Académique Documents
Professionnel Documents
Culture Documents
ENT
GEM
ANA
M
S A M D AV I S
POWER
SUPPLIES
SYSTEMS
APPLICATIONS
COMPONENTS
SEMICONDUCTORS
ID
VGS
VD
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
MANAGEMENT
0V
-1V
-1.5V
-2V
-2.5V
-3.3V
BY SAM DAVIS
VIN 100.0
Linear VGS
Regulators HSFET 80.0
Driver 5V
Control 60.0 4V
Anti-Cross- SW 3V
BP6
20.0
Conduction
RT 2V
Oscillator Pre-Bias LSFET 40.0 1V
SYNC/RESET_B ID 00.0 0V
Ramp
PWM
S Q GND
OC Event -20.0
RESET_VOUT + Current
R Q Average IOUT Sense, -40.0
COMP
OC
Error Amplifier Detection CLK -60.0
FB Fault
+ VREF for -80.0
Soft-Start and OC Threshold DATA
Reference VOUT_COMMAND -100.0
DAC -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0
SMBALERT
PMBus Engine VD
VSET Courtesy of Efficient Power Conversion
ADC, PMBus Commands, CNTL
VOUT Sense
IC Interface, EEPROM
OV/UV Detection
ADDR0
CHAPTER 10:
SILICON POWER-MANAGEMENT
POWER
SEMICONDUCTORS
S
ilicon power semiconductors +v 10-1. Power management power
are employed in power man- semiconductors duplicate the action of
agement systems. They are a mechanical switch in which a control
found in two different forms: Load operation turns the power semiconductor
Discrete Power Semi- switch on and off to control power applied
conductors (single type to a load.
housed in a package.)
Integrated Power Semiconductors Control Power turned off so zero current flows to the load.
MOSFETs and BJTs (bipolar junction Signal Switch It should also have zero impedance when
transistors) can be integrated with other turned on so that the on-state voltage drop is
circuits in a single package. In addition, zero. Another idealistic characteristic would
various types of power semiconductors be that the switch input consumes zero pow-
may housed in a hybrid (multi-chip mod- er when the control signal is applied. Howev-
ule, or MCM) package, that is, interconnected with other er, these idealistic characteristics are unachievable with
monolithic discrete devices in the same package. the present state of the art.
Power switches are actually the electronic equivalent In the real world, actual power semiconductor switch-
of a mechanical switch, except for much faster switching es do not meet the ideal switching characteristics. For
speed. Fig. 10-1 is the representation of a mechanical
switch. The individual power semiconductor switch
applies power to a load when a control signal tells it to
do so. The control signal also tells it to turn off. Ideally, (a) Control Signal
the power semiconductor switch should turn on and off
+V
in zero time. It should have an infinite impedance when
2.0 12
RDS(ON), Drain-to-Source On Resistance (Normalized)
ID = 11A
VGS = 4.5V
1.5 9
0.5 3
0.0 0
-60 -40 -20 0 20 40 60 80 100 120 140 160 25 50 75 100 125 150
TJ, Junction Temperature (C) TC, Case Temperature (C)
10-5. Normalized Variation of on-resistance vs. junction 10-6. Maximum Drain Current for a typical N-Channel
temperature for an N-Channel MOSFET. Power MOSFET.
properly. Some gate drivers also have protection circuits vices are rated in terms of EAR, the repetitive avalanche
to prevent failure of the power semiconductor switch and energy.
also its load. Trench technology provides the desirable charac-
MOSFET characteristics include several parameters teristics of low on-resistance sometimes at the expense
critical to their performance: of high avalanche energy. Trench power MOSFET tech-
Blocking voltage (BVDSS) is the maximum voltage nology provides 15% lower device on-resistance per
that can be applied to the MOSFET. When driving an unit area than existing benchmark planar technologies
inductive load, this includes the applied voltage plus any but usually at the cost of higher charge. And, the trench
inductively induced voltage. With inductive loads, the technology allows 10% lower on-resistance temperature
voltage across the MOSFET can actually be twice the coefficient. Fig. 10-4 is plot of single pulse avalanche
applied voltage. energy for a MOSFET.
Maximum single pulse avalanche energy (EAS) deter- On-resistance (RDS(ON)) for both planar and Trench
mines how much energy the MOSFET can withstand un- MOSFETs is important because it determines the power
der avalanche conditions. Avalanche occurs if the max- loss and heating of the power semiconductor. The lower
imum drain-to-source voltage is exceeded and current the on-resistance the lower the device power loss and
rushes through the device. The higher the avalanche val- the cooler it will operate. Low on-resistance drastically
ue the more rugged the device. The avalanche condition reduces heat-sinking requirements in many applications,
can cause two possible failure modes that can destroy which lowers parts count and assembly costs. In many
a MOSFET. The most destructive is bipolar latching applications, the low on-resistance also eliminates the
that occurs if the device current causes a voltage drop need to parallel MOSFETs for low on-resistance, which
across its internal device resistance, resulting in transis- leads to improved reliability and lower overall system
tor action and latching of the parasitic bipolar structure cost than previous MOSFET generations. In virtually all
of the MOSFET. A second failure mode is thermal, which MOSFETS, the n-channel versions have lower on-resis-
occurs if the avalanche condition raises the device tem- tance than p-channel devices with the same operating
perature above its maximum junction temperature. voltages.
