Taufik Tutorial

Practical Design of Buck ConverterDr. Taufik Associate Professor Electrical Engineering Department California Polytechnic State University, USA [email protected] http://www.ee.calpoly.edu/faculty/taufik

Tutorial Outline

Practical Design of Buck Converter

Brief Review of DC-DC Converter Design Equations Loss Considerations Layout Considerations Efficiency Improvement Synchronous Buck Resonant Buck PWM Controller MultiphaseTaufik | Page 2

PECON 2008, Johor Bahru, Malaysia

Review: DC-DC Converter Basics

A circuit employing switching network that converts a DC voltage at one level to another DC voltage Two basic topologies:

Non-Isolated Buck, Boost, Buck-Boost, Cuk, SEPIC Isolated Push-pull, Forward, Flyback, Half-Bridge, Full-BridgePractical Design of Buck Converter Taufik | Page 3



When ON: The output voltage is the same as the input voltage and the voltage across the switch is 0. When OFF: The output voltage is zero and there is no current through the switch. Ideally, the Power Loss is zero since output power = input power Periodic opening and closing of the switch results in pulse outputPractical Design of Buck Converter Taufik | Page 4



T

DT

1 1 V 0 = vo ( t ) dt = T 0 T

V dt = V Di i

0

ton Dutycycle = D = ton f s = T

Duty Cycle range: 0 < D < 1 Two ways to vary the average output voltage:

Pulse Width Modulation (PWM), where ton is varied while the overall switching period T is kept constant Pulse Frequency Modulation (PFM), where ton is kept constant while the switching period T is variedPractical Design of Buck Converter Taufik | Page 5





Taufik | Page 6




Taufik | Page 7

Review: DC-DC Converter BasicsLoad/User SideDC-DC Converter

Source Side

Wants:to switch

Wants: DC Voltage and DC Current

DC Voltage and DC CurrentWants:

No AC Component No HarmonicsPractical Design of Buck Converter Taufik | Page 8


Review: DC-DC Converter BasicsLoad/User SideDC-DC Converter

Source Side

Gets:

Wants: Voltage with some ripple

Current with some ripple Needs Filtering



Taufik | Page 9

What is Buck Converter?

A dc-dc converter circuit that steps down a dc voltage at its input Non-isolated hence ideal for board-level circuitry where local conversion is needed

Cell-phones, PDAs, fax machines, copiers, scanners, computers, anywhere when there is the need to convert DC from one level (battery) to other levels

Widely used in low voltage low power applications Synchronous version and resonant derivatives provide improved converters efficiency Multiphase version supports low voltage high current applicationsPractical Design of Buck Converter Taufik | Page 10


The Basic Topology

Controller

Two types of Conduction Modes Continuous Conduction Mode (CCM) where Inductor current remains positive throughout the switching period Discontinuous Conduction Mode (DCM) where Inductor current remains zero for some time in the switching periodPractical Design of Buck Converter Taufik | Page 11


The Basic Topology



Taufik | Page 12

Steady State Analysis of CCM Buck: Transfer Function

Inductor is the main storage element Transfer function may be derived from Volt Second Balance:

Average Voltage across Inductor is Zero in steady state Inductor looks like a short to a DC

VL = vLonton + vLoff toff = 0Practical Design of Buck Converter Taufik | Page 13


CCM Buck: Transfer Function

When the switch is closed or ON Diode is reverse biased since

Cathode (at Positive of Input) more positive than Anode (at 0 volt)

vLon =VS VOton = DTPractical Design of Buck Converter Taufik | Page 14

Voltage across inductor: Recall that: D = ton/T Then, duration of on time, ton:



When the switch is OPEN or OFF Inductor discharges causing its voltage to reverse polarity Diode conducts since Anode (0 volt) is more positive than the Cathode (at some negative voltage) vLoff = VO Voltage across inductor: Recall that: toff = T ton = T DT t = 1 D Toff

(

)



Taufik | Page 15


vLonton + vLoff toff = 0

(VS VO ) DT + ( VO )(1 D ) T = 0VS D VO D VO + VO D = 0

VO = DVS

Average output voltage is LESS than Input VoltagePractical Design of Buck Converter Taufik | Page 16


CCM Buck: Sizing ComponentsL and Ipk

For MOSFETs: Vds and Id

Vrrm, and If

C, V and Irms



Taufik | Page 17

CCM Buck: Inductor Current

When switch is ON, Inductor is charging: di L diL VS VO = = v L = V S VO = L dt L dtiLon iLon

