Frequency and Cost Boundaries of the PFC Stage in Industrial Power Supplies
Industrial power supply PFC SiC diodes: recovering costs by replacing silicon FRDs with SiC
The unit cost of a SiC diode is about 3–5 times that of a silicon FRD, but in the PFC boost stage, the switching-loss savings from zero reverse recovery scale with frequency. Using a 3 kW UPS as a model, SiC conduction loss is only about 0.5 W higher, while above 65 kHz the switching-loss savings can already cover this increment. Whether the premium can be recovered depends on how well the switching frequency matches the power profile.
Industrial PSU PFC selection has long been dominated by a simple question: SiC diodes cost several times more than silicon fast-recovery diodes, so what system-level benefit justifies the premium?
The answer is not a general assertion that “SiC is more efficient.” It lies in a specific engineering variable: switching frequency. The SiC premium space is, fundamentally, a function of frequency.
Loss Accounting: What Conduction Costs, Switching Returns
Take a 3 kW double-conversion online UPS as a model, with a 400 V PFC bus and an average diode current of about 4.2 A. Under this condition, SiC Schottky conduction loss is approximately 0.5 W higher than that of a silicon FRD at the same current class—a deterministic cost on the conduction side.
The switching side moves in the opposite direction. Every time a silicon FRD transitions from conduction to blocking, stored charge is swept out, and the reverse-recovery current spike acts on both the diode and the main switch. This loss scales linearly with frequency, whereas SiC’s Qrr approaches zero, leaving switching loss nearly frequency-independent.
From 65 kHz to 100 kHz, the switching-loss savings from SiC already cover the 0.5 W conduction increment, and total loss begins to favor SiC. Above 150 kHz, the silicon FRD’s reverse-recovery loss deteriorates sharply, and the SiC benefit widens rapidly. In more aggressive totem-pole topologies, SiC solutions achieve 98.8%–99.3% full-load efficiency at 200 kHz–500 kHz+.
In other words, the SiC premium payback boundary sits roughly around 65 kHz. Below it, the conduction increment may consume the switching savings; above it, the benefit scales with frequency.
The System-Level Ledger: Three Payback Paths Beyond the Device Price Gap
The device cost differential is the starting point, not the endpoint. At the system level, there are three payback paths for the SiC premium.
The first is heatsink cost compression. SiC Tjmax is typically 175°C, 25°C higher than silicon FRD, and switching loss is substantially lower. At the same output, the heatsink can be reduced or simplified. For rack-mount UPS or compact industrial PSUs, a downsize in heatsink and fan specification directly offsets part of the device cost.
The second is magnetic component miniaturization. Higher switching frequency permits a smaller PFC inductor. Pushing frequency from 65 kHz to 150 kHz or above yields a deterministic reduction in inductor volume—for space-constrained industrial PSU enclosures, this means higher power density or layout simplification.
The third is lifetime electricity cost. Analysis of nearly 20 typical industrial PSU cases shows that system-level cost amortization and long-term energy savings bring the payback period into the 1.5–3 year range. For continuously operating UPS or industrial PSUs, this period is financially viable.
The Cross-Boundary of Power Class and Frequency
Premium payback is not universal; it varies with the combination of power class and frequency.
The low-to-medium power range (3–10 kW) , represented by UPS and general industrial PSUs, typically operates at 20–100 kHz. The SiC payback boundary begins to appear above 65 kHz. If the design stays at 20–40 kHz, the conduction increment may dominate, and the economic case for SiC is incomplete. But if the product requires higher frequency to reduce size or meet efficiency grades, the SiC premium acquires system-level justification.
The medium-to-high power range (above 10 kW) , typified by welding machines and large UPS, is different. Higher power means larger average diode current, and the absolute SiC conduction increment grows accordingly. The payback boundary shifts to higher frequency, typically above 100 kHz, before switching-loss savings reliably cover the conduction cost. In this range, localized replacement is the more pragmatic path: use SiC only at the boost diode position where reverse-recovery loss is most concentrated, rather than across the entire bridge.
Welding machine PFC presents an intermittent high-current load profile with large swings in average diode current. SiC’s zero reverse recovery pays off most during peak-current intervals, while the conduction increment becomes more prominent during light-load periods. Selection should be based on weighted loss over the actual welding duty cycle, not the rated operating point alone.
Model Entry Points: Matching Frequency and Current
Hypersemi’s SiC SBD portfolio provides comparable entry points at the current classes common in industrial PSU PFC stages.
CR30120L2X100 is a 1200 V / 30 A / VF 1.4 V device in TO-247-2L, covering 800 V bus PFC boost stages. Its through-hole package aligns with the heatsink mounting interface of mainstream silicon FRDs, so replacement evaluation does not require mechanical redesign. CR2065SX100 is a 650 V / 20 A / VF 1.25 V device in TO-263, suited to 400 V-class buses or compact PFC stages with surface-mount layout requirements; its lower VF narrows the conduction-loss gap with silicon FRDs.
Decision Variable | Silicon FRD Advantage Range | SiC Payback Boundary | Selection Implication |
Switching frequency | Below 65 kHz | Above 65 kHz; benefit widens above 150 kHz | Calculate conduction increment first in low-frequency designs |
Power class and current | High light-load share, low average current | Continuous high load, pronounced peak current | Use weighted loss, not rated point |
System justification | No thermal or volume constraints | Hard requirements on heatsink compression or power density | Premium needs system-level benefit to back it |
The full range of 1200 V / 650 V SiC SBD voltage, current, and package combinations can be located on the SiC diode selection page according to actual PFC topology conditions.
Payback Is a Frequency Question
The economics of SiC diodes in industrial PSU PFC cannot be answered apart from switching frequency. The 3 kW UPS model gives the boundary: about 0.5 W more conduction loss, switching savings beginning to cover above 65 kHz, and the benefit widening significantly above 150 kHz. System-level payback falls in the 1.5–3 year range, provided the design actually operates in SiC’s advantageous frequency range and the thermal or volume compression materializes as calculable cost savings.
If the target design remains in the low-frequency, low-load-factor, thermally unconstrained range, silicon FRDs remain a reasonable choice. The SiC premium is recoverable only in scenarios where frequency and power density have already pushed silicon devices against the loss wall.