,diode) = 2×(128,098W + 75,451W) = 407,096W (32) The Boost-converter losses with this calculation model were considered for components in listed table 4.
Due to relatively high switching losses for the internal diode in IKW40N120H3 IGBT, a calculation was
also conducted for IGW40N120H3 IGBT and IDW30S120 silicon-carbide diode. The IGW40N120H3 is the same IGBT as IKW40N120H3 but without the internal diode.
Table 4: Comparison of Boost-converter losses with different components Component / Technology Losses η SKiiP39AC12T4V1 / Si IGBT module 831,484 W 85,7 %
IKW40N120H3 / Si IGBT 407,096 W 92,5 %
IGW40N120H3 / Si IGBT and IDW30S120 / SiC Diode 271,39 W 94,9 %
5.1.2 Buck/Boost -converter with MOSFET loss model
The losses are calculated using the MOSFET loss model with the same current and voltage values as in section 5.1.1. This model is also calculated for two parallel MOSFETs, calculating the losses for a single component and the multiplying to get the full losses. For MOSFET semiconductors the conduction losses are determined by the On-state resistance of the component which can be found from the component datasheet[9]. For worst-case calculation, the maximum value is chosen. The losses are calculated in calculation 33.
Pcond loss, MOSFET=RDS,on×IMOSFET,RMS2 = 120×10−3W×(11,548A)2= 16,004W (33) The switching losses of MOSFET the consist of the linear losses of current times voltage during the rise and fall times of switching and the and charging and discharging of the output capacitor. The switching energies are calculated in equation 34.
Esw, MOSFET= IMOSFET,RMS×UDC×(trise+tfall)
2 +Coss×UDC2
2
= 11,548A×800V×(97×10−9s + 75×10−9s)
2 +120×10−12F×(800V)2
2
= 7,945×10−4J + 3,84×10−5J
= 8,329×10−4J
(34)
The switching power losses are calculated in calculation 35.
Psw loss,MOSFET=Esw,MOSFET×fsw= 8,329×10−4J×40000Hz = 33,317W (35) The total losses for the MOSFET are shown in calculation 36.
Ptotal,MOSFET=Pcond loss, MOSFET+Psw loss,MOSFET= 16,004W + 33,317W = 49,321W (36) The diode conduction losses are calculated the same way as with the IGBT model, the curve for the internal diode forward voltage is not provided in the component datasheet[9], and therefore no curvefitting can be done. Conduction losses for the diode are calculated using the forward voltage value provided in the datasheet, this is shown in calculation 37.
Pcond loss,diode=Vf(IDiode,average)×IDiode,average+Rlead×IDiode,RMS,HB2
= 3,5V×3,289A + 0,18W×(9,429A)2
= 27,517W
(37)
The switching loss for the internal diode of the component can be calculated the same way as in the IGBT-model. This is calculated with the reverse recovery charge given in the datasheet [9]. This is shown in calculations 38 and 39.
Esw, diode= 1
2 ×Qrr×UDC= 1
2 ×142×10−9C×800V = 5,68×10−5J (38)
Psw loss,diode=Esw,diode×fsw= 5,68×10−5J×40000Hz = 2,272W (39) Total losses for the Boost-converter diode are shown in calculation 40.
Ptotal,diode=Pcond loss,diode+Psw loss,diode= 27,517W + 2,272W = 29,789W (40) The total losses for the number of components in the converter are shown in calculation 41
Ptotal,boost,MOSFET=nMOSFET×(Ptotal,MOSFET+Ptotal,diode)
= 2×(49,321W + 29,789W)
= 158,219W
(41)
The efficiency of the Boost-converter with silicon-carbide MOSFETs is shown in calculation 42.
ηboost,MOSFET= Pout,Boost
Pout,Boost+Ptotal,boost,MOSFET
= 5000W
5000W + 158,219W= 96,9% (42) 5.1.3 H-bridge -converter with IGBT loss model
The output current for the low-voltage side is calculated for 48 V version with the nominal output power of 5 kW in equation 43.
Iout,lv= Pout
Uout =5000W
48V = 104,17A (43)
With the transformer ratio 10, the primary current is calculated in calculation 44.
