How is a phase-shifted full-bridge converter controlled?

Instead of varying the pulse width, both legs of the full bridge switch at a fixed 50% duty and the output is controlled by the phase shift φ between the two legs. The overlap of the two legs sets the effective duty; the output is Vo = 2·n·Vs·D_eff, where D_eff = φ/360 (φ from 0 to 180 degrees). Fixed-frequency phase-shift control enables zero-voltage switching.

What is zero-voltage switching (ZVS) in a PSFB converter?

ZVS means each MOSFET turns on only after its drain-source voltage has already been driven to zero by resonance between the transformer leakage inductance and the MOSFET output capacitance, so there is no capacitive turn-on loss. The energy stored in the leakage inductance (½·L·I²) must exceed the energy needed to charge the switch-node capacitance (≈ Coss·Vs²). This lets the PSFB run at high frequency with high efficiency.

What is duty loss in a phase-shifted full-bridge?

During each commutation the primary current has to slew from one polarity to the other through the leakage inductance before the output diode can take over. During that time no voltage is applied to the output, so the effective duty seen by the output is less than the phase-shift duty — this is duty loss, roughly ΔD ≈ 4·Llk·fsw·n·Io/Vs. It is the price paid for the leakage inductance that enables ZVS.

Phase-Shifted Full-Bridge (PSFB) ZVS Converter Simulator

Q: Why does the lagging leg lose ZVS at light load?

The leading leg transitions with the full reflected load current plus the magnetizing current, so it almost always has enough energy for ZVS. The lagging leg transitions with only the leakage-stored current, which falls with load — so at light load ½·Llk·I² drops below Coss·Vs² and the lagging leg loses ZVS. This simulator shows the ZVS margin for each leg.

Q: Where are phase-shifted full-bridge converters used?

PSFB ZVS converters are the standard high-power isolated DC-DC stage: EV and battery chargers, telecom and server rectifiers, welding supplies and industrial power from roughly 500 W to several kW, where the soft switching gives high efficiency at high frequency and low EMI.

An advanced, physics-accurate simulator of the phase-shifted full-bridge (PSFB) — the high-power soft-switching workhorse. Both legs switch at a fixed 50% duty, and the output is set by the phase shift φ between them; the transformer leakage inductance resonates with the MOSFET output capacitance to give zero-voltage switching (ZVS). Sweep φ, the turns ratio n = Ns/Np, leakage, filter and load, add a real device model, and watch every waveform on a real-time oscilloscope — validated live against Vo = 2·n·Vs·D_eff (D_eff = φ/360), with a leading/lagging-leg ZVS window analyzer, leakage duty loss, CCM/DCM detection, efficiency and a ripple-spectrum FFT.

Phase-shifted full-bridge (PSFB) ZVS isolated DC-DC converter circuit diagram: DC input voltage Vs, four MOSFETs in a leading and a lagging leg driven with a phase shift, resonant leakage inductance, transformer, full-wave secondary rectifier and an output inductor-capacitor LC filter with output voltage Vo = 2·n·Vs·D_eff — Phase-shifted full-bridge — both legs run at 50%, the phase shift φ sets the output, and leakage inductance + MOSFET C_oss give ZVS. V_o = 2·n·V_s·D_eff, D_eff = φ/360.

Parameters

Phase shift φ 120°

D_eff = φ/360. φ = 180° is maximum output.

Input voltage V_s

Turns ratio n = N_s/N_p 0.20

Switching frequency 100 kHz

Resonant & output stage (ZVS)

Leakage inductance L_lk 18.0 µH

Resonates with C_oss for ZVS — but causes duty loss.

MOSFET C_oss 300 pF

Magnetizing inductance L_m 1200 µH

Output inductor L_o 47 µH

Output capacitor C_o 470 µF

Load resistance R 4 Ω

Device & parasitics model

MOSFET R_ds(on) 0 mΩ

Diode V_f 0.0 V

Inductor DCR 0 mΩ

Capacitor ESR 0 mΩ

Ideal (all 0, L_lk=0) gives Vo = 2·n·Vs·D_eff exactly; leakage adds duty loss and parasitics add droop.

Sampling & display

Sampling rate 700 pts/cycle

Points plotted per switching period

Switching periods shown

Run speed (DSO) 1.0×

Show sample points

Presets

Waveforms to display

Waveforms — one steady-state switching period —

rectified node Vo output i_Lo output inductor i_mag magnetizing

LIVE

ZVS window analyzer (PSFB-specific)

Leading leg

—

Lagging leg

—

ZVS is achieved when ½·L·I² at the transition exceeds the energy to charge the switch node (≈ C_oss·V_s²). The leading leg has the magnetizing energy too; the lagging leg relies only on leakage and loses ZVS at light load.

Leakage duty loss ΔD—

Effective duty D_eff = φ/360—

Efficiency & conduction-loss breakdown

Estimated efficiency (conduction losses)—

Conduction losses from the true integrated currents. ZVS removes the capacitive turn-on (switching) loss that a hard-switched bridge would add on top of this.

Ripple spectrum analysis

Harmonics shown 16

Output-voltage ripple / distortion (relative to |DC|)—

FFT of the output voltage — dominant ripple at the 2nd harmonic (twice f_sw). Toggle to the inductor-current spectrum above.

