For a PDF verion of this article, click here.
The notebook computer is fast becoming an appendage for the 21st century professional. While maximum functionality (such as a large screen, easy viewing, high computing power and long battery life) from laptops is highly desirable, it often comes with penalties in the weight and size of the machines. Power electronics professionals have an opportunity to confront this paradox by making the notebook power adapter lighter yet functionally robust.
In the recent past, notebook functionality was limited such that the power requirements were in the 50-W to 70-W range. In recent years, the power requirements have climbed past the 100-W range, while the weight and size expectations have essentially remained the same. Additionally, the regulatory need to meet the low standby power performance, external power supply (EPS) efficiency requirements and IEC1000-3-2 harmonic requirements for more than 75 W of input power compound this challenge. However, recent trends have enabled power-supply manufacturers to address these challenges and provide alternative solutions.
As notebook computers become functionally richer, their power demand increases. In addition, as battery capacity (or density) increases, the charging requirements also go up; thus, the notebook adapter sees a two-fold increase in power requirements. However, the option of a bigger or hotter power adapter is not appealing to users. Users might also want the same notebook adapter to function anywhere in the world without a voltage-select switch for the 110-Vac/220-Vac operation; thus, notebook adapters have to be designed for true universal line-voltage operation.
The regulatory and OEM expectations are also coming into play. The regulatory environment expects that the notebook adapter does not waste energy or inject harmonics into the utility line. When the adapter is plugged in without any load, it should draw as little power as possible (standby requirement). A more recent stipulation requires certain active-mode efficiency, which is averaged over different load conditions (25%, 50%, 75% and 100%). This requirement is driven by data collected on the usage patterns of computers, and is being implemented in a harmonized manner by regulatory agencies around the world to ease compliance and reduce overhead.
Finally, the harmonic reduction requirement mandated by the European Union and Japan has started becoming applicable to the notebook adapters, because they have crossed the compliance threshold of 75-W input power for these standards. In a sense, the mobile, global nature of the notebook adapters makes them the first mass-market power supply to fall under the IEC1000-3-2 umbrella.
For OEMs, the notebook adapter is nothing more than a commodity requirement. As such, they impose stringent internal specifications and derating guidelines while requiring low costs. However, recent events indicate they are getting more concerned about complying with the adapter requirements. This has led them to explore possible alternatives and also to reevaluate requirements to achieve a higher power density in the adapters. One approach is to revisit the notebook adapter architecture, and adjust the performance and efficiency for a realistic set of specifications.
Existing Solutions and Approaches
The existing notebook converters typically use flyback topology for the pulse-width modulated (PWM) conversion. This has been the most effective solution, both economically and technically, for many years. As shown in Fig. 1, a typical flyback converter used for notebook adapters operates over a universal input voltage range of 90 Vac to 265 Vac, and employs few components thanks to highly integrated PWM controller solutions. If the power level of the adapter is below 75 W, there are no additional power stages involved.
The key performance criteria for these adapters are power density (driven by package size requirements), safety and low case temperature. In terms of control methodology, the dominant approach is the current-mode control as enabled by the UC384x series. However, as shown in Fig. 1, today's notebook adapter uses a different PWM controller, such as ON Semiconductor's NCP1200. The benefits offered by the new-generation PWM controllers over the UC384x are:
Higher level of integration eliminating many external components, while still supplying current-mode control benefits
Ability to go into cycle-skip mode to reduce standby and light load losses
Ability to start from the high-voltage (HV) input, eliminating losses and component costs in startup circuit
Elimination of the error amplifier circuit from the IC as the error processing is done in the secondary
Ease of safety and EMI compliance for quicker time to market.
However, for power adapters with power requirements greater than 75 W, the notebook manufacturers face a choice. Either they can design adapters for universal applications and include a power factor correction (PFC) stage, or they can implement a separate design for markets that require PFC such as Europe and Japan. However, the addition of the PFC stage adds complexity and cost, making the design more challenging. Typical designs in this application use a critical-conduction mode (CRM) boost PFC stage as shown in Fig. 2.
