Feature

FROM EDN EUROPE: Active power factor correction

Explosive growth in consumer and office electronics presents electricity generators with increasingly nonlinear loads. Poorly behaved loads reduce supply capacity and can even threaten supply integrity. The European Commission introduces standards to combat the problem, but how do you ensure that your products meet new regulations?

David Marsh, Contributing Editor -- EDN, 1/6/2000

Until recently, virtually any domestic appliance you switched on looked very much like a resistor to the ac line supply. The explosive growth in consumer electronics has radically changed this picture, and it's causing the electricity supply industry considerable concern. Most contemporary appliances employ offline supplies that draw current toward the peak of the sine wave, typified by rectifier-bridge/smoothing-capacitor circuits. Paradoxically, today's drive for energy-efficient lighting circuits and variable-speed motor controls compounds the electricity generators' problems by presenting yet more nonlinear loads. In sufficient magnitude, these poorly behaved loads reduce the electricity-supply system's capacity and can even cause instability and resonance that can induce supply failures.

Acknowledging the potential for emerging problems that the widespread use of electronics could present, the European Union published its Directive 89/336/EEC for electromagnetic compatibility (EMC) in 1989. Apart from limiting RF emissions and specifying standards for RF immunity, the directive states that "Member States are also responsible for ensuring that electric energy distribution networks are protected from electromagnetic disturbance, which can affect them and, consequently, equipment fed by them." Accordingly, one of the key requirements for European Community (CE) approval is limiting harmonic emissions that couple into the ac line supply. This requirement has been steadily evolving, with tightening standards due to take effect on Jan 1, 2001 in the shape of European Normative (EN) EN61000-3-2. Although earlier and similar standards, such as EN60555-2, targeted the domestic environment, EN61000-3-2 extends its scope to include common office equipment and tightens limits for devices that draw maximum power around the crest of the sine wave.

The standards issue is an international bone of contention, with some offshore companies believing that European standards are designed to make their products uncompetitive. (From this viewpoint, EMC really means "evil market conspiracy.") In the United States, the authorities have chosen to make the electricity supply network more robust rather than put the onus on the consumer. But here in Europe, the national electricity generators have different capabilities, and authorities believe it's better to make the load easier to drive. The European approach is more efficient and therefore "greener"; it helps meet the global reduction in greenhouse emissions that's already an international agreement (Reference 1). But whatever view you take, it's up to you to select the standards that might apply to your products and ensure that your products comply (see sidebar "Navigating the standards maze").

Power factor steps in

Although the EN standards don't specifically mention power factor correction (PFC), designers frequently refer to EN61000-3-2 as the power factor correction standard. Power factor is the ratio of real power to apparent power, or the cosine of

(for a pure sine wave) that represents the phase angle between the current and voltage waveforms (see sidebar "So, what's power factor?"). Devices such as fluorescent lamp ballasts present the line supply with "peaky" loads, but the worst offenders are the uncorrected offline supplies that use rectifier bridge/smoothing capacitor circuits, which appear in nearly every domestic and office appliance. These supplies invariably draw nonsinusoidal currents as the ac line's instantaneous voltage exceeds the storage capacitor's voltage. The electricity generator must supply energy around the top of the sine wave rather than throughout the cycle, which can cause the sine wave to collapse around its peak (Figure 1). Another offender is the triac-based phase-angle control circuit that ac heater controls, lighting dimmers, and low-tech variable-speed motors often use.

From the generators' viewpoint, the phase lag between current and voltage, together with harmonics from peaky loads, combine to produce additional rms currents, reducing the real power that the network can supply. The supply lines must be heavier to accommodate these loads, and the resulting distortion induces resonant overvoltage and overcurrent in the neutral line that disturbs devices such as zero-crossing detectors. In three-phase systems—in which frequently little or no PFC exists, even in powerful rotating machines—the self-heating that distortion currents produce is a prime cause of burnout in the neutral line conductor.