VS VO = ton L VS VO = DT L



Taufik | Page 18

CCM Buck: Inductor CurrentdiL VO = = dt L V iLoff = O toff L VO iLoff = (1 D ) T L

When switch is OFF, Inductor is discharging: diL vL = VO = L dt



Taufik | Page 19

CCM Buck: Inductor CurrentAverage Inductor Current = Average Output Current = Vo/R

iL

We can then determine ILmin and ILmax

I L min iL 2 =

iL V0 1 V0 1 (1 D) =IL = (1 D)T = V0 2 2 Lf R 2 L R 1 (1 D ) V0 1 V0 + (1 D )T = V0 + R 2L R 2 Lf Practical Design of Buck Converter Taufik | Page 20

I L max = I L +


Sizing Inductor: Critical InductanceMinimum Load (output current)

ILmin is used to determine the Critical Inductance (Minimum Inductance value at which the inductor current reaches Boundary Conduction Mode) Any inductance lower than critical inductance will cause the buck to operate in Discontinuous Conduction Mode Requirement is set either by means of maximum iL or by specifying the minimum percentage load where converter still maintains CCM Set ILmin = 0, then solve for L = LC, then choose L > 1.05*LC

I L min = 0 = I L

iL 1 (1 D) = V0 Rmax 2 LC f 2 Practical Design of Buck Converter

LC =

(1 Dmax ) Rmax 2fTaufik | Page 21


Sizing Inductor: Critical Inductance(1 Dmax ) Rmax LC = 2f

Calculated at Minimum Input Voltage

Switching frequency normally chosen by the designer The higher the switching frequency, the smaller the required critical inductance, i.e. beneficial for reducing size of Buck

Calculated at Minimum Output Current = Rmax = Vo/Iomin Iomin is either given as percentage of load to maintain CCM, e.g. 10% load with CCM Or, Iomin is calculated as specified by maximum iL, such that Iomin = iL/2



Taufik | Page 22

Sizing Inductor: Peak CurrentMaximum Load (output current)

ILmax is used to determine peak current rating of Inductor Worst case maximum inductor current occurs at maximum load Maximum output power rating per specified required output voltage

I L max

iL 1 (1 Dmin ) =IL+ = V0 + Rmin 2 2 Lf

Calculated from Highest Input Voltage Chosen inductance value as discussed previouslyPractical Design of Buck Converter Taufik | Page 23


Sizing Switch: Voltage Rating

With ideal diode, the Vswitch-max = Vinmax For non-ideal diode, Vswitch-max = Vinmax + VF where VF is the maximum forward drop across the diode (calculated at maximum load current) Use safety factor of at least 20% For MOSFET, the rating would be VDSmaxPractical Design of Buck Converter Taufik | Page 24


Sizing Switch: Current Rating

Switch current rating is calculated based on average value Draw switch current waveform and then compute the average value

By KCL, Inductor Current = Switch Current + Diode Current During tON, Inductor current equals switch current During tOFF, Inductor current equals diode currentPractical Design of Buck Converter Taufik | Page 25


Switch Current Waveform

iL

ON T 2T

OFF

ON

OFF

ON

OFF 3T

t

iSwitch t

iDiode tPractical Design of Buck Converter Taufik | Page 26


Switch Current Waveform for Current RatingiLmax iLmin Average Value ON T 2T ON t ON

iSwitch

0

I Switch 2 T

( iL min + iL max ) ton = iL ] + iL max ) DT 2 T D = I L D = Io D

I Switch

([i =

L max

( 2iL max iL ) D =2

I Switch

iL = iL max 2

I Switch max > I o max DmaxPractical Design of Buck Converter Taufik | Page 27


MOSFET Rating Example



Taufik | Page 28

Sizing Diode (Schottky): Voltage Rating

Known as PIV (Peak Inverse Voltage) or VRRM is the maximum voltage across the diode With ideal switch, the VRRM = Vinmax For non-ideal diode, VRRM = Vinmax + VSW where VSW is the maximum forward drop across the switch (calculated at maximum load current) Allow at least > 20% safety factorPractical Design of Buck Converter Taufik | Page 29


Sizing Diode (Schottky): Current Rating

Same approach as that for the switch current

iDiode OFF T 2T OFF OFF 3T

Average Value

t

IF =

( iL min + iL max ) toff( 2iL max iL ) (1 D ) =2 2 T

IF =

2 T ([iL max iL ] + iL max ) (1 D ) T

I F = I L (1 D ) = I o (1 D )

