Application NoteAPT-0403 Rev B

March 2, 2006

1

Power MOSFET TutorialJonathan Dodge, P.E.

Applications Engineering ManagerAdvanced Power Technology

405 S.W. Columbia StreetBend, OR 97702

IntroductionPower MOSFETs are well known for superiorswitching speed, and they require very little gate drivepower because of the insulated gate. In these respects,power MOSFETs approach the characteristics of an“ideal switch”. The main drawback is on-resistanceRDS(on) and its strong positive temperature coefficient.This application note explains these and other mainfeatures of high voltage N-channel power MOSFETs,and provides useful information for device selectionand application. Advanced Power TechnologyMOSFET datasheet information is also explained.

Power MOSFET StructureGate Metal

Gate Poly Source Metal

P- BodyP- Body

Source MetalN+ N+

N- Drain

J-FET

N+ Substrate

RBchannel

BodyDiode

Parasitic BJT

Rdrift

Figure 1 N-Channel MOSFET Cross Section

Figure 1 shows a cross section of an APT N-channelpower MOSFET structure. (Only N-channelMOSFETs are discussed here.) A positive voltageapplied from the source to gate terminals causeselectrons to be drawn toward the gate terminal in thebody region. If the gate-source voltage is at or abovewhat is called the threshold voltage, enough electronsaccumulate under the gate to cause an inversion n-typelayer; forming a conductive channel across the bodyregion (the MOSFET is enhanced). Electrons can flowin either direction through the channel. Positive (orforward) drain current flows into the drain as electronsmove from the source toward the drain. Forward draincurrent is blocked once the channel is turned off, and

drain-source voltage is supported by the reverse biasedbody-drain p-n junction.

In N-channel MOSFETs, only electrons flow duringforward conduction – there are no minority carriers.Switching speed is only limited by the rate that chargeis supplied to or removed from capacitances in theMOSFET. Therefore switching can be very fast,resulting in low switching losses. This is what makespower MOSFETs so efficient at high switchingfrequency.

RDS(on)

The main components of on resistance RDS(on) includethe channel, JFET (accumulation layer), drift region,and parasitics (metallization, bond wires, and package).At voltage ratings above about 150V, drift regionresistance dominates RDS(on).

Figure 2 RDS(on) vs. Current, APT50M75B2LL

The effect of current on RDS(on) is relatively weak inhigh voltage MOSFETs . Looking at Figure 2,doubling the current results in only about a 6% increasein RDS(on).

Application NoteAPT-0403 Rev B

March 2, 2006

2

Figure 3 RDS(on) vs. Temperature, APT50M75B2LL

Temperature on the other hand has a strong effect onRDS(on). As seen in Figure 3, on resistanceapproximately doubles from 25ºC to 125ºC. Thetemperature coefficient of RDS(on) is the slope of thecurve in Figure 3 and is always positive because ofmajority-only carriers. The strong positive RDS(on)temperature coefficient compounds the I2·R conductionloss as temperature increases.

The positive RDS(on) temperature coefficient is a nicefeature when paralleling power MOSFETs because itensures thermal stability. It does not however ensureeven current sharing. This is a common misconception[1]. What really makes MOSFETs so easy to parallelis their relatively narrow part-to-part parameterdistribution, particularly RDS(on), combined with thesecurity from current hogging provided by the positiveRDS(on) temperature coefficient.

For any given die size, RDS(on) also increases withincreasing voltage rating V(BR)DSS, as shown Figure 4.

RDS(on) versus VDSS

02468

100 300 500 700 900 1100

V(BR)DSS (Volts)

RD

S(on

)

Nor

mal

ized

to50

0V

Figure 4 Normalized RDS(on) vs. V(BR)DSS

A curve fit of rated RDS(on) versus V(BR)DSS for PowerMOS V and Power MOS 7 MOSFETs reveals thatRDS(on) increases as the square of V(BR)DSS. This non-linear relationship between RDS(on) and V(BR)DSS is acompelling reason to research ways to reduce theconduction loss of power transistors [2].

Intrinsic and Parasitic ElementsJFETWithin the structure of a MOSFET, you can imagine anintegral JFET shown in Figure 1. This JFET has asignificant influence on RDS(on) and is part of thenormal operation of the MOSFET.

