A Case Study on HuaXuanYang HXYS110N120MPI
HuaXuanYang Electronics HXYS110N120MPI is a 1200V/110A/26mΩ silicon carbide N-channel MOSFET in a TO-247-4L four-pin Kelvin-source package. It is designed for 800V DC fast-charger power modules, serving in the isolated DC-DC stage (LLC resonant half-bridge) and the three-phase active front end (AFE). Its 1200V blocking voltage provides roughly 1.5x headroom over the 800V bus, the 26mΩ on-resistance cuts conduction loss at full load, and the four-pin Kelvin source suppresses high-frequency switching ringing, helping charging modules exceed 97% efficiency.
Stand in front of a 350kW ultra-fast charging station and most people see a liquid-cooled cable thicker than an arm and the current numbers flickering at the vehicle connector. But what truly determines charging speed and operating cost is the row of hot-swappable power modules inside the cabinet. A 350kW charging stack is typically built from twelve 30kW standard power modules in parallel. Each module draws three-phase grid power, passes it through an AC-DC power factor correction (PFC) stage and an isolated DC-DC stage, and outputs adjustable DC from 200V to 1000V to match everything from legacy 400V platforms to next-generation 800V battery packs.
Across these two power conversion stages, semiconductor selection sets the ceiling for module efficiency, power density, and lifetime reliability. The 800V DC bus figure turns the question from "is a silicon IGBT good enough" into "how to pick the right 1200V SiC" — and that is exactly what this article breaks down.
Pain point one: blocking-voltage headroom anxiety. The 800V DC bus is a starting point, not a ceiling. Grid transients, switching overshoot, and fault conditions can push the bus briefly toward 900V or even 1000V. With a 1200V device, the margin is only 1.2 to 1.5 times. Three-level topologies like Vienna PFC halve the stress and allow 650V devices, but a two-level six-switch active front end (AFE) demands 1200V parts. Pick too low and the device avalanches; pick too high and both cost and conduction loss climb. Voltage-class selection is the first gate.
Pain point two: power density and magnetic volume. Chargers must fit standard cabinets, and 30kW modules are pushed to roughly one kilowatt per liter of power density. Silicon IGBTs cap switching frequency at 10 to 20kHz, leaving transformers and inductors large and heavy. Only by pushing frequency to 50 to 150kHz can magnetic component volume be cut by 40% or more. The prerequisite for high-frequency operation is that switching loss must not run out of control — and that is where SiC holds a fundamental edge over silicon IGBTs.
Pain point three: EMI and reliability at high frequency. Once frequency rises, dv/dt routinely hits 50 to 100V/ns. Any stray inductance in the gate loop invites ringing and false turn-on. In a conventional three-pin TO-247 package, the source inductance is shared between the power loop and the drive loop, creating common-source inductance coupling at high frequency — an invisible killer of devices and an EMI offender. Upgrading from three pins to a four-pin Kelvin source is not just adding one more lead; it is a structural fix.
HuaXuanYang HXYS110N120MPI is a 1200V/110A/26mΩ silicon carbide N-channel MOSFET in a TO-247-4L four-pin Kelvin-source package. Placed in the 800V fast-charger power chain, four parameters each map to a concrete engineering benefit.
| Parameter | Value | Engineering Benefit |
|---|---|---|
| VDS blocking voltage | 1200V | ~1.5x headroom over the 800V bus, absorbing fault overshoot; suits two-level AFE full-voltage swing and LLC primary half-bridge |
| ID continuous current | 110.4A | 30kW module at 800V bus draws ~37.5A continuous, ~2.9x margin, ample thermal headroom at full load |
| RDS(on) on-resistance | 26mΩ | At 37.5A full load, conduction loss ~36.6W (P = I squared times R), markedly lower than a same-voltage silicon IGBT |
| Package | TO-247-4L Kelvin | Separate Kelvin source pin isolates drive loop stray inductance, cutting ringing and false turn-on risk at high frequency |
The first benefit is in blocking voltage. For an 800V-bus charger, the LLC resonant half-bridge primary switches see nearly the full bus voltage, and 1200V is the industry-accepted class. When the bus briefly overshoots under fault conditions, 1.5x headroom is far more reassuring than stretching a 650V part to its limit.
