Maximising AC/DC efficiency from full-load to no-load

While maximising power efficiency at full-load is a top priority in the design of an AC/DC supply, standby power considerations and new power-efficiency standards loom larger. Thus, apart from the general subject of “high efficiency,” designers are looking harder at ways to maximize energy-savings from end to end. Indeed, saving milliwatts is of particular concern with designs that use AC power adapters, which are proliferating worldwide.

Quasi-resonant control, valley-voltage switching, and multi-mode operation (i.e., pulse-skipping) offer one solution. In this article, we’ll summarize the techniques being used in some of today’s green-type IC controllers to minimize energy losses across the converter’s entire load range.

Limiting standby power
In a design philosophy that embraces smart electronics and “instant on” response, today’s AC/DC power converter often spends a good amount of its time in the standby mode, where there’s always some sort of power drain. It’s basically the same problem whether we’re talking about a remote-controlled TV or video equipment, an external low-voltage power supply for cordless telephones or wireless routers, office equipment (copiers and printers), or devices such as battery chargers for laptop computers. The actual power drawn in the standby mode of individual converters is small, typically 0.3 to 20 watts. But however small the standby power, the sum total becomes large when you multiply it by the number of consumer, commercial, and industrial systems in use.

Indeed, standby power accounts for perhaps 10 per cent of the electricity used in homes and offices in the European Union, and about four per cent of the total electricity used in the United States. Developing standards such as Energy Star focus on energy-savings at no-load and light-load conditions, higher efficiency in normal operation, less total harmonic distortion (THD), and close to unity power factor (PF). The Table summarizes the Energy Star standards for external single-voltage AC/DC and AC/AC power supplies.

Meeting the standards
How do system designers meet Energy Star and other developing international standards? They apply active clamp and reset techniques, transition-mode and interleaved multi-phase PFCs, pulse-skipping, quasi-resonant control and valley-voltage switching. Flyback converters with quasi-resonant control, valley-voltage switching and pulse-skipping offer one of the best solutions.

Flyback converters, widely used in consumer applications, are low cost, easy to control, and support multiple output rails (see Fig. 1, in this case using the UCC28600 quasi-resonant chip). Quasi-resonant control facilitates the use of soft-switching, which improves efficiency and energy-savings. In quasi-resonant operation, the primary main switch has a much lower turn-on voltage. The energy that charges the switch capacitance when it’s in the off-state is regenerated to the source.

In contrast, continuous and discontinuous current mode (CCM and DCM) operations in hard-switching topologies suffer high turn-on losses. To achieve better energy-savings across the entire load range, the flyback converter can be operated in the frequency foldback mode (FFM) and the pulses kipping mode, depending on load conditions. FFM circuitry cuts back on the switching frequency as the load becomes less, thus lowering switching losses. When the load becomes very light, the hysteretic mode (also known as green-mode or burst operation) comes into play to initiate pulse-skipping. The pulse-skipping reduces switching losses and achieves better energy-savings for lightload and no-load conditions.

For applications with frontend PFC pre-regulators, additional energy-savings can be obtained by turning off the PFC operation at very light load.

The circuits
Quasi-resonant control describes a flyback converter operated in critical conduction mode (CrCM) with zero voltage switching (ZVS), or so-called valley-voltage switching (VVS). The ZVS/VVS operation results from a LC resonance formed by the primary winding inductance of the flyback transformer and the total equivalent capacitance across the primary main MOSFET switch (CDS). The voltage across the MOSFET drops during the resonant switching process. The flyback controller detects the drop and turns on the primary switch at valley points. Valley-voltage switching must meet two conditions. The first is:

Vin≤ N (Vout + VD )

where N is the transformer turns ratio. In this situation, the reflected secondary voltage is high enough to force the primary VDS voltage to zero. As a result, the primary side MOSFET can be turned on with zero voltage across it. Secondly,

Vin > N(Vout + VD )

Under this condition, the reflected secondary voltage is not able to steer the VDS voltage to zero. Instead, we have a “voltage valley.” Figure 2 shows typical VVS operation from quasi-resonant flyback converters. The valley voltage would be extended to zero-valley if the first equation is met. Then we achieve zerovoltage switching.

ZVS/VVS brings high energysavings and an improvement in efficiency. For a given capacitance, the switching power, Psw, is determined by the voltage across the capacitor, CDS as well as the switching frequency, fs:

Psw = 0.5 CDS VDS2 fs

A flyback converter with hard-switching turns on the switch at a high voltage, which results in high switching-power. The energy stored in the capacitor CDS is dissipated through the MOSFET channel resistance during the next switching cycle, thus manifesting as a switching-power loss. Such a power loss is especially significant in off line AC/DC applications where a high DC-link voltage results from the rectified 85-285 VAC line voltage.

Conversely, the same flyback converter, if operated in the quasi-resonant mode with valley-voltage switching, will turn on the switch at a lower voltage. The voltage is reduced through LC resonance as the energy stored in the capacitor is discharged and recycled back to the DC link capacitor, CBLK, instead of being dissipated through the MOSFET channel resistance.

In general flyback operation, quasi-resonant control from a small load percentage to full rated load implies multi-mode operation for best efficiency. That is, we subdivide converter operation into two modes: normal quasi-resonant mode with variable on-time control; and constant on-time control, the aforementioned frequency foldback mode (FFM). For example, a quasi-resonant controller may be designed for operation from 15 per cent to 50 per cent load during which time it’s in the FFM mode. The frequency decreases as the load becomes less and so switching power losses are reduced further. From 50 per cent to full-rated load, the controller cuts down on its frequency as loading increases. Usually the switching frequency is clamped to below 150kHz to minimize EMI and meet EMI requirements.

Pulse-skipping, also defined as green-mode or burst operation, offers the best energy-savings at ultra-light loads. At such load levels, it’s easier to maintain output voltage regulation. So there’s switching only when the output voltage is moving out of regulation. Extra switching actions just contribute to energy-waste. For example, energy is wasted in the dissipative snubber circuit every switching cycle. Such energy losses are eliminated when we apply pulse-skipping.

Pulse-skipping initiates switching only if the output voltage drops below a certain threshold value. During this time, the controller (on the primary side) applies a pulse packet to the transformer to bring up the output voltage to the upper limit of the hysteretic window to keep the output in regulation. Then the switching circuitry is disabled. This correction circuitry comes back into play when the output voltage again approaches the lower limit of the hysteretic window.

Turning off PFC at light-load saves energy
Power-factor-correction (PFC) brings no real benefit at light load. Essentially, all the circuitry does is consume power. A properly configured flyback quasiresonant controller may contain a dedicated pin to conveniently implement such function and automatically turn off PFC circuitry at a predetermined load condition. By adding minor external circuitry (a diode and a resistor, for example Ds and Rs as shown in Fig. 1), the designer can use the status pin as an indicator to reduce the primary peak-current. This design technique helps to reduce harmonic power at light loads, and thus power losses are less. In addition, we can reduce audible noise.

In summary, the flyback converter, using quasi-resonant control and pulse-skipping techniques, will maintain high efficiency over the entire load range. Figures 3 and 4 present typical test results from a 65-watt flyback converter.

Figure 3 shows typical efficiency from a quasi-resonant flyback converter; Fig. 4 shows how pulse-skipping can minimise losses at standby power levels.