I F > I o max (1 Dmin )Practical Design of Buck Converter Taufik | Page 30


Schottky Diode Rating Example



Taufik | Page 31

Sizing Output Capacitor: Voltage Rating

Capacitor Voltage should withstand the maximum output voltage

Ideally: Vcmax = Vo + Vo/2 More realistic: Capacitor has ESR (Equivalent Series Resistance) which worsens Vo Output voltage ripple contributed by ESR is (ESR * IL) Suppressing ripple contribution from ESR

Reduce ESR (Paralleling Caps, Low ESR Caps) Reduce IL by increasing L or increasing switching frequencyPractical Design of Buck Converter Taufik | Page 32


Sizing Output Capacitor: Minimum Capacitance

The AC component (ripple) of inductor current flows through the capacitor, leaving the average flowing through the load Capacitor current waveform will look like:

iC OFF T ON OFF 2TPractical Design of Buck Converter

ON

ON

OFF t 3TTaufik | Page 33


Sizing Output Capacitor: Minimum Capacitance+Q -Q T/2 T t

iC

VO (1 D ) T (1 D )VO 1 T iL iL q = Area = = L = = 2 2 2 8 f 8f 8 Lf 2

(1 D )VO = (1 D ) q q = C Vo C = = Vo 8 Lf 2 Vo 8 Lf 2 ( Vo VO )(1 Dmin ) C= 8 Lf 2 ( Vo VO )Practical Design of Buck Converter

Percent VoppTaufik | Page 34


Sizing Output Capacitor: RMS Current RatingiLmax Io = iL/2

iC OFF T 2T ON OFF ON OFF t 3T

ON

iCrms =

iCpk

iL 2 (1 D ) VO = = 3 3 2 3Lf

iCrms

(1 Dmin )VO =2 3LfPractical Design of Buck Converter Taufik | Page 35


Sizing Input Capacitor: Voltage Rating

Capacitor Voltage should withstand the maximum input voltage

Ideally: Vcmax = Vinmax More realistic: Capacitor has ESR (Equivalent Series Resistance) contributes to capacitor loss Minimizing loss contribution from ESR

Reduce ESR (Paralleling Caps, Low ESR Caps)Practical Design of Buck Converter Taufik | Page 36


Sizing Input Capacitor: Minimum CapacitanceiLmax D*IO

iC OFF ON 2T 3T T ON OFF OFF D*IO t

ON

q = Area

= toff D I O = (1 D ) T D I O

(1 D ) D IO =f

(1 D ) D IOf Vin

q = q = C Vin C = Vinf Vin

(1 D ) D I O =f Vin

(1 D ) D IO max C=Practical Design of Buck Converter Taufik | Page 37


Sizing Input Capacitor: RMS Current RatingiLmax D*IO

iC OFF ON 2T 3T T ON OFF OFF D*IO t

ON

I Crms =2

( I Switchrms ) ( I switchavg )2

2

I Crms

iL I o D 1 + ( D I o )2 = 2 Io Practical Design of Buck Converter Taufik | Page 38


To SummarizeI Switch max > I o max DmaxVswitch-max = VinmaxLC = (1 Dmax ) Rmax 2f

1 (1 Dmin ) I L max = V0 + Rmin 2 Lf

C=

I F > I o max (1 Dmin )VRRM = VinmaxPractical Design of Buck Converter

(1 Dmin ) 8 Lf 2 ( Vo VO )Vcmax = Vo + Vo/2

iCrms =

(1 Dmin )VO

2 3LTaufik | Page 39


Simple Buck Design: 12V to 2.5V 1AVs := 12V Ioccm := 0.1A %Vo := 1% f := 50kHz Vo := 2.5V Iomax := 1A

Given:

Solution: D := Inductor: Lcrit := 2 f Ioccm6

Vo Vs

D = 0.208

( 1 D) H Vo

Lcrit = 1.979 10

4

H

Choose: ILmax := Iomax + IL := ( 1 D) Vo L f

L := 200 10

( 1 D) Vo 2 L f

ILmax = 1.099 A IL = 0.198 APractical Design of Buck Converter Taufik | Page 40


Simple Buck Design: 12V to 2.5V 1AMOSFET: Vds := Vs Id := DIomax Diode: Vrrm := Vs If := ( 1 D) Iomax Capacitor: Vcmax := Vo + Vcmax = 2.513 V C = 1.979 106 5

Vds = 12 V Id = 0.208 A

Vrrm = 12 V If = 0.792 A

%VoVo 2 ( 1 D) 1 C := 2 %Vo 8 Lf F ( 1 D)2

F

Choose

Co := 50 10 %Vo :=

%Vo = 0.396 % 8 Lf CoPractical Design of Buck Converter Taufik | Page 41


Simple Buck Design: 12V to 2.5V 1A1+

L1 2 200u

Sbreak C1 50u 2.5

S1 D1 Dbreak R1

V1 12

+ -

0V2 V1 = 0 V2 = 1 TD = 0 TR = 10n TF = 10n PW = {(3.185/12)*(1/50k)} PER = {1/50k}