The second benefit is conduction loss. 26mΩ is not an extreme low figure, but at a 30kW module‘s current level, full-load conduction loss is about 36.6W (P = 37.5 squared times 0.026). A same-voltage silicon IGBT at this frequency often has switching loss several times its conduction loss. SiC‘s real advantage is not in a single metric but in simultaneously holding both conduction and switching losses low — the prerequisite for high-frequency operation.
The third benefit is packaging. The fourth pin of the TO-247-4L is an independent Kelvin source, so drive current no longer shares source inductance with the power loop. During high-dv/dt switching transients, the voltage drop across common-source inductance superimposes on the gate-source voltage, causing ringing or even negative spikes that prevent clean turn-off. The four-pin structure physically separates the drive and power loops — a near-must for SiC drives above 50kHz.
The fourth benefit is system-level miniaturization. SiC can be pushed to 50 to 150kHz without switching loss running away. Doubling frequency means transformer and resonant inductor core volume can be trimmed by 30% to 40%, which is how a 30kW module fits into a standard 1U cabinet — something silicon IGBTs cannot achieve at this density.
Note: gate charge Qg, threshold voltage VGS(th), and junction-to-case thermal resistance for HXYS110N120MPI should be taken from the official HuaXuanYang datasheet. This article gives only order-of-magnitude references and does not state specific values.
A typical 30kW charging power module chain runs: three-phase grid input → active front end PFC → 800V DC bus → isolated DC-DC → vehicle output. HXYS110N120MPI can appear in two positions: the two-level six-switch AFE of the PFC stage, or the LLC resonant half-bridge primary of the DC-DC stage. The signal chain below follows the DC-DC LLC half-bridge as the example.
Control unit. Charger DC-DC stages typically use a DSP or dedicated power MCU (such as the C2000 family) to generate phase-shift or frequency-modulated PWM commands. An LLC resonant converter operates in frequency modulation, sweeping the switching frequency around resonance to regulate output voltage. Typical operating frequency is 50 to 100kHz; dual-active-bridge (DAB) phase-shift designs run in the 10 to 20kHz mid-band. The control loop also handles soft start across a wide output range (200V to 1000V), constant-current/constant-voltage handoff, and overcurrent protection.
Drive stage. SiC MOSFET gate drive differs fundamentally from silicon. HXYS110N120MPI requires an isolated gate driver delivering +18V turn-on and -3V to -5V negative turn-off bias. The positive voltage ensures full enhancement to minimize RDS(on); the negative bias clamps the gate during dv/dt transients to prevent Miller-induced false turn-on. Typical isolated gate driver choices include NCP51563, UCC21520, or equivalents, with CMTI (common-mode transient immunity) rated at 50 to 100V/ns. Gate resistance Rg is typically 10 to 22 ohms, balancing switching speed against ringing — refer to the datasheet and bench measurement.
Power stage. The LLC resonant half-bridge consists of two HXYS110N120MPI devices as upper and lower switches, with the midpoint feeding the resonant inductor and capacitor. It operates in ZVS (zero-voltage switching) soft-switching mode: after the lower switch turns off, resonant current commutates through the output capacitance and stray inductance, and the upper switch turns on as its drain-source voltage reaches zero, driving switching loss toward zero. This is the core reason LLC achieves 97% to 98% efficiency above 50kHz. SiC‘s low Qg keeps switching speed fast enough to satisfy the ZVS window; the body diode freewheels during dead time with near-zero reverse recovery charge, yielding clean commutation.
Load stage. After high-frequency transformer isolation, the secondary rectifies (using SiC diodes or synchronous rectifier MOSFETs) and feeds the vehicle battery. The 800V bus steps down through the transformer to match battery voltage (200V to 1000V wide range), with output ripple kept below 1% to protect the pack. The battery is a capacitive load; the control loop needs enough bandwidth to handle the dynamic response at constant-current/constant-voltage transitions, and the power devices must withstand inrush current stress at startup.