Apart from benefiting the electricity generators, implementing PFC benefits consumers. Because every ac line source has some source impedance, varying loads at the consumer end produces fluctuations throughout the local supply. In sufficient magnitude, these fluctuations cause tungsten lighting circuits to flicker and can interfere with electronic devices. From the efficiency viewpoint, consider that a typical uncorrected switched-mode power supply has a power factor of around 0.6, which effectively reduces the current available from the ac line socket from 13 to 7.8A. The electricity generator sums real power and apparent power and charges you for it in kilovoltamperes, rather than the kilowatts of real power you're actually using.

ICs ease PFC implementation

Assuming that your products are within the scope of EN standards, you may be able to add a passive filter network at the equipment's ac line input. This low-cost approach generally suits only simple loads of about 75 to100W. Typically, you need a choke in series with both live and neutral lines, together with capacitor networks for low-frequency balancing and high-frequency interference attenuation. You may also need another choke to help smooth the dc output rail to further limit reflections into the ac line. Apart from being technically inelegant, this method may supply you with a product that's compliant but too bulky and heavy for its target application.

It's better to condition the equipment's input load power so that it appears purely resistive using active PFC techniques. Table 1 presents a selection of ICs that make it easy for you to implement PFC for supplies from electronic ballast power levels, to about 3 kW. Today's dominant PFC designs employ a boost preconverter ahead of the conventional voltage-regulation stage, effectively cascading two switched-mode power supplies. The boost preconverter raises the full-wave rectified, unfiltered ac line to a dc input rail at a level slightly above rectified ac line, typically to around 375 to 400V dc for European supplies. By drawing current throughout the ac line cycle, the boost preconverter forces the load to draw current in phase with ac line voltage, quashing harmonic emissions.

According to Ian Diaper, European field applications engineer at Micro Linear, "Other topologies are available, but the boost preconverter configuration provides the best attributes for meeting multiple power supply requirements, including universal input voltage operation (85 to 265V ac). You only have to design one supply for different continents—no circuit changes, 110/240V switches that users can set incorrectly with disastrous consequences or differently populated circuit boards for manufacturing to control." Diaper continues that the boost preconverter topology also resists power-line disturbances, such as brown-outs and dropouts, and simplifies overvoltage protection.

You can build a PFC supply using offline switched-mode power-supply (SMPS)ICs. Originally targeting flyback buck converters, Power Integrations' TopSwitch devices also suit a boost converter configuration. (For more on offline SMPS ICs, see Reference 2.) Two TopSwitch ICs suit PFC supplies from less than 25W to about 150W. The simplest configuration runs at constant duty cycle throughout the ac line cycle and illustrates the boost preconverter's basic operation (Figure 2). Because the IC switches at around 100 kHz, the ac line input voltage remains nearly constant during each switching period. Inductor current builds linearly during the switch's on time and discharges when the switch turns off through the (now) forward-biased diode. Inductor current is the sum of the average sinusoidal switch current and nonsinusoidal diode current, so the ac input current is nonsinusoidal; performance measurements indicate 18% harmonic distortion but a respectable power factor of 0.978.

Modulate the control loop

You can improve the basic circuit's harmonic distortion performance and also enhance its power-factor efficiency by varying the switch duty cycle with the incoming ac line waveform. Power Integrations' application note DN-7 illustrates how to perform this task for the company's TopSwitch series devices; other vendors apply the same principle in their dedicated PFC ICs by including multiplier circuits. On Semiconductor's MC33262 typifies the single-quadrant multiplier control system; it is one of several pin-compatible ICs that suit low-power loads, such as electronic lighting ballasts and SMPSs to about 450W (Figure 3). Similar ICs include Fairchild's KA7526, Linfinity's LX1562, and STMicroelectronics' L6561.

The MC33262 operates in critical-conduction mode, working in the area between continuous- and discontinuous-mode operation. Continuous-mode operation typically suits SMPS power levels higher than 400W, in which current in the energy-transfer inductor never reaches zero during the switching cycle; inductor charge and discharge rates depend on input- and output-voltage levels. Discontinuous-mode operation suits lower power SMPSs and signifies that the control circuit allows inductor current to fall to zero and remain there during each switching cycle, making it easy to regulate output voltage cycle by cycle. Because of stability considerations, designers typically avoid operating in the crossover region between these two modes. But in the MC33262, the control loop's zero-current detector starts the switching cycle, and an error amplifier-multiplier switches off the output MOSFET when the multiplier's output reaches a threshold. There are no dead-time gaps between cycles, so ac line current appears continuous, and peak switch current is twice the average input current. Also, because the switch can't turn on until inductor current falls to zero, the output rectifier's reverse recovery time is less important, allowing you to use cheaper diodes.

