Comparing Synchronous to Non-Synchronous Converters Comparing Size, Cost, Efficiency, & More Anston Lobo Texas Instruments 1

Synchronous and non-synchronous buck 2 Synchronous Buck Non-Synchronous Buck Sync-FPWM Sync-DCM (Diode Emulation) Heavy Load Non-sync Light Load

Comparison – Solution size & cost 3 Non-SyncSync Solution Size Larger IC + Power Diode Smaller Both FETs Integrated Layout Cost IC costs less + Diode cost IC costs more Other Considerations  Integrated sync rectifier  More Sophisticated controller & driver  Needs a thermally optimized package Synchronous Buck Non-Synchronous Buck

Comparison – Ease of use 4 SyncNon-Sync DesignEasierNeed to select Schottky diode External Diode No need Selection Considerations  Reverse voltage rating V R >V IN max  Forward current rating I FWD >Load max + Iripple/2  Forward voltage drop V FWD, over current and temp Conduction loss = V FWD * I FWD  Reverse leakage current over temp Loss = V IN * Ileak  Parasitic capacitance Switching loss  Package for heat dissipation  Size increase with V R and I F Synchronous Buck Non-Synchronous Buck

Power diode example – DIODES B340, 40V 3A 5 V FWD I REVERSE Package 5.59mm×6.6mm

6 Critical path Switching Current on the input side Switch Node Synchronous Buck Non-Synchronous Buck For Any Buck Converter Reduce Critical Path Area  Reduce EMI Key of PCB Layout Comparison – Layout to optimize EMI

7 Non-SyncSync Layout for EMI Harder to optimize EMI by layout (Depends on pin out of IC) Easier (Depends on pin out of IC) Synchronous Buck Non-Synchronous Buck

8 VIN VOUT SW 14.5V max VOUT 47mV pp 41dBµV/m 8 EMI mitigation by PCB layout Critical Path Area Reduction

9 SW 18.1V max VOUT 75mV pp 44dBµV/m EMI mitigation by PCB layout Critical Path Area Reduction Now shown with single C IN with 2.5 times larger area Comparison Results SW max (V) Vout p2p (mV) EMI peak (dBµV/m) Smaller Area Larger Area

10 Non-SyncSync - FPWM Light Load Current Waveform Negative Current?NoYesNo HS Conduction Losses Lower I RMS HS: (I RMS-HS 2 )*R ON-HS *D Higher I RMS HS: (I RMS-HS 2 )*R ON-HS *D Lower I RMS HS: (I RMS-HS 2 )*R ON-HS *D LS Conduction LossDiode: I Diode *V F *t Diode *f S LS: (I RMS-LS 2 )*R ON-LS *(1-D)LS: (I RMS-LS 2 )*R ON-LS *t LS *f S Switching FrequencyReduced at very light loadConstant over loadReduced at very light load Switching LossesLessMoreLess Sync - DCM Comparison – Light load efficiency

11 Non-SyncSync – FPWM and DCM Heavy Load Current Waveform Low Side Conduction Losses Diode: I Diode *V F *(1-D)LS: (I RMS-LS 2 )*R ON-LS *(1-D) HS FET Switching Losses Less ½*(V IN *I OUT )*(t rise +t fall )*fs More ½*(V IN *I OUT )*(t rise +t fall )*fs + Qrr*fs*V IN Qrr: LS Body diode Reverse recovery charge Parasitic Body Diode Comparison – Heavy load efficiency

Comparison – Efficiency 12 Synchronous Buck Parasitic Body Diode Conduct during dead-time Reverse Recover Charge increases switching loss Increase ringing in SW node To have the best of both worlds Parallel a SMALL Schottky diode Bypass body diode during dead-time No reverse recover charge Current rating can be a fraction of the power diode for a non-sync buck – only conducts during deadtime Parasitic Body Diode

Efficiency improvement with small Schottky 13 Improvement depends on switching loss from Qrr

Comparison – Efficiency summary 14 Non-SyncSync-FPWMSync-DCMSync + Small Schottky Light Load Efficiency BetterWorseBetter Heavy Load Conduction Losses MoreLess Heavy Load Switching Losses LessMore Less Fixed Switching Frequency NoYesNo Lower V OUT Worse (V F )Better

Comparison – Thermal performance 15 Non-SyncSyncSync + Small Schottky ThermalBetter Two packages LMR14030 Harder One package LM43603 Loss and heat reduced by Schottky diode LM43603 Example V IN = 24V V OUT = 5V Load = 3A 500kHz Synchronous Buck Non-Synchronous BuckSynchronous + small Schottky IC Diode

SyncNon-SyncSync + Small Schottky V OUT SpikeMore - Body diode reverse recovery Less - Schottky diodeLess – small Schottky diode SW Ringing Peak V IN = 24V V OUT = 5V Load = 3A 500kHz V OUT Spikes V IN = 24V V OUT = 5V Load = 3A 500kHz 8 V 4.5 V 2 V Comparison – Noise

Comparison – Unloading transient Sync-FPWMSync-DCMNon-Sync Unloading Transient Overshoot Good Allow neg current to discharge C OUT Worse Depend on load to discharge C OUT Worse Depend on load to discharge C OUT Example V IN = 8V V OUT = 5V Load = 3A F S =2.2MHz LM53603 set at FPWM and DCM modes Same as Sync-DCM 17

Comparison – Controllability Non-SyncSync Controllability Limited Only controls one FET Better Controls both FETs Info from both FETs Current LimitPeak current only Peak current Valley current Average current (depends on controller) OVP Cannot actively pull down V OUT Can actively pull down V OUT (depends on controller) Control architectureMore options 18

Comparison – Current limit SyncNon-Sync DC current limiting Control of Peak current and Valley current  more accurate DC current limiting Only has control of Peak current DC current limit depends on ripple Fast slew rate Slow slew rate 5A 4A 3A 5A 4A 3A 5A

Comparison – Summary 20 Non-SynchronousSynchronous Size / Ease of useLarger sizeSmaller size, Easier to Use Light Load EfficiencyBetterWorse with FPWM, better with DCM Heavy Load EfficiencyMore conduction loss Less switching loss Less conduction loss More switching loss due to reverse recovery Adding small Schottky gives highest efficiency Lower V OUT Lower Efficiency (V F /V OUT )Higher Efficiency ThermalBetter with two packagesHarder with one package Adding small Schottky reduces power loss a lot Fixed F SW NoYes with FPWM, No with DCM NoiselessMore, due to body diode reverse recover charge Adding small Schottky diode reduces noise a lot Unloading TransientWorseGood with FPWM, worse with DCM ControllabilityNo LS controlMore control flexibility

For more information, visit TI Information – Selective Disclosure