Taufik | Page 42

Simple Buck Design: 12V to 2.5V 1A

1.25A

Inductor Current

1.00A

SEL>> 0.59A

I(L1)

2.0A Input Current

1.0A

0A 3.7000ms 3.7200ms 3.7400ms Time 3.7600ms 3.7800ms 3.8000ms 3.8182ms

3.6800ms -I(V1)



Taufik | Page 43

Non-ideal Buck: Loss Considerations

When efficiency estimation is required in the design, losses in Buck circuit should be considered Several major losses to consider:


Static loss of MOSFET Switching loss of MOSFET MOSFET Gate Drive Losses Static loss of diode Switching loss of diode Inductors copper loss Capacitors ESR lossTaufik | Page 44


Static Loss of MOSFET

With MOSFET, its on resistance RDSon directly impacts the static loss RDSon depends on applied gate voltage and MOSFETs junction temperature



Taufik | Page 45

Static Loss of MOSFETiLmax iLmin ON T 2T ON ON Average Value t

Recall, switch current:

iSwitch

0

Static loss for MOSFET with RDSon:2 Pstatic = I switch rms RDSon

2

Pstatic

iL = I o D 1 + RDSon 2 Io Practical Design of Buck Converter Taufik | Page 46


Switching Loss of MOSFET

turn on Io Io

The switching loss depends on how the voltage and current overlaps May be approximated with a scenario where voltage and current start moving simultaneously and reach their endpoints The overlap causes power loss (V x I) Will assume to occur both at turn-on and turn-off transitionsturn off VIN

VIN

0 tonPractical Design of Buck Converter

toffTaufik | Page 47


Switching Loss of MOSFETturn on Io Io turn off VIN

VIN

0 ton toff

I oVinton P(ton ) = 6T

P (toff ) =

I oVin toff 6T

PswitchingPswitching

I oVinton I oVintoff = P(ton ) + P(toff ) = + 6T 6TI oVin = ( ton + toff 6TPractical Design of Buck Converter

)Taufik | Page 48


Switching Loss of MOSFET & Gate Drive Loss

When MOSFET is off, its output capacitance Coss is being charged translates to loss

PC os s

1 2 = COSSVin f 2

Gate drive loss comes from the total gate charge Qgate and the gate drive voltage Vgate used

Pgate

1 = QgateVgate f 2



Taufik | Page 49

Static Loss of Diode: Forward Loss

Losses that occur during diodes fully on (forward loss) and fully off (reverse loss) conditions Forward loss come from the product of diodes forward voltage (VF) and forward current (IF), in addition to the rms loss due to diode dynamic resistance, rd

Pforward = V f I f + I rd2 f



Taufik | Page 50

Static Loss of Diode: Forward Loss

From datasheet

Pforward = V f I f + I 2 rd f

I f = (1 D ) I o

If =

(1 D ) I 23Practical Design of Buck Converter

2 + I min + I max I min max


Taufik | Page 51

Static Loss of Diode: Reverse LossPreverse = Vr I r (1 D )

Loss occurs when the diode is in the fully off or nonconducting condition



Taufik | Page 52

Switching Loss of Diode: Turn On Loss

The switching behavior at turn-on is characterized by a low value of peak forward voltage (VFP) and forward recovery time (tfr)

PON = 0.4 (VFP V f ) t fr I f fPractical Design of Buck Converter Taufik | Page 53


Switching Loss of Diode: Turn On Loss

Both VFP and tfr are normally plotted against dId(t)/dt in the datasheet, whereas dId(t)/dt itself is also available in the datasheet for a given set of conditions



Taufik | Page 54

Switching Loss of Diode: Turn Off Loss

Turn-off loss constitutes appreciable switching losses due to the overlapping of diode voltage and current at turn-off with its associated reverse-recovery timeId0 t rra = I rrm

(

dI d dt t1( rr )

)t rrb = 1.11 (t rr t rra )

[t 2rr , t3rr ] = k rr t rrbId0t1( rr )

Poff = 0.5Vds I d 0dI d dt

(

)

2 + 0.033Vd 0 I rrm t rra + Vd 0 I rrm 0.467 0.433k rr + 0.15k rr t rrb

(

)



Taufik | Page 55

Inductors Copper Loss

Inductors winding is made of copper and hence inherently it will have resistive lossAverage Inductor Current, I