The single-quadrant, dual-input multiplier is critical to power factor and THD performance. The multiplier monitors ac full-wave rectified half-sines together with the feedback error amplifier's output. The error amplifier is a transconductance amplifier whose output also feeds an external compensation capacitor to set loop bandwidth at or below one-third line frequency. The error amplifier then monitors the converter's average output voltage over several cycles. The multiplier's output modulates a current sense comparator at ac line frequency so that the output MOSFET's on-time tracks ac line voltage. This action fixes output driver on-time and makes the preconverter's load appear resistive to the ac line. By applying an offset to the multiplier and current-sense comparator, the preconverter switches as the ac line passes through zero, which reduces line harmonic distortion. Measurements for the circuit in Figure 3 report a power factor of 0.996, THD of 4.8%, and 98% conversion efficiency with 240V ac input and 395V dc/450W output.

Fix divide-and-conquer designs

You might consider cascading two independent SMPSs as a divide-and-conquer approach to offline PFC and voltage regulation issues. Apart from parts count and space considerations, two SMPSs in series may suffer from high-frequency emissions that result from their internal oscillators beating or interfering with sensitive circuits, such as CTV video amplifiers. Vendors such as Infineon, Micro Linear, Texas Instruments, and Toshiba increasingly offer combination PFC/SMPS controllers that avoid these problems. It's also reasonable to expect that companies with offline SMPS IC expertise that don't appear in Table 1 will soon join the competition (Reference 2).

Infineon's TDA16888 is one example of this new combination IC class, controlling a two-stage converter topology that also supports low-power standby operation (Figure 4a). The PFC part operates as a pulse-width-modulation (PWM)- controlled boost or flyback converter; the SMPS part operates as a PWM forward or flyback converter. Both stages operate from a common clock that you can set at 15 to 200 kHz with one resistor. You can also synchronise the clock with an external reference, which reduces system noise in applications such as CTVs and PC monitors. PWM frequencies for either stage run at half the master oscillator frequency, with the PFC stage using leading-edge triggering and the SMPS stage using trailing-edge triggering (Figure 4b). Because the SMPS stage's current consumption normally coincides with the PFC stage's energy transfer phase, this configuration lets you use a smaller bulk-smoothing capacitor.

A variation on current-mode control makes the circuit stable even with no load. The SMPS stage avoids transformer saturation by limiting duty cycle to 50%; the PFC stage operates to as much as 94% duty cycle to ensure that sufficient energy is always available without distorting the source. In standby operation, the SMPS stage turns off, and the circuit provides power from auxiliary windings in the PFC stage. Measurements for a 150W supply demonstrate 76.7% efficiency and 99.4% power factor with 230V ac input and full output power; in standby mode, a 2.5W input supports 500 mW of isolated output power.

Don't neglect individual strengths

Many of the ICs in Table 1 have similar characteristics and capabilities, but you shouldn't neglect their individual strengths. For example, all ICs contain a voltage reference that's typically accurate to less than 2%, but if you want to use it externally, consider STMicroelectronics' L4981. And if you need combination-controller functions in an eight-pin device, see Micro Linear's ML4803. This IC reduces pin count with a single-pin voltage error amplifier and a current wave-shaping circuit that senses the boost inductor's current. While the main power switch is off, the system compares the input inductor's current down-slope with a ramp that the PFC section's output voltage variation programs. When the two signals meet, the main switch turns on for the rest of the cycle. Line or load transients that vary the PFC output voltage also change the programmed ramp shape, maintaining close-to-unity power factor.

At light loads, the ML4803's control technique demands that you ensure that the programmed ramp always starts from a nonzero value to prevent the output voltage from running away. Adding an offset to the current-sense signal avoids discontinuous-mode operation by forcing the PFC converter to operate in continuous-conduction mode and then pulse-skip when there's little output load. ICs such as Infineon's TDA16846 support variable loads and low-power standby operation by optionally varying the PFC switching frequency; the similar TDA16847 has a temporary high-power mode. In high-power mode, current-limited maximum power is available for a period you program with a resistor/capacitor, followed by indefinite operation at a lower current-limited level. The temporary high-power mode suits applications such as starting motors and permits less derating in the output power MOSFET.

