5 Bidirectional DC DC converter topologies for EV and storage systems

Key Takeaways

• The isolation requirement decides the topology family first, because it rules out either the transformer-based or the non-isolated candidates before any other comparison.
• Voltage range and time spent at light load separate the dual active bridge from the CLLC, with wide ranges favouring the first and near-constant ratios favouring the second.
• Each topology carries its own control burden, so testing your controller against switching-accurate models of the candidates belongs ahead of the hardware order.

The bidirectional DC/DC converter topology you pick sets the efficiency, isolation and control effort of every charger or storage system built on it.

A bidirectional DC/DC converter moves energy between two DC voltages in either direction, so one power stage both charges a battery and discharges it. The volume of power electronics that needs this is large. United States grid operators had 43.6 gigawatts of battery storage running by the end of 2025, and every one of those batteries sits behind a converter that passes current both ways.

Five topologies cover most EV chargers, storage racks and DC microgrids. They differ in switch count, voltage range and how hard the control problem gets. You’ll choose faster when you compare them on the same terms.

How isolation splits bidirectional DC/DC converters into two families

The main difference between isolated and non-isolated bidirectional converters is a high-frequency transformer. Isolated designs place one between the two sides, which blocks fault current and sets a large voltage ratio. Non-isolated designs share a common ground, use fewer parts and lose less energy at the same power.

Consider an onboard EV charger that exports power to a house. The pack must stay galvanically separated from the grid side, so the DC/DC stage needs a transformer. A storage rack tied to a DC bus that’s already isolated upstream doesn’t carry that requirement, and a transformer there only adds cost, weight and loss. 

“Settle this question first, because it removes half the candidates before you compare anything else.”

5 bidirectional DC/DC converter topologies compared by where each fits

Three of these power converter topologies are non-isolated and two are isolated. Each entry covers how the circuit moves power, where it fits and what it costs you in control work.

1. Half-bridge buck-boost converter

The half-bridge buck-boost converter is the simplest bidirectional stage, with two switches and one inductor. It bucks from the bus to charge the battery and boosts from the battery to discharge. Each direction uses the same two switches and the same inductor, so one current loop handles both. The battery voltage must always stay below the bus voltage.

Picture a 350 V pack on a 750 V DC bus. The stage runs near 47% duty, the part count is minimal and efficiency is high. It’s the default for storage racks and DC microgrids with a fixed voltage relationship. Trouble starts when the two voltages approach each other, because duty cycle hits its limit and control authority disappears.

2. Four-switch buck-boost converter

The four-switch buck-boost converter places a half-bridge on each end of one inductor, so either side can sit above or below the other. That removes the voltage ordering rule of the simpler stage without adding a transformer.

A pack that swings from 300 V to 500 V against a 400 V bus is the textbook fit, since the pack crosses the bus voltage during every cycle. You pay with four switches and inductor current that always flows through two devices in series, so conduction loss runs higher than the two-switch stage. The handover between buck and boost operation needs careful modulation, or the current will show a visible disturbance each time the voltages cross.

3. Interleaved buck-boost converter

The interleaved buck-boost converter runs several half-bridge phases in parallel, each shifted in time, so the phases share the current. Ripple at the battery terminals partly cancels, which shrinks the inductors and capacitors.

A three-phase version carrying 300 A puts about 100 A through each phase, with the phases spaced 120 degrees apart. That makes it the practical choice for high current storage racks and fast charger output stages. Shedding phases at light load recovers some of the efficiency the extra switches cost. The added work is current balancing. Each phase needs its own sensing, and a controller that lets one phase drift will overheat that inductor long before the total current looks wrong.

4. Dual active bridge converter

The dual active bridge converter links two full bridges through a high-frequency transformer and steers power with the phase shift between them. It gives galvanic isolation, symmetric operation in both directions and a wide voltage range from one control variable.

That range makes it a common pick for chargers serving both 400 V and 800 V packs. Its weak spot is light load, where the bridges lose soft switching and circulating current climbs. Fixing that takes extra modulation modes, such as dual or triple phase shift, and tight timing between the two bridges. OPAL-RT runs switching-accurate models of this kind of stage in real time, so your controller meets those edge cases before a transformer is wound.

5. CLLC resonant converter

The CLLC resonant converter adds a matched resonant tank on each side of the transformer and regulates with switching frequency. Devices switch softly across the load range, so losses stay very low close to resonance.

One university prototype, a 25 kW silicon carbide design, measured a peak efficiency of 98.8% when discharging. The same study found lower total device loss than a dual active bridge. The limit is the gain range. A battery that swings widely pushes the frequency far from resonance, so designers often vary the DC link voltage to keep the tank at its best point. It suits onboard chargers and storage modules with a narrow battery window.

How to choose a bidirectional topology before you buy hardware

Choose the topology from your voltage range, your isolation requirement and your load profile, in that order. Isolation narrows the field to two or three candidates. Voltage range and the share of time spent at light load then separate the dual active bridge from the CLLC. Five questions settle most projects.

• Does a safety standard require galvanic isolation?
• Does the battery voltage ever cross the bus voltage?
• How much current must a single stage carry?
• How many hours will the converter spend below 20% load?
• Can your control hardware hold the timing each option needs?

An EV charger for mixed 400 V and 800 V fleets points to the dual active bridge. A battery energy storage product with a narrow pack window points to the CLLC. The last question is the one teams skip, and it’s where schedules slip.

“Paper comparisons can’t show how your own control code behaves on each circuit.”

Paper comparisons can’t show how your own control code behaves on each circuit. OPAL-RT runs the candidate topologies as real-time models against the same controller, so you compare measured behaviour and commit to hardware once.