Professional Documents
Culture Documents
fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2776898, IEEE Access
Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.Doi Number
The temperature-dependent static simulation results could simulate the actual measured values well.
Furthermore, based on the comparison between the dynamic simulation results involving the device’s turn-
on and turn-off process losses and the actual transient state losses derived from a double-pulse test under
the conditions of Vds=300 V, Id=15 A, Rg_exton=18 Ω, and Rg_extoff=12 Ω at 200 °C, the maximum error was
6.7%.
INDEX TERMS SiC MOSFET, MATLAB/Simulink, whole working region, temperature-dependent model
2169-3536 © 2017 IEEE. Translations and content mining are permitted for academic research only.
VOLUME XX, 2017 Personal use is also permitted, but republication/redistribution requires IEEE permission. 1
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
2169-3536 (c) 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2776898, IEEE Access
controlled current sources in parallel with the standard are difficult to obtain by data fitting, thus reducing the
MOSFET model, the influences of temperature on the carrier complexity of the model and enhancing its portability.
and avalanche effects were simulated. In [6–7], the Nth-
power model was established to simulate the high voltage II. STATIC CHARACTERISTIC MEASUREMENT METHOD
SiC power MOSFET, and the precision of the output curve The SiC MOSFET selected for modeling was SCT20N120
was improved. The model presented in [8] divided the from ST Microelectronics. The initial measurement of the
device’s working region into three parts, calculated the static characteristic curve adopted the test platform shown
voltage of each part, and analyzed the carrier drift in each in Fig. 1. The device was placed on the heating plate, and a
part. Reference [9] established the physical model of SiC PT100 resistor was attached to the heating plate as a
VDMOSFETs and simulated the uneven current distribution temperature reference. According to the test conditions
in the JFET region by a nonlinear voltage source and resistor offered by the datasheet, the temperature-dependent curves
network. The main advantage of this model was to describe of SCT20N120’s on-state resistance and threshold voltage
the characteristics of the device in the linear and saturated at different temperature points were measured by an
regions by a set of simple functions. In [10], a new model Agilent B1505A power device analyzer and curve tracer.
parameter extraction strategy was proposed. The parameters
could be extracted from the device’s datasheet without
additional experimental testing. Reference [11] explained the SCT20N120
low carrier mobility in the inversion layer by simulating the Test
Box
interface trap. Reference [12] established the 2D model by
Heating Platform
SILVACOATLAS software, which included the composite
effect, bandgap narrowing, collision ionization, and lattice PT100
heating effect.
The modeling methods for SiC Power MOSFETs are Agilent B1505A
mainly based on physical modeling or behavioral modeling. Multimeter
Power Device Analyzer
The former simulates the electrical characteristics and Curve Tracer
thermal behavior of the device by solving the physical
equations. Its advantage is that the model has high precision,
and the correlative parameters have corresponding physical
meanings, whereas its disadvantage is that the model is FIGURE 1. Static characteristic test platform for SiC MOSFET.
complex, and the parameters are numerous and difficult to
obtain. The latter usually adopts the method of mathematical However, due to the limitation of the measuring
fitting, without considering the complex physical instrument’s power and range, the whole output
mechanisms of the device, so the relevant model parameters characteristic curve and transfer characteristic curve could
have no actual physical meanings. Its advantage is that the not be obtained completely by the measuring instrument.
model is simple and time-saving, while its disadvantage is The device’s working point is located in the cut-off
that the accuracy of the simulation is low, especially when region when it works in the off-state, whereas the device’s
the device’s operation is beyond the curve fitting conditions. working point is located in the linear region when it works
In this paper, a temperature-dependent SiC MOSFET in the on-state. In the SiC MOSFET’s turn-on and turn-off
modeling method based on MATLAB/ Simulink was processes, the working point inevitably passes through the
proposed. It took into account the working characteristics of saturation region. Although the resident time in the
the device in the saturation region. By supplementary saturation region is very short, the voltage and current in
experiment, the saturation region part of the output the saturated region are high, so the loss is greater. In order
characteristic curve, from which some essential parameters to better simulate the characteristics of the device and
were extracted for modeling, was made available, so that the establish an accurate model in the whole working region,
model could simulate the device’s working characteristics in the output characteristics of the device in the saturation
the whole working region. In this model, a compensation region need to be obtained by a supplementary experiment.
voltage source was added to simulate the temperature The supplementary test circuit is shown in Fig. 2. To
characteristic of the threshold voltage, and a compensation provide enough transient current, the DC power supply uses
current source was attached to compensate for the material rectification circuit with a 1500-μF electrolytic capacitor
and structural differences between the lateral double-diffused pack in parallel with the output side, as shown in Fig. 3.
