Power supplies rarely operate at full rated output continuously, so optimizing efficiency at only one load point can miss important operating conditions. For engineers designing AC-DC power supplies, PFC efficiency optimization means managing losses across the load range while meeting power-factor and harmonic-current requirements. Control mode and converter architecture both influence that balance, but the best choice depends on line voltage, output power, component stress and cost priorities.
Why power factor correction affects efficiency
A nonlinear load can draw a distorted current from the AC supply, while reactive loads can shift current out of phase with voltage. Both effects reduce power factor. For a given real-power demand, a lower power factor requires higher RMS input current, increasing conduction losses in the distribution path and limiting the power available from a supply connection.
An active PFC pre-regulator is commonly placed between the input rectifier and the bulk capacitor. A boost converter can shape the input current to follow the line voltage while regulating a higher-voltage DC link. That link provides a more consistent input to the downstream converter and supports hold-up requirements. The PFC stage itself, however, adds losses and must be designed for efficiency as well as current shaping.
Match the operating mode to the load
The loss profile changes with operating conditions. At low line and heavy load, current-related conduction losses become especially important. Resistance in the current path—including the boost inductor, MOSFET, sensing element and filter—produces losses that rise with current squared. At light load, switching and other relatively constant losses take a larger share of the processed power, reducing efficiency.
Critical conduction mode
In critical conduction mode (CrM), each switching cycle begins when the boost-inductor current has fallen to zero. This naturally avoids diode reverse-recovery current at turn-on and allows the MOSFET to turn on near a voltage valley. These characteristics make CrM attractive in lower-power applications, and its relatively simple control is another advantage.
The trade-off is a large inductor-current ripple and a variable switching frequency. The frequency can rise at light load, increasing switching losses just when those losses have a greater effect on efficiency. The large ripple also raises RMS current and associated inductor and converter losses, limiting the practical power range of a single-channel CrM design.
Frequency clamp and foldback
Frequency-clamped critical conduction mode (FCCrM) limits excessive switching frequency by allowing discontinuous-conduction operation when needed. Maintaining a correctly shaped line current during these intervals requires compensation for the dead time introduced between cycles.
Current-controlled frequency foldback takes a related approach: the converter operates in CrM when line current is above a selected threshold, then reduces switching frequency as current falls. This can improve light-load efficiency and support stable operation at low power. It can also reduce the need to oversize the inductor solely to constrain switching frequency. Frequency reduction is not automatically beneficial, however: the controller must balance lower switching losses against the increased conduction losses associated with changed current waveforms.
Use CCM when ripple becomes the constraint
Continuous conduction mode (CCM) reduces inductor-current ripple, which can lower RMS current and ease input filtering. Lower ripple can also reduce stress and heating in the bulk capacitor. These advantages become valuable as power rises and a CrM design’s ripple would otherwise drive component size, losses or cost too high.
CCM brings its own switching-loss considerations. Unlike CrM, it does not inherently turn on at the inductor-current valley, and boost-diode reverse recovery can contribute to turn-on losses. Diode recovery characteristics and MOSFET switching behavior therefore matter when selecting components. The appropriate operating mode depends on whether ripple-related conduction losses or switching-related losses dominate in the intended application.
Interleaving shares current and reduces ripple
An interleaved PFC uses two or more converter branches operating with phase shifts between them. Each branch handles part of the power, and their switching-frequency ripple currents partially cancel at the input and output. This can reduce the RMS current in the bulk capacitor and ease EMI filtering. It can also distribute heat across branches and extend the useful power range of CrM.
Interleaving requires additional components and control complexity, but those components can be smaller than the parts needed for a comparable single-channel stage. The approach can therefore be useful when thermal distribution, ripple reduction or a low-profile form factor matters. Its efficiency benefit is not identical at every operating point: conduction losses and switching losses must be assessed for the selected devices and line and load conditions.
When a bridgeless PFC is worth considering
A conventional input bridge places two rectifier diodes in the current path during each half-cycle. Their forward-voltage drops create conduction losses, with a more pronounced impact at low line. A bridgeless design changes the input arrangement to reduce the number of diodes in the active current path.
In a two-boost bridgeless implementation, separate PFC branches serve the two line half-cycles, with one branch active at a time. Removing a diode drop can improve efficiency, particularly at low input voltage. The gain depends on operating conditions and the rest of the design, including MOSFET conduction losses. The architecture also requires careful attention to current sensing and implementation. It is therefore a potential choice when efficiency targets justify added complexity, rather than a universal replacement for a bridged or interleaved design.
Choose the architecture around the real operating profile
For a practical design review, compare candidate solutions under the conditions that drive component stress and efficiency:
- Light load: Examine switching frequency, constant losses and the stability of low-power operation.
- Heavy load at low line: Assess input current, resistive conduction losses and thermal requirements.
- Ripple and filtering: Consider whether CCM or interleaving can reduce inductor ripple, capacitor RMS current or EMI-filter demands.
- Efficiency targets and cost: Weigh the diode-loss reduction of bridgeless PFC against added design complexity, and evaluate component choices as part of the complete loss picture.
These comparisons should reflect the supply’s actual load range and input-voltage conditions, not just its rated output point.
Conclusion
PFC efficiency optimization is a system-level decision. CrM offers attractive turn-on behavior but may suffer from high light-load switching frequency and substantial ripple. Frequency foldback can address light-load losses, while CCM and interleaving help manage ripple as power increases. Bridgeless designs can reduce rectifier conduction losses, especially at low line, but add implementation considerations. Matching control strategy, topology and components to the application’s operating profile is the practical route to balancing efficiency, thermal performance and cost.