Trench technology offers an avalanche capability ap- RDS(ON) decreases with increasing cell density. The
proaching industry-leading planar technology. To ensure cell density has increased over the years from around
satisfactory performance, devices in this technology can half a million per square inch in 1980 to around eight
be fully characterized for single pulse avalanche ener- million for planar MOSFETs and around 12 million and
gy (EAS) up to their maximum junction temperature. The higher for trench technology.
higher the EAS, the more rugged the device. Some de- Maximum junction temperature, TJ(max), is a function of
1000 6
ID = 7.0A
Operation in This Area Limited
by Rds(On)
5
VDS = 6.0V
100
4
10sec VDS = 12V
10 1 msec 3
10 msec
2
1
TC = 25C 1
TJ = 150C
Single Pulse
0.1 0
0.1 1 10 100 0 5 10 15 20
VDS, Drain-to-Source Voltage (V) QG, Total Gate Charge (nC)
10-7. Typical Maximum Safe Operating for an N-Channel 10-8. Total Gate Charge (Qg) of an N-Channel Power
Power MOSFET MOSFET Varies With the Drain-Source Voltage
the electrical characteristics of the device itself, as well mechanism called secondary breakdown that significant-
as the package employed. Package thermal properties ly reduces the SOA.
determine its ability to extract heat from the die. The junc- Total Gate charge (QG) The charge on the gate termi-
tion-to-ambient and junction-to-case thermal resistance nal of the MOSFET as determined by its gate-to-source
is a measure of the MOSFETs ability to extract heat. Data capacitance. The lower the gate charge, the easier it is
sheets rate thermal resistance in terms of either C/W or to drive the MOSFET. Total gate charge, QG, affects the
K/W. The lower the thermal resistance, the more efficient highest reliable switching frequency of the MOSFET. The
the package is in eliminating heat. In some cases, a heat lower the gate charge, the higher the frequency. Oper-
sink may be required to maintain the device junction ation at higher frequencies allows use of lower value,
temperature below its maximum rating. Fig. 10-5 shows smaller size capacitors and inductors, which can be
the variation of RDS(ON) with junction temperature for VGS = significant factors in system cost. A low gate charge also
4.5 V and 10 V. VGS is gate-to-source voltage. makes it easier to drive the MOSFET, however, designers
Drain current (ID) establishes the ability of the MOS- sometimes need to trade-off switching frequency with
FET to drive a specific load. This value can be limited by EMI considerations. Some New trench devices exhibit
the MOSFETs package. When operated in the pulsed lower gate charge than some existing planar technolo-
mode, the MOSFETs drain current can be several times gies by replacing larger die with new smaller die devices
its continuous rating. In the pulsed mode the pulse width that have been optimized to offer a lower charge version
and duty cycle determine safe drain current and device of the trench devices. Fig. 10-8 shows the gate charge
power dissipation. Fig. 10-6 shows the maximum drain for a typical power MOSFET, which is specified in nC,
current vs. case temperature. nano-coulombs.
Safe operating area (SOA) for a MOSFET is a function Although input capacitance values are useful, they
of the voltage and current applied to the device. Power do not lend themselves to calculation of the gate current
semiconductor manufacturers include a curve in their required to switch the device in a given time and they do
power transistor data sheets (Fig. 10-7) that defines the not provide accurate results when comparing the switch-
allowable combination of voltage and current, which ing performance of two devices. A more useful parame-
is called the devices safe operating area (SOA). The ter from the circuit design point of view is the total gate
product of the voltage and current represents the watts charge. Most manufacturers include both parameters on
dissipated in the chip. If you exceed the SOA, the chip their data sheets. Using gate charge, QG, the designer
will get too hot and fail. MOSFET devices are limited can calculate the amount of current required from the
by the SOA; bipolar devices have an additional failure drive circuit to switch the device on in a desired length of
1.6 100
1.4
VGS(th), Gate Threshold Voltage (V)
TJ = 150C
TJ = 25C
1
1.0
VGS = 0V
0.8 0.1
-75 -50 -25 0 25 50 75 100 125 150 0.0 0.5 1.0 1.5 2.0
TJ, Temperature (C) VSD, Source-to-Drain Voltage (V)
10-9. The Gate Threshold Voltage of this N-Channel 10-10. Body-Diode Forward Voltage for an N-Channel
Power MOSFET Is About 1.28 V for ID = 250A at 25C. Power MOSFET.
time because QG = current time. For example, a device MOSFETs have typical values of about 2V to 3V, whereas
with a gate charge of 20nC can be turned on in 20msec other devices can have higher values. In Fig. 10-9 is the
if a current of 1mA is supplied to the gate, or it can turn threshold voltage plotted against junction temperature.
on in 20nsec if the gate current is increased to 1A. These Power Loss MOSFETs are expected to have low
simple calculations would not have been possible with conduction and switching losses. For power manage-
input capacitance values. ment applications, conduction losses, ruggedness and
Gate charge and on-resistance are inter-related. That avalanche capability are important features. Conduction
is, the lower the gate charge, the higher the on-resis- losses are determined by the product of operating cur-
tance and vice versa. Historically, MOSFET manufac- rent and on-resistance (I2R) of the power MOSFET.
turers have focused on reducing RDS(on) without paying Maximum Allowable Power Dissipation (PD) is the max-
much attention to gate charge. This has changed in the imum allowable power dissipation that raises the MOS-
last several years, with new designs and processes be- FETs die temperature to the maximum allowable junction
coming available that offer reduced gate charge devices. temperature, Tjmax, when the case temperature is held
Figure of Merit (FOM) relates to the tradeoff between at 25C. Tj max is normally 150C or 175C.
