Is it really off?

Addressing the poor efficiency of power circuits in standby mode

Mats Nilsson, Field Applications Engineer, Future Electronics (Sweden)

READ THIS TO FIND OUT ABOUT: Regulations regarding standby power Power design for binary low-load/high-load operation Pulse-skipping methods to boost low-load efficiency

With dramatic growth in the use of electronic equipment in the home and at work, the cumulative power consumption of electronic products in standby mode has increased hugely in recent years. Although eco-design regulations are now placing pressure on manufacturers to slash standby power consumption, it is a difficult technical challenge to design a power circuit that achieves high efficiency when the device is in standby mode. Mats Nilsson, Field Applications Engineer, Future Electronics (Sweden), examines effective techniques to reduce the amount of energy lost in standby mode.

The typical approach to power supplies taken by engineering teams today is to optimise the design for efficiency when the device is operating normally. Unfortunately, the same circuit that operates efficiently at full load will not normally be efficient in the low-load, standby state.

The emphasis on improving efficiency at full load made sense in the past: after all, the biggest energy savings can be made when energy consumption is at its peak. As a by-product, improvements in energy efficiency at full load also reduce the amount of heat to be dissipated. If the system is designed to manage the heat generated at full load, it will certainly be able to handle the heat generated in standby. There was therefore less impetus to improve the power efficiency of standby operation in the past.

But new energy-efficiency regulations in various parts of the world are now raising the stakes for designers of power circuits; it is no longer safe to ignore the efficiency of standby power supplies. The most important of these new regulations are:

• Energy Star – a joint programme of the US Environmental Protection Agency and the US Department of Energy, giving a consumer-friendly rating to products and practices.
• 80 PLUS – an incentive programme funded by North American electric utility companies promoting more energy-efficient power supplies for desktop computers and servers. There are also rumours that the European Union will soon issue a directive that is more stringent than 80 PLUS.
• China Standard Certification Centre – a voluntary programme with the aim of stimulating manufacturers to produce more resource-efficient products and helping consumers to make more sustainable purchasing decisions.
• Blue Angel – a programme in Germany, which gives practical guidance to consumers to help them in product selection.

For engineers charged with complying with these kinds of regulations, however, there is no single solution or design that optimises the efficiency of standby power supplies. The correct engineering approach depends on the requirements of the application.

 

 

Binary low-/high-load applications

In some applications, there is either a constant high load or a constant low load, however the load does not vary on a sliding scale between high and low. A good example would be a flat-screen TV, which is either fully on or in standby mode.

Decisions over power-supply architectures can often be a matter of balancing complex factors or making nuanced judgements. In the above case, however, the choice is very simple to make: such applications should have two separate power supplies, one optimised for the high load and the other optimised for the low load (see Figure 1).

The designer has broadly two categories of component to choose from for the smaller, low-load power supply: one option is that chip manufacturers have in recent years developed new families of devices that feature an integrated MOSFET and are able to operate efficiently at very low power. Good examples are the VIPer family from STMicroelectronics, the NCP10xx family from ON Semiconductor, the GreenChip™ range from NXP Semiconductor and the Green Mode series from Fairchild Semiconductor.

The other option is to avoid using integrated circuits and instead use a rechargeable battery. Whichever the designer chooses, this architecture avoids operating a large power supply in low-load conditions. The high-load power supply will contain large components such as MOSFETs with relatively high gate charge, large magnetising transformers and energy-consuming EMI filters, and therefore has poor efficiency when not operating at full power.

This dual-supply architecture does, however, raise the issue of how to minimise losses arising from switching between the two supplies. One well-known technique is resonant switching, in which a low voltage across two MOSFETs turns a supply on or off by resonant ringing. But there is a problem with true resonant converters: the PWM frequency has to be fixed in order for the switching to take place when the voltage across the MOSFETs is at a minimum. This means that output voltage regulation can only be done by phase shifting in bridge converters.

The limitations of resonant switching can, however, be overcome in smaller flyback power supplies with a technique called quasi-resonant switching, in which the normal switch-off ringing in a flyback supply is used during switch-on. This is illustrated in Figure 2.

The red line shows how the switch turns on during the first ringing period, and the blue line shows the switch on during the second ringing period; both events occur with the voltage at its low point. For this reason, this method is sometimes called valley switching.

There are various ways to implement quasi-resonant switching so that switching takes place at a defined point in time. For example, Fairchild Semiconductor’s FSQ0565R, a member of its Green Mode family of power switches, has a time window in which switching takes place. If the first valley occurs too early – outside the time window – the switch will take place at the second or third valley. If the valley occurs too late the FSQ0565R implements a hard switch.

By contrast, ON Semiconductor’s NCP1205 SMPS controller accomplishes quasi-resonant operation by using the completion of transformer core demagnetisation as a trigger for initiating a new cycle. This circuit is able to increase frequency when powering heavy loads. Hence, unlike resonant switching, the frequency in quasi-resonant switching is variable. This, combined with the low voltage over the switch, serves to reduce EMI emissions, and therefore enable the use of smaller, more efficient EMI filters.

 

 

Increasing the efficiency of single power supplies at low load

If the flat-screen TV is a good example of an application that is best served by two power supplies, there are many others in which the load varies on a sliding scale from high to low, and in which a single power supply is the most appropriate power architecture. In this case, the design engineer will need to adopt techniques that boost the efficiency of the power supply at low load.

One such method is pulse-skipping regulation. It can be implemented in various ways: one is to halve the switching frequency when the current falls below a fixed threshold.

Another method is burst-mode regulation, in which the normal switching frequency is maintained, but several switching cycles are skipped periodically. The VIPer17 family from STMicroelectronics is suitable for use in implementing this technique.

A third method of implementing pulse skipping is hysteretic-mode regulation, in which the switching frequency changes according to the load: the greater the load, the higher the frequency. This can be implemented using a comparator to compare the output voltage with a reference in order to control the switch (see Figure 3).

Every kind of converter implementing a reduced-pulse mode saves energy by lowering the MOSFET switching frequency. The faster a MOSFET is switched, the more often the MOSFET gate charge has to be cycled, and thus the more energy it consumes.

 

 

Other design considerations

Besides implementing an efficient and suitable regulation technique, the designer can also address other parts of the power-supply circuit to achieve further efficiency gains.

For instance, in a transformer it is important to ensure the windings are optimised for minimum leakage inductance. Using the right core material, and running several wires in parallel to reduce skin and proximity effects can also be beneficial. The use of a good software design tool is a valuable investment for this purpose.

Start-up circuits can often be a source of power loss in low-power conditions, especially if there is a resistor connected directly to the rectified mains voltage; it is important to supply sufficient current to feed the gate on the switching transistor. Often an auxiliary winding is needed on the transformer to supply the primary-side circuits.

Careful design of filters can also help to increase efficiency. In most cases, filters on both the mains and secondary side are required. To minimise losses in the filter, it is important to use low-resistance inductors.

Finally, the switching frequency and board layout should be considered carefully. While a high switching frequency enables the use of smaller magnetic and capacitive components, it also increases power losses in all other components. In general, a lower switching frequency will be more efficient.

 

Conclusion

With the use of standard components from manufacturers such as STMicroelectronics, ON Semiconductor, NXP, International Rectifier and Fairchild Semiconductor, it is now possible to design circuits that achieve better efficiency than ever before in standby power supplies. This is true whether the application calls for separate power supplies for full power and standby power, or whether there is a single power supply.