What Changed When Chargers Went to 800 V
A 350 kW DC fast charger module is a different magnetics problem from an on-board charger. The DC bus moved to 800 V, the switching frequency rose to shrink the passives, the module is liquid-cooled at 60–70 °C coolant temperature, and the isolation barrier now carries a working voltage that makes partial discharge a first-order requirement rather than a formality. Designs carried over from 400 V platforms fail these modules for predictable reasons.
The Magnetics Inside a Charging Module
| Position | Function | Dominant constraint |
|---|---|---|
| PFC inductor | Boost or totem-pole PFC at 800 V DC output | Core loss at high ripple with large flux swing; powdered or gapped ferrite |
| DC-DC transformer | Isolated LLC or CLLC stage, 800 V to 200–500 V | Isolation and creepage at 800 V working voltage; leakage inductance set for resonance |
| Resonant inductor | Sets the resonant tank with the transformer | Gap stability and loss at high frequency, plus fringing loss near the gap |
| Output inductor | Filtering at the charge port | High DC bias with saturation margin, and thermal path to the cold plate |
| Gate drive transformers / current sense | Isolation for SiC gate drivers | Common-mode transient immunity, creepage at high dv/dt |
Isolation at 800 V Is Not a Hipot Number
At 800 V working voltage the design question is not whether the part passes a routine hipot test, it is whether the insulation has voids that will discharge in service. Practically that means: a specified insulation system qualified for the class and the construction, a winding arrangement that keeps the highest field out of air pockets, vacuum impregnation or potting to eliminate voids, and partial discharge testing at a stated inception and extinction limit. A part that passes 3 kV hipot and shows PD at 900 V will fail in the field on a humid day.
Thermal Design Follows the Coolant, Not the Air
- Conduction, not convection. In a liquid-cooled module the part is potted to a cold plate, so thermal resistance is dominated by the potting layer and the interface, typically 2–5 K/W rather than 10–15 K/W in free air.
- Coolant at 65 °C leaves little margin. With a 65 °C coolant and a Class H system at 180 °C, the permitted rise is around 110 K — generous on paper, but the hotspot inside a potted winding is often 20–30 K above the measured surface, so the effective budget is smaller than it looks.
- Potting compound choice matters. Thermal conductivity, modulus and CTE all affect both the hotspot and the mechanical stress on the winding during thermal cycling.
Design Points Buyers Should Specify
- Working voltage across the barrier, and the standard the module must meet.
- Partial discharge limit and the test method.
- Coolant temperature, flow, and the mounting interface (cold plate flatness and thermal interface material).
- Switching frequency range, because core loss and AC copper loss both scale with it and the supplier needs the worst-case point.
- Leakage inductance target with tolerance, not just a maximum.
- Duty profile: continuous at rated power, or a charging profile with a long plateau.
Related reading
- EV On-Board Charger Magnetics
- Solar PV Boost Inductor Design
- Energy Storage PCS Magnetics
- Hipot and Partial Discharge Testing
We quote DC fast charging magnetics with the PD test plan, the potting and cold-plate interface and the loss budget attached. Magnetic Core team: sales@sirui-tech.com.