RDS(ON) and gate charge. The product of RDS(ON) QG Body-Diode Forward Voltage (VSD) is the guaran-
is a figure of merit (FOM) that compares different power teed maximum forward drop of the body-drain diode at
MOSFETs for use in high frequency applications. a specified value of source current. The value of VSD
Threshold Voltage (VGS(TH)) is the minimum gate- is significant and must be low in applications where
source electrode bias required to form a conducting the source-drain voltage may extend into the negative
channel between the source and the drain regions. It is range, causing forward biasing the body-drain diode. If
usually measured at a drain-source current of 250A. A this happens, the source-drain current flows from drain
value of 2-4V for high voltage devices with thicker gate straight to the source contacts, across the forward bi-
oxides, and logic-compatible values of 1-2V for lower ased body-diode p-n junction.
voltage devices with thinner gate oxides are common. In A second and more dominant current conduction path
battery-based applications where power is a premium, will exist through the channel if the gate-source voltage,
the trend is towards lower values of RDS(on) and Vgsth. VGS >Vgsth. Low voltage and low RDS(on) power MOSFETs
Gate oxide quality and integrity become major issues as are used in such synchronous rectifier modes since their
gate oxide thickness is reduced to achieve lower Vgsth, forward voltage drop can be as low as 0.1V versus the
the minimum voltage is required between the gate and typical Schottky diode forward voltage drops of 0.4-
source that enables the MOSFET to turn on. Logic level 0.5V. Maximum values of 1.6V for high voltage devices
CISS
associated with the depletion region of the body diode,
extending into the drift region is denoted as CDB and 1000 COSS
appears between the base of the BJT and the drain of
the MOSFET. This capacitance gives rise to a current that
flows through the base resistance, RB, when a voltage
ramp appears across the drain-source terminals.
Static Electricity (ESD) Effects is another way to kill
CRSS
semiconductors. The static charge accumulated by a
person handling an MOSFET semiconductor is often 100
enough to destroy the part. Therefore, manufacturers of 1 10 100
semiconductors have instituted static discharge ratings VDS, Drain-to-Source Voltage (V)
that range from 3000V to 5000V. Handlers of MOSFET 10-12. Typical Plot of N-Channel MOSFETs Parasitic
semiconductors use grounding straps and conductive Capacitances vs. Drain-to-Source Voltage.
gate-to-drain capacitance. CRSS, often referred to as uration of the older planar MOSFETs. And, new Trench
the Miller capacitance, is one of the major parameters MOSFETs offer significant advantages over the older
affecting rise and fall times of the output voltage during generation Trench MOSFETs and also some improve-
switching. Plus, it also affects turn-off delay time. The ments over the older planar MOSFET technology.
capacitances decrease over a range of increasing drain- Among the other technologies are MDMesh . ST-
source voltage, especially the output and reverse trans- Microelectronics says that the improvement in RDS(ON)
fer capacitances. achieved with MDmesh V will significantly reduce losses
Important parameters for MOSFETs are listed in Table in line-voltage PFC circuits and power supplies, which
10-1. will in turn enable new generations of electronic products
offering better energy ratings and smaller dimensions.
Power MOSFET Fabrication Technologies This new technology should help designers with high
The trench MOSFET has replaced the planar device efficiency targets and also save power.
in many applications because it extends the cell density MDmesh V achieves its RDS(ON) per area performance
limit. Trench technology allows a higher cell density but by improving the transistor drain structure to lower the
is more difficult to manufacture than the planar device. drain-source voltage drop. This reduces the devices on-
Process refinements have yielded devices with steadily state losses while also maintaining low gate charge (Qg),
increasing density and lower on-resistance. TrenchFET enabling energy-efficient switching at high speeds and
devices have achieved on-resistance less than 1mW for delivering a low RDS(ON) x Qg Figure of Merit (FOM). ST
a 25mm2 silicon die, exclusive of lead resistance. claims that the breakdown voltage of 650V is also high-
Trench MOSFETs employ the same schematic config- er than competing 600V devices, delivering a valuable
safety margin for designers. A
TABLE 10-1. MOSFET PARAMETERS further advantage of STs MD-
Symbol Parameter Description mesh V MOSFETs is a cleaner
V(BR)DSS Breakdown voltage The MOSFETs maximum operating voltage, where the turn-off waveform, enabling
reverse-biased body-drift diode breaks down and current easier gate control and simpler
flows between the source and drain. filtering due to reduced EMI.
VGS(TH) Threshold voltage Minimum gate electrode voltage required to cause the STMicroelectronics STrip-
MOSFET to conduct. FET technology uses an
RDS(ON) On-resistance RDS(ON) = RSOURCE +RCH+RA+RJ+RD+RSUB+RWCML optimized layout and updat-
RSOURCE = Source diffusion resistance ed manufacturing process to
RCH = Channel resistance
improve the gate charge, gate
RA = Accumulation resistance
RJ = JFET component-resistance of the region between resistance and input capaci-
the two body regions tance characteristics. The low
RD = Drift region resistance gate charge enables excellent
RSUB = Substrate resistance switching behavior and the
RWCML= Sum of Bond Wire resistance, contact resistance
low gate resistance means fast
between the source and drain metallization and silicon,
metallization and leadframe transient response. The tech-
contributions. nology also offers an extremely
IDS(MAX) Maximum drain Maximum drain-to-source output current. low figure-of-merit, meaning re-
current duced conduction and switch-
QG Total gate charge Gate charge allows calculation of the amount of current ing losses.
required from the drive circuit to switch the device on in a Among STMicroelectron-
desired length of time. ics introductions is a series
PD The maximum Maximum allowable power dissipation that raises the die of 30V surface-mount power
allowable power temperature to the maximum allowable when the case transistors, achieving on-resis-
dissipation temperature is held at 25oC. tance as low as 2 m (max) to
TJ ( MAX) 25 increase the energy efficiency
PD = of products such as comput-
R JC
where: ers, telecom and networking
TJ(MAX) = Maximum allowable temperature of the p-n junc- equipment. The latest-gener-
tion (normally 150oC or 175oC) ation STripFET VI DeepGATE
RJC = Junction-to-case thermal impedance of the MOSFET. family process has high equiv-
used in networking and industrial applications. packaging and handling precautions for semiconductors
This integrated device offers continuous current up to and is an electronic standard for the time the device can
40 A in the 5 mm by 5 mm PowerPAK MLP55-31L pack- be exposed to ambient room conditions of approximately
age. Besides its high-current capabilities and footprint 30C/60%RH. The reason this is important is that thin
savings, the power stage also lowers package parasitics fine-pitch devices could be damaged during surface
to enable switching frequencies up to 2 MHz further mount technology (SMT) reflow when moisture trapped
shrinking the overall solution size and profile by reducing inside the component expands. Trapped moisture can
the size of the output filter. Its high- and low-side MOS- damage a semiconductor. In extreme cases, cracks will
FETs utilize Gen IV TrenchFET technology to reduce extend to the component surface.