Iprimary,max= Iout,LV
T T R =104,17A
10 = 10,417A (44)
In order to estimate the maximum RMS-current for each transistor, a calculation of the duty cycle of the full-bridge converter is needed. The converter will be run with dutycycle of 0,5 but the actual dutycycle of the output current is less than this because it takes time to change the direction of the current in the transformer primary. To estimate this it is necessary to calculate the rate at which the current can be changed with the available voltage. This is done in calculation 45
dI
dt =UDC
Lσ
=800V
50µH = 1,6×107A/s (45)
The actual duty-cycle is estimated in calculation 46.
DFB=Dnom×
tperiod−4×Iprimary,max dI dt
tperiod
= 0,5×25µs−4×1,610,417A×107A/s
25µs = 0,448 (46)
The output current of the converter can be estimated to be square wave and the RMS-value of the current is calculated in calculation 47
IRMS,FB,IGBT=Iprimary,max×DFB= 10,417A×0,448 = 4,666A (47) The free-wheeling diodes conduct the current during the off-times of the IGBTs, the RMS-value of the current is calculated in calculation 48.
IRMS,FB,diode= Iprimary,max
2 ×
Iprimary,max dI dt
tperiod
= 10,417A
2 ×
10,417A 1,6×107A/s
25µs = 0,136A (48) The conduction losses for the IGBT are calculated using the same curvetting as is equation 18 in section 5.1.1. The conduction losses are shown in calculation 49.
Pcond loss,IGBT,FB=Vce(IRMS,FB,IGBT)×IRMS,FB,IGBT+Rlead,IGBT×IRMS,FB,IGBT2
= 0,995V×4,666A + 31×10−3Ω×(4,666A)2
= 5,316W
(49)
Curvefitting equation for the IGBT switching energies from the Buck/Boost model (equation 20 from 5.1.1) is also applicable for the H-bridge. The H-Bridge will be run with zero voltage switching, and therefore only the off-switch losses need to be calculated for the IGBT. This is done in calculations 50 and 51.
Esw,IGBT=Eoff(RG) = 3,888×10−4J (50)
Psw,FB,IGBT =Esw,IGBT ×fsw= 3,888×10−4J×40×103Hz = 15,553W (51) Equations 27 and 28 from section 5.1.1 are also applicable for the H-bridge diode conduction losses, which are shown in calulation 52.
Pcond. loss, FB, diode= 0,905V×0,136A + 0,037Ω×(0,0136A)2= 0,123W (52) The switching losses shown in calculations 29 and 30 in section 5.1.1 are independent of current and the same value can be used for the diode switching losses in this case. The complete losses of the H-bridge converter are shown in calculation 53.
Ptotal loss,FB= 4×(Pcond.loss,FB,IGBT+Psw,FB,IGBT+Pcond.loss,FB,diode+Psw,FB,diode)
= 4×(5,316W + 15,553W + 0,123W + 68,8W)
= 383,038W
(53)
Calculation model presented above was used for estimating losses for the IGBT’s listed in table 5 Table 5: Comparison of H-bridge-converter losses with different components
Component / Technology Losses η SKiiP39AC12T4V1 / Si IGBT module 1447 W 77,5%
IKW40N120H3 / Si IGBT 383,038W 92,9 %
5.1.4 H-bridge -converter with MOSFET loss model
The current waveforms and RMS-values used for the IGBT calculations in section 5.1.3 are also used for the MOSFET model. The conduction losses for the primary-side MOSFETs are calculated in calculation 54.
Pcond. loss,FB,MOSFET=Rds,on×IRMS,FB,MOSFET= 120−3W×4,666A = 2,612W (54) The switching losses are calculated with equations 34 and 35 from the section 5.1.2. Due to ZVS, only the fall time of the MOSFET is used in the switching loss calculation. The result is shown in calculations 55 and 56.
Esw,FB,MOSFET=4,666A×800V×(75×10−9s)
2 +120×10−12F×(800V)2 2
= 1,4×10−4J + 3,84×10−5J
= 1,7837×10−4J
(55)
Psw loss,FB,MOSFET= 1,7837×10−4J×40000Hz = 7,135W (56) The total losses for the full-bridge MOSFETS are calculated in calculation 57.
PTotal loss,FB,MOSFET=Pcond. loss,FB,MOSFET+Psw loss,FB,MOSFET
= 2,612W + 7,135W
= 9,747W
(57)
Diode conduction losses are shown in calculation 58.
Pcond loss,FB,diode =Vf(IDiode,FB,RMS)×IDiode,FB,RMS+Rlead×IDiode,FB,RMS2
= 3,5V×0,136A + 0,18W×(0,136A)2
= 0,478W
(58)
With the silicon-carbide semiconductors the switching losses are independent of current, and therefore the losses calculated in equation 39 from section 5.1.2 can also be used for the full-bridge converter. The total losses for a single full-brgide diode are shown in calculation 59.
