|
If you know your equipment’s load current waveshape – whether calculated, simulated, or measured with close to the standard source impedance – you can calculate or simulate the resulting mains voltage fluctuation waveshape. By referring to 4.2.3 and figures 5, 6 and 7 in EN 61000-3-3 you may find that altering the load current waveform can give useful reductions in the measured flicker value, even if the rate per minute, peak and steady-state amplitudes of the fluctuation remain unchanged.
EN 61000-3-3 requires the integration of each measured sample of voltage fluctuation for a period of 10 milliseconds. Voltage fluctuations occurring on shorter timescales are ‘smoothed out’ by this integration process.
The supply impedance has an inductive component (L), so since V = Ldi/dt the rise/fall times of the load’s current waveform could have an effect on voltage fluctuations. The standard total single-phase L is around 796mH and its impedance exceeds the resistive component of the supply impedance above 80Hz. The rate of change of current required to create a 4% voltage fluctuation – due to the standard supply inductance alone – is 11,558 A/s.
80Hz is not a very high frequency, so equipment which draws fluctuating currents with spectral components above 80Hz may find that the inductive component of the standard supply impedance contributes more to the measured voltage fluctuation than the resistive component.
For example, high rates of current change are commonly exceeded by DC storage capacitors charging-up via rectifiers directly from the mains, especially when they are connected at the peak of the mains cycle – the worst-case for capacitor inrush. But under such conditions typical capacitors of under 1,000mF require less than 1ms to charge to 370V and although the peak voltage fluctuation caused by the capacitive inrush would be very large indeed, the 10ms integration time ‘smooths it out’ to make the EN 61000-3-3 measured value very much less. Simplified analyses indicate that the rectifier-capacitor power supply input stages typical of switch-mode power supplies risk failing the inrush current limit when capacitor values exceed 600mF.
Where the DC current load on the storage capacitor is significant during the switch-on period, the capacitance values may need to be much less.
For many low-power consumption products the rise and fall times of their load current may not create significant emissions problems due to the supply inductance, at switch-on or during operation. But equipment which has unregulated 400VDC capacitors of over 600mF powered directly from mains rectifiers – and equipment with high levels of pulsed power (e.g. strobe lamps, powerful pulse generators) – should always consider the effects of Ldi/dt on their emissions of voltage fluctuations.
Most DC capacitors aren’t significantly discharged between one half-cycle and the next and only need ‘top-ups’ after their switch-on inrush. The fluctuating currents drawn by their DC load will then be the major contributor to their emissions.
Products such as computers, computer monitors, and TVs were traditionally not a cause of voltage fluctuations and flicker (except at switch-on, when CRT degaussing coils often caused the biggest problems). But they are increasingly adopting ‘Energy Star’ or other energy saving methods which can cause their DC load currents to change by 90% or more in under 1ms between standby and operational modes, making modern versions into significant sources of voltage fluctuations and flicker.
Reducing inrush currents
Where AC-DC power converters draw such high inrush currents at switch-on that their corresponding voltage fluctuation (integrated over 10ms) creates a problem with meeting EN 61000-3-3, steps should be taken to reduce their inrush current.
Inrush current into DC power supplies can be limited and/or have its rate-of-change slowed down by fitting resistors in series with their mains inputs. These resistors are usually shorted-out by relay contacts or triacs after the first second or two to permit normal operation of the equipment.
An alternative to an expensive resistor/relay combination is to use negative temperature coefficient (NTC) devices instead of a series resistor. NTCs have an initially high resistance, reducing to a low value as they heat up due to the passage of the equipment’s mains current. Their benefit is that they do not need to be shorted-out by a relay or triac for full operation of the equipment – but their dimensioning can sometimes be tricky. Their thermal inertia and hot/cold resistances need to be taken into account with the operational current consumed by the equipment and the permitted maximum inrush current. NTC devices can get very hot, and must be positioned so they cannot melt plastics, damage cables or components, or burn operators or service technicians.
