wide bandgap semiconductors -...
TRANSCRIPT
1
Wide Bandgap Semiconductors
Burak Ozpineci
Oak Ridge National Laboratory
2
Why not Si?
Today ⇒ All Si-based power semiconductor switches
• Limited breakdown voltages and limited power ratings• Limited operation temperature (<150°C)• Limited switching frequency (≤ 20kHz) for power levels
of more than a few tens of kW
∴ The current Si technology is approaching the material’s theoretical limits, and it cannot meet all the requirements of the transportation industry
3
Why Wide Bandgap Semiconductors
• Advantages– Wide bandgap semiconductor based power
devices,• Operate at higher temperatures, 600 °C (SiC) reported
compared to 150°C of Si ⇒ less cooling requirement; saves space and weight; no heatsink? (1/3 the converter size)
– They have a higher electric breakdown field; much higher doping levels can be achieved; therefore they are
• Thinner ⇒ occupy less space• Low on resistance (~a few hundred times less)
⇒ lower conduction losses, and higher efficiency• Higher breakdown voltages (no need to connect devices in
series)
4
Why Wide Bandgap Semiconductors (cont’d)
– They have higher thermal conductivities ⇒ lower thermal resistance; the device dissipates heat to the surrounding faster
– Forward and reverse electrical characteristics vary only slightly with temperature and time ⇒ reliable
– Excellent reverse recovery characteristics ⇒ less EMI; less snubbing required; lower switching losses; higher efficiency
◊ High temp operation capability and lower switching losses ⇒ high frequency operation (more than 20 kHz at > 1 MW) ⇒ less filtering; smaller passive components; saves space
∴ WBG semiconductor-based converters are compact, light, reliable, efficient, and have a high power density
5
Physical Characteristics
2.72.22211Saturated Electron Drift Velocity, vsat (×107 cm/s)
221.34.94.90.461.5Thermal Conductivity, λ(W/cm⋅K)
850850115101400600Hole Mobility, µp (cm2/V⋅s)
220012501000 5008085001500Electron Mobility, µn
(cm2/V⋅s)
10000200022002500400300Electric Breakdown Field, Ec(kV/cm)
5.5910.19.6613.111.9Dielectric constant, εr1
5.453.453.263.031.431.12Bandgap, Eg (eV)
DiamondGaN4H-SiC6H-SiCGaAsSiProperty
6
Figures of Merit
1426711560.5458.1748.91.41.0BTFM
59410.735.457.30.91.0BPFM
53044591973.63424.81470.540.71.0FTFM
147630.456.048.33.61.0FPFM
240252.512.913.11.61.0BSFM
359565.061.230.511.41.0FSFM
25106186.7223.1125.314.81.0BFM81000215.1215.1277.81.81.0JFM
DiamondGaN4H-SiC6H-SiCGaAsSi
7
Breakdown Voltage
d
crB qN
EV2
2ε≈
where
q is the charge of an electron andNd is the doping density
8
Drift RegionWidth Resistance
( )c
BB E
VVW 2≈ ( )
ncs
Bspon E
VR
µε 3
2
, )(
4=
To calculate the device on resistance, contact resistance and channel resistance are also required.
9
SiC
• Most mature technology among WBG semiconductors
• SiC Schottky diodes are commercially available
• Many companies and universities all over the world are working on SiC power devices: 4H-SiC and 6H-SiC PiN diodes, Schottky diodes, IGBTs, thyristors, BJTs, various MOSFETs, GTOs, MCTs, MTOs, etc. in kV range with reduced on-resistances.
10
SiC Availability
• As of today, only four companies have advertised the commercial availability of SiC power devices:• Infineon (Schottky diodes, 600V up to 12A or
300V up to 10A)• Microsemi (Schottky diodes, 200/400/600V,
1A/4A)• Cree (Schottky diodes, 600V up to 20A)• IXYS (H-bridge Schottky diodes, 600V 3A/5A)
– Experimental controlled SiC VJFET cascodeswitches (4A max) can be obtained from SiCED of Germany.