Ptotal loss,FB,diode=Pcond loss,FB,diode+Psw loss,FB,diode= 0,478W + 2,272W = 2,75W (59) The total losses of the high-voltage side of the H-bridge converter are calculated in the calculation 60.
Ptotal loss,FB=ncomponents×(PTotal loss,FB,MOSFET+Ptotal loss,FB,diode)
= 4×(9,747W + 2,75W)
= 49,99W
(60)
H-bridge converter efficiency is shown in calculation 61.
ηFB= Pout,FB
Pout,FB+Ptotal loss,FB
= 5000W
5000W + 49,99W = 99.01% (61)
5.2 Inverter brigde
The losses in the inverter are calculated in the same way as in the DC/DC converter with igbt model in section 5.1, but in the case of the inverter bridge the losses need to be integrated over the whole period of the output sine wave. This produces the average losses of the inverter. The peak losses are greater and due to that, the low output frequencies need to be considered separately. The output of the inverter is assumed symmetrical and therefore it is enough to calculate losses for a single IGBT and diode pair, and then multiply the result by six.[10]
For this calculation output frequency (fe) of 50 Hz is selected. The corresponding period (te) is 20 ms.
As can be seen from the main schematic in figure 6 the inverter bridge consists of six pairs of IGBTs and inverse diodes, each of the pairs sees identical waveform at a different phase. The figure 10 shows the output current waveform and the dutycycles for one IGBT and diode.
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
−35
−30
−25
−20
−15
−10
−5 0 5 10 15 20 25 30 35
Current [A]
Time [s]
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Duty
Current IGBT dutycycle IGBT dutycycle diode dutycycle diode dutycycle
Figure 10: Current waveform and switching component dutycycles The current of a single phase of the inverter as a function of time is shown in equation 62.
iinverter(t) = ˆi×sin(ω×t) =√
2×iRMS×sin(2×π×fe×t) (62) As can be seen from figure 10, the positive part of the sine flows through the IGBT and the negative part flows through the inverse diode. For calculations the IGBT part and the diode part are separated. To estimate the switching losses of the IGBT, the switching energies need to be calculated from the curve in the component datasheet [11]. The curvefitting is done in the same way as in section 5.1.1 and it’s shown in figure 11.
Figure 11: Curvefitting for the IGBT switching energies, left side calculated, right side datasheet Gate resistor value of 2 Ω is selected, which results in turn on and turn off energies of 25mJ and 14mJ respectively. The linear scaling for the energies used in section 5.1.1 cannot be used directly because the inverters current is sinusoidal instead of DC. The modified functions for the turn on and turn off energies are shown in equations 63 and 64.
Eon,IGBT(t) =Eon, IGBT, fitted× UDC
Udatasheet×iinverter(t) Idatasheet
= 25mJ×200V 600V× i(t)
150A (63)
Eoff,IGBT(t) =Eoff, IGBT, fitted× UDC
Udatasheet ×iinverter(t) Idatasheet
= 14mJ×200V 600V× i(t)
150A (64)
Based on the datasheet [11], the switching energy of the IGBT at zero current is negilibible and in this calculation it is ignored. The function for the IGBT switching losses is shown in equation 65.
Psw loss,IGBT,inv(t) =
(Eon,IGBT(t) +Eoff,IGBT(t))×fsw , t= 0..tperiod
2
0 , t= tperiod
2 ..tperiod
(65)
The average switching losses for the IGBT can be calculated by integrating the switching loss function over one period of the output current wave and then multplying the result with the output frequency.
Only the positive part needs to be calculated. This is shown in calculation 66.
Psw loss,IGBT,inv
=fe× Z
tperiod 2
0
Psw loss,IGBT,inv(t)dt
=fe× Z
tperiod 2
0
(Eon,IGBT(t) +Eoff,IGBT(t))×fswdt
= 50Hz×(25mJ + 14mJ)×200V 600V× 1
150A×√
2×23A×8000Hz
× Z 20ms2
0
sin(2×π×50Hz×t)dt
= 50Hz×(25mJ + 14mJ)×200V 600V× 1
150A×√
2×23A×8000Hz
×
−cos(2×π×50Hz×10×10−3s)
2×π×50Hz −−cos(2×π×50Hz×0s) 2×π×50Hz
= 7,179W
(66)
The losses for the IGBTs and diodes in the inverter bridge are shown in figure 15. The collector-emitter voltage drop (Vce) is needed to calculate conduction losses for the IGBT, in order to estimate Vce the lead resistance of the component is needed, equation 17 from section 5.1.1 is used. Figure 12 shows the curvefitting for the component.