A useful technique is the soft-start, sometimes called a ramp start. This gradually increases a DC or AC voltage from zero to the full value over a second or two. As well as helping to reduce inrush current soft-starts also help reduce the stresses on components. (They also allow protection devices time to operate before components are damaged when switch-on occurs during a fault or misuse situation, such as an incorrect mains voltage.)
Many types of switch-mode controller ICs have soft-start features designed into them. These help inrush currents by ramping the DC load on the power supply. ‘Active PFC’ types of controllers don’t have large DC storage capacitors after their mains rectifiers, so using soft-starts with these controller types can reduce switch-on inrush to negligible amounts.
Mains-powered motors, transformers and other inductive components can draw large inrush currents, which can vary over a 2:1 range depending on the phase angle of the AC mains voltage at the instant of switch-on. It is often found that switching on when the mains voltage is at a maximum gives the lowest inrush currents with inductive loads. When testing such parts it is important to find their worst-case switch-on phase angle.
Soft-starting/stopping mains-powered inductive loads in the past often used motorised variable transformers, but are now more likely to use phase-angle-controlled triacs. Because such triacs are only used intermittently they often need little or no heatsinking. The conducted and radiated RF emissions from the triacs during starting or stopping might not need to be suppressed if they are infrequent enough. An added advantage of soft-starting motors is that it helps prevent sudden application of torque.
Series resistors and NTCs can often be used with inductive and motor loads, but the reactive nature of the load might make them more difficult to design.
Reducing emissions in normal operation
Where time-proportioning on/off control (sometimes called bang-bang control) is used, it is sometimes practical to replace electro-mechanical switches or burst-fired triacs with electronic power converters. These can be designed to vary the load power in a continuous manner and so emit no voltage fluctuations or flicker. Microwave ovens have been known to use this technique.
If time-proportioning control must be used, load switching rates of between 5 and 20Hz are best avoided altogether. Reducing the switching rate of a load below 5Hz (or increasing it above 20Hz) allows it to benefit from the higher limits permitted by figure 4 of EN 61000-3-3. For heaters, chillers, and motors decreasing the switching rate may require additional thermal or mechanical inertia to achieve the desired degree of control.
A supplier of a system must meet EN 61000-3-3 for the whole system. In computer systems where the power-down times of the individual computers are set to be quite low, the system as a whole might create significant amounts of flicker at quite a high rate. It may be enough to merely set the power-down energy-saving times to be longer, to take advantage of the more relaxed limits allowed at lower rates of flicker.
Apparatus with switched loads can also benefit from soft-start/stop techniques – typically using phase-angle-controlled triacs. Figure 6T shows that a ramp up/down time of 100ms reduces the measured emissions to 70% of the corresponding step-change voltage fluctuation. Longer ramp times give correspondingly lower measured emissions, with 1 second reducing the measured value to just 20% of what would be measured on a step-change of the same magnitude.
In equipment with multiple smaller heating elements or motors, sometimes all that is necessary is to make sure that more than one or two heaters or motors cannot be switched on or off at the same time. Splitting a large load into two or more smaller loads, each individually controlled so that they do not switch on or off at the same times can be a useful technique. Figure 6T shows that splitting a voltage fluctuation step change into two half-height step changes with a delay of more than 100ms between them reduces the measured emissions to about 70% of the single-step value. Longer delays do not result in very much lower emissions (a 1 second delay only reduces emissions to 50%). Splitting a load into three or more and delaying their switching by at least 100ms each should bring greater reductions in emissions.
Most power supply designers design unregulated rail voltages so that they exceed the dropout voltage of the following regulator by the minimum amount possible, to improve efficiency and reduce size and cost. Unfortunately, this approach makes it difficult for the unregulated DC storage to provide much benefit in reducing the effects of load current fluctuations. (In the section on immunity to supply dips and dropouts it is shown that such design also makes products more susceptible to supply quality issues such as dips and dropouts.)