Measurements

Live accuracy check — simulation vs closed-form theory

Output voltage

Inductor ripple

What is a phase-shifted full-bridge converter?

The phase-shifted full-bridge (PSFB) is the standard high-power, soft-switching isolated DC-DC converter. It uses the same four-switch H-bridge as a hard-switched full-bridge, but instead of pulse-width modulation both legs switch at a fixed 50% duty and the output is controlled by the phase shift φ between the two legs. Crucially, the transformer leakage inductance resonates with each MOSFET's output capacitance C_oss during the switching transitions to drive the drain voltage to zero before turn-on — zero-voltage switching (ZVS) — eliminating capacitive turn-on losses and allowing high frequency with high efficiency and low EMI.

How phase-shift control works

Call one leg the leading leg and the other the lagging leg. The overlap between their 50%-duty square waves determines how long the full ±V_s is applied to the transformer. Increasing the phase shift φ increases that overlap and the output. The effective duty is D_eff = φ/360 (φ from 0° to 180°), and:

Vo = 2 · n · Vs · D_eff with D_eff = φ/360 (n = Ns/Np)

ZVS: ½·L·I² ≥ C_oss·Vs² · duty loss ΔD ≈ 4·L_lk·f_sw·n·Io / Vs

This simulator numerically integrates the real PSFB output stage — the phase-controlled ±V_s drive, the leakage-induced duty loss, the magnetizing current and the output LC — to steady state, then compares the measured V_o to the equation above (validated at zero leakage / ideal devices) in the accuracy panel.

Zero-voltage switching (ZVS) & the two legs (topic-specific)

Each leg turns on only after resonance has swung its switch-node voltage to zero. The energy available is the inductive energy at the transition, ½·L·I²; the energy required is what it takes to charge and discharge the two MOSFET output capacitances, ≈ C_oss·V_s².

Leading leg: transitions while carrying the full reflected load current plus the magnetizing current, so it has plenty of energy — ZVS almost always succeeds.
Lagging leg: transitions relying only on the leakage-stored energy, which falls with load. At light load ½·L_lk·I² drops below C_oss·V_s² and the lagging leg loses ZVS — the classic PSFB limitation.

The ZVS window analyzer above shows the energy balance and pass/fail for each leg live — drop the load and watch the lagging leg fall out of ZVS.

Duty loss — the price of the leakage inductance

The same leakage inductance that enables ZVS also causes duty loss: at each commutation the primary current must slew from one polarity to the other through L_lk before the output rectifier can conduct, and during that slew no voltage reaches the output. The effective duty is therefore less than φ/360, by roughly ΔD ≈ 4·L_lk·f_sw·n·Io/Vs. Larger leakage gives easier ZVS but more duty loss — the central PSFB design trade-off, shown live here.

Advanced options in this simulator

Phase-shift control: set the output with φ at fixed frequency, the way a real PSFB is regulated.
ZVS window analyzer: leading/lagging-leg energy balance and pass/fail, with C_oss and L_lk controls.
Leakage duty loss: watch Vo droop as leakage or load rises.
Device model: MOSFET R_ds(on), diode V_f, inductor DCR and capacitor ESR; the accuracy check stays locked to the ideal Vo.
Ripple spectrum & export: FFT at twice f_sw, plus CSV/report/PNG.

The six isolated DC-DC converters

Converter	Switches	Ideal Vo	Switching
Flyback	1	n·Vs·D/(1−D)	Hard
Forward	1	n·Vs·D	Hard
Push-Pull	2 (center-tap)	2·n·Vs·D	Hard
Half-Bridge	2 + split caps	n·Vs·D	Hard
Full-Bridge	4	2·n·Vs·D	Hard
Phase-Shifted FB	4	2·n·Vs·D_eff	ZVS (soft)

Explore the others: Flyback, Forward, Push-Pull, Half-Bridge and Full-Bridge simulators. For the full theory see the Phase-Shifted Full-Bridge tutorial.

Applications

High-power isolated conversion where efficiency matters: EV and battery chargers, telecom and server rectifiers, welding supplies and industrial power from roughly 500 W to several kW.

Frequently asked questions

How is a PSFB converter controlled?

Both legs run at fixed 50% duty; the phase shift φ between them sets the output: Vo = 2·n·Vs·D_eff with D_eff = φ/360. Fixed-frequency phase-shift control enables ZVS.

What is ZVS?

Turning each MOSFET on only after resonance has driven its drain voltage to zero, so there is no capacitive turn-on loss. It needs ½·L·I² ≥ Coss·Vs² at the transition.

Why does the lagging leg lose ZVS at light load?

It relies only on leakage-stored energy, which falls with load. At light load ½·Llk·I² drops below Coss·Vs² and ZVS is lost — the analyzer shows this.

What is duty loss?

Time lost each commutation while the primary current slews through the leakage inductance before the rectifier conducts: ΔD ≈ 4·Llk·fsw·n·Io/Vs. It lowers Vo below 2·n·Vs·D_eff.

Why use a PSFB over a hard-switched full-bridge?

Same power stage, but ZVS eliminates capacitive turn-on losses, so it runs at higher frequency with higher efficiency and lower EMI — ideal above ~500 W.

Power4All · Phase-Shifted Full-Bridge interactive simulator. All waveforms are produced by numerical integration of the actual switching circuit and validated against closed-form theory.