switching and conduction losses. The integrated power According to IPC/JEDECs J-STD-20: Moisture/Reflow
driver IC is compatible with a wide range of PWM control- Sensitivity Classification for Plastic Integrated Circuit (IC)
lers and supports tri-state PWM logic of 5 V. SMTs, there are eight levels of moisture sensitivity. Orig-
The SiC632CD is optimized for synchronous buck inally MOSFETs was rated at MSL 3, which allowed 168
converters, DC/DC voltage regulation modules, and hours of moisture testing. MSL 1 allows an unlimited time
multiphase VRDs for CPUs, GPUs, and memory. To for the moisture test.
increase light-load efficiency in these applications, the Today, there are new applications and power MOS-
driver IC incorporates diode emulation mode circuitry FETs continue to grab them. . This includes Electric
and zero-current detect. An adaptive dead time control Power Steering (EPS) and Micro Hybrid Vehicles, and
helps to further improve efficiency at all load points. To chassis, drive train and power train systems. Besides
support PS4 mode light-load requirements for IMVP8, meeting the AEC Q101 standard, they must meet the
the power stages reduce current consumption to 5 A cost constraints imposed by automotive manufacturers.
when systems are operating in standby mode, and they For EPS, an electric motor driven by a power MOS-
can awake from this state within 5 s. Protection features FET provides steering assist to the driver of a vehicle. A
for the RoHS-compliant, halogen-free devices include typical system employs sensors detect the motion and
undervoltage lockout (UVLO). torque of the steering column, and a computer module
controls system performance. Software allows varying
Power MOSFETs for Synchronous Rectifiers amounts of assistance to be applied depending on driv-
Fig. 10-17 shows a simplified synchronous rectifier ing conditions.
circuit. Typical synchronous rectifiers consist of high-side Electric systems have a fuel efficiency advantage
and low-side MOSFETs, which require different charac- over conventional hydraulic power steering. The electri-
teristics for an optimum design. Generally, the best high cal approach eliminates the belt-driven hydraulic pump
side MOSFET is one with the lowest Qswitch RDS(ON) constantly running, whether assistance is required or not.
figure-of-merit. Qswitch is defined as the post gate thresh- Another major advantage is the elimination of a belt-driv-
old portion of the gate-to-source charge plus the gate-to- en engine accessory, and several high-pressure hydrau-
drain charge (Qgs2 + Qgd). In contrast, the best high side lic hoses between the hydraulic pump, mounted on the
MOSFET must exhibit very low RDS(ON) coupled with good engine, and the steering gear, mounted on the chassis.
Cdv/dt immunity. This simplifies manufacturing and maintenance.
A micro-hybrid system performs a stop-start function
Power MOSFETs for Automotive Applications completely transparent to the driver, during idling, like
Over the last two decades Power MOSFETs have waiting for a traffic light, a starter-alternator turns off the
evolved as a necessary power handling component is engine. Then, the engine restarts very quickly and silently
virtually all automobiles. To be eligible for use in automo- when the drive steps on the accelerator. This technique
tive electronic systems these power MOSFETs must meet cuts fuel consumption and gas emissions at standstill.
the AEC Q101 standard. Some new power MOSFETs are Tests have shown that this can cut fuel consumption
AEC Q101 qualified and will fit in the growing use of elec- about 6%.
tric motors, solenoids and fuel injectors. Power MOSFETs Power MOSFETs have played a major role make
have low on-resistance, 40 V and 100 V maximum oper- automotive systems more reliable. Among the traditional
ating voltage and the ability to tolerate the high-voltage mechanical components that have been eliminated are
transients such as load dump that can occur in automo- shafts, pumps, hoses, fluids, coolers, etc., which reduces
tive electrical systems. the weight of the vehicle and improves fuel efficiency.
An important feature of these new MOSFETs is their Safety improvement is another feature of electronic con-
Moisture Sensitivity Level, or MSL. This relates to the trols that provide more automated functions that cannot
be achieved by mechanical techniques. Compared with junction. Breakdown voltage is primarily determined by
mechanical systems, the electronics trend also allows the resistivity of the epitaxial layer.
easier modification or upgrade of automotive systems. All applications of power MOSFET switches require
MSL-1 preconditioning is required for surface mount some guardbanding when specifying BVDSS rating. It is
capable devices that are put on Temperature Cycling, important to remember that there is a price to be paid for
H3TRB, IOL, and Autoclave tests. If straight leaded this in the form of either higher RDS(on) or larger die. There
devices (such as I-PAK or TO-262) are used, then the de- may be applications where a reduction of conservative
vices are required to undergo the preconditioning with a guardbanding on BVDSS can be justified by an improved
third reflow exposure in lieu of the surface mounting step RDS(on) specification or lower cost without jeopardizing
performance or reliability.
MOSFETs and BJT Comparison Bipolar transistors have ratings for maximum current
Power MOSFETs are capable of operating at very high under continuous and pulsed conditions. Exceeding
frequencies compared with Bipolar Junction Transistors these ratings usually result in device failure. Current rat-
(BJTs) whose switching speed is much slower than for a ings on MOSFET transistors have a different meaning be-
power MOSFET of similar size and voltage rating. Typical cause they behave as a resistor when they turn on. This
rise and fall times of power MOSFETs are of the order of means that the maximum voltage drop or heat generated
several nanoseconds which is two orders of magnitude determines the maximum current. Turning the current on
faster than bipolar devices of similar voltage rating and and off at high speeds reduces the average power or
active area. BJTs are limited to frequencies of less than heat generated, thereby increasing the maximum allow-
100kHz whereas power MOSFETs can operate up to able current.