Where the fluctuating load current is drawn from a DC rail, increasing the size of the unregulated DC storage capacitors can help by reducing the di/dt of the load current. This can reduce the contribution to the fluctuations from the inductive part of the supply impedance and (if di/dt is low enough) reduce the measured emissions by achieving a ramp start and stop instead of a step-change (see Figure 6T). This technique also requires some inductance or other current limiting between the mains rectifiers and the storage capacitors – and often needs a higher unregulated voltage too – because the storage capacitors need to experience a large ripple voltage in order to ‘smooth out’ the energy demands of the fluctuating DC load.
‘Smoothing out’ the mains current variations by increasing the size of DC storage capacitors would need very much larger capacitance values than are generally used to control DC ripple, especially when the rate of fluctuations is not very high. However, modern developments in ‘supercapacitors’ (also known as ‘ultracapacitors’ and ‘boostcaps’)mean that such options may not now be impractical and novel solutions may now be possible to previously intractable problems.
Where larger storage capacitors follow immediately after an AC rectifier they can mean larger inrush currents at switch-on, and also an increase in the harmonic emissions. This is another reason for adding inductance or other current limiting devices between the rectifier and the storage capacitor.
Using unregulated rails with a very much increased voltage makes it easier to achieve enough energy storage to ‘smooth out’ DC load changes (and improve immunity to supply dips and dropouts) at reasonable cost. This is because the energy stored in a capacitor is proportional to the square of its voltage. So where large capacitive energy storage is required, using higher voltages can reduce size and cost. Also, operating the unregulated DC storage at a much higher voltage than is needed for circuit operation makes it possible for all the excess voltage to be used for DC ripple caused by the variable load current demands.When linear regulation is used the higher unregulated voltages make efficiency much worse, so this method is more suited to switch-mode regulators.
‘Active PFC’ switch-mode boost converters (see the section on reducing harmonic emissions) can also be used to reduce voltage fluctuations and flicker. They typically set the time-constant of their input current demand to 0.5s or so and will make step-changes in DC load appear as ramp-like changes in mains current, giving a lower measurement on a flickermeter (as shown by Figure 6T, useful benefits occur when ramp times >100ms ).
Where the rate of occurrence of load fluctuations exceeds 120 per minute a 0.5 second time-constant in the boost circuit will ‘smooth out’ the fluctuations to some degree. With the same time-constant, greater reductions will be achieved for higher rates. Longer boost-circuit time-constants will give greater reductions at fluctuation rates above 120/min and/or help achieve some useful ‘smoothing’ at lower rates.
For ’smoothing’ to work in an active PFC boost circuit the values of the storage capacitors and the unregulated voltage need to be dimensioned correctly for the size and rate of the DC load current fluctuations, and the time constant of the boost circuit.
Beware - most active PFC control ICs will suddenly switch off the input current completely when the maximum voltage on their storage capacitor is exceeded (this usually occurs just after a heavy DC load current has been removed). So if the unregulated capacitors don’t have enough stored energy these active PFC circuits can sometimes make emissions of voltage fluctuations worse.
If circuit techniques fail or aren’t appropriate for some reason, system-level approaches can help reduce flicker although they won’t help the products concerned to meet EN 61000-3-3.
One (expensive) solution may be to run the problem equipment from its own low-voltage distribution transformer, so that their voltage fluctuations aren’t applied to equipment powered from the public mains supply. The much lower impedance of the high-voltage distribution network attenuates the effect of their fluctuations considerably. Equipment run from a private LV supply is not covered by EN 61000-3-3 at all.
Motor-generator sets or continuous double-conversion on-line UPS can also be used to reduce voltage fluctuations, if they are dimensioned correctly. At least they can reduce the di/dt of the load current fluctuations. At most – with sufficient energy storage given the rate of the current fluctuations – they may be able to ‘smooth out’ the mains current so that the peak amplitudes of each fluctuation are reduced.
Some types of power factor correction equipment used in systems and installations may also be able to reduce the levels of voltage fluctuations and flicker caused by equipment.
Every conductor stores energy in its intrinsic inductance, and inductive devices such as motors also store energy in their magnetic fields. When the flow of current is suddenly interrupted by breaking an electro-mechanical contact, such as a switch, relay, commutator, or slip-ring, the ‘flyback’ of this stored energy causes a spark due, to breakdown of the air as the circuit-interrupting contact first opens (or when it bounces after closing).