The CRM PFC can be designed by using a controller such as the MC33262 as shown in Fig. 2. This simple eight-pin controller enables an easy implementation of PFC. In critical-conduction mode, the circuit level benefits, such as no diode recovery issues and lower switching losses, can be realized.
While this approach has yielded acceptable performance and cost for existing designs, more improvements are needed to meet emerging requirements.
Improved Two-Stage Solutions
In recent years, many targeted solutions have emerged to provide better alternatives for the two-stage approach. On the PFC side, despite the prevalent acceptance of the CRM approach, there are recognized limitations. First, it operates under variable switching-frequency mode and the switching frequency can go quite high during zero crossings and light load conditions. This variable frequency creates problems with filtering and low standby power.
A new solution that takes the best traits of the critical mode approach, yet limits the switching frequency and improves standby performance, was introduced with ON Semiconductor's NCP1601 (Fig. 3). In this solution, the controller is allowed to go seamlessly between the discontinuous mode (DCM) and CRM without compromising the PFC. At or near the zero crossings of the line voltage, and under lighter load conditions, the controller functions in a true fixed-frequency DCM PFC mode. At around the peak of the line waveform at full load, the controller shifts into critical mode, thus limiting peak current values from getting too high.
The discrete components available for the PFC implementation also have come a long way in recent years. The PFC MOSFET requires a 500-V rating, and the introduction of low RDS(ON), low gate-charge MOSFETs has helped tremendously. The boost diode in CRM or DCM circuits requires different characteristics; therefore, manufacturers have started catering to these needs by introducing devices such as ON Semiconductor's MUR450PF.
The end result of these advances is that the addition of the PFC section can be implemented at low cost, while delivering a low-line voltage efficiency of approximately 95% at full load and consuming less than 300 mW of standby power. These performance points clearly allow adapter designers to meet the OEM and regulatory requirements.
On the switched-mode power supplies (SMPS) side, one of the significant trends has been to use valley-switching flyback converters instead of the traditional fixed-frequency flyback topology. This approach yields better efficiency and lowers the electromagnetic interference (EMI). Similar to the issues faced by the CRM boost topology, the valley-switching topologies suffer from significant frequency variation as a function of line and load that may create high EMI and standby power dissipation. Recent innovations in control techniques have enabled the industry to address these issues — broadening the acceptability of the valley-switching approach for notebook adapter applications.
The traditional valley-switching algorithm works by sensing when the MOSFET drain voltage goes to its minimum point and turning the FET on at that point. However, depending on the load and line conditions, that valley point may be reached fast, resulting in high switching frequency and high switching losses. Unfortunately, this phenomenon occurs at light load when low power consumption is critical. New controllers such as ON Semiconductor's NCP1337 incorporate valley-skipping algorithms to address this problem. As shown in Fig. 4, if the switching frequency at the valley point is higher than a set switching frequency, the MOSFET turn-on is delayed to the next available valley point.
Finally, due to the special startup, fault mode and standby requirements, it is required that the interfacing (including sequencing and handshake) between the PFC stage and the SMPS stage be carefully managed. For example, many designers prefer to keep the PFC stage off during the startup, standby and fault modes. The power dissipation attributable to the PFC stage is eliminated by this approach. However, it places an additional burden on the SMPS stage, because it has to be able to support full output power without the help of the boosted input that is normally available from a PFC front end.
Typically, the flyback converter handles the wider input range well, and since the modes when PFC is off and output is delivering full power are not sustained for longer periods, no significant overdesign of the power stage is required. The interface requirement is easily handled by an innovative solution contained in many recent controllers (Fig. 5). This solution incorporates smarts into the PWM controller that recognizes all the modes in which the PFC must be off. The PWM controller has an output pin that provides the bias voltage to the PFC, resulting in turn-off of the PFC controller and PFC stage as required. This simple mechanism works with any PFC controller, and hence is nonrestrictive.
