Access Official Technical PDF

Share Summary:LinkedInEmail

A 48V gate drive power supply for an automotive inverter has to do more than provide the voltages needed to switch power devices. It must maintain isolation, limit conducted and coupled electromagnetic interference (EMI), and move heat away from the converter. These requirements become especially important in hybrid and electric vehicle systems, where switching power stages and sensitive control electronics share a demanding electrical environment.

A Fly-Buck supply developed for a 24V input can provide a useful starting point, but adapting it for 48V calls for changes to the converter, input components, transformer, and feedback network. Careful PCB design is just as important: current-return paths, switch-node area, secondary-side copper, and thermal spreading all influence the result.

Adapting a gate drive supply for 48V

A multi-output isolated supply can serve three high-side gate drivers and a shared low-side drive. The high-side supplies are isolated from one another because each driver is referenced to a different switching emitter. The low-side drivers share a reference, so one supply can serve all three. In the IGBT design described here, each output provides a 23V span that a Zener network divides into +15V, 0V, and −8V. The negative turn-off voltage helps guard against unintended turn-on caused by the Miller effect.

Moving from a 24V to a 48V input requires a converter with suitable voltage capability and margin for automotive input conditions, including load dump. The design replaces a 65V converter with a 100V device and uses 100V-rated input capacitors to accommodate the higher voltage and capacitance derating.

The transformer and feedback network also need attention. The 48V design targets a 24V primary output to maintain a suitable Fly-Buck duty cycle, while retaining the required 23V secondary output. That calls for a revised transformer turns ratio and feedback resistor values. These are design-specific choices; engineers should verify component ratings and operating conditions for their own application.

PCB layout strategies for automotive inverter EMI

EMI performance depends not only on filter values but also on how components and current paths are arranged. A typical input filter combines an LC section, a common-mode choke, and a ferrite bead with capacitors. The LC network attenuates noise over its effective frequency range, while the choke targets common-mode current and the ferrite bead helps suppress higher-frequency energy. Parasitic inductance and capacitance limit real-world filter behavior, so the layout must support the circuit rather than undermine it.

Keep filter paths controlled

Placing the filter on the board side opposite noisy switching circuitry can let a grounded plane provide shielding. A carefully managed ground layout helps ensure that return current passes through the filter in the intended sequence. Keep filter connections compact to limit loop area, but consider spacing inductors where necessary to reduce magnetic coupling between them. Route the input supply with a nearby, continuous return path; reducing the loop width helps limit inductive coupling and radiated magnetic fields.

Place input capacitors with their frequency behavior and connection paths in mind. A smaller capacitor can be positioned close to the converter input, with larger capacitance nearby and a low-impedance return to ground. Similarly, keep the switch node compact: its fast voltage transitions can couple capacitively into nearby circuitry. Short switch-node routing, separation from sensitive or filter circuitry, and appropriate grounding around the area can help manage that coupling.

Manage transformer coupling and common-mode current

In an isolated Fly-Buck supply, parasitic capacitance between transformer windings can transfer switching energy to the secondary. The secondary circuitry can then couple current to a chassis or ground plane, creating common-mode noise that bypasses the input filter. Transformer shielding can reduce primary-to-secondary capacitance, but it can also increase leakage inductance and affect converter operation. Evaluate that trade-off with the transformer design and operating requirements.

A Y-capacitor between the low-side secondary and primary ground can reduce the secondary voltage swing caused by capacitive coupling. Its value must remain compatible with the applicable isolation and leakage-current requirements. This approach is not automatically suitable for every isolated output: high-side secondaries referenced to switching nodes may not be appropriate places to add such capacitors.

Secondary-side layout also matters. Minimize trace length and copper surface area where practical, while preserving regulation and electrical requirements. A smaller secondary area reduces capacitive coupling to nearby ground structures. Transformer placement, winding-pin assignment, and any external shielding are further variables to investigate during EMI troubleshooting.

Thermal design: give heat a path

The converter IC’s exposed die-attach pad transfers much of its heat into the PCB. Copper planes and thermal vias can then spread that heat across board layers. A small, isolated ground area may limit heat spreading; a larger plane can improve it, provided clearances maintain the required isolation between primary and secondary circuits.

Component placement can either support or obstruct heat flow. Avoid placing parts directly in the path where heat needs to spread from the IC. Thermal vias beneath the exposed pad and copper on multiple layers provide additional routes through the board. The reference design’s optimized layout measured an IC temperature 12°C lower than the modified original layout, illustrating the impact of board-level thermal changes in that comparison—not a guaranteed outcome for other designs.

Thermal imaging also requires care. Differences in surface emissivity, including from solder, can make areas appear cooler or warmer than they are. Interpret images in the context of the materials and measurement setup.

Adapting outputs for SiC and GaN

Changing the power device can change the gate-drive voltage requirements. In the described design, SiC devices use a +20V, 0V, and −4V output, rather than the IGBT supply’s +15V, 0V, and −8V. Achieving that output requires adjusting the feedback resistor, replacing the 15V Zener with a 20V Zener, and changing the associated resistor values. The transformer can retain the required voltage characteristics while using a lower current rating when the load requirement is lower.

GaN devices require a different voltage arrangement in this design: +6V, 0V, and −2V. The output network and transformer design must therefore be reviewed for the specific device requirements rather than copied directly from an IGBT implementation.

Validate performance in the intended application

Layout refinements improved conducted EMI in the reported comparison, but the first optimized board did not meet the cited CISPR 25 Class 5 limits. This is a useful reminder that good design practices do not replace testing. Transformer shielding, via stitching, Y-capacitor use, and filter adjustments can be assessed systematically to identify the dominant noise paths. Thermal behavior should likewise be checked under relevant operating conditions.

For engineers and system integrators, the practical lesson is to treat the 48V gate drive supply as a coordinated power, layout, isolation, and thermal design problem. Select components for the input and output requirements, control high-frequency current loops, examine transformer coupling, and provide copper and vias for heat spreading. Then validate EMI and temperature on the actual board, with the intended switching devices and installation constraints.

Access Official Technical PDF

About The Author