With inductors dc resistance of RL and inductors rms current, the copper loss of inductor is:2

PL = I L RL1 iL IL = I 1+ 3 2I Practical Design of Buck Converter

2


Taufik | Page 56

Inductors Core Loss

Factors affecting core loss: switching frequency F, temperature, flux swing B General form:

Core Loss = Core Loss/Unit Volume x Volume

Where,

Core Loss/Unit = k1 x Bk2 x Fk3

Constants k1, k2, and k3 are normally provided by the core manufacturers



Taufik | Page 57

Capacitors ESR Loss

Real world capacitors posses ESR (Equivalent Series Resistance) ESR can measured with, for example, Capacitor Wizard

iC OFF T 2T ON OFF ON OFF 3T t

ON

Loss due to Capacitors ESR is:

PESR = I C ESR

2

iL IC = 2 3Practical Design of Buck Converter Taufik | Page 58


Buck Design With LossesTaufik Pomax := 120W Vonom := 12V Vinom := 24V Fs := 250kHz Iccm := 10% Vopp := 2% m 1 103

Buck Design with LossesMaximum Output Power: 1 10 Nominal Output Voltage: Nominal Input Voltage: Switching Frequency: Minimum Percent CCM: Maximum Ripple Percentage: Design Calculations and Sizing Components: Nominal Duty Cycle: D := Vinom ( 1 D) Critical Inductance: Lc := Lo := 200H Choose L > Lc Peak Inductor Current: = 0.5 Vonom6

Vonom2 Iccm Pomax = 12.000H 2 Fs RLo := 100m 1Vonom2

with assumed DC resistance of: ILopk := Vonom +

( 1 D)

2 Lo Fs Pomax

Pomax Id := D Vonom

= 10.06A

Vswmax:= Vinom = 24V = 5APractical Design of Buck Converter Taufik | Page 59

Switch Voltage: Switch Current:


Choose MOSFET IRF7471 40V 10A Rdson 13mO Diode Vrrm: Vrrm:= Vinom = 24V If := ( 1 D) Vonom = 5A Pomax Diode Forward Current: Choose MBR3040 Capacitor Voltage Rating: Vcap := Vonom + Co := 8 Lo Fs Vopp RMS Current Rating: Icaprms := 2 3 Lo Fs = 0.035A Choose a 25V 50uF capacitor ( 1 D) Vonom2

Vopp Vonom = 12.12V 2 = 250 103

Capacitance: F

( 1 D)

Power Loss Calculations MOSFET Loss Calculations: Rdson := 13m n := 0 .. 11 Load :=n

Output Current Array:

Loadn Pomax 100 Io := n Vonom

Static Loss: Idrms := Io D 1 + Iccmn n

Pmos1 := Io D 1 + Iccm Rdsonn n

(

)2

0.01 5 10 20 30 40 50 60 70 80 90 100



Taufik | Page 60

Switching Loss: ton1 := 12ns Vg := 12V toff1 := 15ns Coss := 700pF Qg := 21nC Io Vinom ( ton1 + toff1) Fsn n

Pmos2 := 6 Coss Vinom Fs = 0.05W 22

Pcoss := 2n n

1 Pgate := 1 Qg Vg Fs = 0.032W

Pmostot := Pmos1 + Pmos2 + Pcoss + Pgaten

Diode Loss Calculationsn n

Ifavg := Io ( 1 D) Rd := 4A 0.5A Vonom ( 1 D) Lo Fs IL := = 0.12A 0.62V 0.4V = 0.063

From Diode Datasheet:

Vf :=n

Dynamic Resistance:

Peak to peak Inductor Current ( 1 D) 3 Pd1 := Vf Ifavg + Ifrmsn n n

Ifrms :=

n

2 2 IL Io IL + Io IL Io + IL Io + + n n 2 2 2 n 2 n

(

)2Rd nIr := 0.00015 A

0.02V 0.46V 0.5V 0.58V 0.61V 0.64V 0.66V 0.68V 0.7V 0.71V 0.73V

From Datasheet:

Vr := Vinom Vonom

Pd2 := Vr Ir ( 1 D) = 0.001W



Taufik | Page 61

Efficiency of 12V 120W0.95 0.905 0.86 0.815 0.77 n 0.725 0.68 0.635 0.59 0.545 0.5 0 10 20 30 40 50 Loadn 60 70 80 90 100 Efficiency

Assume: tfr := 500ns Vfp := 10Vn

Pd3 := 0.4 Vfp Vf tfr Ifavg Fsn n n

n

(

)

