Engineers building 220V-input AC-DC power supplies know the drill: efficiency mandates tighten every year — server PSUs chase 80 PLUS Titanium, telecom rectifiers face half-load efficiency audits, LED drivers compete on energy-efficiency tiers. The PFC stage is the first gate, and one that gets optimized over and over. Many teams have already upgraded the main switch from plain silicon MOSFETs to superjunction devices, or even to SiC MOSFETs, yet full-load efficiency still falls half a point to a point short. When the investigation finally converges, the culprit is often the unassuming boost diode sitting next to the switch — still a silicon ultrafast recovery diode (FRED). The reverse recovery of that diode is one of the most consistently underestimated loss sources in CCM PFC. This article walks the full accounting.

HXY Electronics HC1D08065E is a 650V/8A silicon carbide Schottky diode with a typical forward voltage of 1.3V in a TO-252-2L surface-mount package, and near-zero reverse recovery charge. It serves as the boost rectifier in CCM Boost PFC and bridgeless totem-pole PFC stages, replacing 600V/650V silicon ultrafast diodes. By the industry selection rule it suits PFC stages of roughly 600W to 1.2kW. Its unipolar Schottky conduction mechanism stores no minority carriers: the only switching-related charge is a nanocoulomb-class junction capacitance charge, essentially independent of junction temperature and di/dt.

1. The Three Bad Line Items of the Silicon Diode in CCM PFC

First, where the diode sits. In a classic CCM Boost PFC power path — after the EMI filter and rectifier bridge — the PFC inductor connects the input to the switch node; the boost diode runs from the switch node up to the 400V bus capacitor. While the switch conducts, the inductor stores energy and the diode blocks the bus voltage. When the switch turns off, inductor current freewheels through the diode into the bus. When the switch turns on again, the diode is forcibly reverse-biased — and that is the moment the trouble begins.

The first bad line item is the reverse recovery current spike. A silicon ultrafast diode is a bipolar device: during forward conduction its drift region stores a large population of minority carriers. When the voltage reverses, that stored charge must be swept out before the device can block again. Industry-public figures put Qrr of 600V/650V silicon FREDs in the 100~500nC class, and it rises sharply with junction temperature — full-load operation at 125°C degrades measurably versus room temperature. The recovery current spike rides directly on top of the switch‘s turn-on current: turn-on loss balloons, the switch junction heats up, and the recovery current interacting with loop stray inductance produces ringing that dirties the EMI spectrum. The physics of one diode inflates costs in three places at once — the switch, the thermal design, and the filter.

The second bad line item is a temperature positive-feedback loop. Qrr is a strong function of temperature: the hotter the junction, the heavier the reverse recovery, the larger the loss, and the hotter the junction gets. A heatsink with comfortable margin on the room-temperature bench can enter a vicious cycle during full-load high-temperature burn-in, failing both the efficiency and thermal numbers. This is one of the classic reasons silicon FRED designs struggle to pass burn-in tests in enclosed power supplies.

The third bad line item is the frequency ceiling. Conducted EMI regulation starts measuring at 150kHz. The ideal play is pushing the main switching frequency to 100~140kHz and beyond, trading a smaller PFC inductor and a more compact filter for power density. But reverse recovery loss scales linearly with frequency, so silicon FRED designs mostly stay parked at 65~100kHz, leaving the inductor volume, copper loss, and board area all constrained. For 300kHz+ operation, silicon diodes are simply out of the running.

All three line items point to the same physical root cause: minority carrier storage. The cure is one road only — switch to a unipolar device with no minority carrier storage. Silicon Schottky diodes have no reverse recovery, but beyond roughly 200V they lose practical value; they cannot reach a 400V bus. The production device that satisfies "no reverse recovery" and "650V blocking" simultaneously is the silicon carbide Schottky diode.

2. HC1D08065E: Wiping the Reverse Recovery Line Item to Zero

The HXY HC1D08065E is a 650V/8A SiC Schottky diode with typical VF of 1.3V in a TO-252-2L surface-mount package. A Schottky barrier conducts with majority carriers only; reverse commutation merely charges the junction capacitance — no stored carriers, no recovery current. The table below puts it head-to-head against a same-class silicon ultrafast diode:

Now the line-by-line accounting, for a 1.2kW CCM Boost PFC (220V input, 400V bus, 100kHz switching):

Conduction: the boost diode‘s average current equals the bus output current — 1200W divided by 400V gives 3A. At typical VF of 1.3V, conduction loss is about 3.9W; the silicon FRED at a 1.8V class figure is about 5.4W. SiC saves roughly 1.5W on conduction at room temperature — and silicon VF climbs when hot while the SiC positive temperature coefficient is gentle, so the gap widens further at temperature.