Figure 12: Vcecurvefitting for SKiiP39AC12T4V1 IGBT, left side calculated, right side from datasheet The function for theVce is shown in equation 67.
Vce(t) =Vce,0+Rlead×i(t) = 0,9V + 8,4mΩ×√
2×23A×sin(2×π×50Hz×t) (67) The conduction losses for the IGBT are intergrated over half of the period of the output wave because
only the positive part of the wave is needed. This is done in calculation 68.
Pcond loss, IGBT,inv
=fe× Z
tperiod 2
0
[Vce,0+Rlead×i(t)]×i(t)dt
=fe×
"
2׈i2×Rlead×tperiod2 ×ω−ˆi2×Rlead×sin(2×ω×tperiod2 )−4׈i×Vce,0×cos(ω×tperiod2 ) 4×ω
−2׈i2×Rlead×0×ω−ˆi2×Rlead×sin(2×ω×0)−4׈i×Vce,0×cos(ω×0) 4×ω
#
= 50Hz×[0,138J−(−0,093J)]
= 11,54W
(68) The inverse diode conducts when the phase current is negative, for calculations the absolute value is used.
For the conduction losses the curvefitting for the diode forward voltage (Vf) is needed, figure 13 shows the curvefitting.
Figure 13: Vfcurvefitting for SKiiP39AC12T4V1 inverse diode, left side calculated, right side datasheet The function for the forward voltage used in loss calculations is shown in equation 69.
Vf,Diode,inv=
0 ,0≤t < tperiod
2 Vf,0+Rlead× |iinverter(t)| ,tperiod
2 ≤t≤tperiod
(69)
The conduction losses for the diode are calculated in equation 70. The diode conducts when the phase
current is negative and therefore only that part of the period is calculated.
Pcond loss, Diode, inv= Z tper
tper 2
Vf,Diode,inv× |i(t)|dt×fe
= Z tper
tper 2
[Vf,0+Rlead×(−i(t))]×(−i(t))dt×fe
=
"
Rlead× Z tper
tper 2
i(t)2dt−Vf,0× Z tper
tper 2
i(t)dt
#
×fe
=
"
Rlead׈i2× Z tper
tper 2
sin2(ω×t)dt−Vf,0׈i× Z tper
tper 2
sin(ω×t)dt
#
×fe
= [0,034J−(−0,186J)]×50Hz
= 11,038W
(70)
Estimating the switching losses in the inverter diodes requires curvefitting for the switching energy, this is done in figure 14.
Figure 14: Eon andEoff curvefitting for SKiiP39AC12T4V1 inverse diode, left side calculated, right side datasheet
The equation for the switching energy of the diode is shown in equation 71.
Esw,Diode,inv(t) =
UDC
UDC,datasheet ×(Err+ 0,0015)×e−500400 ,0≤t≤tperiod
2 UDC
UDC,datasheet ×(Err+ 0,0015)×e−i−500(t)+400 ,tperiod
2 < t≤tperiod
(71)
The switching power loss for the diode is calculated in calculation 72.
Psw,Diode,inv=
"
Z
tper 2
0
UDC
UDC,datasheet×(Err+ 0,0015)×e−500400 dt
!
×fsw
+ Z tper
tper 2
UDC
UDC,datasheet×(Err+ 0,0015)×e−i(t)×400−500 dt
!
×fsw
#
×fe
= 2,4899×10−5J×8000Hz + 2,6472×10−5J×8000Hz
×50Hz
= 20,548W
(72)
The total losses of the inverter bridge are calculated in calculation 73.