Intrinsic Body DiodeThe body-drain p-n junction forms an intrinsic diodecalled the body diode (see Figure 1). Reverse draincurrent cannot be blocked because the body is shortedto the source, providing a high current path through thebody diode. Enhancing the device reduces conductionloss when reverse drain current flows because electronsflow through the channel in addition to electrons andminority carriers flowing through the body diode.

The intrinsic body diode is convenient in circuits thatrequire a path for reverse drain current (often calledfreewheeling current), such as bridge circuits. Forthese circuits, FREDFETs are offered with improvedreverse recovery characteristics. FREDFET is simply atrade name Advanced Power Technology uses todistinguish a MOSFET that has additional processingsteps to speed up the reverse recovery of the intrinsicbody diode. There is no separate diode in a FREDFET;it is the MOSFET intrinsic body diode. Either electronirradiation (which is usually used) or platinum dopingis used for minority carrier lifetime control in the bodydiode, greatly reducing the reverse recovery charge andtime.

A side effect of FREDFET processing is higherleakage current, particularly at high temperature.However, considering that MOSFETs have lowleakage current to begin with, the added leakagecurrent of a FREDFET is normally of no concernbelow 150°C junction temperature. Depending on theirradiation dose, a FREDFET may have a higher RDS(on)rating than a corresponding MOSFET. The body diodeforward voltage is also slightly higher for a FREDFET.Gate charge and switching speed are identical betweenMOSFETs and FREDFETs. From here on, the termMOSFET will be used for both MOSFETs andFREDFETs unless specifically stated otherwise.

Application NoteAPT-0403 Rev B

March 2, 2006

3

The reverse recovery performance of a MOSFET oreven of a FREDFET is “crummy” compared to adiscrete fast recovery diode. In a hard switchedapplication operating at 125°C, the turn-on loss in theswitch due to the reverse recovery current of the bodydiode is about five times higher than if a discrete fastrecovery diode is used. There are two reasons for this:

1. The area of the body diode is the same as thearea of the MOSFET or FREDFET, whereas thearea of a discrete diode for the same functioncan be much smaller and hence have muchlower recovery charge.

2. The body diode of a MOSFET or even aFREDFET is not optimized for reverse recoverylike a discrete diode is.

Like any conventional silicon diode, body diodereverse recovery charge and time depend ontemperature, di/dt, and current. The forward voltage ofthe body diode, VSD, decreases with temperature byabout 2.5 mV/°C.

Parasitic Bipolar TransistorThe layered MOSFET structure also forms a parasiticNPN bipolar junction transistor (BJT), and turning it onis definitely not part of normal operation. If the BJTwere to turn on and saturate, it would result in acondition called latchup, where the MOSFET cannotbe turned off except by externally interrupting the draincurrent. High power dissipation during latchup candestroy the device.

The base of the parasitic BJT is shorted to the source toprevent latchup and because breakdown voltage wouldbe greatly reduced (for the same RDS(on)) if the basewere allowed to float. It is theoretically possible forextremely high dv/dt during turn-off to cause latchup.For modern, conventional power MOSFETs however,it is very difficult to build a circuit capable ofachieving such high dv/dt.

There is a risk of turning on the parasitic BJT if thebody diode conducts and then commutates off withexcessively high dv/dt. High commutation dv/dtcauses high current density of minority carriers(positive carriers, or holes) in the body region, whichcan build up enough voltage across the body resistanceto turn on the parasitic BJT. This is the reason for thepeak commutating (body diode recovery) dv/dt limit inthe datasheet. Peak commutating dv/dt is higher for aFREDFET compared to a MOSFET because ofreduced minority carrier lifetime.

Temperature EffectsSwitching SpeedSwitching speed and loss are practically unaffected bytemperature because the capacitances are unaffected bytemperature. Reverse recovery current in a diodehowever increases with temperature, so temperatureeffects of an external diode (be it a discrete diode or aMOSFET or FREDFET body diode) in the powercircuit affect turn-on switching loss.

Threshold VoltageThe threshold voltage, denoted as VGS(th), is really aturn-off specification. It tells how many milliamps ofdrain current will flow at the threshold voltage, so thedevice is basically off but on the verge of turning on.The threshold voltage has a negative temperaturecoefficient, meaning the threshold voltage decreaseswith increasing temperature. This temperaturecoefficient affects turn-on and turn-off delay times andhence the dead-time requirement in a bridge circuit.