11
GaN
• GaN device applications focus on optoelectronics and radio frequency uses
• Some experimental GaN Schottky diodes available
• Better reverse recovery compared with SiC; lower switching losses
• Higher forward voltage drop compared with SiC because of wider bandgap; higher conduction losses (diodes)
12
GaN compared with SiC
Malay Trivedi and Krishna Shenai, J. Appl. Phys., Vol. 85, No. 9, 1 May 1999
13
GaN compared with SiC (cont’d)
• GaN does not have a native oxide (for MOS devices); SiC uses the same oxide as Si, SiO2.
• GaN boules cannot be grown; instead GaN is grown on sapphire or SiC.
• Thick GaN substrates are not commercially available; GaN wafers are more expensive.
• GaN thermal conductivity is one-fourth of that of SiC
14
Diamond
• Best theoretical performance.• Its processing problems have not been
solved yet.• Processing is more difficult than the
other WBG materials• In the literature, diamond is used in
only sensors and field emission devices.
15
Conclusions• WBG materials have the promise to surpass Si in
the near future.• Diamond is the ultimate material for the power
devices in utility applications. Diamond power devices are expected to be abundant in 20-50 years.
• GaN and SiC power devices have similar performance improvements compared to Si power devices; however, they both have processing issues.
• SiC power devices are at a more advanced stage than GaN power devices.
• SiC is the best suitable transition material for future power devices.
16
System Impact of Silicon Carbide Power Electronics on Hybrid Electric Vehicle
Applications
Burak Özpineci
Electrical and Computer Engineering DepartmentThe University of Tennessee, Knoxville
17
Outline
SiC MATERIALDEVICESSYSTEMS
CONCLUSIONS
18
Transportation Requirements
• Compactness• Lightweight• High power density• High efficiency• High reliability under harsh conditions• Low cost
19
SiC compared with Si (cont’d)
• Disadvantages– Low processing yield – micropipes (<1/cm2)– Expensive - $7 for 600 V, 4 A SiC Schottky diode
(similar Si pn diode << $1)– Limited availability – High temperature packaging techniques not yet
developed– Compatible passive components and gate drivers do
not exist
20
SiC vs. Si diode I-V characteristics
0.6 0.8 1 1.2 1.4 1.6 1.70
1
2
3
4
5
6
7
Diode Forward Voltage, V
Dio
de F
orw
ard
Cur
rent
, A
Si SiC
0.5
Arrows point at the increasingtemperature 27-250C
R
DUTVdc
IF
+
VF
-
IDUT
CurrentProbe
oven
• Forward voltage drop of Si diode is smaller than that of the SiC diode at the same forward current.
• I-V characteristics of the SiC diodes do not vary much with temperature
21
SiC vs. Si diode I-V characteristics (cont’d)
• Find the diode parameters IS , RS , and n using the genetic algorithm (GA) curve fitting method
• Use the piece wise linear (PWL) approximate model of a diode and find the PWL parameters, RD and VD using GA curve fitting method
( )
−=
− 1nkTIRVqs
seII
A
K
VD
RD
+
vF
-
iF
vF
iF
1
RD
VD
22
SiC vs. Si diode PWL Model
0 50 100 150 200 2500
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Toven, °C
V D, V
SiC
Si
• RD and VD expressed as functions of temperature
7042.02785.0 0046.0 += − TSiCD eV
2023.01108.0 0072.0 +−= − TSiCD eR
5724.03306.0 0103.0 += − TSiD eV
0 50 100 150 200 2500
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Toven, °C
RD, Ω
SiC
Si
0529.02136.0 0293.0 += − TSiD eR
• VD of the Si diode is smaller than that of the SiC diode at any temperature.
• Si : RD decreases • SiC: RD increases with temperature
23
SiC vs. Si diode conduction lossesDrmsDDavDcond RIVIP ⋅+⋅= 2
,,Diode conduction loss expression
DDCDDCcond RIVIP ⋅+⋅= 2For dc operation, ID,av=ID,rms=IDC
•For low temperatures the conduction loss of the SiC diode is less than that of the Si diode and vice versa for higher temperatures.
