Highly efficient ATX power supply solution

As personal computers (PCs) are increasingly used, their energy consumption is also increasing. For desktop computers, ATX power supplies are usually used. In the past, typical ATX power supplies have always used traditional forward topology (1 or 2 switches), with an energy efficiency of about 70%. In recent years, with the increasing pressure on energy conservation and environmental protection, the need to improve the energy efficiency of ATX power supplies has become more and more urgent. To this end, many government agencies or industry organizations in the world have formulated corresponding computer power supply specifications and standards, hoping to improve the efficiency of power use, reduce power consumption as much as possible, and avoid unnecessary power waste. The topology and efficiency development trend of computer power supplies are shown in Figure 1.

For example, Energy Star, an energy-saving project jointly promoted by the US government and the industry, has issued the Energy Star Computer Specification 4.0. This specification includes an energy efficiency requirement of more than 80% for desktop computer power supplies, and it came into effect on July 20, 2007. According to this requirement, the energy efficiency of computer power supplies under 20% light load, 50% typical load and 100% full load conditions must be higher than 80%, and its power factor PF must be higher than 0.9.

In addition, the industry has also put forward newer and higher energy-saving requirements for computer power supplies. For example, the Computing Industry Climate Change Initiative (CSCI) proposed that computer power supplies under 20%, 50% and 100% load conditions must not only reach 80% in July 2007, but also meet higher requirements in subsequent periods, as shown in Table 1.

Sources of power loss in computer power supplies and high-efficiency design strategies

To improve the energy efficiency of computer ATX power supplies and meet the increasingly high energy-saving standards, it is very important to analyze the sources of power loss and take targeted measures to reduce energy consumption. Common ATX power supplies usually include many components such as EMI filters, rectifiers, PFC controllers, power switches, transformers, and switching power supply controllers. Figure 2 is a schematic diagram of the structure of an ATX switching power supply.

Assume that the output power of a computer power supply is 300W, the power supply efficiency is 75%, and its total power loss is 100W. According to calculations, the loss of the power factor correction (PFC) section is about 40W, accounting for 40% of the total loss; and the loss of the switching power supply section is about 60W, accounting for 60% of the total loss.

If the energy efficiency of the power supply is to be improved, it should be considered in different power sections, the number of power sections should be minimized, and the energy efficiency of each section (such as PFC section, switching power supply section, etc.) should be improved. In addition, other factors need to be considered, such as the limitations of different topologies, the complexity of the design, the improvement of energy efficiency at light loads, and the total cost of the power supply solution. For the aforementioned 300W


power

supply, assuming that the goal of increasing the energy efficiency from 75% to 82% is set, and the power loss is reduced from 100W to 66W, the power factor of the PFC section can be set from 90% to 93%, and the corresponding power loss is reduced from 40W to 25W, while the energy efficiency of the switching power section is increased from 83% to 88%, and the power loss is reduced from 60W to 40W.

Among them, for the PFC section, to achieve the corresponding energy efficiency improvement goals, it is necessary to first select the appropriate working mode of the PFC controller, such as continuous conduction mode (CCM) and critical conduction mode (CRM). For both CCM and CRM applications, ON Semiconductor can provide solutions with a power factor higher than 93%, such as NCP1606 and NCP1654, which exceed the requirements of many regulations.

For CCM mode, to achieve higher energy efficiency, the following strategies can be adopted:
(1) Optimize switch selection (when light load, switching loss is dominant, it is more recommended to sacrifice the on-resistance Rds-on to obtain faster switching speed);
(2) Use soft recovery boost diode;
(3) Select the right size of inductor to reduce the copper wire loss in the inductor (smaller core loss).

ON Semiconductor's NCP1654 is a PFC controller designed for CCM mode. It has the characteristics of fast transient response, very few peripheral components, extremely low startup current (<7.5μA), extremely low shutdown current (<400μA), low operating power consumption, and many safety protection features, such as inrush current detection, overvoltage protection, undervoltage detection for open-loop detection, soft start, accurate overcurrent limit, true overload limit, etc. It integrates all the features required to build a compact and stable PFC stage, and is very suitable for system applications with high requirements for cost performance, reliability and high power factor. Figure 3(a) is a schematic diagram of the energy efficiency of NCP1654 in a 300W computer power supply application, showing that its maximum energy efficiency is close to 96%.

For CRM or discontinuous conduction mode (DCM), the following strategies are recommended to achieve higher energy efficiency:
(1) Optimize the inductor core to reduce core loss and high-frequency winding loss;
(2) Select a lower Rds-on switch;
(3) Do not pay too much attention to the choice of boost diode.

ON Semiconductor's NCP1606 is a cost-effective PFC controller with an embedded CRM mechanism. Its main features include no input voltage sensing, extremely low startup current consumption (<40μA), and low typical operating current (2.1mA). In terms of safety protection, it also provides programmable overvoltage protection, undervoltage protection, accurate and programmable on-time limit and overcurrent limit. Figure 3 (b) shows the energy efficiency of NCP1606 in a 240W computer power supply application.


Energy efficiency improvement of the switching power supply stage and comparison of different topologies

As mentioned above, it is assumed that a 300W power supply needs to achieve 88% energy efficiency in the DC-DC switching power supply stage. To achieve this goal, we can start from many aspects, such as reducing primary side losses, reducing switching losses, reducing secondary side losses and reducing core losses.

Taking reducing primary side losses as an example, it can be achieved by reducing on-resistance and/or reducing primary side peak current and root mean square (RMS) current. To reduce switching losses, soft switching technology can be considered. In terms of reducing secondary side losses, the rectifier voltage drop can be reduced (using diodes or FET rectifiers with low forward voltage Vf). As for reducing core losses, it can be achieved by using better core materials.