1MHz before switching losses become unacceptably
high. Recent advances in the design and processing of Power semiconductor Reliability
MOSFETs are pushing this frequency limit higher. Excessive operating voltage can cause power
Power MOSFETs are voltage controlled devices with semiconductor failures because the devices may have
simple drive circuitry requirements. Power BJTs on the small spacing between their internal elements. An even
other hand are current controlled devices requiring large worse condition for a power semiconductor is to have
base drive currents to keep the device in the ON state. high voltage and high current present simultaneously. A
Power MOSFETs have been replacing power BJTs in few nanoseconds at an excessive voltage or excessive
power application due to faster switching capability and current can cause a failure. Most power semiconductor
ease of drive, despite the very advanced state of manu- data sheets specify the maximum voltage that can be
facturability and lower costs of BJTs. applied under all conditions. The military has shown very
BJTs suffer from thermal runaway. The forward volt- clearly that operating semiconductors at 20% below their
age drop of a BJT decreases with increasing tempera- voltage rating provides a substantial improvement in their
ture potentially leading to destruction. This is of special reliability.
significance when several devices are paralleled in order Another common killer of power semiconductors is
to reduce forward voltage drop. Power MOSFETs can be heat. Not only does high temperature destroy devices,
paralleled easily because the forward voltage increases but even operation at elevated, non-destructive tempera-
with temperature, ensuring an even distribution of current tures can degrade useful life. Data sheets specify a max-
among all components. They can withstand simultane- imum junction temperature, which is typically between
ous application of high current and high voltage without 100C and 200oC for silicon. Most power transistors
undergoing destructive failure due to second breakdown. have a maximum junction rating of 125C to 150C, the
However, at high breakdown voltages (>~200V) the safe operating temperature is much lower.
on-state voltage drop of the power MOSFET becomes
higher than that of a similar size bipolar device with Transient Effects
similar voltage rating, making it more attractive to use the Power semiconductors can be destroyed by very
bipolar power transistor at the expense of worse high-fre- short pulses of energy. A major source of destructive
quency performance. transients is caused by turning on or off an inductive
Breakdown voltage (BVDSS) is the drain-to-source volt- load. Protection against these problems involves a care-
age at which a current of 250A starts to flow between ful combination of operating voltage and current margins
source and drain while the gate and the source are short- and protective devices.
ed together. With no bias on the gate, the drain voltage is
entirely supported by the reverse-biased body-drain p-n Power dv/dt and di/dt
The terms dv/dt and di/dt reflect a E silicon wafer, usually called yield. Not
time rate of change of voltage (dv/dt) only does a larger die size mean a
or current (di/dt) describe their reac- PNP disproportionately larger cost, but key
tion to turning on or off a reactive load. parameters may not be the same for
These problems can occur in power JFET all functions of each device on the die.
G
semiconductor switches because all
sections of the device do not behave MOSFET NPN
RB IGBT
in an identical manner when subjected An insulated-gate bipolar transistor
to very high rates of change. It is not C (IGBT) is a three-terminal power semi-
only important to look at the dv/dt and 10-18 Simplified IGBT equivalent conductor device primarily used as an
di/dt values generated within a circuit, circuit electronic switch that combines high
but also turn on and turn off times as C efficiency and relatively fast switch-
well. ing. The IGBT provides the simple
gate-drive characteristics of MOSFETs
G
EMI with the high-current and low-satura-
Switching power on and off at a rapid rate can cause tion-voltage capability of bipolar tran-
E
electromagnetic interference (EMI) that can affect nearby sistors. It combines an isolated-gate
electronic systems. Domestic and international standards 10-19. IGBT FET for the control input and a bipolar
define the amount of EMI that can be emitted. circuit symbol power transistor as a switch in a single
Unclamped Inductive Switching (UIS) device.
Whenever current through an inductance is turned off The IGBT is used in medium- to
quickly, the resulting magnetic field induces a counter high-power applications like switch-mode power sup-
electromagnetic force (CEMF) that can build up surpris- plies, traction motor control and induction heating. Large
ingly high potentials across the switch. With transistor IGBT modules typically consist of many devices in paral-
switches, the full buildup of this induced potential may lel and can have very high current-handling capabilities
far exceed the rated voltage breakdown of the transistor, in the order of hundreds of amperes with blocking volt-
resulting in catastrophic failure. ages of 6000 V. These IGBTs can control loads of hun-
There are two failure modes when subjecting a MOS- dreds of kilowatts. It is equally suitable in resonant-mode
FET to UIS. These failure mechanisms are considered as converter circuits.
either active or passive. The active mode results when The main advantages of IGBT over a Power MOSFET
the avalanche current forces the MOSFETs parasitic bi- and a BJT are:
polar transistor into conduction. In the passive mode the 1. Low on-state voltage drop due to conductivity mod-
instantaneous device temperature reaches a critical val- ulation and on-state current density. So smaller chip size
ue. At this elevated temperature the MOSFETs parasitic is possible, reducing cost.
bipolar transistor causes catastrophic thermal runaway. 2. Low driving power and a simple drive circuit due
In both cases the MOSFET is destroyed. to the input MOS gate structure, which allows relatively
easy control compared with current controlled devices in
Cost Considerations high voltage and high current applications.
As a semiconductor chip gets larger its cost grows 3. Compared with a bipolar transistor, it has better
exponentially. And, there is the cost of the package that current conduction capability as well as forward and
houses the integrated power device and the cost of in- reverse blocking capabilities.
terconnections. In deciding whether to integrate a power Main disadvantages are:
semiconductor into an integrated circuit or use two sepa- 1. Slower switching speed compared with a power
rate devices, look at the die size of each. If an integrated MOSFET, but better than a BJT.
power semiconductor and a discrete power semicon- 2. Possibility of latchup due to the internal PNPN thy-
ductor have large die, the die cost dominates the overall ristor structure.
cost, it would be cheaper to use two parts. A simple equivalent circuit model of an IGBT is shown
Integrated power semiconductors make sense when in Fig. 10-18. It contains MOSFET, JFET, NPN and PNP
the die sizes are moderate, or there are multiple outputs. transistors. The collector of the PNP is connected to the
This is so because the package and handling costs base of the NPN and the collector of the NPN is connect-
offset the increased silicon cost. A major impact on cost ed to the base of the PNP through the JFET. The NPN
is the number of good devices that can be obtained from and PNP transistors represent the parasitic thyristor that
60
voltage and current ratings similar to that of the bipolar
TJ = 25C
transistor. However, the presence of an isolated gate in
50 VGE = 17 V to 13 V
an IGBT makes it a lot simpler to drive than the BJT as it
requires much less drive power.
