Access Official Technical PDF

Share Summary:LinkedInEmail

GaN power supply design can combine high switching frequencies with compact power stages, but realizing those benefits depends on more than selecting gallium nitride transistors. A 300 W adapter reference design illustrates how topology, control, PCB layout, thermal planning and protection work together in a high-density converter. The isolated design accepts 90–265 V RMS input and delivers a nominal 19 V output at 16 A, with a maximum output current of 18 A and peak power capability up to 340 W.

Its architecture pairs a synchronous power factor correction (PFC) boost stage with a half-bridge LLC converter and secondary-side synchronous rectification. GaN HEMTs are used as primary-side switches in both power stages. The design is intended as an engineering reference, not a universal implementation: its documented performance reflects specific design choices and trade-offs among size, efficiency, ripple, noise and protection.

How the PFC and LLC stages work together

The front end uses a synchronous boost PFC converter operating in critical conduction mode (CrM), also known as transition mode. Depending on loading, operation can move between CrM and discontinuous conduction mode. The NCP1616 controller manages the PFC stage to support unity power factor and low input-current total harmonic distortion. A GaN HEMT acts as the synchronous boost switch, while an NCP4306 controller provides its drive and synchronization.

Energy from the PFC stage is stored in the bulk capacitor bank, then supplied to the LLC power stage. The LLC converter operates at 500 kHz at nominal load and uses an NCP13992 current-mode controller. Its primary-side half bridge uses GaN HEMTs driven by an NCP51820 high-speed half-bridge driver. On the isolated secondary side, synchronous rectification uses NCP4306 controllers and parallel 60 V MOSFETs on each branch. Placing the rectification circuitry on a dedicated daughter card separates it from the main board and supports a modular layout.

Why modular construction matters

The reference design divides key functions among a main board and daughter cards for the bridge rectifier, bulk capacitors, LLC stage and synchronous rectification. The main board carries the input and output connections, EMI filter, PFC circuitry, output capacitors and module interfaces.

This arrangement provides practical development options. Engineers can evaluate or update an individual module without redesigning the complete board, and the interfaces leave room to investigate alternative daughter-card implementations. The boards use a two-layer construction with 70 μm copper plating, a choice intended to support thermal management and reduce conduction losses, particularly on the high-current secondary side. The complete board measures 55 × 159 × 18 mm, and the stated power density is up to 32 W/in³.

GaN power supply design depends on PCB layout

Fast-switching GaN devices make parasitic inductance and capacitance important design concerns. The reference design emphasizes minimizing the power-loop area and using a flux-cancellation approach: arrange the forward and return paths so their magnetic fields oppose one another. Reducing loop inductance helps limit stored energy and voltage spikes at the switching devices, while compact placement can also support EMI performance.

Several layout practices follow from this principle:

  • Keep high-frequency decoupling close. Place the half-bridge ceramic capacitor near the high-side switch drain and provide a short path to the switching devices. The manual recommends at least 100 nF for this decoupling role.
  • Control the return path. Route the ground return beneath the GaN devices and minimize overlap with the switching node. Use multiple vias to connect the return path where needed.
  • Separate gate-drive and power currents. Keep drive traces close together, minimize their loop length and connect driver ground to the device’s Kelvin-source connection where available.
  • Plan for heat removal. GaN HEMTs in this design are cooled through their source pads, so the copper area and thermal path must reflect the losses generated in the application.

These details are particularly relevant because GaN devices have low parasitic capacitance and can switch rapidly, but the reference manual cautions that their ruggedness is lower than that of standard silicon MOSFETs. Careful implementation and handling are therefore essential; device selection alone does not ensure reliable operation.

Control, EMI and protection trade-offs

The design’s EMI strategy spans the input filter, PFC pre-filter, local ceramic decoupling and physical grounding. Common-mode chokes address common-mode noise, while capacitors and differential-mode inductance help limit differential noise. A short, low-impedance protective-earth connection to output ground also serves as an EMI path in the documented implementation.

At the controller level, separating quiet ground from power or current-sense ground helps keep switching currents away from sensitive feedback and compensation signals. The manual recommends placing controller decoupling near supply pins, routing feedback dividers away from high-voltage switching nodes and filtering setup or protection pins. These practices aim to reduce noise injection and unwanted control behavior.

Protection settings are application-specific. The LLC controller configuration includes current limiting, short-circuit response, maximum on-time protection and brownout handling. Some available features require additional components; for example, the stated over-temperature protection requires an NTC thermistor to be assembled. The board was not optimized for surge or lightning protection, and the documentation recommends considering additional thermal management, especially at low input voltage.

Ripple and operating conditions

Output ripple is another design trade-off. The general parameters list 500 mV peak-to-peak at full load, while the manual notes that ripple can reach 550 mV peak-to-peak during skip-mode operation. An additional output filter can reduce ripple to below 220 mV in the skip range and below 120 mV at full load. The design also specifies standby input power below 150 mW at both 110 V RMS and 230 V RMS, with an operating temperature range of 0–50°C under the stated evaluation-board conditions.

Practical implications for power-conversion teams

This reference design demonstrates that high-density power conversion is a system-level engineering exercise. The PFC, LLC and synchronous-rectifier stages must be coordinated with their controllers, magnetics, decoupling, feedback, EMI filtering and thermal paths. Modular construction can make development and testing more flexible, while disciplined layout helps manage the fast switching behavior of GaN devices.

For engineers adapting the approach, the key lesson is to evaluate the complete operating envelope rather than focus on peak power density alone. Input range, load behavior, ripple, standby consumption, EMI, thermal conditions and protection requirements all influence the final implementation. The board provides a useful architectural and layout reference, but its stated trade-offs and protection limitations must be considered before applying the design to a product or industrial power-conversion system.

Access Official Technical PDF

About The Author