Connecting the four links: DSP issues frequency-modulated PWM → the isolated gate driver amplifies it into +18V/-4V drive pulses → HXYS110N120MPI completes power transfer under ZVS soft switching → the transformer isolates and rectifies to charge the battery. If any link fails, efficiency drops a couple of percentage points at best, or the device avalanches and the station goes down at worst.
One: gate drive must include negative turn-off bias — do not shortcut to 0V off. SiC dv/dt can reach 50 to 100V/ns. The Miller capacitance couples the drain voltage jump to the gate during switching. At 0V off, the gate can be pushed above VGS(th), causing shoot-through between upper and lower devices. Use -3V to -5V negative bias to lock the gate, paired with an isolated driver that has active Miller clamp capability.
Two: keep the TO-247-4L Kelvin source trace short and straight. The four-pin package‘s value is an independent drive loop, but if the PCB routes the Kelvin source in a long detour or runs parallel to the power loop for too long, you are adding stray inductance right back. Keep the Kelvin trace under 10mm, route it away from the power loop, and return through nearby vias.
Three: dead time can be shorter than silicon IGBTs, but it cannot be zero. In an LLC half-bridge, dead time allows ZVS commutation. SiC switches fast, so dead time can be compressed to the 100 to 300ns range. Too short and commutation is incomplete, forcing hard switching and spiking loss; too long and body-diode conduction loss during dead time adds up. Tune dead time against measured ZVS waveforms, not datasheet estimates alone.
Four: PCB stray inductance is the number-one enemy of high-frequency SiC. Above 50kHz, every nanohenry of stray inductance produces a non-trivial voltage drop during switching transients. Minimize power loop area, shorten commutation paths, and place the bus capacitor tight against the switches. These three layout principles matter more than upgrading to a more expensive device. Consider laminated busbars or embedded copper plates at this power level.
Five: thermal design is more than Rth calculation — look at the whole cooling architecture. Chargers are outdoor sealed or semi-sealed cabinets, with ambient temperatures potentially above 50 degrees C and prolonged full-load duty. HXYS110N120MPI‘s junction-to-case thermal resistance is per the datasheet, but the system-level question is the cooling path: forced air, cold-plate liquid cooling, or evaporative cooling — the choice depends on power density and reliability targets. Every 10 degrees C reduction in junction temperature roughly doubles device lifetime. Thermal design is the foundational investment in reliability.
Back to the opening question: how to select 1200V SiC on the 800V bus without stepping into pitfalls. HXYS110N120MPI‘s answer aligns four things: 1200V blocking voltage secures bus headroom, 26mΩ on-resistance controls full-load conduction loss, the TO-247-4L Kelvin package tames high-frequency ringing, and SiC‘s inherently low switching loss opens the door to high-frequency operation. These four parameters are not isolated nice-to-haves; they form a closed engineering loop of "withstand the voltage, conduct with low loss, switch fast, control the ringing" that supports 30kW modules exceeding 97% efficiency.
As an enterprise focused on advanced power semiconductors, HuaXuanYang Electronics continues to invest in its silicon carbide MOSFET lineup, covering 650V to 3300V across multiple voltage classes and package types, serving new energy, data center, and industrial power applications in high-voltage, high-frequency scenarios. HXYS110N120MPI is one of HuaXuanYang‘s 1200V SiC series devices aimed at high-power applications such as fast-charging piles and energy storage PCS. Specific gate charge, threshold voltage, and thermal resistance parameters should be verified against the official datasheet.
This article is for reference only and does not constitute any purchasing or design commitment. Design selection should be based on the latest HuaXuanYang official datasheet. For application support, contact HuaXuanYang technical support (sales@hxymos.com).
声明 / Disclaimer
本文部分参数引自华轩阳产品手册或第三方规格书,数据有更新可能,仅供选型参考,最终设计请以最新版规格书与实测为准。
联系 Contact:sales@hxymos.com