Future Electronics — Maximizing the Efficiency of DC-DC Converters: How to Take Advantage of the Latest Topologies and Techniques

FTMSept2018-pg22-FE-Header
future-logo
By Robert Gabrysiak
Applications Engineer, Future Electronics (Poland)

Read this article to find out about:

  • The main characteristics of the most widely used DC-DC converter topologies
  • The performance benefits of zero-voltage switching and synchronous rectification
  • Components which enable designers to achieve efficient conversion of a rectified mains input to a point-of-load output as low as 2.5V in only two stages

For some years, the designers of medium- and high-power electronics systems have been under intense pressure to improve efficiency at full and light load, in order both to ensure compliance with increasingly tight limits to power consumption imposed by governments, and to help end users to reduce energy costs.

The industry’s responses to this pressure have included efforts to reduce or eliminate power losses in high-power DC-DC converters; and at the same time to extend the use of high-voltage DC in place of AC in power distribution systems, with the effect of reducing distribution losses.

Electronics OEMs in sectors such as telecoms and networking servers, battery-charging systems and renewable energy generation are also looking to improve efficiency by implementing direct conversion from a 48V DC distribution bus to Points-of-Load (PoL) operating at a voltage as low as 2.5V.

This article explains important power-conversion breakthroughs which have enabled the development of more efficient DC-DC converters, both for stepping down 400V inputs to a 48V output, and also for converting the 48V distribution bus to a PoL voltage. It also introduces two new off-the-shelf solutions that an OEM designer can implement quickly and easily.

The Choice of DC-DC Converter Topologies

The most common DC-DC converter types can be divided into two categories depending on how they transfer power: energy can go from the input through the magnetics to the load simultaneously, or the energy can be stored in the magnetics for later release to the load. These types are shown in Figure 1.

TopologyPower RangeTransformer UtilisationNumber of Active SwitchesVoltage Stress on
Active Switches
Type of Power Transfer
Buck (multi-phase)>1kWSingle ended1VINEnergy flow
Flyback<150WSingle ended1>VIN + n*VOUTEnergy storage
Forward50W – 200WSingle ended1>VIN x 2 (DMAX = 0.5)Energy flow
Active Clamp Forward (ACF)50W – 300WDouble ended2VIN/(1-D)Energy flow
Push-Pull (P-P)100W-500WDouble ended2>VIN x 2Energy flow
Half-Bridge100W – 1kWDouble ended2≥VIN/2Energy flow
Full-Bridge>1kWDouble ended4≥VINEnergy flow

Fig. 1: Common DC-DC converter topologies and the power ranges to which they are suited

Multi-phase buck controllers offer higher efficiency than single-phase converters because of their low transitional losses. They also feature a lower output-ripple voltage, better transient performance, and a lower ripple-current rating at the input capacitor.

A full-bridge converter configuration retains the voltage properties of the half-bridge topology, and the current properties of the push-pull topology. The push-pull, half-bridge and full-bridge configurations use smaller input filter and output inductors than single-ended converters use. They also provide up to 40% better transformer utilization, and balanced semiconductor power dissipation over the input-voltage range.

Different application requirements, then, will call for the use of different converter topologies. But across the board, the realization of high-efficiency DC-DC converter designs has been made possible by the development of two important techniques: Zero-Voltage Switching (ZVS), and synchronous rectification.

Benefits of Implementing Zero-Voltage Switching

ZVS is an extreme form of ‘soft switching’. In a hard-switching system during turn-on, each primary-side MOSFET in turn is exposed to a voltage equal to at least the supply voltage, and current tends to flow from drain to source, resulting in high turn-on losses.

In soft switching, the current at the turn-on point is oriented from source to drain, which discharges the MOSFET’s output capacitance before turning the device on, thus eliminating turn-on losses. It should be noted that ZVS operation affects turn-on losses only; the converter will still suffer from switching losses at turn-off, both due to current overlap and to charging of the output capacitor.

Why Synchronous Rectifiers are Replacing Diode Rectifiers

Alongside ZVS, synchronous rectification on the secondary side is also helping to improve converter performance in important ways. As shown in Figure 2, synchronous rectifiers may be used to replace diode rectifiers, with the drivers of the synchronous rectifiers behaving as diode rectifiers. This approach is more efficient because the product of the average current and on-resistance is much lower than the forward voltage of the diode. In a design using synchronous rectification, however, the MOSFET must be carefully controlled so that the timing of the MOSFET’s turn-on and operation is synchronized with the diode’s turn-on period.

FTMSept2018-pg23-FE-Fig-02
Fig. 2: Simplified circuit diagram of a phase-shifted full-bridge converter with synchronous rectification on the secondary side. (Image credit: Vishay Application Note 833)

New digital Implementation of Phase-Shifted Full Bridge

For high-power converter applications, as Figure 1 shows, the Phase-Shifted Full Bridge (PSFB) topology is the most suitable. When the PSFB topology is implemented with both ZVS and synchronous rectification, power-system designers can achieve excellent power density and efficiency. This is particularly important in DC power distribution applications, for instance when stepping down 400V DC to 48V to supply a distribution bus.