Figure 5 Transfer Characteristic, APT50M75B2LL

Transfer CharacteristicFigure 5 shows the transfer characteristic for anAPT50M75B2LL MOSFET. The transfercharacteristic depends on both temperature and draincurrent. In Figure 5, below 100 Amps the gate-sourcevoltage has a negative temperature coefficient (lessgate-source voltage at higher temperature for a givendrain current). Above 100 Amps, the temperaturecoefficient is positive. The gate-source voltagetemperature coefficient and the drain current at whichit crosses over from negative to positive are importantfor linear mode operation [3].

Application NoteAPT-0403 Rev B

March 2, 2006

4

Breakdown VoltageBreakdown voltage has a positive temperaturecoefficient, as will be discussed in the DatasheetWalkthrough section.

Short Circuit CapabilityShort circuit withstand capability is not typically listedin the datasheet. This is simply because conventionalpower MOSFETs are unmatched for short circuitwithstand capability compared to IGBTs or otherdevices with higher current density. It goes withoutsaying that MOSFETs and FREDFETs are short circuitcapable.

Datasheet WalkthroughThe intent of datasheets provided by APT is to includerelevant information that is useful and convenient forselecting the appropriate device as well as predictingits performance in an application. Graphs are providedthat can be used to extrapolate from one set ofoperating conditions to another. It should be noted thatthe graphs represent typical performance, notminimums nor maximums. Performance also dependssomewhat on the circuit; different test circuits yieldslightly different results.

Maximum RatingsVDSS – Drain-Source VoltageThis is a rating of the maximum drain-source voltagewithout causing avalanche breakdown, with the gateshorted to the source and the device at 25°C.Depending on temperature, the avalanche breakdownvoltage could actually be less than the VDSS rating. Seethe description of V(BR)DSS in Static ElectricalCharacteristics.

VGS – Gate-Source VoltageVGS is a rating of the maximum voltage between thegate and source terminals. The purpose of this rating isto prevent damage of the gate oxide. The actual gateoxide withstand voltage is typically much higher thanthis but varies due to manufacturing processes, sostaying within this rating ensures application reliability.

ID – Continuous Drain CurrentID is a rating of the maximum continuous DC currentwith the die at its maximum rated junction temperatureTJ(max) and the case at 25°C and sometimes also at ahigher temperature. It is based on the junction-to-case

thermal resistance rating RθJC and the case temperatureTC as follows:

J (max)

J(max) C 2D D DS(on)@ T

JC

T TP I R

Rθ

−= = ⋅ (1)

This equation simply says that the maximum heat that

can be dissipated, J(max) C

JC

T TRθ

−, equals the maximum

allowable heat generated by conduction loss,

J (max)

2D DS(on)@ TI R⋅ , where

J (max)DS(on)@TR is the onresistance at the maximum junction temperature.Solving for ID:

J (max)

J(max) CD

JC DS(on)@ T

T T 1IR Rθ

−= ⋅ (2)

Note that there are no switching losses involved in ID,and holding the case at 25°C is seldom feasible inpractice. Because of this, actual switched current istypically less than half of the ID @ TC = 25°C rating ina hard switched application; one fourth to one third iscommon.

Graph of ID versus TC

This graph is simply the solution to (2) over a range ofcase temperatures. Switching losses are not included.Figure 6 shows an example. Note that in some cases,the package leads limit the continuous current(switched current can be higher): 100 Amps for TO-247 and TO-264 packages, 75 Amps for TO-220package, and 220 Amps for the SOT-227 package.

Figure 6 Maximum Drain Current vs. CaseTemperature, APT50M75B2LL

Application NoteAPT-0403 Rev B

March 2, 2006

5

IDM – Pulsed Drain CurrentThis rating indicates how much pulsed current thedevice can handle, which is significantly higher thanthe rated continuous DC current. The purposes of theIDM rating are:• To keep the MOSFET operating in the Ohmic

region of its output characteristic. See Figure 7.There is a maximum drain current for acorresponding gate-source voltage that a MOSFETwill conduct. If the operating point at a givengate-source voltage goes above the Ohmic region“knee” in Figure 7, any further increase in draincurrent results in a significant rise in drain-sourcevoltage (linear mode operation) and a consequentrise in conduction loss. If power dissipation is toohigh for too long the device may fail. The IDMrating is set below the “knee” for typical gate drivevoltages.