–Note that Si diode cannot withstand Tj > 150 °C
0 1 2 3 4 5 6 7 8 9 100
0.25
0.5
0.75
1
1.25
1.5
Dio
de C
ondu
ctio
n Lo
ss, W
Diode Forward Current, A
Si
25°C
225°C
0 1 2 3 4 5 6 7 8 9 100
0.255
0.5
0.75
1
1.25
1.5
Dio
de C
ondu
ctio
n Lo
ss, W
Diode Forward Current, A
SiC
25°C
225°C
0 1 2 3 4 5 6 7 8 9 10 0
0.25
0.5
0.75
1
1.25
1.5
Dio
de C
ondu
ctio
n Lo
ss, W
Diode Forward Current, A
Si
SiC25°C
225°C
25°C
225°CDirection of temperature
increase
24
SiC vs. Si diode switching losses
Derivation of the switching losses
∫=b
cddrr dtivE
ab
-dIF/dt trrta tbc
IR
-VRVRM
IF
0
0
Turn-onloss
Turn-offloss
Reverse recoveryloss
c-a region a-b region
( )
2
1
bRR
b
a bRR
tIV
dtt
atIV
=
−+−⋅⋅−= ∫
2bRR
srrtIVfP =
a
b
ttS ≡
2
12
+
=
SSt
dtdI
SVfP rrFR
srr
25
SiC vs. Si diode switching losses
R1
L1
D=DUTVdc
iDUT=id
CurrentProbe id
+vd-
Q
iL
oven
+
vQ
- iQ
1kHz25% duty
SiC Schottky diode
Si pn diode
Typical reverse recovery waveforms of the Si pn and SiC Schottky diodes
Reverse recovery loss measurement circuit
26
SiC vs. Si diode switching losses (cont’d)
1 1.5 2 2.5 3 3.5 4 4.50
1
2
3
4
5
6Pe
ak R
ever
se R
ecov
ery
Cur
rent
, A
Peak Forward Current, A
Si
SiC
27°C
61°C
107°C
151°C
27, 61, 107, 151, 200, 250°C
1 1.5 2 2.5 3 3.5 4 4.50
0.25
0.5
0.75
1
1.25
1.5
1.75
2
2.25
2.5
Peak Forward Current, A
Dio
de S
witc
hing
Los
s, W
Si
151°C
61°C
107°C
27°C
27, 61, 107, 151, 200, 250°C
SiC
• Switching characteristics of the SiC diode do not change much with temperature
• Maximum operating conditions– Si diode: 150°C and 4.5 A– SiC diode: 250°C and 4 A ⇔ (datasheet: 175 °C )
27
SiC vs. Si diode switching losses (cont’d)
• Modeling
SiC Schottky
Si pn diode
∫=b
cddsrr dtivfP
∫=b
adRsrr dtiVfP
diode
Calculated experimentally
• The integral above varies linearly with the forward current
βα +⋅=∫ F
b
ad Idti
31.2138 105.2105.3 T⋅×+×= −−α53.3158 103.21025.1 T⋅×+×= −−β
SiC: α=2.167×10-8
β=2.33×10-8
— Assume 20 of 10A diodes in parallel. This means 20 times the losses
+⋅⋅⋅= βα
2020200 F
RsA
rrIVfP
α-β model
• For 200A diode
Si:
28
SiC vs. Si MOSFET conduction lossesonDSrmsQQcond RIP ,
2,, ⋅=MOSFET conduction loss expression
• At 300°K for a 1 cm2 device Ron,sp = 0.18Ω for Si andRon,sp = 0.61x10-3 Ω for 6H-SiCRon,sp = 0.305x10-3 Ω for 4H-SiC (calculated)
α
=
300300
,,TRR K
sponT
spon
300 320 340 360 380 400 420 440 460 480 500
0.01
1
RD
S-on
, Ω
300 320 340 360 380 400 420 440 460 480 5000.01
1
10
P con
d,Q
1, W
Temperature, K
0.0001
Si
SiC
Si
SiC
423
29
SiC vs. Si MOSFET switching losses• The switching losses are mostly during the charging and
discharging of two device capacitances: drain-to-source and drain-to-gate
Drain
( )
++
−=
+⋅=
⋅=
11
11
31
21
21
1
KKBVVVEf
EEf
Efp
csc
offonc
totcQ
ε
Source
Gate Cds
Cdg
Cgs
Drain-to-sourcecapacitance
Gate-to-sourcecapacitance
Drain-to-gatecapacitance
( )J
VVgK thGHm −=1
( )J
VVgK GLthm −=2
30
1. DC-DC POWER SUPPLY
2. ELECTRIC TRACTION DRIVE
31
Dc-dc Power Supply
Q1
Q4
Q2
Q3
D1
D2
ab N1
N2
N2
vo1
v1
Vdc /2
Vdc /2
Id
Io
IL+
-
+-
+ -vL+
-voC
Input Voltage: 300 - 450VOutput voltage: 42V (regulated)
Output Power: 2 - 5kW
voi
v1
iL
Vo
Io
