Professional Documents
Culture Documents
Design of A 3-Phase MOSFET Inverter by Lovatt H PDF
Design of A 3-Phase MOSFET Inverter by Lovatt H PDF
Aachen 1989 Absmf. The paper discuses, with particular reference to the gate-drive circuit, the design of an inverter to reproduce a stored current
waveform with a hgh degree of accuracy for use in a motor drive system. The gate-drive circuit combines low cost with individual over-
current protection for each switch, thus providing shoot-through and short-circuit protection without the need or expense of a d.c link
choke. In addition to the gate-drive circuit design, the reasons for choosing MOSFETs as the main power devices, alternative gate-drive
circuit confiations and inverter protection techniques are discussed. Although designed for a standard bridge configuration, MOSFET
inverter, the gate-drivc circuit presented is applicable to other MOS input devices and to other power circuit configurations.
Keywords. MOSFETs, inverters, gate-drive circuits
Rerurn-
current
diode
(i) The inverter is to provide an output waveform with a fundamental fre-
quency in the range 0 to ux)Hz.
(i) The d.c. link voltage is variable between 50 and 400 V. Fig. 1: Standard 3-phase bridge configuration
(iv) Protection is to be provided against shoot-through faults and short-
circuits on the inverter’s outputs. me of switching device
(v) The maximum continuous output current is lOA/phase. The specification requires a DC link voltage of up to 400 V and a maximum con-
(vi) The peak output current is 33 A/phase. tinuous output current of 10 A/phase. In addition, a modulation frequency of
(vii) The control and power eledronics are to be electrically isolated from 20 kHz has been chosen. A good choice of power switch to meet these require-
each other. ments is the MOSFET whose advantages and characteristics are outlined below. -
In the following sections the power circuit codiiration, the reasons for choos- (i) They switch between states quickly, thus reducing switching losses and
ing MOSFETs as the main power devices, the gate-drive circuit, inverter protec- allowing a higher switching frequency. A higher switching frequency
tion and practical implementations of the design are discussed. The section on gives better control of the output current waveform.
the gate-drive circuits presents ‘circuit fragments’ without explanation as to how (i) They are easy to drive (seegate-drive circuit options below).
they operate, since they are well known configurations. in each case, reference is
(i) They are economically pried.
made to a description of their operation.
(iv) They require no snubber network, which leads to higher efficiency,
lower cost and a lower component count.
POWER SWITCH CONSIDERATIONS
(vi) They have good overload capabilities which allows the inverter to pro-
Gmif configurofionand modulation strategy duce the specified peak current rating for a short period of time.
The specification demands an accurate reprodudion of a stored current W h e n the decision to use MOSFETs for the main switching dcviccs was taken
waveform to reduce the undesired harmonics present in the current supplied to (early in 1988), the ‘state-of-the-art’ devices were the IRF [International
the motor. The inverter has a maximum continuous peak current rating of Rectifier, 1985) range of MOSFETs. As with all power MOSFETs, IRF devices
10 A/phase and a maximum d.c. link voltage of 400 V and at these voltage and have an anti-parallel diode (often referred to as a ‘parasitic’ diode) bctwcen their
current levels it is not economical to use a linear power amplifer, since the source and drain connections ( f i e 2 ) . This diode is in parallel with the
power dissipation (losses) would be too large. For example, if full current was return-current diode (see f i w e l ) and therefore conducts when the energy
supplied at a maximum d.c. link voltage to an indudive load a linear power stored in the motor windigs is returned to the supply. Since the ‘parasitic’
amplifer would be required to dissipate 4 kW (full current in one phase. and half diode turns off slowly, relative to the turn-on time of the MOSFET, there is a
current in the other two with half the d.c. link voltage across each conducting period of time equal to the difference in these two switching times when there is
device). To reduce the dissipation and hence the device rating and heatsinking a short-circuit acros the d.c. link. This leads to a large instantaneous power dis-
requirements, a switchg type power amplifier using a standard >phase bridge sipation in both the MOSFET and ‘parasitic’ diode. To prevent this, the ‘parasi-
configurationhas been adopted (see f i e 1). tic’ diode must be prevented from conducting. Two possible ways of doing this
There are many ways of modulating the incoming signal to produce a series of are suggested below.
pulses suitable for ampWication by the power electronics. A pulse-width- (i) Use a return-current diode with a lower ‘on’state voltage drop at full
modulation (PWM) strategy has been chosen on the grounds that it can be current than the voltage drop at zero current of the ‘parasitic’diode.
operated at a fiied frequency (which helps in control Imp analysis) and is easily
implemented. (fi) Put a diode in series with the MOSFET (and ‘parasitic’ diode) to
prevent reverse. currents flowing ( f i i e 3).
