Power Electronics
Dual Active Bridge: Bidirectional Power Flow by Phase Angle
Shift two square waves by just 30 degrees against each other across a small 15-microhenry inductor, and 10 kilowatts of power surges from one 800-volt battery to another — reverse the phase and the same power flows back, all with the semiconductors switching at zero volts and burning almost nothing. That is the Dual Active Bridge (DAB), an isolated DC-DC converter built from two full H-bridges facing each other across a high-frequency transformer, where the phase angle between the bridges — not a duty cycle — is the throttle for both the magnitude and direction of power.
Invented by Rik De Doncker and colleagues in 1991, the DAB has become the workhorse topology for bidirectional, galvanically isolated power conversion: it moves energy either direction with a single control knob, achieves soft switching over most of its range, and routinely hits efficiencies above 98% at power levels from a few hundred watts to hundreds of kilowatts.
- TypeIsolated bidirectional DC-DC converter
- Invented1991, De Doncker, Divan & Kheraluwala (IEEE)
- Control variablePhase-shift angle φ between the two bridges
- Key equationP = n·V1·V2·φ·(π−φ)/(2π²·fs·L)
- Used inEV chargers, solid-state transformers, grid storage, DC microgrids
- Typical range20–500 kHz switching, 0.5 kW to 300+ kW, >98% peak efficiency
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
What It Is and Where It Is Used
The Dual Active Bridge is an isolated, bidirectional DC-DC converter. Two active full-bridges (each four controlled switches — MOSFETs or IGBTs) sit on either side of a medium- or high-frequency transformer. Each bridge chops its DC rail into a square-wave AC voltage; the transformer provides galvanic isolation and voltage scaling by its turns ratio n; a series inductor (often just the transformer's leakage inductance) is the energy-transfer element that ties the two AC ports together.
Because both bridges are active, power can flow either way with no hardware change — you simply reverse the sign of the phase shift. That makes the DAB the default topology wherever energy must move in both directions across an isolation barrier:
- EV fast chargers and onboard chargers — vehicle-to-grid (V2G) capable stages.
- Solid-state transformers (SST) — the isolated DC-DC cell inside medium-voltage AC-DC-AC conversion.
- Battery energy storage and DC microgrids — charging and discharging through one converter.
- Aerospace and traction — high-power-density regenerative links.
How It Works: Phase Shift Drives the Inductor
Both bridges run at the same fixed switching frequency fs, each producing a 50%-duty square wave. Refer the secondary square wave to the primary through the turns ratio; you now have two square-wave voltage sources, v1(t) and n·v2(t), connected only through the series inductance L. The instantaneous voltage across the inductor is their difference, and the inductor current is the time-integral of that difference: i_L = (1/L)·∫(v1 − n·v2) dt.
When the two square waves are in phase, the volt-seconds cancel over a cycle and net power is zero. Introduce a phase lag φ so the primary leads the secondary, and during the overlap intervals a net volt-second is impressed on L. The inductor current develops a trapezoidal ramp; averaged over a cycle, real power flows from the leading bridge (source) to the lagging bridge (sink).
- Primary leads → power flows primary → secondary.
- Secondary leads → power reverses, secondary → primary.
- Magnitude scales with the phase angle up to a peak — the single knob does both jobs.
Key Quantities and a Worked Example
Under single-phase-shift (SPS) control the transferred power is:
P = n·V1·V2·φ·(π − φ) / (2·π²·fs·L)
where V1, V2 = the two DC-bus voltages, n = transformer turns ratio (primary:secondary), φ = phase-shift angle in radians (0 to π/2 for normal operation), fs = switching frequency, and L = total series inductance referred to the primary. Power is maximum at φ = π/2 (90°), giving P_max = n·V1·V2/(8·fs·L).
Worked example: Let V1 = V2 = 400 V, n = 1, fs = 100 kHz, L = 15 µH, φ = 30° = 0.524 rad. Then P = (1·400·400·0.524·(π − 0.524)) / (2·π²·100000·15e-6) = (160000·0.524·2.618)/(29.6) ≈ 7.4 kW. Pushing to φ = 90° yields P_max = 160000/(8·100000·15e-6) = 13.3 kW. Note the trade-off: smaller L or higher V raises power but also raises RMS current and circulating energy.
Design and Operation in Practice
Designing a DAB is largely about sizing L, choosing fs, and picking a modulation scheme:
- Inductor sizing: pick L so rated power lands near φ ≈ 30–60°, leaving headroom below the 90° peak (operating past 90° causes power to decrease with more phase — a control instability).
- Switching frequency: 20–50 kHz for IGBT/SiC high-power stacks; 100–500 kHz for GaN, shrinking the transformer.
- Dead time: a few hundred nanoseconds lets the inductor current commutate the switch-node capacitance for ZVS.
- Advanced modulation: Extended-, Dual-, and Triple-Phase-Shift (EPS/DPS/TPS) add inner phase shifts within each bridge to shrink circulating current and extend ZVS to light load and wide voltage ratios.
The classic weakness of plain SPS is at light load and mismatched voltages (V1 ≠ n·V2), where ZVS is lost and reactive circulating current spikes, hurting efficiency. Closed-loop control usually regulates output voltage or current by trimming φ, often with feedforward of the input/output voltages.