Sparks emit electromagnetic disturbances quite literally from DC to daylight, and many microprocessor circuit designers have been surprised by the ease with which their higher-frequency components can couple into their digital circuits (e.g. via coil-to-contact capacitance, or proximity of cables or PCB tracks) and crash their microprocessors.
It is best to avoid the generation of arcs and sparks by avoiding electromechanical switching completely. The use of solid-state relays, brushless DC motors, AC motors, and the like all help eliminate sparking, although some of these will add new EMC problems of their own.
Where sparks cannot be avoided, emissions standards will be easier to meet by making sure there are no more than 5 sparks per minute in the product, with a spark duration of 10ms or less (less than half a mains cycle, typical of a microswitch or fast-acting relay). In heavy industrial applications it also helps meet emissions standards if the total rate of spark production is less than five per minute, but spark durations of up to a second or two may be acceptable. Beware – although these rates and durations of sparks may be allowed by an emissions standard, they can still upset sensitive circuits so may not be desirable for operational reasons, especially where critical functions are being controlled or monitored.
Emissions from arcs and sparks are usually reduced by ‘snubbing’. Simple snubbing involves connecting a series combination of R and C (sometimes just C) near the switching element to slow the rate of rise of inductive flyback voltage and so limit the size of the resulting spark. Connecting a snubber across the contact gap has the disadvantage, in AC circuits, of allowing a leakage current to flow which might shock a person who worked on a circuit expected it to be safe because its relay contacts or switch were open.
Connecting snubbers in parallel with the load’s send and return conductors, close to the switching element, sometimes gives better results than connection across the switched contacts, and does not allow leakage past the contacts. Sometimes two sets of snubbers may be required, one to deal with the flyback of the load’s inductance, and one to deal with the flyback of the supply’s inductance. Figure 6U shows the alternatives for snubbing switch and relay contacts.
Snubbers can also use non-linear devices such as diodes, rectifiers, zeners, and a variety of surge protection devices (see Part 3 of this series) to provide an alternative path for the flyback currents, either on their own or in conjunction with RC snubbers. The higher the turn-on voltage of the device, the faster the stored energy will collapse and the quicker can be the rate of cycling of the load. Unfortunately, the higher the turn-on voltage, the greater will be the spark at the contacts, so this can lead to a compromise between rate of operation and emissions. DC circuits can use unidirectional semiconductor snubbers, remembering that the flyback voltage has the opposite polarity to the applied voltage.
A side-benefit of all spark suppression techniques is that they generally increase contact life.
In general, DC motors are a very serious source of conducted and radiated emissions, and are very difficult to suppress. The filters and other suppression devices required for them to meet emissions standards can cost more than (and sometimes bulk as large as) the motor or bell itself.
Some ‘pancake’ DC motors don’t spark because their brushes connect to a number of rotor windings at once, so there is always one of them in circuit to provide a path for the flyback currents in the windings.
Larger, more industrial DC motors with fully-enclosed metal bodies tend to emit less and be easier to suppress than lower-cost motors. Larger DC motors connected by many metres of cable to their controls or drives are able to reduce their emissions by using good quality screened cable, as long as its screen is 360o bonded at the motor’s metal terminal box (and probably to the earthed cabinet enclosure at the controlling end).
Where this technique is not sufficient, or impossible to apply (as in many motorised toys or domestic equipment such as CD players), it is best to use a motor with transient suppression fitted to its rotor. The rotor is where the energy is stored in a brushed DC motor, and is best dealt with before it causes sparks in the commutator. A ‘varistor disc’ can easily be fitted to most low-voltage DC motors, essentially connecting a varistor (voltage-dependant resistor, described in Part 3 of this series) between each pair of contacts on the commutator. For a 24V motor the varistor disc may be designed to conduct at 30V or so, and only conducts current when flyback occurs. During flyback, it conducts the energy into the neighbouring winding and limits the resulting overvoltage at the commutator to under 45V or so. This still causes sparking, but only small ones with much lower emissions.