The choice of modulation frequency inevitably involves a compromise between The fiist approach of using a return-current diode with a low ‘on’state voltage
reducing the unwanted harmonics at the output (which implies as high a switch- would be ideal, since it would also reduce the inverter losses when compared to
ing frequency as possible) and reducing the losses in the power electronics the sccond option. Unfortunately, no diodes are readily available with low ‘on’
(some of which are proportional to switching frequency). As a compromise state voltage drops and suffiaent reverse blocking voltage to ‘hold off the
between the conflicting requirements of a ‘good quality‘ output and low power specified d.c link voltage of 400V. Therefore the approach of preventing
dissipation, a modulation frequency of 20 kHz has been chosen. This modulation reverse currents with a series diode and suffering the cxtra ‘on-state’losses ((i)
frequency has the added benefit of beiig ultra-sonic and hence reduces the audi- above), has been chasen.
ble noise produced by the motor drive system.
c
Series
d 1 ode
-7 * Rmturn-
currmnt
diode
r a q u i r e hish
dVldt innunitu
I
Protection
over-ride
nonostabla -
-> Q
+ a-
- t!
Fio 6. Six channel natedrive board m r e d from an external mains Fig. 5 A self-pornred hvo-channel gatedrive
Cmfimtioa Circuit DrantRlCkS
bG
iconix, 1988). MOSFETs (which are
costly and have large 'on'
state losses).
I The MOSFET switches
I
* I
I
I Fast high-voltage transis-
tors are costly.
L ~ 0 . r not
reouiro hish
dVldt lnnunitu
4t-I I
n
A.C. coupled (
&? The HF (approximatel)
(Siemens, 1985). 1 MHz) d t o r required
to drive the transformer is
costly (due to the higk
current requirements).
The MOSFET turns-08
slowly since the gate ir
discharged by a high value
resistor (relative to the
output impedance of an
active driver).
4.c. Coupled drivw 0 The HF (approximately
vith additional 1MHz) d a t o r required
pulse amplifier to drive the transformer is
(Siemens,1985). costly (due to the high
current requirements).
-,,-
i u e when
8r~li.d
l l l L
Active
IOU
++Ad isolator
I I
I
II
II
3ptically isolated
An external power supply
rate drive is needed for each MOS-
:Clemente, 1987).
FET switch.
Optical isolators with
sflicient
are costly. dV/dt rejection
Re4wir.s hiah
dV/dt innuni tu
o n t O D Switches
Duos not I I I I
reauire hieh
dWldt Lnnunitu M in inun Mini nun
off-tine on-tine
nonostable nonostable
REFERENCES
Bmet, J. (l988)."Interadin Switching in Bridge Arms,"Ewmwt Ind (Waf
Gu7nUny) VOL 19,no.2,105-110.
Qemente, S. (19sr). 'Gate Drive CharacteriStia and Requirements for Power
m s , ' I m m a f h a J Rrctfim ILEWET Power MOSFET Lh?sipcrs
M M L67-1.74.
div h X M t i o d Rcctifer (1985). HEXFETdoln book
M&uda, Y, F W H, Ammo, H, Okuda, H, Watanabe, S. and Ishibashi, A.
(1983). P m l O p m c n t - O f PWM Invtrter EmploYing GTO,"IEEE Tmn~.011
w1.IA-19, no. 3,335-342
Philips (1988). Tahnicpl-H VOL 1 (Semiamductor~Dcviccs),no. If
ov - (P- MOS tmnsistm).
Preston, MA., Edwards,JD.and Williams, G. (1987). *A N m l PWh4 Inverter
U b g Trdormer-Isolated P m r MOSFETs,' phn. qf 2nd EPE confer-
.V
m c Gmrob4
~ F m u , voL 1,105-110.
Siemens AG (1985). SIPMOS P o w T n m & m : & M m -@.
OA -
Sicmens AG (1987/88). SIPMOS componmrS data bode
Sioonix Incorporated (1983). "Driving Power MOSWWER =%" MOS-
Fig.% MOSFET drainaoum voltage (top) and current
P O m R h p -c, 6.16-6.19.
(bottom) -den turning-on into a short-circuit
Siconix Incorporated (1988). WOSPOWER Gate Driw,' MOSPOVER Data
Book, 9.18-9.24.
Tho- Semiumducteurs(1987), BUF 410 1BUF 41M data sheet
+O .50s /div Toshiba (1988).Applicption Notes: GTR M~niules,BiporOr GMOS IGBT.
Wood,P. (1985). Transformer-Isolated HEXFET Driver Provides Very Large
Duty cyde Ratios," InfnnntioruJ M f i e r HEXFET Power MOSFET
Lksignm Manual, A.141-A.143.
t100V/div
ov
t 40A/div
OA