How the DAB Compares to Its Cousins
The DAB's closest relatives are other transformer-isolated converters, and the choice hinges on directionality and load profile:
- LLC resonant: exceptional efficiency at a fixed conversion ratio and unidirectional flow (server/telecom rails), but its resonant tank makes clean bidirectional operation awkward. The DAB trades a sliver of peak efficiency for native two-way flow.
- CLLC resonant: a symmetric, bidirectional resonant cousin favored in onboard EV chargers; it soft-switches beautifully near resonance but needs two tuned tanks and is sensitive to component tolerance.
- Phase-shifted full bridge (PSFB): also uses phase control, but it is single-active-bridge (secondary is a diode/synchronous rectifier), so it is essentially unidirectional and loses ZVS on one leg at light load.
Unlike a non-isolated buck-boost converter, the DAB provides galvanic isolation and can step up or down through n — a safety and grounding requirement in EV and grid hardware.
Failure Modes, Trade-offs, and Significance
The DAB's elegance carries real costs and hazards:
- Circulating (reactive) current: under SPS with voltage mismatch, large currents slosh back and forth without delivering power, heating the switches and copper and cutting efficiency several points.
- Loss of ZVS at light load: the inductor current can drop below the threshold needed to fully charge/discharge the switch-node capacitance, forcing hard switching, EMI spikes, and hotter dies.
- Transformer DC bias / saturation: asymmetries in dead time or gate timing inject a DC flux offset that can walk the core into saturation — a runaway current failure; volt-second balancing or flux-tracking control is mandatory.
- Right-half-plane behavior past φ = 90°: operating beyond peak inverts the control gain and can destabilize the loop.
Despite these, the DAB remains foundational: it made compact, high-efficiency bidirectional isolated conversion practical, and it is the beating heart of solid-state transformers, V2G chargers, and the DC backbones of modern grids and electrified transport.
| Topology | Bidirectional | Soft switching | Passive count | Best fit |
|---|---|---|---|---|
| Dual Active Bridge | Yes (native, by phase) | ZVS over wide load | Low (1 inductor + transformer) | EV/storage, SST, grid |
| LLC resonant | No (unidirectional typ.) | ZVS + ZCS at resonance | High (Lr, Lm, Cr) | Fixed-ratio server/telecom PSUs |
| Phase-shifted full bridge | No (typically) | ZVS on lagging leg only | Output L + C filter | Mid-power welders, telecom |
| Flyback | No | Hard-switched (usually) | Coupled inductor + Cout | Low-power (<150 W) adapters |
| CLLC resonant | Yes | ZVS + ZCS both directions | High (two resonant tanks) | Bidirectional EV onboard chargers |
Frequently asked questions
Why does the phase angle control power instead of duty cycle?
Both bridges run at fixed 50% duty and the same frequency, so their square waves are identical except for a time offset. That offset (phase φ) sets the net volt-seconds impressed on the series inductor each cycle, which determines the average current and therefore the power. Sliding the phase forward or backward changes both how much power flows and which direction it flows — something a duty cycle alone cannot do.
What is the maximum-power phase angle, and why not run there?
Maximum power under single-phase-shift occurs at φ = 90°, where P_max = n·V1·V2/(8·fs·L). Engineers deliberately size the inductor so rated load sits around 30–60°. Past 90° the power actually decreases as phase increases, inverting the control gain and risking loop instability, and the RMS/circulating currents near the peak hurt efficiency and stress the switches.
How does the DAB achieve zero-voltage switching (ZVS)?
Just before a switch turns on, the leakage-inductor current is arranged to flow in the direction that discharges that switch's output capacitance, pulling its drain-source voltage to zero. If the current magnitude and the dead-time window are sufficient, the device turns on at 0 V, eliminating turn-on switching loss. ZVS is naturally available over much of the load range but is lost at light load or when V1 and n·V2 are badly mismatched.
What sets the choice of the series inductance L?
L is the power-transfer element: for fixed voltages and frequency, smaller L means more power per degree of phase (P ∝ 1/L). You size L so full-rated power occurs at a comfortable phase (well below 90°) while keeping RMS current, circulating energy, and the ZVS margin acceptable. Often the transformer's own leakage inductance is engineered to be exactly this value, saving a discrete component.
What are single-, dual-, and triple-phase-shift modulation?
Single-phase-shift (SPS) uses only the phase between the two bridges — simple but with high circulating current when voltages differ. Dual-phase-shift (DPS) adds an equal inner phase shift within each bridge; extended (EPS) adds it to one bridge; triple-phase-shift (TPS) allows three independent angles. The extra degrees of freedom minimize RMS current and extend ZVS across wider load and voltage-ratio ranges at the cost of control complexity.
Why is the DAB preferred over an LLC resonant converter for EV and grid systems?
LLC converters are extremely efficient but are naturally unidirectional and best at a fixed conversion ratio. EV vehicle-to-grid, battery storage, and DC microgrids require power to flow both directions and to regulate across a wide voltage range. The DAB delivers native bidirectional flow with a single control variable and isolation, which the LLC cannot do cleanly; the symmetric CLLC is the resonant alternative when bidirectional soft switching is paramount.