• A current density limit to prevent die heating thatotherwise could result in a burnout site.

• To avoid problems with excessive current throughthe bond wires in case the bond wires are the“weak link” instead of the die.

Increasing VGSOhmicregion

vDS

iD

0 BVDSS

Activeregion

VGS4

VGS3

VGS2

VGS1

Figure 7 MOSFET Output Characteristic

Regarding the thermal limitation on IDM, temperaturerise depends upon the pulse width, time betweenpulses, heat dissipation, and RDS(on) as well as the shapeand magnitude of the current pulse. Simply stayingwithin the IDM limit does not ensure that the maximumjunction temperature will not be exceeded. See thediscussion on transient thermal impedance underThermal and Mechanical Characteristics for a way toestimate junction temperature during a current pulse.

PD – Total Power DissipationThis is a rating of the maximum power that the devicecan dissipate and is based on the maximum junctiontemperature and the thermal resistance RθJC at a casetemperature of 25ºC.

J(max) CD

JC

T TP

Rθ

−= (3)

The linear derating factor is simply the inverse of RθJC.

TJ, TSTG – Operating and Storage JunctionTemperature RangeThis is the range of permissible storage and operatingjunction temperatures. The limits of this range are setto ensure a minimum acceptable device service life.Operating well within the limits of this range cansignificantly enhance the service life.

EAS – Single Pulse Avalanche EnergyIf a voltage overshoot (typically due to leakage andstray inductances) does not exceed the breakdownvoltage, then the device will not avalanche and hencedoes not need to dissipate avalanche energy.Avalanche energy rated devices offer a safety net forover-voltage transients, depending on the amount ofenergy dissipated in avalanche mode.

All devices that are avalanche energy rated have an EASrating. Avalanche energy rated is synonymous withunclamped inductive switching (UIS) rated. EASindicates how much reverse avalanche energy thedevice can safely absorb.

Conditions for a test circuit are stated in a footnote, and

the EAS rating is equal to2DL i

2⋅ , where L is the value of

an inductor carrying a peak current iD, which issuddenly diverted into the drain of the device undertest. It is the inductor voltage exceeding thebreakdown voltage of the MOSFET that causes theavalanche condition. An avalanche condition allowsthe inductor current to flow through the MOSFET,even though the MOSFET is in the off state. Energystored in the inductor is analogous to energy stored inleakage and/or stray inductances and is dissipated inthe MOSFET.

When MOSFETs are paralleled, it is highly unlikelythat they have exactly the same breakdown voltage.Typically, one device will avalanche first andsubsequently take all the avalanche current (energy).

Application NoteAPT-0403 Rev B

March 2, 2006

6

EAR – Repetitive Avalanche EnergyA repetitive avalanche rating has become “industrystandard” but is meaningless without information aboutthe frequency, other losses, and the amount of cooling.Heat dissipation (cooling) often limits the repetitiveavalanche energy. It is also difficult to predict howmuch energy is in an avalanche event. What the EARrating really says is that the device can withstandrepetitive avalanche without any frequency limitation,provided the device is not overheated, which is true ofany avalanche capable device. During designqualification, it is good practice to measure the deviceor heat sink temperature during operation to see thatthe MOSFET does not overheat, especially ifavalanching is possible.

IAR – Avalanche CurrentFor some devices, the propensity for current crowdingin the die during avalanche mandates a limit inavalanche current IAR. Thus avalanche current is the“fine print” of avalanche energy specifications; itreveals the true capability of a device.

Static Electrical CharacteristicsV(BR)DSS – Drain-Source BreakdownVoltageV(BR)DSS (sometimes called BVDSS) is the drain-sourcevoltage at which no more than the specified draincurrent will flow at the specified temperature and withzero gate-source voltage. This tracks the actualavalanche breakdown voltage.

Figure 8 Normalized Breakdown Voltage versusTemperature, APT50M75B2LL

As shown in Figure 8, V(BR)DSS has a positivetemperature coefficient. A MOSFET can block more

voltage when hot than when cold. In fact, when cold,V(BR)DSS is less than the maximum VDSS rating, which isspecified at 25°C. For the example shown in Figure8, at -50ºC, V(BR)DSS is about 90% of the 25ºCmaximum VDSS rating.

