Electronic Design
PFC - A Little Old School Knowledge - Part II

PFC - A Little Old School Knowledge - Part II

How does active PFC work?

In Part 1 of this article we discussed electric power basics and the need for and advantages of active power factor correction. The next question should be “Okay, then how does active PFC work?” Here’s my answer: The main tool for active power factor correction is a boost converter. This component rates the nod for the job since a boost converter operates continuously over the entire range of the rectified AC waveform. Other topologies can do this, too. You may be wondering if a buck converter is one of them. No, you can’t use a buck converter for active PFC. Here’s why.

The output of the buck has to be something less than Vpeak. Therefore, when the input drops below this output voltage, the buck converter will shut off for this duration creating a dead band in the current waveform and consequently a non-sinusoidal input current. The buck-boost and the boost-buck are examples of converters whose outputs can be greater than, less than or equal to the peak line input voltage and still operate continuously throughout the rectified line voltage excursions. These are also good candidates for active PFC.

We will primarily discuss the boost topology, as this is the main vehicle used for series type PFC. This has been further classified based on the conduction mode of the inductor. There are two main categories worthy of discussion: continuous conduction mode (CCM) and boundary conduction mode (BCM).


The most popular circuit for higher power levels is the continuous conduction mode or CCM boost. These circuits are typically operated with fixed-frequency PWM controllers, such as the IR1150. This particular topology will have lower ripple currents in the output capacitor and switches at the expense of a slightly larger core to handle the higher total flux in the inductor. 

A discussion of inductors is in order here. For the CCM PFC, assuming the same core size and output power level, Btotal is always higher than a comparable circuit running in discontinuous or boundary conduction mode (recall, in an inductor Btotal = Bdc + ΔB/2). A CCM boost converter has a high DC flux and a lower ΔB as seen by the core. The output of the converter commands the DC flux (DC flux is proportional to ampere-turns; this time-averaged current is drawn by the load). The AC flux or ΔB component relates directly to Ohm’s law for an inductor. The larger the inductor value, the lower the ripple current (V/L=di/dt).

A CCM inductor will have a larger inductance value than the alternate boundary conduction mode solutions assuming similar switching frequency and line/load conditions. This larger inductance can be accommodated with more turns on the same air gap or a smaller airgap and the same number of turns. In either case, the total flux will be higher than the BCM solution. In most cases this drives the magnetics designer to use a slightly larger core. The tradeoff for this is lower ripple in the charge and freewheel switches.

Now let’s talk about diodes. It is noteworthy to mention that in continuous conduction mode, the freewheel diode (Fig. 1) is hard switched at turn off by the MOSFET turning on.The channel of the MOSFET has to sweep out the minority carriers in the PN junction of the freewheel diode that was conducting immediately before the MOSFET was turned on. At higher power levels, this is of particular concern and the trr characteristics (both time and energy) need to be carefully understood. To minimize this impact on overall efficiency, many vendors make PN diodes with soft recovery characteristics. The stealth diodes from Fairchild are a good example of this. Further, there are silicon carbide solutions that are high-voltage Schottkies. Recall, a Schottky is a metal/silicon junction that is a majority carrier device. The reverse recovery event for these types of diodes is almost negligible; however, this is traded for a higher price than the PN junction devices.

In either case, the designer should consider using a good, general purpose, high Ifsm rated diode as a “precharge” switch placed between bridge + and the output capacitor. This diode does not need to be fast. When the boost is up and running, this device is reverse biased and has virtually no interaction with the circuit operation. This diode handles the inrush current to the capacitor when power is applied to the input. It also keeps the inductor out of saturation for this brief time and keeps the faster, lower Ifsm rated freewheel diode from conducting this slug of current.

In most applications, the CCM PFC is the clear winner in a single-stage converter above 300 W. At this power level, the cost savings of lower current in the freewheel diode and MOSFET outweigh the cost of a slightly larger core for the inductor. CCM PFC is typically a fixed-frequency PWM or integral cycle control.


