Zero Voltage Switching Resonant Converter Schematic and Design Guide

zero voltage switching circuit diagram

Implement a phase-shifted full-bridge converter with snubber capacitors across each power transistor. Select values between 1–10 nF depending on the device’s output capacitance and switching edge slew rate. For a 100 kHz half-bridge using 650 V GaN FETs, 4.7 nF X7R ceramic capacitors typically eliminate turn-on losses while keeping dead-time under 100 ns. Place capacitors as close as possible to the die–ideally within 2 mm–to prevent parasitic inductance from reigniting the switch.

Drive the gate with a resonant gate driver IC such as the TI UCC27714 or ST STDRIVEG600. Configure the driver dead-time through adjustable external resistors; 10 kΩ sets ≈80 ns for 12 V gate rails. Ensure the driver output impedance is ≤1 Ω to maintain sub-50 ns rise/fall times. Bypass the driver’s VCC pin with 1 µF X5R and 100 nF ceramic capacitors; noisy rails can cause false turn-on even when the main FET is off.

Add a small magnetizing inductance on the primary transformer to create a zero-crossing condition before the next gate pulse arrives. For a 36 V–75 V push-pull stage, a 3 µH inductor allows the drain-source voltage to ring below 1 V, enabling lossless turn-on. Size the core–typically EE16 or RM8–to prevent saturation at maximum load; verify with a 10 Ω series resistor to damp excessive ringing.

Terminate the secondary side with a Schottky rectifier pair (e.g., Vishay VS-20CWQ10FN) and a 10 nF snubber across each diode. This shifts the reverse-recovery current away from the switch transition. At 6 A output, expect 1 W for ultrafast recovery types. Mount diodes on the same heatsink as the secondary windings to improve thermal coupling.

Measure efficiency across load with a precision power analyzer. At 300 W, typical soft-transition designs achieve 96 %–98 % efficiency; any dip below 95 % usually indicates excessive drain-source overlap or improper gate timing. Log drain waveforms with a 500 MHz differential probe; a clean waveform should show

Soft-Commutation Power Stage Layout Guide

Begin by selecting a resonant tank configuration that matches your load requirements–series or parallel topology dictates component scaling. For a 100 kHz operation targeting 500 W output, use a 10 nF polypropylene capacitor in series with a 22 μH ferrite-core inductor, ensuring ESR below 0.5 Ω to minimize conduction losses. Bypass the main switching elements with ultra-fast recovery diodes (trr < 35 ns) to prevent reverse recovery spikes during commutation intervals. Position the snubber capacitor directly across the MOSFET drain-source terminals, maintaining a trace inductance under 2 nH to preserve turn-off transient integrity.

Gate drivers must deliver &geq;2 A peak current with <20 ns propagation delay; isolation via galvanically separated transformers or optocouplers (e.g., ISO77xx) prevents ground bounce. VGS thresholds vary–set 10–12 V for standard MOSFETs, reducing to 5–6 V for GaN devices to avoid false triggers. Table 1 outlines optimal driver ICs for different power ranges.

Power Range Recommended IC Key Spec Layout Tip
<100 W UCC21710 4 A sink/source Thermal pad vias under IC
100–500 W LM5114 3 A sink, 1.2 A source Star ground at driver GND pin
>500 W ISOP4220 ±10 A, 5 kV isolation Differential pairs for gate signals

Trace routing limits loop area: keep high di/dt paths (<2 cm) between the resonant capacitor and semiconductor die, using 2 oz copper weight or thicker. vias should be &geq;0.5 mm diameter, spaced no farther than 2 mm apart to reduce inductance. Thermal vias under FET pads must connect to an internal or bottom-layer heatsink plane; use solder mask-defined pads to prevent short circuits. Avoid right-angle bends in high-frequency traces–mitre corners at 45° or use round arcs to reduce impedance discontinuities.

Compile an EMI filter before the DC bus, using a common-mode choke (e.g., WE-CMB series) followed by differential-mode capacitors (X7R, 22–47 nF). Mount filter components within 1 cm of the power input connector to suppress conducted noise before it propagates. Probe waveforms with a 500 MHz bandwidth oscilloscope using a 10x passive probe with <2 pF tip capacitance; connect ground lead as close as possible to the measurement point to avoid ringing artifacts. Validate soft-commutation by ensuring drain-source voltage rings down to <5% of VDS(max) before the FET turns on.

Firmware adjustments compensate for component tolerances: sample VDS at turn-off using an ADC with <1 μs latency to dynamically adjust dead-time between 200–500 ns. Calibrate against temperature using look-up tables–MOSFET body-diode recovery characteristics degrade above 100°C, requiring 30% longer dead-time. Store compensation values in EEPROM after initial calibration to maintain consistency across power cycles.