Pdtot := Pd1 + Pd2 + Pd3

n

Inductor Loss Calculation 1 PLo := ILrmsn

ILrms := Io 1 +

n

n

3 2 Io n IL

2

(

)2 RLo n

Percent Load

Capacitor Loss Calculation ESR := 150m 2

Assume: Pc := Icrms ESR

Icrms :=

IL

2 3

= 0.035A

Total Loss Calculationn n n

Ptotal := Pmostot + Pdtot + PLo + Pc Po :=n n n

n

Po := Vonom Ion

n

Efficiency ==> Po + Ptotaln



Taufik | Page 62

Another Example



Taufik | Page 63



Taufik | Page 64



Taufik | Page 65



Taufik | Page 66



Taufik | Page 67



Taufik | Page 68

PARAMETERS:

Vg+ -

Vref = 2.5 Vo = 12 ratio = {Vref/Vo} Rtop = 10k mult = 0.992 percentload = 10I

S1 Sbreak 50u C1 50u R1V

+

1 R1a {Rtop} 2

0L1 Vout

V1 D1 Dbreak

48

Vfb

R1b {mult*((ratio*Rtop)/(1-ratio))}

0

{(100/percentload)*(Vo/10)}

4 100 0

{Vref}

COMPARATOR Vg

Vfb

IN+ OUT+ IN- OUTEVALUE LIMIT(1MEG*V(%IN+, %IN-),5,0)

Vtri V1 = 5 V2 = 0 TD = 0 TR = {5u-10n} TF = {5u-10n} 0 PW = 20n PER = {1/100k}

0



Taufik | Page 69

20V

Output Voltage at Maximum Load

(12.008 V)

10V

SEL>> 0V

30A

V(Vout)

Inductor Current at Maximum Load

20A (10.000 Amps)

10A

0A 0s 1.5ms Time 2.0ms 2.5ms 3.0ms 3.5ms 4.0ms 4.5ms 5.0ms

I(L1)

0.5ms

1.0ms

Output Voltage Ripple at Maximum Load 12.0200V

(12.023 V) 12.0000V

11.9800V (11.977 V) SEL>> 11.9695V 11A V(Vout) Inductor Current Ripple at Maximum Load

10A

(9.1136 Amps) (10.883 Amps)

9A 4.68517ms I(L1)

4.69000ms

4.69500ms

4.70000ms Time

4.70500ms

4.71000ms

4.71500ms



Taufik | Page 70



Taufik | Page 71

Efficiency Improvement

Ways to improve converters efficiency:

MOSFET

Low Rdson for High Duty Cycle Low Gate Charge for Low Duty Cycle Paralleling for High Current

Schottky Diode

Low forward drop Short recovery time

Inductor

Multiple parallel winding such as Bifiliar (two windings), Trifiliar (three windings)Practical Design of Buck Converter Taufik | Page 72


Efficiency Improvement

Capacitors

Low ESR Paralleling caps (increasing capacitance while reducing ESRs)

Lower inductor current ripple

Reduce rms loss (inductor and output capacitor) Increase switching frequency or inductance Switching loss and real-estate trade off

Lower gate drive voltage Use of Synchronous MOSFET in place of diode, especially for low voltage and high current outputPractical Design of Buck Converter Taufik | Page 73


Synchronous Rectification

Replaces freewheeling schottky with MOSFET Especially beneficial on low duty cycle and high current applications Due to required dead time and slow MOSFETs body diode, a Schottky is connected across the Synchronous MOSFET MOSFET + Schottky = FETKY combo such as IRF7326D2



Taufik | Page 74

Soft-Switching

Prevents hard-switching or the overlapping of switchs voltage and current during turn-on and turn-off transitions

switching losses which is proportional to switching frequency

Use of resonant circuit to shape switch voltage and/or current waveforms to inherently go to zero at which switching transition is initiated zero switching losspswitching(t) Ion Ion

Voff

Voff

0 turn onPractical Design of Buck Converter

turn offTaufik | Page 75


Soft-Switching

Quasi-resonant buck topologies such as Zero-Voltage and Zero Current Resonant Switch Buck converter Needs constant-on or constant-off controllers such as UC1865 - UC1868, UC1861 UC1864, MC34067 and MC33067, TDA4605-3, TDA4605-2