Where a varistor-disc motor cannot be obtained, it is usually necessary to shield the motor and filter after the commutator, not always very easy to do at low cost. Metal shielded motor bodies are preferred to (crudely) catch the radiated emissions from the sparks and return them back into the motor where they came from, via the filter. The filter is also needed to reduce the conducted emissions. Since DC motor emissions are still going strong at 1GHz (and also, in fact, at 10GHz), motor shielding needs to have very few very small gaps. Motors with metal end caps and metal bodies may appear well-shielded, but the bonds between the metal parts may be poor due to paint or anodising.
A filtering technique which works well is to bond one of the commutator terminals directly to the metal motor body (the shield). The other terminal is decoupled to the motor’s metal body by a capacitor with very good high-frequency characteristics, such as an 820pF multilayer ceramic with a COG or NPO dielectric and very short leads. The capacitor must be rated to cope with the transient voltages caused by commutation. Where it is not possible to bond one of the brushes directly to the metalwork, it should be decoupled in the same way as the other brush. A low self-inductance is very important for these bonds and decoupling, and even 5mm of length or distance can be crucial. Feedthrough capacitors of around 1nF, screwed into the body of a fully metal enclosed motor and used as the brush terminals to the motor cable, often work very well indeed, although they are not inexpensive.
Correct application of shielding, bonding, and decoupling, may make the motor’s emissions low enough. If not, the next step is to add chokes to the brush leads, as close as possible to the decoupling capacitors but immediately outside the motor body. Differential chokes and across-the-line capacitors may be needed to reduce low-frequency emissions, whereas common-mode chokes and line-to-chassis capacitors are usually best at suppressing high-frequency emissions. A multi-stage filter using both types of choke may be needed in difficult cases, and is often best implemented with a PCB mounted directly on the motor end-cap at the commutator end, to keep all lead lengths low and to permit low-inductance bonding of capacitors to the motor body.
A varistor-disc motor with only very tiny sparks on its commutator should last longer before it commutator wears out, whereas a shielded and filtered motor will not benefit in this way because its sparks have not been made any smaller.
Like commutator motors, electric bells create emissions from DC to daylight. The best technique is to remove their spark gap and use an oscillator (astable) circuit to pulse current through the hammer solenoid at the hammer’s natural frequency. This is usually very much cheaper than any filtering methods. Such an electric bell could be much more reliable, and of course would require no adjustment to its spark-gap during manufacture. It may be that this is the first significant improvement in the design of electric bells since the 1880s.
EN 61000-3-2 came into force on 1/1/2001 under the EMC Directive for all equipment consuming up to 16Amps/phase and connected to public low-voltage mains supplies. It limits the harmonic (non-sine wave) currents drawn by products, for all lighting equipment consuming above 25W and all other products consuming above 75W. Professional equipment rated at over 1kW has no limits to meet at the time of writing.
The problem for typical rectifier-capacitor AC-DC power converters is that they appear to the power distribution as non-linear loads because they only top up their DC storage capacitors at the peaks of the AC supply waveform. Their supply currents are discontinuous, non-sine wave, and rich in harmonics (as shown by Figure 6V).
The special problem for single-phase power supplies is that they emit triple (or triplen) harmonics (3rd, 9th, 15th, etc.), which are a particular nuisance since they add linearly in neutral conductors (no cancellation) and are a major cause of cable and transformer overheating.
In a larger installation with a lot of single-phase electronic loads (typical of a modern office) the neutral currents can reach over 1.7 times the size of the phase currents. Since many older buildings are wired with half-size neutrals, and since building neutrals aren’t fused, the fire hazard is clear.
Harmonic emissions create a number of problems for power generation and distribution, not least of which is overheating and fire (something that fire insurers are becoming increasingly aware of). There are a number of ways of dealing with this problem at the equipment and installation levels. Electronic solutions at the equipment level are the main concern here.