Every IC features comprehensive protection against potential circuit fault conditions. Most ICs derive their start-up supply from the rectified ac line input through a current-limiting resistor, with internal zener clamps to protect against input overvoltage. All devices feature undervoltage lockout protection with or without hysteresis. Fairchild, Linfinity, and Texas Instruments offer devices with undervoltage lockout thresholds that suit alternative start-up circuits (resistor/zener configurations or combination bootstrap supplies). Output overvoltage protection normally uses a voltage feedback comparator that shuts down the output under fault conditions. Some devices, such as Linfinity's LX1562, include PWM duty cycle limiting and current clamps around the MOSFET drivers to limit output voltage and overshoot. One advantage of PWM is that it makes current limiting easy, cycle by cycle; also look for leading-edge blanking circuits that suppress high-frequency noise on the current waveform.

How do I test it?

You may have already visualised your circuit's PFC performance using a software tool such as Analogy's Saber simulator, but you'll certainly want to test the real hardware. When you've come to grips with the EN standards, you'll know that the EC permits a Declaration of Conformity route, which means that you can self-certify products if you follow established procedures. You need to use the correct harmonised EN standards, record measurements at each stage, and build a test file that shows what you have done. Many newcomers first approach a dedicated EMC test facility to help them understand the steps involved and how to proceed. Intertek Testing Services (ITS) is the world's largest product and commodities testing company, with EU-notified competent body facilities that specialise in safety and EMC testing. Graham Hicks, an EMC consultant based at ITS's Leatherhead, UK, office, says that he's always happy to advise on the latest legislative and standards issues before testing starts. ITS also publishes the useful quarterly EMC Newsletter, which will help you to keep up to date.

You'll probably want to invest in your own test facilities, and you have several options for power-supply testing. A power measurement mode augments Tektronix's TDS3000-series oscilloscopes to monitor power and line harmonics, instantaneous power, and modulation effects. Or, you can buy a power analyser from companies such as Gould Nicolet (see sidebar "For more information"). Gould Nicolet's DataSYS 7100 power analyser provides live, oscilloscopelike displays of current, voltage, and power waveforms in a four-channel, 200-MHz DSO package (Figure 5). The instrument automatically selects the appropriate EN-standard test limits for the load under test and prints a test report with pass/fail indication. Electronic equipment designers find the Class D test limit most challenging, describing loads that take most of their current during the top one-third of the ac sine wave—from 60 to 120°.

But according to Leslie Green, senior principal engineer at Gould Nicolet, you must first make sure that the ac line supply is up to the job: "You might think that testing a product with a distorted mains supply would worsen your power supply's performance, but that's not necessarily the case. We found the peak currents into one of our early model oscilloscopes doubles when the mains supply is harmonically pure, compared with our normal factory mains supply that has quite flat tops." For this reason, Gould Nicolet developed PowerSource 1000 to provide a harmonically pure ac power source for loads reaching 1 kW. Vendors such as Agilent offer integrated test sets that include a line power conditioner; the company's HP6812 suits loads as great as 750W, and other models suit single- and three-phase applications at 375W to 4.5 kW.

So, what's power factor? Textbooks conventionally describe "power factor" as "the ratio of real (or working) power to apparent power." This definition appears simple until you consider distortions in the current or voltage waveforms. "Real" power, which you measure in watts, is the energy that performs the useful work. "Apparent" power is the sum of real power plus reactive components; you measure apparent power in volt-amperes. Power factor is then watts/voltamperes. For sinusoidal waveforms, real power is the product of the applied voltage times the in-phase current component. Again for sinusoids, you can model apparent power with a component that's in phase with the input voltage and a 90, out-of-phase quadrature component. Following EN-standard stipulations for measurement configuration, you assume that the voltage waveform remains sinusoidal and distortions apply only to the current waveform. Including distortions, the total current becomes a root-sum-of-squares function of any dc component, I0 (zero in a pure ac waveform), together with the fundamental 50-Hz component I1RMS , and all its harmonics InRMS :   If the fundamental rms current comprises an in-phase component, I1RMS(I) , and a 90° quadrature component, I1RMS(Q) , you can express the total rms current as:   Real power is the product of rms voltage and the in-phase current component (P = VRMS *I1RMS(I) ). If describes the angular displacement between the rms voltage and the in-phase current components: I1RMS(I) = I1RMS *cos and P=VRMS *I1RMS cos . Apparent power S=VRMS *IRMS(total) , and   If represents the phase angle between I1RMS and IRMS(TOTAL) ,   So, if represents the phase lag between the voltage and the fundamental current component, and represents the distortion angle that current harmonics cause, PF=cos *cos . As the total current's harmonic components approach zero, also approaches zero, and cos approaches 1—the ideal unity power factor and active power factor correction's goal.