Si MOSFET and the vertical double-diffused SiC MOSFET. Because the device’s power dissipation is large in the
The model belonged to the semi-physical model, which not saturation region, it is more appropriate to conduct the test
only kept the physical meanings of most parameters, but also with a single narrow drive pulse, to ensure the safety of the
eliminated the dependence on some physical parameters that device.
2169-3536 (c) 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2776898, IEEE Access
Compensation
_ Current Source
G +
Compensation
2.5
Voltage Source
2.0
S 1.5
20 40 60 80 100 120 140 160 180 200 220
Temperature ( )
FIGURE 4. Static model of SiC MOSFET.
2169-3536 (c) 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2776898, IEEE Access
Based on the temperature characteristic scatter diagram Vds_sat,T p p10T p01Vgs p20T 2 p11TVgs p02Vgs2
of the threshold voltage, shown in Fig. 5, the following , (6)
polynomial fitting function is used to obtain a fitted curve p30T 3 p21T 2Vgs p12TVgs2 p03Vgs3
of Vth,T. Where p00 67.86 , p10 0.2199 , p01 13.43 ,
5
Vth,T 1.784 10 T 0.01417T 3.958
2
(5) p20 9.313 104 , p11 0.02321 , p02 0.9072 ,
As the ambient temperature changes, Vds_sat,T is no longer p30 1.737 106 , p21 4.537 105 ,
a single-variable function of driving voltage Vgs, but rather p12 8.812 104 , and p03 0.02496 .
a two-variable function of Vgs and T.
Fig. 6 shows the saturated drain–source voltage
diagrammatic curve with varying gate voltage at 25 °C,
which also denotes the boundary between the linear region Measured Scatters
and saturation region, and the saturated drain–source Fitted Curve
voltage diagrammatic curve with a gate voltage of 18 V at
Vds_sat (V)
various temperatures is shown in Fig. 7. With varying
temperatures (25–200 °C) and gate voltages (10–20 V), a
series of Vds_sat,T’s 3D scatters are extracted for fitting.
100
20V
90
Linear Saturated Drain-Source
80
Region Voltage Curve
70
18V
60
Id (A)
50
Saturation Region Temperature ( ) Vgs (V)
16V
40
FIGURE 8. Saturated drain–source voltage fitting curve.
30
14V
20
12V As T, Vgs, Vth,T and Vds_sat,T are available, the parameter, α,
10
Cutoff Region Vgs=10V could be obtained according to (7).
0
0 10 20 30 40 50 60 70 80 V V
Vds (V) 1 gs th,T (7)
Vds _ sat ,T
FIGURE 6. Saturated drain–source voltage curve with varying gate The parameter Kp is the second to be determined.
voltage at 25 °C.
Because both the conductivity channel width W and the
conductivity channel length L are constant, and W/L is a
100 coefficient of Kp, it is not necessary to know the specific
T=200 values of W and L; instead, the parameter K is introduced as
90
the drain current scale coefficient to integrate the above
80
three parameters, as shown in (8).