There are many other non-linear loads which also cause harmonic currents in the supply, such as transformers and motors; arc furnaces and welding equipment. Fluorescent lamps with magnetic ballasts have harmonic emissions too, and although they include even-order harmonics they usually don’t extend to very high frequencies. High-frequency ballasts for fluorescent lamps (including the popular ‘low energy’ filament bulb replacement products) are simply single-phase AC-DC switch-mode power supplies – with all their harmonic problems. Three-phase power converters (sometimes called 6-pulse converters) are also a source of harmonic emissions, but if operated with balanced loads they produce low triplen levels.
When an item of equipment draws (‘emits’ in EMC terminology) harmonic currents from a sine-wave AC supply, the harmonic currents are reactive and increase the VA consumption of the equipment without affecting its consumption when measured in Watts. The ratio of Watts to VA consumed by a load is known as its Power Factor (PF), so where an equipment has significant emissions of harmonics it also has a poor power factor.
A PF of 1 means that the Watts consumed equals the VA of the equipment, in which case it looks like a pure resistive load and has no harmonic emissions. AC-DC power converters with no harmonic reduction techniques tend to have PFs of around 0.6. Techniques which reduce the emissions of harmonic currents into the AC supply also improve the equipment’s PF, so they are usually called Power Factor Correction (PFC) techniques.
Don’t confuse real Power Factor (= W / VA) with the power factor traditionally used by electrical generation and distribution engineers, which is the cosine of the angle between the sine-wave supply voltage and the load current and can be adjusted by either adding capacitance or inductance to a power line. The electrical engineers’ traditional PF is based on sine wave voltages and linear loads (resistive, inductive, or capacitive) and so is actually a special case of real PF. Few, if any, power distributions these days have linear loads, and you cannot correct the PF of an electronic AC-DC power supply using the traditional methods for linear loads.
There are a number of techniques for reducing harmonic emissions (improving PF) for items of electronic equipment:
· Filtering
· Passive PFC using an inductor between bridge rectifier and DC storage capacitor
· Passive PFC using a charge pump with a suitable SMPS controller
· Active PFC using a boost regulator after the bridge rectifier
· Increasing 3f rectifiers to 6f
Filtering means simply connecting filters at the AC input to the power converter to limit the emissions of some or all of the harmonics. Because the frequencies are so low, and the currents involved are often measured in Amps, these filters can be physically large, heavy, and expensive.
Small linear power supplies have relatively high impedances in their mains transformers, which spreads their pulses of supply current in time and so reduces their harmonic content. They sometimes meet the harmonic limits without modification. As power transformers get bigger their impedance drops and the resulting current pulses into their bridge rectifiers are sharper and contain more troublesome harmonics. Larger linear supplies therefore emit harmonic currents as readily as do switch-mode power supplies, which no transformer between their bridge rectifier and their unregulated DC storage capacitor.
One solution is to add a series inductance either before or after the bridge rectifier, as in Figure 6W. This widens the conduction angle of the rectifiers and so reduces their harmonic emissions. The lowest harmonics are realised when the choke has a very large inductance, but these can be comparable in size with a mains transformer rated for the product’s full power.
It appears that single-phase rectifiers with constant-inductance choke input filters can be designed to meet the toughest harmonic limits in EN 61000-3-2 for power ratings <1500W. Choke values of 7mH to 70mH are approximately required for direct-on-line voltage-doubler rectifiers, the higher values applying at lower powers and currents. Transformers with the requisite inductance could be made, to save adding a separate component for applications where low-voltage high-current supplies are required and switch-mode techniques are not favoured.
The inductor in series with the unregulated DC capacitor resists rapid changes in current, so makes the rectifiers conduct for longer – reducing harmonic currents. The inductor must be sized so its flux does not become discontinuous at any point during a cycle. As the inductor is used on DC, care must be taken to ensure it does not saturate. The large air-gaps used to prevent saturation will emit quite strong ‘hum’ magnetic fields locally and if the inductor is not shielded these may affect the placement of other devices and the routing of cables and PCB traces nearby. It may be possible to adapt this circuit to put the inductor on the AC side, so that it sees an AC current and saturation is less of a problem.