Switching: at an industry-median 250nC for the silicon FRED, the reverse-recovery-related loss is on the order of Qrr times bus voltage times frequency — 250nC, 400V, 100kHz computes to roughly 10W, split between the diode itself and, mostly, the turn-on current spike dumped into the switch. With SiC, only the junction charge Qc remains (industry-class reference 10~20nC; exact value per datasheet); by the capacitive loss formula of one-half times bus voltage times Qc times frequency, 20nC corresponds to about 0.4W. Swapping one for the other changes the switching-path loss by an order of magnitude — and SiC‘s Qc does not move with temperature or di/dt, so this accounting does not turn hostile at full-load high temperature.

Industry test endorsement: Infineon‘s public application literature provides a classic comparison — a 400W CCM PFC platform where swapping only the silicon ultrafast diode for a SiC Schottky, at the same 140kHz, cut total system loss by about 8.7W and lifted efficiency by roughly 2 percentage points; raising frequency from 70kHz to 350kHz with the SiC diode kept loss and heatsink size essentially flat while shrinking the PFC inductor volume by about 65%. The Infineon CCM Boost PFC design guide also gives a selection rule of thumb: 1A of diode rating per 150W of output power (cost-optimized), or 1A per 75W (premium, where lower VF at actual operating current pays off at low-line input). By this rule, the 8A HC1D08065E covers 600W (premium) to 1.2kW (cost-optimized) — squarely the mainstream power band of single telecom power modules, server PSUs, LED street-light drivers, industrial auxiliary supplies, and high-power chargers.

A note on data discipline: Qc, junction capacitance, reverse leakage, IFSM surge rating, and thermal resistance are all dynamic or thermal parameters — please refer to the datasheet; the figures above are industry-class references.

3. Typical Application Circuit: The Full Signal Chain of a 1.2kW CCM Boost PFC

Putting the HC1D08065E back into its system context, here is how the upstream and downstream cooperate. Power path: EMI filter, rectifier bridge, PFC inductor L1 (hundreds of microhenries, sized for ripple current), main switch Q1 (either a 650V superjunction or SiC MOSFET — superjunction when cost leads, SiC MOS when performance leads), boost diode D1 (HC1D08065E), bus capacitor Cbus (400V, sized for hold-up time), and the LLC or other DC-DC downstream stage.

Control loop: average-current-mode CCM control, using either a dedicated PFC controller (UCC28180, NCP1654 class) or a DSP digital scheme. Switching frequency is set at 65~140kHz — below the 150kHz EMI measurement threshold with harmonic margin. The current-loop bandwidth should be at least 5kHz to meet THD and power-factor targets; the controller also manages soft-start, controlling the bus capacitor charge ramp.

Drive loop: the switch gets a totem-pole output gate driver with 1~2A peak capability, with drive loss budgeted against the chosen switch‘s Qg. The diode itself needs no drive — but there is one easily overlooked coupling point: the SiC diode‘s junction capacitance charges and discharges at every commutation, and that Qc displacement current adds to the switch‘s turn-on current. It is nanocoulombs in magnitude, negligible next to a silicon FRED‘s recovery spike — which is exactly the physical meaning of "clean commutation."

Power loop: Q1 conducts, inductor current ramps up, D1 blocks the bus voltage; Q1 turns off, inductor current freewheels through D1 into the bus capacitor (1.3V forward drop); Q1 turns on again and D1 commutates reverse — with no recovery current, the switch current waveform shows only the small step corresponding to Qc. At 1.2kW the input RMS current is about 5.7A, the inductor peak current about 9A, and D1‘s average current 3A, so the 8A rating leaves generous thermal margin. On layout, the D1 TO-252 tab (cathode) solders directly to a large thick copper pour — around 4W of conduction loss at the 1.2kW level is absorbed by the copper alone, and the airflow requirement drops noticeably versus a silicon FRED design.