PTot. loss, inv= 6×(Psw loss,IGBT,inv+Pcond loss, IGBT,inv+Pcond loss, Diode, inv+Psw,Diode,inv)
= 6×(7,179W + 11,54W + 11,038W + 20,548W)
= 294,82W
(73)
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
0 10 20 30 40 50 60 70 80 90
time [s]
Loss [W]
IGBT average switching loss IGBT instantaneous switching loss IGBT average conductions loss IGBT instantaneous conduction loss IGBT average total losses IGBT instantaneous total losses Diode average conduction losses Diode instantaneous conduction losses Diode average switching losses Diode instantaneous switching losses Diode average total losses Diode instantaneous total losses Leg average total losses Leg instantaneous total losses
Figure 15: Losses for individual IGBT’s and diodes in the inverter bridge
5.3 Whole system
The following table lists the power losses with different component selections. The comparison is for the DC/DC- converter, therefore the inverter bridge losses are for the Semikron SKiiP39AC12T4V1. The low-voltage active rectifier (LV-bridge) is not in the scope of this thesis, the losses are coarsely estimated with calculations 74, 75 and 76 according the rectifier mosfet datasheet [12]. This calculation assumes two parallel mosfets for each four switches of the rectifier and component temperature of 150°C.
PCond. loss,rectifier=Dnom×
"
Iout
2 2
×Rds,ON,150
#
×nMOSFETs
= 0,5×
"
100A 2
2
×8,8×10−3Ω
#
×8 = 88,0W
(74)
PSW. loss,rectifier=nMOSFETs×
"I
out
2 ×Uout×tfall×fsw
2 +Coss×Uout2 ×fsw
2
#
= 8×
"100A
2 ×48V×14×10−9s×40000Hz
2 +1140×10−9F×(48V)2×40000Hz 2
#
= 5,796W
(75)
PTotal loss,rectifier= 88,0W + 5,796W = 93,80W (76) In the case of SKiiP39AC12T4V1, the losses are calculated so that 2 IGBT’s of the module are used for the Buck-Boost and the remaining four for the H-bridge. The inductor losses are estimated by the inductor manufacturer, datasheets of these inductors are proprietary to Visedo.
Table 6: Comparison of total system losses with different components
Component / Technology Buck- Boost
H-brigde Inverter LV-bridge Inductor losses
Total η
SKiiP39AC12T4V1 / Si IGBT module
831,48 W 1447 W 294,82 W 93,80 W 102,0 W 2769,10 W 64,4 % IKW40N120H3 / Si IGBT 407,09 W 383,04 W 294,82 W 93,80 W 102,0 W 1280,75 W 79,6 % CMF20120D / SiC MOSFET 158,22 W 49,99 W 294,82 W 93,80 W 102,0 W 698,83 W 87,7 %
6 MECHANICS DESIGN
This section describes the mechanical design of the multiconverter. The main focus is the required cooling for the semiconductors, and only parts of the design relevant to cooling are presented. Table 7 lists specifications used to determine the maximum allowed thermal resistance for the heatsink. The thermal resistance of the SIL-PAD insulator is 0,3 K-in2/W [13], and the area of TO-247 case is 0,508 square-inches [9].
Table 7: Thermal specifications of semiconductors
Component Rth
CMF20120D MOSFET 0,51 K/W CMF20120D Diode 0,51 K/W SKiiP39AC12T4V1 IGBT 0,33 K/W SKiiP39AC12T4V1 Diode 0,52 K/W SIL-PAD thermal pad 0,591 K/W
The maximum temperature of the enclosure of the converter (Tcase,max) is determined first. For this calculation maximum allowed semiconductor junction temperature (Tj,max) of 130 °C is chosen. The maximum temperature is calculated for each of the semiconductors separately with equation 77 [14].
Tcase,max=Tj,max−Rth,component×Ploss,component (77) An electrical insulator is needed between CMF20120D MOSFET and the frame. The thermal resistance of the insulator is added to the thermal resistance of the component.Tcase,max-values for each semiconductor component are shown in table 8.
Table 8: Maximum allowed case temperatures Component Tcase,max
Buck-Boost MOSFET 75,74°C Buck-Boost Diode 97,23°C H-bridge MOSFEET 119,28°C
H-bridge Diode 126,975°C Inverter IGBT 123,82°C Inverter Diode 113,575°C
As can be seen in table 8, the limiting component in this case is the Buck-Boost-converter MOSFET.
To ensure proper function of the device, a maximum case temperature of 70°C is chosen. Calculation 78 shows the maximum allowed thermal resistance between the heatsink and cooling air. Ambient tem- perature for this calculation is 40°C. The total losses for the converter are shown in table 6 in section 5.3.