The boundary conduction mode or BCM boost circuit is set up to operate the inductor at critical or boundary conduction mode. For a given power level on a given size core at similar switching frequencies, the flux in the BCM inductor will be less than the CCM PFC circuit. The inductor current decays to and starts from zero in these topologies. The ripple currents are thereby higher requiring higher current ratings in the switches and capacitors. The inductor can be slightly smaller per unit output power due to lower total flux from the absence of the DC offset in the flux waveform. IR currently does not offer a BCM PFC solution. From a control standpoint, most BCM PFC stages run either a fixed on time or a slightly modified hysteretic control. These types of controls allow the frequency to vary a little to allow zero current transitions over wide ranging line and load conditions.

There are then combinations of these ideas. There are interleaved boost converters that use two stages running 180 degrees out of phase. This decreases the ripple currents in the input and output capacitors and effectively doubles the output frequency. These are popular in higher power applications above 300 W. The tradeoff is more inductors and power switches. When space becomes a concern or there is a restriction in one dimension, these converters are often advantageous.


The control careabouts in a PFC circuit are both stable, regulated output voltage and a sinusoidal input current. Many PFC controllers in the market today use a multiplier arrangement to achieve good power factor over varying line and load.

In this topology, the output of the voltage error amplifier is multiplied by the reciprocal of the average input voltage squared. This gives a larger DC signal at low line and a smaller signal at high line conditions for a given load.

This DC value is then multiplied by the rectified input voltage signal (absolute sine). This signal behaves as a current over line and load. At high line, the DC portion of the multiplier input is low and the absolute sine input is high. The squaring of the DC portion keeps the multiplier output waveform amplitude low in these conditions, as it should be. At low line, we see the inverse effect, the multiplier output is higher—again, as it should be to reflect the true line current. The resultant multiplier output is then used as a reference for the current amplifier. The inverting terminal of the current amplifier is then tied to the shunt that measures current through the boost power switch. The output of the current amplifier then drives the PWM. The main drawback to these types of controllers is price.

A part like the IR1150 can help with regards to price. It is a one-cycle controller that doesn’t use expensive multipliers like older parts do. IR’s One Cycle Control mechanism makes control much simpler. The output of the boost is sensed via an error amplifier. This signal is then split over two paths.

The first path integrates this signal to produce a ramp. The period of the ramp is determined by the high cycle of the clock signal. The deadtime in the ramp is determined by the off time of the clock signal. This is fed into the positive (+) input of the comparator.

The second path adds the switch current to the EA signal and presents it to the negative (–) input of the comparator. The output of the comparator (Fig. 2) is buffered to drive the boost switch.

The elegance of this solution lies in its simplicity. Similar to traditional controls, the voltage loop rolls off at a frequency much lower than the input perturbation. The comparator then sees a DC value with the peak switch current waveform summed in. at low line, the current needs to ramp up for a longer amount of time to reach the turn off value. At high line, the inverse is true. On a cycle-by-cycle basis, when the input voltage at a low value, the boost switch simply stays on longer to ramp the current to the appropriate value. When the input voltage is at a higher value, the current will ramp up faster and the duty cycle will be shorter. This simple method accomplishes precisely the same control, duty cycle variation, and input current shaping as its more expensive predecessors. When the appropriate level is reached, the boost switch is turned off. This control mechanism regulates the output while dynamically tracking the input voltage waveform. (An implementation of the IR1150 controller—AN1077—can be found on www.irf.com.) 

Hopefully this has provided some guidance and made some sense out of a topic that has been so overemphasized that the salient points may have been forgotten.


IR APEC 2005 Paper “One Cycle Control IC Simplifies PFC Designs”

Unitrode U134, “UC3854 Controlled Power Factor Correction Circuit Design,” Philip C. Todd

Navpers “Basic Electricity 10086A” US Navy publication, 1960

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.