Pout (kW) Vdc (V) VMOSFET (V) IMOSFET (A) VDIODE (V) IDIODE (A)2 300 300 6.67 84 472 450 450 4.44 84 475 300 300 16.67 84 1195 450 450 11.11 84 119
Isolated full-bridge dc-dc converter
32
Device losses• MOSFET conduction loss
d is the duty ratio
• MOSFET switching loss – same as the loss of a single MOSFET given earlier
• Diode conduction loss
• Diode switching loss – same as the loss of a single diode (α-β model )given earlier with VR = 84V
( )onDSQ
onDSQonDSrmsQcond
RId
RIdRIP
,2
,
2
,2
)(
⋅⋅=
⋅⋅==
( )( )DDDD
DDDDDrmsDDavDcond
RIVId
RIdVIdRIVIP
⋅+⋅=
⋅⋅+⋅⋅=⋅+⋅=2
22,,
33
SIMULINK model of the dc-dc converter
.5
d
TjQ
TjD
Ptotal D4
TATj
Thermal Model w/ Heatsink1
Ptotal Q1
TATj
Thermal Model w/ Heatsink
I PswQ1
Q1 switching
Tj
I
d
PcondQ1
Q1 conduction
Ptotal
PQ1
PD4
7 N1/N2
16.67
IMOSFET
6
Gain
I
TjPswD4
D4 switching
Tj
I
d
PcondD4
D4 conduction
TA
Constant7273
Constant6
273
Constant5
TA
Constant4
150
Constant3
150
Constant2
w/
34
Simulation results of the dc-dc converter
412347
549
41
1197
775
626
205
0
200
400
600
800
1000
1200
1400
1 2
Si 20kHzSiC 20kHzSi 100kHzSiC 100kHz
Diodes MOSFETs
Heatsink Volume
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
50
100
Con
duct
ion
loss
, W
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
10
20
Switc
hing
lo
ss, W
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
50
100
Tota
l lo
ss, W
SiC
Si
SiC
Si
Si
SiC
Time, s
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
50
100
Con
duct
ion
loss
, W
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
20406080
Switc
hing
lo
ss, W
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
50
100
150
Tota
l lo
ss, W
SiC
SiC
SiC
Si
Si
Si
Time, s
Diode losses
412 549775 626
347 41
1197
205
0
500
1000
1500
2000
2500
1 2 3 4
20kHz 100kHz
35
Passive Components• High frequency transformer
– SiC power devices can be operated at higher switching frequencies.
– As the switching frequency increases, the size of the high frequency transformer decreases, because
– fc⋅N ⋅A needs to stay constant to keep the maximum flux density without saturation.
– If the switching frequency is increased from 20kHz to 100kHz, then N ⋅A has to be decreased five times.
∴ The size of the high frequency transformer is inversely proportional to the switching frequency
NAfVB
c
14max ⋅=
36
Passive Components (cont’d)
• Output filter capacitor
– For a constant dc , C ∝ A.– C is inversely proportional to the
switching frequency∴ The size of the filter capacitor is
inversely proportional to the switching frequency
• Output filter inductor– The same reasoning as the high
frequency transformer with
C
L
Vin(jω) Vout(jω)
0 50 100 150 200 250 300 0
50
100
150
200
250
300
350
400
fc, kHz L
( µH
) and
C ( µ
F)
L
C
cdAC ε=
NALIB max
max =
37
Electric Vehicle Traction Drive
Q1
Q4
Q3
Q6
Q5
Q2
D1
D4
D3
D6
D5
D2
Vdc /2
AC MOTOR
Vdc /2a
bco
PHASE LEG A
ibia
ic
Federal Urban Driving Schedule
• The electric traction drive system is analyzed for system level benefits such as reduction in size and volume, increase in efficiency, etc.
Spee
d, m
ph
60
00 1369Time, s
38
Why average modeling?• Because of the huge difference between the required sampling times
of the simulation components.
