Designing offline power supplies using power-factor correction

To develop offline power supplies that use PFC (power-factor correction), you must first understand the meaning of PFC and the regulations that surround the design of offline power supplies.

PFC aligns the current waveform with the voltage waveform. If the voltage and current waveforms are not aligned, the PF (power factor) is less than 1 (Figure 1); otherwise, the PF is equal to 1 (Figure 2). In most cases, PF-corrected designs have PFs of 0.95 to 0.98. Most supplies that are not PF-corrected have PFs of approximately 0.6. In a three-phase system, if the voltage and current waveforms are not aligned, nonsinusoidal currents at three times the line frequency flow in the neutral line, requiring companies that supply ac power to use larger neutral conductors than would otherwise be necessary and to install more capacity to produce the required number of kilovolt-amperes. These harmonics are tough on power systems and can lead to overload and overheating of transformers, capacitor banks, and other power-system elements, sometimes tripping circuit breakers.

With this basic understanding of PFC, your next step is to determine when to use the PFC technique to comply with regulations in the regions where the product you are designing will be used. In 2001, the European Union put into effect EN 61000-3-2 to establish limits on harmonics as high as the 40th of ac-line-powered equipment's input current. Amendment A14, which took effect on January 1, 2001, has eased the impact of power-harmonics requirements in Class D, a group of equipment now defined far more narrowly than in the original standard: PCs, PC monitors, and television sets (Table 1).

On a circuit level, the main cause of PFs of less than 1 in SMPSs (switch-mode power supplies) is that the main capacitor, C_IN (Figure 3), charges only when V_IN is close to V_PEAK or when V_IN is greater than the capacitor voltage, V_CIN (Figure 4). To achieve a PF-corrected power output, designers often use a boost converter to maintain a voltage higher than the line voltage's peak value. The boost topology maintains this increased voltage; the PFC controller makes the average inductor current proportional to the input-voltage waveform. The boost topology also allows the average inductor to charge the input capacitor, C_IN, now on the output side of the boost controller rather than after the bridge-diode device (Figure 5).

Several IC manufacturers provide PFC boost converters. For example, Fairchild offers discontinuous- and continuous-mode devices. Some of these, such as the FAN7527B, are stand-alone PFC controllers (Figure 6a). Others, such as the FAN4803, integrate PFC and PWM operation into one packaged device that supplies more than 500W (Figure 6b). This variety allows designers to develop offline power supplies that use fewer components and require less space. These combo devices not only achieve PFC control, but also use flyback, forward, or other types of voltage-conversion topology to convert from large dc voltages, such as 385V dc, to 12V dc.

Discontinuous-mode controllers, such as the FAN7527B, offer PFC control at output-power levels up to and including 200W. Such devices operate in the discontinuous mode in which the boost converter's MOSFET turns on at zero inductor current and turns off when the current meets the desired input-reference voltage (Figure 7). In this way, the input-current waveform follows that of the input voltage and achieves an average inductor current that is in phase with the input voltage.

Compared with continuous-mode devices, discontinuous-mode units have higher I²R and skin-effect losses, use larger magnetic cores, and require larger input filters because of the larger inductor-current swings. On the positive side, use of these devices is less complex because they switch the boost MOSFET on when the inductor current is zero, so the boost diode needs no reverse-recovery-current specification, which allows the use of a less expensive diode. Also, the use of variable switching frequencies spreads the spectrum of the EMI (electromagnetic-interference) signature that results from the boost switch's zero-current turn-on.

Continuous- or hard-switcher modes typically suit SMPS power levels greater than 200W. In this mode, the boost converter's MOSFET does not switch on when the boost inductor is at zero current. Instead, the current in the energy-transfer inductor never reaches zero during the switching cycle (Figure 8).

In consequence, the current swing is less than in the discontinuous mode, resulting in lower I²R losses, lower inductor-core losses, and reduced EMI, which allows the use of a smaller input filter. However, because the MOSFET turns on when the inductor and diode currents are nonzero, a fast reverse-recovery diode is necessary to minimize losses.

Designers can use various techniques to reduce the hard transition of a continuous-mode PFC controller. Fairchild's FAN4822 implements one such technique, ZVS (zero-voltage switching). The primary purpose of using ZVS is to reduce the power loss or power dissipation in the PWM switch during turn-on. More losses occur in the external MOSFET with the increase in switching frequency. This technique uses an LC-resonant tank circuit to discharge the MOSFET's C_OSS (output capacitance, Figure 9). This capacitance is the sum of the C_DS (drain-to-substrate capacitance) and C_GD (gate-to-drain capacitance). ZVS works well with power supplies that switch at frequencies greater than 100 kHz. It also reduces the MOSFET's gate-drive requirements because the C_GD charge does not exist when the V_DS (drain-to-source voltage) equals zero.

When the switch is off, you have a charge, E, on C_OSS of E=1/2C_OSSV_DS ² with V_GS=0V. Once the switch turns on, this energy dissipates and can become a limiting factor in SMPS topologies. As the switching frequency increases, so does the power dissipation:

where P_D is the power dissipation, E is the energy stored within C_OSS, and f_Sis the switching frequency.

In its basic form, ZVS works as follows (Figure 10): When switch Q₁ turns on, V_CR=V_DS=0, and V_DO=0.When switch Q₁ turns off, the voltage across the C_R (resonant capacitor) increases linearly and works with the L_R (resonant inductor) to dissipate the charge stored in C_OSS. Note: L_O>>>L_R. The voltage across L_O then reverses polarity, and D₀ begins to conduct, thereby expending the energy stored in L_O. The resonant inductance-capacitance network components, C_R and L_R, are selected based on the maximum input voltage V_INMAX and minimum output current I_OMIN for the circuit to remain resonant over all operating conditions of line and load.

In its combo mode (PFC/PWM) devices, Fairchild uses a patented LEM/TEM (leading-edge modulation/trailing-edge modulation) technique, in which the PFC and PWM switches are synchronized so that the PFC switch turns off just as the boost switch turns on. This technique allows the PFC bulk capacitor to be smaller than normal because it does not power the output all by itself; the PFC inductor helps out. Typically, PFC/PWM controllers use TEM/TEM, which results in an additional step as well as a larger PFC bulk capacitor (Figure 11).

Figure 11a shows the energizing of the PFC inductor. Figure 11b shows the energy from the inductor transferring into the PFC bulk capacitor. When you close the PWM switch, the energy within the PFC bulk capacitor drives the load (Figure 11c). Every time this cycle repeats, the PFC bulk capacitor has to be fully charged because it is fully discharged when the PWM switch closes.

In LEM/TEM, the PFC and PWM switches are tied together such that, when one is opening, the other is closing, so when the PFC switch is open, the PWM switch is closed, and vice versa. Initially, when the PFC switch is closed, the PFC inductor is energized. Once the PWM switch is closed, both the output and the PFC bulk capacitor are energized. Figures 12a and 12b show that upon repetition of this cycle, the PFC bulk capacitor does not have to be as large as that in TEM/TEM because it does not power the output all by itself; the PFC inductor helps out.

The potential for reducing operating costs by minimizing wasted power explains why device-side PFC has become important in so many products' power systems and is expected to increase the PFC-controller market to $175 million in 2006. Many standards (for example, EN 61000-3-2) are driving PFs toward 1 and are requiring minimum total-harmonic distortion in systems that consume ever-smaller amounts of power.