VGS(th) – Gate Threshold VoltageThis is the gate-source voltage at which drain currentbegins to flow, or stops flowing when switching off theMOSFET. Test conditions (drain current, drain-sourcevoltage, and junction temperature) are also specified.All MOS gated devices exhibit variation in thresholdvoltage between devices, which is normal. Therefore,a range in VGS(th) is specified, with the minimum andmaximum representing the edges of the VGS(th)distribution. As discussed previously underTemperature Effects, VGS(th) has a negative temperaturecoefficient, meaning that as the die heats up, theMOSFET will turn on at a lower gate-source voltage.

RDS(on) – On ResistanceThis is the drain-source resistance at a specified draincurrent (usually half the ID current) and gate-sourcevoltage (usually 10 Volts), and at 25°C unlessotherwise specified.

IDSS – Zero Gate Voltage Drain CurrentThis is the drain-source leakage current at a specifieddrain-source voltage when the gate-source voltage iszero. Since leakage current increases with temperature,IDSS is specified both at room temperature and hot.Leakage power loss is IDSS times drain-source voltageand is usually negligible.

IGSS – Gate-Source Leakage CurrentThis is the leakage current that flows through the gateterminal at a specified gate-source voltage.

Dynamic Characteristics

n+ n+p

n-

n+

SourceGate

Drain

Cgd CgdCds

Cgs

Figure 9 Power MOSFET Structure Capacitances

Application NoteAPT-0403 Rev B

March 2, 2006

7

Figure 9 shows the locations of power MOSFETintrinsic capacitances. The values of the capacitancesare determined by the structure of the MOSFET, thematerials involved, and by the voltages across them.These capacitances are independent of temperature, soMOSFET switching speed is also insensitive totemperature (except for a minor effect related to thethreshold voltage changing with temperature).

Capacitances Cgs and Cgd, vary with the voltage acrossthem because they are affected by depletion layerswithin the device [4]. However, Cgs has only a smallvoltage change across it compared to Cgd andconsequently a small capacitance change. The changein Cgd with drain-to-gate voltage can be as much as afactor of 100 or more.

Figure 10 shows the intrinsic capacitances from acircuit perspective. The gate-to-drain and gate-to-source capacitances impact the susceptibility ofunwanted dv/dt induced turn-on in bridge circuits.

Cgd

Cds

Cgs

D

S

GGate drivecircuit

Gatesupplyvoltage

RG

Optionalnegative gatesupply voltage

+-

Minimize this area

Figure 10 Power MOSFET Intrinsic Capacitances

In a nutshell, the smaller Cgd is the less susceptibilitythere is for dv/dt induced turn-on. Also, Cgs and Cgdform a capacitive voltage divider, and a large ratio ofCgs to Cgd is desirable for dv/dt immunity. This ratiomultiplied by the threshold voltage is a figure of meritfor dv/dt turn-on immunity, and APT power MOSFETslead the industry with this and other figures of merit.

Ciss – Input CapacitanceThis is the input capacitance measured between thegate and source terminals with the drain shorted to thesource for AC signals. Ciss is made up of the gate todrain capacitance Cgd in parallel with the gate to sourcecapacitance Cgs, or

iss gs gdC C C= +

The input capacitance must be charged to the thresholdvoltage before the device begins to turn on, anddischarged to the plateau voltage before the deviceturns off. Therefore, the impedance of the drivecircuitry and Ciss have a direct effect on the turn on andturn off delays.

Coss – Output CapacitanceThis is the output capacitance measured between thedrain and source terminals with the gate shorted to thesource for AC voltages. Coss is made up of the drain tosource capacitance Cds in parallel with the gate to draincapacitance Cgd, or

oss ds gdC C C= +

For soft switching applications, Coss is importantbecause it can affect the resonance of the circuit.

Crss – Reverse Transfer CapacitanceThis is the reverse transfer capacitance measuredbetween the drain and gate terminals with the sourceconnected to ground. The reverse transfer capacitanceis equal to the gate to drain capacitance.

res gdC C=

The reverse transfer capacitance, often referred to asthe Miller capacitance, is one of the major parametersaffecting voltage rise and fall times during switching.It also effects the turn-off delay time.