FUDS Cyclesampled at 1Hz
(1 s)
0s 1369s
InductionMachine
Three-PhaseInverter
Electric Traction Drivefo=0-200 Hz
SiC Power devicesfc>20kHz
(switching period<50 µs)
Battery
ADVISOR
ωr, Te I, M
– Vehicle simulation~seconds
– Three phase inverter simulation with switching frequency of 20kHz
~microseconds– Device loss simulation included
~nanoseconds
• For device loss simulation, – without average modeling ~109 simulation points per second of
simulation is required.– with average modeling 2×104 simulation points per second of
simulation is required (for 20kHz switching). ~50000 times less points
39
Verification of average modeling
0.8 0.81 0.82 0.83 0.84 0.85 0.86 0.87 0.88 0.89 0.9-
0
200
i a, i b
,and
i c (
A)
0.8 0.81 0.82 0.83 0.84 0.85 0.86 0.87 0.88 0.89 0.9-50
0
50
T e (N
.m)
0.8 0.81 0.82 0.83 0.84 0.85 0.86 0.87 0.88 0.89 0.9200
250
300
Time, s
ωr (
rad/
s)
1 1.02 1.04 1.06 1.08 1.1 1.12 1.14 1.16 1.18 1.2
-
0
200
i a, i b
,and
i c (
A)
1 1.02 1.04 1.06 1.08 1.1 1.12 1.14 1.16 1.18 1.2
0
100
200
T e (N
.m)
1 1.02 1.04 1.06 1.08 1.1 1.12 1.14 1.16 1.18 1.2200
250
300
Time, s
?r (
rad/
s)
2*pi*40we
vco*
vbo*
vao*
300/2
Vdc/2
TorqueLoad
Scope
vref
vao*
vbo*
vco*
vao
vbo
vco
Inverter
vao
vbo
vco
Tl
we
ia
ib
ic
Te
wr
InductionMachine Model
vao
vbo
vco
Tl
we
ia
ib
ic
Te
wr
InductionMachineModel
vref1k
40
Average device loss modelingMOSFET Losses
•Conduction Losses
Dn = duty ratio in the nth intervalio,n = average output current in the nth interval
I = peak output current
φ = phase angle of the current
onDSrmsQQcond RIP ,2
,11, ⋅=
∑=−
=
1
0
2,,1
1 N
nnnormsQ Di
NI
co
oc
TT
ffN ==
−= φθnInoi sin,
If N >> 1, then
Thus,
D
+= nMn θsin121
φπ
φππ
φπ
φθθφθπ
cos31
81
cos34
241
sin12sin221
,1
MI
MI
dMII rmsQ
+=
+=
∫+
+−⋅
≅
+⋅⋅= φπ cos3
1812
,1, MRIP onDSQcond
41
Average device loss modeling (cont’d)
Averaging over the outputperiod, To,
MOSFET Losses (cont’d)
•Switching Losses
Energy loss during switching
Q1 switching loss in one Tcperiod is
( ) ( )
++
−=
++
−=
+=
JG
J
JG
JH
BV
VVcEs
KBV
VVcEs
K
offEonEtotE
21
21
2
21
1 13
1
13
1εε
totEcfcToffEonE
pQ =+
=1
( )
−
−+−−
+
−
−+−
≅
∫+
≅∑=
=
222
22
2
221
221
1
1,
'
'1tan22'
'
'1tan2'2
2
1
1
1
JG
J
JG
G
JG
J
JG
GcHf
dtotEcfN
nntotEcf
NP Qsw
π
ππ
φπ
φθ
π
21
31
=
BVVVEfH csc ε
( )thGHm VVgG −=1 ( )GLthm VVgG −=2
42
Average device loss modeling (cont’d)
Thus,
Diode Losses
•Conduction Losses
The diode conducts when the MOSFET is not conducting; therefore, the duty ratio of the diode is
And,
DrmsDDavDDcond RIVIP ⋅+⋅= 2,4,44,
−=− nMnD θsin1211
φπ cos31
81
,4 MII rmsD −=
−=
∫+
−−≅
∑ −=−
=
8cos
21
sin121sin2
1
1,1 1
0,4
φπ
φπ
φθθφθπ
MI
dMI
nDnoiNIN
navD
−⋅⋅+−⋅⋅= φπφπ cos8
121cos3
181
4,2 MVIMRI
DcondP DD
•Switching Losses
−Use the α-β model
43
Algorithm to find I, M, and φ1. Get Te and we profiles from ADVISOR
2. Machine input power, 3. Output frequency, 4. V/Hz constant,
5. where fb is the base frequency
6.7.8.