Voff

Ion

Ion

Voff

0 turn onPractical Design of Buck Converter

turn offTaufik | Page 76


Soft-Switching

Zero-Current Resonant Switch Buck

Turns switch OFF at zero current

Zero-Voltage Resonant Switch Buck

Turns switch ON at zero voltage



Taufik | Page 77

PWM Controller

Current Mode Controller will be used due to many of its advantages

Easy Compensation With voltage-mode, the sharp phase drop after the filter resonant frequency requires a type III compensator to stabilize the system Current-mode control looks like a single-pole system, since the inductor has been controlled by the current loop Improves the phase margin, makes the converter much easier to control A type 2 compensator is adequate, greatly simplifies the design process With voltage-mode control, crossover has to be well above the resonant frequency, or the filter will ring. CCM and DCM Operation It is not possible to design a compensator with voltage-mode that can provide good performance in both CCM and DCM With current-mode, crossing the boundary between the two types of operation is not a problem Having optimal response in both modes is a major advantage, allowing the power stage to operate much more efficiently Line Rejection Closing the current loop gives a lot of attenuation of input noise Even with only a moderate gain in the voltage feedback loop, the attenuation of input ripple is usually adequate with current-mode control With voltage-mode control, far more gain (or feed forward) is needed in the main feedback loop to achieve the same performancePractical Design of Buck Converter Taufik | Page 78


PWM Controller

For the sake of example, well use UC184x or MIC38HC4x family



Taufik | Page 79

PWM Controller



Taufik | Page 80

PWM Controller

Selecting Timing Resistor and Timing Capacitor

Maximum Duty Cycle and Switching Frequency have to be determined first Percent Dead time would then be computed from Dmax Using % Dead time along with Switching Frequency, we can then use plots provided in the data sheet to determine the required timing capacitor and timing resistor Example: Lets say that Dmax was calculated to be 70% or 0.7. Add safety factor to Dmax. Say 10% such that Dmax = 0.8

The dead time is therefore = 100% - 80% = 20% If switching frequency used is 80 kHz, then the value for % dead time along with switching frequency can be used to determine the required Timing Capacitor This is done by using the plot provided in the data sheet.Practical Design of Buck Converter Taufik | Page 81


PWM Controller

From plot, 80 kHz intersects the 20% dead time at approximately Timing Capacitor value of 10 nF. Next, the timing resistor is found from the plot which is also provided in the data sheet



Taufik | Page 82

PWM Controller

Plot shows that 80kHz intersects the timing capacitor plot for 10 nF at timing resistance approximately equals to 2k So, in order to provide the 20% dead time at 80 kHz switching frequency, the timing components are: CT = 10 nF and RT = 2 k

= 2k

= 10nF



Taufik | Page 83

PWM Controller

Feedback Compensation As a start, typically a small capacitor is placed on ZF (such as 2200 pF) for feedback compensation Once a prototype is built, the feedback compensation will be investigated to give the desired gain and phase margin and stability (over wide range of load) Involves decision of whether to use type I, II, or III error amplifier



Taufik | Page 84

PWM Controller



Taufik | Page 85

PWM ControllerFp 0 =

Steps for selecting components in Type 2

Choose cross-over frequency Fcross to be around 1/3 of switching frequency Fswitch The required pole frequency Fp0 that yields the desired crossover frequency of the open loop gain (where H0 is dc gain of the plant)

Calculate capacitor C1 where R1 should have been selected when setting the voltage divider

Calculate R2 using the previously calculated C1 and the output pole of the plant Fp

Fcross Ho 1 C1 = 2 R1 Fp 0 1 R2 = 2 C1 FpC3 = 1 2 ESR Fesr

Calculate capacitor C3 where Fesr is the location of the ESR zero



Taufik | Page 86

PWM Controller

Current Sensing Resistor

Need to calculate power rating of the sensing resistor. This involves calculating worst case Irms through the sensing resistor, and then compute P = (Irms)2*Rsense A low pass RC filter circuit is also needed to eliminate leading spike on the pulse voltage resulted from current being sensed Ensure that voltage out of the filter is less than 1V (for this controller). If not, then reduce the value of Rsense



Taufik | Page 87

Layout Considerations

Keep trace inductance low (preferably by reducing length, not increasing width) for the critical path (switch and diode paths)

Noise spikes may appear in input and output, and to the controller chip Avoid using a current probe (a loop of wire) for diode and switch due to additional inductance it will produce

Provision of good Input decoupling since input capacitor is in the critical path

Besides the usual bulk capacitor, also put a small ceramic capacitor at the supply end to ground, and another one close to the switch to ground

Provision of good decoupling with a small ceramic capacitor between input and ground pins Try using shielded inductor, and position the inductor away from the controller and feedback trace In multi-layer boards, dedicate one layer for ground Keep the feedback trace as short as possible to minimize noise pickup and place it away from noise sourcesPractical Design of Buck Converter Taufik | Page 88