Navigating the standards maze If you're a power-supply designer, you're probably feeling that you bear the brunt of the European Commission's (EC's) attention. You've had to come to grips with the safety measures that European Normative (EN)- 61010 embodies; meanwhile, you've been receiving a stream of confusing information that spawns a flood of related electromagnetic-compatibility (EMC) standards. Helping to further confuse the issue, national standards bodies have historically published subtly different specifications. The European Union's (EU's) Directorate General III "New Approach" directives set out to harmonise these specifications. The commission recognises the difficulties that designers face and is sponsoring the (simpler legislation in the internal market (SLIM) group to revise legislation including the EMC directive; the group is expected to introduce recommendations over the next two years. The Official Journal of the European Communities (OJEC) provides first-line information on the status of today's standards. When OJEC publishes a European standard, the standard is ready for use. The EC's Alejandro Ulzurrun observes that, because OJEC published EN 61000-3-2, manufacturers can use this standard, when appropriate, "to obtain presumption of conformity to the protection requirements of the EMC Directive" (EMCD). Ulzurrun continues that harmonised standards under the EMC Directive are not meant to become law; manufacturers may choose other ways—under article 10(2) of the EMCD—to achieve compliance with the protection requirements of the EMCD. But legislation requires your product to have a European Communicty (CE) mark. To obtain this mark, you need to demonstrate good design practice, which is where standards are useful. It's your responsibility to select standards that suit your equipment. For example, you might select EN61326-1 if you manufacture electrical equipment for measurement, control, and laboratory use. Depending on your circumstances, you may choose to combine the standard that directly applies to your equipment with generic standards that are still in force, such as EN60555-2, which tackles harmonic disturbances in supply systems for the domestic environment. But you'd be better off considering the EN61000-3 series, which will replace 555 on Jan 1, 2001. Robert De Vré, EMC specialist at the EC, notes that the scope of EN61000-3-2 and -3 (for voltage fluctuations) applies to all equipment that you connect directly to public low-voltage power distribution systems. This equipment excludes all purely industrial equipment. For industrial environment use, refer to the standards that apply to the relevant equipment class, such as EN50081-2. Requirements also exist for RF interference on the power supply port. Other standards published in OJEC cover these requirements, including EN55011, EN55022, and EN55014. You can obtain more information on the application of standards for the EMC directive in a guidance document from the European Committee for Electrotechnical Standardisation(CENELEC). A visit to the EU's Directorate General III Web site, www.europa.eu.int/comm/dg03/, currently shows more than 100 EMC-related entries. EU's New Approach Web site, www.newapproach.org also contains the EN61000 series of harmonised specifications. You can also navigate the standards maze by following links such as www.elmac.co.uk, the Web site of test-service house Elmac Services. The site contains resources such as a critique of the EMC directive that discusses issues including the technical construction file route to gaining a "presumption of conformity." You'll find a presentation that discusses conducted emissions; Mathcad-format files to download for calculation and tutorial uses; and links to other resources, including lists of standards that appear in OJEC .

---

Author info

You can reach Contributing Editor David Marsh at forncett@compuserve.com.

REFERENCE

1.  "Kyoto Protocol to the United Nations Framework Convention on Climate Change", Kyoto, Japan, Dec 1 to 10, 1997.

2. Marsh, David, " SmartPower ICs simplify offline SMPS design", EDN Europe, August 1999, pg 33.