Load loop: behind the 400V bus capacitor sits the LLC resonant stage. A 1~2 percentage-point PFC efficiency gain plus cleaner bus ripple current improve the downstream stage‘s input conditions — the margin that carries a design through efficiency certification is accumulated one component at a time.

The advanced variant: bridgeless totem-pole PFC. Drop the rectifier bridge and replace it with two legs — the high-frequency leg does the switching (a 0V-turn-off 650V SiC MOSFET is a natural companion), while the line-frequency leg commutates at 50Hz with grid polarity and uses same-voltage-class SiC diodes (again 650V), eliminating two diode drops from the bridge. Combining the SiC switch path and the SiC diode path is the full recipe for moving a 400V-bus supply from "adequate" to "efficient and dense."

4. Design Pitfall Guide: Five Things Not to Overlook When Moving to a SiC Diode

First, keep inrush surge off this device. The bus capacitor charges from zero at cold start, and the surge far exceeds normal operating current. Industry application literature (the Infineon CCM Boost PFC design guide) explicitly recommends a dedicated line-frequency pre-charge diode (high I²t) bypassing the HF rectifier, with NTC or relay current limiting. The HC1D08065E‘s IFSM surge rating is per the datasheet — separating surge stress from the normal rectification path is reliability engineering basics.

Second, Qc is not zero. Swapping to SiC eliminates reverse recovery, but the remaining junction charge Qc still accumulates loss linearly with frequency — one-half times bus voltage times Qc times frequency. At the 65~140kHz band this loss is nearly negligible; pushing toward 250~350kHz, first check the datasheet‘s Qc and capacitance curves and add the capacitive loss into the switch‘s turn-on loss before evaluating the benefit.

Third, account for high-temperature reverse leakage. A Schottky device‘s reverse leakage rises with junction temperature — industry-class 650V devices sit at microamp level at room temperature and tens of microamps at 175°C. Under high temperature at full reverse voltage, leakage loss (leakage times reverse voltage) joins the thermal balance equation. Most applications will never notice, but the worst-case corner — full load, enclosed chassis, high ambient — deserves a calculation; don‘t do thermal design on VF alone.

Fourth, the TO-252 tab is the cathode. The heatsinking copper is itself an electrical node: the copper tied to the tab carries both heat and current, so use a large thick pour with vias to a backside thermal layer. Remember that this copper sits at a switching-node-related potential (swinging at high frequency) — keep it clear of sensitive nearby traces and respect creepage and spacing per the voltage class.

Fifth, re-check ringing and EMI after replacing a silicon FRED. SiC commutation is cleaner, but the switching-edge dv/dt character changes, and the snubber and EMI filter values tuned for the silicon diode may need trimming — measure the spectrum before deciding; do not assume the old values carry over. Also, when paralleling for higher current, the SiC Schottky‘s positive VF temperature coefficient supports automatic current sharing without ballast resistors — but keep the two devices‘ layout and thermal environment symmetrical, or the sharing degrades.

5. Vendor and Summary

Closing the account: moving the CCM PFC boost diode from silicon ultrafast to SiC Schottky eliminates reverse recovery as a physical loss source, and buys a clean switch-current waveform, a 1~2 percentage-point efficiency gain, an unlocked frequency ceiling, and smaller magnetics and filters — the per-piece price premium comes back through the heatsink and the magnetics.

HXY Electronics‘ SiC diode lineup spans 2A to 50A at the 650V class in mainstream packages including TO-220, TO-252, TO-247, DFN, and SOT-227 (HC1D and HC3D series); the HC1D08065E (650V/8A/typical VF 1.3V/TO-252-2L) is the entry workhorse for kilowatt-class PFC. A 1200V class covers 800V-bus rectification and freewheeling needs. Datasheets, production availability, and technical support: sales@hxymos.com.

One-sentence summary: if your PFC efficiency is stuck at 96%, don‘t reach for a bigger heatsink first — look at the switch‘s turn-on current waveform. If the spike is tall and wide, that silicon ultrafast diode is the next component to replace.

Disclaimer: The industry comparison data cited in this article comes from public sources and serves as technical selection reference only; specific electrical parameters are subject to HXY‘s official datasheets. For selection and technical support, contact 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