With the right kind of switch-mode power supply (SMPS) controller it is possible to use passive ‘charge pump’ circuitry to correct power factor and reduce harmonics, as shown in Figure 6X.
An example is the Infineon TDA 1684X family (e.g. TDA 16846, TDA16847).Their load dependent frequency response for free running SMPS’s allows them to control sinusoidal switched mode power supplies featuring a PFC charge pump circuit (as well as provide a low power standby operation). For more on this technique refer (for example) to the Infineon Application Note: AN- TDA 1684X (version 1.2, dated June 2000).
Traditional SMPS converters do not have the charge-pump circuitry shown in the dotted box of Figure 6W. With the charge-pump circuitry, the current (IL) in the charge pump choke (L) becomes higher with a longer switch-on time of the switching device (T). The controller operates so as to make T’s switch-on time increase with a decreasing mains voltage, thereby increasing the current through the PFC charge pump circuit. With suitably dimensioned components this makes the load on the AC mains appear resistive. The additional stored energy in capacitor Cp makes the SMPS capable of stabilising the ripple voltage during mains zero crossings.
Adding the charge-pump circuitry does not worsen any other EMC emissions (unlike the ‘Active PFC’ technique described later). But it does increase the dissipation in T by a significant amount and so can worsen efficiency and make larger devices and/or heatsinks necessary. The waveforms of Figure 6Y show the various currents and voltages in the charge pump and SMPS circuit for an example cycle, to help understand the operation of the circuit.
Different waveforms obtain under different conditions of supply voltage and load current, although they still retain the same principle (for more on this refer to the Infineon Application Note: AN- TDA 1684X (version 1.2, dated June 2000).
The charge pump’s capacitor current peaks when the voltage across T is high. This considerably increases the dissipation in the switching transistor T, when compared with the same converter without the charge pump, requiring larger devices and/or larger heat sinks for reliable operation. However, it is possible to feed the charge pump’s capacitor from a separate winding on the transformer instead of from the switched primary winding. This can be designed in such a way as to reduce the dissipation in T to close to what it would be without a charge pump – saving cost and PCB area.
The ‘Active PFC’ technique interposes a switch-mode boost converter between the bridge rectifier and the storage capacitor, as shown in Figure 6Z. It boosts the full-wave rectified supply when it is lower than the voltage on the storage capacitor.
The rectified mains voltage is boosted under the control of an IC that causes the current into the storage capacitor to approximate a full-wave rectified sine wave. Thus the storage capacitor plus active PFC circuit looks like a resistive load to the bridge rectifier, so the whole lot appears to the AC supply like a resistor (although with little notches around zero-crossings due to the rectifiers). The boost circuit typically operates at a high frequency, even several MHz, so a filter capacitor (usually around 1mF) is required to convert the fast-switching current pulses into a reasonably-looking rectified sine wave.
Active PFC circuits have a time constant of around 0.5 seconds to smooth out load current fluctuations so that their AC supply’s simulated resistance appears to change in value slowly, not to cause harmonic emissions. This needs to be taken into account when designing the size of the DC storage capacitor. The PFC boost circuit used generates high-frequency conducted and radiated emissions which need suppressing. Usually these boost circuits are added to existing switch-mode power supplies where some filtering and shielding will already be in place, although they are likely to need modifying.
Figure 6AA is taken from an EPCOS application note on their active PFC products. The waveforms shown for VC, IL, VE and IE correspond to the points marked on the overview circuit on Figure 6Z. Figure 6AA shows how the inductor current is pulse-width-modulated and smoothed by the input capacitor to simulate a sine-wave input current. The voltage on the DC storage capacitor will have 100Hz ripple on it (or 120Hz).
Most PFC circuits will be followed by voltage regulation, usually another switch-mode converter. Control ICs are now becoming available that will manage both functions: PFC and regulation.
Figure 6BB shows an example application circuit for an EPCOS PFC controller, complete with input filtering and surge protection.
Active PFC has many advantages apart from helping to meet harmonic emissions standards, including...