Critical Elements for Resonant Power Stage Construction

Select inductors with a saturation current rating at least 30% above the maximum expected operating current to prevent core degradation under high-frequency operation. Ferrite materials like N87 or 3F3 offer optimal performance between 50-500 kHz, while powdered iron cores (e.g., -52 mix) suit applications requiring distributed air gaps. The inductance value should create a resonant period 2-5 times longer than the transition time of the semiconductor devices to ensure complete charge/discharge cycles.

MOSFETs with ultra-low output capacitance (Coss) below 200 pF enable faster transitions and reduced switching losses. Devices featuring trench technology (e.g., Infineon CoolMOS CFD7) provide superior body diode recovery characteristics. For frequencies above 200 kHz, consider GaN HEMTs, where the absence of reverse recovery losses allows tighter dead-time margins. Always match the device’s avalanche energy rating to the expected load conditions.

  • Snubber capacitors: Use high-pulse current polypropylene types (e.g., WIMA MKP) rated at 1.5x the bus potential, positioned within 5 mm of the switching nodes to minimize parasitic inductance.
  • Resonant capacitors: Select mica or NP0 ceramics with a dissipation factor below 0.1% at operating frequency, ensuring dielectric stability under temperature swings.
  • Gate drivers: Opt for isolated drivers (e.g., Si827x) with a propagation delay under 30 ns and common-mode transient immunity exceeding 50 kV/µs.

Thermal and Layout Considerations

Thermal vias with 0.3 mm diameter spaced 1.5 mm apart beneath semiconductor pads reduce junction temperatures by up to 25%. Copper pours on both PCB sides should connect via multiple such vias, with the top layer width 3x the pad area for optimal heat spreading. For forced-air cooling, a heatsink with a thermal resistance below 0.5°C/W per 10W dissipation prevents thermal runaway during transient overloads.

Ground return paths must form a continuous star topology to prevent circulating currents from coupling into control signals. High-current traces should use 2 oz copper weight with 3 mm width per ampere, routed on outer layers to leverage enhanced heat dissipation. The resonant tank components should be placed in a tight loop, with the inductor and capacitor leads no longer than 10 mm to limit parasitic series inductance.

  1. Avoid right-angle bends in high-current paths; use 45° miters to reduce impedance discontinuities.
  2. Separate analog control grounds from power grounds by at least 5 mm to prevent noise coupling.
  3. Use fiducial markers at 0.5 mm diameter for automated assembly to ensure consistent component placement.

Step-by-Step Soft-Transition Power Assembly Guide

Begin with a ferrite transformer core rated for high-frequency operation–EFD20 or EE16 types work well for most low-power designs. Wind the primary in bi-filar configuration using 0.5mm enameled copper wire, ensuring even spacing to minimize leakage inductance. For a 12V input, 8–12 turns per winding typically suffice; verify inductance with an LCR meter to confirm it falls within 20–50 µH. Use heat-shrink tubing to insulate the start and finish points of each coil, as exposed connections may arc under rapid transitions.

Attach IRFZ44N MOSFETs (or equivalent fast-recovery types like IRFB4110) to a heatsink with thermal paste–even a small 40×40mm aluminum plate improves reliability. Connect the drains to the transformer’s outer taps and the sources to the resonant capacitor bank. Use polypropylene film capacitors (0.33µF–1µF, 250VAC) for resonance; ceramic types introduce parasitic effects at frequencies above 100 kHz. Solder leads directly to the traces with 60/40 rosin-core solder–avoid acid flux, which corrodes over time.

Wire the gate resistors (10–47Ω, ¼W) in series with each MOSFET gate, linking them to an MCT2E optocoupler for isolation. The optocoupler’s LED side connects to a 555 timer or PWM controller set to 30–50 kHz; use a 10kΩ potentiometer to fine-tune frequency. Ground the transformer’s center tap to the input negative rail through a 1N4007 diode–this clamps back-EMF spikes. Add a snubber network (10Ω + 0.1µF film capacitor) across each MOSFET drain-source to dampen ringing.

Test with a current-limited bench supply (0.5A max) before applying full load. Monitor waveforms with a 10x probe on an oscilloscope–ideal signals show smooth half-sine transitions at the MOSFET gates and trapezoidal peaks at the transformer output. If ringing exceeds 20% of the peak amplitude, adjust the snubber values or reduce gate resistor resistance by 5Ω increments. For high-power builds, replace the IRFZ44Ns with IXFH10N100 and upscale the capacitor bank to 2.2µF per leg.