Figure 11 Capacitance vs Voltage, APT50M75B2LL

Figure 11 shows an example graph of typicalcapacitance values versus drain-source voltage. The

Application NoteAPT-0403 Rev B

March 2, 2006

8

capacitances decrease over a range of increasing drain-source voltage, especially the output and reversetransfer capacitances.

Qg

Qgs

Qgd

Charge (nc)

Gate-sourcevoltage = peakdrive voltage

VGS(pl)

Figure 12 vGS as a Function of Gate Charge

Qgs, Qgd, and Qg – Gate ChargeGate charge values reflect charge stored on the inter-terminal capacitances described earlier. Gate charge isoften used for designing gate drive circuitry since ittakes into account the changes in capacitance withchanges in voltage during a switching transient [5, 6].

Referring to Figure 12, Qgs is the charge from theorigin to the first inflection in the curve, Qgd is thecharge from the first to second inflection in the curve(also known as the “Miller” charge), and Qg is thecharge from the origin to the point on the curve atwhich vGS equals a specified gate drive voltage.

Gate charge values vary slightly with drain current anddrain-source voltage but not with temperature. Testconditions are specified. A graph of gate charge istypically included in the datasheet showing gate chargecurves for a fixed drain current and different drain-source voltages. The plateau voltage VGS(pl) shown inFigure 12 increases slightly with increasing current(and decreases with decreasing current). The plateauvoltage is also directly proportional to the thresholdvoltage, so variation in threshold voltage correlates tovariation in the plateau voltage.

Resistive Switching TimesFor purely historical reasons, resistive switching data isincluded in MOSFET datasheets.

td(on), – Turn-on Delay TimeTurn-on delay time is the time from when the gate-source voltage rises past 10% of the gate drive voltageto when the drain current rises past 10% of thespecified current.

td(off), – Turn-off Delay TimeTurn-off delay time is the time from when the gate-source voltage drops below 90% of the gate drivevoltage to when the drain current drops below 90% ofthe specified current. This gives an indication of thedelay before current begins to transition in the load.

tr – Rise TimeRise time is the time between the drain current risingfrom 10% to 90%, start to stop of the specified current.

tf – Fall TimeFall time is the time between the drain current fallingfrom 90% to 10%, start to stop of the specified current.

Inductive Switching EnergiesBecause resistive switching data is so difficult to usefor predicting switching losses in a real powerconverter, Advanced Power Technology includesinductive switching energy data in many MOSFET andFREDFET datasheets. This makes it convenient for apower supply designer to directly compare theperformance of a MOSFET or FREDFET with anothertransistor, even one of a different technology such asan IGBT, and to successfully utilize the mostappropriate power transistor.

Figure 13 Inductive Switching Loss Test Circuit

Figure 13 shows the inductive switching test circuitschematic. It is a pulsed test with very low duty cyclesuch that the energy in the inductor collapses longbefore the next set of pulses, and self heating can beignored. The temperature of the device under test

Application NoteAPT-0403 Rev B

March 2, 2006

9

(DUT) and the clamp diode are regulated by atemperature forcing system.

The following test conditions are specified in theDynamic Characteristics table: VDD in Figure 13, testcurrent, gate drive voltage, gate resistance, andjunction temperature. Note that the gate resistance mayinclude the resistance of the gate driver IC. Sinceswitching times and energies vary with temperature,mostly because of the diode in the test circuit, data isprovided both at room temperature and hot, with thediode heated along with the DUT. Graphs are alsoprovided showing the relationships between switchingtimes and energies to drain current and gate resistance.Delay time and current rise and fall time definitions arethe same as for resistive switching.

Figure 14 Turn-on Waveforms and Definitions

Actual switching waveforms are used in the datasheetto define the various measured parameters. Figure 14shows turn-on waveforms and definitions.

Switching energy can be scaled directly for variationbetween application voltage and the datasheetswitching energy test voltage. So if the datasheet testswere done at 330 Volts for example, and theapplication is at 400 Volts, simply multiply thedatasheet switching energy values by the ratio 400/330to scale.

Switching times and energies vary strongly with othercomponents and stray inductances in the circuit. Thediode has a particularly strong effect on turn-on energy.Stray inductance in series with the source that is part ofthe gate drive return significantly affects switchingtimes and energies. Therefore, switching time andenergy values and graphs in the datasheet arerepresentative only and may vary from observed resultsin an actual power supply or motor drive circuit.