Universal input
Having active PFC boost circuitry makes it easier to design power supplies which will cope with a wide range of supply voltages, such as 85 to 264V, DC to 400Hz, allowing operation from the public low-voltage mains (and some aircraft supplies) anywhere in the world.
This reduces the need for ‘country variants’, reduces stockholding, and allows faster order fulfilment.
Full power available from wall sockets
Active PFC can draw full power from wall sockets (e.g. 3kW from a UK 230V 13A plug, 1700W from a US 110V 15A plug). Previously they were typically limited to < 50% of the possible power before the fuse in the plug or the building’s over-current protection operated.
Waveform insensitivity
Some countries, especially in the developing world, can have very badly distorted supply waveforms. ‘Traditional’ power supplies are sensitive to crest factor (the ratio of peak to RMS) so do badly on distorted supplies. Active PFC helps cope with these situations.
Universal-input PFC supplies help cope with dips, sags, brownouts, and dropouts
Mains supplies everywhere suffer from dips, sags, brownouts, and dropouts. These can cause microprocessors to reset frequently, causing annoyance to users even if they don’t cause data to be lost or control systems to go out of control. A universal-input power converter running on a nominal 230V will run perfectly well, keeping its storage capacitors fully charged, on -50% mains.
The harmonic emissions from three-phase power rectifiers (AC-DC converters) can be reduced by making them six-phase. Six-phase rectifiers are often called 12-pulse converters and use the fact that the phase outputs from star and delta wound 3f transformer secondaries are 60o apart, so use both star and delta secondaries each feeding a three-phase rectifier which feeds into the DC rail, as shown by Figure 6CC.
Like three-phase converters, six-phase converters naturally have low levels of triplen harmonics (into balanced load) but the conversion to six-phase reduces their emissions of 5th and 7th harmonics too. A series inductor can be added, in the AC or DC sides, to reduce the harmonics even more (these are often called ‘line reactors’ or ‘DC link reactors’) and operate in the same way as the series inductors described earlier, but as they do not have to deal with very high levels of current ripple their design constraints are eased.
Three-phase ‘boost’ rectifiers use PWM control of power transistors (usually IGBTs) instead of ordinary rectifiers, using ‘rectifier’ circuits similar to that of Figure 6DD. When the switching rate of the PWM is much faster than 50Hz the switching functions can be replaced by their average values in the switching period.
By appropriate control of the PWM switching patterns these boost rectifiers can achieve a near sine-wave input current from the mains supply, hence PFC. They can also achieve bi-directional power transfer. They have a tendency to become unstable which is usually overcome by using a large value for the DC capacitor.
There are mains harmonic reduction techniques which can be used at the level of the system or installation, but of course they don’t help equipment meet EN 61000-3-2.
Some harmonic problems in systems and installations are simply dealt with by up-rating conductors (neutrals may need double the cross-sectional area of the phase conductors) and transformers.
Star-delta transformers help remove the triplen harmonics typical of single-phase mains distributions. Their loads must be well balanced and they must be rated for the appropriate levels of zero-phase magnetic flux in their delta windings
Filters (series and parallel types) can restrict the flow of harmonic currents in areas of the mains distribution. Filters are ‘tuned’ to each problem frequency, and can interact unpredictably with existing network resonances. Filters should be left to power system harmonics experts.
‘Active filters’ are now available which are much easier to use than passive filters. A better name for these types of equipment is ‘active harmonic conditioners’ as they are not really filters but ‘anti-harmonic generators’ that take energy from the mains supply and add it back in as necessary to preserve its sine waveform. They only use the energy associated with their internal inefficiencies, since on over periods of one second or more the power they need to extract to preserve the waveshape is exactly equal to the power they need to inject.
Uninterruptible power supplies (UPSs) can be used providing they themselves have sufficiently low levels of harmonic emissions into their mains supply (check manufacturers’ data carefully).
Motor-generator sets are a traditional way of confining harmonic currents to mains conductors which are isolated form the public supply.
· Operating equipment from its own low-voltage distribution transformer is another traditional technique. The low impedance at the common point of connection to the high voltage supply reduces the effect of the harmonic currents on the public low-voltage network. |
