Optimized Synchronous Rectification Drives Up DC-DC Efficiency

The desire to squeeze more data bits through smaller systems at faster rates has brought about a dramatic rise in power consumption. At the same time, there's been a proliferation of voltages that address the needs of microprocessors, memories, I/Os, and other functions in modern computers and communications equipment. With multimode configurations being the current the trend, power supplies must respond instantly to changing load requirements. The result is the rapid adoption of distributed power architectures, with dc-dc converters in close proximity to the load. These improvements are imposing new challenges in cost and efficiency, as well as power density, on dc-dc converters.

In response, manufacturers continue to miniaturize the converter size while boosting the supply's power density. Users are demanding higher and higher power-conversion efficiencies from tinier and tinier packages, even as output voltages drop below 3.3 V. In the mean time, output voltages drop to 2.0 V and below while output currents surge to tens of amperes, and the input voltage remains as high as 48 V. This makes those manufacturer's tasks even more arduous.

Power-supply designers are responding to these problems by slowly migrating to synchronous-rectification techniques. Of course, low-loss, MOSFET-based synchronous rectification has helped beat the limitations of conventional rectifiers employing Schottky diodes. Those dc-dc converters thereby achieved further improvements in efficiency. There's still ample room for improvement over the current approaches that are self-driven, employ some form of discrete control, and use paralleled MOSFETs for higher output currents.

The self-driven design brings its own set of problems, while the MOSFETs implemented are inappropriate for the application. Specifically, for higher output currents, paralleling power MOSFETs has resulted in an increased footprint, as well as higher I²R losses to degrade the performance of the synchronous rectifier. Higher input voltages and lower output voltages have brought about very low duty cycles, increasing switching losses and decreasing conversion efficiency. In fact, the ratio of V_OUT to V_IN dictates the duty cycle and, therefore, the timing of the MOSFETs. With lower duty cycles, the MOSFETs must transition or switch faster, so they end up with higher switching losses. Though more prevalent in synchronous buck, this phenomenon is less severe in isolated versions.

So the pressure to improve efficiency without hiking the solution's price tag is enormous. At power outputs of 100 W and above, every percent increase in conversion efficiency makes a difference. Present-day solutions afford nearly 85% efficiency, which is considered high for designs that step down voltages from 48 V to 3.3 V and below.

Boosting that efficiency number to 90% and over for high-power dc-dc converters isn't easy. That reality really puts the spotlight on International Rectifier Corp.'s claims of a ready-made solution. The company has generated a dedicated synchronous-rectification solution that promises to further augment the conversion efficiency by 4% to 5%, thereby pushing that bar above 90%.

To beat out problems with present methods and accomplish more in power-conversion efficiencies, the company has developed an optimized gate-drive scheme. Using stripe-planar process technology, IR specifically crafted low on-resistance HEXFET power MOSFETs that provide the best combination of on-resistance, gate charge, and die size for this application. According to the company, these elements make this a dedicated solution for synchronous rectification.

The current optimized solution is actually tailored for isolated systems employing single-ended forward-converter topology. But IR intends to extend the concept to other topologies like buck, boost, and multiphase, including multi-output configurations. The advantage of the isolated-converter topology is that it employs the transformer's turns ratio to balance the converter duty cycle. That ratio minimizes the switching losses when V_OUT drops significantly lower than V_IN.

The company states that the dedicated, high-speed CMOS controller chip, IR1175, implements a gate-drive technique to significantly lower losses in isolated topologies implementing synchronous rectification (Fig. 1). To alleviate some of the drawbacks of earlier schemes, the controller IC incorporates several key features. It can compensate for finite MOSFET switching times, so they switch on with the actual transition of the transformer instead of having to do it later. This is accomplished by using a modified phased-locked loop (PLL) with delay block to pre-fire the n-channel MOSFETs.

By providing necessary dead times/overlap between gate drives, the IC also minimizes the effect of stray inductance. It maintains a steady gate drive throughout both power-transfer and reset cycles, regardless of transformer output status. Those gate drives are compatible with logic-level power MOSFETs, independent of transformer output levels.

Finally, the controller follows the transitions of the transformer and maintains the duty-cycle requirements set by the system's pulse-width-modulated controller. Turning the po-wer MOSFETs on or off before the transfor-mer's transitions minimizes body-drain conduction and cuts associated losses, according to Edgar Abdoulin, power systems design engineer.

He continues to state that the turn-on and turn-off lead time is adjustable and can be set to accommodate a variety of power-MOSFET sizes and circuit conditions. The IR1175 also provides gate-drive overlap/dead-time control via external components. These features further reduce diode conduction by nulling the effects of secondary-loop and device package inductance. Above all, Abdoulin emphasizes that the IR1175 is a standalone circuit that has no ties to the primary. The controller derives its operating power directly from the secondary winding of the transformer.

To enable the controller to work in noisy environments, the IC employs Schmitt-trigger inputs incorporating double pulse-suppression techniques. The device also boasts high-current drive capability and high-speed operation. Plus, it's adaptable to multiple topologies. As per the data sheets, the IR1175 offers up to 2-A output drive current and 2-MHz switching capability. Abdoulin cautions, however, that the controller is best suited for designs that operate below 500 kHz and require less than 2-A drive current. For very high current requirements, he recommends external gate drivers.

Selecting the right MOSFETs from the maze out there is equally challenging. To more easily choose the most appropriate parts that will work efficiently with the controller, IR has unveiled optimized MOSFET families. Using its stripe-planar process, the manufacturer developed application-specific HEXFETs for both the primary and secondary side of the transformer.

At the primary end, for instance, 200-V and 150-V HEXFET power MOSFETs have been crafted to offer an optimal combination of on-resistance, gate charge, and die size. The technology applied gives the best tradeoff between the switching and conduction losses, says the company. Plus, by packing these devices in miniature surface-mountable D2PAK and DPAK packages, they offer power-density and cost-per-watt benefits.

Initially, four parts have been readied for the primary side. These include the 200-V, 30-A IRFS30N20D, the 200-V, 13-A IRFR13N20D, the 150-V, 41-A IRFS41N15D, and the 150-V, 18-A IRFR18N15D. For the secondary side, the supplier released three optimized devices that provide a combination of on-resistance and gate charge based on the output power rating and circuit topology. The two 30-V power MOSFETs include IRFBL3703 and IRF7455. The 20-V version is called the IRF7456 (see the table). Because it comes housed in a Super-D2PAK package, the IRFBL3703 is able to target high-current applications in which space is of utmost importance.

To demonstrate the impact of the dedicated solution, the company developed an experimental, single-ended forward converter with synchronous-rectifier control. It uses IR1175 and optimized primary and secondary power MOSFETs (Fig. 2). Compared to the self-driven, cross-coupled design, this solution shows remarkable improvement in the overall efficiency of the dc-dc converter. Internal tests indicate that it achieves about 88.4% efficiency for a 200-W brick (Fig. 3). IR engineers believe that further matching the primary- and secondary-side MOSFETs closer to the dedicated controller, as well as tweaking the design for lower losses due to the magnetics, will bring another 1% to 2% enhancement.

Meanwhile, the company continues to refine the synchronous-rectification controller to completely match the device with the MOSFETs. The controller line might be expanded to extend the solution to other topologies, including non-isolated versions. Outputs could eventually drop down to 1.8 V and below.

International Rectifier Corp., 233 Kansas St., El Segundo, CA 90245; (310) 252-7105; www.irf.com