70
W
60 K Kp (8)
T=25 L
Id (A)
2169-3536 (c) 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2776898, IEEE Access
function (10) is used to obtain the fitted curve shown in Fig. the saturation region is defined as:
9. tan GDS . (16)
I ds _ sat ,T p00 p10T p01Vgs p20T p11TVgs p02Vgs
2 2
A conversion is made to (15) for (17), shown as follows:
, (10) I d _ sat ,T
p30T 3 p21T 2Vgs p12TVgs2 p03Vgs3 tanθ GDS . (17)
1
where p00 66.29 , p10 0.2487 , p01 15.19 ,
120
p20 2.653 104 , p11 0.03345 , p02 0.9227 ,
p30 7.898 108 , p21 1.405 105 , 100
4
p12 6.779 10 , p03 0.005493 . θ
80
Id (A)
60
Measured Scatters
Fitted Curve
40
Id_sat (A)
20
1/λ
0
-200 -150 -100 -50 0 50 100
Vds (V)
By taking (10) into (9), the drain current scale B. COMPENSATION FOR THRESHOLD VOLTAGE
coefficient K is: To simulate the temperature characteristic of the threshold
I d_sat,T voltage, based on Vth,T derived from (5), a temperature-
K
(Vgs Vth,T )Vds_sat,T 0.5(1 )Vds_sat,T
2
1 Vds_sat,T controlled voltage source is added to the model, as shown
in Fig. 4.
(11)
The parameter λ is the third to be determined. The C. COMPENSATION FOR DRAIN CURRENT IN LINEAR
channel length modulation effect makes the drain current in REGION
the saturated region increase with the rising drain–source The differences in the structure and working process
voltage, rather than remaining constant. Thus, λ could be between LDMOSFET and VDMOSFET are difficult to
obtained by the output characteristic curve in the saturation compensate by precise analytic functions. In addition, the
region. difference in the materials between SiC and Si, and the
Define the saturation region output conductance GDS as immature processing technology of SiC power MOSFETs,
dI make the compensation problem more complicated. In this
GDS d,T . (12) paper, the mathematical method is used for compensation
dVds,T saturation region
from the aspect of device behavior, which simplifies the
The saturation-region drain current Id,T and the saturated problem.
drain current Id_sat,T are respectively expressed as: Because the error mainly results from the gate–source
K
I d,T (1 Vds,T )(Vgs Vth,T ) 2 (13) voltage, drain–source voltage, and temperature, the
2(1 ) following compensation function involving the above three
K factors is used to establish a compensation current source.
I d_sat,T (Vgs Vth,T ) 2 . (14)
2(1 ) (T 273) f (V ) (T 273) g (V )
gs ds
I d I d' 1 , (18)
According to (12)–(14), GDS could also be expressed as: 25 273 25 273
GDS I d_sat,T . (15)
As shown in Fig. 10, the geometric definition of GDS in
2169-3536 (c) 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2776898, IEEE Access
where Id is the actual drain current in the linear region; Id' is The turn-on process of a SiC MOSFET with an inductive
the mode’s drain current in the linear region without load is shown in Fig. 12.
compensation; f(Vgs) and g(Vds) are the correction functions, During Period 1 (0–t1), Vgs rises to Vth from 0 V. t1 could
the former of which can be obtained from the transfer be expressed as follows:
characteristic curve to guarantee the fitting accuracy of the Vgs _ drive
transfer characteristic and the threshold voltage, and the t ( Rg _ int Rg _ exton )(C gs C gd ) ln( ) . (20)
Vgs _ drive Vth
latter can be obtained from the output characteristic curve
to ensure the fitting accuracy of the output characteristic During Period 2 (t1–t2), Vgs rises to the miller platform
and the conduction resistance. voltage Vgs_miller from Vth, and Id increases to the steady state
value from 0 A. t2 could be expressed as follows:
IV. DYNAMIC MODEL ESTABLISHMENT PRINCIPLE Vgs _ drive
t ( Rg _ int Rg _ exton )(Cgs Cgd )ln( ) (21)
The dynamic model of the SiC power MOSFET uses the Vgs _ drive Vgs _ miller
gate equivalent circuit, as shown in Fig. 11. In the circuit, During Period 3 (t2–t3), Vgs stays in Vgs_miller, and drain–
Vgs_drive is the gate driving voltage; Rg_int is the gate interior source voltage Vds begin falling at t2 as the Miller
parasitic resistance; Rg_exton is the gate exterior turn-on capacitance Cgd is charged until Vds reaches the on-state
resistance; Rg_extoff is the gate exterior turn-off resistance; Ig voltage. Here,
is the gate drive current; Cgd and Cgs respectively represent
Id
the gate–drain parasitic capacitance and the gate–source Vgs_miller Vth , (22)
parasitic capacitance; Igd and Igs represent the current gm
through Cgd and Cgs, respectively; Vd is the drain voltage; and t3 could be expressed as follows:
and Vgs is the voltage on Cgs. ( R Rg_exton )Qgd
t g_int t . (23)
Vd (Vgs_drive Vgs_miller )
Igd During Period 4 (t3–t4), Vgs rises to the gate drive voltage
Vgs_drive from Vgs_miller. t4 could be expressed as follows:
Rg_exton Cgd t ( Rg_int Rg_exton )(Cgs Cgd ) (t t ) . (24)
Rg_int
Ig Id
Vgs_drive
Rg_extoff +
Vgs _ Cgs Vds
Vgs
Igs
Vgs_miller
t
If Vgs_drive is equivalent to the step excitation signal, then 0 t5 t6 t7 t8
Vgs can be expressed as follows:
FIGURE 13. SiC MOSFET turn-off process under inductive load.
