Magazine Homepage
“It’s the skill of the designer that separates a very good power supply from an average one.” Gary Bocock, Technical Director, XP Power

Fig. 1. The Architecture of a power supply optimised for small size.

Fig. 2. XP Power’s EMA212 power supply.

 
Ac/dc power supplies

Like most things in electronics, there’s a trend towards ever-smaller power supplies. These leave room in the system for added functions and processing power and they often have to fit physical formats that are already in use to avoid system re-design costs.

Smaller designs are being achieved through combining a range of techniques, rather than through dramatic technology breakthroughs. It’s the skill of the designer that separates a very good power supply from an average one.

In most applications the power supply should be as small as possible and generate as little heat as possible.

For a 100W to 200W power supply, efficiency of 90percent is achievable. A 1percent efficiency improvement represents 10percent less heat dissipation at the upper end of the range and this can make a significant difference to the degree of cooling needed for the power supply.

Cost, both in terms of bill of materials and manufacturing complexity, is always of prime concern. Keeping the design simple helps in this respect.

Finally, functionality should not be compromised. Control and alarm signals, current sharing with similar units, and the ability of the power supply to maintain its performance over a wide range of ac input conditions are all important (Fig.1).

u Input filter. A two-stage filter design using high permeability cores will minimise size while providing high common mode and differential noise reduction. Stacking some components vertically can save board space and improve cooling.

u Power factor correction circuit (PFC). The use of silicon carbide diodes has become economically feasible in the last two years as component prices have fallen. Their reverse current characteristics mean that they do not require a snubber circuit, saving on five or six components.

Furthermore, they contribute to a 1 per cent typical efficiency boost. Using a stepped gap inductor provides high inductance at high input line and supports maximum flux density at low line.

Using continuous conduction mode (CCM) operation throughout the input range keeps the peak switching current and input filter requirements to a minimum.

u Main converter. Here, a resonant topology can virtually eliminate switching losses. This not only improves power supply efficiency but also enables smaller heat sinks to be used. In fact, compact ceramic heatsinks can sometimes be used for power transistors, rather than metal ones. Their advantages include a reduction in noise and consequently simplified filtering. This is because the heatsinks do not have capacitive coupling with the drain connections of the switching MOSFETS. In addition, smaller creepage distances, compared with those needed for metal heatsinks, can be used. This gives further savings in board space.

u Output rectifier. Opt for synchronous rectification here, using switched MOSFETS rather than output rectifier diodes. This improves efficiency through a significant reduction in power dissipation. For example, at 20Amps a diode with 0.5V forward voltage gives a power dissipation of 10W. Using a MOSFET with an ‘ON’ resistance of, say, 14mOhms at 100°C dissipates just 5.6W – a 44percent improvement. Once again, ceramic substrates can replace conventional heatsinks.

u Control circuit. Semiconductor manufacturers have been developing increasingly integrated control circuits for power supplies in recent times. This means savings in component count, manufacturing costs and board space, even where the integrated circuits themselves may be more expensive than a discrete component approach. One example is the IR1150 – a PFC chip that operates as a one-cycle control (OCC) device, which allows major reductions in component count without reducing power system performance. Similar, application-specific chips can provide main converter voltage control plus over-current protection, over-voltage protection and over-temperature protection. They can also control the output rectifier switching. Other desirable control options for increased application flexibility include power sharing with synchronous monotonic start-up, an inhibit circuit to shut down the power supply via logic control, a ‘power good’ signal, and the control functionality needed for a standby converter. The standby converter provides an independent 5V output whenever ac power is present.

Today’s best-in-class ac/dc switchers are typified by XP Power’s EMA212 power supply, shown in Fig.2. Using some the techniques described above, this packs 212.5W output from a 3x5inch footprint with a maximum height of 1.34inches.

That’s a power density of 10.55 W per cubic inch in an industry standard footprint that fits within a 1U high enclosure.

It delivers 200W from its main 12V or 48V output, plus 12V at 1A for driving fans and a 5V standby output. The unit needs just 12CFM of forced-air cooling, which is easily achievable using standard 40x40mm fans.

Forced-air cooling

Forced-air cooling is now the norm in many communications systems and 12CFM is easily achievable without complex mechanical arrangements. Finally, it achieves an efficiency of 91percent at full rated load.

Designing the optimum power supply for the space available is a key skill, and one that is being stretched every day as equipment gets smaller and more powerful and power densities increase.

Meeting the power, thermal and performance requirements within the cost budget are key capabilities for today’s power supply designer.

Gary Bocock is Technical Director at XP Power, Reading, UK. For more information, visit www.xppower.com