Why we don’t need silicon carbide diodes for PFC

The introduction of silicon carbide (SiC) diodes has been a welcome solution to the reverse recovery losses in continuous conduction mode (CCM) boost power factor corrector (PFC) converters. While SiC diodes offer negligible reverse recovery charge (Q_rr), the forward voltage drop and temperature coefficient often increase the conduction losses in the PFC boost diodes to an unacceptable level.

Fortunately, the recent introduction of interleaved boundary conduction mode (I-BCM) integrated circuit controllers offer a topological solution to the reverse recovery losses in the CCM boost diodes, which eliminates the need for SiC diodes in many applications from 200 to 800 W.

The CCM boost converter has traditionally been the topology of choice for high-power PFC converters operating in excess of 1 kW. It offers excellent total harmonic distortion (THD), universal input voltage range and reasonable efficiency.

At lower power levels, however, the boundary conduction mode (BCM) topology becomes a viable alternative. In BCM PFC converters, the inductor current is allowed to return to zero every switching cycle, which offers many benefits to the designer. The most important benefit is a reduction in power loss by soft-switching the output diode and power MOSFET.

This soft-switching increases the power-stage efficiency while using inexpensive silicon diodes. The primary disadvantage of the BCM converter is the high ripple current carried by the boost inductor, as well as the high peak currents in the diode and MOSFET. The high peak currents in the power semiconductors can easily be overcome by using devices with slightly larger ratings, while the high ripple current in the boost inductor is harder to deal with.

Since the inductor ripple current needs to be filtered before it reaches the line, this has traditionally limited the use of the BCM converter in applications to those under a few hundred watts. Figure 1 shows a comparison of inductor currents in the CCM and BCM topologies (not to scale).

It is apparent from Figure 1 that the BCM converter has a variable switching frequency. Every switching cycle, the MOSFET is turned on for a fixed time determined by the control loop. The time to ramp down the inductor current is dependent on the input-output voltage difference and the peak current; both continuously change during the line cycle. When the diode current decreases to zero, the converter enters a resonance time interval where the MOSFET drain voltage resonates toward ground. At the resonant valley of the MOSFET drain voltage, a turn-on signal for the MOSFET is generated by an auxiliary winding on the PFC inductor.

Two soft-switched operating modes exist for the MOSFET turn-on event. First, whenever the instantaneous line voltage is less than half the output voltage, the MOSFET is fully soft-switched, i.e. turned on with both zero current and voltage. This mode is shown in Figure 2, where the MOSFET drain voltage has resonated until the MOSFET body diode conducts.




A short time afterwards, the MOSFET gate is switched on and soft-switching results. The second mode is shown in Figure 3, where the line voltage is greater than half of the output voltage and the MOSFET is turned on in a zero-current switched transition.

In this case the MOSFET drain voltage resonates downwards to a valley, where it is switched on. While the MOSFET could be switched on at any time after the diode current has reached zero, switching it on at the valley greatly reduces the power dissipated from the parasitic capacitance on the MOSFET drain node. Since this loss is proportional to the square of the drain voltage, even turning on when the drain voltage is one half the output voltage reduces the power dissipated by a factor of four. In contrast, the CCM converter turn-on event is hard-switched up until the converter enters discontinuous conduction mode around line crossings.


By paralleling two BCM converters and operating them 180 degrees out of phase, the peak and average currents in the power stage components is halved. This interleaving of two power stages also reduces the need for large EMI filters, as the ripple current cancellation effect doubles the effective switching frequency while reducing the peak-to-peak ripple current. In short, this interleaved boundary conduction mode (I-BCM) converter allows for CCM-like line ripple current without the need for relatively more costly SiC diodes. Figure 4 shows the effect of ripple cancellation and frequency doubling in the inductor currents for the I-BCM converter.


There are other efficiency advantages to the I-BCM topology. Because there are two power stages, when operating into a light output load, one of the converters can be disabled, and the other will process the entire output power. This “phase management” allows for even higher efficiency compared to a CCM converter because a smaller MOSFET and diode of the single stage are being switched on and off during each switching period.

Thermal management is always an issue with power conversion and PFC converters are no different. CCM converters are usually a single power stage, which concentrates the heat sources in the power stage components- the MOSFET, diode and inductor. When using an I-BCM converter, the power dissipation of these components is divided between the two stages. For example, there are now two MOSFET packages to dissipate the same power as that in a single stage converter; likewise the same applies to the diodes and inductors. This spreads the power dissipation across the printed circuit board and also allows for a lower profile construction.

Author Bio:


Christopher Bridge is a principal systems engineer  in the Mid-Power Analog Group at Fairchild Semiconductor. He has over 14 years of experience in Power Electronics and Analog IC definition and is the holder of five US patents. Chris is currently working on definition of DC/DC controllers, multi-chip modules, and supporting circuits. He holds a B.S.E.E. from Virginia Tech.