Eon – Turn-on Switching Energy with DiodeThis is the clamped inductive turn-on energy thatincludes a commutating diode reverse recovery currentin the DUT turn-on switching loss. Note thatFREDFETs in a hard switched bridge circuit where thebody diode is hard commutated off have about 5 timeshigher turn-on energy than if a discrete fast recoverydiode is used, like in the test circuit of Figure 13.

Turn-on switching energy is the integral of the productof drain current and drain-source voltage over theinterval from when the drain current rises past 5% or10% of the test current to when the voltage falls below5% of the test voltage. The 5% to 10% of current riseto 5% of voltage fall definitions for the integrationinterval in the waveforms in Figure 14 accommodatethe resolution of the instrumentation while providing areliable means of duplicating the measurement thatdoes not compromise accuracy.

Figure 15 Turn-off Waveforms and Definitions

Eoff – Turn-off Switching EnergyThis is the clamped inductive turn-off energy. Figure13 shows the test circuit and Figure 15 shows thewaveforms and definitions. Eoff is the integral of theproduct of drain current and drain-source voltage overthe interval starting from when the gate-source voltagedrops below 90% to when the drain current reacheszero. This is in accordance with JEDEC standard 24-1for measuring turn-off energy.

Thermal, Mechanical CharacteristicsRθJC – Junction to Case ThermalResistanceThis is the thermal resistance from the junction of thedie to the outside of the device case. Heat is the resultof power lost in the device itself, and thermalresistance relates how hot the die gets relative to the

Application NoteAPT-0403 Rev B

March 2, 2006

10

case based on this power loss. Note that tests at APTrevealed the plastic temperature to be the same as themetal portion of the case for discrete parts.

The maximum RθJC value incorporates margin toaccount for normal manufacturing variation. Due tomanufacturing process improvements, the industrytrend is toward decreasing the margin between themaximum RθJC value and the typical value. Theamount of margin is usually not published.

ZθJC – Junction to Case Transient ThermalImpedanceTransient thermal impedance takes into account theheat capacity of the device, so it can be used toestimate instantaneous temperatures resulting frompower loss on a transient basis.

Thermal impedance test equipment applies powerpulses of varying duration to the DUT, waiting for thejunction temperature to settle back down between eachpulse. This measures the ‘single pulse’ transientthermal impedance response. From this, a curve fit isused to create a resistor-capacitor (RC) model. Figure16 shows such a transient thermal impedance RCmodel. Some datasheets may show the capacitors andresistors in parallel, but this is an error. The capacitorsare “grounded” as shown in Figure 16, and thecomponent values remain the same. There is nophysical significance to the intermediate nodes in themodel. Various numbers of resistor-capacitor pairs areused simply to create a good fit to actual measuredthermal impedance data.

Figure 16 Transient Thermal Impedance RC Model

To simulate temperature rise with the RC model, youapply a current source whose magnitude is the powerbeing dissipated in the MOSFET. Thus you can usePSPICE or other electronic simulator to apply anarbitrary power loss input. From this, you can estimatethe junction-to-case temperature rise as the voltageacross the ladder by setting ZEXT to zero (a short), asshown in Figure 16. You can extend the model toinclude a heat sink by adding additional capacitorsand/or resistors.

The transient thermal impedance “family of curves”that is published in the datasheet is simply arectangular pulse simulation based on the RC thermalimpedance model. Figure 17 shows an example. Youcan use the ‘family of curves’ to estimate peaktemperature rise for rectangular power pulses, which iscommon in a power supply. However, because theminimum pulse width is 10 µs, the graph is onlyrelevant for switching frequencies less than 100 kHz.At higher frequency you simply use the thermalresistance RθJC.

Figure 17 Thermal Impedance 'Family of Curves'

Datasheet Extrapolation ExampleSuppose in a switch mode power supply application wewant to hard switch 15 Amps at 200 kHz, 400 Volts,and 35% duty cycle on average. The gate drive voltageis 15 Volts, and the gate drive resistance is 15?duringturn-on and 5?c during turn-off. Also, suppose wewant to let the junction reach 112ºC maximum and canmaintain the case at 75ºC. With a 500 Volt device,there is only a 100 Volt margin between the applicationvoltage and VDSS. Barring surges in the 400V bus, thenarrow voltage margin is sufficient because the largeavalanche capability of a MOSFET provides at “safetynet”. It is a boost configuration with continuousconduction, so the faster body diode recovery of aFREDFET is not needed; a MOSFET will work fine.Which device should we choose?

