All Categories
banner

Industry News

Home >  News  >  Industry News

Comparing Bidirectional DC Fast Charger Options

Nov 18, 2026

The Decision Starts With Isolation, Not With the Bridge

A bidirectional DC fast charger is not a single product category. It is a family of power architectures that happen to share one capability: moving energy in both directions across a high-voltage DC link. The first fork in the road is galvanic isolation. An isolated design places a high-frequency transformer between the grid-side converter and the vehicle battery, which breaks the DC path and keeps fault currents from traveling where they should not. A non-isolated architecture skips the transformer and connects the two sides through a shared DC bus. The non-isolated route is lighter, cheaper, and often more efficient, but it shifts the safety burden onto the vehicle's own isolation and the charger's grounding scheme.

For a public DC fast charger tied to a medium-voltage grid, isolation is usually non-negotiable because grid codes and safety standards expect a physical barrier between the utility interface and the vehicle. For a wallbox mounted inside a garage, a non-isolated design can be defensible if the upstream AC-DC stage already provides isolation. The architecture choice drives everything downstream, including the converter topology, the magnetics, and the thermal envelope.

Isolated Topologies Split Into Two Camps

Once isolation is fixed, the DC-DC stage in a bidirectional charger tends to land in one of two well-traveled camps. The dual active bridge, or DAB, uses two full bridges linked by a transformer and a series inductor, and it transfers power by phase-shifting the bridges against each other. It is intrinsically bidirectional, handles wide voltage ratios reasonably well, and has a mature control literature behind it. The trade-off is that light-load efficiency can sag because the phase shift becomes small and the circulating current does not shrink in proportion.

The CLLC resonant converter takes a different route. It uses a symmetric resonant tank with capacitors on both sides of the transformer, which lets it achieve soft switching in both power directions without an auxiliary circuit. That symmetry is why CLLC has become a reference point for bidirectional on-board and off-board charging: zero-voltage switching on the primary side and zero-current switching on the secondary side reduce turn-on losses and keep efficiency stable across a wider load band. The cost is complexity. The dual resonant networks demand precise parameter tuning, and the design does not forgive a sloppy transformer build. A third option, the DAB operated in CLLLC mode, blends the two ideas and has demonstrated around 94% efficiency with improved zero-voltage-switching behavior in high-power prototypes, but it sits closer to the research edge than the production mainstream.

Isolated DC-DC topology Bidirectional mechanism Strength Honest limitation Typical fit
Dual Active Bridge (DAB) Phase shift between bridges Intrinsically bidirectional, wide voltage range Light-load efficiency can drop High-power chargers with mature control
CLLC resonant Symmetric resonant tank, soft switching both ways Stable efficiency across load band, ZVS/ZCS Precise tuning, complex magnetics Bidirectional OBC and DC wallboxes
DAB in CLLLC mode Hybrid modulation with resonant tank Improved ZVS, lower secondary current stress Less production maturity High-power prototypes and next-gen designs

The table is a starting point, not a ranking. A 30 kW wallbox and a 350 kW corridor charger place different demands on the same topology family, and the right answer depends on the battery voltage range, the grid interface, and how often the unit will actually run in discharge mode.

Wide-Bandgap Devices Change the Calculus but Not the Architecture

Silicon carbide MOSFETs and gallium nitride HEMTs have moved from lab demonstrations into commercial charger designs, and their effect on the topology conversation is easy to misread. These devices do not replace the need for a good topology. They make a given topology behave better at higher switching frequencies and higher junction temperatures. A CLLC stage built around SiC MOSFETs can push switching frequencies into the hundreds of kilohertz while keeping conduction losses in check, which shrinks the transformer and the resonant tank. A 6.6 kW bidirectional on-board charger designed with SiC and GaN devices has been shown to reach peak efficiency above 97% with careful co-design of the gate driver, the EMI filter, and the thermal path. Non-isolated bidirectional converters, by contrast, often land in the 83% to 88% range without wide-bandgap devices, while resonant designs using SiC or GaN can reach 95% to nearly 99% in optimized prototypes.

The practical lesson is that wide-bandgap devices raise the ceiling, but they do not lower the floor. A DAB with poor transformer design and a long gate loop will still underperform, even with expensive switches. The device choice has to follow the topology and the control strategy, not lead them.

Standards Decide Which Options Are Actually Deployable

A bidirectional charger that cannot communicate its intentions to the vehicle and the grid is a science project. ISO 15118-20 is the standard that makes bidirectional power transfer a managed service rather than an uncontrolled backfeed. It extends the communication layer to support charge and discharge sessions, grid-support settings, and the certificate handling that utilities expect. The European Union's Alternative Fuels Infrastructure Regulation ties public charging deployment to ISO 15118 compliance on a fixed timeline, and that regulatory pull is pushing bidirectional capability from pilot projects into procurement requirements. CHAdeMO remains the other major DC protocol with native bidirectional support, and its later revisions extend power levels well beyond the original 62.5 kW envelope. A charger designed around one protocol may not serve a mixed fleet, and a design that supports both CCS and CHAdeMO adds connector and communication complexity that has to be planned at the architecture stage, not bolted on later.

A School Bus Depot Shows Where the Real Constraints Live

The Fremont Unified School District project in California is a useful reality check because it involved actual buses, actual grid interconnection, and actual money. The site paired fourteen existing low-power chargers with six high-power bidirectional DC fast chargers, and the utility upgraded the service with a 750 kVA transformer and 2,500 A switchgear to support the load. The bidirectional chargers let the buses discharge into the grid during peak demand and participate in an emergency load reduction program. Reported charging cost reductions from similar managed deployments range from 30% to 70%, and the value came from demand-charge avoidance and grid compensation rather than from the charger's peak efficiency number.

What the case illustrates is that a bidirectional charger is only as good as its grid interface and its dispatch logic. A DAB or a CLLC stage can be perfectly efficient on a test bench, but if the site transformer cannot absorb the reverse power flow or the communication stack cannot respond to a utility signal, the bidirectional capability sits unused. The charger topology is one layer in a stack that includes the site electrical infrastructure, the protocol stack, and the energy management platform.

Matching the Option to the Duty Cycle, Not the Datasheet

The most common selection mistake is comparing chargers by their peak efficiency or their maximum power rating without mapping those numbers to how the unit will actually run. A fleet depot that charges overnight and discharges for two hours during a peak event has a different duty cycle from a public corridor charger that may never export power at all. For the depot, light-load efficiency and standby losses matter more than a headline number at full load. For the corridor charger, thermal headroom at rated power and grid-code compliance carry more weight. A CLLC-based design with a wide soft-switching range tends to suit the depot's variable load profile, while a DAB with robust phase-shift control and a proven high-power transformer may be the lower-risk choice for a 350 kW corridor unit. Neither is universally better. The comparison only becomes meaningful when the load profile, the site infrastructure, and the protocol requirements are placed alongside the topology table.

Shenzhen Wengao (szwengao) works with charger developers on the manufacturing side of this decision, covering magnetics sourcing, thermal validation, and production consistency for bidirectional power stages. The engineering conversation is most productive when the duty cycle and the grid interface are defined before the topology is locked, because that is the point where a manufacturing partner can flag build constraints that a simulation will not surface.

Recommended Products

Contact Us x

Email Address *
Phone *
Message