In the switching power supply segment, ON Semiconductor provides a series of power ICs that can be used to improve power supply efficiency, such as NCP1562, NCP1395/1396, NCP1027/1028, etc. for the primary side, and NCP1582/1583, NCP5425/5427, NCP4331 and NCP4350, etc. for the secondary side.

For DC-DC conversion on the primary side, different topologies can be used, such as dual-switch forward, active clamp forward (ACF), and dual inductor plus single capacitor (LLC). The dual-switch forward is a traditional topology, where components are easily available and MOSFET stress is low. However, it also has its disadvantages, namely, high switching losses and difficulty in applying synchronous rectification. In comparison, the active clamp forward topology (as shown in Figure 4) has lower switching losses and can perform self-driven synchronous rectification. However, the rated voltage of the primary switch is higher in this structure.

ON Semiconductor's NCP1562 is a voltage-mode controller with an active clamp topology designed for DC-DC converter applications that require high energy efficiency and a low number of components. This controller integrates two in-phase outputs with overlapping delay functions to prevent simultaneous conduction and facilitate soft switching. The main output of this controller is designed to drive the primary MOSFET of the forward converter, and the second output is designed to drive the active clamp circuit, the synchronous rectifier on the secondary side, or the asymmetric half-bridge circuit. The NCP1562 series integrates many features, such as maximum duty cycle limitation, undervoltage detection, and overcurrent threshold, thereby reducing the number of components and reducing the system size. NCP1562 includes two models, NCP1562A and NCP1562B. The former has a current limit voltage threshold (VILIM) of 0.2V and the latter is 0.5V. Two features of NCP1562 are soft stop and a cycle-by-cycle current limit detector with a time threshold. The technology used in this device and its many features can help it reduce the power loss on the primary side and improve the efficiency of the switching power supply.

The NCP1395/NCP1396 is a dual inductor plus single capacitor (LLC) half-bridge resonant converter.

Taking the NCP1396 as an example, this high-performance resonant mode controller provides all the performance required for a reliable and rugged power supply. Its unique architecture includes a 1.0MHz voltage-controlled oscillator and protection functions with multiple reaction times, making the converter safer without increasing the complexity of the circuit. This LLC half-bridge resonant converter provides higher energy efficiency. In a smaller input and load range, especially in high output voltage applications, the half-bridge resonant converter is a better choice. It has low switching losses, does not require an output inductor, and is a low-component topology. The converter also has the advantages of lower primary conversion voltage stress, resonant operation to minimize switching losses, constant duty cycle operation, and simplified high-side switch drive. Its structural diagram is shown in Figure 5.

Among the power ICs available on the primary side, the NCP1027/NCP1028 are used as standby controllers. They are optimized for ATX power supplies and integrate high-voltage MOSFETs and startup current sources. Under low peak current conditions, skip cycle operation is performed to help reduce energy consumption and improve energy efficiency.

On the secondary side, the NCP158x is a low-cost step-down PWM controller designed to work on a 5V or 12V power supply. This device can generate an output voltage below 0.8V, which is suitable for today's applications with a voltage below 1V. The NCP5425 is a highly flexible dual-step-down controller. This device can work on a single 4.6V to 13.2V power supply and supports a single two-phase or two single-phase outputs. The NCP4331 is a synchronous step-down controller for high-efficiency secondary regulation. It packages two MOSFET drivers together and uses it as a companion chip. The device keeps power dissipation to a minimum while reducing the number of external components. The NCP4350 is a power monitoring IC that provides the necessary functions for monitoring and controlling multiple output power supplies. The device can monitor +3.3Vdc, +5Vdc and +12Vdc (A and B) outputs. Figure 6 shows the block diagram of the 305W ATX power supply reference design based on ON Semiconductor power IC.

Design considerations for improving energy efficiency under light load conditions

For computer power supplies, in addition to considering energy efficiency under full load, typical load and standby conditions, energy efficiency improvement under light load conditions has also attracted greater attention in the industry. There are many tips or ideas to follow in improving the light load energy efficiency of ATX power supplies.

For example, field effect transistors (MOSFETs) with smaller capacitance can be selected to reduce switching losses (trade-off between low on-resistance Rds-on). In addition, switching losses can also be reduced by adopting soft switching operation mode.

When reducing light load losses, even if the loss can only be reduced by 0.1W, it should not be ignored; taking a 240W power supply as an example, under 20% light load conditions, reducing power loss by 0.6W can produce a 1% energy efficiency improvement.

Not only that, you can also try to reduce some unnecessary components. For example, you can eliminate the startup resistor and leakage (preload) resistor, eliminate unnecessary snubbers, and eliminate unnecessary Zener diodes because Zener diodes need to consume bias current. As for the bias current, you can also use integrated circuits with smaller bias currents. The application of all these techniques will help to improve the energy efficiency of computer ATX power supplies under light load conditions.

The energy efficiency challenges faced by computer power supplies are becoming increasingly severe and more urgent. To meet these challenges, a system-level approach can be taken without adding too much cost. For example, the computer power supply can be divided into different power segments for consideration, and the loss sources of each power segment can be analyzed clearly. By using more advanced power ICs or devices and other design techniques, the power loss of each segment can be reduced in a targeted manner, thereby improving the overall energy efficiency of the power supply. In view of the actual application of computers, it is also very important to improve its energy efficiency under light load conditions. It is necessary to reduce switching losses through various ways to improve light load energy efficiency. As a leading global power supply solution provider, ON Semiconductor provides corresponding solutions for different segments or applications of computer power supplies to facilitate customers to develop high-efficiency computer power supplies.