IC, COLLECTOR CURRENT (A)
switching loss. Therefore the IGBT is well suited for motor 5. Jeff Perry, MOSFET Selection For Switching Power Supply
drive control and other hard switching applications. Incor- Systems, PET, April, 2012.
porated into the device is a rugged copackaged re- 6. Avery Lu, PowerStage MOSFETs - More Power in Less
verse recovery diode with a low forward voltage. Figure Space, PET, March, 2012.
10-20 is this IGBTs output characteristics at 25C. 7. Shah, Hemal, Demystifying Power MOSFETs Avalanche
Features Ruggedness, PET, August, 2011.
Low Saturation Voltage Resulting in Low Conduction 8. Havanur, Sanjay, Designing with DrMOS Part 1: Concept and
Loss Features, PET, February, 2011.
Low Switching Loss in Higher Frequency Applications 9. Havanur, Sanjay, Designing with DrMOS, Part II: Application
Soft Fast Reverse Recovery Diode Guidelines, PET, March, 2011.
10 _s Short Circuit Capability 10. Satyavrat Laud, IGBT FAQs, PET, May, 2014.
Excellent Current versus Package Size Performance 11. Klaus Vogel, Impact of IGBT and Diode Evolution on Design
Density of High Power Density Inverter Modules, PET, November, 2013.
This is a PbFree Device 12. Benjamin Jackson, High Speed IGBTs Take on the Super
Junction MOSFET, PET, October, 2012.
Related Articles 13. Clemente, Steve, Webtools Speed Up IGBT Selection and
1. Sam Davis, Unique MOSFETs Automatically Regulate and Design, PET, February, 2011.
Balance Series-Connected Supercaps, PET, December, 2013. 14. Sam Davis, LT4320 Ideal Diode Bridge Controller, PET,
2. Kim Gauen, Intelligent Power Switches for 24V Vehicular August, 2013.
Systems, PET, January, 2013. 3. 15. Davis, Sam, Schottky Diodes: the Old Ones Are Good, the
3. Chris Swartz, Hot-Swap Controller and Circuit Breaker New Ones Are Better, PET, March, 2011.
Protects External MOSFETs, PET, September, 2012.
4. Sam Davis, Semiconductor Back-to-Basics: Power MOSFETs, BACK TO TABLE OF CONTENTS
PET, September, 2012.
Sponsored by
LEARN MORE
LEARN MORE GET THE
APP NOTE
CHAPTER 11:
WIDE BANDGAP
SEMICONDUCTORS
W
ide bandgap (WBG) semiconductor making possible more compact, less costly product
materials allow smaller, faster, more designs.
reliable power electronic components As manufacturing capabilities improve and market ap-
and with higher efficiency than their plications expand, costs are expected to decrease, mak-
silicon-based counterparts. These ing WBG-based devices competitive with less expensive
capabilities make it possible to re- Si-based devices. Several manufacturing challenges
duce weight, volume, and life-cycle costs in a wide range must be addressed to make WBG materials more cost
of power applications. Harnessing these capabilities can effective, including:
lead to dramatic energy savings in industrial processing Cost of producing larger-diameter wafers needs to be
and consumer appliances, accelerate widespread use of reduced.
electric vehicles and fuel cells, and help integrate renew- Novel device designs that effectively exploit the prop-
able energy onto the electric grid. erties of WBG materials are needed to achieve the volt-
WBG semiconductors permit devices to operate at age and current ratings required in certain applications.
much higher temperatures, voltages, and frequencies Alternative packaging materials or designs are also
making the power electronic modules using these mate- needed to withstand the high temperatures in WBG
rials significantly more powerful and energy-efficient than devices.
those made from conventional semiconductor materials. Existing systems may have to be redesigned to inte-
Figure 11-1 compares the breakdown voltages of silicon grate the WBG devices in ways that deliver their unique
and WBG semiconductors SiC and GaN. capabilities.
WBG semiconductors are expected to pave the way
for exciting innovations in power electronics, solid-state TABLE 11-1. WBG MATERIAL COMPARISON
lighting, and other diverse applications across multiple Material Chemical Symbol Bandgap Energy
industrial and clean energy devices with vastly superior (eV)
performance compared to current technology. Projected Germanium Ge 0.7
WBG benefits are: Silicon Si 1.1
Elimination of up to 90% of the power losses that
Gallium Arsenide GaAs 1.4
currently occur during ac-to-dc and dc-to-ac power
Silicon Carbide SiC 3.3
conversion.
Gallium Nitride GaN 3.4
Operation up to 10 times higher than Si-based devices,
which will enhance high-power performance. Diamond C 5.5
Operation up to higher maximum temperature of Si-
based devices, which will provide better overall system SiC
reliability. Using SiC (silicon carbide) can reduce on-resistance
Enabling of smaller and lighter systems with reduced to two orders of magnitude in compared with existing
lifecycle energy use, along with opportunities for new Si devices. Use of SiC devices can reduce power loss
applications. extensively, when applied to power conversion systems.