Rth,c-a,max= Tcase,max−Tambient
Ptotal
= 70°C−40°C
698,83W = 0,043K/W (78)
The final product will have an aluminum molded enclosure which includes the required heatsink, but for prototyping a machined enclosure with external heatsink is used. Three Pinbloc 98x98x45 heatsinks connected with Bergquist GF3500 [15] thermal interface material are used for the prototypes. The TO- 247 components (SiC mosfets, brake IGBT and brake free-wheeling diode) are cooled using an extra aluminum heatsink, which is connected to the case with the GF3500 thermal compound. Figure 16 shows the internal view of the device with TO-247 heatsink highlighted in the top left corner of figure.
Figure 16: Internal view of the device
Passive component captionings in figure 16 correspond to captionings in figure 6 in section 4. The yellow component in top right corner of the figure is the inverter IGBT module. Busbars between the transformer secondary and active rectifier (which is on the other side of the device and therefore not shown in figure) are highlighted in red. The two SOT-227 components between the inverter IGBT and TO-247 heatsink are DC-link discharge resistors, both the inverter DC-link and DC/DC-converter DC-link have their own discharge circuits. Discharge circuits activate when the device is powered down, and the purpose is to make the device voltage free for maintenance and installation purposes.
7 RESULTS
Calculations were verified by measuring input and output powers of the device and calculating efficiency from these. The complex structure of the device limited measurement possibilities to only measuring the total efficiency of the device. Due to test setup limitations the efficiency was measured only for the DC/DC-converter without the inverter bridge. Due to limited output current of the power supply used for testing, tests were conducted at 450 V input voltage. Output for the device was set at 48 V and load current was set at 100 A. Since the test point is different from the calculation point, DC/DC-converter losses were re-calculated for this test point and they are shown in table 9.
Table 9: Calculated losses at measured test point
Buck-Boost H-brigde LV-bridge Inductor losses Total η
64,79 W 37,24 W 93,80 W 71,37 W 267,20 W 94,7 %
Equipment used for the efficiency measurement are their measurement uncertainties are listed in table 10.
Table 10: Equipment used for testing
Device Description Measurement error Agilent MSO7054A Mixed signal oscilloscope ±2% of full scale [16]
Agilent N2780B Current probe ±(1% of reading + 500 mA) [17]
Fluke 80i-110s Current probe ±(4% of reading + 50 mA) [18]
Testec SI 9110 Differential voltage probe ±2% [19]
Definition of full scale for the oscilloscope depends on the selected measurement range, and it is defined as number of divisions on screen (8 in the case of MSO7054A) and the scale of division. Measurement error for each quantity is the combined error of the oscilloscope and probe used, and these are shown in table 11. The error percentages are calculated relative to average value of measurement data.
Table 11: Measurement error for each measured quantity
Quantity Oscilloscope scale Probe error Oscilloscope error Total error
Input current 1 A/div. ±0,5146 A ±0,16 A ±5,81 %
Input voltage 20 V/div. ±8,9630 V ±3,20 V ±2,71 %
Output current 10 A/div. ±1,4920 A ±1,60 A ±3,12 %
Output voltage 5 V/div. ±0,9626 V ±0,80 V ±3,66 %
Since calculation for the DC/DC-converter losses were conducted with average values, the measurement results are filtered with a moving average filter over the length of five switching periods. The sampling rate used for the oscilloscope was 50 MHz, and the corresponding filter length is 6250 samples. From the averages, a section of ten switching cycles is plotted in figures 17 - 20. Figure 17 shows the input current and voltage of the device. Output voltage and current are shown in figure 18. Input and output powers are shown in figure 19 and measured efficiency in figure 20. The average values for the measurements compared to calculated averages are shown in table 12.
0 25 50 75 100 125 150 175 200 225 250 10
11 12 13 14 15
Current [A]
Time [s]
440 445 450 455 460
Voltage [V]
Input current Input voltage
Figure 17: Measured input voltage and current
0 25 50 75 100 125 150 175 200 225 250
95 96 97 98 99 100 101 102 103 104 105
Current [A]
Time [s]
45 46 47 48 49 50
Voltage [V]
Output current Output voltage
Figure 18: Measured output voltage and current
0 25 50 75 100 125 150 175 200 225 250 4000
4200 4400 4600 4800 5000 5200 5400 5600 5800 6000
Time [µs]
Power [W]
Input power Output power
Figure 19: Measured input and output powers of the device
0 25 50 75 100 125 150 175 200 225 250
80 82 84 86 88 90 92 94 96 98 100
Time [µs]
Efficiency [%]
Instantaneous efficiency Average efficiency Calculated efficiency
Figure 20: Measured efficiency