Multiphase

The technique used mainly in very low voltage and high power applications such as processors

Next-generation networking ASICs and processors require multiple lower voltages, higher currents, faster dynamic response, greater efficiency and power management solutions that reside close to the load To meet the need of increasing power density through higher efficiencies and higher operating frequencies

A novel power architecture, multiphasing topologies, is emerging to contend with tomorrow's power requirements High-density applications with lower power levels are usually managed with 2-phase solutions, whereas higher power levels can require up to 4-phase solutions



Taufik | Page 89

Multiphase

Multiphasing address 5 key parameters in power conversion Efficiency: The best power efficiency is achieved by converting a voltage in a single stage, rather than double conversion. For example, assume you want to convert 48V to 1.2V at 100W using a 2-phase forward converter In a 2-phase conversion, current is split equally in the two phases. The FET on-losses are I2 R, which equates to a 50% reduction in on-losses Lower peak currents provide lower turn-on and turn-off losses, resulting in lower switching losses Lower turn-on and switching losses provide overall greater efficiency Input/output ripple reduction: Multiphasing PWM controllers increases switching frequency. The resulting frequency is equivalent to the PWM clock frequency times the number of phases. Higher operating frequency equates to less input/output capacitance and smaller input/output inductorsPractical Design of Buck Converter Taufik | Page 90


Multiphase

Fast dynamic response. Improved dynamic response is the result of smaller output inductors allowing for fast response to current changes combined with higher operating frequency, equal to clock frequency times the number of phases, which allows for higher crossover. Ease of manufacturability: Next-generation designs demand smaller form factors and automated assembly, eschewing hand soldering of large transformers, inductors and capacitors. Better thermal management: Thermal management is critical at these new power densities. The challenge is even higher with the emergence of modules operating in extended temperature range. With multiphasing techniques, you spread the heat evenly over the whole converter, avoiding hot spots and improving converter reliabilityPractical Design of Buck Converter Taufik | Page 91


Multiphase



Taufik | Page 92

Multiphase



Taufik | Page 93

Two-Phase vs. 1 Buck1+

L1 2 1 2 200u 200u

L2

-

+

S1

-

V

S2 Sbreak D2 Dbreak

V

+ -

V1 D1 Dbreak C1 50u 2.5 12 R1

12

Sbreak

V3

+ -

C2 50u

R2 2.5

0V4 V1 = 0 V2 = 1 TD = 0 TR = 10n TF = 10n PW = {Duty *(1/50k)} PER = {1/50k} 1+

0

V2 V1 = 0 V2 = 1 TD = 0 TR = 10n TF = 10n PW = {(3.185/12)*(1/50k)} PER = {1/50k}

L3 200u

2

+ -

S3 Sbreak D3 Dbreak

PARAM ET ERS:V5 V1 = 0 V2 = 1 TD = {0.5/50k} TR = 10n TF = 10n PW = {Duty *(1/50k)} PER = {1/50k} Duty = {3.135/12}



Taufik | Page 94

Two-Phase vs. 1 BuckOutput Voltage Just a Buck 2.5 Volts

4.0V

Two-Phase Buck

3.0V

2.0V

1.0V

0V 1.5ms Time 2.0ms 2.5ms 3.0ms 3.5ms 4.0ms 4.5ms 5.0ms

0s

0.5ms 1.0ms V(R1:2) V(L2:2)

2. 520 0V Ou tpu t V olt age Ri ppl e

2. 510 0V Ju st a B uck

2. 500 0V

2. 490 0V

T wo- Pha se

Buc k

2. 480 3V 3. 620 3ms 3 .64 00m s V(L 2:2 )

3 .66 00m s Ti me

3. 680 0ms

3. 700 0ms

3.7 200 ms

V (R1 :2)



Taufik | Page 95

Two-Phase vs. 1 BuckCapacitor Current Two-phase Buck

200mA

0A

Just a Buck

-200mA 4.6900ms Time 4.7000ms 4.7100ms 4.7200ms 4.7300ms 4.7400ms 4.7500ms

4.6705ms 4.6800ms -I(C2) -I(C1)

1.5A Input Currents Just a Buck

1.0A

Two-Phase

0.5A

0A 4.0200ms 4.0400ms Time 4.0600ms 4.0800ms 4.1000ms

3.9886ms

4.0000ms -I(V1) -I(V3)



Taufik | Page 96

Power Electronics Lab at Cal Poly State University 6 Instructional Lab Benches, 2 Project/Thesis Benches For further information, contact Power Electronics Lab Coordinator, Dr. Taufik at [email protected]



Taufik | Page 97

Taufik Tutorial

Documents