Operation at higher frequencies than Si-based devices, These SiC devices such as power MOSFET or IGBT
ment. EPC produces gallium nitride 11-2. Enhancement the power devices or a short circuit
on silicon wafers using standard D mode GaN has a will result. Enhancement mode
MOS processing equipment. GaNs circuit schematic devices do not have this problem:
exceptionally high electron mobility similar to silicon with zero bias on the gate, an en-
and low temperature coefficient al- G MOSFETs with gate hancement mode device is off and
lows very low RDS(ON), while its later- (G), drain (D), and will not conduct current.
al device structure and majority car- source (S). The threshold of enhancement
rier diode provide exceptionally low mode GaN FETs is lower than that
QG (total gate charge) and zero QRR of silicon MOSFETs. This is made
(source-drain recovery charge). As
S possible by the almost flat rela-
a result, GaN devices can handle tionship between threshold and
tasks benefitted by very high switching speeds. temperature along with the very low gate-to-drain capac-
Initially, GaN-on-silicon transistors were depletion itance (CGD). The device starts to conduct significant
mode types. That is, they operated like a normally on current at 1.6V, so care must be taken to ensure a low
power switch that required a negative voltage to turn impedance path from gate-to-source when the device
them off. The ideal mode for designers is an enhance- needs to be held off during high speed switching in a
ment mode transistor that is normally non-conducting rectifier function.
and requires a positive voltage to turn it on, like the pres- The threshold of depletion mode GaN HEMTs ranges
ent silicon-only N-channel MOSFETs. EPC produces an from -5 V to -20 V.
enhancement mode GaN transistor using a proprietary Besides its low RDS(ON), the lateral structure of the
process with a GaN-on-silicon structure. In operation, a enhanced GaN FET also makes it a very low-capacitance
positive gate voltage turns the enhancement mode GaN device. It can switch hundreds of volts in nanoseconds,
transistor on. giving it multi-megahertz capability. With a lateral struc-
An advantage of the GaN transistor is that its block- ture, CGD comes only from a small corner of the gate and
ing voltage rating depends on the distance between the is much lower than the same capacitance in a vertical
drain and gate; the longer the distance, the higher the MOSFET.
voltage rating. Another GaN advantage is its very low Gate-to-source capacitance (CGS) consists of the
on-resistance. junction from the gate in channel, and the capacitance of
GaN transistors borrowed the same nomenclature as the dielectric between the gate and the field plate. CGS
their silicon brethren: gate, drain, and source, as shown is large compared with CGD, giving GaN FETs good dv/dt
Fig. 11-2. In addition, on-resistance and breakdown immunity, but still small compared with silicon MOSFETs.
voltage of a GaN device have a similar meaning as their The drain-to-source capacitance (CDS) is also small,
silicon counterparts. On-resistance (RDS(ON)) versus gate- being limited to the capacitance across the dielectric
source voltage curves are similar to silicon MOSFETs. from the field plate to the drain. Capacitance versus volt-
The temperature coefficient of GaN FETs on-resistance age curves for GaN FETs are similar to those for silicon,
is similar to the silicon MOSFET as it is positive, but the except that for a similar resistance, its capacitance is
magnitude is somewhat less. significantly lower.
GaN has a higher critical electric field strength than The GaN transistor structure is a purely lateral device,
silicon. Its higher electron mobility enables a GaN de- without the parasitic bipolar junction common to silcon
vice to have a smaller size for a given on-resistance and MOSFETs. Therefore, the enhancement GaN reverse bias
breakdown voltage than a silicon semiconductor. Com- or diode operation has a different mechanism, but a
pared to silicon devices, this also allows devices to be similar function. With zero bias gate-to-source there is an
physically smaller and their electrical terminals closer absence of electrons under the gate region. As the drain
together for a given breakdown voltage requirement. voltage decreases, a positive bias on the gate is created
The two types are the depletion mode and enhance- relative to the drift region, injecting electrons under the
ment mode. The depletion mode transistor is normally on gate. Thus, there are no minority carriers involved in con-
and is turned off with a negative voltage relative to the duction, and therefore no reverse recovery losses.
drain and source electrodes. In contrast, the enhance- Although QRR is zero, output capacitance (COSS)
ment mode transistor is normally off and is turned on has to be charged and discharged with every switch-
by positive voltage applied to the gate. Depletion mode ing cycle. For devices of similar RDS(ON), enhancement
transistors are inconvenient because at start-up of a GaN FETs have significantly lower COSS than silicon
power converter, a negative bias must first be applied to MOSFETs. It takes a bias on the gate greater than the
D
VDD
HV GaN HEMT
HSB
UVLO HSGPU
Level Output
Shifter Driver
HSGPD
Dead HSS LV Si MOSFET
IN Time Logic
Controller
LSB G
EN LSGPU
Level Output K S
RDTL Shifter Driver 11-5. Transphorm employs a
LSGPD
cascode circuit to drive the GaN
RDTH LSS device. Drain, gate, and source
are similar to a silicon MOSFETs
GND LSO D, G, and S, and K is the Kelvin
11-4. PE29100 functional diagram. contact for the gate return.
-80.0 LMG5200
-100.0 Figure 11-8 shows the LMG5200 from
-3.0 -2.0 -1.0 0.0 1.0 2.0 3.0
Texas Instruments, a half-bridge, GaN
VD Courtesy of Efficient Power Conversion power stage with a highly integrated
11-6. Output characteristics of a depletion mode GaN transistor. high-side and low-side gate drivers that
includes built-in UVLO protection circuit-
100.0 ry and an overvoltage clamp circuitry.
VGS The clamp circuitry limits the bootstrap
80.0
5V refresh operation to ensure that the
60.0 4V high-side gate driver overdrive does
20.0 3V not exceed 5.4 V. The device integrates
2V two 19-m GaN FETs in a half-bridge
40.0 1V configuration. The device can be used
ID 00.0 0V in many isolated and non-isolated topol-
ogies, allowing very simple integration.
-20.0
The package is designed to minimize
-40.0 the loop inductance while keeping the
-60.0 PCB design simple. The drive strengths
for turn-on and turn-off are optimized to
-80.0 ensure high-voltage slew rates without
-100.0 causing any excessive ringing on the
-3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 gate or power loop.
