Most applications that use power factor correction (PFC) are mandated by regulatory requirements like the IEC1000-3-2 harmonic reduction requirement in force since January 1, 2001. The regulation places certain limits on harmonic currents drawn by power supplies. This, in turn, requires an active PFC circuit. Some low-power, cost-sensitive applications still manage to use a passive PFC circuit. Comprised of only inductors and capacitors, a passive PFC circuit is simple and efficient for low-power applications. However, it requires a large bulk capacitor, can only be used in a narrow range input, can provide only a small hold-up time, and carries low-frequency ripple on the output. Hence, for applications over 200 W, active PFC is mandatory.
The easiest and most cost-effective way to implement an active PFC is via a boost circuit at the input end of the power supply. The two most popular types of boost topologies for implementing active PFC are transition mode (TM) and continuous-current mode (CCM) PFC.
TM and CCM refer to the way current flows in the boost inductor. In the CCM topology, the cycle-by-cycle inductor current doesn't fall to zero. The opposite of CCM mode is discontinuous-current mode. Here, cycle-by-cycle current falls to zero at the end of the cycle, and there's a dead time before the next cycle starts from a zero current. The TM topology is a discontinuous current-mode topology in which the next cycle starts just at the point where the current reaches zero. Thus, there's no dead time in each cycle.
TM- Vs. CCM-PFC
TM-PFC is simpler and cheaper than CCM-PFC. It's widely used in lighting ballasts; adapters; and low-power, switching-mode power supplies (SMPS). This simpler control technique makes it possible to use low-cost, 8-pin PFC controllers, or an MCU where available. Control-loop implementation can take place by multiplying the instantaneous line voltage by the output of the error amplifier and generating a variable frequency-switching pattern. The other method is to set a fixed on time for the boost MOSFET. In both of these approaches, the zero current of the inductor is detected to start the next cycle. Hence, the peak inductor current is proportional to the instantaneous input line voltage, making the average input current similar to the input voltage for achieving a high power factor (Fig. 1a) .
CCM-PFC uses a fixed-frequency control. The sum of on and off times is constant. Relative to TM-PFC, inductor size is larger for a given output power-frequency combination. The slope of switching currents in CCM topology is smaller, and so is the peak inductor current, resulting in lower EMI and lower rms current through the boost inductor and MOSFET (Fig. 1b) . Lower rms current, in turn, reduces conduction losses. Hence, the CCM topology is the preferred option for higher-power applications (typically over 300 W).
The TM topology offers several advantages:
- Because the control technique is simpler, TM controllers typically cost half the price of CCM topology controllers.
- As the boost inductor needn't store energy at the end of each switching cycle, it can be smaller and cheaper.
- Constant on-time implementation allows use of an MCU, saving the PFC controller cost in some applications.
- Because the commutation of boost diode from on to off happens at zero current, slower diodes can be used, slashing cost.
- Inner current feedback loop is much faster with feedback received every cycle by the inductor current's falling to zero.
But the CCM topology also has advantages:
- Current slopes are much smaller compared to the TM topology for the same average input current, creating lower peak currents through the boost inductor. A lower slope and a lower peak current reduce the cost of EMI filtering circuits. Also, lower peak current means lower RMS current for a given input power and frequencies, resulting in lower conduction losses in the inductor and the boost MOSFET.
- A constant switching frequency allows synchronization with downconverters. Also, for big enclosed systems like computers, the variable frequency of TM may not be acceptable.
Transition-Mode PFC Controllers
The biggest variable that affects system cost in a TM-PFC implementation is the controller, even though it may cost less than the MOSFET or even the diode. Consider the features of some of the most-used controllers for TM-PFC.
Maximum Supply-Voltage Rating: While startup and nominal operating voltage for most controllers are the same, the maximum allowable VCC in some controllers is over 30 V. For a standard PFC application, this may not be beneficial. But it could be important for applications where PFC is implemented along with the dc-dc converter stage, such as in a single-stage converter. For example, in a constant-current-output power supply for a battery charger, output voltage may vary beyond the normal tolerance of 10 %, making the supply voltage on the IC change along with it. Hence, a larger VCC voltage range can be advantageous.
Error Amp: Some controllers contain transconductance amplifiers, as opposed to the more often seen simple voltage amplifiers. Transconductance amplifiers enable isolating the voltage feedback pin from the compensation pin, creating slightly more design flexibility. They also simplify using an optocoupler when necessary for feedback. On the other hand, they're more prone to pick up noise, particularly the one produced by the high di/dt of the drain current.
Overvoltage Protection (OVP)/Protection Against Feedback Loop Opening: Most TM-PFC controllers provide OVP. Some of them, however, have a differentiating element that protects against the feedback-loop opening. To implement this protection, the voltage feedback pin can't be the same as the OVP pin.
Startup And Quiescent Current: With many new standards requiring higher low-load efficiency, this parameter must be compared thoroughly.
Output Driver: Because the TM topology is intended for lower-power applications, a gate-drive current in the range of 0.5 to 1 A is usually available. The important thing to note, however, is switching off speed, which has a major impact on switching off losses. Fortunately, due to zero-current turn-on in a TM topology, switch-on losses don't depend on the driver.
Internal Reference: Usually a 1% or 2% internally trimmed reference is supplied. Better precision of VREF over the range of operating temperature allows for precise setting of the output voltage, and optimizes the cost of the inductor and bulk capacitor.
Current-Sense Blanking: With careful PFC design, built-in current-sense blanking can eliminate the need for an external R and C.
