20W GaN Fast-Charging Design: Architecture, Efficiency, and Selection Notes

    This article reviews a 20W GaN fast-charging design based on a high-frequency quasi-resonant (QR) flyback converter with an integrated GaN primary controller, synchronous rectification, and PD/PPS/QC protocol control. It covers the hardware architecture, efficiency measurements at 120 V and 230 V, thermal design considerations, and key selection and test caveats.

    1. Design Positioning and Approach

    20 W is a common fast-charging power level for phones, tablets, and Bluetooth earphones. Charging is fast enough, while the adapter can still remain portable. GaN devices allow engineers to operate at higher switching frequencies, reducing the size of the transformer and filter components.

    This design uses a high-frequency quasi-resonant (QR) flyback architecture with GaN power devices and synchronous rectification. The design approach is straightforward: increase switching frequency to reduce magnetic components; use the low gate charge and low output capacitance of GaN to offset the switching-loss increase caused by higher frequency; and use synchronous rectification to reduce secondary conduction loss. The design supports PD3.0, PPS, QC3.0, and other protocols, and can adjust output voltage according to device requests.

    2. Hardware Architecture

    The system is divided into three parts: primary power conversion, secondary synchronous rectification, and protocol control.

    2.1 Primary Side: QR Controller with Integrated GaN

    The primary side uses a high-frequency QR flyback controller with an integrated GaN power transistor. The drive circuit, protection logic, and power device are integrated into the same chip, simplifying PCB layout and making parasitic parameters easier to control. The controller uses valley switching to turn on the power transistor at the lowest drain voltage, reducing turn-on loss and helping improve EMI performance.

    The supply range supported by the chip’s high-voltage startup circuit must be based on the selected model’s datasheet. During selection, verify the nominal values and test conditions directly in the datasheet.

    For peripheral components, the input uses an SMD fuse and an SMD bridge rectifier to reduce component height and facilitate automated placement. A high-voltage electrolytic capacitor and an inductor form a pi-type filter network that provides the DC bus for the flyback converter. The transformer typically uses an RM or EP core and Litz wire winding to reduce skin-effect and proximity-effect losses at high frequency.

    2.2 Secondary Side: Synchronous Rectification

    The secondary side uses a synchronous rectification control chip with an integrated power MOS; the nominal on-resistance is 16 mΩ. The chip can achieve 3 V low-voltage self-supply without an auxiliary winding, so the synchronous rectification controller can still obtain a stable supply at light load or low output voltage. During selection, verify the minimum self-supply voltage and the on-resistance test conditions in the datasheet.

    The chip determines turn-off timing by detecting the drain-source voltage. In both QR and DCM modes, it enables the synchronous MOS to operate at the appropriate timing, reducing additional losses caused by body-diode conduction.

    The secondary output filter uses a solid capacitor and ceramic capacitors in parallel. The solid capacitor suppresses low-frequency ripple, while the ceramic capacitors absorb high-frequency switching noise. Output current purity is ensured by both.

    2.3 Protocol Control

    The protocol chip supports PD3.0 and is compatible with QC3.0, Huawei FCP, and other mainstream protocols. It provides PPS programmable voltage regulation. The device performs a handshake through the CC pin; the protocol chip identifies the device type and requests the appropriate voltage level. In PPS mode, the output voltage can be adjusted in steps, making it suitable for devices with voltage-accuracy requirements.

    3. Efficiency Test Results

    The test platform consists of an AC input power meter, a DC output meter supporting PD triggering, and a host PC. The PC controls output-setting switching and records input and output parameters. Testing covers 120 V and 230 V inputs, with output settings including 5 V, 9 V, and 12 V, and current swept from 0.5 A to full load.

    The main test data are as follows:

    Input Voltage Output Setting Load Efficiency (Test Record)
    230 V (measured approx. 228 V) 5 V 0.5 A light load Approx. 91%
    230 V 5 V 1.5 A half load Approx. 87.3%
    230 V 5 V 3 A full load Approx. 85%
    230 V 9 V Half load Approx. 89.5%
    230 V 9 V Full load Approx. 89%
    230 V 12 V Full load (20 W, approx. 1.67 A) 89.77%
    120 V (measured approx. 121 V) 5 V Full load 84%
    120 V 5 V Light load Approx. 90%
    120 V 9 V Full load Approx. 89%
    120 V 12 V Full load Approx. 89.4%

    The test basis for these data is not complete. The test records do not specify the line-loss compensation method, meter model and accuracy, load step interval, or stability criterion. Therefore, the values in the table should be used as a relative reference and should not be used directly for certification or external promotion. In the 12 V full-load test record, the current was once marked as 2 A, which exceeds the 20 W power limit; this article consistently interprets it as approximately 1.67 A.

    At 230 V input, 5 V efficiency first decreases and then stabilizes as load increases, which is consistent with the typical behavior of a flyback converter. At high input voltage, a higher reflected voltage helps reduce primary peak current, thereby balancing conduction loss and turn-off loss.

    At 120 V input, efficiency is slightly lower. The main reason is that input current increases, so conduction loss in the bridge rectifier and ripple-current loss in the bus electrolytic capacitor both increase. Therefore, wide-voltage designs should focus on bridge-rectifier thermal dissipation and electrolytic-capacitor ripple-current margin.

    4. Temperature Rise and Thermal Design

    The temperature-rise test description states 25°C ambient, 230 V input, continuous full-load aging, and thermal-imager monitoring of the GaN device, synchronous MOS, and transformer core. The test power in that description is marked as 12 V/2 A, which does not match the 20 W design; this article interprets it as 20 W full load (12 V/1.67 A). The test record does not provide specific case temperature, temperature-rise values, run duration, or measurement point locations, so it is not possible to judge whether device temperatures meet specification requirements.

    Thermal design can focus on the following aspects: the heat-dissipation path of the GaN package, the copper area on the PCB, transformer core loss, and heat accumulation inside a closed enclosure. Temperature rise differs significantly between open environments and closed enclosures, so testing should be performed with the actual enclosure state. It is recommended to record key measurement points with a thermocouple or thermal imager, and to record ambient temperature, run duration, and load conditions at the same time.

    5. Selection and Test Considerations

    1. Unify the power basis. For a 20 W design, test current should be calculated according to the power upper limit, avoiding over-power conditions such as 12 V/2 A being mixed into the data.
    2. Clarify the efficiency test basis, including input voltage accuracy, meter accuracy, line-loss handling, and stability criterion.
    3. Verify device parameters, including power-device withstand voltage, synchronous-rectification minimum self-supply voltage, and controller supply range.
    4. Verify EMI margin and synchronous-rectification timing at different input voltages.
    5. Verify protocol handshakes, covering PD fixed voltage settings, PPS, and legacy protocol devices.

    6. Conclusion

    By combining high-frequency QR flyback, GaN devices, and synchronous rectification, this design maintains relatively high efficiency at both 120 V and 230 V input, with notable light-load efficiency. The highly integrated control chip reduces peripheral components, helping miniaturization.