Vgs Vgs _ drive (1 et / ) (19)
where the time constant τ= (Cgs+Cgd)(Rg_int+Rg_ext). The turn-off process of the SiC MOSFET with an
inductive load is shown in Fig. 13. Similarly, the relevant
Id
time points are expressed as follows:
V
Vds t5 ( Rg _ int Rg _ extoff )(Cgs Cgd )ln( gs _ drive ) (25)
Vgs Vgs _ miller
( Rg _ int Rg _ extoff )Qgd
t6 t5 (26)
Vgs _ miller
Vgs_miller
Vgs _ miller
Vth t7 ( Rg _ int Rg _ extoff )(Cgs Cgd ) ln( ) t6 (27)
Vth
0 t1 t2 t3 t4 t t8 ( Rg_int R g_ extoff)( C gs C gd) ( t6 t5 ) . (28)
2169-3536 (c) 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2776898, IEEE Access
Id (A)
25
0 M
Id(m) i
Mode D _ + 20
T=200
25 T Id Meas 15
Temperature 10
+
Continuous Vds_Power T=25
12 Vgs 5
Ideal Switch
s
_
Vgs_Signal S
powergui 0
G 0 2 4 6 8 10 12 14 16 18 20
Vds_Signal Vgs (V)
SiC power MOSFET
+
Vds(m)
Vgs=10V 225
0 220
0 2 4 6 8 10 12 14 16 18 20
Vds (V) 215
210
FIGURE 15. Comparison of output characteristic curves between
experiment and simulation at 200 °C. 205
200
Fig. 16 shows the transfer characteristic curves of the 195
SCT20N120 model in comparison with the actual measured 190
values at 25 °C, 100 °C, and 200 °C, respectively. 20 40 60 80 100 120 140 160 180 200 220
Temperature ( )
2169-3536 (c) 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2776898, IEEE Access
Vgs(10V/div)
Vgs
t(50ns/div)
150
Vgs Vgs(10V/div)
100
Id Id(10A/div) 50
Vds(200V/div) 0
Vds 5 10 15 20
Id (A)
Experiment Value
200
Simulation Value
Turn-off Losses (μJ)
Vgs(10V/div)
Vgs 150
Id(10A/div) 100
Id
50
Vds
0
Vds(200V/div) 5 10 15 20
Id (A)
t(50ns/div)
FIGURE 24. Comparison of turn-off losses between experimental
FIGURE 21. Simulation waveform of SCT20N120’s turn-on process. results and simulation results.
2169-3536 (c) 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2017.2776898, IEEE Access
VI. CONCLUSION transistor in 4H-silicon carbide,” Int. J. of High Speed Electron. and
This paper proposed a modeling method based on Syst., vol. 16, no. 2, pp. 733–746, 2006.
[9] R. Fu, A. Grekov, J. Hudgins, et al., “Power SiC DMOSFET model
MATLAB/Simulink for a temperature-dependent SiC accounting for nonuniform current distribution in JFET region,”
MOSFET model in the whole working region. IEEE Trans. on Ind. Appl., vol. 48, no. 1, pp. 181–190, 2012.