The Single-Stage Alternative
The obvious question to address when designing notebook adapters with the PFC requirements is how to eliminate the burden of two power-processing stages. Achieving an efficient and cost-effective single-stage power conversion has been a common goal of the design community. While many single-stage solutions exist, they all have some limitations.
One such limitation is that the output-voltage ripple contains a low-frequency component that cannot be inherently eliminated without using additional energy-storage capacitance. A second limitation lies in the fact that many schemes try to use current steering to achieve a better tradeoff between low output-voltage ripple and low total harmonic distortion (THD). These tradeoffs require extra engineering effort for each design. Yet another limitation to consider is that certain specifications such as output ripple, transient response and hold-up time are harder to meet than when the design is implemented with the two-stage solutions.
For the single-stage solution to see widespread use in notebook adapters, collaboration among the OEMs, power-supply designers and vendors of critical circuit components will be needed to provide an optimal solution. OEMs need to determine if they can make any concessions on the specifications such as allowable output ripple, transients and hold-up time in order to gain system-level savings. Power-supply designers need to invest time and effort into developing and optimizing innovative single-stage solutions. Finally, component suppliers such as semiconductor companies and magnetics companies must understand the system requirements and provide appropriate solutions.
An example of one such collaboration is an adapter design solution developed to meet the specifications of a leading notebook OEM. This approach uses ON Semiconductor's NCP1651 as the core controller, and meets all the performance targets at a significant cost savings. During the development, interactions with the OEM also led to some specification tradeoffs that could realize further cost and space savings. The prototype is shown in Fig. 6.
One parameter that can be negotiated between OEMs and power-supply developers is output-voltage ripple. While this is usually specified as less than 3%, the real concern is the effect of high-frequency ripple components. However, single-stage adapters introduce a low-frequency component, which is not an EMI hazard. This permits OEMs to consider relaxing their output ripple requirements.
Other negotiable parameters include hold-up time and transient response. In the case of the former, the presence of a battery makes this a redundant requirement for most situations. For the latter, agreement between OEMs and power-supply vendors can usually be reached through better clarification and validation of step loads and step times.
Whether using the single-stage or traditional two-stage architectures, the performance requirements from the stakeholders will continue to grow. To meet this challenge, power-supply companies and their partners must continue to innovate. Developments that can contribute to these advances are synchronous rectification of the output, innovative constant current-constant voltage (CC-CV) circuits and better solutions for input line rectification. In addition, for higher power two-stage solutions, other topologies such as resonant or clamped-mode topologies can be considered.
The final decision on what approach is best for the notebook adapter depends on factors such as designer familiarity, component availability and costs, and OEM specifications. Several options are summarized in the table. However, when notebook manufacturers interact at a system level with power-supply manufacturers and key component suppliers, such as semiconductor vendors, better solutions can emerge to meet the end customer's demands and advance the technology of notebook adapters.
|Traditional Two-Stage||Improved Two-Stage||Single Stage|
|PFC Stage||CRM boost||DCM/CRM Boost||CCM Flyback|
|SMPS Stage||DCM Flyback||Valley Switching|
|Efficiency Range||82% to 85%||84% to 87%||84% to 87%|
|Standby Power||~ 500 mW||< 200 mW||< 500 mW|
|Power Stage||2 FETs, 2 magnetics, 2 diodes, 2 capacitors||2 FETs, 2 magnetics, 2 diodes, 2 capacitors||1 FET, 1 magnetic, 1 diode, 1 capacitor|
|Possible Future Improvements||Minimal performance or cost gain (discrete component based)||Some recent gains demonstrated (systems level, synch. rect., etc.)||Performance and cost gains on the horizon (synch. rect., clamp, etc.)|
|Improvement Targets||85% to 87%, < 300 mW at equivalent cost||86% to 88%, < 150 mW at equivalent cost||> 90%, < 150 mW at lower/equivalent cost|
|The efficiency range is taken at the conditions of low line voltage under full load and at the end of the cable. The standby power is taken at the input voltage of 230 Vac and under no load.|