VD Courtesy of Efficient Power Conversion Propagation delays between the
11-7. Output characteristics of an enhancement mode GaN transistor. high-side gate driver and low-side gate
driver are matched to allow very tight control of dead 11-10. The all-SiC
time. Controlling the dead time is critical in GaN-based 300A, 1.2kV half-
applications to maintain high efficiency. HI and LI can bridge module
be independently controlled to minimize the third quad- is packaged in
rant conduction of the low-side FET for hard switched industry-standard
buck converters. A very small propagation mismatch 62mm housing.
between the HI and LI to the drivers for both the falling
and rising thresholds ensures dead times of <10 ns.
Co-packaging the GaN FET half-bridge with the driver
ensures minimized common source inductance.
This minimized inductance has a significant perfor- UVLO, if there is sufficient voltage (VCC > 2.5 V), the driv-
mance impact on hard-switched topologies. er actively pulls the high-side and low-side gate driver
The built-in bootstrap circuit with clamp prevents the output low. The UVLO threshold hysteresis of 200 mV
high-side gate drive from exceeding the GaN FETs maxi- prevents chattering and unwanted turn-on due to voltage
mum gate-to-source voltage (VGS) without any additional spikes. Use an external VCC bypass capacitor with a
external circuitry. The built-in driver has an undervoltage value of 0.1 F or higher. A size of 0402 is recommended
lockout (UVLO) on the VDD and bootstrap (HB-HS) rails. to minimize trace length to the pin. You should place the
When the voltage is below the UVLO threshold voltage, bypass and bootstrap capacitors as close to the device
the device ignores both the HI and LI signals to prevent as possible to minimize parasitic inductance.
the GaN FETs from being partially turned on. Below
All-SiC 300A
LMG5200 Crees all-SiC 300A, 1.2kV half-
UVLO and 2 HB bridge module circuit (Fig. 11-9)
Clamp
1 VIN is packaged in industry-standard
62mm housing (Fig. 11-10). The
Level
VCC 6 Shifter module reduces energy loss due to
switching by more than five times
HI 4 compared to the equivalent silicon
3 HS
solution. This efficiency enables an
all-SiC high power converter rated up
to the megawatt level.
UVLO 8 SW
The all-SiC 62mm half-bridge
module allows designers to reduce
the amount of magnetic and cooling
LI 5 9 PGND
elements, delivering double the pow-
7 AGND er density and a lower system cost
while also reducing end user cost
of ownership. Offering a simplified
11-8. LMG5200 half-bridge with GaN output transistors and internal gate two-level topology that is feasible at
drivers. higher frequencies, the new module
can also eliminate the need to invest
1 in multi-level silicon-based solutions.
This Cree SiC power module is available with multiple
gate driver options and is pin-compatible to standard
62mm half-bridge modules, including IGBT modules rat-
3 2 ed at 450A or more. This allows designers to quickly and
easily evaluate the modules unparalleled capabilities.
The all-SiC 300A, 1.2kV half-bridge module is avail-
able as part number CAS300M12BM2. Companion gate
drivers are also available.
4 5 6 7 A newer module design also configured as in Fig.
11-9. Configuration of the Cree half-bridge. 11-9 is said to be the industrys most optimized to
achieve the unique benefits of SiC technologywith a 5. Sam Davis, Gallium Nitride Transistors Switch in the Sub-
66% reduction in module inductance to 5.5nH, compared Nanosecond Timeframe, powerelectronics.com, October 2013.
to competitive power products at 15nH. This reduction 6. Sam Davis, GaN Basics: FAQs, powerelectronics.com,
in module inductance enables faster switching speeds, October 2013.
higher frequency operation, and ultra-low losses. 7. Michael A Briere, Characterizing Performance of Mid-voltage
Available as part number CAS325M12HM2, the GaN-on-Si Devices, powerelectronics.com, July 2013.
high-performance power module is configured in a 8. Michael de Rooij, eGaN FET - Silicon Power Shoot Out: A
half-bridge topology comprised of seven 1.2kV 25m Retrospective, powerelectronics.com, July 2013.
C2M SiC MOSFETs and six 1.2kV 50A Z-Rec Schottky 9. Robinson Law, SiC MOSFET Gate Drive Optocouplers,
diodes. The companion gate driver (CGD15HB62LP) is powerelectronics.com, June 2014.
specifically designed for integration with the module to fit 10. Sam Davis, SiC Transistor Basics: FAQs, powerelectronics.
within the 62mm mounting footprint. An engineering com, October 2013.
evaluation kit that includes both the module and the gate 11. Mark Loboda, Design Considerations for SiC-based Power
driver is also available so design engineers can quickly Electronics, powerelectronics.com, November 2012.
and easily test the performance of the new device in their 12. Roger Allan, SiC: A Rugged Power Semiconductor
systems. Compound To Be Reckoned With, powerelectronics.com,
February 2012.
Related Articles 13. Deepak Veereddy, SiC Super Junction Transistors Offer
1. Eric Faraci, Enabling Industrial and Automotive Breakthrough High Temp Performance, powerelectronics.com,
Multimegahertz Buck Converters with GaN, powerelectronics. November 2011.
com, September 2015. 14. Sam Davis, 1200V SiC MOSFET Poised to Replace Si
2. Sam Davis, Half-Bridge GaN FET Module Comes In QFN, MOSFETs and IGBTs, powerelectronics.com, February 2011.
powerelectronics.com, March 2015. 15. Sam Davis, SiC and GaN Vie for Slice of the Electric Vehicle
3. Sam Davis, Fourth Generation Boosts eGaN FET Pie, powerelectronics.com, November 2009.
Performance, powerelectronics.com, July 2014.
4. Sam Davis, Wide Bandgap Semiconductors, BACK TO TABLE OF CONTENTS
powerelectronics.com, June 2014.
Sponsored by