The 1200-V, 20-A SiC MOSFET, SCT20N120 from ST [10] M. Mudholkar, S. Ahmed, M. N. Ericson, et al., “Datasheet driven
silicon carbide power MOSFET model,” IEEE Trans. Power
Microelectronics, was selected as the target device for Electron., vol. 29, no. 5, pp. 2220–2228, 2014.
modeling. In order to obtain sufficient simulation precision, [11] S. Potbhare, N. Goldsman, A. Lelis, et al., “A physical model of high
the output characteristics of the device in the saturated temperature 4H-SiC MOSFETs,” IEEE Trans. on Electron. Devices,
vol. 55, no. 8, pp. 2029–2040, 2008.
region were extracted using a supplementary test circuit, [12] B. N. Pushpakaran, S. B. Bayne, and A. A. Ogunniyi, “Electro-
and thereby the whole working region output characteristic thermal transient simulation of silicon carbide power MOSFET” 19th
curves of device were made available. From the curves, IEEE Pulsed Power Conf (PPC), 2013, pp. 1–6.
three main parameters essential for modeling were derived.
Based on the traditional Si lateral double-diffused
MOSFET model, a temperature-dependent static model of
SCT20N120 was established by adding a temperature-
dependent compensation voltage source to simulate the
temperature characteristics of threshold voltage, and a
temperature-dependent compensation current source to
compensate for the drain current error in the linear region
due to the differences in structure and material. In addition,
based on the gate equivalent circuit, a dynamic model of
SCT20N120 was established.
The simulation results of the temperature-dependent
static model involving the output characteristics, transfer
characteristics, threshold voltage, and on-state resistance
could simulate well the actual measured values.
Furthermore, the simulation results of the dynamic model
under the conditions of Vds=300 V, Id=15 A, Rg_exton=18 Ω,
and Rg_extoff=12 Ω at 200 °C indicated that the temperature-
dependent dynamic model could simulate SCT20N120’s
turn-on and turn-off transient states losses with a maximum
error of 6.7%.
REFERENCES
[1] P. Alexakis, O. Alatise, L. Ran, et al., “Modeling power converters
using hard switched silicon carbide MOSFETs and Schottky barrier
diodes,” in 15th European Conf. on Power Electron. and Appl. (EPE),
Lille, France, 2013, pp. 1–9.
[2] A. Merkert, T. Krone, and A. Mertens, “Characterization and
scalable modeling of power semiconductors for optimized design of
traction inverters with Si-and SiC-devices,” IEEE Trans. Power
Electron., vol. 29, no. 5, pp. 2238–2245, 2014.
[3] K. Sun, H. Wu, J. Lu, et al., “Improved modeling of medium voltage
SiC MOSFET within wide temperature range,” IEEE Trans. Power
Electron., vol. 29, no. 5, pp. 2229–2237, 2014.
[4] S. Yin, T. Wang, K. J. Tseng, et al., “Electro-thermal modeling of
SiC power devices for circuit simulation,” 39th Annu. Conf. of the
IEEE Ind. Electron. Soc. (IECON), 2013, pp. 718-723.
[5] V. d'Alessandro, A. Magnani, M. Riccio, et al., “SPICE modeling
and dynamic electrothermal simulation of SiC power MOSFETs,”
IEEE 26th Int. Symp. on Power Semicond. Devices and ICs (ISPSD),
2014, pp. 285–288.
[6] N. Phankong, T. Funaki, and T. Hikihara, “A static and dynamic
model for a silicon carbide power MOSFET,” 13th European Conf.
on Power Electron. and Appl. (EPE), 2009, pp. 1–10, 2009.
[7] Y. Cui, M. Chinthavali, and L. M. Tolbert, “Temperature dependent
Pspice model of silicon carbide power MOSFET,” 27th Annu. IEEE
Appl. Power Electron. Conf. and Expo. (APEC), 2012, pp. 1698–
1704.
[8] M. Hasanuzzaman, S. K. Islam, L. M. Tolbert, et al., “Design,
modeling, testing, and spice parameter extraction of DIMOS
2169-3536 (c) 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.