In the past, the implementation of such a circuit would have called for deep power-design expertise and the ability to create complex PWM waveforms and manage tight timing constraints. More recently, however, microcontroller manufacturers have introduced solutions which provide an easy-to-use platform for a fully digital implementation. Here, MCUs control the gate drivers of the MOSFETs in the full bridge, and for synchronous rectification. Digital control offers advantages such as programming flexibility, high integration, and the ability to program in-rush controls and soft-start operations.

Such an approach is demonstrated by STMicroelectronics in its STEVALISA172V2 board, which implements a 2kW AC-DC converter with one STM32F334 MCU for the AC-DC converter and power factor correction, and another to perform the power-control functions in a PSFB DC-DC converter providing an output of 48V or 52V, as shown in Figure 3. The STM32F334 MCU features a high-resolution timer (217ps maximum resolution) which enables very accurate current regulation in the PSFB circuit.

The control algorithm is based on a simple voltage loop realized with a traditional Proportional Integral (PI) regulator, see Figure 3. The algorithms used in the STM32F334 are explained in detail in Application Note 4856 from ST, which supports the STEVAL-ISA172V2 demonstration board.

FTMSept2018-pg23-FE-Fig-03
Fig. 3: The control loop in the STEVAL-ISA172V2 PSFB converter circuit. (Image credit: STMicroelectronics from AN4856)

Easy 48V-to-PoL Conversion

The STEVAL-ISA172V2 steps a rectified mains input of 400V DC to a distribution bus voltage of 48V. Typically, this distribution bus would be stepped down to an intermediate voltage of 12V or lower, before being stepped down again to PoL voltages of 5V or less. This reflects the ready availability of existing buck converter ICs, which work well over a narrow input-to-output voltage ratio of up to 6:1. But the performance of conventional hard-switching converter ICs falls below acceptable limits, in terms of both efficiency and output power, at step-down ratios above around 12:1. Direct conversion from 48V to a PoL voltage, which involves much wider ratios up to 36:1, is made possible through the use of a ZVS topology.

And looked at logically, it makes sense to eliminate one of the intermediate conversion stages, normally the 48V-to-12V step: in a typical two-stage conversion when both converters offer efficiency of 90%, the overall efficiency is 81% (0.9 x 0.9). So a single-stage conversion, 48V to the PoL voltage, which can achieve average efficiency higher than 81% is better than two stages each with high 90% efficiency.

A particularly efficient implementation of ZVS is enabled by a new buck-conversion topology developed by Vicor in its Cool-Power® ZVS series of regulator modules, as shown in Figure 4. The ZVS technique cuts turn-on switching losses and gate-driver losses, as well as eliminating FET body-diode conduction.

Part numberInput voltage [V]Output Voltage [V]Maximum Current [A]
PI3525-00-LGIZ48 (30 – 60)5.0 (4.0 – 6.5)20
PI3542-00-LGIZ48 (36 – 60)2.5 (2.2 – 3.0)10
PI3543-00-LGIZ48 (36 – 60)3.3 (2.6 – 3.6)10
PI3545-00-LGIZ48 (36 – 60)5.0 (4.0 – 5.5)10
PI3546-00-LGIZ48 (36 – 60)12 (6.5 – 14.0)9

Fig. 4: Specifications of the Vicor Cool-Power ZVS regulators for 48V-to-PoL conversion

The Vicor ZVS buck topology is in fact the same as a conventional synchronous buck regulator, except for the addition of a clamp switch across the output inductor, as shown in Figure 5: energy stored in the output inductor is directed so as to ensure that switching takes place under nominally zero-voltage conditions.

FTMSept2018-pg23-FE-Fig-05
Fig. 5: The Vicor ZVS buck-converter topology, showing the clamp switch across the output inductor. (Image credit: Vicor, ZVS white paper)

The high-side power MOSFET in the Vicor PI35xx parts always turns on at zero current and at nearly zero drain-source voltage. Power wastage and heat still arise from conduction losses, but switching losses are reduced by almost eliminating turn-on losses thanks to the ZVS operation.

With the addition of the clamp phase, there is no body diode conduction in the low-side MOSFET, hence the high reverse-recovery current prior to turning on the high-side MOSFET, a feature of conventional synchronous buck regulators, is not a feature of these Vicor parts. Each power MOSFET thus produces much less heat, eliminating the need for expensive and bulky heat-sinks, and helping to reduce the size and weight of the system, and its bill-of-materials cost. Peak efficiency is typically higher than 95% for all of the devices in the range.

Conclusion

This Vicor solution is provided as a fully integrated surface-mount module, making it quick and easy for the power-system designer to implement. Together with new, high-efficiency approaches to 400V-to-48V conversion such as that demonstrated in the STEVAL-ISA172V2 board, system designers can now more quickly and easily than ever before achieve two-stage conversion from the rectified mains to a PoL voltage.

This approach enables high voltage and low current to be distributed throughout a system, minimizing distribution losses while providing a low-voltage and high-current supply directly from a 48V input in a highly efficient manner. A space-saving solution, this approach also reduces development time since it eliminates one entire power stage from the power-system design.

Distribution at 48V DC requires less cabling, and reduces crowding, cost, weight and conduction losses, typically offering a 16x reduction in power losses and a 4x reduction in capacitor volume.

This combination of benefits suggests that 48V power distribution will rapidly spread from its stronghold today in telecoms and networking equipment to other market sectors including industrial and automotive systems.