ηωrein
TP ⋅=
πω22
ropf ⋅=
bf
dcV
vK
24
2
3
π=
voL KfV ⋅=
LII 2=
L
inL V
PI3
cos =⋅ φ
mL II =⋅ φsin
( ) ( )
2
2
22
3
sincos
mL
in
LLL
IV
P
III
+
⋅=
⋅+⋅= φφ
9. Peak line current
10.
11. Modulation index L
L
II φφ coscos 1−=
b
o
bv
ov
dc
L
ff
fKfK
VVM =
⋅⋅
==
24
23
π
44
SIMULINK model of the traction drive
TjQ
TjD
Ptotal D4
TATj
Thermal Model w/ Heatsink1
Ptotal Q1
TATj
Thermal Model w/ Heatsink
I PswQ1
Q1 switching
TjIphiM
PcondQ1
Q1 conduction
Ptotal
PQ1
PD4
Te
wr
I
M
phi
I, M, & φ
6
Gain
[t wr]
[t Te]
I PswD4
D4 switching
TjIphiM
PcondD4
D4 conduction
TA
Constant7 273
Constant6
273
Constant5
TA
Constant4
175
Constant3
175
Constant2w/
45
Si vs. SiC - Device Losses for HEV
• SiC diode losses are lower mostly because it has lower reverse recovery losses
• SiC MOSFET losses are lowerbecause– Switching losses are similar– SiC MOSFET conduction losses
are lower (Ron,sp (Si)=180×10−3 Ω-cm2
Ron,sp (SiC)=0.3×10−3 Ω-cm2)
0 200 400 600 800 1000 12000
20
40
60
Dio
de lo
sses
, W
0 200 400 600 800 1000 12000
200
400
600
Time, s
MO
SFET
loss
es, W
Si
SiC
Si
SiC
1369
1369
46
Si vs. SiC - Efficiency
• Energy loss in inverter for FUDS cycle:– Si: 925 W⋅sec – SiC: 338 W⋅sec
• Efficiency:– Si: 80 – 85%– SiC: 90 – 95%
0 200 400 600 800 1000 12000
2000
4000
Tota
l inv
erte
rlo
sses
, W
0 200 400 600 800 1000 12000
50
100
Effic
ienc
y (S
i)
0 200 400 600 800 1000 12000
50
100
Effic
ienc
y (S
iC)
Time, s
SiCSi
Si
SiC
1369
1369
1369
47
Si vs. SiC - Heatsink Requirements
• If natural air cooled heatsinks are used, then – Si inverter needs a heatsink
with a volume of 1998 cm3
and a weight of 5.4 kg.
– Si inverter needs a heatsink with only a volume of 606 cm3
and weight of 1.65 kg.
1200450
4200
1200
5400
1650
0
1000
2000
3000
4000
5000
6000
Diodes MOSFETs Inverter
Heatsink Mass (g)
SiSiC
0 200 400 600 800 1000 12000
50
100
150
200
Dio
de J
unct
ion
Tem
pera
ture
, °C
0 200 400 600 800 1000 12000
50
100
150
200
MO
SFET
Jun
ctio
nTe
mpe
ratu
re, °
C
150 C
175 C
150 C
175 C
Si
SiC
Si
SiC
1369
1369Time, s
48
Conclusions
• SiC devices are in their infancy but they are already surpassing the mature Si device technology
• By replacing Si power devices with their SiC counterparts– power converter losses are decreased– the efficiency of the inverter is increased– less cooling is required– the volume and weight of the converter are decreased– higher switching frequencies can be used; consequently,
the volume and weight of the passive device decrease• The averaging model of an inverter is
– similar to a ‘moving average filter’– a good approximation, which can be used to shorten the
long simulation time.
49
Current and Future Projects
• High frequency gate drivers for SiC switches (JFET) operating at high temperatures (Madhu)
• A 55kW Si-SiC hybrid AIPM/inverter (Cree-Semikron)
• A 7.5kW all-SiC inverter (Rockwell)• High temperature packaging forSiC power devices