Since this is a relatively high frequency application thePower MOS 7 series would be the best choice. Let’slook at the APT50M75B2LL. Its current rating is 57Amps, more than three times the switched current,which is a good starting point considering the highswitching frequency and hard switching. We willestimate the conduction loss, switching losses, and thensee if the heat can be dissipated fast enough. The totalpower that can be dissipated is

J C

JC

T T 112 C 75 C 168WCR 0.22

Wθ

− ° − °= =

°.

Application NoteAPT-0403 Rev B

March 2, 2006

11

At 112°C, RDS(on) is roughly 1.8 times higher than atroom temperature (see Figure 3). So conduction loss is

2conductionP 1.8 0.075 15A 0.35 10.6W= ⋅ Ω ⋅ ⋅ =

For turn-on switching loss, we can look at the figure ofswitching loss versus current at 125°C shown in Figure18. Even though our application requires a maximumof 112°C junction temperature, this graph is accurateenough because MOSFET switching energy is notsensitive to temperature except for the temperature-related behavior of the diode in the circuit, and thisshould not have a big change between 112°C and125°C. In any case, our estimates will be conservative.

Figure 18 Inductive Switching Loss, APT50M75B2LL

From Figure 18, at 15 Amps, Eon is about 300 µJ, andEoff is about 100 µJ. These were measured at 330Volts, and our application has a 400 Volt bus. So wecan simply scale the switching energies by voltage:

on400VE 300 J 364 J330V

= ⋅ µ = µ

off400VE 100 J 121 J330V

= ⋅ µ = µ

Data in Figure 18 were also measured with a gateresistance of 5?æ, and we will use 15?æat turn-on. Sowe can use the switching energy versus gate resistancedata shown in Figure 19 to scale again.

Figure 19 Switching Energy vs. Gate Resistance,APT50M75B2LL

Even though the test current in Figure 19 is higher thanour application current, it is reasonable to apply theratio of change in switching energy from Figure 19 toour case. From 5?Uto 15?U, Eon changes by a factor ofabout 1.2 (about 1500µJ / 1250µJ, just looking atFigure 19). Applying this to the voltage-corrected datawe looked up in Figure 18, we get

onE 1.2 364 J 437 J= ⋅ µ = µ

Switching losses are

( )switch switch on offP f E E= ⋅ +

( )200kHz 437 J 121 J 112W= ⋅ µ + µ =

conduction switchP P 123W+ = , which is within what isallowed to keep the junction below 112°C with the casecooled to 75°C. So the APT50M70B2LL will meet therequirements of this example application. The sametechniques can be used to see if a smaller MOSFETwill work.

As a practical matter, the losses are high for a singledevice, dominated by far by the turn-on switching loss.To keep the case at 75°C will probably require aceramic insulator (for electrical isolation) between thecase and high capacity heat sink. An advantage of aMOSFET is that snubbers and/or resonant techniquescan be applied to reduce switching losses withoutworrying about voltage or temperature-dependentswitching effects in the MOSFET.

Application NoteAPT-0403 Rev B

March 2, 2006

12

References[1] R. Severns, E. Oxner; “Parallel Operation of

Power MOSFETs”, technical article TA 84-5,Siliconix Inc.

[2] J. Dodge; “Latest Technology PT IGBTs vs.Power MOSFETs”, application note, AdvancedPower Technology

[3] R. Frey, D. Grafham - APT, T. Mackewicz - TDI-Dynaload; “New 500V Linear MOSFETs for a120 kW Active Load”, application note APT0002,Advanced Power Technology

[4] N. Mohan, T. Undeland, W. Robbins; “PowerElectronics – Converters Applications, andDesign”, text book published by Wiley

[5] K. Dierberger, “Gate Drive Design for Large DieMOSFETs”, application note APT9302, AdvancedPower Technology

[6] R. McArthur, “Making Use of Gate ChargeInformation in MOSFET and IGBT Datasheets”,application note APT0103, Advanced PowerTechnology