Continuous-Current-Mode PFC Controllers
The range of usage for CCM-mode PFC is much wider, because almost any system over 300 W uses this topology. Consequently, a wide range of PFC controllers exist, from a very simple 8-pin device that provides just basic functions, to more-complex devices that can drive an additional active snubber or implement a zero-current switching topology. In the case of CCM-PFC controllers, some considerations remain identical to TM-PFC, such as:
- OVP/protection against opening of the feedback loop
- startup and quiescent current
- output driver
- internal reference
- current-sense blanking
CCM-PFC complexity brings in more variables, including:
Package: Eight-pin packages save board space and look very innovative. Relatively speaking, though, space isn't a major issue in most off-line power supplies. Therefore, attention should be paid to the additional complexities and costs introduced by eliminating some features, or combining some features into a multi-feature pin. For example, the small pin count and smaller package size may consequently mean lower gate-drive capability due to higher thermal resistance.
Frequency modulation: Also called dithering, frequency modulation enables varying of the switching frequency with respect to input line voltage. This switching-frequency variation distributes the EMI noise, resulting in cost savings on EMI filtering components.
Cycle-by-cycle peak current limiting: Smaller packages usually combine the current-sense pin for the feedback loop with the peak-current-limiting function. This will cause over-designing of the MOSFET in some applications.
Programmable oscillator frequency: A programmable switching frequency allows for better optimization of the inductor size and switching losses.
Brown-out: Brown-out protection prevents the PFC controller from restarting during the falling edge of the input line voltage. This prevents a glitch from appearing on the output voltage during switch-off. It also avoids overstress on the PFC and downconverter components.
Synchronization: With a synchronization feature, both the PFC and the downconverter can operate at the same frequency, preventing noise issues related to two nearly equal switching frequencies. If a leading-edge synchronization is available, it can produce further cost savings by reducing ripple current in the bulk capacitor and, hence, its size.
Boost diodes: After the controller, the boost diode is perhaps the second most important component of the PFC circuit. For transition-mode PFC, a diode with a recovery time ( TRR) of less than 200 ns is recommended. A wide range of input applications use 600-V diodes. For the CCM mode, several choices are available.
Fast-recovery diodes: Many manufacturers make diodes with recovery times of under 50 ns, and a dropout of less than 1 V. For most of the mid-range, below-100-kHz applications, these diodes are an optimum choice.
Tandem diodes: For PFC switching at 100 kHz or higher, the losses due to the diodes' TRR become significant. However, in most cases, the available high-voltage diode technologies can't achieve less than 50-ns of recovery time. Lower-voltage diodes (for example, 300 V) can offer a TRR of 30 ns. This led to the creation of a high-voltage diode that uses two diodes in series. For example, connecting two 300-V diodes in series creates a 600-V tandem diode with a TRR equal to that of a single diode (30 ns). The drawback of this technique is that forward voltage drop doubles to around 2 V.
High-voltage Schottky diodes: Some newer diode technologies, such as SiC or GaAs, offer 600-V devices with a Schottky-diode-like performance, meaning very-low forward voltage drop and zero recovery time. These diodes are ideal for applications that require the maximum possible efficiency and aren't so cost-sensitive. Because the process of making these high-voltage Schottky diodes is very complex, they usually cost three to five times more than comparable tandem or fast-recovery diodes. One major drawback of these diodes is their susceptibility to spikes. Thus, they require an RC snubber across them.
MOSFET: Losses in the MOSFET consist of three major components: the charging and discharging of its output capacitance during every cycle, gate-charge losses, and conduction losses. Typically, a 600-V MOSFET is used for a wide-input-range PFC.
Cost Analysis Of An 80-W Transition-Mode PFC Circuit
In a typical 80-W transition-mode PFC circuit, 90% of system cost is due to the input capacitor (C1), boost inductor (T), boost diode (D1), MOSFET, bulk capacitor (C5), and the controller IC (Fig. 2). The optimal design choice is to have the converter's switching frequency at maximum input line voltage and maximum rated output power. A higher switching frequency means lower efficiency, but also lower cost of C1, C5, and T. A higher switching frequency also increases the cost of the MOSFET and diode (or their heatsink) because of higher switching losses.
One very important consideration in selecting a switching frequency is EMI. FCC limits start at 150 kHz, in a way dictating some sweetspots in switching frequency selection. For example, with a 40-kHz switching frequency, the third harmonic will be at 120 kHz. Thus, to save cost, the EMI filters must be designed to suppress only fifth- and higher-order harmonics.
The next highest levels of switching frequencies span 100 to 130 kHz, depending on efficiency and size requirements. Typically, at 200 kHz and above, the advantage of the smaller size of C1, C5, and T is eliminated by the heatsinking requirement for the MOSFET and diode. Numerous application notes are available on TM-PFC ICs that can give precise data on the switching losses in the above mentioned six parts.
Cost Analysis Of A 400-W CCM PFC Circuit
A CCM topology is used in high-power systems where the MOSFET usually is the most expensive component, followed by the diodes and then the controller (Fig. 3). Switching frequency is the first design choice. The same considerations hold true for a TM-PFC circuit. Based on the switching frequency, one or two types of diodes should be tested in the actual circuit, and the selection can be made based on cost versus performance requirements.
Increasing the MOSFET size reduces the conduction losses. But it also increases losses due to its output capacitance and gate charge. Newer MOSFET technologies with lower output capacitance and lower gate charge for a given on-resistance would provide the best efficiency, without much cost increase.
For applications operating at very high power and requiring the highest possible efficiency, you can look into more-complicated current-mode topologies. A tapped-inductor boost topology, for example, can relieve some stress on the boost diode. A magnetic snubber will work too. Some advanced PFC controllers have zero-current or zero-voltage switching topologies to reduce switching losses.All of these advanced topologies improve efficiency at the cost of extra components and complexity. It's arguable how much value these advanced topologies bring to the table. Perhaps the old-fashioned way of using a better diode, better MOSFET, and lower switching frequency is still a less expensive way to go in many applications.