problem is that not all solutions are qualified, and engineers who design PC power supplies need to compare different designs on the market to select the most energy-efficient solution. In particular, the fact that a power supply design demonstrates compliance with a given standard does not automatically mean that the power supply design will provide the target energy efficiency that designers expect when the PC is put into actual use. Therefore, it is necessary for engineers to carefully review the data sheets and the underlying technology and fully understand how the supplier achieves the claimed energy efficiency.
Data Sheets and Standards Versus the Real World
To illustrate the challenge, let's consider an engineer tasked with designing a desktop PC with an ATX power supply that meets the 80 PLUS Silver performance specification. The 80 PLUS program is funded by North American electric utilities to encourage the use of more efficient power supplies in desktop PCs and servers. To achieve this goal, the 80 PLUS Silver specification requires that multi-output power supplies in computers and servers have a minimum efficiency of 85% at 20% and 100% of rated output power, and a minimum efficiency of 88% at 50% of rated output power.
Table 1: 80 PLUS multi-output desktop computer power supply efficiency targets |
This wide range of input voltage conditions raises a series of challenges that cannot be underestimated, especially the impact of conduction losses in the input stage. At the low end of the input voltage range, these losses can seriously affect the overall efficiency of the power supply, making it difficult to maintain the efficiency rating to meet the relevant standards.
Engineers must also consider the "actual power difference" problem. An ATX power supply may be rated for 255 W operation, but in real-world operation, it will only be operated at nearly full load in highly processor-intensive applications. In reality, most desktop PCs are rarely operated under full load, and it is estimated that more than two-thirds of the time are typically operated under "light load" conditions. Under light load conditions, the losses associated with "housekeeping" work account for a larger percentage of the total losses, with a corresponding negative impact on operating efficiency.
Cable Length Issues
Another factor that engineers should be aware of when considering manufacturers' claimed energy efficiency is the length of the cables used during the compliance testing and certification process. Cable length is not specified in certification requirements, resulting in some manufacturers claiming efficiency levels measured directly at the output of the power supply (or with unrealistically short cables). The
form factor and design of real-world desktop PCs means that the cable length between the power supply and the point where power is provided is typically measured at about 16 inches (41 cm), and is therefore a factor in the total losses. For this reason, lab simulations of real-world energy efficiency require testing with cables of similar lengths.
Matching real-world operation with energy efficiency claims
ON Semiconductor has taken these issues into consideration, so it has developed a public high-efficiency ATX power supply reference design to provide desktop PC and server manufacturers with real-world energy efficiency consistent with 80 PLUS Silver, Energy Star 5.0 and CSCI Phase III energy efficiency standards under different operating conditions. At the same time, the design must also meet IEC61000-3-2 requirements related to power factor correction (PFC) and also cover a wide range of real-world operating conditions.
Figure 1: ON Semiconductor 255 W ATX power supply reference design |
This new design combines multiple semiconductor technologies, each of which is optimized to provide the highest possible performance/energy consumption ratio.
At the AC power input or primary side, ON Semiconductor's NCP1654 continuous conduction mode (CCM) power factor correction controller reduces the number of components required to add PFC functions and provides a robust, cost-effective front end for an input voltage range of 90 to 265 Vac. This PFC stage provides a constant 385 V output voltage to the resonant half-bridge dual-inductor plus single-capacitor (HB LLC) converter in the secondary stage. This topology optimizes energy efficiency and minimizes electromagnetic interference (EMI) signals. This resonant half-bridge LLC converter is built using the NCP1396. Thanks to its proprietary high-voltage technology, this converter includes high-side and low-side drivers for half-bridge applications, accepting bulk voltages of up to 600 V.
On the secondary side, the architecture uses a synchronous rectification mechanism implemented with the ON Semiconductor NCP4302 controller to generate a 12 V output. The NCP4302 is specifically designed to simplify the use of synchronous rectification in switching power supplies, improving overall system efficiency by 2% to 4%. In addition to synchronous rectification control, the device also integrates the TL431 function into a precision shunt regulator, which reduces both board area and system cost.
The design uses the MBR20L45 dual Schottky rectifier, which operates with an extremely low forward voltage drop, further reducing power losses.
Finally, the design uses two identical DC-DC controllers to convert the 12 V down to +5 V, +3.3 V, and -12 V. The DC-DC controller is the NCP1587, a low-voltage synchronous buck controller in an SOIC-8 package. Each DC-DC controller uses a synchronous rectification mechanism to drive two NTD4809N (30 V, 58 A single N-channel power MOSFET). The MOSFET has low on-resistance (RDS(ON)) to minimize conduction losses, low capacitance to reduce driver losses, and optimized gate charge to minimize switching losses.
In addition, the compact flyback converter built on the highly integrated switching regulator NCP1027 in this design provides 15 W of standby power to another isolated 5 V input. This device contains a 700 V MOSFET and integrates a proprietary skip cycle technology to improve energy efficiency under light load conditions.
Energy Efficiency Testing
Figure 2 shows the results of energy efficiency tests of this new 255 W ATX power supply reference design at 100 Vac, 115 Vac, 230 Vac, and 240 Vac input voltages and various load conditions. To simulate real-world conditions, all test measurements were obtained at the end of a 16-inch (41 cm) cable.
Figure 2: Energy efficiency of ON Semiconductor's 255 W ATX power supply reference design under different load and voltage conditions |
Conclusion
Today's consumers and businesses are increasingly concerned about the power consumption of desktop PCs and servers from a business and environmental perspective. As a result, modern systems must not only be certified to meet relevant efficiency standards, but must also demonstrate that they can achieve the efficiency levels specified by these standards in real-world, not laboratory conditions. Given this, engineers must carefully evaluate the specifications of the design and fully understand how the vendor achieves the efficiency claims in the datasheet, especially the efficiency at different rated voltage conditions, the distance from the power supply where the test results were obtained, and other factors. With a complete understanding, engineers can accurately evaluate the efficiency under real-world conditions and evaluate the rationality of the claimed "green" energy efficiency in the end product.
ON Semiconductor carefully selected semiconductor components from controllers to MOSFETs to create the ATX 255W public reference design, which is configured to be put into production immediately and provides real-world energy efficiency levels that exceed the energy efficiency levels required by 80 PLUS Silver, Energy Star Version 5.0 and CSCI Phase III desktop PC power supply efficiency standards.